{ "schema_version": "1.0.0", "catalogue": "Digital Fly Lab project catalogue", "version": "2026-10-06-run9", "publisher": "Digital Fly Lab (independent Shaduf resource)", "last_verification": "2026-10-06", "record_count": 116, "schema": "projects.schema.json", "projects": [ { "id": "banc", "name": "BANC brain-and-nerve-cord connectome", "type": "dataset", "author_or_org": "Lee lab (Harvard) and the BANC community; Bates, Phelps, Kim, Yang et al.", "summary": "Connectome of one adult female fly's brain and ventral nerve cord in a single animal, so signals can be traced from sensors through the neck to motor circuits. The linked repository holds the paper's analysis code and a metadata snapshot.", "claim": { "text": "the first synapse-resolution connectome that unites the brain and ventral nerve cord of an animal", "url": "https://github.com/htem/BANC-project" }, "dataset": "BANC", "release": "v888 (Codex, 2026-05-20); paper analyses on v626", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "Dataset entry: README of htem/BANC-project (licence section, Dataverse deposit) [direct]; Codex tile and FAQ [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://codex.flywire.ai/banc", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "try_note": "Codex (Google sign-in)", "code_url": "https://github.com/htem/BANC-project", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit e31a2e26b9); GitHub API spdx_id=None", "data_licence": "CC-BY-4.0", "platform": [ "browser", "Linux", "macOS", "Windows" ], "gpu": "none", "download_size": "Harvard Dataverse deposit: 379 files, 536.1 GB in total, of which 277 influence-matrix chunks (308.3 GB) are available on request; the v3 edge list is 359 MB and the neuron metadata 58 MB.", "last_commit": { "date": "2026-07-20T18:48:11-04:00", "hash": "e31a2e26b9937dca72e5ca1c1960df6454d76114", "branch": "main" }, "pushed_at": "2026-07-20T22:48:43Z", "last_release": { "tag": "v2.0.5", "date": "2026-05-23T03:05:35Z", "url": "https://github.com/htem/BANC-project/releases/tag/v2.0.5" }, "created_at": "2025-07-31T01:37:26Z", "stars": 14, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41586-026-10735-w", "peer_review": "peer-reviewed", "link_status": { "url": "https://codex.flywire.ai/banc", "status": "redirect", "http_code": 200, "final_url": "https://codex.flywire.ai/?dataset=banc", "checked_at": "2026-10-06T09:40:02.041Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-026-10735-w", "status": "blocked", "http_code": 200 }, { "url": "https://github.com/htem/BANC-project", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/htem/BANC-project [direct]", "https://codex.flywire.ai/faq [direct]", "https://www.virtualflybrain.org/docs/concepts/neuron-counts/ [direct]" ], "notes_limitations": "Counts disagree by scope: 158,262 (Codex v888), 155,916 (paper, v626-based), README 'approximately 188,000' (scope not stated). Lamina and ocellar ganglion absent. README says repo code is CC BY 4.0 but no LICENSE file exists at HEAD." }, { "id": "fanc", "name": "FANC female adult nerve cord connectome", "type": "dataset", "author_or_org": "Lee lab (Harvard), Tuthill lab (UW) and the FANC community", "summary": "Connectome reconstruction of an adult female fly's ventral nerve cord. The newest reconstruction is shared with community members only; the linked repository is the Python toolkit for working with it.", "claim": { "text": "FANC (pronounced \"fancy\") is the Female Adult Nerve Cord, a GridTape-TEM dataset of an adult Drosophila melanogaster's ventral nerve cord", "url": "https://github.com/htem/FANC_auto_recon" }, "dataset": "FANC", "release": "not stated (tool package release v3.2.3, 2026-05-01)", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "Dataset entry: htem/FANC_auto_recon README (access restriction) [direct]; data availability and counts [search-summary]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "try_note": "No open browser access to the latest segmentation; EM images viewable via BossDB (search-summary)", "code_url": "https://github.com/htem/FANC_auto_recon", "code_licence": "GPL-3.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 89d9769457; GitHub API spdx_id=GPL-3.0", "data_licence": "no open licence found (community access rules)", "platform": [ "Linux", "macOS" ], "gpu": "none", "download_size": "No public download: access to the latest reconstruction is restricted to authorised users.", "last_commit": { "date": "2026-05-06T16:24:33+02:00", "hash": "89d9769457583051e3ec9ec7b34a5f3575537bf5", "branch": "main" }, "pushed_at": "2026-05-06T14:24:37Z", "last_release": { "tag": "v3.2.3", "date": "2026-05-01T12:54:04Z", "url": "https://github.com/htem/FANC_auto_recon/releases/tag/v3.2.3" }, "created_at": "2020-08-04T18:28:44Z", "stars": 16, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41586-024-07389-x", "peer_review": "peer-reviewed", "link_status": { "url": "https://github.com/htem/FANC_auto_recon", "status": "ok", "http_code": 200, "final_url": "https://github.com/htem/FANC_auto_recon", "checked_at": "2026-10-06T09:40:02.683Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-024-07389-x", "status": "blocked", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/htem/FANC_auto_recon [direct]", "https://doi.org/10.1038/s41586-024-07389-x [direct: DOI resolves]" ], "notes_limitations": "Restricted: 'Access to the latest reconstruction of FANC is restricted to authorized users.' Counts (~14,600 cell bodies, ~45M synapses) only seen via search summary. For an open female nerve cord use BANC." }, { "id": "flywire-fafb", "name": "FlyWire FAFB whole-brain connectome", "type": "dataset", "author_or_org": "FlyWire Consortium (Princeton: Murthy and Seung labs; Janelia; Cambridge and others)", "summary": "Proofread wiring diagram of one adult female fruit fly brain, including both optic lobes but not the nerve cord. It is the data behind the Shiu whole-brain model and most FlyWire-based demos.", "claim": { "text": "Whole-Brain Connectome of an adult female Drosophila.", "url": "https://flywire.ai/" }, "dataset": "FlyWire FAFB", "release": "v783 (public release, Oct 2023 snapshot); v630 used by the original Shiu model", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "Dataset entry: counts and versions from Codex dataset tiles and FAQ [direct] https://codex.flywire.ai/api/download ; licence from https://flywire.ai/guidelines [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://codex.flywire.ai/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "try_note": "Codex explorer; Google sign-in required for apps and downloads", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "Official v783 connectivity files on Zenodo (doi:10.5281/zenodo.10676866): 5 files, 10.6 GB in total; the connection table is 852 MB and the full synapse table 9.5 GB. The Shiu model's copies: Connectivity_783.parquet 101 MB, Completeness_783.csv 3.3 MB.", "last_commit": "n/a", "pushed_at": "n/a", "last_release": { "tag": "v783", "date": "2023-10 (snapshot date per flywire.ai/guidelines)" }, "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": "https://doi.org/10.1038/s41586-024-07558-y", "peer_review": "peer-reviewed", "link_status": { "url": "https://codex.flywire.ai/", "status": "ok", "http_code": 200, "final_url": "https://codex.flywire.ai/", "checked_at": "2026-10-06T09:40:02.370Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-024-07558-y", "status": "blocked", "http_code": 200 }, { "url": "https://flywire.ai/", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://flywire.ai/guidelines [direct]", "https://codex.flywire.ai/api/download [direct]", "https://codex.flywire.ai/faq [direct]", "https://www.virtualflybrain.org/docs/concepts/neuron-counts/ [direct]", "https://doi.org/10.1038/s41586-024-07558-y [direct: DOI resolves]" ], "notes_limitations": "139,255 neurons in v783 (Codex); 127,400 in v630 (Shiu model); 138,639 in the Shiu repo v783 file. Codex counts a connection at 5+ synapses (3,732,460 connections). Data is non-commercial (CC BY-NC 4.0). Brain only: no ventral nerve cord." }, { "id": "hemibrain", "name": "Hemibrain connectome", "type": "dataset", "author_or_org": "Janelia FlyEM and Google Connectomics", "summary": "Earlier dense connectome of a large part of one female fly's central brain (no optic lobes, no nerve cord), including the mushroom body and central complex. Still widely used for learning and navigation circuits.", "claim": { "text": "This connectome reconstruction contains around 25,000 neurons", "url": "https://www.janelia.org/project-team/flyem/hemibrain" }, "dataset": "hemibrain", "release": "v1.2.1", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "Dataset entry: Janelia hemibrain page (licence, count) [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://neuprint.janelia.org/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "try_note": "neuPrint explorer", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY (version not stated)", "platform": [ "browser" ], "gpu": "none", "download_size": "Compact connection summary (v1.2 release, CSV files in a tar.gz): 46 MB.", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": "https://doi.org/10.7554/eLife.57443", "peer_review": "peer-reviewed", "link_status": { "url": "https://neuprint.janelia.org/", "status": "ok", "http_code": 200, "final_url": "https://neuprint.janelia.org/", "checked_at": "2026-10-06T09:40:02.914Z", "other_links": [ { "url": "https://doi.org/10.7554/eLife.57443", "status": "blocked", "http_code": 406 }, { "url": "https://www.janelia.org/project-team/flyem/hemibrain", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://www.janelia.org/project-team/flyem/hemibrain [direct]", "https://www.virtualflybrain.org/docs/concepts/neuron-counts/ [direct]" ], "notes_limitations": "~25,000 neurons in part of one hemisphere's central brain; not a whole brain and should not be scaled up. The eLife DOI returns HTTP 406 to our script (bot protection); resolves in browsers." }, { "id": "larva-l1", "name": "Larval L1 brain connectome (Winding et al. 2023)", "type": "dataset", "author_or_org": "Winding, Pedigo, Barnes et al. (Cambridge, MRC LMB, Janelia, Johns Hopkins)", "summary": "Complete synapse-level wiring of the brain of a first-instar fruit fly larva: about 3,000 neurons, small enough to run whole on any laptop. It is the usual choice for reservoir-computing experiments.", "claim": { "text": "The connectome of an insect brain", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC7614541/" }, "dataset": "larval L1", "release": "paper supplementary data (2023)", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "Dataset entry: PMC7614541 data availability and licence notice [direct]; counts via VFB page [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://catmaid.virtualflybrain.org/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "try_note": "CATMAID (L1 Larval CNS) per the paper's data statement", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "article CC BY 4.0; no separate data licence found", "platform": [ "browser" ], "gpu": "none", "download_size": "Not stated (paper supplementary files).", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": "https://doi.org/10.1126/science.add9330", "peer_review": "peer-reviewed", "link_status": { "url": "https://catmaid.virtualflybrain.org/", "status": "redirect", "http_code": 200, "final_url": "https://virtualflybrain.org/about/hosted/", "checked_at": "2026-10-06T09:40:03.865Z", "other_links": [ { "url": "https://doi.org/10.1126/science.add9330", "status": "blocked", "http_code": 403 }, { "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC7614541/", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://pmc.ncbi.nlm.nih.gov/articles/PMC7614541/ [direct]", "https://www.virtualflybrain.org/docs/concepts/neuron-counts/ [direct]" ], "notes_limitations": "3,016 neurons and ~548,000 synapses (brain only). The supplementary all-to-all matrix used by many projects has 2,952 nodes (third-party figure, not verified). Science DOI returns 403 to scripts (Cloudflare); PMC copy is open." }, { "id": "malecns", "name": "MaleCNS connectome (male central nervous system)", "type": "dataset", "author_or_org": "Janelia FlyEM with Cambridge (Zoology), MRC LMB and Google Research", "summary": "Proofread connectome of one adult male fly's whole central nervous system: brain, optic lobes and ventral nerve cord joined through the neck. Its September 2026 paper set off the wave of game and meme projects.", "claim": { "text": "With over 166,000 neurons and 125 million synaptic connections, this is the largest brain map by number of neurons to date", "url": "https://research.google/blog/a-connectomics-milestone-mapping-the-complete-male-fruit-fly-brain/" }, "dataset": "MaleCNS", "release": "v1.0 (2026-06-08); v0.9 (2025-10-05)", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "Dataset entry: release notes, licence and access route from https://male-cns.janelia.org/ , /release/ and /download/ [direct]; v1.0 count from Codex tile [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://neuprint.janelia.org/?dataset=male-cns%3Av1.0&qt=findneurons", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "try_note": "neuPrint explorer (account needed for API token); also Codex MCNS v1.0", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "Flat connectome in a public Google bucket: 11 files, 31.3 GB in total; connection weights between traced neurons 508 MB, all connection weights 1.1 GB.", "last_commit": "n/a", "pushed_at": "n/a", "last_release": { "tag": "v1.0", "date": "2026-06-08" }, "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": "https://doi.org/10.1016/j.cell.2026.08.015", "peer_review": "peer-reviewed", "link_status": { "url": "https://neuprint.janelia.org/?dataset=male-cns%3Av1.0&qt=findneurons", "status": "ok", "http_code": 200, "final_url": "https://neuprint.janelia.org/?dataset=male-cns%3Av1.0&qt=findneurons", "checked_at": "2026-10-06T09:40:03.078Z", "other_links": [ { "url": "https://doi.org/10.1016/j.cell.2026.08.015", "status": "redirect", "http_code": 200 }, { "url": "https://research.google/blog/a-connectomics-milestone-mapping-the-complete-male-fruit-fly-brain/", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://male-cns.janelia.org/ [direct]", "https://male-cns.janelia.org/release/ [direct]", "https://male-cns.janelia.org/download/ [direct]", "https://pmc.ncbi.nlm.nih.gov/articles/PMC12636603/ [direct]", "https://research.google/blog/a-connectomics-milestone-mapping-the-complete-male-fruit-fly-brain/ [direct]" ], "notes_limitations": "166,700 neurons in v1.0 (Codex); 166,691 in the preprint (v0.9); 165,122 = v1.0 neurons with status 'Traced', the subset many simulators load (counted by igdigitallab/bioreservoir from the official files). Official wording is 'licensed under CC-BY' linking to CC BY 4.0." }, { "id": "manc", "name": "MANC male adult nerve cord connectome", "type": "dataset", "author_or_org": "Janelia FlyEM, Cambridge Connectomics Group and Google Research", "summary": "Dense connectome of the ventral nerve cord of one adult male fly, the part that runs legs, wings and most motor output. Used for walking and descending-neuron studies.", "claim": { "text": "With about 23,000 neurons, 10 million pre-synaptic sites, and 74 million post-synaptic densities", "url": "https://www.janelia.org/project-team/flyem/manc-connectome" }, "dataset": "MANC", "release": "v1.2.1", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "Dataset entry: Janelia MANC page (licence, counts, access) [direct]; Codex tile [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://neuprint.janelia.org/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "try_note": "neuPrint explorer (account needed for API token); also Codex MANC v1.2.1", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY (version not stated)", "platform": [ "browser" ], "gpu": "none", "download_size": "Flat files in a public Google bucket (flyem-manc-exports); size not stated.", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": "https://doi.org/10.7554/eLife.97769.1", "peer_review": "preprint", "link_status": { "url": "https://neuprint.janelia.org/", "status": "ok", "http_code": 200, "final_url": "https://neuprint.janelia.org/", "checked_at": "2026-10-06T09:40:03.473Z", "other_links": [ { "url": "https://doi.org/10.7554/eLife.97769.1", "status": "blocked", "http_code": 406 }, { "url": "https://www.janelia.org/project-team/flyem/manc-connectome", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://www.janelia.org/project-team/flyem/manc-connectome [direct]", "https://codex.flywire.ai/api/download [direct]" ], "notes_limitations": "23,665 neurons in Codex v1.2.1. Nerve cord only (plus neck connective). eLife reviewed preprint; no version of record per VFB." }, { "id": "caveclient", "name": "CAVEclient", "type": "data-tool", "author_or_org": "CAVE developers (CAVEconnectome)", "summary": "Python client for the CAVE microservices that store versioned segmentation, annotation and synapse tables. It is used to query FlyWire, BANC, FANC and MICrONS data by materialization version or timestamp.", "claim": { "text": "This repository supplies client-side code to easily interact with the microservices in CAVE.", "url": "https://github.com/CAVEconnectome/CAVEclient" }, "dataset": "several", "release": "n/a (any CAVE datastack and materialization version)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "n/a (API client)" }, "trained_class": "not-assessed", "grade_basis": [ "caveclient/materializationengine.py - queries versioned annotation/synapse tables [direct]", "caveclient/chunkedgraph.py - segmentation (proofreading graph) queries [direct]", "pyproject.toml:21-24 - caveclient 8.2.1, requires-python >=3.9 [direct]", "README.md - describes CAVE and links the CAVE bioRxiv preprint [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/CAVEconnectome/CAVEclient", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit c57c15f55b; GitHub API spdx_id=MIT", "data_licence": "several (see dataset entries)", "platform": [ "Linux", "macOS", "Windows" ], "gpu": "none", "download_size": "not stated", "last_commit": { "date": "2026-07-10T21:20:20+00:00", "hash": "c57c15f55b6fe9c6dff5e8c6199df57b77b59c2d", "branch": "master" }, "pushed_at": "2026-09-05T21:26:02Z", "last_release": { "tag": "v8.2.1", "date": "2026-07-16T00:49:08Z", "url": "https://github.com/CAVEconnectome/CAVEclient/releases/tag/v8.2.1" }, "created_at": "2018-10-10T14:47:22Z", "stars": 40, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://www.biorxiv.org/content/10.1101/2023.07.26.550598v1", "peer_review": "preprint", "link_status": { "url": "https://github.com/CAVEconnectome/CAVEclient", "status": "ok", "http_code": 200, "final_url": "https://github.com/CAVEconnectome/CAVEclient", "checked_at": "2026-10-06T09:40:04.312Z", "other_links": [ { "url": "https://www.biorxiv.org/content/10.1101/2023.07.26.550598v1", "status": "blocked", "http_code": 429 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/CAVEconnectome/CAVEclient [direct]", "https://caveconnectome.github.io/CAVEclient/ [direct]" ], "notes_limitations": "Infrastructure, not graded. Needs an auth token for most datastacks. Not fly-specific (also used for mouse MICrONS). peer_review is set from the bioRxiv link in the README; a later journal version was not checked. OS list is inferred and not stated by the project." }, { "id": "connectome-interpreter", "name": "connectome_interpreter", "type": "data-tool", "author_or_org": "Yijie Yin et al.", "summary": "A Python library (on PyPI) that turns wiring diagrams into testable hypotheses: effective connectivity across several synapses, activation spreading and path finding, sized to run whole-brain (about 140,000 neurons) analyses on a laptop or in Colab.", "claim": { "text": "connectome_interpreter turns synaptic wiring diagrams into testable hypotheses about circuit function.", "url": "https://github.com/YijieYin/connectome_interpreter" }, "dataset": "several (FlyWire FAFB, MaleCNS and others through user data)", "release": "not tied to one release", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "README.md: purpose, pip install connectome-interpreter, Colab, bioRxiv preprint 10.1101/2025.09.29.679410 [direct]", "LICENSE: MIT [direct]" ], "grade_date": "2026-09-30", "measured_result": null, "try_url": "https://colab.research.google.com/drive/145Td8_fFTPwTsDdEkQGAgdZ7CFQ8qXnr", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/YijieYin/connectome_interpreter", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit d212f86ef3; GitHub API spdx_id=MIT", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0; MaleCNS: CC-BY-4.0", "platform": [ "Colab", "Linux", "macOS", "Windows" ], "platform_note": "Python package", "gpu": "none", "download_size": "pip package; connectome data supplied by the user", "last_commit": { "date": "2026-08-27T11:17:48+01:00", "hash": "d212f86ef32318657dc27fa65e7e8ff604bffac9", "branch": "main" }, "pushed_at": "2026-08-27T10:17:58Z", "last_release": { "tag": "v2.9.0", "date": "2025-06-17T17:12:41Z", "url": "https://github.com/YijieYin/connectome_interpreter/releases/tag/v2.9.0" }, "created_at": "2024-02-16T02:59:30Z", "stars": 37, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": "https://www.biorxiv.org/content/10.1101/2025.09.29.679410v2", "peer_review": "preprint", "link_status": { "url": "https://colab.research.google.com/drive/145Td8_fFTPwTsDdEkQGAgdZ7CFQ8qXnr", "status": "ok", "http_code": 200, "final_url": "https://colab.research.google.com/drive/145Td8_fFTPwTsDdEkQGAgdZ7CFQ8qXnr", "checked_at": "2026-10-06T09:40:03.895Z", "other_links": [ { "url": "https://github.com/YijieYin/connectome_interpreter", "status": "ok", "http_code": 200 }, { "url": "https://www.biorxiv.org/content/10.1101/2025.09.29.679410v2", "status": "blocked", "http_code": 429 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/YijieYin/connectome_interpreter [direct: clone HEAD d212f86, 2026-09-30]" ], "notes_limitations": "Infrastructure, not graded for behaviour. Analysis library, not a simulator of spiking dynamics." }, { "id": "fafbseg", "name": "fafbseg", "type": "data-tool", "author_or_org": "natverse (Gregory Jefferis and contributors)", "summary": "An R package for working with the FlyWire reconstruction of the FAFB adult female brain. It sets up access tokens, downloads the canned FlyWire connectivity and cell-type release data (versions 783 and 630), queries the CAVE materialisation service, fetches meshes, skeletons and synapses, and plugs into the natverse toolkit. It is one building block of coconatfly, which the authors recommend as the easier starting point.", "claim": { "text": "Support for analysis of segmented EM data, focussed on the full adult female brain (FAFB) dataset, with FlyWire as the principal target.", "url": "https://github.com/natverse/fafbseg" }, "dataset": "FlyWire FAFB", "release": "FlyWire public releases 783 and 630 (download_flywire_release_data); also live FlyWire production data through CAVE; legacy Google FAFB segmentation", "evidence_grade": "n/a", "grade_note": "The file citations behind this entry were spot-read by us, not re-verified line by line.", "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "DESCRIPTION:1-6 R package fafbseg 0.15.19, 'Support Functions for Analysis of FAFB EM Segmentation' [direct]", "DESCRIPTION: License GPL-3; LICENSE.md is the GNU GPL v3 text [direct]", "R/release-data.R:248 download_flywire_release_data(which=c('core','all'), version=c(783L, 630L)) [direct]", "R/flytable.R:1388-1393 static data supported for versions 630 and 783, default 783 [direct]", "NEWS.md:1-20 0.15.19 adds flywire_synapse_query() and batched L2/supervoxel lookups [direct]", "README.md: FlyWire token setup, simple_python(), recommends coconatfly for most users [direct]" ], "grade_date": "2026-09-29", "measured_result": "Not applicable (analysis library; no behaviour or model).", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/natverse/fafbseg", "code_licence": "GPL-3.0", "code_licence_source": "LICENSE.md is the GPL v3 text in Markdown (our file heuristic does not recognise it); DESCRIPTION declares GPL-3; checked 2026-09-29", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux", "macOS", "Windows" ], "platform_note": "R package installed with natmanager; some functions need a Python environment set up by simple_python()", "gpu": "none", "download_size": "package is small; canned FlyWire release data are downloaded separately (size not stated)", "last_commit": { "date": "2026-10-02T20:16:50+01:00", "hash": "4fc26c98dd5641f414a6d955090100a404fa5afc", "branch": "master" }, "pushed_at": "2026-09-28T14:13:36Z", "last_release": { "tag": "fafbseg 0.15.17", "date": "2026-09-11T22:38:31Z", "url": "https://github.com/natverse/fafbseg/releases/tag/v0.15.17" }, "created_at": "2018-07-08T10:32:55Z", "stars": 14, "stars_date": "2026-09-29", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-29", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "peer_review_note": "software package; the FlyWire data papers it serves are peer-reviewed", "link_status": { "url": "https://github.com/natverse/fafbseg", "status": "ok", "http_code": 200, "final_url": "https://github.com/natverse/fafbseg", "checked_at": "2026-10-06T09:40:04.553Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/natverse/fafbseg [direct: clone HEAD a21d33f, 2026-09-28]" ], "notes_limitations": "A data-access and analysis library, not a model or demo. Needs a FlyWire access token for live data. Lifecycle badge says 'experimental', but the package is long-running and actively maintained (last commit 2026-09-28). The authors point most new users to coconatfly instead." }, { "id": "fly-connectome-data-tutorial", "name": "Fly Connectome Data Tutorial (SJCABS)", "type": "data-tool", "author_or_org": "Sven Dorkenwald & Alexander Bates (SJCABS winter school)", "summary": "Course material in Python notebooks and R Markdown for loading, analysing and plotting fly connectome data. It covers BANC, FAFB, MANC, hemibrain and male CNS through harmonised tables in a Google Cloud bucket. Optional notebooks add a simple LIF model and a weighted linear model.", "claim": { "text": "We will work with all the major, dense connectome datasets for the fruit fly.", "url": "https://github.com/sjcabs/fly_connectome_data_tutorial" }, "dataset": "several", "release": "BANC 888, FAFB 783, MANC 1.2.1, hemibrain 1.2.1, MaleCNS 0.9 (from file names)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "n/a (tutorial)" }, "trained_class": "not-assessed", "grade_basis": [ "README.md:49,63,75,87,99 - sections for BANC, Male CNS, FAFB, MANC, Hemibrain [direct]", "README.md:277,294,308,319,330 - data files banc_888, fafb_783, manc_121, hemibrain_121, malecns_09 [direct]", "python/fly_connectome_01..05_*.ipynb and R/01..05_*.Rmd - access, morphology, connectivity, indirect connectivity, transcriptomics tutorials [direct]", "python/fly_connectome_06_LIF_model.ipynb - teaching LIF example (activates sugar GRNs, reads descending neurons) [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/sjcabs/fly_connectome_data_tutorial", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 85c66244cf; GitHub API spdx_id=MIT", "data_licence": "several (see dataset entries)", "platform": [ "Linux", "macOS", "Windows" ], "gpu": "none", "download_size": "varies: metadata/edgelists ~10-500 MB per dataset; full synapse tables 1-10 GB each", "last_commit": { "date": "2026-06-07T10:32:12-04:00", "hash": "85c66244cfb2c02b2442905c4428b42a0d8e6656", "branch": "main" }, "pushed_at": "2026-06-07T14:32:12Z", "last_release": { "tag": "v1.0.3", "date": "2026-05-23T02:54:29Z", "url": "https://github.com/sjcabs/fly_connectome_data_tutorial/releases/tag/v1.0.3" }, "created_at": "2025-12-11T17:58:52Z", "stars": 63, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/sjcabs/fly_connectome_data_tutorial", "status": "ok", "http_code": 200, "final_url": "https://github.com/sjcabs/fly_connectome_data_tutorial", "checked_at": "2026-10-06T09:40:04.599Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/sjcabs/fly_connectome_data_tutorial [direct]" ], "notes_limitations": "Infrastructure/teaching material, not graded (the LIF notebook is a teaching example, not a claimed result). Small inconsistencies in the README: the BANC text cites ~114,000 neurons and Neuroglancer links for versions 626/746, while the data files are version 888 with 188,153 metadata rows; Male CNS is v0.9 here while v1.0 is now public. Needs gsutil / Google Cloud access for the data. The repo is large (~437 MB, many rendered HTML/PNG files). OS support is not stated." }, { "id": "fly-connectome-template", "name": "fly-connectome-template", "type": "data-tool", "author_or_org": "Mert Cobanov (cobanov)", "summary": "A React/Three.js/Vite browser template that shows 124,289 measured MaleCNS v1.0 soma positions next to a static Flybody mesh and an exchangeable environment panel. Users can load their own model's activity as JSON replays keyed by MaleCNS body IDs. It contains no neural model, synaptic graph or physics.", "claim": { "text": "A browser workbench for building your own fly-connectome experiments: real anatomy, a replaceable environment, and model outputs mapped by neuron ID.", "url": "https://github.com/cobanov/fly-connectome-template" }, "dataset": "MaleCNS", "release": "v1.0 (soma locations from body-annotations-male-cns-v1.0-minconf-0.5.feather, SHA-256 pinned)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "n/a", "neuron_model": "n/a", "input_mapping": "n/a", "output_mapping": "n/a", "trained_parts": "n/a", "body": "n/a", "scripted_parts": "n/a" }, "trained_class": "not-assessed", "grade_basis": [ "docs/MODEL-INTEGRATION.md:3-4 'a viewer and integration starting point, not a neural model' [direct]", "README.md:70-80 soma positions only, not a synaptic graph; Flybody is a surface mesh, not a physics simulation; no neural simulator included [direct]", "src/lib/replay.ts replay format validator (dataset male-cns:v1.0, source kind synthetic/predicted/measured) [direct]", "public/data/brain-atlas/NOTICE.md source URL, pinned hash and reproduction script scripts/build-brain-atlas.py [direct]", "LICENSE: Cobanov Template Attribution License 1.0, custom and not OSI-approved; UI and README credit required [direct]", "Re-checked 2026-09-29: HEAD unchanged since the 2026-09-28 grading (gh_meta-2026-09-29.json) [direct]" ], "grade_date": "2026-09-29", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/cobanov/fly-connectome-template", "code_licence": "Cobanov Template Attribution License 1.0 (custom; SPDX LicenseRef-Cobanov-Template-Attribution-1.0): use, modification and commercial deployment allowed; every web interface built on it must show a linked 'Built with fly-connectome-template by Mert Cobanov' attribution", "code_licence_source": "LICENSE file read at commit 38f5533 on 2026-09-29 (sections 1-3)", "data_licence": "Template code: Cobanov Template Attribution License 1.0 (custom source-available; a linked credit is required in both web UI and README). Bundled MaleCNS soma data: CC BY 4.0. Flybody mesh: Apache-2.0.", "platform": [ "browser" ], "platform_note": "browser (local Node.js 22.18+ build)", "gpu": "none", "download_size": "About 5 MB repository (atlas binaries ~2.4 MB, Flybody mesh ~1.7 MB) plus npm dependencies", "last_commit": { "date": "2026-09-12T12:26:44+03:00", "hash": "38f55332055328d38c29e72474c4ad5b6876101f", "branch": "main" }, "pushed_at": "2026-09-12T09:26:46Z", "last_release": "none", "created_at": "2026-09-12T09:17:25Z", "stars": 79, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/cobanov/fly-connectome-template", "status": "ok", "http_code": 200, "final_url": "https://github.com/cobanov/fly-connectome-template", "checked_at": "2026-10-06T09:40:04.605Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/cobanov/fly-connectome-template [direct: clone HEAD 38f5533]" ], "notes_limitations": "It makes no claim of fly-brain behaviour, so it is not graded. The bundled example activity (public/examples/model-output.example.json) contains authored synthetic values on real IDs, not simulated or measured activity. No hosted demo URL is given. The custom licence is stricter than MIT: the credit cannot be hidden. Same author as the catalogued 'fly-dino'. The Flybody mesh conversion comes from PinFly." }, { "id": "flybrainlab", "name": "FlyBrainLab", "type": "data-tool", "author_or_org": "FlyBrainLab (Lazar lab, Columbia University)", "summary": "An older interactive platform for exploring fly brain data and executing circuit models in a JupyterLab-based environment. The main repository is an installer and meta-package; its last default-branch commit is from September 2025.", "claim": { "text": "FlyBrainLab is an interactive computing platform for studying the function of executable circuits constructed from fruit fly brain data.", "url": "https://github.com/FlyBrainLab/FlyBrainLab" }, "dataset": "several (FlyWire FAFB, Hemibrain, larva and others through NeuroArch)", "release": "not tied to one release", "evidence_grade": "n/a", "grade_note": null, "mechanism": null, "trained_class": "not-assessed", "grade_basis": [ "README.md: platform description and install instructions [direct]", "LICENSE: BSD 3-Clause [direct]", "git log: last default-branch commit 5dd079a on 2025-09-28 [direct]" ], "grade_date": "2026-09-30", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/FlyBrainLab/FlyBrainLab", "code_licence": "BSD-3-Clause", "code_licence_source": "LICENSE file 'LICENSE' read at commit 5dd079afd0; GitHub API spdx_id=BSD-3-Clause", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0; hemibrain: CC-BY (version not stated)", "platform": [ "Linux", "macOS", "Windows" ], "platform_note": "JupyterLab-based platform", "gpu": "optional", "download_size": "not stated (conda environment plus several services)", "last_commit": { "date": "2025-09-28T20:20:11-04:00", "hash": "5dd079afd0ac0a39659082d5ac95c330d0fa6bde", "branch": "master" }, "pushed_at": "2025-09-29T00:20:15Z", "last_release": "none", "created_at": "2018-08-05T20:40:39Z", "stars": 109, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/FlyBrainLab/FlyBrainLab", "status": "ok", "http_code": 200, "final_url": "https://github.com/FlyBrainLab/FlyBrainLab", "checked_at": "2026-10-06T09:40:04.833Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/FlyBrainLab/FlyBrainLab [direct: clone HEAD 5dd079a, 2026-09-30]" ], "notes_limitations": "Infrastructure, not graded for behaviour. Maintenance is uncertain: the last commit on the main repository is a year old. We did not install it." }, { "id": "flywire-annotations", "name": "FlyWire neuron annotations", "type": "data-tool", "author_or_org": "flyconnectome (Jefferis lab, MRC LMB / University of Cambridge)", "summary": "Systematic neuron annotation tables for the FlyWire FAFB v783 release: cell class/type, hemibrain type, hemilineage, neurotransmitter prediction, side, and (from v3.0.0) dimorphism, fru/dsx and MaleCNS supertype. It accompanies Schlegel et al. 2024 (Nature) and is used by the fafbseg Python and R packages.", "claim": { "text": "Systematic neuron annotations and other data products for the 783 public release of the FlyWire female adult fly brain connectome.", "url": "https://github.com/flyconnectome/flywire_annotations" }, "dataset": "FlyWire FAFB", "release": "Annotations v3.1.0 (tag) on FlyWire materialization 783; v2.1.0 matches Schlegel et al. 2024", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "n/a", "neuron_model": "n/a", "input_mapping": "n/a", "output_mapping": "n/a", "trained_parts": "n/a", "body": "n/a", "scripted_parts": "n/a" }, "trained_class": "not-assessed", "grade_basis": [ "README.md:8-12 annotations for the FlyWire 783 release, first reported in Schlegel et al. Nature 2024 [direct]", "README.md:146-164 changelog: v3.1.0 JO retyping and fru_dsx update; v2.1.0 is the Nature 2024 version [direct]", "supplemental_files/README.md column definitions for Supplemental_file1-5 [direct]", "No LICENSE file; README has no licence statement (grep 'licen' = 0 hits) [direct]" ], "grade_date": "2026-09-28", "measured_result": "n/a (infrastructure)", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/flyconnectome/flywire_annotations", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit a83b2776d6); GitHub API spdx_id=None", "data_licence": "Not stated in the repository (no licence file or statement). The underlying FlyWire data is CC BY-NC 4.0 per flywire.ai (not checked here).", "platform": [ "other" ], "platform_note": "any (TSV/CSV files)", "gpu": "none", "download_size": "Main neuron annotation table (Supplemental_file1_neuron_annotations.tsv) 32 MB at commit a83b2776; skeletons and NBLAST scores are on Zenodo (10.5281/zenodo.10877326).", "last_commit": { "date": "2026-09-29T10:53:05+01:00", "hash": "a83b2776d60d5764cef36b927f5f9679c16c47a2", "branch": "main" }, "pushed_at": "2026-07-21T17:29:07Z", "last_release": { "tag": "Version 3.2.0", "date": "2026-09-29T09:56:31Z", "url": "https://github.com/flyconnectome/flywire_annotations/releases/tag/v3.2.0" }, "created_at": "2023-06-13T13:19:36Z", "stars": 76, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41586-024-07686-5", "peer_review": "peer-reviewed", "link_status": { "url": "https://github.com/flyconnectome/flywire_annotations", "status": "ok", "http_code": 200, "final_url": "https://github.com/flyconnectome/flywire_annotations", "checked_at": "2026-10-06T09:40:04.843Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-024-07686-5", "status": "blocked", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/flyconnectome/flywire_annotations [direct: clone HEAD 8587524, tag v3.1.0]", "https://doi.org/10.1038/s41586-024-07686-5 [indirect: cited in README]" ], "notes_limitations": "Infrastructure, not a model. The content has changed since the Nature paper: v3.x adds annotations cross-checked against MaleCNS (Berg et al., bioRxiv 2025, 'under revision' according to the 3.1.0 changelog), so pin a tag for reproducibility. The README says Codex shows a mix of annotation sources that may differ from these tables. The paper link in the README goes through a Cambridge library proxy; the plain DOI is used here. The repository states no licence." }, { "id": "flywire-codex", "name": "Codex (FlyWire Connectome Data Explorer)", "type": "data-tool", "author_or_org": "Murthy Lab, Princeton", "summary": "Flask web app behind codex.flywire.ai for searching neurons, cell types, connectivity, pathways and statistics in the FlyWire whole-brain connectome.", "claim": { "text": "Codex is a web application for exploring and analyzing neurons and annotations from the FlyWire Whole Brain Connectome.", "url": "https://github.com/murthylab/codex" }, "dataset": "FlyWire FAFB", "release": "v783 (only version in repo code)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "n/a (data browser)" }, "trained_class": "not-assessed", "grade_basis": [ "codex/data/versions.py:1-8 - only snapshot '783' (Oct 2023) listed; default and testing version 783 [direct]", "codex/data/local_data_loader.py:12,160-165 - loads and pickles data per snapshot version [direct]", "README.md:9-10,14 - web app for FlyWire; Python 3.9+; scripts/make_data.sh downloads data [direct]", "https://codex.flywire.ai/ - live site lists FAFB v783, BANC v888, MANC v1.2.1, MAOL v1.1, MCNS v1.0 [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://codex.flywire.ai", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/murthylab/codex", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 3477b6cfcc; GitHub API spdx_id=Apache-2.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "not stated", "last_commit": { "date": "2025-06-03T08:31:15-04:00", "hash": "3477b6cfccaf3bf976e6af0d40513ad221311a36", "branch": "main" }, "pushed_at": "2026-09-08T20:25:09Z", "last_release": "none", "created_at": "2022-07-31T23:41:11Z", "stars": 140, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41586-024-07558-y", "peer_review": "peer-reviewed", "link_status": { "url": "https://codex.flywire.ai", "status": "ok", "http_code": 200, "final_url": "https://codex.flywire.ai", "checked_at": "2026-10-06T09:40:04.751Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-024-07558-y", "status": "blocked", "http_code": 200 }, { "url": "https://github.com/murthylab/codex", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/murthylab/codex [direct]", "https://codex.flywire.ai/ [direct]" ], "notes_limitations": "Infrastructure, not graded. Conflict: the public repo (last commit 2025-06-03) serves only FlyWire v783, but the live site now also offers BANC v888, MANC v1.2.1, MAOL v1.1 and MaleCNS v1.0, so the live service likely runs newer, unpublished code. Sign-in is needed for most features. paper_url points to the FlyWire whole-brain paper (Dorkenwald et al. 2024), which describes Codex; there is no separate Codex paper." }, { "id": "natverse-malecns", "name": "malecns (R package)", "type": "data-tool", "author_or_org": "natverse / Gregory Jefferis (Cambridge) with Janelia FlyEM", "summary": "R package that gives natverse access to the Janelia male CNS connectome. It is a thin wrapper around malevnc and neuprintr, and adds metadata, meshes, skeletons and brain-space transforms.", "claim": { "text": "The goal of malecns is to provide natverse access to the whole male central nervous system dataset.", "url": "https://github.com/natverse/malecns" }, "dataset": "MaleCNS", "release": "male-cns:v1.0 (default; v0.9 also selectable)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "n/a (data access package)" }, "trained_class": "not-assessed", "grade_basis": [ "R/zzz.R:4,10 - default dataset option 'male-cns:v1.0' [direct]", "tests/testthat/test-dataset.R:2-12 - switching between male-cns:v0.9 and v1.0 [direct]", "DESCRIPTION:12-14 - thin wrapper around malevnc; License: GPL (>= 3) [direct]", "LICENSE.md:1-4 - full GNU GPL version 3 text in Markdown (595 lines) [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/natverse/malecns", "code_licence": "GPL-3.0-or-later", "code_licence_source": "LICENSE.md is the full GPL v3 text in Markdown (our heuristic missed it); DESCRIPTION says 'GPL (>= 3)'; GitHub API spdx_id=GPL-3.0", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS", "Windows" ], "gpu": "none", "download_size": "not stated", "last_commit": { "date": "2026-08-23T22:13:53+01:00", "hash": "daf8e2a9849cc77695b14bb6b9d4c02456cd3b3c", "branch": "master" }, "pushed_at": "2026-08-23T21:21:54Z", "last_release": { "tag": "malecns 0.4.2", "date": "2026-08-18T10:21:00Z", "url": "https://github.com/natverse/malecns/releases/tag/v0.4.2" }, "created_at": "2021-10-31T10:12:13Z", "stars": 717, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/natverse/malecns", "status": "ok", "http_code": 200, "final_url": "https://github.com/natverse/malecns", "checked_at": "2026-10-06T09:40:05.262Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/natverse/malecns [direct]", "https://natverse.org/malecns/ [search-summary]" ], "notes_limitations": "Infrastructure, not graded. Licence resolved: there is no LICENSE stub. LICENSE.md holds the full GPL v3 text in Markdown (usethis style), and DESCRIPTION:14 says 'License: GPL (>= 3)'. So the licence is GPL-3.0-or-later, which matches the GitHub API 'GPL-3.0'; our heuristic said 'custom/unrecognised' only because it did not recognise the Markdown-formatted text. Needs a neuPrint token, and some features need Clio auth. README marks the package as 'experimental'. OS list is inferred from a pure R package." }, { "id": "navis", "name": "NAVis", "type": "data-tool", "author_or_org": "navis-org (Philipp Schlegel et al.)", "summary": "Python library for loading, analysing, comparing and plotting neuron morphology (skeletons, meshes, dotprops), with NBLAST, transforms between brain templates, and loaders for neuPrint, MICrONS, H01 and others.", "claim": { "text": "NAVis is a Python 3 library for Neuron Analysis and Visualization.", "url": "https://github.com/navis-org/navis" }, "dataset": "several", "release": "n/a", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "n/a (analysis library)" }, "trained_class": "not-assessed", "grade_basis": [ "navis/interfaces/ - neuprint.py, microns.py, h01.py, neuromorpho.py, vfb.py, insectbrain_db.py, cave_utils.py loaders; neuron/ wraps the NEURON simulator [direct]", "setup.py:59-71 - GPLv3+, Python 3.10-3.13, python_requires >=3.10 [direct]", "README.md:27,63 - natverse .rds exchange; FlyWire tools live in the separate fafbseg package [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://colab.research.google.com/github/navis-org/navis/blob/master/examples/colab.ipynb", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/navis-org/navis", "code_licence": "GPL-3.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit cb9a5915b6; GitHub API spdx_id=GPL-3.0", "data_licence": "several (see dataset entries)", "platform": [ "Linux", "macOS", "Windows", "Colab" ], "gpu": "none", "download_size": "not stated", "last_commit": { "date": "2026-09-06T11:19:41+01:00", "hash": "cb9a5915b6b3587cb81154f4f77ffc62fe12b03a", "branch": "master" }, "pushed_at": "2026-09-06T10:30:49Z", "last_release": { "tag": "Version 1.12.0", "date": "2026-07-13T09:06:12Z", "url": "https://github.com/navis-org/navis/releases/tag/v1.12.0" }, "created_at": "2019-01-29T11:19:00Z", "stars": 136, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://colab.research.google.com/github/navis-org/navis/blob/master/examples/colab.ipynb", "status": "ok", "http_code": 200, "final_url": "https://colab.research.google.com/github/navis-org/navis/blob/master/examples/colab.ipynb", "checked_at": "2026-10-06T09:40:05.029Z", "other_links": [ { "url": "https://github.com/navis-org/navis", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/navis-org/navis [direct]", "https://navis-org.github.io/navis/ [search-summary]" ], "notes_limitations": "Infrastructure, not graded. FlyWire-specific access is in the separate fafbseg package. Cited via Zenodo DOI, no journal paper. GPL-3.0 (setup.py says GPLv3 or later). OS list is inferred." }, { "id": "neuprint-python", "name": "neuprint-python", "type": "data-tool", "author_or_org": "Janelia FlyEM (connectome-neuprint)", "summary": "Python client for the neuPrint connectome service. It runs queries for neurons, connections, synapses, ROIs and skeletons, and returns pandas tables. It also has a small passive single-neuron electrical simulation module.", "claim": { "text": "Python client utilties for interacting with the neuPrint connectome analysis service.", "url": "https://github.com/connectome-neuprint/neuprint-python" }, "dataset": "several", "release": "n/a (any neuPrint dataset, e.g. hemibrain:v1.2.1, manc:v1.2.3)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "n/a (query client)" }, "trained_class": "not-assessed", "grade_basis": [ "neuprint/client.py:20,190-193 - client examples use hemibrain:v1.2.1 and manc:v1.0 [direct]", "neuprint/queries/connectivity.py, neuprint/queries/neurons.py - query functions for connectivity and neuron tables [direct]", "neuprint/simulation.py:2-3 - SPICE-based passive simulation of one neuron from its skeleton and synapses [direct]", "setup.py:26 - python_requires >=3.9 (classifiers 3.9-3.12) [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/connectome-neuprint/neuprint-python", "code_licence": "BSD-3-Clause", "code_licence_source": "LICENSE file 'LICENSE' read at commit 25f7029051; GitHub API spdx_id=BSD-3-Clause", "data_licence": "several (see dataset entries)", "platform": [ "Linux", "macOS", "Windows" ], "gpu": "none", "download_size": "not stated", "last_commit": { "date": "2026-09-28T16:08:04-04:00", "hash": "25f70290517e8def275bd4d4dabde88108354911", "branch": "master" }, "pushed_at": "2026-07-20T22:01:39Z", "last_release": { "tag": "0.6.4", "date": "2026-09-28T20:10:42Z", "url": "https://github.com/connectome-neuprint/neuprint-python/releases/tag/0.6.4" }, "created_at": "2018-11-15T23:41:45Z", "stars": 82, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/connectome-neuprint/neuprint-python", "status": "ok", "http_code": 200, "final_url": "https://github.com/connectome-neuprint/neuprint-python", "checked_at": "2026-10-06T09:40:05.746Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/connectome-neuprint/neuprint-python [direct]" ], "notes_limitations": "Infrastructure, not graded. Needs a neuPrint API token. The simulation module models single neurons only (passive, linear), not networks. OS list is inferred from a pure-Python package and is not stated by the project. The neuPrint service has its own paper (Plaza et al. 2022), but this client has none." }, { "id": "virtual-fly-brain", "name": "Virtual Fly Brain (VFB)", "type": "data-tool", "author_or_org": "Virtual Fly Brain consortium (University of Edinburgh, University of Cambridge and partners)", "summary": "An online atlas and query hub for Drosophila neuroanatomy that brings light-microscopy images, neuron types and connectome data from several datasets into one browsable, searchable site. It offers a 3D browser, Python and R clients, and a Model Context Protocol (MCP) server so that LLM tools can query it.", "claim": { "text": "A hub for Drosophila melanogaster neural anatomy, imaging data and connectomics; query VFB from an LLM with the hosted VFB MCP server.", "url": "https://www.virtualflybrain.org/" }, "dataset": "several (hosts and cross-links FlyWire, hemibrain, MANC, FANC, BANC and light-microscopy datasets)", "release": "rolling (web service)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "n/a (atlas and query service)", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "n/a" }, "trained_class": "not-assessed", "grade_basis": [ "https://www.virtualflybrain.org/ home page: atlas scope, '699,318 registered 3D images from 153 datasets', workshop and MCP tool mentioned [direct, fetched 2026-09-28]", "https://github.com/VirtualFlyBrain organisation page: VFB3-MCP named as the canonical MCP server repository; hosted service needs no install or API key [direct, fetched 2026-09-28]", "GitHub API org repository list (2026-09-28): VFB3-MCP (MIT, pushed 2026-09-25), VFBquery (Apache-2.0), Docker-VFB-Neo4j-ProductionDB (CC-BY-4.0), geppetto-vfb (no SPDX match) [direct]" ], "grade_date": "2026-09-28", "measured_result": "n/a", "try_url": "https://www.virtualflybrain.org/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/VirtualFlyBrain/VFB3-MCP", "code_licence": "several: VFB3-MCP MIT; VFBquery Apache-2.0; database and ontology repositories CC-BY-4.0 (per GitHub licence detection)", "code_licence_source": "LICENSE file 'LICENSE' read at commit 5e1c3e280f; GitHub API spdx_id=MIT", "data_licence": "per source dataset (VFB aggregates data under the licences of the contributing datasets); not re-checked in this run", "platform": [ "browser", "other" ], "platform_note": "web browser; Python (VFB_connect, VFBquery); R (natverse); MCP clients", "gpu": "none", "download_size": "none for the web atlas; client libraries are small", "last_commit": { "date": "2026-09-25T16:51:00+01:00", "hash": "5e1c3e280f01b63515fb4bff635a0ce76e9d62a5", "branch": "main" }, "pushed_at": "2026-09-25T15:51:04Z", "last_release": { "tag": "Release 1.11.2 — per-caller GA4 client IDs and MCP client info tracking", "date": "2026-09-10T15:55:25Z", "url": "https://github.com/VirtualFlyBrain/VFB3-MCP/releases/tag/v1.11.2" }, "created_at": "2026-02-07T05:52:26Z", "stars": 9, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "peer-reviewed", "peer_review_note": "peer-reviewed (VFB platform papers; not re-checked in this run)", "link_status": { "url": "https://www.virtualflybrain.org/", "status": "ok", "http_code": 200, "final_url": "https://www.virtualflybrain.org/", "checked_at": "2026-10-06T09:40:05.238Z", "other_links": [ { "url": "https://github.com/VirtualFlyBrain/VFB3-MCP", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://www.virtualflybrain.org/ [direct]", "https://github.com/VirtualFlyBrain [direct]", "https://github.com/VirtualFlyBrain/VFB3-MCP [direct: clone HEAD 5e1c3e2]" ], "notes_limitations": "Infrastructure, not a digital-fly simulation: it helps you find and cross-reference neurons and connectome data, it does not simulate behaviour. It is a research tool aimed at neurobiologists. last_commit tracks the VFB3-MCP repository only, not the whole VFB code base. The MCP server was not tested by us." }, { "id": "dotfly", "name": "dotFly", "type": "model", "author_or_org": "kkokosa", "summary": "A C#/.NET engine that runs FlyWire v630/v783 and MaleCNS v1.0 as Shiu-style LIF networks, validated against Brian2. It also has Godot demos: readouts of real neurons feed a hand-written decoder that flies a 3D fly around a room.", "claim": { "text": "A native .NET inference engine for fly-connectome spiking models.", "url": "https://github.com/kkokosa/dotFly" }, "dataset": "several", "release": "MaleCNS v1.0 (demos); FlyWire v630/v783 (Shiu reproduction)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0, 166,700 neurons (demo, with 4,225 Kenyon cells and lLN neurons silenced); FlyWire v630 for validation", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none in the demo; an optional trained readout sample (logistic regression / ML.NET / ONNX) in samples/DotFly.Sample.TrainReadout", "body": "custom", "scripted_parts": "A decoder state machine (Flying/Landing/Landed/Feeding/TakeOff) with scripted casting, tumbling, surge upwind, vertical search and a feeding animation; the neurons supply rates and thresholds" }, "trained_class": "none", "trained_class_note": "optional trained readout sample, not in the demo", "grade_basis": [ "samples/DotFly.Sample.Godot3D/Behaviour.cs:7-22 the decoder is a hand-written state machine; 'every rule here is the demo's choice' [direct]", "samples/DotFly.Sample.Godot3D/Behaviour.cs:124 escape yaw = gain x (DNp04 left - right) [direct]", "samples/DotFly.Sample.Godot3D/Behaviour.cs:141-273 casting, tumbling, surge, landing and feeding are decoder rules; feeding starts when MN9 is above a threshold (line 248) [direct]", "samples/DotFly.Sample.Godot3D/FlyBrain3D.cs:28,61-70,188 MaleCNS v1.0 checkpoint; Kenyon cells and lLN* are silenced for stability [direct]", "tests/DotFly.Tests.Unit/Reference/Brian2GoldenTests.cs:10-18,88-97 the reference backend must match Brian2 spikes and v/g traces to 1e-11 mV on small fixtures [direct]", "samples/DotFly.Sample.SugarExperiment/Program.cs:1-6,116-128 the Shiu sugar experiment is compared with the published sugarR.parquet (Jaccard, rate correlation) [direct]", "PLAN.md:183 the degree-preserving shuffle control is only planned; there is no shuffle code in src/DotFly.Data or the CLI [direct]" ], "grade_date": "2026-09-28", "measured_result": "Engine validation only: the Brian2 golden fixtures are reproduced spike for spike (Brian2GoldenTests.cs:18,96-97). Against the published Shiu sugar rates: MN9 83 Hz, rate correlation 0.995, Jaccard 0.95 (docs/findings.md:8-11). Speed: 19.6x real time on FlyWire v630, 6.2-14.5x on MaleCNS (README.md:219-223). No shuffled or no-graph control of the demo behaviour.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/kkokosa/dotFly", "code_licence": "GPL-3.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 8411c87cb3; GitHub API spdx_id=GPL-3.0", "data_licence": "several (see dataset entries)", "platform": [ "Windows", "Linux" ], "gpu": "none", "download_size": "~300 MB FlyWire v630 files, or ~3 GB MaleCNS v1.0 plus checkpoints", "last_commit": { "date": "2026-09-22T10:23:46+02:00", "hash": "8411c87cb3cf9bb2f5b8c3f8bf35b11de5ec45dd", "branch": "main" }, "pushed_at": "2026-09-22T08:23:49Z", "last_release": "none", "created_at": "2026-09-21T17:22:07Z", "stars": 8, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/kkokosa/dotFly", "status": "ok", "http_code": 200, "final_url": "https://github.com/kkokosa/dotFly", "checked_at": "2026-10-06T09:40:05.969Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/kkokosa/dotFly [direct, local clone 8411c87]", "https://kkokosa.github.io/dotFly/ (docs site; sources read from docs/ in clone) [direct]" ], "notes_limitations": "The engine is a careful re-implementation of Shiu et al. and is checked against Brian2 and the published rates. That is engineering validation. Most of the room demo's behaviour comes from a hand-written decoder state machine (casting, tumbling, landing, feeding animation), and the author says so openly (README.md:33-35, docs/findings.md). The demo also silences 4,225 neurons to stop runaway activity, and findings.md:19-23 notes that the steering circuit stays silent. An 'R' key cuts recurrent transmission live, but no measured control is reported. Shuffled controls are described in PLAN.md but not implemented. Platform: CI runs on ubuntu and windows; the README uses PowerShell; the demos need .NET 11 and the Godot 4.7 .NET edition. Licence GPL-3.0." }, { "id": "drosophila-brain-mlx", "name": "drosophila-brain-mlx", "type": "model", "author_or_org": "Kiyan Doguc (Kisame76)", "summary": "A port of the Shiu et al. LIF whole-brain model to Apple Silicon (MLX and Metal) for FlyWire v630 and MaleCNS v1.0. It is checked against Brian2 and tested with a degree-preserving shuffled-wiring control.", "claim": { "text": "The published Shiu et al. leaky integrate-and-fire model ... runs at 0.29 seconds per biological second on an M4 Pro", "url": "https://github.com/Kisame76/drosophila-brain-mlx" }, "dataset": "several", "release": "FlyWire v630 (Brian2 parity and shuffle control); MaleCNS v1.0 (activity film)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire v630 whole brain (127,400 neurons, 14,687,178 edges); also MaleCNS v1.0 (166,700 neurons)", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none", "body": "none", "scripted_parts": "none found" }, "trained_class": "none", "grade_basis": [ "src/lif/compile_pack.py:27,430 builds the pack from Shiu's 'Excitatory x Connectivity' column, dataset flywire-v630 [direct]", "src/lif/shuffle_pack.py:1-16,56-75 degree-preserving shuffle: each neuron keeps its in/out degree and signed output, and targets are permuted [direct]", "src/lif/control_demo.py:1-12,28,145-163 runs the same sugar input on the real pack and the shuffled pack and reads out MN9 [direct]", "docs/figures/control-demo.svg (text labels): real wiring MN9 67.3 Hz, 408 neurons active; shuffled seed 0 MN9 0.0 Hz, 96 neurons; sugar GRNs 99.4 Hz in both [direct]", "README.md:47-58 the same control; seeds 1-4 also leave MN9 silent (89-99 neurons active) [direct]", "README.md:515-540 and src/lif/validate_brian2.py: the float64 oracle matches Brian2 spike for spike on an 800-neuron subnetwork [direct]" ], "grade_date": "2026-09-28", "measured_result": "Shuffle control (README.md:50-57; docs/figures/control-demo.svg), 21 right sugar GRNs at 100 Hz, 30 x 1 s trials: real v630 wiring gives MN9 67.30 Hz with 408 active neurons; the degree-preserving shuffle gives MN9 0 Hz with 96 active (seeds 1-4: MN9 silent, 89-99 active); input GRNs fire at 99.40 vs 99.38 Hz. Brian2 parity (README.md:522-527): ref64 is identical to Brian2 in all 4 configs (38 to 24,700 spikes); MLX float32 counts match in 3 of 4 and are 2,974 vs 2,973 in one. Against the upstream Brian2 notebook files (README.md:558-566): MN9 67.03 Hz (Brian2) vs 67.30/67.67 Hz; spikes per trial within |z| 1.90. Speed: 0.29 s vs Brian2 2.07 s per biological second on M4 Pro (README.md:8-9,159).", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Kisame76/drosophila-brain-mlx", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit e417b33616; GitHub API spdx_id=MIT", "data_licence": "several (see dataset entries)", "platform": [ "macOS" ], "gpu": "required", "download_size": "~90 MB upstream FlyWire v630 files + ~114 MB compiled pack; ~1.1 GB for MaleCNS", "last_commit": { "date": "2026-09-26T18:34:12-04:00", "hash": "e417b33616513ef350b1b1c3cdf2b5b7a1799c8e", "branch": "main" }, "pushed_at": "2026-09-26T22:34:16Z", "last_release": "none", "created_at": "2026-09-14T01:58:50Z", "stars": 2, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Kisame76/drosophila-brain-mlx", "status": "ok", "http_code": 200, "final_url": "https://github.com/Kisame76/drosophila-brain-mlx", "checked_at": "2026-10-06T09:40:06.474Z" }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/Kisame76/drosophila-brain-mlx [direct, local clone e417b33]" ], "notes_limitations": "Not a new model: the equations, constants and v630 wiring are Shiu et al.'s, and the MN9 result itself is Shiu's. The new parts are the engine and a shuffled-wiring control, which shows that the sugar-to-MN9 signal needs the real wiring (this mirrors the shuffle control in the Shiu paper). The control uses one readout neuron, one stimulus and 5 shuffle seeds. The control numbers come from the README and the figure; we did not re-run them. There is no comparison with fly data beyond Shiu's. Apple Silicon only (Metal GPU). No body. The README reports that the upstream sugarR reference file was run at 200 Hz, while upstream default_params now say 150 Hz (README.md:569-573). Licence MIT." }, { "id": "eon-fly-brain", "name": "fly-brain", "type": "model", "author_or_org": "eonsystemspbc (Eon Systems PBC)", "summary": "A benchmark suite. It runs the Shiu et al. whole-brain LIF model on FlyWire v783 in six simulators (Brian2 CPU, Brian2CUDA, PyTorch, NEST GPU, GeNN, Brian2GeNN) and compares their speed and spike output with Brian2 CPU as the reference.", "claim": { "text": "Emulation of the Drosophila Fly brain: Brian2, Brian2CUDA, PyTorch, NEST GPU, and neuromorphic chips", "url": "https://github.com/eonsystemspbc/fly-brain" }, "dataset": "FlyWire FAFB", "release": "v783 (benchmarks); v630 kept in data/archive for the original Shiu notebooks", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "FlyWire v783, whole brain (~138k neurons)", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "none found" }, "trained_class": "not-assessed", "grade_basis": [ "code/benchmark.py:46-47 loads data/2025_Completeness_783.csv and 2025_Connectivity_783.parquet [direct]", "code/benchmark.py:58-60,86 default experiment is sugar GRNs driven at 200 Hz (the Shiu stimulus) [direct]", "code/run_brian2_cuda.py:42-56,91 Shiu LIF equations and w_syn=0.275 mV x signed synapse count [direct]", "code/compare_ground_truth.py:1-17,86-95 engine-vs-Brian2 metrics (Jaccard of active neurons, rate Pearson r, spike-count ratio); its output JSON is not committed [direct]", "data/benchmark-results.csv:3,23,43,63,83,103 spike totals and active neurons per engine for the same 1 s sugar run [direct]", "scripts/setup_WSL_CUDA.sh:43-51 the only body reference: it installs an external 'fly-body' (flybody/MuJoCo) package from a folder that is not in the repo; no code imports it [direct]" ], "grade_date": "2026-09-28", "measured_result": "Engine parity only (sugar GRNs 200 Hz, 1 s, 1 trial, round 1, data/benchmark-results.csv:3-103): Brian2 CPU 16,567 spikes / 380 active neurons; Brian2CUDA 16,418/405; PyTorch 17,429/415; NEST GPU 17,233/412; GeNN 16,978/386; Brian2GeNN 16,738/383. Speed: GeNN 1.94x faster than real time, Brian2 CPU 0.38x (same rows). No Jaccard or correlation numbers are committed. No comparison with fly data and no shuffled control.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/eonsystemspbc/fly-brain", "code_licence": "GPL-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit a3db62f943; GitHub API spdx_id=GPL-2.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux" ], "gpu": "required", "download_size": "~100 MB connectome data in repo; ~580 MB weight caches generated; spike bundle (600 parquet files) on Google Drive", "last_commit": { "date": "2026-08-29T13:48:23+01:00", "hash": "a3db62f9436074e485c0278290c2164ed6150808", "branch": "main" }, "pushed_at": "2026-08-29T12:48:26Z", "last_release": "none", "created_at": "2026-03-05T13:24:16Z", "stars": 921, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/eonsystemspbc/fly-brain", "status": "ok", "http_code": 200, "final_url": "https://github.com/eonsystemspbc/fly-brain", "checked_at": "2026-10-06T09:40:06.106Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/eonsystemspbc/fly-brain [direct, local clone a3db62f]" ], "notes_limitations": "This project re-runs the Shiu et al. model and checks the other engines against Brian2 CPU. That is engineering validation, not a behaviour or biology test. The README's '91% accuracy' is Shiu's result and is not measured again here. Body code: NO. The 41-file tree has no FlyGym, NeuroMechFly, MuJoCo or flybody code. The only mention is a setup script that installs an external flybody env (scripts/setup_WSL_CUDA.sh:43-51). Neuromorphic chips: the repo description names them, but there is no code for them. The code has GeNN and Brian2GeNN, which the description does not list. The README project tree names code/paper-brian2/, but the folder is code/paper-phil-drosophila/. data/ground-truth-comparison.json is listed but not committed. The results folder says it supports a 'Nature paper' but gives no link. Licence GPL-2.0-or-later (Shiu code MIT). Brian2 CPU path may run without GPU, but the benchmark suite targets GPUs." }, { "id": "fasterfly", "name": "fasterfly", "type": "model", "author_or_org": "franciscocarloserra", "summary": "Event-driven Triton GPU kernels that step a LIF network built from the MaleCNS v1.0 connectome. It can batch many independent 'flies' in one launch. It is a speed tool, not a biological model.", "claim": { "text": "Event-driven Triton LIF kernels for the full MaleCNS fly connectome: 4.9x faster than torch.sparse, 72x batched", "url": "https://github.com/franciscocarloserra/fasterfly" }, "dataset": "MaleCNS", "release": "v1.0", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0, traced-only bodies (~165k neurons, 24.5M synapses)", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "none found" }, "trained_class": "not-assessed", "grade_basis": [ "params.json:2-7 MaleCNS v1.0 feather tables (traced-only weights) [direct]", "params.json:16-24 hand-set transmitter signs (dopamine, octopamine, serotonin = 0; histamine = -1) [direct]", "params.json:29-38 dimensionless LIF constants, dt 1 ms, gain 0.003 (not Shiu parameters) [direct]", "event_lif.py:19-46 Triton scatter kernel (atomic_add from fired neurons) and LIF update kernel [direct]", "runs/event_kernel/summary.md:3-23 speed per variant and scatter allclose checks; raster diff 0 events vs CSR [direct]", "runs/batched/summary.md:6-16 batch isolation checks; the seed 13 raster differs from CSR after a 1-ulp threshold tie [direct]" ], "grade_date": "2026-09-28", "measured_result": "Speed (RTX 3090, README.md:17-20,37-44,57-62): torch.sparse COO 708 steps/s; event_fused 3,455 steps/s (4.9x); 64 flies batched 51,159 fly-steps/s (72x). Correctness is checked against CSR only: scatter allclose (max |diff| up to 1.8e-3); the 500-step raster has 0 differing events at 100 and 1000 Hz (runs/event_kernel/summary.md:12-23); batched flies are bit-identical to CSR in 1/1, 4/4, 15/16 and 63/64 (runs/batched/summary.md:8-11). No behaviour or biology comparison.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/franciscocarloserra/fasterfly", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit e97e05bdec); GitHub API spdx_id=None", "data_licence": "CC-BY-4.0", "platform": [ "Linux" ], "gpu": "required", "download_size": "not stated (needs MaleCNS v1.0 feather tables, traced-only weights)", "last_commit": { "date": "2026-09-13T14:47:21-03:00", "hash": "e97e05bdeca387b56132811ab6d196d29f4c814d", "branch": "main" }, "pushed_at": "2026-09-13T17:47:22Z", "last_release": "none", "created_at": "2026-09-13T17:40:07Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/franciscocarloserra/fasterfly", "status": "ok", "http_code": 200, "final_url": "https://github.com/franciscocarloserra/fasterfly", "checked_at": "2026-10-06T09:40:06.525Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/franciscocarloserra/fasterfly [direct, local clone e97e05b]" ], "notes_limitations": "Pure speed and engineering work. Correctness is checked only against torch.sparse/CSR on the same network. That is fine for a kernel, but it says nothing about fly behaviour. The LIF constants are dimensionless and are not Shiu et al.'s (README.md:116). runs/batched/summary.md:20-32 reports a 'server backend (Brain.advance)' and a todelete/summary.md; neither is in the repo. The README gives both 3,455 and 3,485 steps/s for one fly (different variants). No LICENSE file in the tree (confirmed: 19 files, none is a licence), so reuse rights are unclear. All commits are from one day; 0 stars." }, { "id": "flyvis", "name": "flyvis", "type": "model", "author_or_org": "Turaga Lab (HHMI Janelia) / Lappalainen et al.", "summary": "PyTorch model of the fly visual system (optic lobe motion pathways). Wiring and synapse signs come from an EM column connectome; unknown neuron and synapse parameters are trained on an optic-flow task, and the model responses are then compared with published recordings.", "claim": { "text": "A connectome-constrained deep mechanistic network (DMN) model of the fruit fly visual system in PyTorch.", "url": "https://github.com/TuragaLab/flyvis" }, "dataset": "other (FlyEM FIB-25 + FIB-19 optic-lobe column reconstruction)", "release": "fib25-fib19_v2.2.json (FlyEM FIB-25 + FIB-19 optic lobe reconstructions, compiled per column; not FlyWire)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyEM FIB-25 + FIB-19 medulla/optic lobe column connectome (file fib25-fib19_v2.2.json), 65 node types tiled on a hexagonal lattice of radius 15 columns (721 columns, ~45,669 neurons per paper)", "neuron_model": "other: graded/rate passive point neurons with instantaneous graded-release synapses and ReLU output (PPNeuronIGRSynapses)", "input_mapping": "hand-made", "output_mapping": "learned", "trained_parts": "per-cell-type time constants and resting potentials (bias), per type-pair synapse strength scale, plus a convolutional optic-flow decoder; synapse counts and signs are fixed from the connectome", "body": "none", "scripted_parts": "none found" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "per-cell-type and per type-pair parameters trained; counts and signs fixed", "grade_basis": [ "flyvis/config/network/connectome/connectome.yaml:1-4 - loads ConnectomeFromAvgFilters from fib25-fib19_v2.2.json, extent 15 columns [direct]", "flyvis/connectome/connectome.py:109-160 - builds the network graph (cells + synapse sets with column offsets and synapse counts) from the JSON connectome [direct]", "flyvis/network/dynamics.py:152-218 - PPNeuronIGRSynapses: weight = sign * syn_count * syn_strength; dV/dt = (-V + bias + sum(w*ReLU(V_pre)) + input)/tau [direct]", "flyvis/config/network/edge_config/sign/sign.yaml:3 - synapse signs fixed (requires_grad: false); syn_strength.yaml:3 and node_config/time_const/time_const.yaml:6 - strength scale and time constants trained [direct]", "flyvis/config/task/task.yaml:1-4,30-39 - training task is optic flow on Sintel movies with a learned DecoderGAVP readout [direct]", "flyvis/network/stimulus.py:128-131 - visual input is written only to the input cell types (photoreceptors R1-R8) on the hex lattice [direct]", "flyvis/utils/groundtruth_utils.py:13-60,484-530 - literature ground truth: ON/OFF polarity per cell type, known direction-selective types, T4/T5 tuning curves from Maisak et al. 2013 [direct]", "flyvis/analysis/flash_responses.py:101-105 and flyvis/analysis/moving_bar_responses.py:442-507 - model flash/moving-edge responses scored against that ground truth (polarity, DSI, tuning-curve correlation) [direct]", "Europe PMC full-text XML PMC11525180 read 2026-10-03: 64 cell types, 45,669 neurons, 1,513,231 connections, 721 columns; 734 free parameters; ensemble median FRI right for all 32 characterised cell types, best model 30 of 32; r = 0.60, P = 2.6e-6 (task performance vs DSI prediction); 26 studies; random-parameter and ablation statements (quotes on https://shaduf.ai/p/digital-fly-catalog/projects/flyvis/) [direct]", "results_pretrained_models.zip (sha256 71c78d40...281db1, equal to upstream's checksum) read 2026-10-03: 50 models, validation loss median 5.300 (IQR 5.279-5.333), from validation_loss.h5 [direct]", "Pretrained-model licence: license.txt (MIT) in the authors' public Drive folder, created 2026-09-29 [direct]" ], "grade_date": "2026-09-28", "measured_result": "Paper (PMC11525180, quotes checked 2026-10-03): the ensemble median predicts the preferred contrast (ON/OFF) of all 32 cell types with known polarity, the best single model 30 of 32; models with better task performance predict T4/T5 direction selectivity better (r = 0.60, P = 2.6e-6); predictions agree with measurements from 26 studies. Controls: random parameters on the full connectome still get contrast right but direction selectivity and preferred direction poorly; with only cell-type connectivity, trained models predict activity poorly; contrast needs only synapse signs, and direction selectivity (not preferred direction) is possible without synapse counts. No shuffled-wiring network was tested. The authors' 50 pretrained models have validation losses from 5.137 to 5.678 (median 5.300), read by us from their files.", "try_url": "https://colab.research.google.com/github/TuragaLab/flyvis/blob/main/examples/01_flyvision_connectome.ipynb", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/TuragaLab/flyvis", "code_licence": "MIT", "code_licence_source": "LICENSE file 'license' read at commit 92b3845cc4; GitHub API spdx_id=MIT", "data_licence": "FlyEM FIB-25/FIB-19 reconstructions (Janelia), compiled per column by the authors; no separate data licence stated. Pretrained models: MIT (license.txt in the authors' Drive folder, 2026-09-29)", "platform": [ "Linux", "Colab" ], "gpu": "optional", "download_size": "pretrained models 3.4 MB (results_pretrained_models.zip, 50 models); the Python install is not small: flyvis's direct dependencies are about 654 MB of wheels with PyPI's default torch, or about 290 MB with the 196 MB CPU-only torch (PyPI and PyTorch index, 2026-10-03; not installed by us)", "last_commit": { "date": "2026-08-06T00:09:38-04:00", "hash": "92b3845cc426dd309a1a0e1b3890156c42e14021", "branch": "main" }, "pushed_at": "2026-08-18T21:25:52Z", "last_release": { "tag": "v1.2.0", "date": "2026-08-06T04:17:35Z", "url": "https://github.com/TuragaLab/flyvis/releases/tag/v1.2.0" }, "created_at": "2023-03-12T00:52:30Z", "stars": 190, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41586-024-07939-3", "peer_review": "peer-reviewed", "link_status": { "url": "https://colab.research.google.com/github/TuragaLab/flyvis/blob/main/examples/01_flyvision_connectome.ipynb", "status": "ok", "http_code": 200, "final_url": "https://colab.research.google.com/github/TuragaLab/flyvis/blob/main/examples/01_flyvision_connectome.ipynb", "checked_at": "2026-10-06T09:40:05.992Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-024-07939-3", "status": "blocked", "http_code": 200 }, { "url": "https://github.com/TuragaLab/flyvis", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "project_page": "/p/digital-fly-catalog/projects/flyvis/", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/TuragaLab/flyvis [direct]", "https://turagalab.github.io/flyvis/ [direct]", "https://pmc.ncbi.nlm.nih.gov/articles/PMC11525180/ [fetch-summary]" ], "notes_limitations": "Connectome is a small, column-averaged optic lobe model from FlyEM FIB-25/FIB-19 data (paper credits Shinomiya/Takemura et al.), not FlyWire or a whole-brain release; the file name 'v2.2' is the authors' compilation version, not a public dataset release. The dataset enum has no FIB-25/FIB-19 value, so 'several' is used. Comparison is with published literature values (polarity, DSI, tuning curves), not new recordings. Neuron model is a passive graded rate model; many parameters are trained (flagged). Docs say tested on Linux, Python 3.9-3.12 (pyproject.toml:67 requires >=3.9,<3.13). GPU helps for training; inference with pretrained models can run on CPU (not verified here). Could not open nature.com directly (login redirect); paper numbers come from the PMC copy." }, { "id": "shiu-drosophila-brain-model", "name": "Drosophila_brain_model (Shiu et al. 2024)", "type": "model", "author_or_org": "Philip K. Shiu et al. (Scott lab, UC Berkeley, with FlyWire)", "summary": "A leaky integrate-and-fire model of all proofread neurons in the FlyWire adult brain, run in Brian2. You activate or silence chosen neurons by FlyWire ID and read out spike rates of every other neuron; the paper tests these predictions against fly experiments.", "claim": { "text": "Across 164 predictions we were able to test empirically, 91% were consistent with our empirical results (Supplementary Table 9).", "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC11446845/" }, "dataset": "FlyWire FAFB", "release": "v630 (default, used in the paper); v783 files also in repo", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire FAFB materialization v630 (127,400 proofread neurons per paper Methods); v783 files provided as an option", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none; one free parameter w_syn = 0.275 mV per synapse, tuned by hand", "body": "none", "scripted_parts": "none found" }, "trained_class": "none", "trained_class_note": "one hand-set parameter w_syn", "grade_basis": [ "model.py:15-53 default_params: v_0 = v_rst = -52 mV, v_th = -45 mV, t_mbr = 20 ms, tau = 5 ms (synaptic), t_rfc = 2.2 ms, t_dly = 1.8 ms, w_syn = 0.275 mV (free parameter), r_poi = 150 Hz, f_poi = 250 (Poisson input weight = 250 x 0.275 mV) [direct]", "model.py:44-52 equations: dv/dt = (v_0 - v + g)/t_mbr, dg/dt = -g/tau, spike when v > v_th, reset v and g [direct]", "model.py:156-183 loads completeness CSV + connectivity parquet; synapse weight = 'Excitatory x Connectivity' (signed synapse count) x w_syn; on_pre g += w [direct]", "figures.ipynb cell 0 config uses 2023_03_23_completeness_630_final.csv and 2023_03_23_connectivity_630_final.parquet; cells run the sugar/water/bitter/Ir94e GRN and MN9 experiments of the paper figures [direct]", "Readme.md 'Version 783': code is set up for FlyWire v630 as in the paper; for v783 change config path_comp='./Completeness_783.csv' and path_con='./Connectivity_783.parquet' [direct]", "PMC11446845 Discussion: 91% of 164 empirically testable predictions consistent with experiments (84% when excluding the split-GAL4 optogenetic screen) [direct]", "PMC11446845 Results: with shuffled connectivity weights only 1 of 100 simulations activated MN9 vs 100% with the real connectome at 100 Hz sugar GRN input (Supplementary Table 1d) [direct]" ], "grade_date": "2026-09-28", "measured_result": "Paper (PMC11446845): 91% of 164 testable predictions matched experiments (84% excluding the split-GAL4 optogenetic activation screen); split-GAL4 screen >90% accuracy; shuffled-weight control activated MN9 in 1/100 runs vs 100% for the real connectome. Robustness: +/-30% w_syn and +/-50% inhibition changes keep 88-96% agreement with the default model ('95%' appears here, not as an accuracy figure).", "try_url": "https://colab.research.google.com/github/philshiu/Drosophila_brain_model/blob/main/example.ipynb", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/philshiu/Drosophila_brain_model", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 91bdd1e7dc; GitHub API spdx_id=MIT", "data_licence": "CC-BY-NC-4.0", "platform": [ "macOS", "Windows", "Linux", "Colab" ], "gpu": "none", "download_size": "~90 MB (v630 parquet 87 MB + CSV 3 MB); ~104 MB for v783 files; whole repo ~200 MB. Raw paper outputs (several GB) are in a separate archive (doi 10.17617/3.CZODIW).", "last_commit": { "date": "2024-09-14T09:47:44-04:00", "hash": "91bdd1e7dcf193f3e7ca5a8933497fcef63b7960", "branch": "main" }, "pushed_at": "2024-09-14T13:47:44Z", "last_release": "none", "created_at": "2023-02-25T21:59:35Z", "stars": 349, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41586-024-07763-9", "peer_review": "peer-reviewed", "link_status": { "url": "https://colab.research.google.com/github/philshiu/Drosophila_brain_model/blob/main/example.ipynb", "status": "ok", "http_code": 200, "final_url": "https://colab.research.google.com/github/philshiu/Drosophila_brain_model/blob/main/example.ipynb", "checked_at": "2026-10-06T09:40:06.218Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-024-07763-9", "status": "blocked", "http_code": 200 }, { "url": "https://github.com/philshiu/Drosophila_brain_model", "status": "ok", "http_code": 200 }, { "url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC11446845/", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "project_page": "/p/digital-fly-catalog/projects/shiu-drosophila-brain-model/", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/philshiu/Drosophila_brain_model [direct]", "https://pmc.ncbi.nlm.nih.gov/articles/PMC11446845/ [direct]", "https://doi.org/10.1038/s41586-024-07763-9", "https://www.biorxiv.org/content/10.1101/2023.05.02.539144v1" ], "notes_limitations": "The comparison with fly experiments lives in the paper (supplementary tables), not in repo code: the repo produces the model predictions only. The '91%' is the fraction of 164 predictions consistent with experiments; the '95%' figures in the paper are agreement between perturbed and default model (robustness), not accuracy against flies. The paper states it does not trust absolute firing rates; no gap junctions, neuromodulation, or baseline activity. Environment pins Python 3.10 and Brian2 2.5.1; runs on CPU (paper: about 5 min per 1 s trial per CPU thread for sugar activation). Readme links the bioRxiv preprint; the Nature 2024 version is the final one. Last commit 2024-09-14." }, { "id": "flybody", "name": "flybody", "type": "body", "author_or_org": "Turaga Lab (HHMI Janelia) and Google DeepMind", "summary": "Detailed MuJoCo body model of the fruit fly with reinforcement-learning tasks for walking, flight and vision-guided flight. Trained policies are standard neural networks, not connectome-based.", "claim": { "text": "an anatomically-detailed body model of the fruit fly Drosophila melanogaster for MuJoCo physics simulator and reinforcement learning applications.", "url": "https://github.com/TuragaLab/flybody" }, "dataset": "none", "release": "n/a", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "RL policies (DMPO via Acme/TensorFlow) in optional extras; not connectome-constrained", "body": "flybody", "scripted_parts": "n/a (body model and tasks)" }, "trained_class": "not-assessed", "grade_basis": [ "pyproject.toml:8-15 - core deps numpy==1.26.4, dm_control (MuJoCo), h5py, mediapy; requires-python >=3.10 [direct]", "pyproject.toml:35-46 - optional 'tf' extra: dm-acme[tf,envs,jax], tensorflow==2.8.0, tensorflow-probability 0.16.0, dm-reverb 0.7.0, nvidia-cudnn-cu11; 'ray' extra for distributed training [direct]", ".github/workflows/pyversions.yml:18 - CI tests Python 3.10, 3.11, 3.12 [direct]", "README.md:76,102 - recommended conda env uses python=3.10 and cudatoolkit=11.8.0 [direct]", "flybody/agents/agent_dmpo.py, flybody/train_dmpo_ray.py - RL training code (DMPO) [direct]", "flybody/download_data.py:2 - trained policies/data from Janelia figshare doi 10.25378/janelia.25309105 [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/TuragaLab/flybody", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit d015e9bfe4; GitHub API spdx_id=Apache-2.0", "data_licence": "n/a", "platform": [ "Linux" ], "gpu": "optional", "download_size": "not stated (optional figshare data)", "last_commit": { "date": "2025-07-30T18:49:01-04:00", "hash": "d015e9bfe441bd90ae431bac24c55cb74bdbce26", "branch": "main" }, "pushed_at": "2026-02-07T03:55:45Z", "last_release": { "tag": "v0.1.0", "date": "2024-05-16T19:18:50Z", "url": "https://github.com/TuragaLab/flybody/releases/tag/v0.1.0" }, "created_at": "2024-01-30T18:23:14Z", "stars": 946, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41586-025-09029-4", "peer_review": "peer-reviewed", "link_status": { "url": "https://github.com/TuragaLab/flybody", "status": "ok", "http_code": 200, "final_url": "https://github.com/TuragaLab/flybody", "checked_at": "2026-10-06T09:40:06.940Z", "other_links": [ { "url": "https://doi.org/10.1038/s41586-025-09029-4", "status": "blocked", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/TuragaLab/flybody [direct]", "https://doi.org/10.1038/s41586-025-09029-4 [direct: title checked via Europe PMC]" ], "notes_limitations": "Infrastructure, not graded. Core install (MuJoCo via dm_control) runs on CPU; GPU is useful only for RL training with the TF extra. The TF extra pins tensorflow==2.8.0 with CUDA 11 cuDNN, which in practice limits ML use to Python 3.10 even though core allows >=3.10 and CI tests 3.10-3.12. JAX appears only through the dm-acme[jax] extra; no MJX use found. Paper: Vaxenburg et al., 'Whole-body physics simulation of fruit fly locomotion', Nature 2025. Last commit in clone 2025-07-30 (API pushed_at 2026-02-07). Other OS support not stated." }, { "id": "flygym", "name": "FlyGym (NeuroMechFly v2)", "type": "body", "author_or_org": "Neuroengineering Laboratory (Ramdya lab), EPFL", "summary": "Python library for NeuroMechFly, a MuJoCo body model of the adult fly with vision, smell, leg adhesion and joint sensing. Users compose flies and scenes and plug in their own controllers; version 2 also offers GPU batch simulation through MuJoCo Warp.", "claim": { "text": "Simulating embodied sensorimotor control with NeuroMechFly v2", "url": "https://github.com/NeLy-EPFL/flygym" }, "dataset": "none", "release": "n/a", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "none", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none in the core library (optional RL extra with Stable-Baselines3)", "body": "FlyGym", "scripted_parts": "n/a (simulator; controllers are user code)" }, "trained_class": "not-assessed", "grade_basis": [ "pyproject.toml:3-5 - flygym 2.1.0, requires-python >=3.12,<3.15 [direct]", "pyproject.toml:23-24,39-43 - core dependency mujoco 3.9.x; optional 'warp' extra (warp-lang, mujoco_warp) for GPU [direct]", "README.md:1-5 - FlyGym 2.x API introduced March 2026, full rewrite, not backward compatible; ~10x CPU and ~300x GPU (Warp/MJWarp) speed-up claimed [direct]", "docs/migration.md:5,19 - 1.x Gymnasium API moved to separate package flygym-gymnasium [direct]", "docs/changelog.md:7,41,48 - 2.1.0 drops dm-control PyMJCF for MuJoCo MjSpec; adds FlyBody model; supports Python 3.12-3.14 [direct]", "docs/installation.md:53 - Warp extra needs an NVIDIA GPU; skip it on Mac [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://neuromechfly.org/wasm/viewer/viewer.html", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/NeLy-EPFL/flygym", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 38c8ec6103; GitHub API spdx_id=Apache-2.0", "data_licence": "n/a", "platform": [ "browser", "Linux", "macOS", "Colab" ], "gpu": "optional", "download_size": "not stated (large meshes fetched lazily from S3)", "last_commit": { "date": "2026-06-29T01:43:47+02:00", "hash": "38c8ec61034cd59bc5ba0de20688d4a3c0000d60", "branch": "main" }, "pushed_at": "2026-08-24T11:57:38Z", "last_release": { "tag": "Version 2.1.0", "date": "2026-06-24T15:26:40Z", "url": "https://github.com/NeLy-EPFL/flygym/releases/tag/v2.1.0" }, "created_at": "2023-03-16T10:03:01Z", "stars": 386, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.1038/s41592-024-02497-y", "peer_review": "peer-reviewed", "link_status": { "url": "https://neuromechfly.org/wasm/viewer/viewer.html", "status": "ok", "http_code": 200, "final_url": "https://neuromechfly.org/wasm/viewer/viewer.html", "checked_at": "2026-10-06T09:40:06.750Z", "other_links": [ { "url": "https://doi.org/10.1038/s41592-024-02497-y", "status": "blocked", "http_code": 200 }, { "url": "https://github.com/NeLy-EPFL/flygym", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/NeLy-EPFL/flygym [direct]", "https://neuromechfly.org/wasm/viewer/viewer.html [direct]", "https://neuromechfly.org/wasm/game/game.html [direct]" ], "notes_limitations": "Infrastructure, not graded. v2.x (March/April 2026) is a breaking rewrite: no Gymnasium interface (old API lives in flygym-gymnasium), and 2.1.0 replaced dm-control PyMJCF with MuJoCo MjSpec (e.g. .full_identifier -> .name). Python 3.12-3.14 only. GPU path uses MuJoCo Warp (NVIDIA), not MJX; no MJX reference found in code or docs. Small date mismatch: README says 2.x came in March 2026, migration.md says Gymnasium API was followed until April 2026. Browser viewer and in-browser game pages load (HTTP 200) but were not tested in use. Windows support not stated." }, { "id": "bioreservoir", "name": "BioReservoir", "type": "research", "author_or_org": "IG Digital Lab (igdigitallab)", "summary": "Runs the MaleCNS v1.0 connectome as a fixed spiking reservoir (Brian2 LIF, Shiu et al. port). Question text is embedded, projected at random onto sensory neurons, and the left/right bias of descending neurons is read as a yes/no answer. Runs are pre-registered and compared with rewired, random-graph, no-brain and coin controls.", "claim": { "text": "Public connectomes as fixed reservoirs: load a real brain's wiring, simulate it as a spiking network, drive it, read it out — with controls and pre-registration.", "url": "https://github.com/igdigitallab/bioreservoir" }, "dataset": "several", "release": "MaleCNS v1.0 (used for all batch-001 runs); BANC materialization 626 (Dataverse doi:10.7910/DVN/8TFGGB), loaded but excluded from headline claims", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0, 165,122 neurons / 6,235,682 connections after min_syn=5 filter; BANC 626 (114,456 neurons) as second brain", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none in the brain; text encoder is a pretrained sentence model (all-MiniLM-L6-v2) with a fixed seeded random projection to sensory-neuron Poisson rates", "body": "none", "scripted_parts": "none found; answer = descending-neuron L/R spike bias minus a handedness offset from neutral sentences" }, "trained_class": "front-end", "trained_class_note": "pretrained sentence encoder in front of the wiring", "grade_basis": [ "src/bioreservoir/connectomes/malecns.py:31-33 loads the three MaleCNS v1.0 feather files [direct]", "src/bioreservoir/sim/lif.py:124 LIF equation dv/dt=(v_rest-v+g)/tau_m, Brian2 port of Shiu et al. [direct]", "src/bioreservoir/oracle/encode.py:60-92 text -> MiniLM embedding -> seeded Gaussian projection -> Poisson rates (hand-made input map) [direct]", "src/bioreservoir/oracle/controls.py:165,180,276,291 degree-preserving rewire, Erdos-Renyi, no-brain baseline and coin controls [direct]", "experiments/001-fly-oracle/RESULTS.md:16-20 429 runs: real 99, rewired 99, er 99, no_brain 99, coin 33 [direct]", "experiments/001-fly-oracle/RESULTS.md:35-41 real-brain p(yes) mean 0.5001, range 0.4930-0.5053 [direct]", "experiments/001-fly-oracle/RESULTS.md:49-52 no outcomes yet; no accuracy/Brier numbers exist [direct]", "docs/MODEL.md:74-90 calibration: sugar GRN -> MN9 128.5 Hz, bitter suppression 96.8% (reproduces Shiu direction) [direct]", "docs/MODEL.md:143-155 BANC fails the side-tracking check and is excluded from headline claims [direct]" ], "grade_date": "2026-09-28", "measured_result": "No outcome scoring yet (questions resolve by 2026-12-15, RESULTS.md:49-52). Per-run control values are in experiments/001-fly-oracle/results.csv; our own aggregate of that file: p(yes) real 0.4996 (0.4905-0.5053, sd 0.0028), no_brain 0.4999 (0.4923-0.5082), rewired 0.5009 (0.4801-0.5195), Erdos-Renyi 0.5027 (0.3939-0.6192), coin 0.576. So far the real wiring gives answers close to 0.5, like the no-brain baseline. Calibration: bitter input cuts MN9 firing by 96.8% (docs/MODEL.md:85-87).", "try_url": "https://fly.igdigi.com", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/igdigitallab/bioreservoir", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit ac0253e20b; GitHub API spdx_id=Apache-2.0", "data_licence": "several (see dataset entries)", "platform": [ "browser", "Linux" ], "gpu": "none", "download_size": "~1.1 GB MaleCNS v1.0 data (weights file 1.05 GB) plus BANC files (scripts/datasets.json)", "last_commit": { "date": "2026-09-24T04:39:08-07:00", "hash": "ac0253e20b41182b03a7e280ac06d1f747bf137f", "branch": "main" }, "pushed_at": "2026-09-24T11:39:14Z", "last_release": { "tag": "snapshot-e2d7e7e95b3e", "date": "2026-09-24T06:10:06Z", "url": "https://github.com/igdigitallab/bioreservoir/releases/tag/snapshot-e2d7e7e95b3e" }, "created_at": "2026-09-22T06:03:37Z", "stars": 3, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://fly.igdigi.com", "status": "ok", "http_code": 200, "final_url": "https://fly.igdigi.com", "checked_at": "2026-10-06T09:40:07.148Z", "other_links": [ { "url": "https://github.com/igdigitallab/bioreservoir", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "not-tested", "sources": [ "https://github.com/igdigitallab/bioreservoir [direct]", "https://fly.igdigi.com [fetch-summary]" ], "notes_limitations": "Controls are built and run, and all per-run values are public, but the project has not yet published a real-vs-control comparison or any outcome score; RESULTS.md only summarises the real condition. The aggregate numbers above are ours, computed from results.csv. Borderline B/A: could move to A once the pre-registered scoring is published. The task (forecasting questions from their text) has no biological ground truth. Input and output maps are chosen by the authors. The BANC run uses materialization 626, not 888; the authors say 888 exists and has not been tried yet (docs/MODEL.md:131-139). The live site runs fresh simulations (~66 s per question). The repo is published as squashed snapshots. Created 2026-09-22." }, { "id": "doom-fly-control", "name": "Is the fly brain actually playing DOOM? (control experiments)", "type": "research", "author_or_org": "gabrycina (GitHub)", "summary": "A third-party control study of two viral 'fly brain plays DOOM' projects. It re-runs the untrained 166,700-neuron DOOMFLY simulation with shuffled wiring, no connections, blind input and a brainless fixed autopilot, and re-trains the doomfly-rl network with real, rewired and absent wiring. It also re-runs a sugar/bitter taste test on the Shiu et al. whole-brain model with real versus shuffled wiring.", "claim": { "text": "The fly brain isn't what's playing DOOM: a constant turn-right, walk, shoot autopilot survives as long as the untrained spiking fly (52 s vs 53 s), and trained agents score the same with real, random or no wiring.", "url": "https://github.com/gabrycina/doom-fly-control" }, "dataset": "MaleCNS (DOOM experiments); FlyWire FAFB (taste experiment via the Shiu model)", "release": "MaleCNS v1.0 (SHA-256 matched, via nftechie/DOOMFLY); FlyWire v630 (Shiu model commit 91bdd1e)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "Upstream projects' graphs: MaleCNS v1.0 full graph (166,700 neurons, 25.6 M edges) for DOOMFLY; a 49,393-neuron MaleCNS subset inside doomfly-rl; FlyWire v630 for the Shiu model", "neuron_model": "LIF (DOOMFLY's own NativeBrain kernel and the Shiu Brian2 model, both unchanged); doomfly-rl's trained network as published", "input_mapping": "hand-made (DOOMFLY retina sampling, unchanged) / learned (doomfly-rl CNN encoder)", "output_mapping": "hand-made (DOOMFLY fixed 'bci' decoder, unchanged) / learned (doomfly-rl policy head)", "trained_parts": "none added by this project; doomfly-rl arms are re-trained end to end by the author with each wiring", "body": "none", "scripted_parts": "the 'spray' arms are a deliberately brainless fixed autopilot used as a control" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "uses doomfly-rl arms trained end to end", "grade_basis": [ "src/run_a2.py:1-11 arms for DOOMFLY: real graph.npz, shufK (postsynaptic targets permuted, presynaptic neuron and weight kept), blind (photoreceptors zeroed), noconn (all weights zero) [direct]", "src/run_a2.py:31-66 uses DOOMFLY's NativeBrain, retinal_samples and NeuralControls(mode='bci') unchanged; SPRAY arm = fixed turn/forward with trigger held, no brain [direct]", "results/a2/summary.json: real n=20 survival 53.3 s, 12.1 kills; shuffled (3 seeds pooled) n=12 6.7 s; noconn n=8 5.5 s; blind n=8 10.5 s; spray_matched n=20 51.9 s, 11.75 kills [direct]", "results/a1/summary.json: doomfly-rl, 100 seed-matched games per arm; e.g. basic real 80.57 vs shuf0 80.72 vs noconn 80.72 (Welch p 0.92); rewired after training -148.44 [direct]", "results/shuffle_check.json: shuffle verification for the 49,393-neuron graph (about 1.3% of shuffled edges coincide with real edges) [direct]", "partB/summary.json: Shiu model, sugar 100 Hz -> MN9_L 66.1 Hz (30 trials) with real wiring; bitter -> 0 Hz; shuffled arms in the same file [direct]" ], "grade_date": "2026-09-28", "measured_result": "Untrained DOOMFLY (combat_survival): real wiring 53.3 s survival and 12.1 kills (n=20); shuffled wiring 6.7 s (n=12, 3 seeds); no connections 5.5 s (n=8); blind 10.5 s (n=8); brainless autopilot with the fly's average turn/forward 51.9 s and 11.75 kills (n=20). doomfly-rl: real, shuffled and no-wiring arms score the same across 5 ViZDoom scenarios (100 games each); rewiring after training collapses performance. Shiu model: sugar drives MN9 at 66 Hz with real wiring, 0 Hz shuffled. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/gabrycina/doom-fly-control", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 6b2292286a; GitHub API spdx_id=NOASSERTION", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0; MaleCNS: CC-BY-4.0", "platform": [ "macOS", "Linux" ], "platform_note": "macOS (Apple M3 Max per README); Linux (untested)", "gpu": "optional", "gpu_note": "optional (author used Apple MPS; CPU and MPS results reported equal)", "download_size": "repo about 16 MB including a 12 MB video; reproducing needs the upstream projects, MaleCNS v1.0 tables and doomfly-rl checkpoints from Hugging Face", "last_commit": { "date": "2026-09-27T09:03:22+01:00", "hash": "6b2292286af9634bb967fcc672b47ff0ad89cbb6", "branch": "main" }, "pushed_at": "2026-09-27T08:03:36Z", "last_release": "none", "created_at": "2026-09-27T08:03:23Z", "stars": 1, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/gabrycina/doom-fly-control", "status": "ok", "http_code": 200, "final_url": "https://github.com/gabrycina/doom-fly-control", "checked_at": "2026-10-06T09:40:07.623Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/gabrycina/doom-fly-control [direct: clone HEAD 6b22922]", "https://github.com/nftechie/DOOMFLY (upstream under test; catalogued as doomfly)", "https://github.com/nonatofabio/doomfly-rl (upstream under test; not catalogued yet)" ], "notes_limitations": "A control study, not a demo: its result is negative for the claim that the fly wiring plays DOOM. The matched autopilot's turn and forward constants are taken from the real fly's average output, so it shows the behaviour is reproducible by a constant policy, not that no policy could exist. Shuffled-wiring arms are small (4 runs per seed) and shuffled brains are much quieter overall, so an activity-matched null is still missing (the author says so). Upstream code is not redistributed; reproduction needs three other repositories. Measured by the author, not reproduced by us." }, { "id": "doomfly-rl", "name": "doomfly-rl", "type": "research", "author_or_org": "Fabio Nonato (nonatofabio)", "summary": "A Doom-playing network in the style of chessfly: a learned conv encoder feeds visual sensory neurons of a real fly connectome (FlyWire v783 or a 49k-neuron MaleCNS subgraph), the graph is unrolled for 5 rate steps, and a learned decoder reads central and descending neurons to pick actions. The student is distilled from per-scenario PPO teachers and then fine-tuned with GRPO. The repo includes degree-preserving shuffle and no-connectome controls trained with the same recipe, and reports that the wiring gives no measurable advantage.", "claim": { "text": "A neural network wired like a fruit fly's brain, trained with gradient descent to play Doom; the connectome is treated as an inductive bias and the encoder, decoder and one gain per synapse are learned.", "url": "https://github.com/nonatofabio/doomfly-rl" }, "dataset": "FlyWire (FAFB) and MaleCNS", "release": "FlyWire v783 (Shiu et al. Connectivity_783.parquet, SHA-256 checked) and a 49,393-neuron MaleCNS subgraph from Hugging Face dataset fernandofernandes/fly-connectome-49k", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire v783, 138,639 neurons / 15.09 M edges, or MaleCNS 49k subgraph, 49,393 neurons / 9.05 M edges; wiring and synapse signs frozen, one learned log-gain per synapse", "neuron_model": "rate: leaky h <- (1-a)h + a*relu(scale*batchnorm(Wh+u)+shift), 5 steps, learned per-neuron per-step scale/shift", "input_mapping": "learned (conv stem + MLP from 4 stacked 72x96 grey frames to currents on 10,855 visual sensory neurons)", "output_mapping": "learned (MLP decoder from central + descending + motor neuron states to 512 units, then policy/value heads over 22-23 actions with per-scenario masks and scenario embedding)", "trained_parts": "encoder, decoder, heads, per-synapse gains, per-neuron scale/shift; distilled from PPO teachers (soft cross-entropy), then GRPO/RLOO fine-tuning", "body": "none (ViZDoom first-person game)", "scripted_parts": "none found in the policy; teachers are ordinary Stable-Baselines3 PPO CNNs" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "encoder, decoder, per-synapse gains trained (distillation + RL)", "grade_basis": [ "doomfly/connectome/build.py:1-54 downloads Shiu et al. Connectivity_783 and Schlegel annotations with SHA-256 checks, marks visual sensory inputs and central/descending/motor readout [direct]", "doomfly/model/flynet.py:135-168 5-step connectome dynamics between learned stem and learned decoder/heads [direct]", "doomfly/model/flynet.py:45-76,105-120 degree-preserving edge shuffle control and a no-connectome control (stem straight to decoder) [direct]", "doomfly/train.py:4,97 training loss is cross-entropy to the PPO teacher's policy (distillation) [direct]", "docs/controls.md:21-24 end-of-training table (10 eval episodes each): connectome, two shuffles and no-connectome score within one std on all five scenarios [direct]", "docs/results/freeplay_heldout_{conn,shuffled_s0,noconn}_{init,final}.json: 30 held-out Freedoom MAP01 episodes per backbone after GRPO; mean return conn 6.57, shuffled 6.71, no-connectome 7.39 (our recomputation from the episode lists) [direct]" ], "grade_date": "2026-09-28", "measured_result": "Negative result. Distillation controls, step 60,000, 10 eval episodes per scenario (docs/controls.md): MaleCNS-49k connectome basic 78.2+/-7.5, defend_the_center 18.6+/-1.8, health_gathering 1728+/-616, deadly_corridor 2280.5+/-2.4, defend_the_line 21.0+/-5.9; degree-preserving shuffle seed 0: 78.2, 19.0, 1602, 2280.6, 21.7; seed 1: 78.2, 18.4, 1921, 2281.0, 22.2; no connectome (27.7 M vs 48.2 M params, 22x faster training): 77.5, 18.3, 1373, 2279.9, 23.0. All gaps inside one std. Free play on MAP01 after 500 GRPO iterations, 30 held-out episodes: mean return connectome 6.57, shuffled 6.71, no-connectome 7.39 (deaths 13, 14, 10 of 30). Training logs themselves are in S3, not the repo; the tables and held-out JSONs are. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/nonatofabio/doomfly-rl", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 334d915074; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS" ], "gpu": "optional", "gpu_note": "optional for evaluation (CPU command given); training used one NVIDIA L40S per run", "download_size": "not checked (Hugging Face checkpoints plus a MaleCNS-49k connectome file; FlyWire build needs the Shiu et al. parquet)", "last_commit": { "date": "2026-09-27T18:46:46-07:00", "hash": "334d9150745d8a65bd03bfaef5bc7bc30cc27d28", "branch": "main" }, "pushed_at": "2026-09-28T01:46:48Z", "last_release": "none", "created_at": "2026-09-18T02:52:51Z", "stars": 0, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://nonatofabio.github.io/blog/posts/doomfly_autonomous.html", "peer_review": "none", "peer_review_note": "none (blog post)", "link_status": { "url": "https://github.com/nonatofabio/doomfly-rl", "status": "ok", "http_code": 200, "final_url": "https://github.com/nonatofabio/doomfly-rl", "checked_at": "2026-10-06T09:40:07.662Z", "other_links": [ { "url": "https://nonatofabio.github.io/blog/posts/doomfly_autonomous.html", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "wiring-null", "wiring_effect": "no-difference", "sources": [ "https://github.com/nonatofabio/doomfly-rl [direct: clone HEAD 334d915]", "https://huggingface.co/fabiononato/doomfly-rl [not checked]" ], "notes_limitations": "Well-controlled negative result: the fly wiring gives no measurable benefit over a degree-preserving shuffle or over removing the neuron layer; the learned conv encoder and MLP decoder (about two thirds of parameters) do the work. The controls were run on the MaleCNS-49k backbone only; the FlyWire v783 backbone has no shuffle control of its own. One training seed per condition (two shuffle seeds), 10-episode evals on saturated scenarios. The 'control' credited to this author by doom-fly-control corresponds to these 10-episode evaluations in docs/controls.md. Neurons are rate units, not spiking. The README says the project was built largely by an AI coding agent (Strands harness). Not the same project as nftechie/doomfly." }, { "id": "embodied-fly-brain-erojas", "name": "Emergent Individuality in Whole-Brain Connectome Simulations of Drosophila (fly-brain)", "type": "research", "author_or_org": "Enrique Manuel Rojas Aliaga (erojasoficial-byte)", "summary": "The full FlyWire v783 graph runs as a PyTorch LIF network (the Shiu et al. model) with Hebbian plasticity, coupled to a NeuroMechFly v2 (flygym/MuJoCo) body. About 18 named descending neurons are read out and mapped by hand onto left/right drive for flygym's CPG turning controller. Grooming, flight and feeding are scripted motor programs that start when readout rates cross thresholds, and several sensory paths skip the brain altogether.", "claim": { "text": "The fly sees, walks, grooms, and escapes, driven entirely by 138,639 spiking neurons from the FlyWire connectome; two flies with identical connectomes develop distinct behaviour through Hebbian plasticity.", "url": "https://github.com/erojasoficial-byte/fly-brain" }, "dataset": "FlyWire FAFB", "release": "v783 (2025_Connectivity_783.parquet, 138,639 neurons)", "evidence_grade": "C", "grade_note": "Borderline C: the neural part alone could argue for B, but the behaviour shown is triggered, scripted motion. The file citations behind this grade were spot-read by us, not re-verified line by line.", "mechanism": { "wiring": "FlyWire v783, whole graph loaded from the Shiu-model parquet files; weights are changed online by a Hebbian rule (sign-preserving, capped at 3x the original)", "neuron_model": "LIF (PyTorch port of the Shiu et al. model), 0.1 ms step", "input_mapping": "Hand-made: named sensory neurons (sugar/bitter GRNs, LC4, JO, ORNs, photoreceptors and early visual layers) get Poisson rates from simulated senses or keyboard/auto-demo stimuli", "output_mapping": "Hand-made: normalised rates of about 18 DNs (P9, DNa01/02, MDN, GF, aDN1, MN9) go into fixed formulas for two drive values and threshold-based mode selection", "trained_parts": "No trained readout. Hebbian plasticity changes the connectome weights during a run.", "body": "NeuroMechFly v2 (flygym, MuJoCo), with the HybridTurningController CPG for walking", "scripted_parts": "Grooming is a fixed front-leg oscillation. Flight is a state machine with external forces and a direct orientation override. Feeding sets the proboscis joint. Escape runs at a fixed speed. Tactile, bitter and olfactory escape are 'bridge-level backups' that skip the brain. Olfactory attraction and sound orientation are added straight to the turn command. The auto-demo drives forward walking by stimulating the P9 readout neurons themselves." }, "trained_class": "other", "trained_class_note": "Hebbian plasticity during a run", "grade_basis": [ "brain_body_bridge.py:286-302 loads 2025_Completeness_783.csv and 2025_Connectivity_783.parquet into TorchModel [direct]", "brain_body_bridge.py:385-398 Hebbian update dW = eta*pre*post*sign - alpha*W on all synapses [direct]", "brain_body_bridge.py:37-74 hand-picked DN readout IDs; 94-101 'p9' stimulus drives the same P9 cells used as the forward readout [direct]", "brain_body_bridge.py:609-731 compute_drive(): fixed formulas and if/elif mode selection with thresholds; 664-669 tactile and bitter escape marked 'bridge-level backup'; 721-729 olfactory_attraction_bias (gain 10) and sound bias added directly to turn [direct]", "fly_embodied.py:680-694 olfactory attraction/repulsion biases computed from odour geometry and passed to the bridge, bypassing the network [direct]", "fly_embodied.py:59-74 auto-demo cycles scripted stimuli (p9, lc4, sugar, jo, bitter, or56a) [direct]", "fly_embodied.py:82-95 GroomingController front-leg oscillation; 764-795 flight forces, frozen legs in flight, scripted grooming, proboscis set during feeding [direct]", "flight.py:27-40 flight is a force state machine with direct quaternion override [direct]", "No shuffled-wiring, no-graph or real-fly comparison found; compare_plasticity.py only compares weights of two runs to the starting weights [direct]" ], "grade_date": "2026-09-28", "measured_result": "Reported in the preprint and README: after 24 h, two flies starting from the same connectome showed 81% vs 47% escape, CI 0.221 vs 0.198, and 76,034 divergent synapses. Measured by the author, not reproduced by us. The plastic weight files in the repo are Git LFS pointers.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/erojasoficial-byte/fly-brain", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 27cec28d5d; GitHub API spdx_id=MIT", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux", "Windows" ], "platform_note": "Linux; Windows (WSL2)", "gpu": "optional", "gpu_note": "CUDA GPU recommended (PyTorch); falls back to CPU, which is much slower", "download_size": "About 270 MB with Git LFS (README); connectivity parquet ~101 MB, annotations TSV ~33 MB", "last_commit": { "date": "2026-03-21T12:31:05-05:00", "hash": "27cec28d5d202eb004683fb4c1a1033eec8deea0", "branch": "main" }, "pushed_at": "2026-03-21T17:31:57Z", "last_release": "none", "created_at": "2026-03-11T03:25:17Z", "stars": 64, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.5281/zenodo.19152238", "peer_review": "preprint", "link_status": { "url": "https://github.com/erojasoficial-byte/fly-brain", "status": "ok", "http_code": 200, "final_url": "https://github.com/erojasoficial-byte/fly-brain", "checked_at": "2026-10-06T09:40:08.056Z", "other_links": [ { "url": "https://doi.org/10.5281/zenodo.19152238", "status": "redirect", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/erojasoficial-byte/fly-brain [direct: clone HEAD 27cec28]", "https://doi.org/10.5281/zenodo.19152238 [indirect: cited in README]" ], "notes_limitations": "Grade C: The whole v783 graph is simulated and continuously steers walking through DN rates, but most of the behaviours shown are scripted programs that start at readout thresholds (grooming, flight, feeding, fixed-speed escape). Several sensory-to-motor paths skip the network: tactile and bitter escape, olfactory attraction (gain 10 on turning), sound orientation, and a T2 visual threat fallback computed outside the brain. The README says 'no hardcoded rules, no if-else chains', but the code contradicts this. In the auto-demo, forward walking is produced by stimulating the P9 readout neurons directly. No controls (shuffled or random wiring, no-graph baseline) and no comparison with real-fly data. The 'consciousness'/integration metrics (phi, CI) are the author's own measures. The neuron model and data files come from the Shiu et al. Drosophila brain model." }, { "id": "eon-embodied-fly", "name": "Embodied brain emulation (Eon Systems)", "type": "research", "author_or_org": "Eon Systems PBC", "summary": "Eon Systems links the Shiu et al. whole-brain LIF model of FlyWire to the NeuroMechFly body in MuJoCo, exchanging signals every 15 ms. The fly follows taste cues to food, grooms off dust and feeds. A few hand-picked descending neurons steer walking and grooming controllers trained by imitation.", "claim": { "text": "We do think it is the first embodied fly upload, the first to close a sensorimotor loop in a simulated body.", "url": "https://eon.systems/updates/embodied-brain-emulation" }, "dataset": "FlyWire FAFB", "release": "not stated for the embodied demo (the public brain-only repo ships v783)", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "FlyWire whole central brain (~140k neurons) via the Shiu et al. LIF model; vision from the connectome-constrained flyvis model inside NeuroMechFly", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "Walking controllers are modified NeuroMechFly controllers 'trained to imitate' walking; low-level body controllers are trained by imitation learning; flyvis is task-trained", "body": "FlyGym", "scripted_parts": "A few hand-picked neurons drive pre-trained controllers: DNa01/DNa02 for steering, oDN1 for forward speed, antennal DNs for grooming, MN9 for feeding. Sensory rates and DN-to-command gains are 'chosen by hand'. Vision is called 'somewhat decorative'." }, "trained_class": "not-assessed", "grade_basis": [ "https://eon.systems/updates/embodied-brain-emulation: a 'small number of descending outputs' influence body controllers 'trained by imitation learning'; mappings 'chosen by hand'; visual activations 'somewhat decorative' [direct]", "Same page, raw HTML: no GitHub or code link [direct]", "local clone: brain-only LIF benchmark code (Brian2/PyTorch/NEST GPU/GeNN, data/2025_Connectivity_783.parquet); no flygym/neuromechfly/mujoco/flyvis code [direct]", "https://github.com/eonsystemspbc org page: fly-brain, flybody (fork of TuragaLab/flybody), drosophila_brain_model_lif (fork of philshiu/Drosophila_brain_model), pathintegrationBPU, NEURD-sandbox; no embodiment repo [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY-NC-4.0", "platform": [], "gpu": "not stated", "download_size": "not stated", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "related_paper_url": "https://doi.org/10.1038/s41586-024-07763-9", "peer_review": "none", "link_status": { "url": "https://eon.systems/updates/embodied-brain-emulation", "status": "ok", "http_code": 200, "final_url": "https://eon.systems/updates/embodied-brain-emulation", "checked_at": "2026-10-06T09:40:07.770Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://eon.systems/updates/embodied-brain-emulation [direct]", "https://github.com/orgs/eonsystemspbc/repositories [direct]", "https://github.com/eonsystemspbc/fly-brain [direct: local clone]", "https://raw.githubusercontent.com/TheFlyExperiment/Virtual-Embodied-Fly/main/README.md [direct]" ], "notes_limitations": "The body/embodiment code is not public, so the loop cannot be inspected (grade U). By Eon's own description the mechanism would be C: connectome activity in a few hand-picked neurons switches or steers pre-trained, imitation-learned motor controllers. The page (dated 10 Mar 2026) is open about the limits: hand-chosen mappings, a sparse DN interface, weak effect of vision, no validation of internal dynamics. The '91%' figure (fly-brain README:117) is the accuracy of the non-embodied Shiu et al. (Nature 2024) LIF model at predicting experimental activation results. It says nothing about the embodied fly. A third-party README (TheFlyExperiment/Virtual-Embodied-Fly, not by Eon) claims '95% motor behavior accuracy' and says it is not an RL policy, which conflicts with Eon's own account. paper_url points to the peer-reviewed brain model; the embodied work itself is a blog post only." }, { "id": "flight-test-the-fly", "name": "Flight-test the fly", "type": "research", "author_or_org": "Mutaqin Aryawijaya (aryawidjaja)", "summary": "A study that runs a 20,556-neuron FlyWire v783 subcircuit (every neuron within 4 hops of the motion-sensing T4/T5 cells to the steering neurons DNa02 and DNg02) with the Shiu et al. 2024 spiking model in Brian2 as the rudder yaw damper of a simulated aircraft (a 2-state Dutch-roll model and the JSBSim Cessna 172). The yaw rate sets T4/T5 Poisson rates; the rudder follows the left-right difference of DNa02 firing. Every result is compared against degree- and sign-preserving shuffles of the wiring, a type-preserving shuffle, the airframe with no controller and a classical yaw damper, with hypotheses fixed in advance.", "claim": { "text": "\"At 1 Hz the real wiring ranked first against 19 degree-preserving shuffles ... In closed loop it had the lowest mean RMS yaw rate of 11 wirings on both plants (p = 1/11 ...). The fly controller reduces yaw-rate RMS by 14-19% relative to the bare airframe; the classical yaw damper reduces it by 52-57%.\" (README)", "url": "https://github.com/aryawidjaja/flight-test-the-fly" }, "dataset": "FlyWire FAFB", "release": "v783 as packaged by Shiu et al. 2024; extracted subcircuit of 20,556 neurons and 251,358 simulated edges (edges >= 5 synapses, at most 4 hops from T4/T5 to DNa02/DNg02), data/sub/subcircuit_v783_t5.npz", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 subcircuit from the horizontal-motion T4/T5 cells to DNa02 and DNg02 (20,556 neurons, 251,358 edges), with all edges among the selected neurons", "neuron_model": "LIF with the Shiu et al. 2024 parameters, unchanged (scripts/brain.py SHIU dict: v_th -45 mV, t_mbr 20 ms, tau 5 ms, w_syn 0.275 mV), Brian2, dt 0.1 ms", "input_mapping": "hand-made: horizontal T4/T5 cells are Poisson sources with rate = r_base + g * max(0, s * yaw rate), s set by eye and preferred direction (front-to-back motion per side)", "output_mapping": "hand-made: rudder = clip(K * washout(DNa02_R - DNa02_L - b), -1, 1), with the same 1 s washout as the classical damper; one gain K per controller chosen on tuning seeds by the same procedure", "trained_parts": "none (one scalar gain K per wiring and controller, selected by the same tuning procedure as the classical damper; test seeds 100-119 held out from tuning)", "body": "none (aircraft: JSBSim c172x 6-DOF and a linear 2-state Dutch-roll model, moderate MIL-spec Dryden turbulence)", "scripted_parts": "the aircraft simulators, a wings-level roll hold on the 6-DOF model and the washout filter" }, "trained_class": "other", "trained_class_note": "one scalar gain per wiring and controller selected by a tuning procedure", "grade_basis": [ "scripts/brain.py at d3fa996: SHIU parameter dict (line 39), T4/T5 Poisson source equations (lines 160-165), lesions as Shiu's silence() (lines 17, 183), rudder readout docstring (lines 264-273) [direct]", "scripts/nulls.py: degree- and sign-preserving shuffle by Maslov-Sneppen swaps of post endpoints (pre and w fixed), type-preserving variant with classes=(side, cell type), self-tests [direct]", "results/phase1_summary.md (auto-generated by scripts/analyze_phase1.py, 2026-10-04T14:52Z, 510 shard files): 1 Hz coherence real 0.98 vs 19 degree-preserving shuffles (median 0.309), rank 1/20, p = 0.05; gain 0.666 vs median 0.0751; vs 19 type-preserving shuffles 0.98 vs max 0.97, gain 0.666 vs max 0.609 [direct]", "results/phase2b_metrics.json (JSBSim, test seeds): mean RMS yaw rate bare 0.0685 rad/s (3.92 deg/s), fly_real 0.0558 (3.19 deg/s), fly_shuffle_00 0.0688; fly_shuffle_01 picked K = 0.268 and saturated the rudder in 99.97% of steps [direct]", "results/provenance.json: simulation commits per stage (phase 1: 510 files at e5f90c0; phase 2 test: 47 files at 393e70e; phase 3: 104 files at ab35844) [direct]" ], "grade_date": "2026-10-05", "measured_result": "Open loop (Poisson seeds 0-4): DNa02 R-L follows yaw rate with coherence 0.968-0.998 up to 2 Hz; at 1 Hz the real wiring ranks first of 20 against degree-preserving shuffles (coherence 0.98 vs median 0.31) and, only just, against type-preserving shuffles (0.98 vs max 0.97). Closed loop (20 held-out seeds): RMS yaw rate on JSBSim 3.92 deg/s with no controller, 3.19 with the fly wiring, 3.31 for the best shuffle and 1.69 for a classical yaw damper; the real wiring is lowest of 11 wirings on both plants (p = 1/11). Lesion hypothesis untestable (no departures in 1,040 flights).", "try_url": "https://fly.aryawijaya.com", "try_status": "replay app and write-up linked from the README; not loaded by us", "code_url": "https://github.com/aryawidjaja/flight-test-the-fly", "code_licence": "MIT", "code_licence_source": "LICENSE file at d3fa996 (MIT, 2026 Mutaqin Aryawijaya)", "data_licence": "FlyWire v783 derived subcircuit (CC BY 4.0, as stated in the README)", "platform": [ "browser", "Linux" ], "platform_note": "Python 3.11 with Brian2 (pyproject.toml, uv.lock); sweeps run on GitHub-hosted Linux runners (.github/workflows/sweep.yml); the replay viewer is a static three.js page", "gpu": "none", "download_size": "repository about 16.5 MB at HEAD (20 files over 512 KB: paper PDFs and figures, flight recordings); FlyWire/Shiu source files not included", "last_commit": "n/a", "pushed_at": "n/a", "last_release": { "tag": "v1.0.0", "date": "2026-10-05T07:49:24Z", "url": "https://github.com/aryawidjaja/flight-test-the-fly/releases/tag/v1.0.0" }, "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": "https://github.com/aryawidjaja/flight-test-the-fly/blob/main/paper/paper.pdf", "peer_review": "none", "peer_review_note": "write-up in the repository and the app; Zenodo DOI 10.5281/zenodo.23156035 (README badge); the last commit typesets it for bioRxiv, not found posted by us", "link_status": { "url": "https://fly.aryawijaya.com", "status": "ok", "http_code": 200, "final_url": "https://fly.aryawijaya.com", "checked_at": "2026-10-06T09:40:07.944Z", "other_links": [ { "url": "https://github.com/aryawidjaja/flight-test-the-fly", "status": "ok", "http_code": 200 }, { "url": "https://github.com/aryawidjaja/flight-test-the-fly/blob/main/paper/paper.pdf", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:3fed5f73-77cc-4b37-853f-e1e8258b4f9f", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/aryawidjaja/flight-test-the-fly [direct]" ], "notes_limitations": "A subcircuit, not the whole brain (a full-brain check exists in results/phase1_fullbrain_check.json, not read by us). The input is a hand-made yaw-rate to T4/T5 rate map, not images. The fly controller is far weaker than a classical yaw damper (14-19% vs 52-57% reduction), and against type-preserving shuffles (inputs to HS, VS, H2 and DNa02 kept) the real wiring wins only narrowly. In the seed-100 replays the right DNa02 fired no spikes, so the rudder came from one cell. Ten closed-loop shuffles give a minimum p of 1/11. One author; not peer reviewed; repository created 2026-10-04. Relevant to our looming test: an independent, pre-registered Shiu-model study in which DNa02 steering depends on the wiring." }, { "id": "flm", "name": "FLM - Fly Language Model", "type": "research", "author_or_org": "Alex Wormuth (nftechie)", "summary": "A frozen small language model gets a trained adapter that reads the state of a fixed recurrent network built from the full MaleCNS wiring. The adapter nudges the model's next-word scores. The project compares this against a matched adapter that gets no graph input.", "claim": { "text": "A frozen language model with a trained readout of the MaleCNS v1.0 fly connectome: 166,700 retained nodes and 25,582,938 directed connections.", "url": "https://github.com/nftechie/flm" }, "dataset": "MaleCNS", "release": "v1.0 (minconf 0.5 flat-connectome feather, SHA-256 pinned)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0, 166,700 nodes / 25.6 M edges, incoming-normalised contact counts", "neuron_model": "rate: abstract x = tanh(W(0.6x + 0.4 input)), one update per token", "input_mapping": "hand-made (fixed seeded random projection of token embeddings onto nodes)", "output_mapping": "learned (seeded random pooling, then a trained 278,528-parameter adapter)", "trained_parts": "adapter only; LLM backbone and graph frozen", "body": "none", "scripted_parts": "none found" }, "trained_class": "policy-or-llm", "trained_class_note": "LLM with a trained adapter; graph frozen", "grade_basis": [ "scripts/prepare_graph.py:20-23,69 downloads MaleCNS v1.0 feather files from flyem-male-cns bucket with SHA-256 checks, manifest release 'MaleCNS v1.0' [direct]", "flm/graph.py:42-90 Reservoir: tanh recurrence over the full graph; modes 'intact', 'no_edges', 'shuffled' (node relabelling, docstring says it is NOT a random-graph control) [direct]", "flm/graph.py:75 project_input = matched direct-input baseline with no fly graph computation [direct]", "scripts/train_conversation.py:142,160 trains fly adapter and direct-input control side by side; reports base, fly_adapter, direct_input_adapter, relabeled_wiring on held-out test [direct]", "https://artificialscientific.com/papers/flies-are-all-you-need results table: direct-input control 1.359328 vs fly 1.359816 NLL [direct]" ], "grade_date": "2026-09-28", "measured_result": "Paper (3 fit seeds, 32 confirmation conversations, 1,236 target tokens), NLL nats/token: frozen backbone 1.381995; fly readout 1.359816 +/- 0.000110; matched direct-input control 1.359328 +/- 0.000108; relabelled wiring without refit 1.381265; no edges 1.381995. Fly minus direct-input 95% bootstrap +0.000005 to +0.00104 (control slightly better). Source: paper web page, not a file in the repo. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/nftechie/flm", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 7251a8921d; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "macOS", "Linux" ], "gpu": "optional", "download_size": "several GB (LLM weights, corpus, MaleCNS arrays); at least 10 GB free disk", "last_commit": { "date": "2026-09-11T16:51:25-05:00", "hash": "7251a8921db4f891c39bd75ee5ad827f7031a24b", "branch": "main" }, "pushed_at": "2026-09-11T21:51:26Z", "last_release": "none", "created_at": "2026-09-11T21:48:03Z", "stars": 91, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://artificialscientific.com/papers/flies-are-all-you-need", "peer_review": "preprint", "link_status": { "url": "https://github.com/nftechie/flm", "status": "ok", "http_code": 200, "final_url": "https://github.com/nftechie/flm", "checked_at": "2026-10-06T09:40:08.334Z", "other_links": [ { "url": "https://artificialscientific.com/papers/flies-are-all-you-need", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/nftechie/flm [direct]", "https://artificialscientific.com/papers/flies-are-all-you-need [direct]" ], "notes_limitations": "Negative result: the fly graph changes predictions but a parameter-matched adapter without the graph does as well or slightly better, so no advantage from fly anatomy is shown. The only wiring control is a node relabelling without refit, which the code itself says is an interface check, not a random-graph control; there is no degree-preserving shuffle. The headline numbers come from a separate 'frozen study' whose per-token results and fitted artifacts are private; the repo only has the development recipe (it trains the same controls on dev splits). The 'neurons' are abstract tanh units, not spiking. Language ability comes entirely from the frozen LLM (README uses LFM2.5-1.2B via conversation_config; paths.py still names SmolLM2-135M as legacy). Paper lists an AI agent (Codex) as co-author. Measured by the author, not reproduced by us." }, { "id": "fly-api", "name": "fly-api", "type": "research", "author_or_org": "dtch1997", "summary": "Glue code around existing tools. It runs the Shiu et al. whole-brain LIF model (FlyWire v630) to show sugar input driving the MN9 feeding motor neuron. It also runs FlyGym walking and visual-taxis examples that do not use the connectome. A navigation demo uses an 8,991-neuron olfactory/mushroom-body LIF subnetwork with depressed KC-to-MBON synapses: its MBON output sets a hand-made turning command for the FlyGym body.", "claim": { "text": "Sugar input drives the 'eat' motor neuron (0 to 92 Hz, dose-dependent), bitter vetoes it; the conditioned brain steers the body to the rewarded odor.", "url": "https://github.com/dtch1997/fly-api" }, "dataset": "FlyWire FAFB", "release": "v630 (Shiu model data for taste demo); v783 annotations/completeness for the learning and navigation subnet", "evidence_grade": "C", "grade_note": "The file citations behind this C grade were spot-read by us, not re-verified line by line.", "mechanism": { "wiring": "Taste demo: whole-brain Shiu et al. 2024 model (v630). Navigation: 8,991-neuron olfactory/mushroom-body subnet (ORN, ALLN, ALPN, KC, APL, MBON, PAM, PPL1) extracted from the v783-based model.", "neuron_model": "LIF (Brian2, Shiu et al. parameters)", "input_mapping": "hand-made (Poisson drive on listed sugar/bitter/water GRN IDs; odor concentration at each antenna scales ORN class rates)", "output_mapping": "hand-made (MN9 rate read out; navigation: total MBON spikes per antenna -> engineered steering law with gain and arrival rule)", "trained_parts": "none by gradient; 'conditioning' depresses KC->MBON synapses active during rewarded odor pairings (eta 0.5); KC->MBON weights multiplied by a hand-set gain of 20", "body": "FlyGym (NeuroMechFly, HybridTurningController CPG)", "scripted_parts": "Walking demo uses fixed timed turn commands. Visual taxis uses FlyGym's example controller with no connectome. Navigation steering polarity, gains and 'arrived' rule are engineered." }, "trained_class": "other", "trained_class_note": "hand-set plasticity rule on KC->MBON", "grade_basis": [ "demo/track_a_lif.py:3-9,16-34 runs the Shiu et al. Brian2 model with FlyWire v630 sugar/bitter/water GRN IDs and MN9 readout [direct]", "demo/track_a_lif.py:51-56 loads 2023_03_23_connectivity_630_final.parquet from an external Drosophila_brain_model clone [direct]", "demo/track_b_embodied.py:42-50 walking uses fixed time-scheduled [left,right] drives, no brain [direct]", "demo/track_b_embodied.py:81-107 visual taxis maps retina object deviation to leg speeds through a hand-written function, no brain [direct]", "experiments/navigation/nav_demo.py:38-72 builds the olfactory-MB subnet from v783 annotations; KC->MBON weights multiplied by gain 20 (line 72) [direct]", "experiments/navigation/nav_demo.py:117-140 train() depresses active KC->MBON synapses; valence = sum of MBON spikes [direct]", "experiments/navigation/nav_demo.py:179-192 turn = steer_gain*(vL-vR)/total, plus an 'arrived, feeding' rule when valence is silenced [direct]", "experiments/navigation/runs/film/summary.json holds only the naive run (final distance 34.5 mm to A, 36.0 mm to B) [direct]" ], "grade_date": "2026-09-28", "measured_result": "Taste: MN9 0 to 92 Hz with sugar, 93% suppression with bitter (README; the rates would come from results.jsonl, which is not in the repo). Navigation: trained fly ends 2.5 mm from rewarded source vs 16.4 mm from control source; naive fly reaches neither (experiments/navigation/report.md). Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/dtch1997/fly-api", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit a6ad07a810; GitHub API spdx_id=MIT", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux" ], "gpu": "none", "download_size": "repo about 10 MB; needs a clone of philshiu/Drosophila_brain_model and FlyGym/MuJoCo", "last_commit": { "date": "2026-09-08T16:04:35+01:00", "hash": "a6ad07a810b1a43cd0356149c07b32105eb46d2a", "branch": "main" }, "pushed_at": "2026-09-08T15:04:40Z", "last_release": "none", "created_at": "2026-09-08T07:59:21Z", "stars": 10, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/dtch1997/fly-api", "status": "ok", "http_code": 200, "final_url": "https://github.com/dtch1997/fly-api", "checked_at": "2026-10-06T09:40:08.435Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/dtch1997/fly-api [direct: clone HEAD a6ad07a]" ], "notes_limitations": "The author says most of the science and code already existed. The sugar-to-MN9 result reproduces Shiu et al. 2024, running their model from an external clone; it is not copied into the repo. That result has a bitter/water specificity check, but its comparison with real flies is Shiu's, not this repo's. The embodied demos do not use the full brain: walking and visual pursuit have no connectome at all, and navigation uses an 8,991-cell filtered subnet with a hand-set KC->MBON gain of 20. There, the brain supplies only a scalar valence per antenna, and an engineered steering law turns it into movement. The naive-vs-trained comparison controls for learning, not for wiring (no shuffled graph). Only the naive trajectory summary is stored. The trained run's numbers are in the report only." }, { "id": "fly-brain-zero-shot", "name": "Are fruit flies zero-shot adapters?", "type": "research", "author_or_org": "Vibhakar Mohta (vib2810)", "summary": "A closed loop in which the FlyVis eye model sees a dark pole, 27 of its output cell types drive about 31,000 matching FlyWire neurons, the Shiu et al. whole-brain spiking model runs on the GPU with unchanged weights, and the left-right difference of the DNa02 turn neurons sets the turning of a NeuroMechFly body whose legs are moved by a hand-built tripod rhythm. Two legs are removed mid-walk from cloned flies. The same flies are rerun with scrambled wiring and with the turn neurons disconnected, and poles reached per 10 s are compared.", "claim": { "text": "A spiking model of the whole fruit fly brain steers a simulated body toward dark poles; after two legs are cut mid-walk it keeps finding the poles; scramble its wiring and it can't.", "url": "https://github.com/vib2810/fly-brain-zero-shot" }, "dataset": "FlyWire FAFB", "release": "v783 (Shiu et al. Connectivity_783.parquet, pinned commit, checksummed); whole brain, 138,639 neurons", "evidence_grade": "A", "grade_note": "Borderline A: walking comes from a hand-built rhythm generator and the brain only steers through one tuned gain; the scrambled comparison after the leg cut has 16 flies from a single shuffle, and only the per-fly summary table is stored.", "mechanism": { "wiring": "Whole FlyWire v783 brain as used by Shiu et al., unmodified weights", "neuron_model": "spiking: Shiu et al. leaky integrate-and-fire, ported to Warp for batched GPU runs", "input_mapping": "FlyVis (pretrained, connectome-constrained optic-lobe model) output types drive same-type FlyWire neurons as Poisson inputs, column by column; per-type scale set from responses to moving bars", "output_mapping": "hand-set: turn asymmetry = clip(-k*(DNa02 L - R)/100 Hz - b, +-0.8); gain k chosen on separate training flies", "trained_parts": "None in the brain. Two settings tuned (handoff scale per cell type, turn gain k); FlyVis was trained by its authors", "body": "NeuroMechFly (flygym) on MuJoCo-Warp", "scripted_parts": "Leg rhythm is a hand-built tripod pattern generator (the brain only sets turning; walking pace is not from the brain). Pole images are computed analytically. Poles respawn ahead when reached." }, "trained_class": "front-end", "trained_class_note": "pretrained FlyVis front end; two settings tuned", "grade_basis": [ "scripts/download_data.sh:9-20 Shiu Connectivity_783.parquet and Completeness_783.csv at a pinned commit with md5 checks [direct]", "src/flyloop/data.py:1-45 FlyWire v783 edges from Shiu files ('Excitatory x Connectivity') [direct]", "src/flyloop/brain.py:19,153-156 Shiu LIF constants (W_SYN 0.275); wiring argument used for the null model [direct]", "src/flyloop/vision_loop.py:1-6,158,166 Poisson handoff from FlyVis; asym = clip(-k*(L-R)/100 - bias) sets CPG turning [direct]", "src/flyloop/wiring.py:8-12 null model: postsynaptic targets permuted across all connections, keeping presynaptic neuron, sign and synapse count [direct]", "src/flyloop/vision_loop.py:301-322 shuffle_test sweeps k = 1, 2, 4, 8 on scrambled wiring [direct]", "src/flyloop/vision_loop.py:324-362 scramble_at_cut clones each fly at the cut with real vs scrambled wiring [direct]", "assets/results_per_fly.csv: per-fly poles; recomputed means intact 2.96 (n=48) real vs 0.34 (n=32) scrambled vs 0.12 (n=48) unplugged; after cut 3.14 (n=64) vs 0.12 (n=16) vs 0.52 (n=48) [direct]" ], "grade_date": "2026-09-29", "measured_result": "Poles reached per 10 s, from the stored per-fly table (our recomputation matches the README): intact flies 2.96 with real wiring, 0.34 with scrambled wiring, 0.12 with turn neurons disconnected; after the two-leg cut 3.14 real, 0.12 scrambled, 0.52 disconnected. Author reports scrambled brains' DNa02 falls from about 30 Hz to 2 Hz or less for every gain tried. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/vib2810/fly-brain-zero-shot", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 0c63ba5cdd; GitHub API spdx_id=MIT", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux" ], "platform_note": "Python 3.12 with PyTorch, Warp and MuJoCo-Warp; developed on an RTX 3070 Laptop (8 GB)", "gpu": "required", "gpu_note": "NVIDIA GPU", "download_size": "about 100 MB Shiu connectivity plus FlyWire annotations and FlyVis weights via the download script", "last_commit": { "date": "2026-09-29T03:01:48-07:00", "hash": "0c63ba5cdd123f8d38eea62deb574353f0ec2632", "branch": "main" }, "pushed_at": "2026-09-29T10:01:50Z", "last_release": "none", "created_at": "2026-09-29T09:17:29Z", "stars": 1, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/vib2810/fly-brain-zero-shot", "status": "ok", "http_code": 200, "final_url": "https://github.com/vib2810/fly-brain-zero-shot", "checked_at": "2026-10-06T09:40:08.746Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/vib2810/fly-brain-zero-shot [direct: clone HEAD 6c3a5e7, 2026-09-29]" ], "notes_limitations": "Walking itself comes from a hand-built pattern generator; the brain only supplies the turn signal through one hand-set formula on DNa02, with a gain chosen on training flies. The eye model was fitted by its own authors. The control scrambles postsynaptic targets across the whole brain, which shows the mapped wiring is needed to route the eye signal to DNa02, but it is a strong perturbation; the post-cut scrambled condition has only 16 flies from one shuffle. Only the per-fly summary table is stored; raw run files are regenerated by the scripts, and GPU runs are not bitwise reproducible. Builds on the Eon Systems embodied fly and Shiu et al. model, which the author credits. Not a copy of a catalogued project." }, { "id": "fly-cartpole", "name": "fly-cartpole", "type": "research", "author_or_org": "Curt Park", "summary": "A flight-stabilisation circuit cut from MaleCNS v1.0 (5,459 neurons: ocellar, haltere and HS sensory groups down to the wing-amplitude motor neurons, run as linear rate units) balances Gymnasium's CartPole. The circuit's couplings never change; only three to five sensor gains are tuned by reward-modulated random search. The author compares it with degree-preserving shuffled wiring, the untuned circuit, the circuit's own linear map tuned the same way, LQR and random pushes on fresh seeds, with permutation tests. An earlier mushroom-body version (dopamine-gated learning) is still in the repository with its own negative result. A browser viewer simulates the circuit live.", "claim": { "text": "The circuit a fly uses to keep its balance in the air balances the pole: wired from MaleCNS v1.0, 5,459 neurons, no backpropagation; learning only changes five sensor gains.", "url": "https://github.com/Curt-Park/fly-cartpole" }, "dataset": "MaleCNS", "release": "v1.0; flight-stabilisation circuit, 5,459 neurons (data/flight*); the earlier mushroom-body circuits (343 PNs, 1,893 KCs, 48 MBONs, 332 DANs) remain in the repository", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 flight-stabilisation circuit (ocellar, haltere, HS cells to wing-amplitude motor neurons), couplings from synapse counts; never changed by learning", "neuron_model": "rate: linear rate units with leak, settled each step", "input_mapping": "hand-set: pole angle and drift onto ocellar neurons, pole angular rate onto haltere neurons, cart velocity onto HS cells as optic flow, by side", "output_mapping": "hand-set: right minus left wing-amplitude motor activity steers the cart", "trained_parts": "three sensor gains (five with the landmark task) by reward-modulated random search; learning rates from the author's tuning search", "body": "none (Gymnasium CartPole)", "scripted_parts": "CartPole physics and reward are Gymnasium's; the reward signal is computed outside the circuit" }, "trained_class": "other", "trained_class_note": "3-5 sensor gains by reward-modulated random search", "grade_basis": [ "README.md 'Results' at a8c6257 (2026-10-03): the headline experiment is now the flight-stabilisation circuit [direct]", "results/linear-retuned/summary.md at a8c6257: per-seed final-100 means, seeds 160-179: fly-reflex 275.6 ± 9.4, fly-reflex-adaptive 499.9 ± 0.3, fly-reflex-adaptive-shuffled 165.9 ± 182.8, random 22.1 ± 0.8, linear-adaptive 499.6 ± 0.7, lqr 500.0; permutation p: adaptive vs shuffled 0.0001, adaptive vs linear map 0.1797 (two-sided) [direct]", "src/fly_cartpole/reflex.py shuffle_flight + circuit.py shuffle_edges: degree-preserving directed double-edge swaps, couplings travel, sensor and motor identities kept; reflex_report.py line 196 shuffles once per evaluation seed [direct]", "results/lqr-station/summary.md: holding station past the 500-step cap, circuit 0 of 200 track exits vs its linear map 8 of 200; LQR 0.03 m from centre vs circuit 0.12 m [direct]", "results/summary.md (unchanged since 3f676bb, 2026-10-01): the earlier mushroom-body study, fly-bilateral 392.5 vs shuffled 390.3, p = 0.4771 [direct]" ], "grade_date": "2026-10-03", "measured_result": "CartPole mean episode length over the final 100 episodes, 20 fresh seeds (results/linear-retuned/summary.md at a8c6257): self-tuned flight circuit 499.9 ± 0.3, the same circuit with degree-preserving shuffled wiring 165.9 ± 182.8 (author's permutation p = 0.0001), untuned circuit 275.6, the circuit's own linear map tuned the same way 499.6 (p = 0.18), LQR 500.0, random pushes 22.1. Past the 500-step cap the circuit holds station better than its linear map (0 vs 8 track exits of 200) but worse than LQR. The earlier mushroom-body version learned no better than shuffled wiring (392.5 vs 390.3, p = 0.48). Numbers are the author's; not recomputed by us this run.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Curt-Park/fly-cartpole", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit a8c62579bb; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "browser", "Linux", "macOS", "Windows" ], "platform_note": "Python with uv; static web viewer", "gpu": "none", "download_size": "processed mushroom-body data committed (data/*.npz); the full extract downloads 1.1 GB once", "last_commit": { "date": "2026-10-03T18:13:25+09:00", "hash": "a8c62579bb4566baa517e55c63d53f85344018d4", "branch": "main" }, "pushed_at": "2026-09-30T00:31:02Z", "last_release": "none", "created_at": "2026-09-29T23:47:25Z", "stars": 1, "stars_date": "2026-10-05", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Curt-Park/fly-cartpole", "status": "ok", "http_code": 200, "final_url": "https://github.com/Curt-Park/fly-cartpole", "checked_at": "2026-10-06T09:40:08.862Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/Curt-Park/fly-cartpole [direct: clone HEAD 0455e2e, 2026-09-30]" ], "notes_limitations": "Re-graded 2026-10-03 after a redesign: the author's headline is now a MaleCNS flight-stabilisation circuit, and shuffling its wiring does hurt (165.9 vs 499.9). But balancing needs no more than a small linear controller: the circuit's own linear map scores the same and LQR reaches 500 without learning, so the wiring matters mainly through its pathway signs (the author's reading). The real arm sits at the 500-step ceiling. Neurons are linear rate units and the mapping from CartPole to the fly's senses is the author's own; the author says none of this is evidence about real flies. The earlier mushroom-body study (no wiring effect) remains in the repository." }, { "id": "fly-connectome-adds-to-body", "name": "FLY-lab: What a fly connectome adds to controlling a body", "type": "research", "author_or_org": "Recluse (FLY-lab contributors)", "summary": "Couples the Shiu et al. FlyWire v783 LIF brain model to the NeuroMechFly body in FlyGym and tests five action selectors on simple turn-toward-touch tasks: constant command, a two-line rule, the live connectome, a replay of another episode's brain output, and five degree-preserving shuffled connectomes.", "claim": { "text": "on our tasks the connectome showed no advantage whatsoever over a controller two lines long", "url": "https://github.com/Recluse/FLY-lab" }, "dataset": "FlyWire FAFB", "release": "v783 (138,639 neurons, 15,091,983 connections; v630 for the sugar reference check)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 full brain via Shiu et al. model (pinned commit)", "neuron_model": "LIF", "input_mapping": "hand-made (Poisson drive to 150 left / 155 right head-bristle mechanosensory neurons)", "output_mapping": "hand-made (left minus right spike count over 645/646 descending neurons; 3 readout constants grid-searched on calibration seeds to agree with the two-line rule)", "trained_parts": "none (readout constants chosen by grid search)", "body": "FlyGym", "scripted_parts": "walking comes from the stock FlyGym hybrid turning controller (CPG + reflexes); brain only sets speed/turn drives every 15 ms" }, "trained_class": "readout-or-decoder", "trained_class_note": "readout constants chosen by grid search", "grade_basis": [ "brain_loop.py:28-62 shuffle_edges: double-edge swap preserving in/out degree per sign, weight multiset, no self-loops or duplicates; :64-80 self-check asserts these [direct]", "brain_loop.py:85-105 shuffled_connectivity on Connectivity_783; results/shuffled-connectivity-4000.json: 99.997% of 15,091,983 edges changed [direct]", "brain_loop.py:139-155 loads Completeness_783/Connectivity_783, head-bristle inputs, all descending neurons as readout [direct]", "pilot.py:34 policies C (connectome), D (replay of another episode), E (shuffled) [direct]", "results/cd-test-v1/compare-C-vs-B-ab-test-v1/summary.json: C 30/30 and B 30/30 on none/left/right [direct]", "results/e-compare-fixed-v1/summary.json: left E-minus-C -0.69, bootstrap 95% [-0.96, -0.40]; results/e-compare-recal-v1/summary.json: -0.47 [-0.87, -0.08] [direct]" ], "grade_date": "2026-09-28", "measured_result": "Test seeds (30 per condition): connectome 30/30 on no-stimulus, turn-left and turn-right; two-line rule 30/30 on each (difference 0, conservative 95% interval +/-13.6 pp); replay of another episode 0/30 on each; five degree-preserving shuffles: 30/30 no-stimulus, 0-30/30 left (mostly a constant bias), 0/30 right. Shuffled minus connectome: -69 pp (left, fixed readout, hierarchical bootstrap [-96, -40]), -47 pp after readout recalibration ([-87, -8]), -100 pp (right). Files: results/cd-test-v1/, results/e-compare-*-v1/summary.json. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Recluse/FLY-lab", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit c9ee19f4f7; GitHub API spdx_id=MIT", "data_licence": "CC-BY-NC-4.0", "platform": [ "macOS", "Linux" ], "gpu": "none", "download_size": "not stated (Shiu model repo with v783 tables; two Python environments)", "last_commit": { "date": "2026-09-18T02:09:36+03:00", "hash": "c9ee19f4f7704d7bf3a26099719c2b85861911db", "branch": "main" }, "pushed_at": "2026-09-18T03:47:34Z", "last_release": "none", "created_at": "2026-09-12T12:36:45Z", "stars": 2, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://github.com/Recluse/FLY-lab/blob/main/paper/fly-lab-en.pdf", "peer_review": "none", "link_status": { "url": "https://github.com/Recluse/FLY-lab", "status": "ok", "http_code": 200, "final_url": "https://github.com/Recluse/FLY-lab", "checked_at": "2026-10-06T09:40:08.936Z", "other_links": [ { "url": "https://github.com/Recluse/FLY-lab/blob/main/paper/fly-lab-en.pdf", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/Recluse/FLY-lab [direct]" ], "notes_limitations": "Careful, pre-registered negative result: connectome equals a two-line rule on tasks that need one bit; live activity and real topology are needed for the lateralised pathway in this assembly. Tasks are very simple (1 s episodes, turn toward a touch); stopping is not tested for the brain. Readout constants were tuned to agree with rule B on calibration seeds. Brain pass is computed separately from the body (open loop), shown bit-exact for these stimuli. Gait is entirely the stock FlyGym controller. A side SAT/reservoir experiment reports 95-99% decodability on real vs 68-73% on shuffle (docs/sat-experiment.md). Measured by the author, not reproduced by us." }, { "id": "fly-exe", "name": "Fly.exe (MaleCNS Virtual Fly)", "type": "research", "author_or_org": "Ibtisam Mohammad", "summary": "A LIF simulator runs the 'Traced' subset of MaleCNS v1.0 (165,122 neurons, 25,563,197 edges) in closed loop with NeuroMechFly bodies. In the swarm demo, an analytic bearing/size encoder drives lamina cells, and two DNp descending population rates feed FlyGym's turning CPG. Other showcases add declared shortcuts: odour-gradient steering terms and replayed grooming kinematics triggered by neural readouts.", "claim": { "text": "Runs the whole Traced universe of the male CNS connectome, 165,122 neurons and every one of the 25,563,197 edges, inside a physical fly body, in closed loop.", "url": "https://github.com/Ibtisam-Mohammad/Fly.exe" }, "dataset": "MaleCNS", "release": "v1.0 ('Traced' annotation status subset: 165,122 of 166,700 bodies)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0, all Traced bodies and all edges between them; 12 flies share one connectivity allocation", "neuron_model": "LIF (NumPy/GeNN engines, membrane and adaptation state, noise); photoreceptor output edges zeroed by sign policy", "input_mapping": "hand-made (analytic bearing and angular size of objects mapped onto lamina L1-L5 cells; no rendered pixels)", "output_mapping": "hand-made (filtered rates of DNp* populations, 160 left / 158 right, chosen after measuring visual input share on the graph -> two drives with thresholds and yaw gain)", "trained_parts": "none (no learning or plasticity in the swarm run); STP/completeness parameters fitted in separate stage-2 validation work", "body": "FlyGym (NeuroMechFly, HybridTurningController CPG, fixed leg adhesion)", "scripted_parts": "Gait is FlyGym's engineered CPG. Onset timer holds all flies still for 1.5 s. The decoder has no stop state. In the eon showcase: odor-gated direct DNg97 drive, a raw odor-gradient term in the steering controller, a grooming readout that triggers a published kinematic replay, and MN9 triggering an engineered proboscis pose." }, "trained_class": "none", "trained_class_note": "no learning in the run", "grade_basis": [ "src/flysim/connectome.py:239 source_release default 'male-cns:v1.0' [direct]", "src/flysim/engines/lif.py:18-160 LIF engine with membrane tau, reset and input frames [direct]", "src/flysim/swarm3d.py:13-16,388 each fly driven by two normalised descending drives into FlyGym HybridTurningController [direct]", "src/flysim/swarm3d_vision.py:124-134 entry layer lamina L1-L5; all 66,533 photoreceptor output edges zeroed; strongest-object drive per cell [direct]", "configs/experiments/demo01-visual-operating-point-v1.json:127-132 decoded populations DNp* 160/158 chosen by measured visual-input share; DNa01/02 recorded but not decoded [direct]", "artifacts/showcase/swarm3d-v1/run-exact-summary.json: graph 165,122 neurons / 25,563,197 edges, shuffle null; decoder thresholds and yaw gain; 12 flies closed on food vs 8 in run-stimulus-absent-summary.json [direct]", "artifacts/showcase/eon-showcase-v1/acceptance.json: input- and readout-ablation controls pass; shuffled-connectome and zero-weight runs are 'diagnostic only' and do not gate; exact seed 3 incomplete; limitations deny 'autonomous connectome-generated behaviour' [direct]", "artifacts/showcase/eon-showcase-v1/showcase-manifest.json:29-37 declared bridges incl. raw odor-gradient steering term and grooming kinematic replay [direct]" ], "grade_date": "2026-09-28", "measured_result": "Swarm (one seed, one arena): 12/12 flies entered locomotion vs 0/12 in the stimulus-absent control; 12/12 ended within 1 mm of an object vs 0/12; 9,337 vs 164 neurons spiking per fly per interval (README; artifacts/showcase/swarm3d-v1/*summary.json). Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Ibtisam-Mohammad/Fly.exe", "code_licence": "GPL-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit cc25411e00; GitHub API spdx_id=GPL-2.0", "data_licence": "CC-BY-4.0", "platform": [ "Linux" ], "gpu": "required", "gpu_note": "required for the full closed loop (one GPU; CPU engine for tests)", "download_size": "MaleCNS v1.0 tables (about 1 GB); repo about 50 MB", "last_commit": { "date": "2026-09-15T10:08:16+05:30", "hash": "cc25411e00f945e7330cb6f400372cce315c0802", "branch": "main" }, "pushed_at": "2026-09-15T07:05:52Z", "last_release": { "tag": "v0.1.0 — Twelve embodied flies on the full MaleCNS connectome", "date": "2026-09-15T07:08:37Z", "url": "https://github.com/Ibtisam-Mohammad/Fly.exe/releases/tag/v0.1.0" }, "created_at": "2026-09-12T15:20:28Z", "stars": 21, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Ibtisam-Mohammad/Fly.exe", "status": "ok", "http_code": 200, "final_url": "https://github.com/Ibtisam-Mohammad/Fly.exe", "checked_at": "2026-10-06T09:40:08.798Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "wiring-null", "wiring_effect": "mixed", "sources": [ "https://github.com/Ibtisam-Mohammad/Fly.exe [direct: clone HEAD cc25411]" ], "notes_limitations": "The only control for the swarm is removing the stimulus, not a wiring control. Shuffled and zero-weight runs exist for the eon showcase but are marked diagnostic, not activity-matched, and do not gate acceptance. The swarm is one seed in the third arena tried, with headings bounded so food falls in the encoder's field. Vision is analytic (exact object positions), not optical. The readout population was picked after measuring the graph, and walking is FlyGym's engineered CPG. The eon showcase adds shortcuts around the network (a raw odor-gradient steering term, a direct DNg97 drive) and turns grooming and feeding readouts into replayed or engineered motor programs, so those behaviours are C-level. The artifacts themselves disclaim 'autonomous connectome-generated behaviour'. Female body with male CNS. Very large AI-agent-style documentation (AGENTS.md 74 KB), matching HN criticism. Not a copy of a catalogued project; stage-2 work compares against the Shiu model and published physiology." }, { "id": "fly-ocr", "name": "Fly OCR", "type": "research", "author_or_org": "Jerry Liu (FlyOCR contributors)", "summary": "Character images are sampled by a compound-eye-like receptor map, fed as currents into a fixed spiking MaleCNS circuit, and a small trained decoder reads downstream spike counts to recognise printed letters and digits. Pixel baselines, label-shuffle, edge-removal and degree-preserving rewiring controls are reported.", "claim": { "text": "A simulated fly circuit recognizes printed letters and numbers from PDF pixels.", "url": "https://github.com/jerryjliu/fly_ocr" }, "dataset": "MaleCNS", "release": "v1.0 (minconf 0.5 feather, SHA-256 pinned)", "evidence_grade": "A", "grade_note": "Borderline A: the rewiring control was run only on a 200-image pilot, not on the final model.", "mechanism": { "wiring": "MaleCNS v1.0, 166,700 neurons / 25,582,938 connections, frozen", "neuron_model": "LIF-type spiking point neurons (native C++ kernel adapted from DOOMFLY; tests compare against Brian2)", "input_mapping": "hand-made (image pixels -> retinal receptor map -> photoreceptor/lamina drive)", "output_mapping": "learned (decoder on spike counts of up to 1,024 variance-selected downstream neurons)", "trained_parts": "readout/decoder only (letters: 64 hidden units, 266,628 parameters; digits: logistic regression)", "body": "none", "scripted_parts": "segmentation, normalisation, word gaps and table geometry are conventional code outside the circuit" }, "trained_class": "readout-or-decoder", "trained_class_note": "trained readout/decoder", "grade_basis": [ "configs/malecns-v1.json: release 'MaleCNS v1.0', feather URLs with SHA-256 [direct]", "src/flyocr/experiments/controls.py:1,47-48 degree-preserving target permutation with in-degree assertion, heads retrained per seed (41-43) [direct]", "reports/controls/controls.json: edge lesion 10% (20/200); randomized seed 41: 65% (130/200) [direct]", "docs/results.md:7-16 fly digits 89.0% vs raw-pixel linear 99.0%, retinal samples linear 91.8%, CNN 100%, shuffled labels 10.4% [direct]", "docs/research-report.md:176-178 intact pilot 82% vs rewired 65.0/63.5/54.5%; mismatched labels 2.3% on 68-class benchmark [direct]" ], "grade_date": "2026-09-28", "measured_result": "Final letter model: 87.6% on 1,632 glyphs (68 classes), 85.0% on letters; mismatched features 2.3%. Digit task (docs/results.md): fly 89.0% vs raw-pixel linear 99.0%, retinal-samples linear 91.8%, CNN 100%. Digit pilot controls (reports/controls/, n=200): intact 82%, edges removed 10%, three degree-preserving rewirings with retrained heads 65.0% / 63.5% / 54.5%. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/jerryjliu/fly_ocr", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 48cf341e99; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "macOS", "Linux" ], "gpu": "none", "download_size": "several GiB graph cache; at least 10 GiB free (viewer runs without the graph)", "last_commit": { "date": "2026-09-12T20:37:23-07:00", "hash": "48cf341e99c17dc911fb09fdc8419d98b4d0ea86", "branch": "main" }, "pushed_at": "2026-09-13T03:39:07Z", "last_release": { "tag": "Fly OCR: compound eyes, characters, and a fixed connectome", "date": "2026-09-13T02:07:22Z", "url": "https://github.com/jerryjliu/fly_ocr/releases/tag/v0.1.0" }, "created_at": "2026-09-13T01:08:56Z", "stars": 90, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://github.com/jerryjliu/fly_ocr/blob/main/docs/fly-ocr-research-report.pdf", "peer_review": "none", "link_status": { "url": "https://github.com/jerryjliu/fly_ocr", "status": "ok", "http_code": 200, "final_url": "https://github.com/jerryjliu/fly_ocr", "checked_at": "2026-10-06T09:40:09.117Z", "other_links": [ { "url": "https://github.com/jerryjliu/fly_ocr/blob/main/docs/fly-ocr-research-report.pdf", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/jerryjliu/fly_ocr [direct]" ], "notes_limitations": "Controls are real and in code, but the wiring control (rewiring) was run only on a 200-image digit pilot, not on the final letter model; rewiring also changes activity scale, which the author flags. Plain pixel classifiers beat the fly circuit (99-100% vs 89% on digits), so no advantage from the connectome is claimed. The decoder is trained; input mapping and segmentation are engineered. The 3D fly in videos is illustrative animation. Measured by the author, not reproduced by us." }, { "id": "fly-racer", "name": "Fly-Racer", "type": "research", "author_or_org": "Supat Roongpraiwan (supat-roong)", "summary": "A car-racing agent (Gymnasium CarRacing-v3) whose recurrent core is wired like a 3,350-neuron steering pathway from MaleCNS v1.0 (visual projection and ring neurons, central complex, bridging and descending neurons; 164,773 connections). Topology and signs are fixed; synapse strengths, time constants, biases, a CNN encoder and a Beta readout are trained with PPO. The same pipeline trains a degree-preserving shuffle, a sign-shuffled core and a plain MLP.", "claim": { "text": "\"On this evidence the fly's wiring makes learning faster and driving more consistent, though with one seed per core the differences in mean are small.\" fly 903.2 vs signs 896.9, mlp 894.0, random 883.7 mean score over 100 held-out tracks (README)", "url": "https://github.com/supat-roong/fly-racer" }, "dataset": "MaleCNS", "release": "v1.0 via neuPrint (male-cns:v1.0); Tier S subgraph of 3,350 neurons (800 visual input, 2,430 central brain, 120 descending) and 164,773 connections (configs/tier_s.yaml, README)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 subgraph selected by cell-type patterns (LPLC2, LC4, LC9-16, ER ring neurons; EPG, PEN, PFL, PFN, hDeltaB; descending neurons) plus the strongest bridging neurons", "neuron_model": "rate: leaky units v <- v + (dt/tau)(-v + W r + b + I), r = tanh(relu(v)), 6 settling steps per frame (flyracer/models/flyrnn.py)", "input_mapping": "learned: a CNN encoder on 4 stacked 96x96 grey frames projects 256 features onto the input neurons", "output_mapping": "learned: Beta readout from descending neurons to steer, gas and brake, initialised with a left/right steering prior", "trained_parts": "PPO trains synapse strengths (log_w), a global gain, per-neuron time constants and biases, the CNN encoder, the input projection, the readout and an MLP critic", "body": "none (a 2D car in CarRacing-v3)", "scripted_parts": "none beyond the Gymnasium environment" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "PPO trains synapse strengths, time constants and biases on fixed connectome topology, plus a CNN encoder and readout", "grade_basis": [ "flyracer/models/flyrnn.py at 71ede00: sign buffer fixed per presynaptic neuron, trainable log_w, log_gain, log_tau and bias (lines 48-55) [direct]", "flyracer/connectome/shuffle.py: degree_preserving_shuffle (permutes postsynaptic endpoints, keeps in- and out-degree and per-neuron signs) and shuffle_signs [direct]", "configs/tier_s.yaml: dataset male-cns:v1.0, cell-type patterns for inputs, central complex and descending neurons [direct]", "Control scores are given only in the README table and docs/core_comparison.png; no run, eval or episode files are committed (runs/ is not in the repository) [direct]" ], "grade_date": "2026-10-06", "measured_result": "Not verified by us: the four-core comparison (one seed each) exists only as a README table and a figure.", "try_url": null, "try_status": "no no-install option (Python training and evaluation; checkpoints not published)", "code_url": "https://github.com/supat-roong/fly-racer", "code_licence": "MIT", "code_licence_source": "LICENSE file at 71ede00 (MIT, 2026 Supat Roongpraiwan)", "data_licence": "MaleCNS v1.0 (CC BY 4.0), fetched by the user from neuPrint with a personal token; not included", "platform": [ "macOS" ], "platform_note": "Python 3.12 with PyTorch and Gymnasium box2d; the README targets Apple Silicon (about 6-8 hours for 1.5M decisions on an M4)", "gpu": "not stated", "gpu_note": "Apple Silicon assumed by the README; no CUDA requirement found", "download_size": "repository about 41 MB (mostly GIF/PNG figures); connectome subgraph fetched from neuPrint (size not stated)", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "none", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/supat-roong/fly-racer", "status": "ok", "http_code": 200, "final_url": "https://github.com/supat-roong/fly-racer", "checked_at": "2026-10-06T09:56:57.556Z" }, "last_verified": "2026-10-06", "added_in_run": "run:c0ecce13-d7a1-40c8-b30e-661b847df700", "controls": "wiring-null", "wiring_effect": "mixed", "sources": [ "https://github.com/supat-roong/fly-racer [direct]" ], "notes_limitations": "A trained agent with connectome topology, not a simulation of fly driving: most of the skill can sit in the trained encoder, weights and readout. The wiring controls are implemented in code, but their results are only in the README (one seed per core, final means within 20 points, the MLP ahead at the end of training), so they do not raise the grade." }, { "id": "fly-self-driving", "name": "Fly Self Driving", "type": "research", "author_or_org": "suanmiao (built with agents in a Kylon workspace)", "summary": "The traced MaleCNS graph (165,122 neurons, 25.6 M edges) runs as a recurrent rate network. Frozen random maps connect pixels to optic-lobe sensory cells and ventral-cord motor cells to steering. One gain per synapse and one leak per neuron (25.7 M parameters) are trained by imitating a pure-pursuit expert plus DAgger, to drive a ray-cast street with traffic. Code includes a randomly rewired graph control and linear and MLP baselines.", "claim": { "text": "The measured wiring of a fruit fly, run as a recurrent network and trained to steer from a 64 x 32 windscreen; on held-out streets with traffic it completes 20, 19 and 19 of 20 (58/60).", "url": "https://github.com/suanmiao/fly-self-driving" }, "dataset": "MaleCNS", "release": "v1.0 (traced graph via flyhard 'graph-traced-v1', sha256 eff4093b...)", "evidence_grade": "A", "grade_note": "Borderline A: the headline and rewired-graph numbers appear only in docs/results.md, and the stored baseline JSON does not match that document. One training seed per condition.", "mechanism": { "wiring": "MaleCNS v1.0 traced graph, 165,122 neurons, 25,563,197 edges; adjacency fixed", "neuron_model": "rate: signed rate state, 4 graph updates per 50 ms decision, state carried between decisions (flyhard recipe)", "input_mapping": "hand-made (frozen seeded random signed pixel map onto 4,114 optic-lobe sensory neurons)", "output_mapping": "hand-made (frozen random linear readout from 708 VNC motor neurons, tanh to steering)", "trained_parts": "All 25,728,319 per-synapse gains and per-neuron leaks, by truncated BPTT behaviour cloning then 3 DAgger rounds against a pure-pursuit expert. Only the topology comes from the fly.", "body": "none (simulated car; FlyGym only in the reproduced wheel pilot)", "scripted_parts": "The expert teacher is scripted. The Blender video is a replay of logged state. No braking behaviour was learned." }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "all 25.7 M synapse gains trained", "grade_basis": [ "tasks/street/train_street.py:1-9 docstring: measured adjacency frozen, per-edge gains and leaks train, frozen random interfaces; '--graph-mode shuffled is the control' [direct]", "tasks/street/train_street.py:26-47 StreetPolicy: SparseConnectome core, random pixel->sensory map and random motor decoder as fixed buffers, 4 updates per decision [direct]", "tasks/street/train_street.py:171,204-205 shuffled mode permutes column indices of the graph with a fixed seed [direct]", "tasks/flat-road/mlp_baseline.py and tasks/street/train_street.py --model linear/mlp: non-connectome baselines (file present, not opened)", "results/street/street-v5-measured-stateful.json: graph sha256 eff4093b..., 25,728,319 trainable params, cloning-only 12/20 with traffic, 20/20 without [direct]", "results/street/street-v4-linear.json, street-v4-mlp40.json: v4 task baselines complete 1/20 with traffic [direct]", "docs/results.md: DAgger w16 20/19/19 of 20 vs rewired graph 16/20, linear 12/20, MLP-40 18/20; flat-road task rewired graph matched or beat measured [direct]" ], "grade_date": "2026-09-28", "measured_result": "Street task v5 with traffic, held-out streets: fly connectome + DAgger 20, 19, 19 of 20 (three sets); same recipe on a randomly rewired graph 16/20; linear 12/20; MLP-40 18/20; expert 20/20 (docs/results.md). Speed task v6: 18/19/16 of 20 vs rewired 5/20, but a second training seed gave 13/20. Flat road: rewired graph matched or beat measured wiring, MLP beat both. One training seed per condition. The best-model and rewired street metrics are in docs only, not in results/ JSON. Measured by the author, not reproduced by us.", "try_url": "https://fly-self-driving.kylon.app", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/suanmiao/fly-self-driving", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 3516a094a6; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "macOS", "Linux" ], "platform_note": "macOS (Apple silicon); Linux", "gpu": "required", "gpu_note": "required (Apple MPS or CUDA for training)", "download_size": "MaleCNS data about 1 GB via flyhard; checkpoints about 600 MB each (not included)", "last_commit": { "date": "2026-09-13T08:21:34-07:00", "hash": "3516a094a6462e2a4ab7cc517286d5c3bc3ff63e", "branch": "main" }, "pushed_at": "2026-09-13T15:21:36Z", "last_release": "none", "created_at": "2026-09-13T15:11:44Z", "stars": 28, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://fly-self-driving.kylon.app", "status": "ok", "http_code": 200, "final_url": "https://fly-self-driving.kylon.app", "checked_at": "2026-10-06T09:40:10.243Z", "other_links": [ { "url": "https://github.com/suanmiao/fly-self-driving", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "wiring-null", "wiring_effect": "mixed", "sources": [ "https://github.com/suanmiao/fly-self-driving [direct: clone HEAD 3516a09]", "https://fly-self-driving.kylon.app [not opened]" ], "notes_limitations": "A proper wiring control and non-connectome baselines exist in code, but the gap is small and fragile: one training seed per condition, 20 streets per set, and a second seed on the speed task dropped to 13/20. On the first task the rewired graph did as well and an MLP did better. All 25.7 M synaptic gains are trained, so the fly supplies only the topology, with random, non-biological input and output maps. The stored JSON baselines (v4, 1/20) do not match the v5 baseline numbers in the docs, and the headline and rewired street results are not stored as JSON. The browser site replays logged runs; it does not run the network. Built on flyhard by Mark Unthank (MIT) with patches, but adds new tasks and controls, so not a plain copy. Not a copy of a catalogued project." }, { "id": "fly-tennis", "name": "Connectome ping pong (fly tennis)", "type": "research", "author_or_org": "castor639 (Warpfield)", "summary": "Two simulated flies rally a ball on a fly-scaled tennis court. Each runs the MaleCNS v1.0 brain and nerve cord (163,903 neurons, edges with 5 or more synapses) as an untrained flyvis-style rate network. A geometric model of FlyGym's 721-ommatidium eye feeds the medulla columns; a decoder reads the ball's bearing from small-object medulla cells (Mi1, Tm3, T2, T3) and moves the fly along its baseline. Matches are compared with a shuffled-wiring control, a parked fly and a descending-neuron (DNa01/02) decoder.", "claim": { "text": "\"If the real wiring matters, scrambling it should hurt the flies. It does.\" Real wiring 54 hits (longest rally 54) vs shuffled wiring 10 (4), parked fly 14 (6), DNa01/02 steering 12 (6) in one 40 s match each (README)", "url": "https://github.com/castor639/fly-tennis" }, "dataset": "MaleCNS", "release": "v1.0 flat connectome (minconf 0.5, traced only) from the public FlyEM bucket; 163,903 neurons, 6,235,682 edges with at least 5 synapses (89,731,552 synapses); signs from predicted transmitters", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 brain and nerve cord, edges with at least 5 synapses, signs from predicted neurotransmitter (flypp/connectome.py)", "neuron_model": "rate (graded): flyvis formulation, tau dV/dt = -V + b + sum w ReLU(V) + x, inputs scaled by each neuron's total synapse count, firing rate capped (flypp/network.py)", "input_mapping": "hand-made: analytic eye model of FlyGym's 721 ommatidia mapped onto the medulla column grid (flypp/vision.py, RetinotopicMap in flypp/loop.py); a dark ball on a plain background", "output_mapping": "hand-made: BumpDecoder population-vector azimuth of the 16 most deviating Mi1/Tm3/T2/T3 cells against a ball-free twin network (no trained parameters) sets lateral speed along the baseline, up to 16 mm/s", "trained_parts": "none in the match (--ckpt \"\" loads no checkpoint); training scripts from earlier phases are in the repository but unused by the match", "body": "custom kinematic: the fly slides along its baseline (no leg movement); FlyGym meshes for display only", "scripted_parts": "racket swing trigger, return direction (a random landing target inside the court), ball physics, court scaled about 3x smaller so the ball stays visible" }, "trained_class": "none", "trained_class_note": "untrained match; training scripts from earlier phases ship in the repository but are not used", "grade_basis": [ "flypp/tennis.py at c0b5e17: docstring states what is scripted (swing trigger, return velocity); --shuffle calls net.shuffle_edges_(0); return target drawn uniformly in the court (lines 248-262) [direct]", "flypp/network.py: shuffle_edges_ permutes postsynaptic endpoints across edges (weights and signs stay with the presynaptic side; in- and out-degree kept), one seed [direct]", "cache/v2/logs/bump2.log: 40 s, hits 54, max rally 54; open-loop bearing sweep 13, 10, 8, 3, 1, -4, -8, -9, -16 deg; closed-loop steer check corr A = -0.230, B = +0.020 (log says it should be positive) [direct]", "cache/v2/logs/shuffle.log: hits 10, max rally 4; open-loop sweep 4, 6, -16, -15, -12, -9, 2, 26, 18 deg [direct]", "cache/v2/logs/none.log: hits 14, max rally 6; cache/v2/logs/contrast.log (DNa01/02 steering): hits 12, max rally 6 [direct]" ], "grade_date": "2026-10-06", "measured_result": "Author's committed logs, one 40 s match per condition, seed 1: real wiring 54 hits (one unbroken rally), shuffled wiring 10 hits (max rally 4), parked fly 14 (6), DNa01/02 steering 12 (6). Open loop, the decoded bearing falls monotonically across 9 ball positions with the real wiring and is disordered with the shuffled wiring.", "try_url": "https://fly.warpfield.me", "try_status": "video page and a single-file replay (site/fly-tennis.html) linked from the README; not loaded by us", "code_url": "https://github.com/castor639/fly-tennis", "code_licence": "MIT", "code_licence_source": "LICENSE file at c0b5e17 (MIT, 2026 Warpfield)", "data_licence": "MaleCNS v1.0 (CC BY 4.0; downloaded by the user from the public FlyEM bucket, not included)", "platform": [ "Linux" ], "platform_note": "Python with PyTorch, FlyGym 2.1.0 and MuJoCo 3.9.0 (requirements.txt); the replay page needs only a browser; OS not stated", "gpu": "optional", "gpu_note": "CPU is enough for the match; a GPU is used if PyTorch finds one", "download_size": "566 MB of MaleCNS files (README); repository about 20 MB including the replay video", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "none", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://fly.warpfield.me", "status": "ok", "http_code": 200, "final_url": "https://fly.warpfield.me", "checked_at": "2026-10-06T09:56:57.515Z", "other_links": [ { "url": "https://github.com/castor639/fly-tennis", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:c0ecce13-d7a1-40c8-b30e-661b847df700", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/castor639/fly-tennis [direct]", "https://fly.warpfield.me [not loaded]" ], "notes_limitations": "Each condition is one 40 s match with one seed and one shuffle, so the 54 vs 10 hits gap has no spread. In the 54-hit match the logged closed-loop steering check is near zero or negative, so the long rally is not shown to come from tracking; the open-loop bearing sweep is the cleaner evidence. The swing, the return direction and the ball physics are scripted, the fly does not walk, and the court is shrunk so the ball stays visible." }, { "id": "fly-vnc-walking-sim", "name": "Fly VNC Walking Simulation", "type": "research", "author_or_org": "Sadiq Khan (sadiqkhzn)", "summary": "A work-in-progress closed loop between a conductance-style LIF simulation of 24,115 MaleCNS v1.0 nerve-cord and descending neurons (1.7M edges) and the NeuroMechFly body in MuJoCo, with a live Three.js viewer of the spikes. The README lists brain-only checks against published descending-neuron experiments (MDN, DNp09) with pass, partial and fail verdicts. In the committed closed loop, motor-neuron spikes reach the 42 leg actuators through a deterministic random hash.", "claim": { "text": "\"Closed-loop simulation of the Drosophila male ventral nerve cord driving a biomechanical fly in MuJoCo\"; \"Closed loop: fly produces 1.17 mm of movement in 400 ms of MDN drive\" (README)", "url": "https://github.com/sadiqkhzn/fly-vnc-walking-sim" }, "dataset": "MaleCNS", "release": "v1.0 via neuPrint (male-cns:v1.0): 24,115 VNC and descending neurons, about 1.77M edges with weight 3 or more; modulatory neurons left out of the LIF layer", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 ventral nerve cord plus descending neurons (24,115 neurons), split into excitatory (acetylcholine) and inhibitory (GABA, glutamate) sparse matrices", "neuron_model": "LIF with separate excitatory (5 ms) and inhibitory (150 ms) synaptic decay, tau_m 20 ms (src/sim/lif.py)", "input_mapping": "hand-made: constant drive to chosen descending neurons (MDN, DNp09)", "output_mapping": "hand-made, arbitrary: motor-neuron spikes counted over 10 ms and assigned to 42 actuator buckets by a seeded random hash (\"Not biology\", scripts/09_closed_loop.py); the muscle-based MotorDecoder returns 0.0 for every muscle (src/encoding/motor.py)", "trained_parts": "none found in the code", "body": "FlyGym (NeuroMechFly v2, 42 position-controlled leg actuators)", "scripted_parts": "the motor-to-actuator mapping; no proprioceptive feedback yet" }, "trained_class": "none", "trained_class_note": "README mentions a learned command interface; none found in the committed code", "grade_basis": [ "scripts/09_closed_loop.py at c6b50b3: build_motor_to_action_map assigns motor neurons to actuator buckets with a seeded random generator, docstring \"Deterministic hash of motor neuron index -> action bucket. Not biology.\" [direct]", "src/encoding/motor.py: MotorDecoder.decode returns a placeholder 0.0 for every muscle [direct]", "README.md: validation table (MDN raises the motor pool PASS, DNp09 suppression PARTIAL, tripod anti-phase FAIL); the JSON reports and figures are not committed [direct]" ], "grade_date": "2026-10-06", "measured_result": "Not verified by us: the validation verdicts and the 1.17 mm displacement are stated in the README; no result files are committed.", "try_url": null, "try_status": "no no-install option (Python, neuPrint token, local FastAPI viewer)", "code_url": "https://github.com/sadiqkhzn/fly-vnc-walking-sim", "code_licence": "MIT", "code_licence_source": "LICENSE file at c6b50b3 (MIT, 2026 Sadiq Khan)", "data_licence": "MaleCNS v1.0 (CC BY 4.0), fetched by the user from neuPrint; not included", "platform": [ "macOS", "Linux", "Windows" ], "platform_note": "Python 3.12-3.13 with PyTorch and FlyGym; README lists macOS, Linux or WSL on Windows; about 300 MB of cached connectome data", "gpu": "not stated", "download_size": "about 300 MB cached connectome subset (README); repository under 1 MB", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "none", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/sadiqkhzn/fly-vnc-walking-sim", "status": "ok", "http_code": 200, "final_url": "https://github.com/sadiqkhzn/fly-vnc-walking-sim", "checked_at": "2026-10-06T09:56:57.436Z" }, "last_verified": "2026-10-06", "added_in_run": "run:c0ecce13-d7a1-40c8-b30e-661b847df700", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/sadiqkhzn/fly-vnc-walking-sim [direct]" ], "notes_limitations": "The body is not walking under connectome control yet: in the committed loop, motor-neuron spikes move leg joints through a random assignment, so the reported displacement says nothing about gait. The brain-only checks against published experiments are the substantive part, but their outputs are not committed. The README says a thin command interface is learned; no training code was found." }, { "id": "fly-walking-wiring", "name": "How much of fly walking is written in the wiring? (code)", "type": "research", "author_or_org": "I. Guan, Y. Zhao, D. Zhang, S. Lyu, I.-M. Chen (kosmoKwan)", "summary": "Code for a submitted manuscript that builds fixed-weight rate models of the adult leg motor system from MaleCNS v1.0 and MANC v1.0 and compares each real network with six families of rewired networks under pre-specified criteria, plus motor-input reassignment, rescue and sign-perturbation experiments. Results and manuscript are not public yet.", "claim": { "text": "\"It compares each real network with six families of rewired networks under pre-specified criteria.\" (README)", "url": "https://github.com/kosmoKwan/fly-connectome-wiring" }, "dataset": "MaleCNS and MANC", "release": "MaleCNS v1.0; MANC v1.0", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "leg motor subnetworks of MaleCNS v1.0 and MANC v1.0 (src/build_graph.py, build_graph_manc.py)", "neuron_model": "rate model (src/openloop.py, JAX)", "input_mapping": "hand-made: motor-neuron-to-joint map (src/mn_joint_map.py); stimulation as in the scripts", "output_mapping": "hand-made: motor neurons mapped to leg joints", "trained_parts": "none (fixed weights; 132 parameter settings scanned, best eligible setting frozen per network)", "body": "none (open-loop rate model)", "scripted_parts": "none found" }, "trained_class": "other", "trained_class_note": "132 parameter settings scanned, best frozen", "grade_basis": [ "README.md and file list at c84104a (2026-09-30), read 2026-10-02 from a depth-1 blobless clone; src/nulls.py, nulls_recip.py, nulls_w.py and analysis/analyze_exp1.py exist [direct]", "No results, figures data or manuscript text are in the repository ('available from the corresponding author on request'), so the comparison with the rewired networks cannot be inspected" ], "grade_date": "2026-10-02", "measured_result": "Not public: the repository holds code only; results, derived graphs and the manuscript are to be deposited on publication.", "try_url": null, "try_status": "no no-install option (GPU workstation scripts with absolute paths)", "code_url": "https://github.com/kosmoKwan/fly-connectome-wiring", "code_licence": "MIT", "code_licence_source": "LICENSE file at c84104a", "data_licence": "MaleCNS CC BY 4.0; MANC: see neuPrint terms (not checked)", "platform": [ "Linux" ], "platform_note": "GPU (RTX 5090 in the authors' setup), WSL Ubuntu 22.04, JAX", "gpu": "required", "download_size": "code only; connectome flat files downloaded separately", "last_commit": { "date": "2026-09-30T22:00:27+08:00", "hash": "c84104aac4986a27ea371636c2c59258ceb6c1c3", "branch": "main" }, "pushed_at": "2026-09-30T14:00:36Z", "last_release": { "tag": "v1.0", "date": "2026-09-30T14:00:27Z", "url": "https://github.com/kosmoKwan/fly-connectome-wiring/releases/tag/v1.0" }, "created_at": "2026-09-30T13:20:46Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "peer_review_note": "submitted manuscript, not public", "link_status": { "url": "https://github.com/kosmoKwan/fly-connectome-wiring", "status": "ok", "http_code": 200, "final_url": "https://github.com/kosmoKwan/fly-connectome-wiring", "checked_at": "2026-10-06T09:40:09.811Z" }, "last_verified": "2026-10-06", "added_in_run": "run:03aa82ff-d83b-47ec-bb36-c8edc64e696c", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/kosmoKwan/fly-connectome-wiring [direct]" ], "notes_limitations": "Graded B because the controls exist only as code: the outcome of the real-vs-rewired comparison is not published, so nothing about 'how much is in the wiring' can be checked yet. Upgrade to A when the results or manuscript appear (the code already contains six rewiring families and coded decision rules). Candidate for the controls ledger then." }, { "id": "flyaim", "name": "FlyAim", "type": "research", "author_or_org": "0Sakura721 (\"FlyAim contributors\")", "summary": "A pre-registered test of whether the whole MaleCNS v1.0 connectome (166,700 neurons), run as a discrete-time LIF network with screen pixels fed to its real photoreceptor cells and a readout from its 1,360 descending neurons, can steer a crosshair onto targets. Same seeds and frame budgets for the real wiring, a shuffled-wiring null, a PID controller that sees the target and a uniform random walker. The real wiring is no better than the shuffle, and both are worse than random; later arms with plasticity and gradient-trained readouts are also negative. In a final step the author replaced the connectome with a compound-eye servo, which works.", "claim": { "text": "\"该 LIF 简化模型 + 静态连接组配置,不承载可用的视觉伺服信号。\" (\"This simplified LIF model with the static connectome does not carry a usable visual-servo signal.\") and the pre-registered verdict \"fly 相对零模型 shuffle 无显著优势 (p=1.0000)\" (\"no significant advantage of fly over the shuffle null\") (README)", "url": "https://github.com/0Sakura721/flyaim" }, "dataset": "MaleCNS", "release": "v1.0 (CC BY 4.0): 166,700 neurons with a superclass, 25,582,938 edges between them (124,177,616 synapses) after filtering (flyaim/runs/20261003-195410-phase3/REPORT.md)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 whole CNS, signed CSR matrices W_exc and W_inh (rows postsynaptic), never densified", "neuron_model": "other: discrete-time LIF (dt 4 ms, tau_m 20 ms, threshold 1.0, refractory clamp; flyaim/brain/lif.py, flyaim/config.py), not the Shiu model", "input_mapping": "hand-made: an encoder maps screen pixels to ommatidium illumination of the real photoreceptor cells (superclass ol_sensory); colour channels approximate", "output_mapping": "learned: a readout from the 1,360 descending neurons to 2-D crosshair velocity", "trained_parts": "the readout (frozen-random, retrained in the game domain, or gradient-trained in later arms); in the plasticity arm also connectome weights by a three-factor rule; the ANN arms train a connectome-shaped RNN by behaviour cloning and DAgger", "body": "none (a 2-D aiming arena; later a bridge that injects mouse movement into Aim Lab)", "scripted_parts": "the arena and target respawn, the PID and random baselines, the eye-servo replacement controller" }, "trained_class": "readout-or-decoder", "trained_class_note": "readout trained; later arms also train weights or a connectome-shaped RNN", "grade_basis": [ "flyaim/baselines/shuffle.py at 15d800f: shuffle null permutes CSR column indices, keeping every neuron's out-degree and the weight multiset; 'global' mode also keeps the in-degree multiset and shared E/I targeting [direct]", "flyaim/brain/lif.py header and flyaim/config.py: LIF equations, dt 4 ms x 8 steps per frame, tau_m 20 ms [direct]", "flyaim/runs/20261003-195410-phase3/REPORT.md (generated by flyaim/report.py): 10 seeds x 900 frames, mean target distance fly 412.5 px, shuffle 430.9, PID 155.8, random 251.7; verdict 'no causal contribution of the wiring' (p = 1.0) under the rule registered in CONTRACT.md section 3 [direct]", "flyaim/runs/plastic/verdict.json: three pre-registered gates failed (real plastic 432.9 px vs frozen 430.6; shuffle plastic 432.2, p 0.68; tail slope CI includes 0) [direct]", "README: paired tests fly vs random +160.8 px (p 0.0069), fly vs shuffle -18.4 px (p 0.66); gradient-trained (DAgger) arm 391 px vs random walk 278 px (p 0.012) [README; the first two traced to the REPORT above, the DAgger figure not checked against flyaim/runs/ann-dagger/verdict.json]" ], "grade_date": "2026-10-05", "measured_result": "Mean distance from crosshair to target over 10 seeds x 900 frames (lower is better): real MaleCNS wiring 412.5 px, shuffled wiring 430.9 px (not different, p = 0.66), uniform random walker 251.7 px, PID with target access 155.8 px. Hit rate 0 for both network arms. A plasticity arm and gradient-trained readouts were also negative (pre-registered gates failed).", "try_url": null, "try_status": "no no-install option (Python; data and pre-trained artefacts in release v0.1.0)", "code_url": "https://github.com/0Sakura721/flyaim", "code_licence": "MIT", "code_licence_source": "LICENSE file at 15d800f (MIT, 2026 FlyAim contributors); the API reports NOASSERTION", "data_licence": "MaleCNS v1.0 (CC BY 4.0, as stated in the generated report)", "platform": [ "Windows" ], "platform_note": "Python; the Aim Lab bridge uses Windows input APIs (flyaim/bridge, win32/SendInput); the arena simulation itself is plain Python; an optional CuPy/CUDA engine (flyaim/brain/lif_gpu.py)", "gpu": "optional", "download_size": "repository about 3.2 MB at HEAD; connectome data and artefacts from release v0.1.0 (size not measured)", "last_commit": "n/a", "pushed_at": "n/a", "last_release": { "tag": "v0.1.0 — 数据与预训练工件", "date": "2026-10-03T19:24:14Z", "url": "https://github.com/0Sakura721/flyaim/releases/tag/v0.1.0" }, "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/0Sakura721/flyaim", "status": "ok", "http_code": 200, "final_url": "https://github.com/0Sakura721/flyaim", "checked_at": "2026-10-06T09:40:09.913Z" }, "last_verified": "2026-10-06", "added_in_run": "run:3fed5f73-77cc-4b37-853f-e1e8258b4f9f", "controls": "wiring-null", "wiring_effect": "no-difference", "sources": [ "https://github.com/0Sakura721/flyaim [direct]" ], "notes_limitations": "A clean pre-registered negative result for one simplified model: discrete-time LIF with a 4 ms step, not the Shiu model, and an image encoder onto photoreceptors that the author calls a design choice. The README's hit rate is a flat metric by the author's own analysis (ceiling about 6%); mean target distance is the working metric. The later eye-servo success bypasses the connectome entirely and must not be read as a fly result. Chinese-language documentation; one author group; created 2026-10-03 and still changing." }, { "id": "flyarm", "name": "FlyArm", "type": "research", "author_or_org": "yusenthebot", "summary": "A frozen MaleCNS v1.0 connectome (166,700 neurons, 10.5 million connections, run as a rate network on Apple Silicon with MLX) is the recurrent core of a controller for a simulated Franka arm in MuJoCo. Only an encoder into the 1,846 ascending neurons and a linear decoder from the descending and motor neurons are trained, by imitation and then PPO. It reports multi-step manipulation and FrankaKitchen results, lesions against size-matched random sets, and early degree-preserving shuffle comparisons that the author says do not yet establish a wiring advantage.", "claim": { "text": "\"A frozen, complete fruit-fly central nervous system (MaleCNS v1.0: 166,700 neurons, 10.5 M measured connections) is the controller of a simulated Franka arm.\" and \"Whether the measured wiring trains better than another graph is not yet claimed.\" (README)", "url": "https://github.com/yusenthebot/FlyArm" }, "dataset": "MaleCNS", "release": "v1.0 (CC BY 4.0): 166,700 annotated non-glia neurons, 10,520,377 edges with at least 3 contacts, compiled to a CSR pack (recipe whole-malecns-contact-v1)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 whole CNS; weights = synapse counts x transmitter sign (ACh +1, GABA and Glu -1, others 0), row-normalised", "neuron_model": "other: rate model h <- (1 - alpha) h + alpha tanh(I + g W h), alpha 0.5, g 0.8, 3 updates per control step (src/flyarm/whole_brain/backend_mlx.py)", "input_mapping": "learned: a trained encoder writes currents into the 1,846 ascending neurons from simulator state and a task cue", "output_mapping": "learned: a linear decoder from 1,314 descending and 708 VNC motor neurons to the arm action", "trained_parts": "encoder and decoder (behaviour cloning with DAgger, then PPO with a demonstration term); the connectome weights are frozen", "body": "simulated Franka Panda arm (MuJoCo; FrankaKitchen in Gymnasium-Robotics), not a fly body", "scripted_parts": "the scripted teacher used for imitation, task plans and the task cue" }, "trained_class": "readout-or-decoder", "trained_class_note": "encoder and decoder trained (behaviour cloning, PPO); connectome frozen", "grade_basis": [ "src/flyarm/whole_brain/backend_mlx.py at ab3922e: rate update with tanh, frozen connectome, alpha/gain constants [direct]", "src/flyarm/whole_brain/compiler.py and shuffle.py: MaleCNS v1.0 CSR pack; degree-preserving null that keeps each neuron's in- and out-degree, contact multiset, sign and the input/output sets [direct]", "docs/RESEARCH_LOG.md E1-E3: whole-brain check (no input-to-output bypass: with every edge removed outputs are exactly zero); pick-and-place lift 58/72 connectome vs 27/72 shuffle vs 33/72 GRU over 3 seeds (seed-level p = 0.125) [author's log in the repository; run files not read]", "README Results and Lesions tables (manipulation success 76.3% iid with PPO; state cleared every step 0/112; all-but-interface 16/112) and STATUS.md (shuffle seeds for the manipulation protocol stopped 2026-10-03, not finished) [README; numbers not checked against docs/results/*.json]" ], "grade_date": "2026-10-05", "measured_result": "Author's figures: full-task success on held-out articulated manipulation episodes 76.3% (iid), 82.6% (unseen objects), 36.6% (unseen furniture) and 26.6% (unseen compositions) after PPO; FrankaKitchen all four tasks 50/50. Lesions: clearing the recurrent state each step 0/112, silencing everything but the interface 16/112 (intact 87/112). Wiring vs degree-preserving shuffle: pick-and-place lift 58/72 vs 27/72 over 3 seeds, a kitchen ordering 25 vs 0 and 21.25 vs 15, a dexterous-hand tie (62 vs 63 of 64); the author states the wiring advantage is not established.", "try_url": "https://yusenthebot.github.io/FlyArm/", "try_status": "interactive site linked from the README (recorded activity and a lesion lab); not loaded by us", "code_url": "https://github.com/yusenthebot/FlyArm", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file at ab3922e (Apache License 2.0); THIRD_PARTY.md lists other licences", "data_licence": "MaleCNS v1.0 (CC BY 4.0, as stated in the README)", "platform": [ "macOS" ], "platform_note": "Apple Silicon with MLX (custom Metal sparse kernel); about 1.5 GiB", "gpu": "required", "gpu_note": "Apple Silicon GPU through MLX/Metal", "download_size": "repository about 6.3 MB of files at HEAD plus large videos and run files (20 files over 512 KB); connectome pack built locally", "last_commit": "n/a", "pushed_at": "n/a", "last_release": { "tag": "milestone-kitchen", "date": "2026-09-23T22:13:00Z", "url": "https://github.com/yusenthebot/FlyArm/releases/tag/milestone-kitchen" }, "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://yusenthebot.github.io/FlyArm/", "status": "ok", "http_code": 200, "final_url": "https://yusenthebot.github.io/FlyArm/", "checked_at": "2026-10-06T09:40:09.272Z", "other_links": [ { "url": "https://github.com/yusenthebot/FlyArm", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:3fed5f73-77cc-4b37-853f-e1e8258b4f9f", "controls": "wiring-null", "wiring_effect": "no-difference", "sources": [ "https://github.com/yusenthebot/FlyArm [direct]" ], "notes_limitations": "The connectome is frozen but sits between a trained encoder and a trained decoder, so much of the skill can live in the interface; the author's own lesions (all-but-interface 16/112) show the recurrent connectome state matters. The wiring-vs-shuffle comparison is partial and mixed across tracks (clear in pick-and-place, absent or tied elsewhere), and the shuffle runs for the headline manipulation protocol were stopped unfinished. The robot is an arm, not a fly body; inputs are simulator state, not vision. One author; not peer reviewed." }, { "id": "flybench", "name": "flybench", "type": "research", "author_or_org": "Brandon Cho", "summary": "A benchmark of 36 cited fly behaviours (reflexes, sparseness, escape, and more) that scores whole-brain spiking simulations of FlyWire v783 and MaleCNS v1.0. Every task can also be run on degree-preserving, random and sign-shuffled wiring, and the score gap is reported as 'specificity'.", "claim": { "text": "Does the simulated fly still do the things a real fly is known to do?", "url": "https://www.fly-bench.com/bench" }, "dataset": "several", "release": "FlyWire FAFB v783 and MaleCNS v1.0; flybench release 0.2.1 (2026-09-24)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 (139,255 neurons, >=5 synapses) and MaleCNS v1.0 (neuPrint male-cns:v1.0)", "neuron_model": "LIF (Shiu 2024 reference; also adaptive LIF and terminal-aware LIF variants)", "input_mapping": "hand-made (named cell types driven at set rates; optional flyvis front end for vision)", "output_mapping": "hand-made (named readout neurons, e.g. MN9, giant fiber, TTMn)", "trained_parts": "none in the brain; flyvis front end is a pretrained published model", "body": "FlyGym (optional embodied tasks 33-36)", "scripted_parts": "embodied tasks: TTMn/GF spikes trigger a fixed middle-leg extension (jump) program" }, "trained_class": "front-end", "trained_class_note": "pretrained flyvis front end", "grade_basis": [ "flybench/controls.py:59-78 rewire_degree_preserving (permutes postsynaptic endpoints, keeps out-degree, weights, signs); :81-91 Erdos-Renyi matched; :94-105 sign shuffle [direct]", "flybench/bench.py:881-901 runs each task on control connectomes, specificity = score(real) - max(score(control)), flags non_diagnostic [direct]", "flybench/models/reference_lif.py:27-29,42 LIF parameters with provenance [direct]", "results/flywire783_gain0.45.json specificity 0.1998, graded 0.644; results/shiu2024.json specificity 0.180; results/malecns-gain-0.65.json specificity 0.219 [direct]", "pyproject.toml:7 version 0.2.1; CHANGELOG.md 0.2.1 dated 2026-09-24 [direct]" ], "grade_date": "2026-09-28", "measured_result": "From results/*.json (LEADERBOARD.md): reference LIF on FlyWire v783, gain 0.45, 3 seeds: core 1.00, hard 0.51, graded 0.64 [0.64, 0.65], shuffled-wiring specificity +0.20. Shiu 2024 settings (gain 1.0): graded 0.63, specificity +0.18, ~34% of brain active. Adaptive LIF: graded 0.75, specificity +0.31. MaleCNS gain 0.65 (Minecraft demo settings): core 0.57, graded 0.66, specificity +0.22. Terminal-aware LIF: 0.70 on both brains. Through a flyvis eye the loom does not reach the giant fiber (0/3) while a flash does. Measured by the author, not reproduced by us.", "try_url": "https://www.fly-bench.com/bench", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/brandoncho369/flybench", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 3052ce5fa9; GitHub API spdx_id=MIT", "data_licence": "several (see dataset entries)", "platform": [ "browser", "Linux", "macOS", "Windows", "Colab" ], "gpu": "none", "download_size": "not stated (FlyWire files need a free Codex login; ~2 GB RAM per run); toy runs with no download", "last_commit": { "date": "2026-09-23T23:57:27-05:00", "hash": "3052ce5fa9ab8cc0a9e101e8cd8f5bb520a80ff5", "branch": "master" }, "pushed_at": "2026-09-24T05:02:35Z", "last_release": { "tag": "flybench 0.2.1 — a pip install ships the tasks", "date": "2026-09-24T04:57:33Z", "url": "https://github.com/brandoncho369/flybench/releases/tag/v0.2.1" }, "created_at": "2026-09-10T19:07:54Z", "stars": 2, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://doi.org/10.5281/zenodo.22886374", "peer_review": "none", "link_status": { "url": "https://www.fly-bench.com/bench", "status": "ok", "http_code": 200, "final_url": "https://www.fly-bench.com/bench", "checked_at": "2026-10-06T09:40:09.547Z", "other_links": [ { "url": "https://doi.org/10.5281/zenodo.22886374", "status": "redirect", "http_code": 200 }, { "url": "https://github.com/brandoncho369/flybench", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "mixed", "sources": [ "https://github.com/brandoncho369/flybench [direct]", "https://www.fly-bench.com/bench [direct: page title checked, HTTP 200]" ], "notes_limitations": "A benchmark, not a single demo; results are a leaderboard of parameter settings. Most rows are 'self-reported' by the maintainer, who also wrote the reference and adaptive models (declared in the table). The 'rewired' control permutes edge targets (configuration-model style), so in-degree is kept only as an endpoint multiset and duplicates are merged; it is not a strict double-edge swap. Specificity is only computed for rows where controls were run. Tasks measure listed cited manipulations, not overall realism. Zenodo DOI is a software archive, not a paper. Measured by the author, not reproduced by us." }, { "id": "flybrain-connectome-benchmark", "name": "flybrain-connectome-benchmark", "type": "research", "author_or_org": "Vichien Fugsukjit (independent researcher)", "summary": "A pre-registered benchmark of the Shiu et al. whole-brain model (re-implemented in NumPy/SciPy) against published fly experiments: feeding, grooming, escape, smell and navigation criteria set before running, plus a whole-brain single-neuron knockout screen of the feeding circuit. Raw simulation outputs and scores are committed; a MaleCNS v1.0 replication is included.", "claim": { "text": "Benchmarking a whole-brain connectome model of Drosophila against experimental data: diagnosing knockout-prediction failures and a candidate excitatory role for the water-taste neuron Usnea", "url": "https://github.com/vichienFF/flybrain-connectome-benchmark" }, "dataset": "FlyWire FAFB", "release": "v783 via Shiu et al. Connectivity_783.parquet and Completeness_783.csv; annotations v3.1.0; replication on MaleCNS v1.0", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "Whole FlyWire v783 brain as used by Shiu et al.", "neuron_model": "spiking: Shiu et al. leaky integrate-and-fire, NumPy/SciPy port validated against the Brian2 original", "input_mapping": "hand-set: literature-defined sensory neuron sets (sugar, water, bitter GRNs, JO, LC4, LPLC2, ORNs) at 150 Hz Poisson", "output_mapping": "read-out of named neurons (MN9, aBN1, giant fiber, PNs) scored against pre-set thresholds", "trained_parts": "none", "body": "none", "scripted_parts": "none" }, "trained_class": "none", "grade_basis": [ "README.md:3-12 preprint, contents, model/flybrain.py re-implementation of Shiu LIF validated against Brian2 [direct]", "benchmark/BENCHMARK_PLAN.md:1-30 pre-registered criteria written 2026-09-25 before simulation (F1-F9 feeding, G1-G2 grooming, E1-E2 escape, O1 smell, N1 navigation) with literature sources [direct]", "benchmark/results/bm2_scores.json key D0.A_primary: knockout prediction sensitivity 0.769, specificity 0.989, balanced accuracy 0.879; D0.B_pass 11/17; D0.C_pass 3/3 [direct]", "benchmark/results/: raw_bm2..4 and results_raw*.jsonl committed simulation outputs [direct]" ], "grade_date": "2026-09-30", "measured_result": "Pre-registered benchmark v2 (benchmark/results/bm2_scores.json, variant D0): the model predicts published single-neuron knockout effects with sensitivity 0.769 and specificity 0.989 (balanced accuracy 0.879); 11 of 17 activation-change criteria and 3 of 3 grooming-rate criteria pass. The author proposes that the water-taste neuron Usnea is excitatory. No shuffled-wiring control. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/vichienFF/flybrain-connectome-benchmark", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit c4ef087e30; GitHub API spdx_id=MIT", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux", "macOS", "Windows" ], "platform_note": "Python scripts", "gpu": "none", "gpu_note": "CPU; about 50 s and 2.5 GB RAM per simulated second (README)", "download_size": "Shiu v783 files (about 100 MB) and optional MaleCNS v1.0 fetched separately", "last_commit": { "date": "2026-10-06T09:29:53+07:00", "hash": "c4ef087e304e0e6898b12994db37a772e536e22e", "branch": "main" }, "pushed_at": "2026-09-29T03:46:50Z", "last_release": "none", "created_at": "2026-09-28T03:21:00Z", "stars": 0, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": "https://doi.org/10.5281/zenodo.23025560", "peer_review": "preprint", "peer_review_note": "Zenodo preprint", "link_status": { "url": "https://github.com/vichienFF/flybrain-connectome-benchmark", "status": "ok", "http_code": 200, "final_url": "https://github.com/vichienFF/flybrain-connectome-benchmark", "checked_at": "2026-10-06T09:40:10.296Z", "other_links": [ { "url": "https://doi.org/10.5281/zenodo.23025560", "status": "redirect", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/vichienFF/flybrain-connectome-benchmark [direct: clone HEAD 565253e, 2026-09-30]" ], "notes_limitations": "A careful test of the model against fly data rather than a demo: criteria were fixed before the runs, and failures are reported. It tests the Shiu model with its original parameters, so it says how far the model can be trusted, not whether wiring beats a scrambled copy (there is no wiring null). Some scripts still use absolute Windows paths. Plan files are partly in Thai." }, { "id": "flybrain-reservoir", "name": "flybrain-reservoir", "type": "research", "author_or_org": "Milan Kalajdzic", "summary": "Uses the MaleCNS v1.0 connectome (and FlyWire 783 as a replication) as the recurrent matrix of a leaky-tanh echo state network fed with SPY market data, and compares it with four random rewirings on market prediction, memory capacity, NARMA-10 and volatility forecasting. The real wiring is no better, and on memory it is much worse.", "claim": { "text": "Is a real fly brain a better reservoir computer than random wiring? ... Short answer: no, and it replicates on FlyWire.", "url": "https://github.com/MilanKalajdzic/flybrain-reservoir" }, "dataset": "several", "release": "MaleCNS v1.0 (main); FlyWire FAFB 783 (replication)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 whole CNS (166,700 neurons) and a 3,000-neuron sensory-grown subgraph; FlyWire 783 whole brain (~139k neurons)", "neuron_model": "rate", "input_mapping": "hand-made", "output_mapping": "learned", "trained_parts": "linear ridge readout only (standard for reservoir computing); recurrent weights fixed, rescaled to spectral radius 0.9", "body": "none", "scripted_parts": "none found" }, "trained_class": "readout-or-decoder", "trained_class_note": "linear ridge readout", "grade_basis": [ "src/flyres/connectome.py:38-42 MaleCNS v1.0 flat-connectome files; :11-15,49 FlyWire release 783 [direct]", "src/flyres/reservoir.py:3 leaky-tanh ESN x[t]=(1-a)x[t-1]+a*tanh(W x + W_in u + b) [direct]", "src/flyres/controls.py:26,64,113,128 degree-preserving rewire, Erdos-Renyi, weight shuffle (plus sign shuffle) [direct]", "configs/full.yaml:7-12 and configs/flywire_full.yaml:7-12 same five wirings, 10 seeds [direct]", "README.md:79-85 market: IC ~0.015, Sharpe 0.42-0.51 vs 0.59 buy-and-hold, all bootstrap CIs for differences include zero [direct]", "README.md:93-98 memory capacity with readout noise, whole CNS: connectome 2.2 vs degree-preserving 15.5, ER 7.4 [direct]", "README.md:142-145 best valid gain, whole CNS: 2.9 vs 22.8 (degree-preserving) and 30.8 (ER) [direct]", "README.md:350-355 FlyWire 783 whole brain: 3.0 vs 21.8 and 72.4 [direct]", "README.md:398-407 volatility 5-day log MSE vs HAR: connectome -16.5%, controls -16.5% to -17.1%, linear HAR+inputs -14.9% [direct]" ], "grade_date": "2026-09-28", "measured_result": "Negative result against random-wiring controls. Memory capacity (readout noise), whole MaleCNS CNS at standard gain: connectome 2.2, degree-preserving 15.5, weight shuffle 2.4, sign shuffle 2.9, Erdos-Renyi 7.4 (README.md:96); at each wiring's best gain 2.9 vs 22.8 / 2.9 / 3.1 / 30.8 (README.md:145). 3,000-neuron circuit best gain: 6.7 vs 6.0-7.4, a tie (README.md:144). FlyWire 783 whole brain: 3.0 vs 21.8 and 72.4 (README.md:354). NARMA-10 NRMSE whole CNS: 0.77 vs 0.42 (degree-preserving) and 0.48 (ER) (README.md:276). Markets: Sharpe differences at most 0.06, all CIs include zero (README.md:79-85). Volatility 5-day: all wirings about -16.5% to -17.1% log MSE vs HAR, linear HAR+inputs -14.9% (README.md:400-405).", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/MilanKalajdzic/flybrain-reservoir", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 6da6325715; GitHub API spdx_id=NOASSERTION", "data_licence": "several (see dataset entries)", "platform": [ "Linux" ], "gpu": "optional", "download_size": "~1.2 GB MaleCNS v1.0; ~130 MB FlyWire 783 (scripts/download_data.py:3-5)", "last_commit": { "date": "2026-09-26T20:09:59+02:00", "hash": "6da63257154a334006579c692a31f671e572f83e", "branch": "main" }, "pushed_at": "2026-09-26T18:10:02Z", "last_release": { "tag": "Corrections from a final review. The main results don't change: across the whole brain the fly's wiring remembers 8–11× less than random wiring, in both connectomes, and no wiring gives a market edge.", "date": "2026-09-26T18:10:53Z", "url": "https://github.com/MilanKalajdzic/flybrain-reservoir/releases/tag/v1.0.1" }, "created_at": "2026-09-21T04:58:00Z", "stars": 2, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/MilanKalajdzic/flybrain-reservoir", "status": "ok", "http_code": 200, "final_url": "https://github.com/MilanKalajdzic/flybrain-reservoir", "checked_at": "2026-10-06T09:40:10.659Z" }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "worse", "sources": [ "https://github.com/MilanKalajdzic/flybrain-reservoir [direct]" ], "notes_limitations": "A well-controlled negative result: five wirings, several seeds, several benchmarks, two connectomes. The numbers are taken from the README tables; the result folders are not in the repo, but the configs pin data dates and seeds so they can be rerun. It is an abstract rate reservoir (tanh units, one global gain, spectral radius 0.9), not a biophysical fly model, so it tests the wiring graph, not fly behaviour. The whole-brain result depends on dense hub clusters that set the spectral radius; the author discusses this. FlyWire connectivity comes from Shiu et al.'s Connectivity_783 table. Small count mismatch: config says 139,248 FlyWire neurons, connectome.py says 139,255 (probably after filtering). Whole-CNS runs need ~32 GB RAM. Licence: LICENSE file is MIT (API says NOASSERTION)." }, { "id": "flyconnectome-nulls", "name": "Null-model treatment of the sensory-motor boundary changes an evolutionary connectome comparison", "type": "research", "author_or_org": "G. Park (gyujeongion)", "summary": "Foraging agents in a 2D world have a brain built from FlyWire v783, compressed to 512 cell-type groups plus 1,000 sampled Kenyon cells, and their weights are evolved. They are compared against populations evolved on randomised wiring. Standard nulls (column shuffle, degree-preserving swaps) create direct smell-to-descending-neuron shortcuts and out-evolve the connectome; nulls that keep the sensory and motor boundary fixed come out about even.", "claim": { "text": "What the null model preserves at the sensory-motor boundary changes the outcome of an evolutionary comparison between a FlyWire v783 connectome brain and randomised wiring.", "url": "https://github.com/gyujeongion/flyconnectome-nulls" }, "dataset": "FlyWire (FAFB)", "release": "v783 (Connectivity_783.parquet + Schlegel et al. annotations; compressed data/brain.npz included)", "evidence_grade": "A", "grade_note": "Borderline A: the brain is heavily compressed (512 cell-type groups plus 1,000 sampled Kenyon cells), and all weights evolve away from the connectome.", "mechanism": { "wiring": "subset: FlyWire v783 without optic lobes, aggregated to 512 cell-type groups (chosen by sensory-to-descending path score plus mandatory olfactory, gustatory, mushroom-body and named descending groups) plus 1,000 sampled individual Kenyon cells; weights = signed synapse fraction of each group's input", "neuron_model": "rate: leaky tanh(relu) group units, KC layer with threshold, dopamine-gated plastic KC->MBON synapses", "input_mapping": "hand-made (odour, sugar, bitter and looming stimuli drive named ORN, GRN and LC4/LPLC2 groups with evolved gains)", "output_mapping": "hand-made prior plus evolved readout (forward from DNp09/MDN/DNp01, turn from DNa01/DNa02 left-right difference, eat from an innate prior; readout weights mutate)", "trained_parts": "evolved: all group weights (scaled, added, deleted by mutation), time constants, biases, input gains, readout weights, plasticity rates", "body": "simple 2D point agent (speed, turn, eat)", "scripted_parts": "scripted reference policies (random walk, reflex, chemotaxis, oracle memory) exist only as baselines; the tested agents are brain-driven" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "all group weights evolved", "grade_basis": [ "build_brain.py:1-120 reads FlyWire v783 Connectivity_783 and annotations, drops optic lobes, groups by cell type, keeps 512 groups and 1,000 KCs, zeroes recurrent input to sensory rows [direct]", "evo_fix.py:102-114 recurrent rate update with KC layer and dopamine-gated KC->MBON plasticity; 70-79 descending-neuron readout for forward/turn/eat [direct]", "evo_fix.py:425-443 mutation rescales, adds and deletes weights: the connectome is a starting point, not frozen [direct]", "evo_fix.py:242-342 null conditions: N1 column shuffle, N2 degree-preserving swaps, N3 block shuffle, N4/N5 interior-only randomisation that keeps sensory and descending edges [direct]", "results/fix_analysis.txt and results/equivalence.txt: 10 seeds per ecology, connectome minus control at generation 600 with CIs and Holm-corrected p [direct]", "results/shortcut_check.txt: share of olfactory output landing directly on descending neurons, connectome 0.0001 vs N1 0.107 / N2 0.106 [direct]", "protocols/*.md frozen pre-registration protocols with hashes; RESEARCH_LOG.md dated log including a retraction [direct]" ], "grade_date": "2026-09-28", "measured_result": "Corrected grid, generation 600, pooled over four ecologies, 10 seeds (results/fix_analysis.txt): fitness connectome 1.57, column shuffle 1.78, degree-preserving swap 1.77, interior-only swap 1.58, interior-only shuffle 1.62; connectome minus standard nulls -0.224 (p_holm 0.029) and -0.199 (p_holm 0.016), minus boundary-preserving nulls +0.002 and -0.074 (n.s.). Olfactory output landing directly on descending groups: connectome 0.01 %, standard nulls about 10.7 %. README also reports shortcut transplant +0.44 fitness (10/10 seeds) and a dose-response. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/gyujeongion/flyconnectome-nulls", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 2f5683daa7; GitHub API spdx_id=MIT", "data_licence": "unknown", "platform": [ "macOS" ], "platform_note": "macOS (Apple Silicon, MLX)", "gpu": "required", "gpu_note": "required: Apple Silicon GPU via MLX (about 55 GPU-hours on an M1 Ultra for the full set)", "download_size": "small (repo includes data/brain.npz, 154 KB); rebuilding needs the FlyWire v783 connectivity parquet and annotations", "last_commit": { "date": "2026-09-30T01:04:08+09:00", "hash": "2f5683daa7a9619c228c05949d3208f3d32d3926", "branch": "main" }, "pushed_at": "2026-09-29T16:04:12Z", "last_release": "none", "created_at": "2026-09-23T07:41:34Z", "stars": 0, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": "https://doi.org/10.5281/zenodo.22871090", "peer_review": "preprint", "link_status": { "url": "https://github.com/gyujeongion/flyconnectome-nulls", "status": "ok", "http_code": 200, "final_url": "https://github.com/gyujeongion/flyconnectome-nulls", "checked_at": "2026-10-06T09:40:10.923Z", "other_links": [ { "url": "https://doi.org/10.5281/zenodo.22871090", "status": "redirect", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "wiring-null", "wiring_effect": "mixed", "sources": [ "https://github.com/gyujeongion/flyconnectome-nulls [direct: clone HEAD c653377]", "https://doi.org/10.5281/zenodo.22871090 [not checked]" ], "notes_limitations": "The question is methodological (which random-wiring control is fair), not whether the fly brain produces fly-like behaviour; no comparison with real fly behaviour. The brain is a heavy compression (512 cell-type groups selected by a path score, plus 1,000 sampled KCs) and all weights evolve away from the connectome, so wiring is an initial condition. Input and output neuron choices and the readout prior are hand-made. Seeds are not bit-reproducible (clock reseeding, GPU scatter). One 20-seed extension did not replicate an earlier lead. Raw run logs (about 40 GB) are not in the repo; analysis outputs are. The paper states code and manuscript were written with an AI coding assistant. Measured by the author, not reproduced by us." }, { "id": "flydoom", "name": "FlyDoom", "type": "research", "author_or_org": "eganeganegan (FlyDoom contributors)", "summary": "A framework that uses a MaleCNS v1.0 subgraph (about 1,000-10,000 neurons) as the edge set of a sparse recurrent policy trained with PPO (or a three-factor rule) on VizDoom, with Erdos-Renyi, degree-preserving, MLP, GRU and LSTM baselines. No results are published in the repo.", "claim": { "text": "does the topology of the real Drosophila male CNS connectome provide a useful inductive bias for reinforcement learning?", "url": "https://github.com/eganeganegan/flydoom" }, "dataset": "MaleCNS", "release": "v1.0 (male-cns:v1.0 via neuPrint or official feather files)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "subset: up to ~5,000 neurons of MaleCNS v1.0 (sensory/LC4 to descending k-hop subgraph)", "neuron_model": "rate: leaky tanh message passing, 5 steps per observation", "input_mapping": "learned (CNN encoder by default; optional hand-made 'fly-inspired' features) into designated sensory nodes", "output_mapping": "learned (policy/value heads on descending nodes)", "trained_parts": "default connectome_rate config trains internal edge weights plus encoder and heads with PPO; fixed_internal/readout_only/three_factor modes exist", "body": "none", "scripted_parts": "none found" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "default config trains internal edge weights with PPO", "grade_basis": [ "src/flydoom/data/controls.py:11,37,44-60 Erdos-Renyi, weight shuffle and degree-preserving double-edge swap controls implemented [direct]", "src/flydoom/models/connectome_network.py:60-80 leaky tanh sparse message passing on subgraph edges [direct]", "configs/model/connectome_rate.yaml: training_mode trainable_internal, encoder cnn [direct]", "configs/experiment/flydoom_basic.yaml: variants real_connectome, erdos_renyi, degree_rewired, mlp, gru, lstm; env backend mock [direct]", "repo tree: no results, metrics or learning-curve files committed [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/eganeganegan/flydoom", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit b047fabffb); GitHub API spdx_id=None", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS", "Windows" ], "gpu": "optional", "download_size": "not stated (MaleCNS feather files, several GB; neuPrint token for query mode)", "last_commit": { "date": "2026-09-14T17:05:52-04:00", "hash": "b047fabffb62e10268de2fc1f12c765b8b019c0b", "branch": "main" }, "pushed_at": "2026-09-14T21:05:57Z", "last_release": "none", "created_at": "2026-09-12T03:39:07Z", "stars": 8, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/eganeganegan/flydoom", "status": "gone", "http_code": 404, "final_url": "https://github.com/eganeganegan/flydoom", "checked_at": "2026-10-03T09:46:41.971Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/eganeganegan/flydoom [direct]" ], "notes_limitations": "Controls are well designed and implemented in code, but no measured results (connectome vs rewired vs ordinary networks) are reported anywhere in the repo, so there is nothing to grade A on. With default settings the connectome weights are trained by PPO on a small subgraph behind a CNN, so behaviour comes largely from learning, not from the wiring as measured. README says itself it is not Doom in a biological brain and states no result. No licence file (GitHub reports none). Contains a stale build/lib copy of the package. Could move to A if the author publishes the control comparison. Repository eganeganegan/flydoom returns HTTP 404 on 2026-10-01 and 2026-10-02 (git ls-remote asks for credentials: gone or private); the record is kept with the last catalogued commit b047fab (2026-09-14). A third HTTP 404 on 2026-10-03 (git ls-remote again asks for credentials): link status set to gone. The record is kept, with no working code link." }, { "id": "flydoom-mutkuoz", "name": "flydoom", "type": "research", "author_or_org": "mutkuoz", "summary": "The full FlyWire FAFB v783 connectome (139,255 neurons, about 2.7 M connections) is simulated with spiking and graded cells, sees ViZDoom through a 1,581-ommatidium eye model, and plays via 8 descending-neuron populations mapped to game buttons. No weights are trained; only one global synaptic gain is calibrated on a non-visual reflex. Results are tested against shuffled-wiring brains, command-matched random agents and mirrored or frozen eyes, with raw run data and a paper in the repo.", "claim": { "text": "A fruit fly's brain plays Doom: the FAFB v783 wiring diagram is simulated, shown Doom through the fly's eyes, and keypresses are read from descending nerves", "url": "https://github.com/mutkuoz/flydoom" }, "dataset": "FlyWire FAFB", "release": "v783 (139,255 neurons)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FAFB v783 whole brain (connections_princeton_no_threshold with own threshold, NT-signed)", "neuron_model": "LIF plus graded (non-spiking) cells, with compartment options", "input_mapping": "hand-made: 1,581 ommatidial columns to 4,541 lamina L1/L2/L3 inputs; olfaction and mechanosensation channels", "output_mapping": "hand-made: DNa02 L-R to turn, BPN forward / MDN backward, DNp01 lateral, Schmitt-triggered buttons", "trained_parts": "none; one global synaptic gain calibrated on a non-visual reflex", "body": "none (Doom player), with a custom fly-scale arena", "scripted_parts": "motion-detection front end (EMD-like) and game arena design; no aiming or game-state shortcuts found in the files inspected" }, "trained_class": "none", "trained_class_note": "one global gain calibrated", "grade_basis": [ "flydoom/sources.py:1-30 documents the FAFB v783 file inventory used (classification, cell types, connections_princeton_no_threshold) [direct]", "flydoom/motor.py:1-25 DN population rates map to ViZDoom DELTA and Schmitt-triggered buttons [direct]", "experiments/m9_behaviour.py:557, 718-726 comparison with an AR(1)-matched random agent [direct]", "experiments/m10_klinotaxis.py, m13_odour_square.py, m18_stripe.py and others use shuffled-wiring controls [direct]", "paper/data/behav_*/fly_{intact,mirrored,frozen,nosmell}_shards: raw results for the control arms [direct]" ], "grade_date": "2026-09-28", "measured_result": "Measured by the author, not reproduced by us. In a visible arena it beats a command-matched random agent on health and survival, and mirroring the eyes removes that advantage. Smell beats a shuffled brain by about 1000x. Sugar drives proboscis extension (77 Hz) and bitter suppresses it by 99%. Motion detection reaches about 2% of a real fly's, and there is no optomotor counter-turn, looming escape or fixation (README scorecard, paper/data).", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/mutkuoz/flydoom", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit 86d99a4f10); GitHub API spdx_id=None", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux" ], "gpu": "required", "gpu_note": "CUDA GPU with 8 GB or more required", "download_size": "FAFB v783 tables (size not stated) plus repository with media", "last_commit": { "date": "2026-09-29T22:16:38+03:00", "hash": "86d99a4f10193203759c3b4266535190e3b09672", "branch": "main" }, "pushed_at": "2026-09-27T16:34:02Z", "last_release": "none", "created_at": "2026-08-20T16:42:25Z", "stars": 8, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "peer_review_note": "none (26-page PDF at paper/main.pdf in the repository, not peer reviewed)", "link_status": { "url": "https://github.com/mutkuoz/flydoom", "status": "ok", "http_code": 200, "final_url": "https://github.com/mutkuoz/flydoom", "checked_at": "2026-10-06T09:40:10.615Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/mutkuoz/flydoom [direct: clone HEAD 5e53bbf]", "https://github.com/eganeganegan/flydoom [direct: clone HEAD b047fab, compared]" ], "notes_limitations": "This is a distinct project from eganeganegan/flydoom: different author, tree and files, with 2 identical blobs out of thousands, and eganeganegan's project uses MaleCNS. The grade rests on the controls present in the code and results files, not on game skill. The author reports that vision helps only narrowly and in one purpose-built arena, and that classic fly reflexes are absent. It runs at about half real time. Only a short inspection was done; the paper was not read." }, { "id": "flygym-banc-bridge", "name": "FlyGym + BANC v888 bridge (arisliwind/flygym)", "type": "research", "author_or_org": "arisliwind (GitHub fork of NeLy-EPFL/flygym 2.1.0)", "summary": "A fork of FlyGym 2.1.0 that adds scripts linking the BANC v888 connectome to FlyGym walking: one descending neuron's LIF firing rate sets a constant CPG amplitude, and a 50-neuron closed-loop subnet with STDP sets per-leg gains of a sinusoidal leg controller.", "claim": { "text": "\"a direct bridge from the BANC v888 real neural connectome ... into FlyGym's motor control loop, demonstrating a closed-loop pipeline: real neuron spikes -> CPG modulation -> leg actuators -> walking simulation\" (README)", "url": "https://github.com/arisliwind/flygym" }, "dataset": "BANC (brain and nerve cord connectome, Harvard Dataverse)", "release": "v888 (banc_888_meta.feather, banc_888_edgelist_simple_v2.feather)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "BANC v888 subsets: (bridge) one descending neuron with the most outputs plus its top 50 targets; (closed loop) 18 leg sensory neurons, 20 randomly chosen interneurons (np.random.seed(7)), 12 leg motor neurons", "neuron_model": "LIF (numpy, hand-set parameters; bridge: tau 15 ms, syn_gain 60; closed loop: syn_gain 40)", "input_mapping": "hand-made: (bridge) a tonic current I_ext = 25 into the chosen descending neuron, which has no inputs in the sub-circuit; (closed loop) joint angles encoded as currents into sensory neurons assigned to random legs", "output_mapping": "hand-made: (bridge) amplitude gain = clip(rate/60, 0.3, 2.0) applied equally to all six CPG oscillators of FlyGym's tripod CPGController; (closed loop) motor spikes per leg min-max scaled to gains 0.2-2.0 multiplying a hand-written sinusoid", "trained_parts": "STDP in the closed-loop script (A_pre = A_post = 0.01, tau 20 ms)", "body": "FlyGym", "scripted_parts": "the stepping pattern: FlyGym's tripod CPG (bridge) or a hand-written 0.3*gain*sin(phase) joint target (closed loop)" }, "trained_class": "other", "trained_class_note": "STDP in the closed-loop script", "grade_basis": [ "scripts/banc_flygym_bridge.py, scripts/banc_flygym_closed_loop.py, scripts/build_banc_subnet.py, scripts/test_banc.py and README.md at 8a794e33 (2026-09-26), read 2026-10-02 from a depth-1 blobless clone [direct]", "In the bridge, neuron 0 (the descending neuron) receives no synaptic input inside the sub-circuit (recurrent edges are added only among the 50 targets), so its rate, and therefore the CPG gain, is set by the tonic current and the LIF constants, not by the connectome [direct, read from code]" ], "grade_date": "2026-10-02", "measured_result": "None reported beyond printed rates and gains (for example '96 Hz -> 1.60x') and rendered videos; no comparison or control.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/arisliwind/flygym", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file (FlyGym's Apache-2.0) at 8a794e33", "data_licence": "BANC: see Harvard Dataverse terms (not checked)", "platform": [ "Linux", "macOS" ], "platform_note": "Python with uv; FlyGym 2.1.0", "gpu": "none", "download_size": "FlyGym 2.1.0 environment (about 0.6 GB) plus the two BANC feather files", "last_commit": { "date": "2026-09-26T01:41:18+08:00", "hash": "8a794e33bc893aa43686fed38dc098c6ded7ae15", "branch": "main" }, "pushed_at": "2026-09-25T17:41:23Z", "last_release": "none", "created_at": "2026-09-25T16:53:51Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "peer_review_note": "none", "link_status": { "url": "https://github.com/arisliwind/flygym", "status": "ok", "http_code": 200, "final_url": "https://github.com/arisliwind/flygym", "checked_at": "2026-10-06T09:40:11.152Z" }, "last_verified": "2026-10-06", "added_in_run": "run:03aa82ff-d83b-47ec-bb36-c8edc64e696c", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/arisliwind/flygym [direct]" ], "notes_limitations": "Grade C: a small selected or random subset with hand-set parameters, and the walking itself is FlyGym's CPG or a hand-written sinusoid; in the bridge script the connectome cannot change the drive at all. A fork of FlyGym 2.1.0 (not a lure copy: the added scripts are original). Closest existing design to our own 'Add a body' test; differences are in body/results.md of run 6." }, { "id": "flys-hash-function", "name": "The Fly's Hash Function", "type": "research", "author_or_org": "realgauravvyas", "summary": "Tests whether the measured hemibrain projection-neuron to Kenyon-cell wiring works better as a FlyHash similarity-search hash than the random matrix used in the literature, with several matched null matrices. A second study tries a mushroom-body readout on sudoku sub-steps.", "claim": { "text": "On generic similarity search the real connectome is ~20% worse than the idealised random matrix standing in for it", "url": "https://github.com/realgauravvyas/flys-hash-function" }, "dataset": "hemibrain", "release": "v1.2 (traced adjacencies; 130 uniglomerular PNs, 1,745 KCs, >=3 synapses)", "evidence_grade": "A", "grade_note": "Borderline A: a well-controlled negative result, but the model is a single feed-forward step, not a simulation over time.", "mechanism": { "wiring": "subset: 130 PNs x 1,745 KCs of hemibrain v1.2", "neuron_model": "other: one-step feedforward projection + top-5% winner-take-all (FlyHash), no time dynamics", "input_mapping": "hand-made (shared fixed random encoder from data to PN space)", "output_mapping": "hand-made (binary KC code used for nearest-neighbour search)", "trained_parts": "none for hashing (sudoku study trains readouts)", "body": "none", "scripted_parts": "none found" }, "trained_class": "none", "trained_class_note": "none for hashing", "grade_basis": [ "src/extract_connectome.py:5-6,67 builds real PN->KC matrix from hemibrain v1.2 export, 3-synapse threshold [direct]", "src/extract_connectome.py:89-100 random FlyHash null; :102-140 Maslov-Sneppen degree-preserving swap; :142,156 glomerulus-matched and weight-shuffled nulls [direct]", "src/benchmark.py:46,88-104,126 shared encoder, top-5% KC code, precision@k [direct]", "out/benchmark.json mnist precision@16: real 0.4993, degree_preserving 0.5520, random_flyhash 0.6212 [direct]", "out/specialization.json: own-world real 0.2187 vs random 0.1552; uniform world 0.0944 vs 0.1619 [direct]" ], "grade_date": "2026-09-28", "measured_result": "out/benchmark.json, precision@16, 5 seeds (MNIST / Fashion-MNIST / correlated): real 0.499 / 0.480 / 0.368; real binary 0.547 / 0.518 / 0.420; degree-preserving 0.552 / 0.524 / 0.427; random FlyHash 0.621 / 0.581 / 0.511. out/specialization.json: in the connectome's own synthetic input world real 0.219 vs random 0.155 (+41%), degree-preserving 0.230; in a uniform world real 0.094 vs random 0.162. Values match the README. Measured by the author, not reproduced by us.", "try_url": "https://realgauravvyas.github.io/flys-hash-function/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/realgauravvyas/flys-hash-function", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit 271cc854dc); GitHub API spdx_id=None", "data_licence": "CC-BY (version not stated)", "platform": [ "browser", "Linux", "macOS" ], "gpu": "optional", "download_size": "not stated (hemibrain v1.2 adjacency tarball, MNIST)", "last_commit": { "date": "2026-09-13T00:49:15+05:30", "hash": "271cc854dcd2c517a41074c493a504681c12df8a", "branch": "main" }, "pushed_at": "2026-09-12T19:19:18Z", "last_release": "none", "created_at": "2026-09-12T18:42:38Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://realgauravvyas.github.io/flys-hash-function/", "status": "ok", "http_code": 200, "final_url": "https://realgauravvyas.github.io/flys-hash-function/", "checked_at": "2026-10-06T09:40:10.679Z", "other_links": [ { "url": "https://github.com/realgauravvyas/flys-hash-function", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "worse", "sources": [ "https://github.com/realgauravvyas/flys-hash-function [direct]", "https://realgauravvyas.github.io/flys-hash-function/ [direct: page title checked, HTTP 200]" ], "notes_limitations": "Well-controlled negative result for a small circuit: the real matrix hashes worse than random, and even in its 'own world' a degree-preserving shuffle is slightly better, so any specialisation sits in the degree sequence. Model is a single feedforward step with k-winner-take-all, not a dynamic simulation; the 'fly's own world' inputs are synthetic, drawn from the connectome's own covariance. No licence file (GitHub reports none). Sudoku study is a separate readout experiment with trained layers. Measured by the author, not reproduced by us." }, { "id": "haltere", "name": "Haltere", "type": "research", "author_or_org": "skulitom", "summary": "A 30,000-neuron subgraph of the Janelia male CNS v1.0 connectome (flight-relevant sensory, compass, premotor and wing motor populations) is run as a rate RNN. It is trained by imitation and on flight tasks to output throttle, roll and pitch for an FPV drone in the game Liftoff. A separate visual pilot supplies the velocity goal and yaw. The brain has been compared with a PD controller in a few full races.", "claim": { "text": "A connectome-constrained fruit-fly brain controlling an FPV drone in Liftoff", "url": "https://github.com/skulitom/haltere" }, "dataset": "MaleCNS", "release": "v1.0 (filtered subgraph: 30,000 of 71,618 candidate neurons, 2,767,698 edges)", "evidence_grade": "C", "grade_note": "The file citations behind this C grade were spot-read by us, not re-verified line by line.", "mechanism": { "wiring": "MaleCNS v1.0 subgraph selected by population queries (haltere, wing campaniform, LPTC, ocelli, JO, EPG compass, premotor, wing MNs), min 3 synapses, optic-lobe intrinsic excluded", "neuron_model": "rate units (sigmoid), learnable per-neuron time constant, gain and bias", "input_mapping": "learned population encoders from telemetry/vision channels into sensory populations", "output_mapping": "trained readout from wing motor / premotor neurons to 4 stick channels", "trained_parts": "Per-edge weight magnitudes (sign and structure fixed by the connectome), unknown signs, neuron parameters, encoders and readout. Trained by gradient descent and imitation of an MLP teacher.", "body": "simulated quadcopter in the game Liftoff", "scripted_parts": "Race-cue visual pilot (heuristic checkpoint guidance) supplies the velocity goal and yaw; obstacle clearance heuristics" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "per-edge magnitudes trained by gradient descent", "grade_basis": [ "data/built/flight.meta.json: version v1.0, n_candidates 71618, n_nodes 30000, n_edges 2767698, whole_cns false [direct]", "haltere/brain/model.py:1-7, 100-140 ConnectomeRNN: connectome fixes structure and sign; per-edge log-gain, taus, gains and biases are nn.Parameters [direct]", "haltere/brain/model.py:24 BrainConfig model 'mlp' baseline option; artifacts/mlp_baseline.pt used as imitation teacher [direct]", "haltere/liftoff/fast_race_cue.py:1-28 heuristic velocity-level checkpoint guidance that supplies goal and yaw [direct]", "docs/flight_cards/2026-09-23_matched_full_races.md: brain 0/2 finishes vs PD 1/2 [direct]" ], "grade_date": "2026-09-28", "measured_result": "Measured by the author, not reproduced by us. In a frozen full-race comparison the brain finished 0/2 races and the PD controller 1/2 (docs/flight_cards/2026-09-23_matched_full_races.md). The later fast-brain-08 finished Straw Bale twice in about 5:18 with the race-cue pilot, against a user target of 1:34.8; it has not finished Minus Two or Pine Valley (README). Release fast-brain-11 (29 Sep 2026, experimental) flew one full Straw Bale lap in 1:42.988 with a braking motor readout, but every flight ended in a crash and it has not finished a race; fast-brain-08 remains the author's published race result (docs/fast_brain_11_release.md). Round-6 flights (29 Sep 2026, docs/fast_brain_11_release.md) took fast-brain-11 through the Minus Two hairpin for the first time, still with no race finish.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/skulitom/haltere", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 7189690794; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "Windows" ], "gpu": "not stated", "gpu_note": "not stated (PyTorch; GPU helpful for training)", "download_size": "repository about 250 MB including checkpoints (several 11-20 MB .pt files); requires Liftoff on Steam", "last_commit": { "date": "2026-09-29T10:42:56+01:00", "hash": "7189690794153ea7f777c2ea5ed7c38926995a8c", "branch": "main" }, "pushed_at": "2026-09-28T04:19:37Z", "last_release": { "tag": "fast-brain-11 (experimental): first braking brain motor to fly a full Straw Bale lap (1:42.988)", "date": "2026-09-29T09:43:23Z", "url": "https://github.com/skulitom/haltere/releases/tag/fast-brain-11-experimental" }, "created_at": "2026-09-12T12:57:17Z", "stars": 11, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/skulitom/haltere", "status": "ok", "http_code": 200, "final_url": "https://github.com/skulitom/haltere", "checked_at": "2026-10-06T09:40:11.029Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/skulitom/haltere [direct: clone HEAD e2ee246]" ], "notes_limitations": "The network is a filtered subgraph (30,000 neurons) with trained weight magnitudes and a trained readout, and it was taught by imitating a non-connectome MLP. Navigation (velocity goal and yaw) comes from a hand-written visual pilot, so the brain acts as a low-level motor controller. There is a PD baseline, but the n is tiny (2 races each), and there is no shuffled-wiring control for flight. The author openly reports that reliable general racing is unsolved." }, { "id": "larva-vs-shuffles", "name": "Does the larval connectome beat its own shuffles? (connectome-null-models)", "type": "research", "author_or_org": "cqw-acq", "summary": "Uses the larval Drosophila connectome as a frozen recurrent reservoir for MNIST and CIFAR-10 and compares it with degree-preserving, weight-shuffled, Erdos-Renyi and no-recurrence versions under the same training. The real wiring gives no measurable gain over its shuffles.", "claim": { "text": "Against its own degree-preserving shuffle the gap is +0.09 pp, 95% CI [-0.20, +0.38], p = 0.51.", "url": "https://github.com/cqw-acq/connectome-null-models" }, "dataset": "larval L1", "release": "Winding et al. 2023 via netzschleuder 'fly_larva' (no version number); 2,956 neurons, 63,545 axon-dendrite edges", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "larval connectome (Winding 2023), 2,956 neurons, 63,545 'ad' edges, spectral radius pinned at 0.95", "neuron_model": "rate: leaky tanh reservoir, 8 steps", "input_mapping": "learned (trained linear projection onto 434 sensory neurons)", "output_mapping": "learned (trained linear readout from 346 descending neurons)", "trained_parts": "input projection and readout only; recurrent matrix frozen", "body": "none", "scripted_parts": "none found" }, "trained_class": "readout-or-decoder", "trained_class_note": "input projection and readout trained; recurrent frozen", "grade_basis": [ "src/connectome.py:63-99 degree_preserving_swap: directed double-edge swap with self-loop and duplicate checks [direct]", "src/connectome.py:102-115,145-184 Erdos-Renyi, weight shuffle, no-recurrence floor; same Dale sign vector and spectral normalisation for all conditions [direct]", "src/model.py:15-47 frozen reservoir, only inject/readout trainable [direct]", "src/run.py:93-95 same torch seed for every condition at a given seed [direct]", "results/results_cifar10.json: recomputed by us from the file: connectome 43.04% (n=15) vs degree-preserving 42.95% (n=15), gap +0.093 pp, Welch t = 0.66 (consistent with p = 0.51) [direct]" ], "grade_date": "2026-09-28", "measured_result": "CIFAR-10 (results/results_cifar10.json): connectome 43.04% +/- 0.40 (n=15), degree-preserving 42.95% +/- 0.37 (n=15), gap +0.09 pp, 95% CI [-0.20, +0.38], p = 0.51; weight shuffle 42.96%, Erdos-Renyi 42.98%, no recurrence 10.00%. MNIST (results/results.json, n=5): connectome 97.30%, degree-preserving 97.28%, weight shuffle 97.38%, Erdos-Renyi 97.53%, no recurrence 11.35%. We recomputed means and the gap from the committed JSON files; we did not rerun training. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/cqw-acq/connectome-null-models", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit be301d9c2c; GitHub API spdx_id=MIT", "data_licence": "article CC BY 4.0; no separate data licence found", "platform": [ "macOS", "Linux" ], "gpu": "optional", "download_size": "small (netzschleuder fly_larva CSV zip plus MNIST/CIFAR-10 via torchvision)", "last_commit": { "date": "2026-09-12T15:14:23-04:00", "hash": "be301d9c2c5fe481c68db174cbed40f1fb2881c9", "branch": "main" }, "pushed_at": "2026-09-12T19:14:30Z", "last_release": "none", "created_at": "2026-09-12T18:49:47Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/cqw-acq/connectome-null-models", "status": "ok", "http_code": 200, "final_url": "https://github.com/cqw-acq/connectome-null-models", "checked_at": "2026-10-06T09:40:11.607Z" }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "no-difference", "sources": [ "https://github.com/cqw-acq/connectome-null-models [direct]", "https://networks.skewed.de/net/fly_larva [search-summary: named in README, not fetched]" ], "notes_limitations": "Well-controlled negative result, with confounds (spectral radius, shared init, fixed signs) handled in code. Limits: one substrate, image tasks the larva never evolved for, excitatory/inhibitory signs assigned at random (dataset has no transmitter labels), n=15 resolves only d >= 1.02. Absolute CIFAR-10 accuracy (43%) is well below the BPU paper's 58%. The README's MNIST table and CIFAR table match the committed result files. Measured by the author, not reproduced by us." }, { "id": "loihi2-fly-brain", "name": "Neuromorphic Simulation of Drosophila Melanogaster Brain Connectome on Loihi 2", "type": "research", "author_or_org": "Felix Wang, Bradley H. Theilman, Fred Rothganger, William Severa, Craig M. Vineyard, James B. Aimone (Neural Exploration and Research Laboratory, Sandia National Laboratories)", "summary": "The authors port the Shiu et al. whole-brain LIF model of FlyWire to 12 Intel Loihi 2 neuromorphic chips, going through their STACS simulator. They check that spike rates match the Brian2 reference for the sugar-neuron experiment and measure speed.", "claim": { "text": "We demonstrate the first-ever nontrivial, biologically realistic connectome simulated on neuromorphic computing hardware.", "url": "https://arxiv.org/abs/2508.16792" }, "dataset": "FlyWire FAFB", "release": "not stated (about 140K neurons, 50M synapses, condensed to ~15M connections; model replicated from Shiu et al. 2024)", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "FlyWire FAFB, release not stated (~140K neurons)", "neuron_model": "LIF (fixed-point microcode approximation of Shiu et al. model)", "input_mapping": "hand-made", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "none found" }, "trained_class": "not-assessed", "grade_basis": [ "https://arxiv.org/html/2508.16792 Sec. 3.1: network model 'replicated from [7]' (Shiu et al., Nature 2024); ~140K neurons, synapses condensed to ~15M connections [direct]", "https://arxiv.org/html/2508.16792 Sec. 3.1.1 / Fig. 6: STACS vs Brian2 spike-rate parity for the sugar neuron experiment over 10 trials [direct]", "https://arxiv.org/html/2508.16792 Conclusion / Table 1: Loihi 2 ran ~3x to ~350x faster than the Brian2 reference; faster than real time at low activity [direct]", "No code or data link found in the abstract page or full text [direct]" ], "grade_date": "2026-09-28", "measured_result": "Engineering validation only: spike rates compared with the Brian2 reference simulation (sugar experiment, 10 trials). Table 1, wall-clock per 1 s simulated for the sugar experiment: Brian2 4419 ms, STACS 2656 ms, Loihi 2 53.76 ms (0.1 ms step) and 12.40 ms (1 ms step); overall ~3x-~350x faster than Brian2 depending on activity level. No comparison with fly data and no wiring control.", "try_url": null, "try_status": "no no-install option", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY-NC-4.0", "platform": [], "gpu": "not stated", "download_size": "not stated", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": "https://arxiv.org/abs/2508.16792", "peer_review": "preprint", "link_status": { "url": "https://arxiv.org/abs/2508.16792", "status": "ok", "http_code": 200, "final_url": "https://arxiv.org/abs/2508.16792", "checked_at": "2026-10-06T09:40:11.050Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://arxiv.org/abs/2508.16792 [direct]", "https://arxiv.org/html/2508.16792 [direct]", "https://videocast.nih.gov/watch=55007 [search-summary]" ], "notes_limitations": "This is a hardware/simulation-speed paper, not a new model of behaviour: it is a port of the Shiu et al. model and its validation is parity with the Brian2 reference, not with flies. No public code found, so the mechanism cannot be inspected (grade U); as a simulator it would otherwise sit with infrastructure (n/a). Needs Loihi 2 hardware (Intel research access), not available to the public. FlyWire release not named; ~140K neurons suggests v783 but this is our inference. Fixed-point arithmetic, weight quantisation/capping and timestep changes mean small differences from the reference. arXiv v1, 22 Aug 2025 (cs.DC); an earlier version was presented in an NIH lecture in Oct 2024 [search-summary]." }, { "id": "neuroweave", "name": "NeuroWeave", "type": "research", "author_or_org": "Parva Trivedi (Titanium-xd)", "summary": "A small benchmark that uses 150 MaleCNS neurons and their connections as a sparsity mask on a trainable graph network, then compares it with a same-density random graph, a configuration-model graph, a dense layer, an MLP and an LSTM on two toy tasks: static binary pattern classification and delayed recall of a binary cue. A browser site shows the results and a replay page that draws outcomes from the measured accuracies.", "claim": { "text": "Fruit Fly Brain vs AI: a benchmark comparing connectome-topology-constrained neural architectures derived from a real Drosophila nervous system against conventional AI models on controlled tasks.", "url": "https://github.com/Titanium-xd/neuroweave" }, "dataset": "MaleCNS", "release": "v1.0 (stated); 150-neuron subgraph taken as the first 150 rows of a neuron table, 3,029 edges stated (stored T-002 results record 1,681)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "A 150-neuron MaleCNS subgraph (first 150 rows of the author's neuron parquet, not a named circuit) used as a mask: only real edges carry weights.", "neuron_model": "rate: graph layers with learned edge scales and biases; a hand-set leaky/decay state ('SA-010', decay 0.05, leak 0.02) for the recall task", "input_mapping": "learned input projection from a 16-value observation onto the graph nodes", "output_mapping": "learned readout head from node states to the answer", "trained_parts": "Edge scales, biases, input projection and readout (A1-BIO); in the 'frozen' variant the graph weights are fixed but the readout and projections still train (2,852 trainable parameters in the stored result)", "body": "none (abstract classification tasks)", "scripted_parts": "Tasks and training loop are ordinary code. The browser 'Live Arena' does not run the models; it samples outcomes from the reported accuracies." }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "edge scales, biases, projections, readout trained", "grade_basis": [ "scripts/tasks/run_t001_topology.py:32-37 neurons = read_parquet(...).head(max_nodes): subgraph is the first 150 rows [direct]", "abb/models/topology.py:70,101,230,299-345 Erdos-Renyi, configuration-model and weight-shuffle controls implemented [direct]", "artifacts/t001_topology_results.json: T-001 single seed, A1-BIO 0.98 vs A3-ER 1.00, A3-CONFIG 0.99, A3-DENSE 1.00 test accuracy [direct]", "scripts/tasks/run_confirmation_t002.py:77-83 T-002 confirmation compares only A0-Random, A1-BIO, A1-FROZEN and A8-LSTM; no random-graph control [direct]", "artifacts/confirmation_t002/A1-BIO_0.05_0.02.json: 0.948 mean, 5 seeds, graph_edges 1681 [direct]", "artifacts/confirmation_t002/A1-FROZEN_0.05_0.02.json: 0.924 mean with 2,852 trainable parameters (README says 72.2% and 0) [direct]", "artifacts/confirmation_t002/A8-LSTM.json: 0.576 mean, CI95 0.268 (README says 8.9%) [direct]", "abb/models/bio_variants.py:100-125 trained readout head; freezing applies to graph layers only [direct]", "git ls-files: no abb/data package and no data/ folder in the repository, though tests import abb.data.graph [direct]" ], "grade_date": "2026-09-29", "measured_result": "Stored results: T-001 (static classification) all trained models near ceiling; the MaleCNS mask scored 0.98 and the same-density random graph 1.00 (single seed), so no topology effect. T-002 (delayed recall, 5 seeds): MaleCNS mask 94.8%, frozen-mask variant 92.4%, LSTM 57.6%; no random-graph control was run on T-002. Several README figures do not match the stored files. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Titanium-xd/neuroweave", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 518f162ef7; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "Windows", "macOS" ], "platform_note": "Python/PyTorch research code; a results website is linked from the README", "gpu": "optional", "download_size": "small repository; MaleCNS extract must be fetched separately (loader code missing from repo)", "last_commit": { "date": "2026-09-27T15:17:16+05:30", "hash": "518f162ef7a68022d3dfa9466d3b8b73dd92652c", "branch": "main" }, "pushed_at": "2026-09-27T09:47:16Z", "last_release": "none", "created_at": "2026-09-15T09:33:41Z", "stars": 3, "stars_date": "2026-09-29", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-29", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Titanium-xd/neuroweave", "status": "ok", "http_code": 200, "final_url": "https://github.com/Titanium-xd/neuroweave", "checked_at": "2026-10-06T09:40:11.659Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "wiring-null", "wiring_effect": "no-difference", "sources": [ "https://github.com/Titanium-xd/neuroweave [direct: clone HEAD 518f162, 2026-09-27]" ], "notes_limitations": "The 'fly' network is 150 neurons picked as the first rows of a table, not a named circuit, with every weight trained. The only topology control (random graph of the same density) was run on the task where every model hits ceiling, and it did as well as the fly mask. The headline memory result compares against an LSTM, not against rewired wiring, so it does not show that the fly wiring matters. README numbers (frozen 72.2%, 0 parameters, LSTM CI 8.9%, 3,029 edges) differ from the stored result files (92.4%, 2,852 parameters, CI 26.8%, 1,681 edges). The data-loading module is missing from the repository. The 'Live Arena' page replays sampled outcomes, not the models. The author states many of these caveats." }, { "id": "shiu-model-attractor", "name": "Hidden attractor in the Shiu et al. whole-brain model (code and results)", "type": "research", "author_or_org": "xiangdoz (paper authors not named in the README)", "summary": "Code, raw results and pre-registered plans for a paper showing that the Shiu et al. FlyWire v783 LIF model is bistable: a strong enough sugar (or single-glomerulus olfactory) drive switches it into a state in which about 8,100 neurons keep firing for at least 10 s without input, resting on the sign given to antennal-lobe local neurons with no neuron-level transmitter call. Includes a wiring-only screen that predicts and removes the attractor.", "claim": { "text": "\"The leaky integrate-and-fire model of Shiu et al. (2024, Nature), built on the FlyWire v783 connectome, is bistable ... about 8,100 neurons keep firing for at least 10 s without input.\" (README)", "url": "https://github.com/xiangdoz/fly-connectome-attractor" }, "dataset": "FlyWire FAFB", "release": "v783 (Shiu et al. model files)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 whole brain, as in Shiu et al. 2024", "neuron_model": "LIF (Shiu et al. parameters; an exact batched float64 re-implementation plus the original Brian2 model.py for replication)", "input_mapping": "Poisson drive of sugar GRNs, olfactory glomeruli, Johnston's organ or visual neurons (engine/stimuli.py)", "output_mapping": "MN9 rate and the number of active neurons per time bin", "trained_parts": "none", "body": "none", "scripted_parts": "none" }, "trained_class": "none", "grade_basis": [ "README.md, prereg/GATE_PLAN.md (written 2026-09-24 'BEFORE any run'), results/g1c_brian2.jsonl, results/g1c_fast.json, results/e2_persistence.json at 07b01c8 (2026-10-01), read 2026-10-02 from a depth-1 blobless clone [direct]", "g1c_brian2.jsonl: in Shiu et al.'s own Brian2 model, seeds 8101-8103, a first pulse at s1 = 1.2 leaves 8,054-8,179 neurons active in every later bin after the input stops, against 0 with s1 = 0 (the pre-registered control); the fast engine gives identical counts (g1c_fast.json) [direct]", "e2_persistence.json: fresh seeds 8211-8215, about 8,150-8,200 neurons active in each of ten 1 s bins after the pulse [direct]" ], "grade_date": "2026-10-02", "measured_result": "After a strong sugar pulse, about 8,100 neurons stay active for at least 10 s without input in the original Brian2 model and in the re-implementation (identical counts on 3 seeds); none without the pulse.", "try_url": null, "try_status": "no no-install option (Python 3.12 + numba; Brian2 for the replication)", "code_url": "https://github.com/xiangdoz/fly-connectome-attractor", "code_licence": "MIT", "code_licence_source": "LICENSE file at 07b01c8; engine/shiu_code MIT (Shiu et al.)", "data_licence": "FlyWire data: see the Shiu et al. repository terms (not redistributed here)", "platform": [ "Linux" ], "platform_note": "Python 3.12, numpy, scipy, pandas, pyarrow, numba; brian2 for replications", "gpu": "none", "download_size": "repository with raw results; FlyWire files downloaded separately (data/README.md)", "last_commit": { "date": "2026-10-01T18:14:12-04:00", "hash": "07b01c8634bd31500eef37c4015961e816d2d56d", "branch": "main" }, "pushed_at": "2026-10-01T22:14:17Z", "last_release": "none", "created_at": "2026-09-25T17:42:43Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "peer_review_note": "paper (venue not stated)", "link_status": { "url": "https://github.com/xiangdoz/fly-connectome-attractor", "status": "ok", "http_code": 200, "final_url": "https://github.com/xiangdoz/fly-connectome-attractor", "checked_at": "2026-10-06T09:40:11.705Z" }, "last_verified": "2026-10-06", "added_in_run": "run:03aa82ff-d83b-47ec-bb36-c8edc64e696c", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/xiangdoz/fly-connectome-attractor [direct]" ], "notes_limitations": "A claim about the model, not about fly behaviour: grade A because a pre-registered control (no first pulse) and a replication in the original Brian2 code are in the result files. Relevant to Build your own: our 1 s sugar trials at 150 Hz (21 GRNs) stay at about 380 active neurons, below this attractor's regime; longer or stronger drives may not. The paper itself is not linked from the README." }, { "id": "wired-different", "name": "Wired Different (ConnectomeLens)", "type": "research", "author_or_org": "Dhruvin Sarkar", "summary": "A gradient-boosted classifier uses graph and neuropil features of the MaleCNS cell-type graph to rank which cell types are annotated as sex-related. It is tested against 500 degree-preserving randomised wirings.", "claim": { "text": "cross-validated AUC-PR of 0.759 against 0.041 by chance, and none of 500 randomized wirings matched it (p = 0.002)", "url": "https://github.com/dhruvin-sarkar/ConnectomeLens" }, "dataset": "MaleCNS", "release": "v1.0 (neuPrint male-cns:v1.0)", "evidence_grade": "n/a", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 cell-type graph (11,751 types)", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "LightGBM classifier on graph/neuropil/transmitter features", "body": "none", "scripted_parts": "none found" }, "trained_class": "not-assessed", "grade_basis": [ "pipeline/common.py:15-16 neuPrint server, DATASET = 'male-cns:v1.0' [direct]", "pipeline/null_model.py:28-50 rewire(): igraph degree-preserving swaps (10 x |E|), out-strength kept [direct]", "pipeline/null_model.py:66-76,108-140 recompute topology features, retrain same model with same folds, empirical p with +1 correction [direct]", "results/null_model_summary.json: real 0.7589 vs null 0.7116 +/- 0.0049 (full), 0.4888 vs 0.0696 +/- 0.0038 (topology only), 0/500, p = 0.002 [direct]" ], "grade_date": "2026-09-28", "measured_result": "results/null_model_summary.json: all 89 features AUC-PR 0.759 vs 500 degree-preserving rewirings 0.712 +/- 0.005 (range 0.696-0.725), 0/500 >= real, p = 0.002; topology only 0.489 vs 0.070 +/- 0.004, p = 0.002. Chance 0.041. Neuropil features alone reach 0.722. Measured by the author, not reproduced by us.", "try_url": "https://dhruvin-sarkar.github.io/ConnectomeLens/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/dhruvin-sarkar/ConnectomeLens", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 03d9c5c2d1; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "browser", "Linux", "macOS" ], "gpu": "none", "download_size": "not stated (neuPrint queries)", "last_commit": { "date": "2026-09-20T18:22:24+04:00", "hash": "03d9c5c2d194532b661141cf3c60f8e58f0707d7", "branch": "main" }, "pushed_at": "2026-09-20T14:22:43Z", "last_release": "none", "created_at": "2026-09-14T04:01:57Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://dhruvin-sarkar.github.io/ConnectomeLens/report.pdf", "peer_review": "none", "link_status": { "url": "https://dhruvin-sarkar.github.io/ConnectomeLens/", "status": "ok", "http_code": 200, "final_url": "https://dhruvin-sarkar.github.io/ConnectomeLens/", "checked_at": "2026-10-06T09:40:11.278Z", "other_links": [ { "url": "https://dhruvin-sarkar.github.io/ConnectomeLens/report.pdf", "status": "ok", "http_code": 200 }, { "url": "https://github.com/dhruvin-sarkar/ConnectomeLens", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/dhruvin-sarkar/ConnectomeLens [direct]", "https://dhruvin-sarkar.github.io/ConnectomeLens/ [direct: page title checked, HTTP 200]" ], "notes_limitations": "Graded n/a because this is a graph-statistics study, not a neural simulation: no activity or behaviour is modelled, so the A-D scale does not fit. The control itself is real and implemented in code (would meet the 'proper control' bar). Most of the signal is anatomical location (neuropil-only 0.722; randomised wirings still score 0.712), so the extra gain from wiring in the full model is small (+0.047). p = 0.002 is the floor for 500 permutations. Correlational only; labels come from existing annotations. One of three pre-specified checks failed (named types ranked 531-591). Measured by the author, not reproduced by us." }, { "id": "acamilo-flybrain", "name": "flybrain (Game Boy fly)", "type": "demo-game", "author_or_org": "Alex Camilo (acamilo)", "summary": "A spiking model of the whole FlyWire FAFB v783 brain (139,255 neurons, 2.7 million edges) plays Game Boy games, mainly Pokémon Red, for a planned 24/7 stream. Screen pixels are projected onto the fly's L1 optic-lobe columns. Descending neurons are split by index into eight groups, and a hand-made population decoder turns their rates into the eight Game Boy buttons, with hold, fatigue, hysteresis and blocked-direction rules. Game-memory reward detectors nudge a bounded set of Kenyon-cell-to-MBON gains. An optional macro mode (off by default) turns the fly's choices into scripted multi-step actions such as walking to an exit.", "claim": { "text": "A simulated fruit-fly brain (FlyWire connectome) plays Game Boy games on a 24/7 stream.", "url": "https://github.com/acamilo/flybrain" }, "dataset": "FlyWire FAFB", "release": "v783 (Codex export, retrieved 2026-09-13); 139,255 neurons, 2,700,513 edges, 1,572 L1 retina columns", "evidence_grade": "C", "grade_note": "Borderline C: in its default raw-button mode, without macros, it would sit on the B/C border. We graded what the stream shows, which is driven by macros and the decoder.", "mechanism": { "wiring": "Whole FAFB v783 Codex connection export as a CSR graph, with a sign per neurotransmitter (sign table is the author's modelling choice). No subsetting.", "neuron_model": "LIF point neurons at 1 ms: 20 ms decay, threshold 1, 2 ms refractory, weight scale 0.005, random per-neuron baseline drive and noise; TypeScript library with a bit-exact Rust service (optional CUDA kernel)", "input_mapping": "hand-set: each L1 column samples one screen pixel (nearest pixel after normalising the column's 2D position; left eye mirrored), drive = luminance x 0.20. Screen pixels, not game memory.", "output_mapping": "hand-set: descending neurons split round-robin by index into command_0..7; normalised population rates feed a fixed decoder (D-pad as one exclusive group with 800 ms hold, hysteresis 1.05, fatigue; A/B pulses; Start/Select rare pulses with a boot variant)", "trained_parts": "No trained readout. Reward-modulated STDP on a bounded set of Kenyon-cell-to-MBON gains (clamped to 0.9-1.1). Decoder constants were tuned by hand from measured room-escape runs.", "body": "none (Game Boy player)", "scripted_parts": "Reward detectors read game memory (story flags, new maps, exploration tiles, badges, catches, talks, items) and pay a scalar reward. The game layer reports blocked directions from position, which forces fatigue on that direction. Boot mode relaxes Start/Select throttling on title screens. Optional macro mode (default off) maps channels onto scripted actions (A* walks to exits/people/items, best-move attacks, menu handling)." }, "trained_class": "other", "trained_class_note": "reward-modulated STDP on KC->MBON gains", "grade_basis": [ "data/fafb-v783/meta.json:1 'FlyWire FAFB Codex v783', neurons 139255, edges 2700513 [direct]", "packages/brain/src/model/lif.ts:71-72 decayMs 20, threshold 1, refractoryMs 2, synapseScale 0.005; :244 spike propagation weight x plastic gain x scale [direct]", "packages/brain/src/model/retina.ts:41-66 one pixel per L1 column, luminance x gain; docs/limitations.md:25-31 'no claim ... that the retinal mapping resembles fly optics' [direct]", "tools/build_flywire.py:117-119 descending neurons split into N round-robin command_ roles; docs/limitations.md:13-15 buckets are a round-robin partition by neuron index [direct]", "packages/brain/src/readout/presets/gameboy.ts:32-33,61-81 hand-set decoder: exclusive D-pad (hold 800 ms, hysteresis 1.05, fatigue 0.08, blocked-direction fatigue 0.35), A/B/Start/Select pulses with boot variant [direct]", "docs/rewards-learning.md:9-30 reward rules read from game memory, 'Rewards are design' [direct]", "services/flysim/flysim.toml.example:137 mode = \"raw\"; services/flysim/crates/flysim/src/config.rs:601-606 raw is the default; docs/design/macros.md:44-58,170-186 scripted macro and plan modes [direct]", "docs/limitations.md:33-49 learning not shown to improve play; :98-105 after 43 brain minutes direction scores froze into a fixed order and only decoder habituation moved the winner [direct]", "No shuffled-wiring or no-graph comparison found in docs/ or infra/docs/; the random walker (docs/design/room-escape.md:16-20) is used to tune decoder timings, not as a control for the connectome [direct]", "docs/stream-mvp-plan.md:838-852, 901-916 the stream's own run log: progress through Pewter Gym and Mt. Moon is recorded in terms of macro actions (GO OBJECTIVE, GO WARP; '1,500 of 1,500 macros from the checkpoint'), i.e. the streamed runs use the macro mode in which the network picks among scripted multi-step actions (services/flysim/flysim.toml.example:121-126 'while a macro runs it owns the pad') [direct]" ], "grade_date": "2026-09-29", "measured_result": "Author reports decoder tuning runs: with the blocked-direction rule at hysteresis 1.05, 14 of 15 runs left the starting house against 13 without it (gameboy.ts:43-54, from infra/docs/room-escape.md). Throughput about 20 emulator fps in the browser prototype. No comparison against shuffled wiring or a random agent for the connectome itself; learning not shown to help. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/acamilo/flybrain", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 291620f5d5; GitHub API spdx_id=Apache-2.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux" ], "platform_note": "Node.js library (tests and a noise-frame example run in plain Node) plus a Rust service and Linux stream infrastructure; no Raspberry Pi build found", "gpu": "optional", "gpu_note": "CPU kernel by default; optional CUDA LIF kernel in the Rust service (flybrain-core)", "download_size": "connectome artifacts in repo about 11 MB (data/fafb-v783); ROM not included", "last_commit": { "date": "2026-10-02T13:16:49+00:00", "hash": "291620f5d5026ee3430172dd7f752fd945705b4f", "branch": "main" }, "pushed_at": "2026-09-30T03:47:48Z", "last_release": { "tag": "v0.6.7", "date": "2026-09-30T03:47:22Z", "url": "https://github.com/acamilo/flybrain/releases/tag/v0.6.7" }, "created_at": "2026-09-21T15:32:01Z", "stars": 0, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/acamilo/flybrain", "status": "ok", "http_code": 200, "final_url": "https://github.com/acamilo/flybrain", "checked_at": "2026-10-06T09:40:12.291Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/acamilo/flybrain [direct: clone HEAD 9f2345d, 2026-09-29]" ], "notes_limitations": "Grade C: the whole FAFB brain is simulated and screen pixels really reach it, but what the stream shows is not mainly the network's doing. The eye is one pixel per L1 column; the motor side is an arbitrary index-based split of descending neurons into eight buttons; a hand-tuned decoder with fatigue, hysteresis and a blocked-direction rule decides much of the play. The author's own 43-minute check found the four direction scores frozen in a fixed order, with only decoder habituation changing the winner (docs/limitations.md). The stream's run log describes progress through scripted macro actions (walk to an exit, go to a warp) that the network only selects; in the default raw-button mode the project would sit on the B/C border. Reward detectors read game memory. There is no shuffled-wiring or random-agent control for the connectome, and learning has not been shown to improve play. The author documents these limits openly. Not linked to the Raspberry Pi 5 Pokémon stream covered by XDA Developers on 28 Sep 2026: that stream reads game memory into looming-detector neurons, while this project reads screen pixels." }, { "id": "beat-saber-fly", "name": "Fly brain plays Beat Saber (X video)", "type": "demo-game", "author_or_org": "@_lyraaaa_ (lyra)", "summary": "A 31-second video on X in which a simulated fly brain is said to play the VR game Beat Saber. Only the video and the author's replies are public; no code, data or write-up was found.", "claim": { "text": "the fly brain can play beat saber", "url": "https://x.com/_lyraaaa_/status/2097527368919470162" }, "dataset": "unknown", "release": "not stated", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "not stated (third-party coverage links the demo wave to MaleCNS v1.0; the author does not confirm)", "neuron_model": "other: not stated", "input_mapping": "none", "output_mapping": "learned", "trained_parts": "By the author's replies (seen only via search): the motor side was 'overfit to a replay' of one track, the input included replay data, and visual reactivity was still being trained with RL", "body": "none", "scripted_parts": "cannot inspect; the author says the motor output was fitted to a replay of one track" }, "trained_class": "not-assessed", "grade_basis": [ "Post text via https://api.fxtwitter.com/_lyraaaa_/status/2097527368919470162: only 'the fly brain can play beat saber' plus a 31 s video, no code link [direct]", "Author replies reported in search results: 'still getting the visual cortex reactivity trained, this is so far just the motor cortex overfit to a replay'; 'overfit to one track, and has some replay data in the input' [search-summary]", "https://www.neuroai.science/p/are-flies-playing-beat-saber: critiques the demo wave in general (hand-mapped descending neurons, noise-driven output, no real sensorimotor loop); does not analyse this demo's code [fetch-summary]", "GitHub web search 'fly beat saber' found only unrelated Beat Saber mods [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "unknown", "platform": [], "gpu": "not stated", "download_size": "not stated", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://x.com/_lyraaaa_/status/2097527368919470162", "status": "ok", "http_code": 200, "final_url": "https://x.com/_lyraaaa_/status/2097527368919470162", "checked_at": "2026-10-06T09:40:13.014Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://x.com/_lyraaaa_/status/2097527368919470162 (via api.fxtwitter.com) [direct]", "https://www.neuroai.science/p/are-flies-playing-beat-saber [fetch-summary]", "https://raw.githubusercontent.com/townie/awesome-fruit-fly/main/README.md [direct]", "https://www.dexerto.com/gaming/googles-digital-fly-brain-gets-its-own-heaven-after-going-through-beat-saber-hell-3407304/ [search-summary]" ], "notes_limitations": "No code, weights or method description, so the mechanism cannot be checked (grade U). By the author's own replies, the clip shows motor output fitted to a replay of one song, with replay data in the input. It is not a fly reacting to game pixels, and visual/RL training was said to be unfinished. The replies were seen only through search summaries; the reply URLs were not fetched directly. Views were about 23.06 M at fetch time (reported ~22 M earlier), and the post is dated 2026-09-09 03:27 UTC. The dataset (MaleCNS) comes only from third-party coverage." }, { "id": "brain-runners", "name": "Brain Runners", "type": "demo-game", "author_or_org": "zack-maz", "summary": "A lane-runner study in which very different 'minds' play the same seeded tracks: TypeSafe's Jev, Claude Haiku 4.5, an untrained Shiu et al. FlyWire v783 brain (Brian2) whose looming-sensitive visual neurons are driven by gaps ahead and whose descending neurons steer and jump, and bots (solver, random, always-jump). For the fly player 'fly2' the input channel, readout and thresholds were chosen on practice seeds and frozen; it was then compared on held-out seeds with the same mapping and rule with no brain and with shuffled wiring.", "claim": { "text": "an untrained fruit fly: the adult Drosophila connectome as a spiking simulation. Gaps ahead stimulate its looming-sensitive visual neurons and its descending neurons steer. No training, only innate wiring", "url": "https://github.com/zack-maz/brain-runners" }, "dataset": "FlyWire FAFB", "release": "v783 (philshiu/Drosophila_brain_model at 91bdd1e7, sha256-checked)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "Whole FlyWire v783 brain built by the authors' own create_model (Shiu et al.)", "neuron_model": "spiking: Shiu et al. LIF in Brian2, restored per 100 ms decision window", "input_mapping": "hand-made: gaps in own lane drive LPLC2 + LC4 of both eyes; side-lane gaps drive LPLC4 + LC22 (fly2, candidate M3)", "output_mapping": "hand-made: turn = right minus left of DNa02 + DNa01 + DNg13; jump = giant fibre; thresholds chosen on practice seeds (gain 250 Hz, falloff 2, turn 40 Hz, jump 175 Hz)", "trained_parts": "none in the brain; 4 mapping/readout parameters grid-searched on practice seeds and frozen", "body": "none (game runner)", "scripted_parts": "game engine and bots; dodge-before-jump rule" }, "trained_class": "readout-or-decoder", "trained_class_note": "4 mapping/readout parameters grid-searched", "grade_basis": [ "bakeoff/fly/data.py:10-24 (commit 79d8039): Shiu et al. repository at 91bdd1e7 with sha256 checks of the v783 files [direct]", "bakeoff/fly/brain.py:1-10,67-83: untrained Shiu model built by create_model/default_params; shuffle_seed builds the shuffled-wiring control from bakeoff/fly/shuffle.py [direct]", "bakeoff/fly/shuffle.py:4-6,16-24: targets permuted among connections of the same sign; out-degree and in-degree by sign kept; counts travel with the presynaptic side [direct]", "docs/calibration/FLY2_REPORT.md 'Controls' and 'Floors' tables: fly2 82.34 held-out mean rows; no brain 74.05; shuffled wiring (seed 1) 27.84; fly 65.98; random 23.93 and always-jump 32.13 on practice seeds [direct]", "docs/DECISIONS.md:236-275 (decision 43): the authors' own reading ('our part is large'; 'the shuffled control is weak evidence') [direct]" ], "grade_date": "2026-10-01", "measured_result": "Held-out seeds 1200-1399, mean rows survived (of 150): fly2 82.3; the same mapping and rule with no brain 74.1; fly2 on shuffled wiring 27.8 (docs/calibration/FLY2_REPORT.md). The real wiring adds about 8 rows over a brainless rule; on shuffled wiring the readout neurons never fire.", "try_url": null, "try_status": "local Python install (uv); live server for side-by-side runs; Jev and Claude players need API keys", "code_url": "https://github.com/zack-maz/brain-runners", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 99dcc0a4e4; GitHub API spdx_id=MIT", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0", "platform": [ "Linux", "macOS" ], "platform_note": "Python 3.12 (uv) for the fly; browser front end for viewing", "gpu": "none", "download_size": "about 105 MB Shiu et al. data plus Python dependencies", "last_commit": { "date": "2026-10-01T10:35:21-07:00", "hash": "99dcc0a4e4687462c6483e661b3cce43927fa3cf", "branch": "main" }, "pushed_at": "2026-10-01T03:01:25Z", "last_release": { "tag": "Study runs (v1)", "date": "2026-09-30T23:02:05Z", "url": "https://github.com/zack-maz/brain-runners/releases/tag/study-v1" }, "created_at": "2026-09-19T18:23:13Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/zack-maz/brain-runners", "status": "ok", "http_code": 200, "final_url": "https://github.com/zack-maz/brain-runners", "checked_at": "2026-10-06T09:40:12.336Z" }, "last_verified": "2026-10-06", "added_in_run": "run:721af05d-86ce-46b6-9c1d-0038d2469d32", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/zack-maz/brain-runners", "https://github.com/zack-maz/brain-runners/blob/main/docs/calibration/FLY2_REPORT.md" ], "notes_limitations": "A careful, pre-specified comparison by the author, but small: one shuffle seed, no spread reported, held-out seeds had been used earlier in candidate research, and the brainless rule already reaches 90% of fly2. The shuffled control mainly shows that the real wiring routes the chosen eye cells to the chosen readout neurons." }, { "id": "doodle-fly", "name": "Doodle Fly", "type": "demo-game", "author_or_org": "dtecx", "summary": "The whole FlyWire v783 brain (Shiu et al. LIF) runs in the browser and plays a Doodle-Jump-style game: the target platform's horizontal offset drives the LC10a population of each eye as Poisson input, and the smoothed rates of left and right steering descending neurons (DNa02, DNa01 and others), balanced by a start-up calibration, press the left and right buttons. Two built-in controls (swap eyes, blind) break the play.", "claim": { "text": "A fruit fly's entire brain plays a Doodle-Jump-style game — live, in your browser. 138,639 spiking neurons · 15.1 million connections · zero training", "url": "https://dtecx.github.io/doodle-fly/" }, "dataset": "FlyWire FAFB", "release": "v783 (scripts/build_data.py downloads and packs the Shiu et al. files)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "Whole FlyWire v783 brain, signed synapse counts", "neuron_model": "spiking: Shiu et al. LIF (TypeScript, event-driven closed-form updates in a Web Worker)", "input_mapping": "hand-made: platform offset to LC10a Poisson drive per eye", "output_mapping": "hand-made: smoothed left/right steering DN rates, balanced by calibration gains, to buttons (src/control.ts:34-51)", "trained_parts": "none; a start-up calibration sets left/right gains (src/brain/worker.ts calibrate())", "body": "none (2D game sprite; a 3D fly presses arcade buttons as animation)", "scripted_parts": "game physics and rendering" }, "trained_class": "none", "trained_class_note": "start-up left/right gain calibration", "grade_basis": [ "src/control.ts:11,25-29,34-51 (commit 676b4e5): LC10a input, swap-eyes and blind controls, steering readout with calibration gains [direct]", "src/brain/worker.ts:77,130,151: calibrate() at start-up (50 Hz, 600 ms) [direct]", "README 'controls' table (falls in 2 min, typical score, best of 5): intact, eyes swapped, blind; produced by scripts/play.ts (not run by us) [direct: README table; numbers not in a results file]", "GitHub API 2026-10-01: MIT; created and last commit 2026-09-29 [direct]" ], "grade_date": "2026-10-01", "measured_result": "README table (author's runs, not re-run by us): intact brain about 12,500 typical score and 0-1 falls in 2 min; eyes swapped 133 and 41-44 falls; blind about 280 (hops in place).", "try_url": "https://dtecx.github.io/doodle-fly/", "try_status": "GitHub Pages link from README; not loaded by us in run 5 (no browser)", "code_url": "https://github.com/dtecx/doodle-fly", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 2a8a0311e4; GitHub API spdx_id=NOASSERTION", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "34.99 MB transferred in our browser check (6 Oct 2026; largest file graph.bin.gz 31.48 MB). Figure from the code before the check: FlyWire v783 package downloaded by the page (size not measured)", "last_commit": { "date": "2026-10-01T15:19:31+02:00", "hash": "2a8a0311e4e488e9bf9a7f8e79e42ffb8bcaadf5", "branch": "main" }, "pushed_at": "2026-09-29T17:18:34Z", "last_release": "none", "created_at": "2026-09-29T16:49:38Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://dtecx.github.io/doodle-fly/", "status": "ok", "http_code": 200, "final_url": "https://dtecx.github.io/doodle-fly/", "checked_at": "2026-10-06T09:40:11.917Z", "other_links": [ { "url": "https://github.com/dtecx/doodle-fly", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "runs", "checked_at": "2026-10-06", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 34.99, "compute": "in-browser-worker", "first_activity_s": 12.09, "console_errors": 0, "page_errors": 0, "matches_catalogue": "yes", "note": "Ran; in your browser (Web Worker); 34.99 MB in the first 60-90 s (largest: graph.bin.gz 31.48 MB).", "source": "Digital Fly Lab browser check of 2026-10-06 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-06-run9/#browser-heading), result file doodle-fly.json" }, "last_verified": "2026-10-06", "added_in_run": "run:721af05d-86ce-46b6-9c1d-0038d2469d32", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/dtecx/doodle-fly", "https://dtecx.github.io/doodle-fly/" ], "notes_limitations": "Input and readout are hand-made and calibrated; the two controls manipulate the input (eye swap, blinding), not the wiring, so they show that the steering follows the eyes but not that this wiring is needed. No scrambled-wiring or simple-rule baseline. The control numbers are in the README only." }, { "id": "doomfly", "name": "DOOMFLY", "type": "demo-game", "author_or_org": "nftechie", "summary": "Frames from the ViZDoom game drive about 4,100 photoreceptor inputs of a LIF simulation of the full MaleCNS v1.0 graph. Four hand-chosen descending neurons are read out as turn, move and fire buttons. An experimental dopamine-gated learning rule was tested with controls and did not produce learning.", "claim": { "text": "Game frames stimulate modeled sensory neurons; activity propagates through the retained MaleCNS v1.0 wiring, and a fixed neuron-to-button interface turns, moves and fires.", "url": "https://github.com/nftechie/DOOMFLY" }, "dataset": "MaleCNS", "release": "v1.0 (166,700 neurons, 25,582,938 directed edges)", "evidence_grade": "B", "grade_note": "Kept at B: its own controls are a full disconnection and a learning pilot with two episodes per arm. A third-party study (doom-fly-control) found that shuffled wiring stops the fly, but a brainless constant autopilot survives about as long, so there is no evidence of skilled play.", "mechanism": { "wiring": "MaleCNS v1.0, whole retained graph (166,700 neurons, 25.6 M edges), no cropping", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "No trained readout. An optional experimental dopamine-gated plasticity rule acts on 4,184 existing KC->MBON11 edges; health loss triggers a 200 ms artificial pulse into 2 PPL101 cells.", "body": "none", "scripted_parts": "none found. The readout gains are fixed joystick gains, and no game state goes into the decoder." }, "trained_class": "none", "trained_class_note": "optional experimental plasticity rule", "grade_basis": [ "doom/connectome.py:98-174 import_graph() accepts only 'malecns_v1' [direct]", "doom/engine.py:78-100 step(): each frame's luminance drives retina neurons, then the LIF kernel advance() runs over all edges [direct]", "doom/engine.py:121-125 'bci' mode: turn = DNp20 right minus left rate x 0.12, forward = DNpe017 rate x 0.4, attack = any DNpe017 spike. Hand-set gains [direct]", "outputs/doom/bci-validation.json: intact vs black pixels vs retina disconnected vs all edges disconnected. Disconnecting every edge abolishes the controls. Marked 'learning_demonstrated': false [direct]", "doom-ui/public/learning-iterations.json: v6 survival pilot with plastic, frozen and shuffled arms. Plastic mean survival 3.66 s vs frozen 5.83 s (one frozen run censored at 8 s); announcement_ready false [direct]", "docs/doom-learning-review.md: first pilot of 22 episodes; all training runs had zero Kenyon-cell spikes and zero changed memory edges, which is a negative mechanism result [direct]" ], "grade_date": "2026-09-28", "measured_result": "Negative. The learning candidate failed its vision, conditioning and survival gates. In the v6 survival pilot, the plastic arm averaged 3.66 s survival vs 5.83 s for frozen (n=2 test episodes each; doom-ui/public/learning-iterations.json). The first pilot had zero KC spikes and zero edge changes (docs/doom-learning-review.md). Baseline BCI: a 6 s closed-loop test gave 210 non-zero action tics and 3 kills; disconnecting all edges abolished the controls (docs/doom-status.md, outputs/doom/bci-validation.json).", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/nftechie/DOOMFLY", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 71ecf53d78; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS" ], "gpu": "none", "download_size": "MaleCNS v1.0 tables (several GB RAM; download size not stated)", "last_commit": { "date": "2026-09-09T10:11:10-05:00", "hash": "71ecf53d78eaffaf1a57ed7b0ccf5d458abc9f33", "branch": "main" }, "pushed_at": "2026-09-09T15:11:47Z", "last_release": "none", "created_at": "2026-09-06T01:34:18Z", "stars": 404, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/nftechie/DOOMFLY", "status": "ok", "http_code": 200, "final_url": "https://github.com/nftechie/DOOMFLY", "checked_at": "2026-10-06T09:40:12.315Z" }, "last_verified": "2026-10-06", "project_page": "/p/digital-fly-catalog/projects/doomfly/", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/nftechie/DOOMFLY [direct: local clone, HEAD 71ecf53]" ], "notes_limitations": "Grade B: the full MaleCNS v1.0 graph is simulated and drives the game, but the only controls are disconnecting all edges and an n=2 learning pilot, with no shuffled/random-wiring or no-graph baseline for the Doom behaviour. The openly reported negative learning result is recorded in measured_result. The input-output interface is engineered: the four read-out cells (DNp20, DNpe017) were picked after checking which ones respond to the screen. The author says these are not natural motor roles; the biologically motivated readout (DNa02, DNp09, MDN, MN9) produced zero actions. Photoreceptor positions and colour responses are inferred proxies, and graded visual cells are approximated as spiking. The simulation runs slower than real time (about 0.16-0.21x). The controls are small (two seeds, one reconstructed animal, 12 s episodes, some censored), and the first shuffled control failed its dose check. The author records this. Attempts: no count of learning attempts is given as a performance metric. Survival rounds restart on death, with neural state kept. The repo is a fresh snapshot with no history. The public viewer URL in the docs is a placeholder (doomfly.example)." }, { "id": "fly-brain-minecraft", "name": "Fly Brain Minecraft", "type": "demo-game", "author_or_org": "blendi-remade", "summary": "A Fabric mod for Minecraft 1.21.1. Each fly mob runs a LIF simulation of the whole MaleCNS v1.0 neuron set on the CPU. Game senses become Poisson input to sensory neuron types, and descending and motor neuron rates are decoded into walking, feeding, grooming and escape. A hand-built reflex layer and state machines fill the gaps.", "claim": { "text": "A Fabric mod for Minecraft 1.21.1 that runs the complete male fruit fly nervous system inside a fly mob.", "url": "https://github.com/blendi-remade/fly-brain-minecraft" }, "dataset": "MaleCNS", "release": "v1.0 (neuPrint male-cns:v1.0)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0: all 176,422 neurons, edges with >=5 synapses only (6,287,749 edges, about 72% of synapses), autapses dropped, sign from predicted transmitter", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none (global gain 0.65 and Kenyon-cell input gain 0.25 hand-tuned)", "body": "custom", "scripted_parts": "Reflex layer on by default (odour taxis, random exploration bouts, collision turns) runs when the brain gives no locomotor command. There is a hand-coded priority ladder of behaviour modes, a flight state machine and hand-set speed gains. Looming, small-object and optic-flow values are computed from the game's object list and injected into LC4/LPLC2/LC11/HS/VS rather than coming from a retina pathway." }, "trained_class": "none", "trained_class_note": "hand-tuned gains", "grade_basis": [ "tools/fetch_neuprint.py:148-149,164 query male-cns:v1.0 with e.weight >= 5 [direct]", "tools/build_flyb.py:123,172,175 writes src/main/resources/connectome/malecns-v1.0.flyb.gz, keeps weight >= min_weight, drops autapses, excludes no superclass by default [direct]", "docs/connectome-stats.json:2-22 male-cns:v1.0, min_weight 5, 176,422 neurons, 6,287,749 edges [direct]", "src/main/java/com/fruitfly/brain/LifConfig.java:12-39 and LifNetwork.java:326-344: LIF (tau 20 ms, threshold -45 mV, 0.275 mV/synapse) [direct]", "src/main/java/com/fruitfly/brain/SensoryEncoders.java:10-20,303-310 Poisson rates on sensory types; analytic looming injected into LC4/LPLC2 [direct]", "src/main/java/com/fruitfly/brain/MotorMap.java:83-114 and MotorDecoder.java:175-222 hand-weighted DN readouts and priority ladder with thresholds [direct]", "src/main/java/com/fruitfly/entity/FlyBody.java:44-80 and FruitFlyConfig.java:50 reflex layer on by default [direct]" ], "grade_date": "2026-09-28", "measured_result": "Sanity benchmarks only, with no wiring control (README 'Validation', docs/VALIDATION.md). Silent brain gives 0 spikes. Sugar GRNs at 120 Hz drive MN9 to 30-90 Hz, and adding bitter suppresses MN9 to 0-10 Hz. Looming at 150 Hz drives the giant fibre to 330-380 Hz. These are compared with reference values from the Shiu et al. 2024 model, not with fly recordings.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/blendi-remade/fly-brain-minecraft", "code_licence": "MIT (code); LICENSE appends a data note: male-cns:v1.0 data CC BY 4.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 6cfa301750; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "Windows", "macOS", "Linux" ], "gpu": "none", "download_size": "~23 MB connectome inside the mod jar (README); build from source with JDK 21 (no release jar)", "last_commit": { "date": "2026-09-05T20:14:21-07:00", "hash": "6cfa30175003ef25da68a237d5eda958f8047b82", "branch": "main" }, "pushed_at": "2026-09-06T03:14:27Z", "last_release": "none", "created_at": "2026-09-06T03:10:39Z", "stars": 145, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/blendi-remade/fly-brain-minecraft", "status": "ok", "http_code": 200, "final_url": "https://github.com/blendi-remade/fly-brain-minecraft", "checked_at": "2026-10-06T09:40:12.639Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/blendi-remade/fly-brain-minecraft [direct: local clone, HEAD 6cfa301]" ], "notes_limitations": "LICENCE: the API says NOASSERTION because LICENSE lines 23-27 add a data note (male-cns:v1.0, CC BY 4.0) after the MIT text. Code is MIT and data is CC BY 4.0. DATA: 'complete nervous system' holds for neurons, but only edges with >=5 synapses are kept, and gain is tuned to compensate. The bundled .flyb.gz blob was not opened (missing from the shallow clone); its origin and filter were checked through the build scripts and connectome-stats.json. BEHAVIOUR: the README itself marks food-seeking walking and flight as hand-built. The reflex layer is on by default, so some walking seen in play is not from the network (the HUD labels it [REFLEX]). Looming input is analytic. Female mobs use the male brain. There is one commit." }, { "id": "fly-brain-rainbow-six-siege", "name": "Fly brain plays Rainbow Six Siege (claim)", "type": "demo-game", "author_or_org": "Oleksandr Samoilenko (GHOOD_BHOY)", "summary": "A claim, made on X on 26 Sep 2026 and reported by dev.ua on 28 Sep 2026, that an artificial model of the fruit fly brain was built into the shooter Rainbow Six Siege and scored its first kill after a month of training on more than 20 TB of professional gameplay footage. No code, model, dataset release or method has been published.", "claim": { "text": "A digital fruit fly brain with over 1,000,000 simulated neurons, trained for a month on 20+ TB of professional gameplay, scored its first kill in Rainbow Six Siege (as reported by dev.ua, 28 Sep 2026).", "url": "https://dev.ua/en/news/mukha-v-sidzhi-1790577179" }, "dataset": "not stated", "release": "not stated (MaleCNS is mentioned as background by the news report, not as the source used)", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "not stated; the stated size (over 1,000,000 neurons) is about six times the largest fly central nervous system reconstruction (MaleCNS v1.0, 166,700 neurons)", "neuron_model": "not stated", "input_mapping": "not stated (claimed: real-time vision and sound)", "output_mapping": "not stated", "trained_parts": "claimed: trained for a month on 20+ TB of gameplay footage", "body": "none (game character)", "scripted_parts": "unknown" }, "trained_class": "not-assessed", "grade_basis": [ "No code, model or write-up found: web searches for the author's name, handle and project on 2026-09-29 returned only the dev.ua report [search-summary]", "Google News RSS entry 'Ukrainian Man Made Fly's Digital Brain Play Rainbow Six Siege - dev.ua', 2026-09-28T06:44:00Z (leads/raw/news-scan-2026-09-29.json) [direct]", "dev.ua article text (claims of 1,000,000+ neurons, 20+ TB of footage, first kill, X post of 26 Sep 2026) known only through search summaries; the page returned HTTP 403 (bot challenge) to our fetches on 2026-09-29 [search-summary]", "Neuron counts for comparison from our catalogue dataset entries: FlyWire FAFB v783 139,255; MaleCNS v1.0 166,700 [direct]" ], "grade_date": "2026-09-29", "measured_result": "None published. The only reported result is a single in-game kill, described by the author on social media.", "try_url": null, "try_status": "no no-install option", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "unknown", "platform": [ "Windows" ], "platform_note": "Rainbow Six Siege on PC (as claimed); nothing released to run", "gpu": "not stated", "download_size": "nothing released", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "peer_review_note": "social media post, reported by one news site", "link_status": { "url": "https://dev.ua/en/news/mukha-v-sidzhi-1790577179", "status": "blocked", "http_code": 403, "final_url": "https://dev.ua/en/news/mukha-v-sidzhi-1790577179", "checked_at": "2026-10-06T09:40:12.395Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://dev.ua/en/news/mukha-v-sidzhi-1790577179 [search-summary: HTTP 403 to our fetch]", "https://news.google.com/rss/search?q=%22fruit%20fly%20brain%22%20when:7d (RSS entry, 2026-09-28T06:44Z) [direct]", "https://liquipedia.net/rainbowsix/GHOOD_BHOY (identity of the author) [search-summary]" ], "notes_limitations": "Unverified claim. Nothing can be inspected: no code, model, data or method. As described it cannot be a fly connectome simulation on its own: the stated size of over 1,000,000 neurons is about six times the largest fly central nervous system map (MaleCNS v1.0, 166,700 neurons), and training on 20 TB of human gameplay footage means any skill would come from ordinary machine learning. One kill is not a measure of play. The report relies on the author's own post; the author's tone was partly joking." }, { "id": "fly-gorilla-tag", "name": "Fly plays Gorilla Tag (video claim)", "type": "demo-game", "author_or_org": "SuperCatCrazeGT (YouTube creator; Gorilla Tag map maker)", "summary": "A YouTube video, uploaded 26 Sep 2026, claiming that a fly's \"open source brain\" learned to move around as a gorilla in the VR game Gorilla Tag, got good at it and learned to play Infection, and that the creator then played with it in VR. No code, model, map file or method has been published.", "claim": { "text": "\"Using Its OPEN SOURCE BRAIN, The Fly Learned To Move Around As A Gorilla And Got SUPER Good At It... It Even Learned How To Play INFECTION!\" (video description, chapters Phase 0 to Phase 7)", "url": "https://www.youtube.com/watch?v=wK4lixw7p4M" }, "dataset": "not stated (\"open source brain\"; no connectome release named)", "release": "not stated", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "not stated; the description says only \"its open source brain\"", "neuron_model": "not stated", "input_mapping": "not stated", "output_mapping": "not stated (claimed: the fly moves a Gorilla Tag avatar)", "trained_parts": "claimed: the fly \"learned\" over seven phases; method not stated", "body": "none (VR game avatar)", "scripted_parts": "unknown" }, "trained_class": "not-assessed", "grade_basis": [ "Watch-page metadata read 2026-10-01 ~10:58Z: title, description, chapter list, 527,756 views, 7.8K likes, upload date Sep 26, 2026 (leads/raw/youtube-watch-meta-2026-10-01.json) [direct]", "Description links only to linktr.ee/supercatcraze; its 76 links (fetched 2026-10-01) are Gorilla Tag maps on mod.io, Google Drive files, Discord, social accounts and sponsors; none names a fly, brain, connectome or code repository [direct]", "Channel search for \"fly\" on @supercatcrazegt (2026-10-01) lists this video, a short of the same claim (slfEv4o3GVI) and an earlier \"I Trained an AI to Beat Gorilla Tag's Fastest Player\" (3gpJ-X9yvTc); no code link [direct]", "GitHub repository search for \"gorilla tag fly brain\", \"gorilla tag connectome\", \"gorilla tag flywire\" and \"gtag fly brain\" on 2026-10-01: 0 results each [direct]", "The video itself was not watched (no browser in this run); on-screen claims are not recorded" ], "grade_date": "2026-10-01", "measured_result": "None published. The video shows the creator's account of seven training phases; no numbers, baseline or code are given in the metadata.", "try_url": null, "try_status": "no no-install option", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "unknown", "platform": [ "other" ], "platform_note": "Gorilla Tag VR (Meta Quest 3, as stated in the description); nothing released to run", "gpu": "not stated", "download_size": "nothing released", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "peer_review_note": "YouTube video", "link_status": { "url": "https://www.youtube.com/watch?v=wK4lixw7p4M", "status": "ok", "http_code": 200, "final_url": "https://www.youtube.com/watch?v=wK4lixw7p4M", "checked_at": "2026-10-06T09:40:13.135Z" }, "last_verified": "2026-10-06", "added_in_run": "run:721af05d-86ce-46b6-9c1d-0038d2469d32", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://www.youtube.com/watch?v=wK4lixw7p4M [direct: watch-page metadata]", "https://linktr.ee/supercatcraze [direct]", "https://www.youtube.com/@supercatcrazegt [direct: channel search]" ], "notes_limitations": "Unverified claim. Nothing can be inspected: no code, model, map or method. \"Open source brain\" suggests one of the public fly-brain projects, but none is named, and a named base project would not show how the fly was coupled to the game or what \"learned\" means. The channel's earlier video trained an ordinary AI on Gorilla Tag, so a trained controller is a plausible source of any skill. Most-viewed unrefereed fly-brain claim in our scans (527,756 views on 2026-10-01)." }, { "id": "fly-jjs-fight", "name": "Two fly brains fight in Jujutsu Shenanigans (video claim)", "type": "demo-game", "author_or_org": "Evoke (YouTube creator)", "summary": "A YouTube video, uploaded 21 Sep 2026, claiming that the creator made \"two real flies brain\" fight each other in the Roblox game Jujutsu Shenanigans (JJS). No code, model or method has been published.", "claim": { "text": "\"I Made Two Real Flies Brain Fight Each other in JJS\" (title); \"i made two flies fight each other in JJS\" (description)", "url": "https://www.youtube.com/watch?v=KC7-fpjIRN4" }, "dataset": "not stated", "release": "not stated", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "not stated (\"real flies brain\")", "neuron_model": "not stated", "input_mapping": "not stated", "output_mapping": "not stated (claimed: two fly brains control two Roblox fighters)", "trained_parts": "not stated", "body": "none (Roblox avatars)", "scripted_parts": "unknown" }, "trained_class": "not-assessed", "grade_basis": [ "Watch-page metadata read 2026-10-01 ~10:58Z: title, description (Discord and Roblox group links only), 139,364 views, upload date Sep 21, 2026 (leads/raw/youtube-watch-meta-2026-10-01.json) [direct]", "GitHub repository search for \"jjs fly brain\" (0 results), \"jujutsu shenanigans fly\" (9 results, all Roblox script-download repositories unrelated to the creator) and \"roblox fly brain\" (9 results, none linked to the creator) on 2026-10-01 [direct]", "The same creator's \"I Forced FLY to Listen AIZO (Brain Dead)\" (AXiv0AIBnt0, 15,244 views) found in the same scan; no code link [direct: search result and watch-page count]", "The video itself was not watched (no browser in this run); a second watch-page fetch was refused (HTTP 429)" ], "grade_date": "2026-10-01", "measured_result": "None published.", "try_url": null, "try_status": "no no-install option", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "unknown", "platform": [ "other" ], "platform_note": "Roblox (Jujutsu Shenanigans); nothing released to run", "gpu": "not stated", "download_size": "nothing released", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "peer_review_note": "YouTube video", "link_status": { "url": "https://www.youtube.com/watch?v=KC7-fpjIRN4", "status": "ok", "http_code": 200, "final_url": "https://www.youtube.com/watch?v=KC7-fpjIRN4", "checked_at": "2026-10-06T09:40:13.331Z" }, "last_verified": "2026-10-06", "added_in_run": "run:721af05d-86ce-46b6-9c1d-0038d2469d32", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://www.youtube.com/watch?v=KC7-fpjIRN4 [direct: watch-page metadata]" ], "notes_limitations": "Unverified claim. Nothing can be inspected. Roblox games do not accept outside controllers without client scripts, so any coupling would be the creator's own code, which is not published. Found only by the views-sorted month catch-up (139,364 views, uploaded 21 Sep 2026)." }, { "id": "fly-naf", "name": "Fly-NAF", "type": "demo-game", "author_or_org": "ArtyMend07", "summary": "The FlyWire v783 brain (138,639 neurons, LIF in PyTorch) plays Five Nights at Freddy's 1 from screen captures. A pixel difference against an empty-hallway reference drives eye clusters; a giant-fiber spike slams a door, a slow left-right eye-population difference picks which hallway to check, and DNp09 drive raises the camera tablet. The mouse is moved by these readouts.", "claim": { "text": "A whole-brain Drosophila melanogaster connectome simulation that plays Five Nights at Freddy's 1.", "url": "https://github.com/ArtyMend07/Fly-NAF" }, "dataset": "FlyWire FAFB", "release": "v783 (2025_Connectivity_783.parquet / 2025_Completeness_783.csv); 138,639 of 139,255 neurons", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "FlyWire v783, measured weights; every weight multiplied by 3 (arousal_multiplier)", "neuron_model": "spiking: LIF (adapted from eonsystemspbc/fly-brain run_pytorch.py), v_th -45 mV", "input_mapping": "hand-set: screen-region pixel difference above an MSE threshold drives chosen eye clusters at full rate", "output_mapping": "hand-set: DNp01 spike -> door, accumulated left-right eye-population potential -> which hallway, DNp09 -> tablet; fixed click positions", "trained_parts": "none (constants fitted to the simulated network's own measurements)", "body": "none (NeuroMechFly mesh for display)", "scripted_parts": "Scene analysis is a pixel-difference threshold in ordinary code; mouse positions are fixed coordinates. Since 6 Oct 2026, when DNp09 fires on camera 1C, a hand-written memory that fades over about 10 s drives the left looming inputs (LPLC2 and LC4) directly; it replaced the look at camera 2A and keeps the giant fibre firing so the left door stays shut for about 20 s." }, "trained_class": "none", "trained_class_note": "constants fitted to the network's own measurements", "grade_basis": [ "src/config.py:9-10 FlyWire 783 parquet and completeness files; :52 mse_threshold 1500; :80 v_threshold -45.0 [direct]", "src/brain_adapter.py:31-54 FlyWire ids to model indices; spikes accumulated per step [direct]", "README.md:12-30 door via giant fiber, hallway via left-right eye potential, tablet via DNp09; arousal_multiplier x3; no training [direct]" ], "grade_date": "2026-09-30", "measured_result": "The author reports reaching 5 AM on night 3 with 9% of the power left in the best recorded run, before Foxy got in ahead of 6 AM (video R4Ogbf33Gfo on YouTube, added to the README on 6 Oct 2026); the earlier best was 4 AM on night 2 (video 4UNPlA-YJtw). No control run (for example scrambled wiring or a threshold rule without the brain).", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/ArtyMend07/Fly-NAF", "code_licence": "GPL-3.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 38a56113a8; GitHub API spdx_id=GPL-3.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "Windows", "Linux" ], "platform_note": "Windows, or Linux on X11 with the game under Wine; needs Five Nights at Freddy's 1", "gpu": "optional", "gpu_note": "CUDA optional; about 3.6 frames per second on CPU (README)", "download_size": "FlyWire v783 files fetched separately; needs the game", "last_commit": { "date": "2026-10-06T01:52:08-03:00", "hash": "38a56113a8703e8074c017a1e2e5761477c34e95", "branch": "main" }, "pushed_at": "2026-09-29T18:55:29Z", "last_release": { "tag": "The whole brain in the loop", "date": "2026-09-24T16:33:48Z", "url": "https://github.com/ArtyMend07/Fly-NAF/releases/tag/v0.1.0" }, "created_at": "2026-09-01T17:57:05Z", "stars": 3, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/ArtyMend07/Fly-NAF", "status": "ok", "http_code": 200, "final_url": "https://github.com/ArtyMend07/Fly-NAF", "checked_at": "2026-10-06T09:40:13.325Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/ArtyMend07/Fly-NAF [direct: clone HEAD 7cf7d90, 2026-09-30]" ], "notes_limitations": "An honest, untrained whole-brain setup, but the game-relevant seeing is a pixel-difference threshold computed before the brain, and each action is read from one chosen neuron or population. The same threshold could close the door without a brain; no such baseline or wiring control is reported. Weights are scaled by 3. Re-read 2026-10-03 at adf8752d (PR #12, 2 Oct): the tablet camera feed now reaches the brain too. The image distance from a start-of-night reference bank drives LC9 and LC31a (Pirate Cove) or LPLC2, with its growth driving LC4 (west hall), and a DNp04 or DNp09 spike decides the action (ADR 0026). This is still a pixel-difference input with hand-set spans, computed before the brain, and there is still no wiring control, so grade B is unchanged." }, { "id": "fly-plays-games", "name": "fly-plays-games (Pokémon Red chapter; formerly fly-plays-pokemon)", "type": "demo-game", "author_or_org": "blackicon-eth", "summary": "A frozen MaleCNS v1.0 spiking model (loaded from the separate fly.ai flybrain package) sits in a Pokémon Red loop under PyBoy. The position of the nearest on-screen sprite is read from game memory and injected into the LPLC2, LC4, LPLC1 and LC10a visual projection neurons. A logistic readout on descending neurons, trained at startup on a synthetic left/right sweep, picks only left or right. The walk from Pallet Town to Viridian City shown in the video is a scripted planner that reads game memory. The connectome only lights up the brain panel there. The same repository has an Arkanoid-style chapter with wiring ablations.", "claim": { "text": "A real fruit fly connectome (MaleCNS v1.0, 166,700 neurons) running inside Pokémon Red via PyBoy; the README says the long walk is a scripted teacher, not the fly.", "url": "https://github.com/blackicon-eth/fly-plays-games" }, "dataset": "MaleCNS", "release": "v1.0 (166,700 neurons, ~25.6 million connections, as stated; brain.npz/weights.npz downloaded by the external fly.ai package)", "evidence_grade": "C", "grade_note": "The neuron model sits in an external package (fly.ai) that we did not inspect; fly.ai is not in our catalogue because it promotes a crypto token. fly-plays-games itself has no token promotion.", "mechanism": { "wiring": "Whole MaleCNS v1.0, frozen, as stated; the loader and weights live in the external alextitonis/fly.ai package, which is not in this repository", "neuron_model": "LIF (dt 20 ms, tau 100 ms, gain 3.0, tonic 0.14, as stated in the README; kernel code is in fly.ai)", "input_mapping": "hand-set from game memory: nearest sprite's screen dx (or a fixed right-side enemy in battle) drives LPLC2/LC4/LPLC1/LC10a on the left or right side by the sign of dx. The README diagram says 'screen pixels', but the code reads sprite positions from RAM.", "output_mapping": "learned: logistic readout on a decaying descending-neuron trace (PCA-compressed), two classes only (left/right); A is pressed by code when no object is off-centre", "trained_parts": "The readout only (fit at startup on 33 synthetic offsets); connectome weights frozen", "body": "none (stylised 2D fly drawing on a gamepad)", "scripted_parts": "The whole journey shown in the video: a planner builds a walkable-tile map from memory, searches frontier northward and flees battles (render/record_journey.py). In the fly-driven mode, target selection, dead zone, A presses and move timing are code." }, "trained_class": "readout-or-decoder", "trained_class_note": "readout fitted at start-up", "grade_basis": [ "pokemon-red/README.md:14-16,83-85 'the long walk from Pallet Town to Viridian City is a scripted teacher'; 'The navigation is the teacher' [direct]", "pokemon-red/render/record_journey.py:15-22,29-37 walkable-tile planner presses buttons; the brain is stepped on screen luminance only to record the activity panel [direct]", "pokemon-red/pokesim/encoder_b.py:1-10,291-323 input is sprite positions read from RAM, not pixels [direct]", "pokemon-red/play_live.py:56-61,103-110 logistic readout trained on a synthetic dx sweep; fly chooses only left or right, code presses A when no target [direct]", "pokemon-red/pokesim/compare.py:388-392 control is a shuffled-label readout, not shuffled wiring [direct]", "retroid/retroidsim/ablate.py:1-14 and retroid/README.md:79-84,117-128 wiring ablations for the Arkanoid chapter: rewired topology loses the ball in about 204 frames against >1500 for the real connectome, but shuffled weights do as well as real [direct]" ], "grade_date": "2026-09-29", "measured_result": "Author reports a cross-validated AUC of about 0.95 for decoding the object's side from descending-neuron activity when RAM offsets drive the visual neurons, with a shuffled-label control at chance (pokemon-red/README.md:93-98). No shuffled-wiring control for Pokémon. In the Arkanoid chapter, the real and weight-shuffled connectomes both survive over 1500 frames, while rewired or silenced wiring fails at about 204 frames. Measured by the author, not reproduced by us.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/blackicon-eth/fly-plays-games", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit ecb95f7b98; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS" ], "platform_note": "Python with PyBoy on a laptop CPU; setup uses Unix shell commands; Windows not stated", "gpu": "optional", "gpu_note": "FlyBrain accepts --device cpu, cuda or auto; README says it runs on a laptop CPU", "download_size": "brain.npz and weights.npz about 260 MB on first use, plus the fly.ai repository; Pokémon Red ROM must be supplied by the user", "last_commit": { "date": "2026-09-23T13:22:43+02:00", "hash": "ecb95f7b98d273752705942af2405066ff98866b", "branch": "main" }, "pushed_at": "2026-09-24T08:35:29Z", "last_release": "none", "created_at": "2026-09-19T19:57:59Z", "stars": 0, "stars_date": "2026-09-29", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-29", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/blackicon-eth/fly-plays-games", "status": "ok", "http_code": 200, "final_url": "https://github.com/blackicon-eth/fly-plays-games", "checked_at": "2026-10-06T09:40:13.220Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "wiring-null", "wiring_effect": "mixed", "sources": [ "https://github.com/blackicon-eth/fly-plays-games [direct: clone HEAD ecb95f7, 2026-09-23]", "https://github.com/blackicon-eth/fly-plays-pokemon [direct: redirects to fly-plays-games, same HEAD ecb95f7]" ], "notes_limitations": "In the Pokémon demo the connectome does not play. The video's walk is a scripted planner. In the fly-driven mode it gets one number from game memory (where the nearest sprite is) and a trained readout turns its descending-neuron activity into left or right. The README diagram says 'screen pixels', but the code reads RAM. The neuron model and connectome loader are in the external fly.ai package, so they were not inspected here. The Pokémon chapter has only a shuffled-label control. The Arkanoid chapter in the same repository has a real wiring ablation (rewired topology fails, shuffled weights do not), which would merit a separate assessment. blackicon-eth/fly-plays-pokemon now redirects to blackicon-eth/fly-plays-games (same commit), so one entry covers both. Not the Raspberry Pi 5 Twitch stream covered by XDA (different dataset, emulator and input neurons)." }, { "id": "fly-pokemon-pi5-stream", "name": "Fruit fly brain plays Pokémon on a Raspberry Pi 5 (Twitch stream)", "type": "demo-game", "author_or_org": "Leetzerzz (Twitch)", "summary": "A live Twitch stream in which a spiking model of the FlyWire FAFB v783 brain is said to play Pokémon (listed under FireRed/LeafGreen) on a Raspberry Pi 5 alongside the mGBA emulator. According to the creator, walls, doors and people are read from game memory and fed as left/right drive to looming-detector populations, and descending-neuron activity is decoded into button presses. Chat can trigger reward or punishment stimulation. No code has been published, so the mechanism cannot be checked.", "claim": { "text": "All 139,255 neurons of FlyWire FAFB v783 (connections of five or more synapses) simulated on a Raspberry Pi 5 alongside mGBA; the layout of walls, doors and people is read from game memory and fed to LPLC1, LC10 and LPLC4 as left/right drive ('there's no retinotopy'); descending-neuron activity decodes into button presses; viewers can stimulate reward/punishment neurons for about 10 minutes.", "url": "https://www.xda-developers.com/someone-wired-up-a-digital-flys-brain-in-a-pi-5-to-play-pokmon-and-you-can-help-it-progress/" }, "dataset": "FlyWire FAFB", "release": "v783 (as claimed; connections with five or more synapses)", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "claimed: whole FAFB v783, edges of five or more synapses; not inspectable", "neuron_model": "not stated", "input_mapping": "claimed: game-memory layout of walls, doors and people as left/right drive to LPLC1, LC10 and LPLC4; no image processing", "output_mapping": "claimed: descending-neuron activity decoded into button presses; decoder not described", "trained_parts": "unknown; chat commands (!pain, !pleasure) stimulate reward/punishment neurons", "body": "none (Game Boy Advance player)", "scripted_parts": "unknown" }, "trained_class": "not-assessed", "grade_basis": [ "No public code found: GitHub profile github.com/leetzerzz returns 404; searches for Leetzerzz with GitHub, fly and Pokémon return only other projects; the XDA article gives no code link [search-summary]", "acamilo/flybrain and blackicon-eth/fly-plays-games do not match the described setup (neither mentions Leetzerzz, Raspberry Pi, or LPLC4 input; acamilo uses pixel input to L1 columns, blackicon uses MaleCNS with PyBoy) [direct]" ], "grade_date": "2026-09-29", "measured_result": "None published. The Twitch title on 2026-09-29 reported in-game progress (Wartortle level 35, beat Brock, on Route 25), but it is not stated how much of this came from the fly, the decoder, chat or other code.", "try_url": "https://www.twitch.tv/leetzerzz", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "try_note": "A live Twitch stream, not a demo you control. Not watched by us; no code is published.", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY-NC-4.0", "platform": [ "other" ], "platform_note": "Raspberry Pi 5 (as claimed); viewed as a Twitch stream", "gpu": "none", "gpu_note": "claimed to run on a Raspberry Pi 5 CPU", "download_size": "nothing released (stream only)", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://www.twitch.tv/leetzerzz", "status": "ok", "http_code": 200, "final_url": "https://www.twitch.tv/leetzerzz", "checked_at": "2026-10-06T09:40:13.446Z", "other_links": [ { "url": "https://www.xda-developers.com/someone-wired-up-a-digital-flys-brain-in-a-pi-5-to-play-pokmon-and-you-can-help-it-progress/", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://www.xda-developers.com/someone-wired-up-a-digital-flys-brain-in-a-pi-5-to-play-pokmon-and-you-can-help-it-progress/ [fetch-summary]", "https://www.reddit.com/r/raspberry_pi/comments/1wrz7hs/pi_5_running_a_real_fruit_flys_brain_playing/ [search-summary]", "https://www.twitch.tv/leetzerzz [fetch-summary: page title and description only]", "https://news.google.com/rss/search?q=%22fruit%20fly%20brain%22%20when:7d (RSS entry for the XDA article, 2026-09-28T05:29Z; leads/raw/news-scan-2026-09-29.json) [direct]" ], "notes_limitations": "Only the creator's description is available: a Reddit post and the XDA summary of it. Without code there is no way to tell how buttons are decoded, whether menus, battles or movement are helped by scripts, or how much chat stimulation steers progress. The creator says openly that the input is game-memory layout rather than vision." }, { "id": "fly-terraria", "name": "Digital Fly in Terraria (MaleCNS fly with a body)", "type": "demo-game", "author_or_org": "George Ostrobrod (GitLab wdf.gost; YouTube channel George Ostrobrod)", "summary": "A tModLoader mod that puts a fly body in Terraria (two 28x28 compound eyes, smell, touch, pain, wings, six legs, proboscis) and drives it with a sparse spiking simulation of the MaleCNS v1.0 connectome with reward-modulated plasticity. The author's own report says the fly 'learned a partial survival loop, not a solved policy' and lists the missing plasticity-off control.", "claim": { "text": "\"build it a body, give it vision, smell, pain, a sense of balance and put it all in Terraria\" (video description, translated); stream: \"There are no scripted commands ... the nervous system decides how to move the body.\"", "url": "https://www.youtube.com/watch?v=0nniju8lLxk" }, "dataset": "MaleCNS (Janelia/Google male CNS connectome); a FlyWire FAFB brain-only mode also exists", "release": "v1.0 flat connectome, minconf 0.5 (166,700 neurons, 25,088,107 signed connections; SHA-256 of the three input files checked by tools/prepare_malecns.py)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 (minconf 0.5), whole CNS incl. VNC; incoming CNS edges to primary sensory neurons omitted by default", "neuron_model": "other: sparse spiking model with adaptation, short-term depression and delays (CPU/CUDA), 'simplified engineering models' per the author", "input_mapping": "hand-made: game light, chemistry, contact, load, airflow, gravity, temperature and physiology injected into annotated MaleCNS sensory populations (approximate receptor assignment)", "output_mapping": "hand-made: decode_motors() reads turn/forward/backward/feeding/adhesion/grip motor-population rates (per-leg slots in MaleCNS) through fixed activation functions", "trained_parts": "reward-modulated plasticity on eligible synapses, reward from the body's own energy, hydration and harm", "body": "custom (2-D Terraria body, tModLoader)", "scripted_parts": "documented peripheral fall-backs: a tarsal-flexion grip reflex when no grip population exists, a fixed adhesion formula when no adhesion population exists; a diagnostic taxis controller exists but is not the experiment's controller (which mode the video used was not traced)" }, "trained_class": "other", "trained_class_note": "reward-modulated plasticity", "grade_basis": [ "GitLab wdf.gost/terraria-wdf-fly-brain at 9052966d (2026-09-26): README.md, docs/BRAIN.md, docs/EMBODIMENT_AUDIT.md, docs/EXPERIMENT_REPORT.md, data/malecns/PROVENANCE.md, runtime/src/flywire_brain.cpp:148-233 (motor decode) read 2026-10-02 [direct]", "experiments/latest/lives.csv (176 lives) recomputed by us: mean lifespan 278.4 s, quarter means 223.1/325.9/251.8/312.7 s match the report; quarter medians 102.8 -> 163.3 s; our one-sided permutation p = 0.037 (last vs first 44 lives, median); Spearman rho 0.11 [direct]", "Video metadata: 0nniju8lLxk (28,244 views on 2026-10-02) and bZtcBcGVnbU (stream 2026-09-25) [video]; the video is newer (2026-09-29) than the last commit" ], "grade_date": "2026-10-02", "measured_result": "One 13.7-hour run, 176 lives: mean lifespan 278.4 s, longest 1,676.5 s, 3,353 food events; deaths thorn 88, creature 65, starvation 23 (experiments/latest/lives.csv, recomputed by us). Median lifespan of the last 44 lives 163.3 s vs 102.8 s for the first 44 (our recomputation). No plasticity-off or scrambled-wiring control.", "try_url": null, "try_status": "no no-install option (needs Terraria + tModLoader on Windows and a native runtime; CUDA optional)", "code_url": "https://gitlab.com/wdf.gost/terraria-wdf-fly-brain", "code_licence": "MIT", "code_licence_source": "LICENSE file read at GitLab commit 9052966d", "data_licence": "MaleCNS CC BY 4.0", "platform": [ "Windows" ], "platform_note": "Terraria with tModLoader (Windows .cmd launchers); native C++ runtime, optional CUDA", "gpu": "optional", "download_size": "repository plus a 131 MB Git LFS synaptic checkpoint; MaleCNS feather files downloaded separately", "last_commit": { "date": "2026-09-26T15:28:51+10:00", "hash": "9052966d", "branch": "main" }, "pushed_at": null, "last_release": "n/a", "created_at": "2026-09-25T21:59:45Z", "stars": 0, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "peer_review_note": "YouTube video and repository docs", "link_status": { "url": "https://gitlab.com/wdf.gost/terraria-wdf-fly-brain", "status": "ok", "http_code": 200, "final_url": "https://gitlab.com/wdf.gost/terraria-wdf-fly-brain", "checked_at": "2026-10-06T09:40:13.696Z", "other_links": [ { "url": "https://www.youtube.com/watch?v=0nniju8lLxk", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:03aa82ff-d83b-47ec-bb36-c8edc64e696c", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://gitlab.com/wdf.gost/terraria-wdf-fly-brain [direct]", "https://www.youtube.com/watch?v=0nniju8lLxk [video]", "https://www.youtube.com/watch?v=bZtcBcGVnbU [video]" ], "notes_limitations": "Grade B, not A: no control (the author says so and recommends paired plastic vs plasticity-disabled runs). The improvement over lives comes from one uncontrolled run whose world changed as it ran. The video (2026-09-29) is newer than the last public commit (2026-09-26). GitLab, not GitHub: stars and archive state from the GitLab API (0 stars on 2026-10-02)." }, { "id": "fly64", "name": "Fly64", "type": "demo-game", "author_or_org": "Jessica Paquette (GitHub ornata, X @barrelshifter)", "summary": "Runs a simple spiking model on the full MaleCNS v1.0 wiring and plays Super Mario 64 (patched sm64ex). Six hidden camera views around Mario drive the photoreceptor cells, and rates of chosen descending-neuron groups become stick and jump inputs through hand-set rules. No training.", "claim": { "text": "playing mario 64 using a fly's brain", "url": "https://x.com/barrelshifter/status/2097004115826200898" }, "dataset": "MaleCNS", "release": "v1.0 (minconf-0.5; 166,700 neurons, 25,582,938 edges)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0, all 166,700 superclass-annotated neurons; weights signed by transmitter and normalised by total input", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none", "body": "none", "scripted_parts": "Stick smoothing, deadzone, clamping, 0.8 s jump cooldown and a 2-frame A press. No scripted gameplay found." }, "trained_class": "none", "grade_basis": [ "fly64/data.py:16-21 downloads MaleCNS v1.0 flat tables; fly64/data.py:88-98,133-135 keeps 166,700 annotated neurons, NT signs, input normalisation [direct]", "fly64/data.py:139-146 visual inputs R1-6/R7/R8; forward = DNg100; turn = DNa02/DNg13 by side; jump = DNp01/DNp10 [direct]", "fly64/model.py:124-140 LIF step (dt 20 ms, tau 100 ms) with tonic current, noise and visual drive; fly64/model.py:142-157 hand-set rate-to-stick gains and jump threshold [direct]", "scripts/validate_causality.py:1-30 live vs frozen vs vision-disconnected check that only requires outputs to differ [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/ornata/fly", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit f2f4114e53); GitHub API spdx_id=None", "data_licence": "CC-BY-4.0", "platform": [ "macOS" ], "gpu": "none", "download_size": "~1.1 GB brain data, plus a user-supplied SM64 ROM and sm64ex build", "last_commit": { "date": "2026-09-08T21:34:01+09:00", "hash": "f2f4114e53eaa326e54129f27a5383f93c6957af", "branch": "main" }, "pushed_at": "2026-09-08T12:35:15Z", "last_release": "none", "created_at": "2026-09-07T23:34:16Z", "stars": 67, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/ornata/fly", "status": "ok", "http_code": 200, "final_url": "https://github.com/ornata/fly", "checked_at": "2026-10-06T09:40:13.706Z", "other_links": [ { "url": "https://x.com/barrelshifter/status/2097004115826200898", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/ornata/fly [direct: local clone, HEAD f2f4114]", "https://x.com/barrelshifter/status/2097004115826200898 (via https://api.fxtwitter.com/barrelshifter/status/2097004115826200898) [direct]" ], "notes_limitations": "LINK resolved: README.md line 3 links https://x.com/barrelshifter/status/2097004115826200898. The only commit author is 'Jessica Paquette', and the post's author (via api.fxtwitter.com) is Jessica Paquette (@barrelshifter), posted 2026-09-07. So flybrain.info (GitHub ornata) and Fly Brain Hub (X post) point to the same project. There is no licence file (API licence null), so by default all rights are reserved. The user must supply their own US Super Mario 64 ROM and build sm64ex. It was tested only on one M2 MacBook, and the author calls the code '100% vibe coded' and unreviewed. The LIF parameters are the author's own coarse choices (dt 20 ms), not Shiu et al. The eye-to-pixel map is partly estimated. The causality script reports no numbers, and there is no goal, benchmark or control. There is also a separate --demo-model random 4,096-cell fixture, clearly labelled." }, { "id": "flyarena-banc", "name": "FlyArena", "type": "demo-game", "author_or_org": "primaryNK (NKprime)", "summary": "A Windows game where two flies, each running its own copy of the BANC v888 connectome as a GPU LIF network, fight in a 2D physics arena with a sword and shield. Hand-chosen sensory groups get input; descending and motor neuron rates are turned into move, turn, sword and shield signals, and a small reward-trained readout adds corrections.", "claim": { "text": "BANC v888 fruit-fly connectome simulation with real-time neural control, physics combat, and reinforcement learning.", "url": "https://github.com/primaryNK/FlyArena" }, "dataset": "BANC", "release": "materialization 888 (edgelist simple v3, NT prediction v2)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "BANC 888 (README: 188,508 neurons, 13,620,865 directed pairs; not checked, data not in repo)", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "small plastic readout (19 features -> 4 residuals on forward/turn/sword/shield) trained with reward-modulated policy gradient; connectome weights fixed", "body": "custom", "scripted_parts": "none found for actions; hit/block/parry are classified from geometry; tuning gains and thresholds are hand-set" }, "trained_class": "readout-or-decoder", "trained_class_note": "small plastic readout trained by policy gradient", "grade_basis": [ "tools/prepare_banc_latest.ps1:11-17 downloads banc_888 metrics, meta, edgelist_simple_v3 and NT v2 from the BANC GCS bucket [direct]", "tools/convert_banc_latest.py:184-186,384 builds the cache from banc_888 ids, materialization 888 [direct]", "src/neural/gpu_dual_brain.cpp:240-306 HLSL LIF update for two brains (leak, threshold, reset, refractory, synaptic decay) [direct]", "tools/build_banc_io_map.py:166-182 inputs = sensory afferents by side; readouts = descending and motor neurons by side [direct]", "src/sim/v065_learning_main.cpp:879-933 visual/body senses set Poisson rates on sensory groups [direct]", "src/sim/v065_learning_main.cpp:728-876 hand-made map: MN/DN mean -> forward, L-R difference -> turn, right MN -> sword, left MN -> shield [direct]", "src/learning/plastic_readout.h:47-56 and src/sim/v065_learning_main.cpp:2096-2106 learned residual added to the base control [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/primaryNK/FlyArena", "code_licence": "custom: FASL-1.1 (view and run for non-commercial use; modified redistribution or commercial use need the author's written permission)", "code_licence_source": "LICENSE file 'LICENSE' read at commit bd6d00d2c3; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "Windows" ], "gpu": "required", "download_size": "not stated (BANC 888 files downloaded at setup)", "last_commit": { "date": "2026-09-23T19:03:36+09:00", "hash": "bd6d00d2c35b5bbaf9bd103a1ba6a2fc3ba5c47e", "branch": "main" }, "pushed_at": "2026-09-23T10:03:37Z", "last_release": { "tag": "v0.6.8", "date": "2026-09-22T15:18:55Z", "url": "https://github.com/primaryNK/FlyArena/releases/tag/v0.6.8" }, "created_at": "2026-09-22T12:16:48Z", "stars": 1, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/primaryNK/FlyArena", "status": "ok", "http_code": 200, "final_url": "https://github.com/primaryNK/FlyArena", "checked_at": "2026-10-06T09:40:14.142Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/primaryNK/FlyArena [direct]" ], "notes_limitations": "BANC 888 data is really loaded and simulated; the learner only changes a small readout (flagged as a trained part), not the connectome. The input and output maps are coarse (whole sensory/descending/motor superclasses split by side), and sword/shield are simply right/left motor-neuron rates. No control (e.g. rewired graph) or benchmark. The Windows binary was withdrawn after a Defender detection; source build only, D3D12 GPU needed. Made with GPT/Codex help (README). Licence (LICENSE file, 'FASL-1.1') is source-available, not open source: you may view, build and run it for personal, educational or non-commercial research use and share unmodified copies, but sharing modified versions, derivative projects or any commercial use needs written permission from the author." }, { "id": "flybrain-flappy", "name": "FlyBrain · Flappy (escape reflex)", "type": "demo-game", "author_or_org": "programmingWTF", "summary": "A Flappy Bird game played through the looming-escape pathway of the FlyWire FAFB v783 whole brain (144,837 neurons) run as a frozen spiking network in the author's own discrete LIF model (20 ms ticks). Approaching pipes drive LC4 and LPLC2 neurons; a flap follows when the giant-fibre neurons (DNp01) fire. The repository also holds an escape-reflex study (ESCAPE.md) with cut-pathway and shuffled-wiring variants, and a local browser demo served by a Python server.", "claim": { "text": "\"cut and shuffled both make 0 flaps: the brain no longer commands the wings\"; real wiring mean score 100.8 with 256 flaps vs 0.00 and 0 flaps for the cut and shuffled graphs over 40 games (README table, translated; measured under an earlier noise stream)", "url": "https://github.com/programmingWTF/FlyBrain" }, "dataset": "FlyWire FAFB", "release": "v783 (fafb_783_simple_edgelist and fafb_783_meta feather files, scripts/build_spiking_asset.py); whole-brain asset of 144,837 neurons and 15,023,799 synapses (README)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "FlyWire FAFB v783 whole brain as a signed sparse graph built by the author (data/spiking_full.npz)", "neuron_model": "LIF (author's own discrete model: 20 ms ticks, leak exp(-dt/tau) = 0.819, threshold 1.0, background noise; src/fpv/spiking_brain.py), not the Shiu et al. parameters", "input_mapping": "hand-made: looming angular kinematics of the pipes mapped to LC4/LPLC2 target rates through a tuning curve (s50 30, n 3), which the author calls the one external assumption (src/fpv/looming.py)", "output_mapping": "hand-made: a flap when DNp01/DNp04 fire, with a 0.14 s cooldown (demo/server.py)", "trained_parts": "none in the escape-reflex demo; an earlier controller variant with a small trained readout is still in src/fpv/spiking_controller.py", "body": "none (2D Flappy Bird game)", "scripted_parts": "game physics; projection of the pipes into a looming stimulus" }, "trained_class": "none", "trained_class_note": "the escape-reflex demo has no trained parameters; an earlier controller with a small trained readout remains in the code", "grade_basis": [ "scripts/build_spiking_asset.py at 7a8b601: reads fafb_783_simple_edgelist.feather and fafb_783_meta.feather, records source FlyWire FAFB v783 [direct]", "src/fpv/spiking_brain.py: discrete LIF, 20 ms tick, leak 0.81873, threshold 1.0 [direct]", "src/fpv/looming.py: LC4/LPLC2/LPLC1 sense groups, DNp01/DNp04 escape group, looming_drive tuning curve, variant() for shuffled targets (out- and in-degree kept), cut edges and blocked inhibition [direct]", "demo/server.py: loads SpikingBrain with the whole-brain asset and fpv.looming; FLAP_COOLDOWN 0.14 s [direct]", "No committed result files: the README's ablation commands write to output/, which is not in the repository; control numbers exist only in README.md and ESCAPE.md [direct]" ], "grade_date": "2026-10-06", "measured_result": "Not verified by us. The author's report (ESCAPE.md) gives a DNp01 threshold near 29 Hz per LC4 cell, a maximum DNp01 response of about 4.1 Hz at full drive, 11 responding descending neurons of 472 (truncated graph) or 1,301 (whole brain), no DNp01 response after cutting the 102 direct LC4 to DNp01 edges, and all 472 descending neurons silent under the shuffled graph.", "try_url": null, "try_status": "local demo only (Python server plus a three.js page, Windows .bat launchers); no hosted page found", "code_url": "https://github.com/programmingWTF/FlyBrain", "code_licence": "none found", "code_licence_source": "no LICENSE file at 7a8b601; README: no licence yet, all rights reserved by default", "data_licence": "FlyWire v783 derived asset committed in the repository (data/spiking_full.npz, about 50 MB); no data licence stated", "platform": [ "Windows", "Linux" ], "platform_note": "Python 3.10 or newer with numpy, pandas and PyTorch (CPU is enough); launchers are Windows .bat files; the server itself is standard-library HTTP", "gpu": "optional", "download_size": "repository about 53 MB at HEAD, mostly the 50.5 MB whole-brain asset data/spiking_full.npz", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "none", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/programmingWTF/FlyBrain", "status": "ok", "http_code": 200, "final_url": "https://github.com/programmingWTF/FlyBrain", "checked_at": "2026-10-06T09:56:57.566Z" }, "last_verified": "2026-10-06", "added_in_run": "run:c0ecce13-d7a1-40c8-b30e-661b847df700", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/programmingWTF/FlyBrain [direct]" ], "notes_limitations": "The control results (cut and shuffled pathways give zero flaps) are only in the author's documents, measured under an earlier noise stream, so they do not raise the grade. Per the author's own report, DNp01 stays far below its real firing range and does not fire when the stimulus covers a realistic part of the visual field, so the game needs strong, wide drive. Chinese-language documentation; no licence." }, { "id": "flybrain-halflife", "name": "FlyBrain-HalfLife", "type": "demo-game", "author_or_org": "Yusuftmle", "summary": "A leaky integrate-and-fire network reads screen captures from Half-Life (or a ViZDoom/standalone arena) and sends mouse and keyboard input. Despite the MaleCNS v1.0 / FlyWire naming, the default network is generated procedurally from region sizes and index arithmetic. Steering, attacking and escapes are computed from image heuristics and scripted reflexes, and then injected as current into the output neurons.", "claim": { "text": "Autonomous Drosophila (MaleCNS v1.0 / FlyWire) connectome agent playing Half-Life", "url": "https://github.com/Yusuftmle/FlyBrain-HalfLife" }, "dataset": "none (synthetic network labelled MaleCNS/FlyWire)", "release": "none loaded by default; an optional user-supplied DOOMFLY-format npz is accepted", "evidence_grade": "D", "grade_note": null, "mechanism": { "wiring": "Procedural: region blocks (3,600 photoreceptors, 4,800 optic lobe, 1,200 central complex, 2,400 KC, etc.) wired by modular-arithmetic tracts or random draws, not a connectome", "neuron_model": "LIF (PyTorch sparse)", "input_mapping": "hand-made: screen pixels to a 60x60 photoreceptor grid", "output_mapping": "hand-made: rate thresholds on labelled 'DNp20/DNpe017/MDN/DNp09' indices to mouse/keys", "trained_parts": "dopamine-modulated STDP and a mushroom-body avoidance module on the synthetic graph", "body": "none (game character)", "scripted_parts": "Image-derived steer bias (left/right brightness, depth balance) injected into the steering neurons; 'target in crosshair' injects attack current; damage triggers a scripted 180-degree turn, return fire and strafe; obstacle saccades; panic mode; auto-respawn" }, "trained_class": "other", "trained_class_note": "dopamine-modulated STDP", "grade_basis": [ "data_loader.py:129-330 build_canonical_flywire_connectome() builds tracts with add_tract() using index arithmetic ((i*3+shift) % dst_len); real data only if the user supplies an npz [direct]", "core/connectome.py:73-219 _build_biological_connectome(): random positions and rng.integers random synapses per region pair [direct]", "main.py:361-384 steer bias computed from left-vs-right image brightness and depth balance, later added to DNp20 stimulus (main.py:385-390) [direct]", "main.py:486-497 damage or 'is_target_in_crosshair' injects 45-70 units of current straight into the attack readout neuron [direct]", "main.py:289-296 and input_bridge.py:342-348 damage calls trigger_combat_retaliation(): scripted 180-degree flick, counter-fire and evasion [direct]" ], "grade_date": "2026-09-28", "measured_result": "none reported beyond runtime benchmarks", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Yusuftmle/FlyBrain-HalfLife", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 0ba5548be6; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Windows", "Linux" ], "gpu": "optional", "gpu_note": "optional (PyTorch CUDA; CPU lite mode)", "download_size": "repository about 15 MB; no connectome download needed", "last_commit": { "date": "2026-09-28T09:01:11+03:00", "hash": "0ba5548be6fade5e7657b3cea4e8c6177591c7a1", "branch": "main" }, "pushed_at": "2026-09-28T06:01:19Z", "last_release": "none", "created_at": "2026-09-12T15:44:30Z", "stars": 50, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Yusuftmle/FlyBrain-HalfLife", "status": "ok", "http_code": 200, "final_url": "https://github.com/Yusuftmle/FlyBrain-HalfLife", "checked_at": "2026-10-06T09:40:14.296Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/Yusuftmle/FlyBrain-HalfLife [direct: clone HEAD 0ba5548]" ], "notes_limitations": "The network is not a connectome: it is a synthetic graph with region labels, and the MaleCNS/FlyWire and v783 naming in files and cache names does not match what the code builds. The game behaviour comes mainly from image heuristics and scripted reflexes that set the output neurons directly, so the neurons act largely as a relay. There are no controls. The optional FlyWire CAVE downloader is not wired into the default run." }, { "id": "flybridge", "name": "FlyBridge", "type": "demo-game", "author_or_org": "SWOT (swotstudio)", "summary": "A Fabric mod lets Python code see and control Minecraft, and a 1,536-neuron olfactory/mushroom-body subset of FlyWire FAFB v783 is used as a rate-based recurrent core inside a trained policy network. Trainable image and world-fact encoders drive the sensory cells, and a trained readout on output cells picks a farming job and target. A scripted motor controller then walks, aims and clicks. Teaching is by imitation of a scripted teacher, then reinforcement.", "claim": { "text": "A fruit fly's brain wiring, taught to farm wheat in Minecraft (video: I Put a Fly's Brain in Minecraft and Made It Farm).", "url": "https://github.com/swotstudio/FlyBridge" }, "dataset": "FlyWire FAFB", "release": "v783 (via snedea/flybrain mirror); 1,536-neuron, 42,921-edge subset", "evidence_grade": "C", "grade_note": "The file citations behind this C grade were spot-read by us, not re-verified line by line.", "mechanism": { "wiring": "Filtered subset of FAFB v783: the top cells by internal connection strength from six olfactory-pathway groups (256 ORN, 128 LN, 256 PN, 512 KC, 96 MBON, 288 LH). Edges crossing the sample boundary are dropped. Signed log(1+count) weights, normalised by input.", "neuron_model": "rate: leaky tanh units, 4 propagation steps per decision, learned leak and bias per neuron", "input_mapping": "learned (trainable CNN image encoder, a 3-D world-block encoder and a game-fact MLP project onto ORN-group cells)", "output_mapping": "learned (LayerNorm + linear readout on MBON and lateral-horn cells, then actor/critic heads that score candidate jobs)", "trained_parts": "Everything except topology and sign: per-edge gains (0-2x), per-neuron leak and bias, encoders, readout, actor and critic. Trained by imitation of a scripted teacher, then PPO.", "body": "none (Minecraft player)", "scripted_parts": "Candidate jobs are listed by code from the mod's game state. Walking, route finding, aiming, clicks, chest and inventory handling are done by a scripted motor controller. Demonstrations come from scripted teachers. Actor output priors are hand-set." }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "all but topology and sign trained (imitation + PPO)", "grade_basis": [ "scripts/prepare_data.py:24 group quotas {6:256, 8:128, 9:256, 17:512, 19:96, 23:288}; :50 provenance 'FlyWire FAFB v783', edges crossing the sample boundary omitted [direct]", "brain/finale/source/model.py:1-5 docstring 'Engineered sensory mapping and rate dynamics, not a biological fly simulation' [direct]", "brain/finale/source/model.py:38-62 trainable edge_gain, leak, bias, CNN encoder, world/context encoders, readout, actor and critic heads [direct]", "brain/finale/source/model.py:88-96 weight = initial_weight*sign*2*sigmoid(gain); 4 tanh propagation steps; readout from output groups 19/23 [direct]", "brain/finale/source/model.py:24-31 'shuffled' and 'disconnected' variants exist in code [direct]", "brain/finale/source/farm_model.py:1-6,23-53 finale job scorer on the same 1,536-cell circuit; 'Scripted motor control is outside this decision network' [direct]", "brain/finale/source/farm_motor.py:1-4 motor assistance shared by teacher and learner (walking, camera, clicks, inventory) [direct]", "results/ and docs/: no stored fly-vs-shuffled or fly-vs-disconnected comparison for the farming finale found; docs/HUMAN_CONTROLS_EXPERIMENT.md:33 lists the shuffled comparison only as a plan [direct]" ], "grade_date": "2026-09-28", "measured_result": "Author reports the frozen finale checkpoint passed 12 of 12 fresh-world checks (README; brain/finale/summary.json, brain/finale/arena-checks/). Measured by the author, not reproduced by us. No control comparison reported for the finale.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/swotstudio/FlyBridge", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit a8950e001a; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-NC-4.0", "platform": [ "Windows" ], "gpu": "required", "gpu_note": "required (NVIDIA CUDA for training)", "download_size": "repo about 70 MB with checkpoints and demonstrations; Minecraft Java 1.21.4 must be owned separately", "last_commit": { "date": "2026-09-22T15:41:15-04:00", "hash": "a8950e001ab767fe20f63de0fc9caa29634fbd74", "branch": "main" }, "pushed_at": "2026-09-22T19:42:16Z", "last_release": "none", "created_at": "2026-09-22T19:40:38Z", "stars": 10, "stars_date": "2026-10-05", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/swotstudio/FlyBridge", "status": "ok", "http_code": 200, "final_url": "https://github.com/swotstudio/FlyBridge", "checked_at": "2026-10-06T09:40:14.366Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/swotstudio/FlyBridge [direct: clone HEAD a8950e0]" ], "notes_limitations": "The network is about 1% of the brain: a filtered smell-pathway subset, with every edge gain, leak and bias trained, plus trained CNN encoders and a trained policy readout. The fly wiring is a constrained recurrent layer inside an ordinary deep-RL/imitation policy. Movement, aiming and clicking come from scripted controllers, and the candidate jobs come from game state supplied by the mod. Shuffled and disconnected variants exist in code, but we found no stored result comparing them for the farming finale, so the circuit has not been shown to be needed. The README states most of these limits openly. Not a copy of an already catalogued project; circuit data are mirrored from snedea/flybrain." }, { "id": "flycns-tictactoe", "name": "FlyCNS Tic-Tac-Toe", "type": "demo-game", "author_or_org": "50RISHU", "summary": "A tic-tac-toe agent that pulls a small signed sub-circuit (descending neurons and the interneurons that drive them) from MaleCNS v1.0 through neuPrint, spreads activation through it with a leaky integrator, and scores each empty cell from a hand-assigned neuron group. By default a minimax solver first restricts the choice to optimal moves.", "claim": { "text": "A tic-tac-toe agent driven by a small circuit pulled out of the male Drosophila CNS connectome", "url": "https://github.com/50RISHU/FlyCNS-TicTacToe" }, "dataset": "MaleCNS", "release": "male-cns:v1.0 via neuPrint (sub-circuit of about 100 neurons)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "Small signed MaleCNS v1.0 sub-circuit around descending neurons (about 100 neurons), weights normalised by the maximum", "neuron_model": "rate: leaky integrator (the README calls it intentionally simpler than a spiking model)", "input_mapping": "hand-set: board cells -> injected activation on chosen neurons", "output_mapping": "hand-set: mean activation of a neuron group per cell", "trained_parts": "none", "body": "none", "scripted_parts": "minimax solver (unbeatable=True by default) limits moves to optimal ones; the circuit only breaks ties" }, "trained_class": "none", "grade_basis": [ "src/flycns/simulate.py:1-3,31-37 leaky integrator on the signed core graph, weights / max weight [direct]", "src/flycns/agent.py:31,49,87-89 unbeatable defaults to True: candidates = optimal_moves(board) from minimax, the circuit scores choose among them [direct]", "README.md: board-to-neuron mapping is a deliberate engineering convention; \"not a fly brain that plays games\" [direct]" ], "grade_date": "2026-09-30", "measured_result": "No measured result or control in the repository.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/50RISHU/FlyCNS-TicTacToe", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit c3cb9da47b; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS", "Windows" ], "platform_note": "Python command line", "gpu": "none", "download_size": "small; needs a neuPrint token to fetch the sub-circuit", "last_commit": { "date": "2026-09-28T11:22:51+05:30", "hash": "c3cb9da47bafb9e71ca8c516b7a86f2e1b17e875", "branch": "main" }, "pushed_at": "2026-09-28T05:52:58Z", "last_release": "none", "created_at": "2026-09-23T07:06:34Z", "stars": 2, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/50RISHU/FlyCNS-TicTacToe", "status": "ok", "http_code": 200, "final_url": "https://github.com/50RISHU/FlyCNS-TicTacToe", "checked_at": "2026-10-06T09:40:14.607Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/50RISHU/FlyCNS-TicTacToe [direct: clone HEAD c3cb9da, 2026-09-30]", "https://news.ycombinator.com/item?id=49888926 [direct, 2026-09-29]" ], "notes_limitations": "By default the minimax solver decides which moves are allowed, so the play quality is ordinary game code and the fly circuit only chooses among equally good moves. The circuit is a small hand-chosen piece of MaleCNS run as a simple rate model. The README says this plainly." }, { "id": "flycraft", "name": "FlyCraft", "type": "demo-game", "author_or_org": "jjedwards2081", "summary": "Runs the Eon Systems fly-brain PyTorch LIF model of FlyWire v783 (138,639 neurons) on the GPU and connects it by websocket to the Minecraft Education Agent. Senses and code-computed 'drives' are injected as Poisson input, and the most active motor pool picks the Agent command.", "claim": { "text": "A whole Drosophila brain playing Minecraft Education. ... No behaviour is scripted.", "url": "https://github.com/jjedwards2081/FlyCraft" }, "dataset": "FlyWire FAFB", "release": "v783", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 whole brain (138,639 neurons) through the eonsystemspbc/fly-brain submodule", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none (calibrate.py picks the feeding motor neurons most driven by sugar and records baselines)", "body": "custom", "scripted_parts": "body.py computes navigation 'drives' (forward onto DNp09, seek/home/heading onto DNa01/DNa02, landing onto MDN) from game state. The readout uses the same DNp09, DNa01/02 and MDN pools, so these decisions pass almost straight through the network. Sleep and heading memory live in Python. Winner-take-all picks discrete Agent commands, and 'feed' runs a fixed destroy-and-collect sequence." }, "trained_class": "none", "trained_class_note": "calibration picks readout neurons", "grade_basis": [ "flyminecraft/brain.py:1-27 steps the whole-FlyWire LIF from the fly-brain submodule [direct]", "flyminecraft/neurons.py:48-64 vs 66-74: drive groups (DNp09, DNa01/DNa02, MDN) are the same cell types as the motor readout groups [direct]", "flyminecraft/body.py:214-244 stimulus(): code-decided forward, homing, seek, goal and landing drives injected as rates [direct]", "flyminecraft/body.py:325-337 winner-take-all over motor pools gives the Agent command [direct]", "README.md:14 'No behaviour is scripted.' vs README sections 'Why there are drives' and 'Searching new ground' [direct]" ], "grade_date": "2026-09-28", "measured_result": "README 'Searching new ground', against a scripted game stand-in: a 150 s run with nothing to seek walked 141 steps over 136 distinct cells. Seeking sand took 12 steps / 10 blocks out, vs 24 steps / 13 blocks without the heading drive. This ablates a scripted drive, not the connectome.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/jjedwards2081/FlyCraft", "code_licence": "GPL-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 4b9f021f74; GitHub API spdx_id=GPL-2.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "Windows" ], "gpu": "required", "download_size": "~600 MB weight cache built on first run (README) plus CUDA PyTorch", "last_commit": { "date": "2026-09-16T21:30:03+01:00", "hash": "4b9f021f746611de18460fe1aa2a4632ccc701b7", "branch": "main" }, "pushed_at": "2026-09-16T20:30:06Z", "last_release": "none", "created_at": "2026-09-14T22:07:48Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/jjedwards2081/FlyCraft", "status": "ok", "http_code": 200, "final_url": "https://github.com/jjedwards2081/FlyCraft", "checked_at": "2026-10-06T09:40:14.667Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/jjedwards2081/FlyCraft [direct: local clone, HEAD 4b9f021]", "https://github.com/eonsystemspbc/fly-brain (submodule; not fetched)" ], "notes_limitations": "This is not a MakeCode script: a real whole-brain FlyWire v783 LIF runs. But the headline 'No behaviour is scripted' conflicts with the code and the author's own README. Walking, target seeking, homing, exploring and sleep are decided in Python and pushed into the same descending neurons that are read out, so for navigation the connectome mostly relays the decision. Feeding (sugar to feeding motor neurons) and the giant-fibre escape climb are genuinely routed through the network. The fly-brain submodule (eonsystemspbc/fly-brain, gitlink a3db62f) was not in the clone, so it was not inspected here. The licence is GPL-2.0." }, { "id": "flyhard", "name": "Flyhard (The Driving Fly)", "type": "demo-game", "author_or_org": "Mark Unthank", "summary": "A recurrent rate network is built on the wiring of 165,122 traced MaleCNS v1.0 neurons. All its edge gains are trained to copy an inverse-kinematics teacher. It moves a NeuroMechFly foreleg that turns a steering wheel in MuJoCo, and the wheel angle steers a car in CARLA.", "claim": { "text": "A connectome-based fruit fly physically learning to operate a steering wheel in CARLA", "url": "https://github.com/MarkUnthank/flyhard" }, "dataset": "MaleCNS", "release": "v1.0 (minconf-0.5 flat tables, traced neurons only)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 topology: 165,122 traced neurons, 25,563,197 edges; unsigned (transmitter predictions not used)", "neuron_model": "rate", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "All ~25.7 M edge gains plus per-neuron leaks, trained by behaviour cloning of an offline IK teacher (600 Adam steps, one seed). Input and output maps are frozen random projections (10 state features to VNC sensory neurons; VNC motor neurons to 7 foreleg joints).", "body": "FlyGym", "scripted_parts": "Throttle/speed and turn requests in the CARLA videos are scripted. There is an engineered forefoot grip and a passive right-leg grip. In parking and three-point turns (code not in HEAD), velocity regulation and IK are fixed." }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "all edge gains trained by behaviour cloning", "grade_basis": [ "scripts/acquire_connectome.py:2,10-14,32 downloads pinned MaleCNS v1.0 flat tables with hash checks [direct]", "src/flyhard/connectome.py:1-5,44-79 tanh rate model on fixed adjacency with trainable edge_gain and leak [direct]", "src/flyhard/motor_policy.py:8-43 frozen random sensory and motor projections [direct]", "scripts/train_wheel.py:57-98 IK teacher demonstrations and MSE behaviour cloning [direct]", "src/flyhard/cockpit.py:1,11-18 NeuroMechFly/FlyGym body with passive wheel; scripts/capture_carla_cockpit.py:38,206-215 steer = gain x wheel angle, scripted throttle [direct]", "reports/2026-09-09/e03-wheel-pilot/metrics.json initial_successes 0, trained_successes 100 [direct]" ], "grade_date": "2026-09-28", "measured_result": "Wheel steering: 100/100 held-out targets after training vs 0/100 for the untrained core (reports/2026-09-09/e03-wheel-pilot/metrics.json). Disabling the grip removes >99% of the steering response (reports/2026-09-09-carla-video/validation.json). Parallel parking: 0/50 successes for both learned and reset core, mean final error 3.71 m vs 18.31 m (apps/website/public/press/data/parking-notes.md). Three-point turn: 6/8 held-out passes vs 0/2 for the reset core.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/MarkUnthank/flyhard", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 328906f4a0; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux" ], "gpu": "required", "download_size": "~1.1 GB MaleCNS weights table plus CARLA 0.9.16 runtime", "last_commit": { "date": "2026-09-15T16:55:53+02:00", "hash": "328906f4a0e62c8f9fc18805cf6edae6989b82a5", "branch": "main" }, "pushed_at": "2026-09-15T14:55:53Z", "last_release": "none", "created_at": "2026-09-09T19:50:12Z", "stars": 83, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/MarkUnthank/flyhard", "status": "ok", "http_code": 200, "final_url": "https://github.com/MarkUnthank/flyhard", "checked_at": "2026-10-06T09:40:14.603Z" }, "last_verified": "2026-10-06", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/MarkUnthank/flyhard [direct: local clone, HEAD 328906f]" ], "notes_limitations": "The controls compare trained vs untrained gains on the same wiring. They show that training works, not that connectome wiring matters. There is no shuffled-wiring or no-graph baseline and no real-fly comparison, so the grade is B, not A. The whole network is fitted by behaviour cloning, so it is a large trained model that uses connectome topology, with random frozen I/O maps (trained parts flagged). Transmitter signs are not used, and there is one training seed. Driving has no vision: the fly controls only steering, from scripted turn requests. The famous 'parallel parking, 50 attempts' video shows 0/50 successes, and its side-panel fly is an unsynchronised replay. The parking and three-point-turn code is not in HEAD, only result files and press notes. Most of the repo is website and livery assets. Licence MIT." }, { "id": "flykart", "name": "FlyKart", "type": "demo-game", "author_or_org": "ZENinjaneer", "summary": "A browser go-kart game run by a local Python server. The whole MaleCNS v1.0 connectome (traced neurons only) is simulated as a Shiu-style leaky integrate-and-fire network on the GPU. The game computes where the target truck, looming obstacles and sugar are and stimulates chosen cell types (LC10a for the truck, looming and taste neurons); spike rates of DNa02/DNa01, oDN1/BDN2, MDN, the giant fiber and MN9 are mapped by hand to steering, throttle, reverse, jump and feeding. A 3-D view shows the neurons firing.", "claim": { "text": "A real fruit fly's brain drives a go-kart, and you can watch it think; nobody trained anything, the behaviour comes from the wiring.", "url": "https://github.com/ZENinjaneer/flykart" }, "dataset": "MaleCNS", "release": "v1.0 flat-connectome feather files (minconf 0.5) from the official FlyEM bucket; traced neurons only: 165,122 neurons, 25.6 M connections", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "Whole MaleCNS v1.0, traced neurons, signed by predicted neurotransmitter", "neuron_model": "spiking: leaky integrate-and-fire with the Shiu et al. equations and parameters, on GPU (PyTorch)", "input_mapping": "hand-set: truck azimuth computed by the game drives LC10a neurons with a sigmoid receptive field; looming channel gated by computed time to contact; taste neurons for sugar; a user 'drive' slider stimulates oDN1 (DNg97) at up to 300 Hz", "output_mapping": "hand-set: steer = (DNa02+0.5 DNa01 right - left)/60 Hz; throttle from oDN1+BDN2; reverse if MDN > 5 Hz; jump on any giant-fiber spike; feed from MN9", "trained_parts": "none", "body": "kart physics in Python (no fly body)", "scripted_parts": "Scene analysis (truck direction, collision course) is done by game code before reaching the neurons. Forward drive is set by the user's slider stimulating oDN1. An optional autopilot blend exists (0% by default)." }, "trained_class": "none", "grade_basis": [ "flykart/connectome.py:28-32,111-135 downloads MaleCNS v1.0 flat-connectome feathers; keeps status 'Traced' neurons [direct]", "flykart/brain.py:1-10,41-51 Shiu LIF equations and parameters (w_syn 0.275 mV, v_th -45 mV) [direct]", "flykart/interface.py:1-17,151-169 LC10a pursuit input, looming, taste, oDN1 driven at 300 Hz * drive slider [direct]", "flykart/world.py:240-262 looming input gated by computed time to contact [direct]", "flykart/interface.py:176-190 decode: steer from DNa02/DNa01 L-R, throttle from oDN1/BDN2, reverse MDN > 5 Hz, jump on GF spike [direct]", "flykart/world.py:330-336 and flykart/server.py:42 autopilot blend 'assist' defaults to 0.0 [direct]", "Re-opened by the runner at HEAD 82d88a1 (2026-09-29T19:20): flykart/connectome.py:28-32 MaleCNS v1.0 flat-connectome feathers; flykart/brain.py:41-51 v_th -45 mV, w_syn 0.275 mV; flykart/interface.py:176-190 hand-set decode (steer from DNa02/DNa01 L-R /steer_hz, throttle (oDN1+0.5 BDN2)/180, reverse MDN > 5 Hz, jump on any GF spike); flykart/world.py:330-336 and server.py:42 assist 0.0 by default. The new commit adds a CPU kernel (brain.py, cpu_kernels.py, cpu_random.py) and does not change the mapping [direct]" ], "grade_date": "2026-09-30", "measured_result": "No measured result or control in the repository.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/ZENinjaneer/flykart", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 82d88a13ef; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS", "Windows" ], "platform_note": "local Python server opened in a browser; Windows via WSL2; macOS runs on CPU in slow motion", "gpu": "optional", "gpu_note": "NVIDIA GPU for real time; since 2026-09-29 an accelerated CPU path (docs/cpu-performance.md) runs without one, more slowly", "download_size": "about 1.2 GB connectome download; about 8 GB disk including PyTorch/CUDA", "last_commit": { "date": "2026-09-29T19:20:59-04:00", "hash": "82d88a13ef2b5133b9d8812413435971734e5599", "branch": "main" }, "pushed_at": "2026-09-29T23:26:19Z", "last_release": "none", "created_at": "2026-09-28T22:06:51Z", "stars": 0, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/ZENinjaneer/flykart", "status": "ok", "http_code": 200, "final_url": "https://github.com/ZENinjaneer/flykart", "checked_at": "2026-10-06T09:40:14.838Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/ZENinjaneer/flykart [direct: clone HEAD 82d88a1, 2026-09-30]" ], "notes_limitations": "The whole connectome is simulated with no training, but the game does the seeing: it computes the truck's direction and collision timing and injects them into chosen cell types, and output rates are mapped to kart controls by hand-set thresholds and gains. Forward speed follows a slider that stimulates the forward-walking neuron directly. There is no control run (for example scrambled wiring), so how much of the chasing comes from connectome routing rather than the chosen input/output cells is not shown. The README's 'its eyes see the game' overstates the input side." }, { "id": "flypoker", "name": "flypoker", "type": "demo-game", "author_or_org": "0909-BoB", "summary": "A no-limit Texas Hold'em table in the browser (served by a local Python server) where three 'flies' play. Each fly is a 6,194-neuron spiking subgraph of MaleCNS v1.0 (seed cell types per input role, at least 8 synapses) whose input pools receive hand-made game features, including a Monte Carlo hand equity computed outside the brain; a linear softmax decoder over about 90 descending-neuron spike counts, cloned from human hand histories, chooses the action.", "claim": { "text": "\"A subgraph of a real fly brain (MaleCNS v1.0 connectome, Janelia/Google, 2026) plays no-limit Texas Hold'em, via a spiking neural network simulation with a small trainable readout layer on top.\" (README)", "url": "https://github.com/0909-BoB/flypoker" }, "dataset": "MaleCNS", "release": "v1.0 (neuPrint male-cns:v1.0), 6,194-neuron subgraph, min 8 synapses (data/circuit_meta.json, data/circuit.npz)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 subgraph of 6,194 neurons: input pools seeded from chosen cell types (for example LC4/LPLC1/LPLC2 for equity, LC9/LC12/LC17 for pot odds, JO for opponent aggression), 4,000 relay neurons and 90 descending output neurons (DNg types)", "neuron_model": "spiking (src/flypoker/brain.py)", "input_mapping": "hand-made: equity (Monte Carlo, computed outside the brain, gain 8), pot odds, stack depth, street, position and opponent aggression drive chosen populations", "output_mapping": "learned: linear softmax decoder over about 90 output-neuron spike counts, with a 12% uniform floor on every action", "trained_parts": "the decoder only: behaviour-cloned from public hand histories (pretrain.py); self-play REINFORCE collapsed and was not used for the shipped weights (README)", "body": "none", "scripted_parts": "poker rules, Monte Carlo equity and the rule-based EquityBot opponent are ordinary code" }, "trained_class": "readout-or-decoder", "trained_class_note": "decoder behaviour-cloned", "grade_basis": [ "data/circuit_meta.json at e1ed8f6: source male-cns:v1.0, min_synapses 8, seed_types_by_role, 6,194 nodes (4,000 relay, 90 out:action) [direct]", "src/flypoker/encoder.py:8-66: monte_carlo_equity computed before the brain and injected as in:equity with EQUITY_GAIN 8 ('equity is the channel the decision leans on most') [direct]", "src/flypoker/decoder.py:1-110: SoftmaxDecoder, the only thing that learns; README: shipped runs/decoder.npz is pretrain.py output, fold weights untouched, 12% action floor [direct]", "No wiring control or brain-free baseline result in the repository; synthetic_circuit.py is a placeholder random circuit for development [direct]" ], "grade_date": "2026-10-03", "measured_result": "None reported as a number for the brain's contribution. The README documents training failures (fold-collapse, input-independent collapse) and ships behaviour-cloned decoder weights instead of self-play ones.", "try_url": null, "try_status": "no no-install option (Docker or Python 3.10+ local server on port 8420)", "code_url": "https://github.com/0909-BoB/flypoker", "code_licence": "none found", "code_licence_source": "no LICENSE file at e1ed8f6; GitHub API reports none", "data_licence": "MaleCNS v1.0 derived data committed (CC BY 4.0 per Janelia; not stated in the repository)", "platform": [ "Linux", "macOS", "Windows" ], "platform_note": "local server: Docker, or Python 3.10+ with pip install -e .", "gpu": "none", "download_size": "repository about 3 MB of data (circuit.npz 1.6 MB uncompressed arrays, brain.stl mesh)", "last_commit": { "date": "2026-10-01T18:53:49+08:00", "hash": "e1ed8f6129382a2adbf8fd0fae755f8828d98cc3", "branch": "main" }, "pushed_at": "2026-10-01T10:53:52Z", "last_release": "none", "created_at": "2026-10-01T08:01:13Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/0909-BoB/flypoker", "status": "ok", "http_code": 200, "final_url": "https://github.com/0909-BoB/flypoker", "checked_at": "2026-10-06T09:40:15.756Z" }, "last_verified": "2026-10-06", "added_in_run": "run:5bfca8cc-f3cb-4020-83e6-5c1225da9200", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/0909-BoB/flypoker [direct]" ], "notes_limitations": "The poker decisions come mostly from a hand equity computed outside the brain and a decoder cloned from human players; the connectome subgraph sits in between as a feature generator, and no run compares it with a scrambled or no-brain version. The README is candid about the training failures. No licence file." }, { "id": "flywire-gba", "name": "FlyWireGBA", "type": "demo-game", "author_or_org": "LakoMoor", "summary": "A Game Boy Advance ROM with a flybody-derived animated fly, a Tamagotchi care mode and a 128-neuron, 2,048-edge subgraph of FlyWire v783 around the Shiu et al. sugar neurons and MN9, run as an integer LIF network at 30 ticks per second. Sugar contact drives the sugar neurons; MN9 spikes gate feeding. Walking, flight, grooming and obstacle avoidance come from a heuristic controller.", "claim": { "text": "\"A real reduced FlyWire FAFB v783 network: 128 neurons and 2,048 directed signed edges, including 20 surviving sugar sensory IDs and one MN9\" and \"A compact heuristic controller handles locomotion, flight, grooming, and energy. The extracted neural circuit controls the feeding gate through MN9.\" (README)", "url": "https://github.com/LakoMoor/FlyWireGBA" }, "dataset": "FlyWire FAFB", "release": "v783 as packaged by Shiu et al. 2024 (Connectivity_783 parquet; source sha256 recorded in docs/connectome.json): 128 neurons, the 2,048 strongest induced edges, weights sign(w)*round(sqrt(|w|)*7) clipped to +/-96", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 subgraph of 128 neurons (20 sugar GRNs, 1 MN9 and 107 neurons chosen by 3-hop relevance), 2,048 signed edges", "neuron_model": "other: integer LIF, 33.3 ms ticks, threshold 100, leak 1/8, one-tick refractory (src/sim.c; the README states it does not reproduce the Shiu model)", "input_mapping": "hand-made: ground contact with sugar drives the 20 sugar neurons (drive = stimulus x 3)", "output_mapping": "hand-made: MN9 spikes gate food intake; nothing else", "trained_parts": "none", "body": "flybody-derived sprites rendered offline (Apache-2.0 meshes); no physics", "scripted_parts": "locomotion, flight, grooming, obstacle avoidance, energy and all pet needs are a heuristic controller and game rules" }, "trained_class": "none", "grade_basis": [ "tools/import_connectome.py at 0ce81d9: reads the Shiu sugar list and MN9, selects 128 nodes by 3-hop relevance, keeps the 2,048 strongest edges; metadata says 'Illustrative integer LIF, 30 Hz; not the calibrated Shiu model' and 'heuristic locomotion; MN9 gates feeding' [direct]", "src/sim.c lines 18-38 and 105: integer LIF update, sugar-only drive into group 0, MN9 (group 2) spike count gates feeding [direct]", "docs/validation.md: ROM built and checked in mGBA 0.10.5, including a sugar-contact feeding test [direct; not run by us]" ], "grade_date": "2026-10-05", "measured_result": "None (a game; the author's emulator check confirms that sugar contact leads to circuit-driven intake).", "try_url": null, "try_status": "no browser option; the README links a .gba ROM on the repository's GitHub Releases page (built by CI, SHA-256 published) for the mGBA emulator or a flash cartridge; not run by us", "code_url": "https://github.com/LakoMoor/FlyWireGBA", "code_licence": "MIT", "code_licence_source": "LICENSE file at 0ce81d9 (MIT, 2026 LakoMoor); the API reports NOASSERTION; flybody meshes under licenses/flybody-Apache-2.0.txt", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0 (the repository's import metadata also says CC BY-NC 4.0)", "platform": [ "other" ], "platform_note": "Game Boy Advance ROM: mGBA emulator (any desktop OS) or a GBA flash cartridge", "gpu": "none", "download_size": "ROM 16 MiB (16,777,216 bytes, docs/validation.md)", "last_commit": "n/a", "pushed_at": "n/a", "last_release": { "tag": "FlyWireGBA v0.2.1", "date": "2026-10-04T18:20:01Z", "url": "https://github.com/LakoMoor/FlyWireGBA/releases/tag/v0.2.1" }, "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/LakoMoor/FlyWireGBA", "status": "ok", "http_code": 200, "final_url": "https://github.com/LakoMoor/FlyWireGBA", "checked_at": "2026-10-06T09:40:14.697Z" }, "last_verified": "2026-10-06", "added_in_run": "run:3fed5f73-77cc-4b37-853f-e1e8258b4f9f", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/LakoMoor/FlyWireGBA [direct]" ], "notes_limitations": "Only the feeding gate is connectome-driven, by a 128-neuron subgraph with quantised weights and a toy integer LIF; everything the fly visibly does besides eating is a heuristic controller, and the neural map uses a synthetic layout. The ROM is the author's own CI build (an original project, not a copy of another repository); we did not run it." }, { "id": "fruitfly-roblox", "name": "Fruit fly vs Jev: Roblox maze race", "type": "demo-game", "author_or_org": "SolidifiedPlayDoh", "summary": "A Roblox maze race between a character driven by a spiking MaleCNS brain model and one driven by a hosted language model. Roblox sends raycast distances, object lists and an apple 'smell' to a local Python server. Spike rates of a few descending-neuron types are turned into forward and turn commands by hand-set formulas, then blended with a scripted steering routine that picks open space toward the apple and remembers visited cells. The scripted part carries most of the weight in the maze.", "claim": { "text": "MaleCNS v1.0, the published connectome of a male fruit fly, races an LLM through a Roblox maze to an apple; descending neurons do the moving (video 0zY_yzrcF9c).", "url": "https://github.com/SolidifiedPlayDoh/fruitfly-roblox" }, "dataset": "MaleCNS", "release": "v1.0 as stated, loaded through the third-party 'flybrain' Python package (weights not in repo; release not verified in code)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "Whole MaleCNS as supplied by the flybrain package (stated 166,700 neurons), frozen", "neuron_model": "spiking, inside the flybrain package (not inspectable in this repository)", "input_mapping": "hand-set: game objects are sorted into fruit/threat/other by name and passed to the package's FeatureDetectors, which inject into visual feature-detector cells", "output_mapping": "hand-set: forward = tanh(8*(DNg100-MDN)+5*DN mean), turn = tanh(10*(DNa02 R-L)+5*(DN R-L)), jump from DNp01", "trained_parts": "A ridge readout is fitted to a scripted teacher during play and saved, but it is not used to produce the action", "body": "Roblox humanoid rig", "scripted_parts": "Output is 0.35 brain + 0.65 scripted steering whenever the distance ring is present, 0.2 + 0.8 near walls; the scripted steering chooses open space toward the apple smell and penalises visited cells. Rest/walk bouts are chosen by code with random timers. A Lua bumper reverses near walls. The LLM opponent is ordinary API code." }, "trained_class": "none", "trained_class_note": "ridge readout fitted but not used for the action", "grade_basis": [ "src/fruitfly_roblox/malecns.py:278-284 FlyBrain() and FeatureDetectors from the flybrain package; DNg100, MDN, DNa02, DNp01 cell groups [direct]", "src/fruitfly_roblox/malecns.py:603-617 8 brain steps per request, spike rates of DN groups [direct]", "src/fruitfly_roblox/malecns.py:652-661 hand-set tanh mapping from DN rates to forward/turn/jump [direct]", "src/fruitfly_roblox/malecns.py:620-641 scripted teacher_action and _steer_from_ring (open space toward apple) [direct]", "src/fruitfly_roblox/malecns.py:674-678 raw = 0.35*native + 0.65*taught with ring; 0.2/0.8 near walls [direct]", "src/fruitfly_roblox/malecns.py:425-455 visited-cell memory penalises revisits (maze solving in code) [direct]", "src/fruitfly_roblox/malecns.py:669,697,703-731 readout fitted in background but mix=0.0 and never applied to the action [direct]", "src/fruitfly_roblox/malecns.py:402-423 rest/walk bouts from random timers [direct]", "studio/FlyBrainServer.server.lua:471-490 applies forward/turn; bumper raycast forces reverse near walls [direct]" ], "grade_date": "2026-09-29", "measured_result": "No measured result in the repository; the video shows one race.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/SolidifiedPlayDoh/fruitfly-roblox", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 7943476bb0; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Windows", "macOS" ], "platform_note": "Python server plus Roblox Studio; the Studio place file is not included, only the scripts", "gpu": "none", "gpu_note": "the brain runs on CPU (device='cpu')", "download_size": "about 260 MB of connectome data downloaded by the flybrain package on first launch; LLM opponent needs a paid OpenRouter key", "last_commit": { "date": "2026-09-28T20:48:06-07:00", "hash": "7943476bb0729ad4c6e5faf10f860ba6c357e734", "branch": "main" }, "pushed_at": "2026-09-29T03:48:15Z", "last_release": "none", "created_at": "2026-09-29T03:48:11Z", "stars": 0, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/SolidifiedPlayDoh/fruitfly-roblox", "status": "ok", "http_code": 200, "final_url": "https://github.com/SolidifiedPlayDoh/fruitfly-roblox", "checked_at": "2026-10-06T09:40:15.155Z" }, "last_verified": "2026-10-06", "added_in_run": "run:4018cb43-5949-4ecf-911a-0366c3d032ab", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/SolidifiedPlayDoh/fruitfly-roblox [direct: clone HEAD 7943476, 2026-09-28]", "https://youtu.be/0zY_yzrcF9c [not viewed]" ], "notes_limitations": "The maze is mainly solved by code: whenever the distance ring is available, 65% of the command (80% near walls) comes from a scripted steering routine that seeks open space toward the apple and avoids cells already visited. The brain's share is a hand-set formula over a few descending-neuron spike rates. A readout is trained during play but not used. The simulator is a third-party package whose model we did not inspect, and the Roblox place file is not published. No control (for example a run with the brain share removed or rewired). The README presents the race as brain versus LLM without this split." }, { "id": "kick-the-fly", "name": "Kick the Fly", "type": "demo-game", "author_or_org": "legendarylolo318-cloud", "summary": "A Kick-the-Buddy style game in which each fly runs a live LIF simulation of the filtered MaleCNS v1.0 graph (about 166,700 neurons). Player actions drive real sensory neuron groups. When descending-neuron groups rise above thresholds relative to their calm baseline, they trigger game-physics reactions such as jump, run, kick and fly, and the game labels each reaction REAL or RULE. It ships a validation suite that tests circuit effects against matched control neuron sets.", "claim": { "text": "Kick-the-buddy game where the buddy is a live 166,700-neuron connectome", "url": "https://github.com/legendarylolo318-cloud/kick-the-fly" }, "dataset": "MaleCNS", "release": "v1.0 flat connectome (neurons with non-glia superclass, edges with at least 3 synapses; about 166,700 neurons)", "evidence_grade": "C", "grade_note": "Borderline C: the neural part alone could argue for B, but the behaviour shown is triggered, scripted motion. The file citations behind this grade were spot-read by us, not re-verified line by line.", "mechanism": { "wiring": "MaleCNS v1.0 flat-connectome tables downloaded from the flyem-male-cns bucket; filtered by superclass and weight >= 3, signed by predicted neurotransmitter", "neuron_model": "LIF, rate-normalised input, global gain controller", "input_mapping": "hand-made: tools drive named touch, heat/cold, ORN, taste, JO, looming populations", "output_mapping": "hand-made: DN group rate / calm-baseline thresholds (THRESH dict) trigger reactions", "trained_parts": "optional dopamine-gated plasticity on mushroom-body synapses (training mode)", "body": "2D/3D cartoon ragdoll", "scripted_parts": "All movement is game physics. The direction of jumps and runs (away from the hit), stunning, damage, eating, flight paths, drunkenness, death, pain index and several reactions are game rules, which the code labels RULE." }, "trained_class": "none", "trained_class_note": "optional plasticity in training mode", "grade_basis": [ "kickthefly/sim/connectome/loader.py:1-40, 202-300 downloads MaleCNS v1.0 flat-connectome, filters and signs edges (MIN_WEIGHT = 3 at :208) [direct]", "kickthefly/sim/connectome/sim.py:1-40 LIF over the signed CSR connectome [direct]", "kickthefly/game/kick_the_fly.py:407 THRESH dict (jump 3.0, run 2.4, kick 2.0, fly 1.58 ... x calm) gates scripted reactions [direct]", "kickthefly/game/kick_the_fly.py:169-187, 1794-1816 states ragdoll physics, direction of jumps and runs, and movement are game rules; REAL/RULE tagging [direct]", "kickthefly/lab/validation.py and docs/validation.md: driven vs matched control sets, seeds 1000-1009, Wilcoxon tests; some pass (looming to DNp01, sugar to MN9) and some fail (MDN backward walking) [direct]", "Re-grade 2026-09-29 after release 2.12.0 (commit a125c32): kickthefly/sim/connectome/larva_loader.py:1-35 adds the Winding et al. 2023 larval connectome (2,952 neurons) as a headless-only option; docs/larva.md: 'The windowed game (2D and 3D) always runs the adult brain', and both larva validation tests FAIL ('no excitation'); kickthefly/core/individuality.py adds per-fly parameter variation. The game's reactions are still triggered scripted programs, so the grade stays C [direct]" ], "grade_date": "2026-09-29", "measured_result": "Measured by the author, not reproduced by us. Circuit-level validation (n = 10 seeds): looming detectors drive DNp01 x11.8 vs x0.80 for controls (pass); sugar neurons drive MN9 x2.11 vs x1.25 for bitter (pass); JO-C/E drive aDN1/aDN2 x4.87 vs x0.85 (pass); MDN activation does not drive leg motor neurons (fail); aDN drives front-leg MNs only x1.13 (fail) (docs/validation.md). Release 2.12.0 (29 Sep 2026) adds a headless larva-brain mode whose two validation tests fail: nociceptive and chordotonal inputs do not excite their target neurons more than random sensory inputs (docs/larva.md).", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/legendarylolo318-cloud/kick-the-fly", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 2fd9e5b251; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "Windows", "Linux" ], "gpu": "none", "gpu_note": "none (NumPy; optional Numba/torch)", "download_size": "about 90 MB (Windows exe) / 110 MB (Linux AppImage), plus connectome tables on first build", "last_commit": { "date": "2026-10-06T01:09:33-07:00", "hash": "2fd9e5b251268160ee91735cf098c8bfe2af6a4f", "branch": "main" }, "pushed_at": "2026-09-28T06:55:18Z", "last_release": { "tag": "Kick the Fly 2.13.1: tool loadouts, a self-test and a bug report that sends nothing", "date": "2026-09-30T07:00:12Z", "url": "https://github.com/legendarylolo318-cloud/kick-the-fly/releases/tag/v2.13.1" }, "created_at": "2026-09-13T11:36:54Z", "stars": 5, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/legendarylolo318-cloud/kick-the-fly", "status": "ok", "http_code": 200, "final_url": "https://github.com/legendarylolo318-cloud/kick-the-fly", "checked_at": "2026-10-06T09:40:15.113Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/legendarylolo318-cloud/kick-the-fly [direct: clone HEAD d1f504f]" ], "notes_limitations": "The whole filtered graph is simulated and its activity decides when reactions fire. The reactions themselves are scripted motor programs with game physics, so the behaviour shown is connectome-triggered animation. The validation suite is an honest circuit-level check against matched controls, with published failures, but it tests neuron responses, not the game behaviour. The thresholds were chosen after exploratory probing. The code discloses what is a game rule and what is not. Since 2.12.0 a larval connectome can be loaded for headless tests only; the playable game always uses the adult brain, and the larva validation tests fail, so larva reactions would be game rules." }, { "id": "making-fly-play-chess", "name": "making-fly-play-chess", "type": "demo-game", "author_or_org": "mncrftfrcnm", "summary": "A chess player that turns the board into 782 numbers, pushes them for six steps through a fixed 8,192-neuron patch of the FlyWire v783 wiring (tanh rate units), and scores each legal move with a linear readout trained by self-play on 1,024 randomly chosen cells. A Gradio board lets you play it in Colab, and an optional FlyGym view moves a 3-D fly's joints from readout-weighted activity. A comparison notebook trains the same setup on rewired wiring and on classical features.", "claim": { "text": "A chess agent built from the connectivity of the fruit-fly brain, used as a fixed neural reservoir and trained with reinforcement learning.", "url": "https://github.com/mncrftfrcnm/making-fly-play-chess" }, "dataset": "FlyWire FAFB", "release": "v783 (Shiu et al. Connectivity_783.parquet); 8,192-neuron breadth-first patch around the highest-degree neuron, 168,930 edges", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "8,192 neurons chosen by breadth-first search from the highest-degree neuron; signed synapse counts, row-normalised, fixed", "neuron_model": "rate: tanh units, 6 leaky update steps per position", "input_mapping": "hand-set: 782 board features written directly onto the first 782 patch neurons", "output_mapping": "learned: linear readout on 1,024 random patch neurons gives a position score; python-chess lists legal moves and the best-scoring one is played", "trained_parts": "Readout weights only (self-play TD-style updates with a small material reward)", "body": "optional FlyGym fly animation", "scripted_parts": "Legal move generation, search over one ply and game rules are ordinary code. The FlyGym joint angles are readout-weighted activity split into joint groups, an animation, not motor control." }, "trained_class": "readout-or-decoder", "trained_class_note": "readout weights by self-play", "grade_basis": [ "scripts/fly_chess_trainer.py:18-19,72-76 Shiu Completeness_783 / Connectivity_783 files [direct]", "scripts/fly_chess_trainer.py:69,90-100 RESERVOIR_NEURONS = 8192, breadth_first_order from the highest-degree neuron [direct]", "scripts/fly_chess_trainer.py:117-121 1,024 readout neurons chosen at random [direct]", "scripts/fly_chess_trainer.py:127-141 six tanh propagation steps; features = state[readout_indices] [direct]", "scripts/fly_chess_trainer.py:189-221 value = tanh(value_weights @ features) over legal moves [direct]", "scripts/fly_chess_inference_flygym.py:115-138 FlyGym joint angles from readout-weighted features split into groups (animation) [direct]", "notebooks/connectome_vs_classical_architectures.ipynb cell 12: rewire_reservoir shuffles postsynaptic rows of the patch [direct]", "notebooks/connectome_vs_classical_architectures.ipynb cells 4, 22, 24: stored run in 'quick' mode ('for code checks only'); fly vs rewired mean score 0.506 (CI 0.406-0.607, 5 seeds, 80 games); fly vs material-count 0.306 [direct]", "Re-opened by the runner at HEAD 4ddc9b5 (unchanged): scripts/fly_chess_trainer.py:69 RESERVOIR_NEURONS = 8192; :117-121 1,024 readout neurons by rng.choice; notebook cell 4 RUN_MODE = 'quick'; cell 24 output 'Quick mode is for code checks only'; 0.506 in the stored results table [direct]" ], "grade_date": "2026-09-30", "measured_result": "Stored comparison notebook (quick mode, 5 seeds, 80 games per pairing, flagged by the author as a code check): the fly patch scored 0.506 against the same patch with rewired connections (no difference), 0.531 against classical features, 0.738 against random moves and 0.306 against a simple material-count player. Measured by the author, not reproduced by us.", "try_url": "https://colab.research.google.com/github/mncrftfrcnm/making-fly-play-chess/blob/main/notebooks/fly_chess_inference.ipynb", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/mncrftfrcnm/making-fly-play-chess", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit b97dfefbf9; GitHub API spdx_id=Apache-2.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "Colab", "Linux", "macOS", "Windows" ], "platform_note": "Python notebooks and scripts; Gradio interface", "gpu": "optional", "download_size": "trained readout included in repo; Shiu connectivity data fetched separately (about 100 MB)", "last_commit": { "date": "2026-10-06T13:03:05+04:00", "hash": "b97dfefbf9b8850f3cdf6b412c2ed840be94e260", "branch": "main" }, "pushed_at": "2026-09-29T04:24:53Z", "last_release": "none", "created_at": "2026-08-20T13:14:43Z", "stars": 7, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://colab.research.google.com/github/mncrftfrcnm/making-fly-play-chess/blob/main/notebooks/fly_chess_inference.ipynb", "status": "ok", "http_code": 200, "final_url": "https://colab.research.google.com/github/mncrftfrcnm/making-fly-play-chess/blob/main/notebooks/fly_chess_inference.ipynb", "checked_at": "2026-10-06T09:40:14.879Z", "other_links": [ { "url": "https://github.com/mncrftfrcnm/making-fly-play-chess", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "wiring-null", "wiring_effect": "no-difference", "sources": [ "https://github.com/mncrftfrcnm/making-fly-play-chess [direct: clone HEAD 4ddc9b5, 2026-09-30]" ], "notes_limitations": "The fly wiring is a fixed random-feature layer: about 6% of the brain, picked by graph distance rather than by circuit, with only a linear readout trained. The author's own rewired-wiring control, run in a quick mode the author says is only for code checks, shows no advantage for the real wiring, and the agent loses to a simple material counter. Move generation is ordinary chess code. The 3-D fly moves by animation, not by motor neurons. The README is open that only the readout is trained." }, { "id": "neurocraft-fly", "name": "NeuroCraft Fly", "type": "demo-game", "author_or_org": "Evan Sinclair Smith (evnsnclr)", "summary": "A Minecraft demo in which a model built on the MaleCNS connectome is said to control a custom fly through hand-chosen readouts and scripted body programs. The public repository has only a landing page, citation files and demo videos. The simulation code has not been released.", "claim": { "text": "run the full retained MaleCNS v1.0 fruit fly connectome, all 166,700 neurons, inside Minecraft, with its simulated neural activity driving a fly's movement", "url": "https://x.com/evnsnclr/status/2095975490708291948" }, "dataset": "MaleCNS", "release": "v1.0 (launch post); README gives only 'retained MaleCNS graph', 166,700 neurons / 25,582,938 edges", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "not inspectable; README self-reports a retained MaleCNS graph of 166,700 neurons and 25,582,938 edges", "neuron_model": "other: not stated, and no code to check", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none in current demo per README; README mentions an earlier trained readout that performed worse than a direct controller", "body": "custom", "scripted_parts": "Self-reported: readouts 'select and modulate scripted body programs'. The approach stimulus uses scripted scene preparation, and the video includes an operator-selected scripted movement tour. Cannot be checked in code." }, "trained_class": "not-assessed", "grade_basis": [ "git ls-tree HEAD: 10 files, none of them source code. CHANGELOG.md, CITATION.bib, CITATION.cff, LICENSE (MIT), README.md (landing page), ROADMAP.md, media/README.md (recording notes), media/manifest.json (checksums/chapters), media/brush-demo.gif (7 s clip), media/neurocraft-demo.mp4 (2:27 video) [direct]", "README.md:33-35: the mod and companion source are 'being prepared', and 'software downloads and runnable assets are not available here yet' [direct]", "README.md:10-18: pipeline is inputs -> modelled activity -> labelled readouts -> 'scripted body programs' -> movement [direct]", "README.md:90-93: the MIT licence covers 'project-owned documentation'; no bulk connectome data [direct]", "media/README.md:24-29: some responses remain under shuffled weights, and baseline cruise moves the body with sensory input switched off [direct]", "Launch post text (via api.fxtwitter.com, 2026-09-04): 'Code and mod coming soon!' [direct]" ], "grade_date": "2026-09-28", "measured_result": "Self-reported and unverifiable without code (README.md:53-58): an earlier trained readout had 95.66x higher raw error than a direct controller. An offline probe found DNp01 changes about 300x the wing-pool changes, and TTMn activity 35.1x baseline. The video includes shuffled-weights and no-input conditions without reported numbers.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/evnsnclr/neurocraft-fly-public", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit d121466b3b; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [], "gpu": "not stated", "download_size": "not stated (only ~32 MB of demo media available)", "last_commit": { "date": "2026-09-06T14:21:05-04:00", "hash": "d121466b3ba2f11498e6da498c74acb061bb00eb", "branch": "main" }, "pushed_at": "2026-09-06T18:21:06Z", "last_release": "none", "created_at": "2026-09-06T17:13:32Z", "stars": 170, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/evnsnclr/neurocraft-fly-public", "status": "ok", "http_code": 200, "final_url": "https://github.com/evnsnclr/neurocraft-fly-public", "checked_at": "2026-10-06T09:40:15.293Z", "other_links": [ { "url": "https://x.com/evnsnclr/status/2095975490708291948", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/evnsnclr/neurocraft-fly-public [direct: local clone, HEAD d121466]", "https://x.com/evnsnclr/status/2095975490708291948 (via https://api.fxtwitter.com/evnsnclr/status/2095975490708291948) [direct]" ], "notes_limitations": "CONFLICT resolved: both directories are partly right. The repo exists and has an MIT LICENSE, but the licence covers only documentation, and the simulation code and mod are not in it (README.md:33-35; the launch post says 'Code and mod coming soon'). flybrain.info's 'code not released' is correct for the code. Fly Brain Hub's 'repo exists, MIT' is literally true but misleading as a code release. The author's own notes are frank: the body moves through scripted programs, some responses survive weight shuffling, and the body moves even with inputs off. So the connectome's share of the visible behaviour is unclear. 166,700 neurons is fewer than the full MaleCNS v1.0 table (176,422 in fly-brain-minecraft's stats), so a filter was applied but is not described. The video was not reviewed frame by frame. Re-check when the code is released." }, { "id": "stonkfly", "name": "Stonkfly", "type": "demo-game", "author_or_org": "nftechie", "summary": "A BTC-USDC price chart is drawn as an RGB image and fed into the photoreceptors of a LIF simulation of the full MaleCNS v1.0 graph. DNp20 left/right rates, gated by DNpe017, give buy, sell or hold. By default it paper-trades $100; live Coinbase orders need opt-in. Profit and loss trigger artificial dopamine pulses for an experimental memory rule.", "claim": { "text": "A fly-connectome simulation that can operate a crypto trading account. Actual neural output, actual Coinbase integration. Profitable learning has not been demonstrated.", "url": "https://github.com/nftechie/stonkfly" }, "dataset": "MaleCNS", "release": "v1.0 (166,700 neurons, 25,582,938 connections)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0, whole retained graph", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "Experimental dopamine-gated plasticity on 7,835 existing KC->MBON07/MBON11 edges. Gain grows by 20 mV-equivalent pulses into 15 PAM11 cells (profit) or 2 PPL101 cells (loss). No trained readout.", "body": "none", "scripted_parts": "none found. Risk limits and cooldowns can veto orders." }, "trained_class": "other", "trained_class_note": "experimental dopamine-gated plasticity", "grade_basis": [ "stonkfly/neural/controller.py:15-17,26-38 fixed decoder: DNp20 right-minus-left mean rate above a threshold, gated by DNpe017 spikes, gives BUY or SELL, else HOLD [direct]", "stonkfly/neural/kernel.cpp:4-22 LIF kernel (20 ms membrane, 5 ms synapse) over every retained edge; docs/model.md:11 gives the parameters [direct]", "docs/model.md:43-47 profit gives a PAM11 pulse, loss gives a PPL101 pulse; the rule acts on 7,835 KC->MBON edges [direct]", "docs/validation.md: 6-observation paper run gave one BUY fill and five BUY proposals vetoed; 'neither successful learning nor a profitable policy' [direct]", "stonkfly/config.py:28,53-56 default capital $100, max order $10 [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/nftechie/stonkfly", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 78ef3e05ab; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "macOS", "Linux" ], "gpu": "none", "download_size": "MaleCNS v1.0 tables, 'several GB' (README); 16 GB RAM recommended", "last_commit": { "date": "2026-09-09T21:39:57-05:00", "hash": "78ef3e05ab0fa086032098558d893667068944a0", "branch": "main" }, "pushed_at": "2026-09-10T02:40:34Z", "last_release": "none", "created_at": "2026-09-10T02:40:30Z", "stars": 844, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/nftechie/stonkfly", "status": "ok", "http_code": 200, "final_url": "https://github.com/nftechie/stonkfly", "checked_at": "2026-10-06T09:40:15.493Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/nftechie/stonkfly [direct: local clone, HEAD 78ef3e0]" ], "notes_limitations": "The default is paper trading on real public prices, and live trading needs the user's own Coinbase key. The only reported run was six observations, with no benchmark or trading control. Same-checkpoint controls only show that weights change differently with and without reward. The decoder tends to repeat BUY, which the author flags as one-sided exposure. The chart had to be switched to a light background because the dark chart gave no Kenyon-cell spikes. The input display is an engineered adapter, and the 'profit/loss dopamine' assignments are engineered choices. flybook-git/flybook is an exact copy of this repo (same commit 78ef3e05)." }, { "id": "desktop-fly", "name": "DesktopFly", "type": "demo-desktop", "author_or_org": "Denis Shiryaev (DenisSergeevitch)", "summary": "A desktop pet fly for macOS (with an Electron port for Windows). A 668-neuron spiking circuit cut from FlyWire picks its behaviour state, and since v1.1.0 a 1,045-neuron MaleCNS leg circuit drives its walking legs. Flight, grooming, darting and sleep are coded animations that neuron rates (or the clock) switch on.", "claim": { "text": "A 3D fruit fly living on your macOS desktop, driven by a live spiking simulation of the real FlyWire connectome", "url": "https://github.com/DenisSergeevitch/desktop-fly" }, "dataset": "several", "release": "FlyWire FAFB v783 (brain circuit and display); MaleCNS v1.0 (leg circuit, added in release v1.1.0)", "evidence_grade": "C", "grade_note": "Re-graded from B (borderline C) to C on 29 Sep 2026 after a line-by-line check of unchanged code: neuron rates only switch coded flight, grooming and dart programs, and sleep follows the computer clock and idle time. Walking alone is driven by simulated leg neurons.", "mechanism": { "wiring": "subset: 668 neurons / 18,968 connection rows of FlyWire v783 (11 chosen cell types such as LC4, LPLC2, giant fibre, DNa01/02, DNp09, DNg11, MDN, plus their 330 strongest partners) + subset: 1,045 neurons / 17,224 edges of MaleCNS v1.0 (descending -> VNC -> leg motor neurons). The 23,210 points in the brain window are soma positions drawn for display, not simulated", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none (knee torque and damping coefficients hand-calibrated)", "body": "custom", "scripted_parts": "Neuron rates cross hand-set thresholds to switch an ordinary state machine (walk, idle, groom, fly, sleep). Flight is a random-target timed trajectory, grooming a leg sine animation, darting a coded turn away from the cursor with random speed, and some take-offs are random. Sleep follows the computer's idle time and the clock, not a neuron. Only walking in v1.1 is driven by simulated leg motor neurons." }, "trained_class": "none", "trained_class_note": "hand-calibrated body coefficients", "grade_basis": [ "FlyModel.swift:652-714 brainBehavior(): every state change compares a neuron rate with a hand-set threshold, then runs code: giant-fibre spike -> startFlight; DNg11 rate > 0.5 -> grooming; DNp09 rate > 0.22 -> walking; MDN burst -> 0.5 s backward; arousal -> random take-off [direct]", "FlyModel.swift:448-482, 786-817 startFlight/updateFlight: random landing target and timed smoothstep trajectory; FlyModel.swift:861-865 grooming is a leg sine animation [direct]", "main.swift:1003-1007 sleepy = (idle > 600 s and hour 22-6) or idle > 1800 s: sleep comes from user idle time and the clock, not from a neuron [direct]", "etl.py:37, 104-112 MAX_PARTNERS = 330: the FlyWire part is 11 chosen cell types plus their strongest partners (668 neurons, 0.5% of 139,255); Sim.swift:310-322 hand-set x6 boost on looming -> giant-fibre edges [direct]", "Locomotor.swift:122-128, 202-213 and FlyModel.swift:587-593: in the walking state the MaleCNS leg circuit (fed by FlyWire DN rates through a modelled cross-specimen interface) drives the legs and body motion [direct]", "EVALUATION.md:188-189 (author): 'Grooming, flight, sleep and behavioral state selection still use the existing modeled rules and animations.' Only in-model checks (EVALUATION.md:73-80), no comparison with fly data and no wiring control [direct]" ], "grade_date": "2026-09-29", "measured_result": "Only in-model checks (EVALUATION.md, native run 5 Sep 2026): bilateral DNp09 at 40 Hz for 8 s gives 10.64 model units forward; MDN at 70 Hz gives -12.66 backward; removing all synapses gives zero propulsion. The results are in model units and are not compared with fly data.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/DenisSergeevitch/desktop-fly", "code_licence": "MIT (code); LICENSE adds that data/ is CC BY-NC 4.0 (FlyWire-derived) and data/DATA_LICENSE.md puts the MaleCNS extract under CC BY 4.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 32b00011e8; GitHub API spdx_id=NOASSERTION", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0; MaleCNS: CC-BY-4.0", "platform": [ "macOS", "Windows" ], "gpu": "none", "download_size": "~1.2 MB repo tree incl. data (clone metadata tree_bytes)", "last_commit": { "date": "2026-09-05T21:50:03+02:00", "hash": "32b00011e83c3dc85fa3ea0b3934155b04f1635d", "branch": "master" }, "pushed_at": "2026-09-05T19:50:05Z", "last_release": { "tag": "Release v1.1.0: MaleCNS locomotion and smooth state transitions", "date": "2026-09-05T19:50:03Z", "url": "https://github.com/DenisSergeevitch/desktop-fly/releases/tag/v1.1.0" }, "created_at": "2026-08-18T17:29:17Z", "stars": 1053, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/DenisSergeevitch/desktop-fly", "status": "ok", "http_code": 200, "final_url": "https://github.com/DenisSergeevitch/desktop-fly", "checked_at": "2026-10-06T09:40:15.716Z" }, "last_verified": "2026-10-06", "project_page": "/p/digital-fly-catalog/projects/desktop-fly/", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/DenisSergeevitch/desktop-fly [direct: local clone, HEAD 32b0001]", "https://github.com/DenisSergeevitch/desktop-fly/releases/tag/v1.1.0 [direct: releases feed in gh_meta]" ], "notes_limitations": "Grade C: the network's activity mostly switches coded animations. Only 668 FlyWire neurons and 1,045 MaleCNS neurons are simulated; the 23,210 points in the brain window are drawn, not simulated. The two circuits come from different flies (female FlyWire, male MaleCNS) and are joined by a modelled rate interface, not by synapses. Neuron rates cross hand-set thresholds to switch between behaviour states. Walking legs are driven by simulated motor neurons, but flight, grooming and darting are coded animations, and sleep follows the computer's idle time and the clock. The author's own tests check the model against itself, not against real flies, and the author states that grooming, flight, sleep and state selection use modelled rules and animations. Licence: code MIT; the LICENSE file adds that FlyWire-derived data are CC BY-NC 4.0, and data/DATA_LICENSE.md puts the MaleCNS extract under CC BY 4.0 (GitHub's detector reports NOASSERTION because of the added text). There are no official app downloads: build from source. Ports by others exist for Android, Linux and Windows. A repository named caucasiadoublevision2701/desktop-fly contains no DesktopFly code, only a zip with a Windows program; do not download it." }, { "id": "flybrain-robot-bridge", "name": "FlyBrain Robot Bridge", "type": "demo-desktop", "author_or_org": "Frankweb33 (LICENSE and docs name Himas1211)", "summary": "A Python pipeline that turns camera or synthetic frames into optical-flow features, feeds them to a 'brain backend' and sends left/right motor commands over UDP to an ESP32 (M5 Atom Matrix) robot. The only working backend is eight hand-designed leaky activity groups. The MaleCNS backend is an empty stub that raises NotImplementedError.", "claim": { "text": "Connect a Drosophila connectome simulation to a physical robot; the default backend is a small demonstrator and MaleCNS support is an experimental integration target.", "url": "https://github.com/Frankweb33/flybrain-robot-bridge" }, "dataset": "none (MaleCNS named as a future target only)", "release": "none", "evidence_grade": "D", "grade_note": null, "mechanism": { "wiring": "none. 8 hand-designed groups with hand-set drive formulas, no connectome", "neuron_model": "first-order leaky integrators (tau 0.12 s)", "input_mapping": "Hand-made: Farneback optical-flow magnitude in the left and right image halves plus radial expansion ('looming'), and IMU yaw rate", "output_mapping": "Hand-made: two motor groups scaled to wheel speed; looming above a threshold triggers fixed reverse", "trained_parts": "none", "body": "Physical two-motor robot (M5 Atom Matrix firmware); not validated on hardware", "scripted_parts": "The whole controller: forward bias 0.15, crossed motion-to-motor gains 0.7, fixed reverse on escape" }, "trained_class": "none", "grade_basis": [ "src/flybrain_robot/brain/mock.py:1-49 docstring 'Eight hand-designed leaky populations, not a connectome simulation'; drive formulas hard-coded at 33-44 [direct]", "src/flybrain_robot/brain/malecns.py:1-34 'Extension seam only: no graph loader or biological simulation'; every method raises NotImplementedError [direct]", "src/flybrain_robot/vision.py:20-36 optical-flow heuristics for motion and looming [direct]", "src/flybrain_robot/motor_decoder.py:39-56 escape >= threshold gives fixed full reverse, otherwise scaled left/right [direct]", "docs/VALIDATION.md:13-14 webcam, UDP link, firmware and hardware not validated; MaleCNS deliberately unimplemented [direct]" ], "grade_date": "2026-09-28", "measured_result": "none reported", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Frankweb33/flybrain-robot-bridge", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 966e5419ee; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux", "macOS", "Windows" ], "gpu": "none", "download_size": "~3 MB repository (mostly images); no connectome download", "last_commit": { "date": "2026-09-14T23:55:38+03:00", "hash": "966e5419ee64018557da3e31e6c59b58b36b6534", "branch": "main" }, "pushed_at": "2026-09-16T09:21:44Z", "last_release": "none", "created_at": "2026-09-14T11:21:32Z", "stars": 223, "stars_date": "2026-09-28", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-28", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Frankweb33/flybrain-robot-bridge", "status": "ok", "http_code": 200, "final_url": "https://github.com/Frankweb33/flybrain-robot-bridge", "checked_at": "2026-10-06T09:40:15.678Z" }, "last_verified": "2026-10-06", "added_in_run": "run:98ee92ee-abef-44f6-8c98-42ce102025b7", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/Frankweb33/flybrain-robot-bridge [direct: clone HEAD 966e541]" ], "notes_limitations": "No connectome is loaded or simulated. The behaviour comes entirely from hand-written optical-flow heuristics and eight hand-set leaky units, so the neuron names are cosmetic. The README itself says it is a proof of concept with a mock backend. The demo GIF is software output, not a robot recording. The GitHub account was renamed: github.com/Himas1211/flybrain-robot-bridge redirects here (HTTP 301, checked 2026-09-28), which explains the older name in the LICENSE and docs." }, { "id": "pianist-fly", "name": "pianist-fly", "type": "demo-desktop", "author_or_org": "Noir-infini", "summary": "Runs a numpy LIF simulation of the MaleCNS v1.0 connectome next to a flybody MuJoCo fly that presses keys on a piano. The notes, leg choice and leg movements come from a song file and inverse kinematics; the brain's motor-pool firing only changes how hard each key is pressed.", "claim": { "text": "Real-time LIF simulation of the 166,700-neuron MaleCNS fruit fly connectome driving a 3D MuJoCo fly that hunts sugar and strikes piano keys.", "url": "https://github.com/Noir-infini/pianist-fly" }, "dataset": "MaleCNS", "release": "v1.0", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 (166,700 neurons, 25.58M synapses in data/graph.npz)", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none live; an offline ridge readout script (src/train_leg_readout.py) exists but is not used by play_piano.py", "body": "flybody", "scripted_parts": "Key sequence from a song file; leg choice by key position; press motion is a fixed IK raise-hover-descend-hold-release ramp; the 'sugar side' gate compares plume concentrations computed from geometry, not neural activity; 'known' comes from stimulus strength; a hand-set reward gain boosts tarsal drive" }, "trained_class": "none", "trained_class_note": "offline ridge readout script not used", "grade_basis": [ "scripts/fetch_graph.py:33-49 downloads the MaleCNS v1.0 feather files [direct]", "src/native_brain.py:42,185 LIF kernel with bincount spike delivery [direct]", "src/brain_driver.py:58-61 'play_piano.py always runs its own scripted choreography; the brain never picks the note'; it only shapes press force [direct]", "src/play_piano.py:265-268 the side gate uses plume concentrations cL/cR, not spikes [direct]", "src/brain_driver.py:319-320 intent 'known' is stimulus strength above a threshold, not neural output [direct]", "src/brain_driver.py:449-453 motor output = T1 motor-pool rate x a lateral weight computed from the stimulus [direct]", "src/play_piano.py:205-210,310-334 pool rate -> press force; leg chosen from key y position and IK [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Noir-infini/pianist-fly", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 6aebb8aced; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "Linux" ], "gpu": "none", "download_size": "~1.1 GB MaleCNS v1.0 data (not stated exactly)", "last_commit": { "date": "2026-09-23T18:38:42+05:30", "hash": "6aebb8aced550f7bc0c98a7779b591c684666f0d", "branch": "main" }, "pushed_at": "2026-09-23T13:08:43Z", "last_release": { "tag": "demo-assets", "date": "2026-09-20T06:45:02Z", "url": "https://github.com/Noir-infini/pianist-fly/releases/tag/demo-assets" }, "created_at": "2026-09-20T06:41:42Z", "stars": 3, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Noir-infini/pianist-fly", "status": "ok", "http_code": 200, "final_url": "https://github.com/Noir-infini/pianist-fly", "checked_at": "2026-10-06T09:40:16.145Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/Noir-infini/pianist-fly [direct]" ], "notes_limitations": "README says the connectome drives the fly and the antennae decide the sugar side. In the code, the side check uses plume concentrations from geometry (play_piano.py:265-268), so the neural simulation does not gate or choose movements; it only scales press force. The author's own comment in brain_driver.py:58-61 says the choreography is scripted. The whole-brain LIF does run on real MaleCNS v1.0 wiring. No control or benchmark. Linux only; an optional Rust brain viewer lives in a separate repo not included here." }, { "id": "bad-apple-fly", "name": "Bad Apple Fly", "type": "demo-art", "author_or_org": "Kevin Lin (kevinlinxc; X: @linguinelabs)", "summary": "Frames of the 'Bad Apple' video select which MaleCNS neurons get stimulated, based on where their cell bodies sit in the XY plane. A LIF model of the full graph then runs. Descending-neuron rates switch pre-programmed walking and grooming movements on a NeuroMechFly body in MuJoCo.", "claim": { "text": "Bad Apple but it's playing on a fly's brain", "url": "https://x.com/linguinelabs/status/2096487329091441090" }, "dataset": "MaleCNS", "release": "v1.0 (minconf-0.5 flat connectome)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 full flat connectome (minconf 0.5)", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none", "body": "FlyGym", "scripted_parts": "Motor output uses FlyGym PreprogrammedSteps and HybridTurningController (recorded walking steps). Grooming and other movements are hand-written gestures, weighted by the rates of chosen descending neurons (DNg100/DNg97 forward, MDN backward, DNa02/DNg13 steering, grooming DNs)." }, "trained_class": "none", "grade_basis": [ "backend/video.py:1,19-50 VideoProjector: dark pixels (brightness below 96) select neurons whose soma XY falls on them, at every depth. The picture is imposed as stimulation, not seen by the fly [direct]", "backend/app.py:184-193 stimulus_inputs(): each selected neuron is driven at rate_hz (default 100 Hz) [direct]", "backend/brain.py:1 'Sparse, delayed LIF dynamics on MaleCNS'; scripts/download_data.py:8-12 MaleCNS v1.0 feather files [direct]", "backend/body.py:1,19,43-47,65-98 FlyGym PreprogrammedSteps + HybridTurningController; _gesture() is described as 'explicit qualitative motor programs' [direct]", "backend/evidence.py:20-32 hand-made map from descending-neuron types to movement programs, with literature citations [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/kevinlinxc/badapple-fly", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit 6726892e7e); GitHub API spdx_id=None", "data_licence": "CC-BY-4.0", "platform": [ "macOS", "Linux" ], "gpu": "none", "download_size": "~1 GB connectome tables (README)", "last_commit": { "date": "2026-09-13T00:41:26-07:00", "hash": "6726892e7ebd93e6c5ffab056ab8db09433c28a6", "branch": "main" }, "pushed_at": "2026-09-13T07:42:48Z", "last_release": "none", "created_at": "2026-09-13T07:03:45Z", "stars": 3, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/kevinlinxc/badapple-fly", "status": "ok", "http_code": 200, "final_url": "https://github.com/kevinlinxc/badapple-fly", "checked_at": "2026-10-06T09:40:16.095Z", "other_links": [ { "url": "https://x.com/linguinelabs/status/2096487329091441090", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/kevinlinxc/badapple-fly [direct: local clone, HEAD 6726892]", "https://x.com/linguinelabs/status/2096487329091441090 (via https://api.fxtwitter.com/linguinelabs/status/2096487329091441090) [direct]" ], "notes_limitations": "AUTHORSHIP CONFLICT: the X page is titled 'Kevin on X' under the handle @linguinelabs. Resolved: the post's author profile (fetched through the fxtwitter API) is named 'Kevin', says 'bad apple engineer' and links to kevinlinxc.com. The only commit in the repo is by Kevin Lin. So @linguinelabs and GitHub kevinlinxc are very likely the same person. The post (6 Sep 2026, about 20.8 M views at fetch) came a week before the repo was created (13 Sep). The public code may therefore differ from what made the viral clip. MECHANISM: the Bad Apple image in the brain view mostly shows the stimulus itself. The neurons under dark pixels are forced to fire, so the picture is drawn onto soma positions rather than produced by the network. Network propagation does add spread-out activity. The README says plainly that this is 'not a real brain simulation': parameters are invented, the video is not something the fly sees, there is no sensory feedback, and the female NeuroMechFly body is paired with a male connectome. Graded C because connectome activity only switches scripted or pre-recorded motor programs. No licence file, so default copyright applies. A full run of the video can take an hour or more." }, { "id": "fly-with-me", "name": "Fly With Me (Fly Brain DJ)", "type": "demo-art", "author_or_org": "izntariq", "summary": "A local Python server runs the whole FlyWire v783 brain as a real-time LIF network. Music is split into frequency bands that drive Johnston's-organ (antennal hearing) neurons; bursts in named descending-neuron types are mapped by hand to DJ moves (deck choice, scratches, echo, filters) and a synth in the browser.", "claim": { "text": "A real fruit-fly brain that makes music and DJs.", "url": "https://github.com/izntariq/fly-with-me" }, "dataset": "FlyWire FAFB", "release": "v783 via Shiu et al. Completeness_783.csv / Connectivity_783.parquet (138,639 neurons, 15,091,983 connections)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "Whole FlyWire v783 brain, Shiu weights", "neuron_model": "spiking: Shiu et al. LIF, fast real-time reimplementation", "input_mapping": "hand-set: audio frequency bands -> JO-A/JO-B neuron groups (40 Hz to 2.2 kHz bands)", "output_mapping": "hand-set: bursts (fast rate above slow baseline) in named DN types -> DJ actions (DNa01/02 steering -> deck, DNp01 -> loop roll, DNg24/29 -> scratch, ...)", "trained_parts": "none", "body": "none", "scripted_parts": "Beat matching, key lock, set planning, stems and the 29 DJ styles are ordinary audio code; the brain triggers moves" }, "trained_class": "none", "grade_basis": [ "flybrain/brain.py:1-18,35-36 real-time Shiu LIF on Completeness_783 / Connectivity_783 [direct]", "flybrain/mapping.py:18-24 Johnston's organ JO-B/JO-A groups by frequency band [direct]", "web/js/flydj.js:3-13 burst in named DN types -> DJ move table [direct]" ], "grade_date": "2026-09-30", "measured_result": "No measured result or control in the repository.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/izntariq/fly-with-me", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 8454c6ad82; GitHub API spdx_id=Apache-2.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "Linux", "macOS", "Windows" ], "platform_note": "local Python server opened in a browser", "gpu": "none", "download_size": "about 400 MB of connectome data on first run (README)", "last_commit": { "date": "2026-09-28T20:29:56-07:00", "hash": "8454c6ad82b31bd7cba44217413a8b45353b4130", "branch": "main" }, "pushed_at": "2026-09-29T03:41:11Z", "last_release": { "tag": "v1.12", "date": "2026-09-29T03:29:56Z", "url": "https://github.com/izntariq/fly-with-me/releases/tag/v1.12" }, "created_at": "2026-09-29T03:38:57Z", "stars": 1, "stars_date": "2026-10-05", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/izntariq/fly-with-me", "status": "ok", "http_code": 200, "final_url": "https://github.com/izntariq/fly-with-me", "checked_at": "2026-10-06T09:40:16.444Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/izntariq/fly-with-me [direct: clone HEAD 8454c6a, 2026-09-30]" ], "notes_limitations": "The brain really runs and hears the music through its antennal hearing neurons, but which descending neurons trigger which DJ move is the author's choice, and the musical skill (beat matching, key lock, set planning) is ordinary audio code. No scrambled-wiring comparison." }, { "id": "flybrain-bad-apple-doom", "name": "FLYBRAIN Bad Apple x DOOM", "type": "demo-art", "author_or_org": "fazchile17", "summary": "A local 3-D viewer of the FlyWire v783 connectome (139,255 neurons, 15.1 M directed pairs) in which the neurons are used as pixels to play the Bad Apple!! video or the shareware DOOM episode. It also lets you search neurons and list their partners. No neural activity is simulated.", "claim": { "text": "Sobre el mismo mapa puedes reproducir Bad Apple!! o jugar al episodio shareware de DOOM, pintados solo con las neuronas.", "url": "https://github.com/fazchile17/FLYBRAIN-BAD_APPLE_X_DOOM" }, "dataset": "FlyWire FAFB", "release": "v783 (139,255 neurons, 15,091,983 directed pairs, 54,492,922 synapses; about 800 MB download)", "evidence_grade": "D", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 positions and connections for display only", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none", "body": "none", "scripted_parts": "Video frames and the DOOM screen are painted onto neuron positions; the game is ordinary DOOM" }, "trained_class": "none", "grade_basis": [ "README.md (Spanish): 3-D viewer; Bad Apple!! and DOOM painted with neurons [direct]", "app/app.js: three.js rendering, search and partner tables; code_scan found no neuron model (only three.js matches for \"spike\") [direct]" ], "grade_date": "2026-09-30", "measured_result": "No measured result; no simulation.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/fazchile17/FLYBRAIN-BAD_APPLE_X_DOOM", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit 4650db26b2); GitHub API spdx_id=None", "data_licence": "CC-BY-NC-4.0", "platform": [ "Windows", "Linux", "macOS" ], "platform_note": "local Python server and WebGL browser viewer", "gpu": "none", "gpu_note": "WebGL browser", "download_size": "about 800 MB FlyWire download and 460 MB of processed indices; about 3 GB free disk (README)", "last_commit": { "date": "2026-09-12T17:56:32-03:00", "hash": "4650db26b2597a081506bac3ad22038efbb7c60e", "branch": "main" }, "pushed_at": "2026-09-12T20:59:05Z", "last_release": "none", "created_at": "2026-09-12T20:53:43Z", "stars": 2, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/fazchile17/FLYBRAIN-BAD_APPLE_X_DOOM", "status": "ok", "http_code": 200, "final_url": "https://github.com/fazchile17/FLYBRAIN-BAD_APPLE_X_DOOM", "checked_at": "2026-10-06T09:40:16.432Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/fazchile17/FLYBRAIN-BAD_APPLE_X_DOOM [direct: clone HEAD 4650db2, 2026-09-30]" ], "notes_limitations": "A connectome display, not a brain simulation: the fly does not play DOOM, its neurons are used as screen pixels. It is a different project from badapple-fly (kevinlinxc), by a different author with different code." }, { "id": "infinite-sugar", "name": "Infinite Sugar", "type": "demo-art", "author_or_org": "cnqso", "summary": "A browser artwork: a MuJoCo flybody fly sits in a terrarium while a whole-brain spiking model of FlyWire gets constant sweet-taste input. Activity in proboscis, head and antenna motor neurons moves the matching body joints.", "claim": { "text": "A whole-brain emulation using the FlyWire connectome: 139,255 neurons driving a simulated fruit fly in a terrarium.", "url": "https://github.com/cnqso/infinite-sugar" }, "dataset": "FlyWire FAFB", "release": "v783", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "FlyWire FAFB v783, all 139,255 neurons and 2,700,513 edges (Codex >=5-synapse threshold), synapse count x transmitter sign", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none (global gains, noise and baseline drive are hand-tuned)", "body": "flybody", "scripted_parts": "Sugar input is always on. Wing postures and a small foot-lift pattern ('ours, not a reconstructed gait') are supplied by code and only scaled or triggered by descending-neuron rates (DNg11 grooming, DNa01/02 steering). There is no walking. Random noise kicks and a random tonic baseline drive keep the network active." }, "trained_class": "none", "trained_class_note": "hand-tuned gains", "grade_basis": [ "web/brain.ts:1 header: whole-brain LIF over FlyWire FAFB v783, 139,255 neurons / 2,700,513 edges [direct]", "tools/fetch_flywire.sh downloads the Codex FAFB 783 dumps; tools/build_brain.py:2 builds the graph from them [direct]", "web/brain.ts:188 step() updates all N neurons every 1 ms (20 ms tau, threshold, refractory, delayed inhibition); weights loaded from w2.bin.gz at web/brain.ts:135 [direct]", "web/brain.ts:109 sweet stimulus drives the grn_sweet / grn_sweet_leg populations; web/app.ts:802 turns sugar on at start [direct]", "web/app.ts:664 brain.step(1) runs every 10 MuJoCo steps, then applyBrainToActuators (web/app.ts:571) [direct]", "web/app.ts:467 DRIVE table maps motor-neuron pools (mn_proboscis, mn_neck, mn_antenna) to actuators; wing rows from DN pools are marked as supplied 'command' mappings (web/app.ts:458) [direct]", "web/app.ts:593 and :649 foot shuffle is a fixed sin^2 flexion pattern triggered when DNg11 activity builds up [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://infinitesugar.cnqso.com/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/cnqso/infinite-sugar", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit fdbbd866b2); GitHub API spdx_id=None", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "36.04 MB transferred in our browser check (6 Oct 2026; largest file colidx.bin.gz 7.12 MB). Figure from the code before the check: ~47 MB static site (brain data ~10 MB gzipped, MuJoCo WASM ~11 MB)", "last_commit": { "date": "2026-09-14T09:01:11-04:00", "hash": "fdbbd866b203a709a164970b0a8996108edd57a5", "branch": "main" }, "pushed_at": "2026-09-14T13:01:17Z", "last_release": "none", "created_at": "2026-09-08T15:01:33Z", "stars": 25, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://infinitesugar.cnqso.com/", "status": "ok", "http_code": 200, "final_url": "https://infinitesugar.cnqso.com/", "checked_at": "2026-10-06T09:40:16.231Z", "other_links": [ { "url": "https://github.com/cnqso/infinite-sugar", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "runs", "checked_at": "2026-10-06", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 36.04, "compute": "in-browser-wasm", "first_activity_s": 36.84, "console_errors": 0, "page_errors": 0, "matches_catalogue": "yes", "note": "Ran; in your browser (WebAssembly); 36.04 MB in the first 60-90 s (largest: colidx.bin.gz 7.12 MB).", "source": "Digital Fly Lab browser check of 2026-10-06 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-06-run9/#browser-heading), result file infinite-sugar.json" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/cnqso/infinite-sugar (clone: README.md, web/brain.ts, web/app.ts, tools/*) [direct]", "https://infinitesugar.cnqso.com/ (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. Needs WebAssembly (mujoco_wasm), WebGL (three.js) and DecompressionStream; no WebGPU. The brain runs in plain TypeScript on the main thread. FAFB is brain-only, so only proboscis, head and antenna are driven directly by motor neurons, and the README says so. Candidate list typed this browser-demo; set to demo-art because the project presents itself as an artwork. No wiring control or benchmark: tools/shuffle_test.mjs tests the foot-shuffle routine, not shuffled wiring. No licence file." }, { "id": "neurafly", "name": "neurafly", "type": "demo-art", "author_or_org": "emiliano-go", "summary": "A terminal audio visualiser written in Rust. It loads the FlyWire v783 graph (138,333 neurons, 2,052,622 weighted edges after filtering), injects the energy of ten audio frequency bands into neurons chosen by their horizontal position in a 2-D layout, runs a simple leaky threshold model and draws a braille disk whose shape follows each band's mean activity.", "claim": { "text": "\"A terminal audio visualizer that runs your music through the actual FlyWire whole-brain connectome ... Real audio in, real neurons firing, real time.\" (README)", "url": "https://github.com/emiliano-go/neurafly" }, "dataset": "FlyWire FAFB", "release": "v783 (proofread_connections_783.feather, Zenodo 10.5281/zenodo.10676865), filtered to 2,052,622 weighted edges (data/flywire_net.bin, 18 MB)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 whole brain, signed weights from synapse counts and transmitter fractions (tools/preprocess.py, W_SCALE 0.075)", "neuron_model": "other: discrete-time leaky threshold units with refractory steps and hashed threshold jitter (src/sim.rs tick)", "input_mapping": "hand-made: ten tonotopic bands, each injected into the neurons of one horizontal stripe of the 2-D layout (preprocess.py: band = x * 24 // width), not into auditory neurons", "output_mapping": "hand-made: per-band mean activity of the same band's neurons drives disk diameter, rim bumps and anomalies (src/figure.rs:475-489)", "trained_parts": "none", "body": "none", "scripted_parts": "the disk geometry, shockwaves and flares are drawing code; bass onsets also trigger effects" }, "trained_class": "none", "grade_basis": [ "tools/preprocess.py at ab0e853: builds the network from the FlyWire v783 feather; band assignment by layout x position (line 135) [direct]", "src/sim.rs:305-345: leak, threshold, spike delivery along the edges, then audio current into band_members [direct]", "src/figure.rs:475-489: the drawing reads pools.band_act, the mean activity of each band's own injected neurons [direct]", "No comparison with a scrambled or no-network version; the 'r' key rewires to a random reseed but no result is reported [direct]" ], "grade_date": "2026-10-03", "measured_result": "None (a visualiser; no measurement).", "try_url": null, "try_status": "no no-install option (Rust; Linux with PulseAudio or PipeWire; AUR package or cargo install)", "code_url": "https://github.com/emiliano-go/neurafly", "code_licence": "MIT", "code_licence_source": "LICENSE file at ab0e853 (MIT)", "data_licence": "FlyWire v783 derived data committed (CC BY 4.0 per FlyWire; not stated in the repository)", "platform": [ "Linux" ], "platform_note": "Rust (stable), Linux desktop with PulseAudio or PipeWire, Unicode braille terminal", "gpu": "none", "download_size": "18 MB connectome file (data/flywire_net.bin) plus the binary", "last_commit": { "date": "2026-10-03T16:32:55-03:00", "hash": "4cfdc741b1ebef4467bb4274bb8da1dd13137925", "branch": "main" }, "pushed_at": "2026-10-02T20:16:44Z", "last_release": { "tag": "v0.2.0", "date": "2026-09-30T15:50:39Z", "url": "https://github.com/emiliano-go/neurafly/releases/tag/v0.2.0" }, "created_at": "2026-09-17T18:34:18Z", "stars": 6, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/emiliano-go/neurafly", "status": "ok", "http_code": 200, "final_url": "https://github.com/emiliano-go/neurafly", "checked_at": "2026-10-06T09:40:16.846Z" }, "last_verified": "2026-10-06", "added_in_run": "run:5bfca8cc-f3cb-4020-83e6-5c1225da9200", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/emiliano-go/neurafly [direct]" ], "notes_limitations": "The connectome really is loaded and stepped, but the picture is read from the same neuron pools the audio is injected into, chosen by screen position rather than by any auditory pathway, so the network acts mostly as a relay of the music's band energies. Without a scrambled-network comparison there is no sign that the wiring shapes what you see. The crates.io build falls back to a small synthetic network." }, { "id": "chessfly", "name": "ChessFly", "type": "browser-demo", "author_or_org": "Ruben Nugmanov (znatgost; also the author of fly67)", "summary": "A browser game in which a mushroom-body model learns chess by self-play. The 285 olfactory projection neurons, 5,177 Kenyon cells and their 21,509 connections (at least 5 synapses each) come from FlyWire v783; the board is fed onto the projection neurons through a fixed hand-made mapping, Kenyon cells fire when at least 2 inputs are on, and one output neuron's plastic synapses learn from game results. A small alpha-beta search picks the move. The author trained the real wiring, a shuffled-claws control and the original random design for 10,000 games each.", "claim": { "text": "\"The real wiring of a fruit fly's learning centre, taken from the FlyWire connectome and hooked up to a chessboard ... teaches itself chess by playing itself\"; and \"the real FlyWire wiring learns chess exactly as well as the same cells with their inputs shuffled\" (README)", "url": "https://github.com/znatgost/chessfly" }, "dataset": "FlyWire FAFB", "release": "v783 (as packaged by Shiu et al. 2024): 285 PNs, 5,177 KCs, 21,509 PN->KC connections with >= 5 synapses (data/mushroom_body.json)", "evidence_grade": "A", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 PN->KC connections only (285 PNs, 5,177 KCs, 21,509 edges >= 5 synapses); binary (who connects to whom)", "neuron_model": "other: binary threshold Kenyon cells (fire when >= 2 inputs are on) and one rate-like output neuron (tanh-squashed sum)", "input_mapping": "hand-made: 806 board channels (piece-square and piece-count) converge 2-3 per PN by a seeded permutation (src/brain.js)", "output_mapping": "hand-made: the single MBON value is the position evaluation used by an ordinary alpha-beta search (1-3 moves plus captures)", "trained_parts": "KC->MBON weights and a board-channel shortcut, by a dopamine-like prediction-error rule from self-play results", "body": "none (3-D fly carrying pieces is animation)", "scripted_parts": "alpha-beta search and move generation are ordinary engine code; games stopped at 200 moves are scored by material" }, "trained_class": "other", "trained_class_note": "prediction-error plasticity rule on KC->MBON", "grade_basis": [ "data/mushroom_body.json at 52276aa: source 'FlyWire connectome v783 ... as packaged by Shiu et al. 2024', minSynapses 5, 285 PN IDs and KC claw lists [direct]", "src/brain.js:9-14, 85-105: 'flywire', 'shuffled' (same claw count per KC, each claw re-drawn from the pool of real PN partners, so PN popularity is kept in expectation) and 'random' (4,000 cells, 7 random inputs straight from board channels) wirings [direct]", "brains/adult.json, adult-shuffled.json, adult-random.json: exam curves over 10,000 self-play games (seed 1 each); exam rating at 10,000 games: real 1680, shuffled 1675, random design 1850; untrained egg 660 for all [direct]", "README 'Real wiring vs random wiring': bigger 200-game exams 1690 / 1695 / 1755 and a round robin (real vs shuffled 53% to 47%); tools/compare.mjs produces them, but its output is not committed [README; numbers not in files]" ], "grade_date": "2026-10-03", "measured_result": "Exam rating (Elo scale on which the Random bot = 400) after 10,000 self-play games, one fly per wiring, from the committed brain files: real FlyWire wiring 1680, shuffled claws 1675, the original random design 1850; all start at 660. The author's bigger exams (README): 1690, 1695 and 1755; real vs shuffled 53% to 47% head to head. The real wiring is no better than shuffled wiring and a little worse than a random design.", "try_url": "https://znatgost.github.io/chessfly/", "try_status": "page link from README; not loaded by us", "code_url": "https://github.com/znatgost/chessfly", "code_licence": "MIT", "code_licence_source": "LICENSE file at 52276aa (MIT, 2026 Ruben Nugmanov)", "data_licence": "FlyWire v783 derived data (CC BY 4.0 per FlyWire; not stated in the repository)", "platform": [ "browser" ], "gpu": "none", "download_size": "small: brains/*.json about 30 KB each, data/mushroom_body.json (not measured in a browser)", "last_commit": { "date": "2026-09-30T20:58:50+04:00", "hash": "52276aaee3d5f60842cdc5409ad4ae99c788b0b1", "branch": "main" }, "pushed_at": "2026-09-30T16:59:20Z", "last_release": "none", "created_at": "2026-09-30T16:59:01Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://znatgost.github.io/chessfly/", "status": "ok", "http_code": 200, "final_url": "https://znatgost.github.io/chessfly/", "checked_at": "2026-10-06T09:40:16.379Z", "other_links": [ { "url": "https://github.com/znatgost/chessfly", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:5bfca8cc-f3cb-4020-83e6-5c1225da9200", "controls": "wiring-null", "wiring_effect": "no-difference", "sources": [ "https://github.com/znatgost/chessfly [direct]" ], "notes_limitations": "A clean negative result, but thin: one trained fly per wiring and one shuffle (seed 1), with exam ratings from 40-game exams per opponent, so differences of a few tens of Elo are within noise. Only the PN->KC layer is from the connectome; the Kenyon cells are binary threshold units, a single output neuron does the evaluating and an ordinary search picks moves. The shuffle keeps each Kenyon cell's claw count and draws partners from the real PN pool (PN popularity kept only in expectation)." }, { "id": "cyber-larva", "name": "CyberLarva", "type": "browser-demo", "author_or_org": "ChenYvhang", "summary": "A 3D larva crawling in an editable habitat. The online page runs a simple animation: the larva heads for the nearest food with a sine-wave crawl, and no neurons are simulated. A separate local Python/MuJoCo build can load the Winding et al. L1 larval connectome if the user prepares the data.", "claim": { "text": "A sparse neural simulation constrained by the Winding et al. L1 connectome receives environmental sensory signals ... drives an 11-segment MuJoCo neuromechanical body", "url": "https://github.com/ChenYvhang/cyber-larva" }, "dataset": "larval L1", "release": "Winding et al. 2023 Supplementary Data S1 (local build only, user-prepared; the browser demo loads none)", "evidence_grade": "D", "grade_note": null, "mechanism": { "wiring": "none in the browser (0 neurons simulated). Local build only: larval L1, 2,952 nodes from user-prepared data, otherwise a random 512-node scaffold", "neuron_model": "none", "input_mapping": "none", "output_mapping": "none", "trained_parts": "none (local build body/gait parameters calibrated against Schleyer larva tracks)", "body": "custom", "scripted_parts": "In the browser, heading turns toward the nearest food with atan2 and the crawl is a fixed 1.41 Hz sine wave. The wall response is a fixed +0.9 rad turn. 'Knockout' lowers drive by 0.035 per silenced item, whichever neuron it is. The neuron list is 12 hard-coded entries while reporting 'total: 2952', and 'active neurons' is 180*drive. In the local build, a segmental CPG sets the rhythm, and DN activity only adjusts it slightly." }, "trained_class": "none", "trained_class_note": "body/gait parameters calibrated", "grade_basis": [ "tools/build-pages.mjs:5,13-14 the Pages build injects web/demo-api.js, which replaces /api/* with in-browser code (no Python/MuJoCo server) [direct]", "web/demo-api.js:43-47 steers toward the nearest food with atan2, uses a fixed-frequency phase and a scripted bump turn, and has no neural computation [direct]", "web/demo-api.js:36 segment muscle wave from sin(phase+i*.66) [direct]", "web/demo-api.js:44,53 knockout lowers drive by 0.035 per item; 'active' = Math.round(180*drive) [direct]", "web/demo-api.js:12-25,68 hard-coded 12-neuron list reported as total: 2952 [direct]", "cyberlarva/connectome.py:54-72 (local build) loads winding_l1_connectome.npz if present, else a random 512-node scaffold; :92,99-102 random E/I signs and DN groups taken by index stride; :169-175 CPG with DN modulation, turning dominated by odor_lr*1.5 [direct]" ], "grade_date": "2026-09-28", "measured_result": "Local build only: paired MDNa knockout (20 seeds, 5 s trials) gave identical control and knockout values on every metric (0.0% change). The author reports this as a null result showing that the controller bypasses MDNa (experiments/mdna_bilateral_knockout.md). Nothing is measured for the browser demo.", "try_url": "https://chenyvhang.github.io/cyber-larva/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/ChenYvhang/cyber-larva", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 2af0910a07; GitHub API spdx_id=MIT", "data_licence": "article CC BY 4.0; no separate data licence found", "platform": [ "browser", "Windows", "macOS", "Linux" ], "gpu": "none", "download_size": "not stated (browser page is small: three.js + ~180 KB GLB); the local build needs the Winding S1 archive, downloaded by the user", "last_commit": { "date": "2026-09-15T14:42:58+08:00", "hash": "2af0910a0777b61b6557bcffa0ad8a881aaa7928", "branch": "main" }, "pushed_at": "2026-09-15T06:43:00Z", "last_release": "none", "created_at": "2026-09-15T03:07:11Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://chenyvhang.github.io/cyber-larva/", "status": "ok", "http_code": 200, "final_url": "https://chenyvhang.github.io/cyber-larva/", "checked_at": "2026-10-06T09:40:16.497Z", "other_links": [ { "url": "https://github.com/ChenYvhang/cyber-larva", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/ChenYvhang/cyber-larva (clone read at HEAD) [direct]", "https://chenyvhang.github.io/cyber-larva/ (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. Needs WebGL (three.js). No WebGPU or WASM, and MuJoCo does not run in the browser. The README does say the online build uses a 'reduced browser-side segmented controller', but in fact it simulates no neurons, and the page header still reads 'MuJoCo neuromechanics / connectome habitat'. Its neuron count and 'active neurons' numbers are made up. The local build would be about grade C: the connectome data is not shipped (gitignored), transmitter signs are random, DN groups are arbitrary slices, and behaviour comes mostly from the CPG and direct sensor terms. flybrain.info hosts a copy under /play/cyberlarva/." }, { "id": "fly-brain-lulzx", "name": "fly-brain", "type": "browser-demo", "author_or_org": "Lulzx", "summary": "Runs the whole MaleCNS v1.0 connectome (165,122 neurons) as a spiking network for each fly in the browser, with a MuJoCo flybody body and a flyvis compound eye. Descending-neuron activity sets walking, turning, grooming and escape commands. Documented offline experiments score the model against literature targets and against shuffled-wiring controls.", "claim": { "text": "165,122 neurons, 104 million synapses. This project runs that file as a spiking brain, inside a physics-simulated body, in a web browser", "url": "https://github.com/Lulzx/fly-brain" }, "dataset": "MaleCNS", "release": "v1.0", "evidence_grade": "A", "grade_note": "Borderline A: the shuffled-wiring benchmark was also used to fit the model, and it has not been re-run after a later fix.", "mechanism": { "wiring": "MaleCNS v1.0 (Janelia FlyEM), all 165,122 neurons; CSR graph of connections with >=3 synapses (browser run uses minSyn 6); transmitter signs", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "The flyvis optic-lobe model (a separately trained network) drives the optic-lobe input neurons from eye raycasts. The tripod stepping generator is fitted by CMA-ES to FlySuite kinematics. Nine global brain parameters were fitted by cross-entropy search to the same literature benchmark used for scoring. A reafference-cancel model is also fitted.", "body": "flybody", "scripted_parts": "The docs openly list many 'scaffold' plugins outside the graph. These include a tripod CPG that makes all coordinated leg movement (without it the fly freezes), a bout scheduler using ethology statistics, an escape gate, a reafference gain, steering adaptation, a righting reflex, a groom pattern, courtship and female-rejection rules, flight saccades, an octopamine arousal rule and a hand-added GF gap junction. Each can be switched off with ?off=." }, "trained_class": "front-end", "trained_class_note": "flyvis front end; stepping generator and 9 brain parameters fitted by search", "grade_basis": [ "scripts/prep_graph.py:7,10,32 read MaleCNS v1.0 annotation, transmitter and connectome-weight feather files [direct]", "src/wasm/lif.c:29 lif_step, the WASM SIMD conductance LIF kernel over the full graph; src/brainsetup.js:66 uses the WebGPU kernel if navigator.gpu exists, otherwise WASM [direct]", "src/arena.js:47,75 loads lif.wasm into SharedArrayBuffer memory (warns if not cross-origin isolated); src/arena.js:56 loads the flyvis vision model [direct]", "src/sim/motor.js:3,10 default 'descending' mode: DN_ROLES map named descending neurons (DNg100, MDN, DNa02, DNp01, ...) to forward, backward, turn, groom and escape commands [direct]", "src/sim/scaffold/plugins/cpg.js:6,23 the stepping generator is a CMA-ES-fitted two-harmonic tripod, on by default [direct]", "scripts/rungs.mjs:17 and scripts/calib_eval.mjs:38 w_shuffle control permutes synapse counts across retained edges [direct]", "public/data/ablation_ladder.json and ablation_refit.json (generated 2026-09-18): baseline 0.7961 vs w_shuffle 0.4041 (refit 0.5591), sign_free 0.5241, 12 paired seeds [direct]", "public/data/experiments/respond-like-the-fly.md compares ensemble members, including weight-shuffled wiring, with animal perturbation results (MDN, 13B, decapitation) [direct]" ], "grade_date": "2026-09-28", "measured_result": "Ablation ladder (docs/31-ablation-ladder.md, public/data/ablation_ladder.json; 12 paired seeds, 17-assay benchmark built from literature summary statistics): calibrated model 0.796 +/- 0.006; weight-shuffled wiring 0.404 (change -0.392 +/- 0.013), 0.559 after refitting; all-excitatory control 0.524. respond-like-the-fly: only 1 of 5 animal rows is matched by every ensemble member, and 0 of 4 members match all rows.", "try_url": "https://lulzx.com/fly-brain/arena.html", "try_status": "loads its brain data, then passed our 1.6 GB browser memory limit in headless Chromium without a GPU (checked 2026-10-03; not a play test)", "code_url": "https://github.com/Lulzx/fly-brain", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit ba0b4f815c; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "browser" ], "gpu": "optional", "download_size": "~23 MB for the arena (README); ~30 MB for the viewer", "last_commit": { "date": "2026-10-05T11:22:08+05:30", "hash": "ba0b4f815c00e81a2ef1f36165df2b1891e5e39c", "branch": "main" }, "pushed_at": "2026-09-24T15:59:20Z", "last_release": "none", "created_at": "2026-09-10T10:05:59Z", "stars": 27, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": "https://lulzx.com/fly-brain/textbook/", "peer_review": "none", "link_status": { "url": "https://lulzx.com/fly-brain/arena.html", "status": "ok", "http_code": 200, "final_url": "https://lulzx.com/fly-brain/arena.html", "checked_at": "2026-10-06T09:40:16.633Z", "other_links": [ { "url": "https://github.com/Lulzx/fly-brain", "status": "ok", "http_code": 200 }, { "url": "https://lulzx.com/fly-brain/textbook/", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "stopped-memory", "checked_at": "2026-10-03", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 33.93, "compute": "unclear", "first_activity_s": null, "console_errors": 2, "page_errors": 0, "matches_catalogue": "unclear", "note": "Fetched the 14.6 MB MaleCNS graph and WebAssembly through its service worker and started 4-5 workers, then passed our 1.6 GB browser memory limit (both loads, at 29 s and 44 s). Without cross-origin isolation the page warned that SharedArrayBuffer was unavailable.", "source": "Digital Fly Lab browser check of 2026-10-03 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-03-run7/#run-heading), result file fly-brain-lulzx-load2.json" }, "last_verified": "2026-10-06", "controls": "wiring-null", "wiring_effect": "helps", "sources": [ "https://github.com/Lulzx/fly-brain (clone: README.md, docs/03, docs/08, docs/31, src/arena.js, src/brainsetup.js, src/wasm/lif.c, src/sim/motor.js, src/sim/scaffold/*, scripts/rungs.mjs, scripts/calib_eval.mjs, public/data/*) [direct]", "https://lulzx.com/fly-brain/arena.html (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. Needs WebAssembly, SharedArrayBuffer (cross-origin isolation through a service worker) and a module Web Worker per fly. WebGPU is used if present, else WASM (?gpu=0 forces WASM). The README targets desktop Chrome, Edge or Firefox. Grade A rests on the offline benchmark with shuffled-wiring and sign-free controls, not on the arena behaviour. The benchmark targets are literature summary numbers, and the same benchmark was used to fit the global parameters. Doc 31 says its table predates a later objective repair (baseline would drop to 0.697) and has not been re-run. In the arena, coordinated walking comes from the fitted CPG scaffold, not from the connectome; the author documents this openly. The 'paper' is a self-published textbook. Browser check 2026-10-03 (no GPU): the arena fetched its 14.6 MB graph and started 4-5 workers, then passed our 1.6 GB browser memory limit twice; the page also warned that it was not cross-origin isolated (no SharedArrayBuffer)." }, { "id": "fly-dino", "name": "Fly Dino (flyjump)", "type": "browser-demo", "author_or_org": "Mert Cobanov (cobanov)", "summary": "Plays the original Chromium Dino game. Eight hand-made game-state numbers feed an 80-neuron subgraph taken from MaleCNS v1.0 and run as a leaky tanh rate model. The activity of 16 descending neurons goes to a trained 243-parameter readout (16-12-3), which picks run, jump or duck.", "claim": { "text": "An 80-cell measured fly connectome circuit drives the original Chromium Dino through a learned neural readout.", "url": "https://github.com/cobanov/flyjump" }, "dataset": "MaleCNS", "release": "v1.0 (min confidence 0.5)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "subset: 80 neurons of MaleCNS v1.0 (32 visual input cells, 16 descending outputs, 32 bridge cells), 1,296 edges, 26,029 synaptic contacts; chosen by anatomy, not by game results", "neuron_model": "rate: signed, input-normalised leaky tanh units, 3 iterations per decision (ACh +1, GABA/Glu -1, others 0)", "input_mapping": "hand-made", "output_mapping": "learned", "trained_parts": "Only the 243-parameter 16-12-3 MLP readout, trained with CEM (64 candidates, 8 elites). Circuit weights are fixed contact counts.", "body": "none", "scripted_parts": "The 8 inputs are read straight from the game state, not from pixels. Which observation drives which visual cell type is an arbitrary choice. The Flybody fly and the gray 124k-cell atlas are display only (a keyboard-rig animation and anatomical context)." }, "trained_class": "readout-or-decoder", "trained_class_note": "243-parameter MLP readout trained with CEM", "grade_basis": [ "scripts/build-connectome.py:17-47 picks 4 cells for each of 8 visual types (LC4, LC11, LC9, LC15, LC16, LC17, LC21, LPLC2) by synapses onto DNs, 16 DN targets and 32 two-hop bridges, and keeps all internal edges [direct]", "public/data/connectome/manifest.json: 'FlyEM MaleCNS v1.0, min confidence 0.5', 80 nodes, 1296 edges, source SHA-256 hashes [direct]", "src/lib/connectome.ts:3-37 leaky tanh dynamics over the normalised signed graph; 'ablated' returns zeros [direct]", "src/lib/policy.ts:41-53 8 engineered observations (distance, width, height, speed, player y, ...) [direct]", "src/lib/policy.ts:54-81 16 DN activities -> 12 tanh -> 3 scores, argmax gives the key [direct]", "src/lib/training.ts:81-158 CEM neuroevolution of the readout weights only [direct]", "src/lib/benchmark.ts:18-25 controls: silenced connectome, untrained readout, rule, random, idle; no rewired/shuffled circuit and no readout fed the raw observations [direct]" ], "grade_date": "2026-09-28", "measured_result": "public/benchmarks/benchmark.json, 100 held-out 180 s courses: connectome + trained readout 99/100 completed (mean 179.37 s); silenced connectome 0/100 (4.51 s); untrained readout 0/100 (4.49 s); random 0/100 (4.66 s); idle 0/100 (4.51 s); hand-written rule 0/100 (46.49 s). Replicates on the same courses: 85/100 and 100/100.", "try_url": "https://flydino.cobanov.dev/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/cobanov/flyjump", "code_licence": "custom: Cobanov Template Attribution License 1.0 (free use incl. commercial; visible linked credit required; not OSI-approved)", "code_licence_source": "LICENSE file 'LICENSE' read at commit e34c6614e7; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "3.81 MB transferred in our browser check (6 Oct 2026; largest file model.bin 1.71 MB). Figure from the code before the check: ~4 MB assets (brain atlas ~2.4 MB, Flybody mesh ~1.7 MB); not stated", "last_commit": { "date": "2026-10-03T13:55:26+03:00", "hash": "e34c6614e7d13a1018585f705cdeb59b9f38291d", "branch": "main" }, "pushed_at": "2026-09-12T13:37:30Z", "last_release": "none", "created_at": "2026-09-12T09:36:52Z", "stars": 20, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://flydino.cobanov.dev/", "status": "ok", "http_code": 200, "final_url": "https://flydino.cobanov.dev/", "checked_at": "2026-10-06T09:40:16.853Z", "other_links": [ { "url": "https://github.com/cobanov/flyjump", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "runs", "checked_at": "2026-10-06", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 3.81, "compute": "unclear", "first_activity_s": 12.15, "console_errors": 0, "page_errors": 0, "matches_catalogue": "yes", "note": "Ran; unclear; 3.81 MB in the first 60-90 s (largest: model.bin 1.71 MB).", "source": "Digital Fly Lab browser check of 2026-10-06 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-06-run9/#browser-heading), result file fly-dino.json" }, "last_verified": "2026-10-06", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/cobanov/flyjump (clone at HEAD c08c86b) [direct]", "https://flydino.cobanov.dev/ (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. Uses WebGL via three.js, and training runs in a web worker; no WebGPU or WASM. Only 80 neurons are simulated, and the count and selection method match the code. The 124k-cell brain view is static context. Graded C (small subset + trained readout): nearly all control skill sits in the trained readout. The controls only show that the readout needs some input (silencing gives zeros). They do not test whether the fly wiring beats a rewired or random circuit, and the author says so openly (docs/experiment.md:9,95). Licence: the 'Cobanov Template Attribution License 1.0' is custom and not OSI-approved. It allows free use, changes and commercial deployment. But any web interface or repository that uses substantial parts must show a visible, linked credit ('Built with fly-connectome-template by Mert Cobanov'). Rights end if a breach is not fixed within 30 days. Third-party data (MaleCNS CC BY 4.0, Flybody Apache 2.0, Chromium BSD) keeps its own licences. 5 Oct 2026: at commit e34c6614 (2026-10-03) the README adds trained direct-input controls with no connectome (10 seeds each): mean completion 92.2 for the connectome controller and 92.2 for a 243-parameter plain network (78.4 for 147 parameters); the author writes that the task does not need the connectome. There is still no rewired-circuit comparison." }, { "id": "fly-worker", "name": "Fly Worker", "type": "browser-demo", "author_or_org": "hwkim3330", "summary": "A browser tool that plays games with key presses, reads the screen, and records bugs so they can be replayed. One of its steering policies is a spiking model of the whole FlyWire v783 brain in a Web Worker. The author benchmarked it against random-steering baselines, and it does worse.", "claim": { "text": "A QA fuzzer for browser games. One of its policies is a real fruit-fly connectome — and that one loses.", "url": "https://github.com/hwkim3330/flyworker" }, "dataset": "FlyWire FAFB", "release": "v783", "evidence_grade": "A", "grade_note": "Borderline A: a negative result against no-graph and random baselines, but the correlation numbers are not produced by any committed script. A stricter reading gives B.", "mechanism": { "wiring": "FlyWire v783, all 138,639 neurons, pruned to 2,700,513 connections with >=5 synapses; weights rescaled to 0.44 mV", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "learned", "trained_parts": "No weight training. A 1.2 s start-up calibration flashes each eye, picks up to 20 left and 20 right descending neurons that respond differently, and normalises them (a fitted readout). Vision preprocessing gains are hand-set.", "body": "none", "scripted_parts": "No scripted steering found. Steer is the normalised right-minus-left rate of the calibrated DN channels, with an adaptive baseline and smoothing. Thrust is active-DN rate clamped to 0.25-1. The game turns steer/thrust into movement or synthetic keys." }, "trained_class": "readout-or-decoder", "trained_class_note": "start-up calibration picks and normalises readout neurons (a fitted readout)", "grade_basis": [ "tools/00_fetch.sh:6-11 downloads Connectivity_783.parquet, Completeness_783.csv and FlyWire 783 annotations [direct]", "tools/pack.py:4,18-30 packs FlyWire v783 with threshold TH=5, N=138639; docs/data/brain.bin header N=138639, M=2700513 [direct]", "docs/lif.js:10-19,80-127 uniform LIF (Shiu et al. parameters, W_SYN=0.44, Poisson drive on sensory neurons) [direct]", "docs/fly.worker.js:55-127 loads brain.bin and meta.json in a Web Worker, drives left/right LA>ME neurons from a 16x12 view grid, reports steer/thrust every 100 ms [direct]", "docs/calibrate.js:18-87 picks DN channels by flashing each eye; docs/calibrate.js:96-137 computes steer/thrust only from those channels [direct]", "policy_bench.mjs:45-95,97-123 seeded benchmark (24,000 frames): fly vs uniform noise, smoothed noise, straight-ahead, and a paired no-connectome control fed the same noise sequence as the fly+noise hybrid [direct]", "docs/policy.js:75-81 hard-codes the measured table the page shows (fly 42%, hybrid 53%, smooth 61%, uniform 50%, straight 5%) [direct]" ], "grade_date": "2026-09-28", "measured_result": "Steering vs light-direction correlation over 8 seeds is about 0: 0.046 ± 0.272 (ratio decoder), 0.077 (adaptive baseline), −0.042 ± 0.111 (threshold+tanh), 0.046 ± 0.272 (200 ms saccade) (LIMITATIONS.md:15-24; README.md:171 says 'correlation ... is 0 over 8 seeds'). Map-coverage benchmark (policy_bench.mjs; README.md:46-56; LIMITATIONS.md:188-221): fly 42% (35-50, 3 reps), smoothed noise 61% (57-66, 5 reps), uniform noise 50% (46-52), straight 5%. Paired test: fly+noise 53% (49-56) vs the same noise alone 56% (52-63), so the connectome adds −3 points, within run-to-run spread. Static left/right test: separation 0.947, 0.000 after cutting the 40 steering channels, 1.121 after restoring them (README.md:69-72).", "try_url": "https://hwkim3330.github.io/flyworker/", "try_status": "runs in headless Chromium without a GPU (checked 2026-10-03; not a play test)", "code_url": "https://github.com/hwkim3330/flyworker", "code_licence": "none found (README says MIT but no LICENSE file)", "code_licence_source": "no LICENSE/COPYING file at repo root (commit 83b910466e); GitHub API spdx_id=None", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "~13 MB (brain.bin 11.4 MB raw / ~7.45 MB gzip, meta.json 0.48 MB, pos.bin 0.97 MB), plus 4.6 MB doom.wasm for DOOM mode", "last_commit": { "date": "2026-09-27T22:36:56+09:00", "hash": "83b910466ed845a826cb291cb044eeb82dc8b956", "branch": "main" }, "pushed_at": "2026-09-27T13:36:58Z", "last_release": "none", "created_at": "2026-09-15T04:12:43Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://hwkim3330.github.io/flyworker/", "status": "ok", "http_code": 200, "final_url": "https://hwkim3330.github.io/flyworker/", "checked_at": "2026-10-06T09:40:16.908Z", "other_links": [ { "url": "https://github.com/hwkim3330/flyworker", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "runs", "checked_at": "2026-10-03", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 17.55, "compute": "in-browser-worker", "first_activity_s": null, "console_errors": 0, "page_errors": 0, "matches_catalogue": "yes", "note": "The 7.6 MB brain file loads into a Web Worker and the on-page simulation numbers keep changing; 8.2 MB more is the GTA1 game data. The screenshot never changed, so first movement is not seen visually.", "source": "Digital Fly Lab browser check of 2026-10-03 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-03-run7/#run-heading), result file fly-worker.json" }, "last_verified": "2026-10-06", "controls": "baseline-only", "wiring_effect": "not-tested", "sources": [ "https://github.com/hwkim3330/flyworker (clone read at HEAD) [direct]", "https://hwkim3330.github.io/flyworker/ (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. The simulation runs on the CPU in plain JS in a Web Worker. WebGL is used only for the three.js brain view and the GTA1/Carnage3D target, and DOOM mode uses WASM. No WebGPU. Graded A as a well-controlled negative result: a seeded, reproducible benchmark with a paired no-connectome control. Limits: the controls are no-graph and random baselines, not shuffled wiring. There are only 3-5 repeats, on the author's own test game, and the metric is map coverage, not fly behaviour. The readout is fitted at start-up. No script in the repo produces the ~0 correlation numbers; they appear only in LIMITATIONS.md. The page shows hard-coded results. Visual input goes to LA>ME, not to photoreceptors. The README says MIT, but the repo has no LICENSE file." }, { "id": "fly-x-jev", "name": "Fly x Jev", "type": "browser-demo", "author_or_org": "pasangimhana", "summary": "A browser demo of a small robot on a road. MaleCNS v1.0 wiring turns the robot's 1,771-column compound eye into LC/LPLC visual-projection activity and runs a mushroom-body memory (PN to Kenyon cell to MBON, dopamine weakens KC-MBON gains). TypeSafe's Jev model, called over an API several times a second, reads these signals and picks which of six descending or feeding neurons fires next.", "claim": { "text": "A browser demo of a small robot walking down a road, driven by wiring from the real MaleCNS v1.0 fruit fly connectome", "url": "https://github.com/pasangimhana/fly-x-jev" }, "dataset": "MaleCNS", "release": "v1.0 (selected neurons, synapse counts and derived weights committed in arena/data/)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0 subsets: eye columns to 4,664 LC/LPLC neurons (direct or via one optic-lobe intermediate, signed), PN-KC-MBON-DAN mushroom-body circuit, LC/LPLC to five command-neuron pathways", "neuron_model": "rate (precomputed sparse weights, low-pass filtered rates); no spiking whole-brain simulation", "input_mapping": "hand-made: ray-cast eye columns, synthetic odors (random 6-glomerulus draws), pain sources", "output_mapping": "hand-made menu of six actions (DNp09 walk, DNa02 veer left/right, MDN back, DNp01 takeoff, Fdg feed)", "trained_parts": "external model (TypeSafe Jev, System One API) chooses the action from a text-like state; hand-set dopamine plasticity on KC-MBON gains and visuomotor pathway gains", "body": "none (2D/3D robot in a browser scene)", "scripted_parts": "Demo sequence (arena/js/main.js:189-251); the world and pain sources" }, "trained_class": "policy-or-llm", "trained_class_note": "external model chooses the action", "grade_basis": [ "arena/js/jev.js:29,94,108,134-141 (commit 88cd607): one 'choice' question per call to /api/decide, proxied to TypeSafe; the returned choice (or a probability-weighted exploration when stalled) sets the action [direct]", "arena/js/mb.js:7,46,88: KC-MBON gains decay with dopamine and relax with a 420 s time constant; valence label passed to Jev [direct]", "README 'How it works' (same commit): eye columns, LC/LPLC weights, Jev's six-option action menu and state fields [direct]", "GitHub API 2026-10-01: created and pushed 2026-09-29T21:00Z, 4 stars, MIT [direct]", "aryap1804/fly-x-jev (created 2026-09-30T22:37Z, not a GitHub fork) is a later copy of the same project with a 'Memory Lab' of ablation experiments (silence PPL1 / PAM); logged in exclusions.md, not linked [direct: README]" ], "grade_date": "2026-10-01", "measured_result": "None reported as numbers. The demo shows avoidance after toxic contact, seeking after honey and takeoff over a barrier after pain; the action itself is chosen by Jev.", "try_url": null, "try_status": "needs a TypeSafe API key (paid service) and a local Python server; without a key the page loads but the robot does not decide ('jev offline')", "code_url": "https://github.com/pasangimhana/fly-x-jev", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 88cd6075ba; GitHub API spdx_id=MIT", "data_licence": "MaleCNS: CC-BY-4.0", "platform": [ "browser" ], "platform_note": "browser front end plus a local Python 3.9+ server that proxies the TypeSafe API", "gpu": "none", "download_size": "repository with prebuilt arena/data/ (size: see gh_meta)", "last_commit": { "date": "2026-09-30T02:30:48+05:30", "hash": "88cd6075ba806d2a7ccc7d90a9d6516ceaaeea45", "branch": "main" }, "pushed_at": "2026-09-29T21:00:55Z", "last_release": "none", "created_at": "2026-09-29T21:00:50Z", "stars": 6, "stars_date": "2026-10-05", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/pasangimhana/fly-x-jev", "status": "ok", "http_code": 200, "final_url": "https://github.com/pasangimhana/fly-x-jev", "checked_at": "2026-10-06T09:40:17.421Z" }, "last_verified": "2026-10-06", "added_in_run": "run:721af05d-86ce-46b6-9c1d-0038d2469d32", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/pasangimhana/fly-x-jev" ], "notes_limitations": "The connectome supplies sensory features and a learned valence, but an external commercial model picks every action, so the behaviour shown is not driven by a fly-brain simulation. No control (no scrambled wiring, no Jev-only baseline). Requires a paid API key to run." }, { "id": "fly67", "name": "FLY67", "type": "browser-demo", "author_or_org": "znatgost", "summary": "The whole FlyWire v783 brain (138,639 neurons, 15.1 M connections, Shiu et al. LIF parameters) runs in a browser Web Worker and drives an animated 3D fly: sugar under the proboscis drives MN9, bitter suppresses it, an air puff drives antennal grooming via aDN1, and a looming shadow recruits the giant fibre and DNa02. Descending-neuron rates are turned into movement with hand-set gains; a '6-7' meme gesture is a labelled scripted easter egg.", "claim": { "text": "A whole fruit fly brain — all 138,639 neurons and 15.1 million connections of the FlyWire connectome — simulated live in your browser tab and wired to a virtual fly", "url": "https://znatgost.github.io/fly67/" }, "dataset": "FlyWire FAFB", "release": "v783 as packaged by Shiu et al. (data/connectome.bin.gz built by tools/build_data.py)", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "Whole FlyWire v783 brain, signed synapse counts (Shiu et al. files)", "neuron_model": "spiking: Shiu et al. LIF (JavaScript, lazy exact updates; validation/ compares with Brian2 2.9 reference runs)", "input_mapping": "hand-made: sugar, bitter and water GRNs, Johnston's organ, LC4/LPLC2 driven as Poisson inputs by arena events", "output_mapping": "hand-made: MN9, giant fibre, aDN1, P9, DNa02, MDN rates times fixed MOTOR_GAIN values set body kinematics", "trained_parts": "none", "body": "custom (procedural animated 3D fly, not physically simulated)", "scripted_parts": "the '6-7' gesture every 15-25 s (labelled as scripted); gait animation" }, "trained_class": "none", "grade_basis": [ "src/engine.js:17-20,91-98 (commit 45f62fd): Shiu et al. constants (tau 5 ms, refractory 2.2 ms, w_syn 0.275 mV) and closed-form LIF updates [direct]", "src/circuits.js:5,29-45: sensors (sugar/bitter/water GRNs) and readouts (MN9 = CB0701, DNa02 left/right, ...) chosen by the author [direct]", "src/arena.js:9,34-36,94-100: MOTOR_GAIN mapping and the scripted '6-7' easter egg ('NOT produced by the connectome') [direct]", "validation/ folder: Brian2 2.9 reference files and a lazy-vs-dense equivalence check (not run by us) [direct: file list]", "GitHub API 2026-10-01: MIT; last commit 2026-09-29 [direct]" ], "grade_date": "2026-10-01", "measured_result": "README: bitter laced sugar lowers MN9 about 14x (52 to 3.7 Hz); not re-run by us and not compared with a control. 5 Oct 2026, our own test (Brian2 2.9.0, Shiu et al. model, not fly67's engine): fly67's looming rule held on both sides (left loom: giant fibre 111.5 Hz, contralateral DNa02 23.0 vs 0.0 Hz; right loom: 119.0 Hz, 44.0 vs 0.0 Hz; fly67's committed values: 114.8 Hz, 25 vs 0 and 120.3 Hz, 43 vs 0). With degree-preserving shuffles the giant fibre and DNa02 were 0 Hz (n 3). In a FlyGym body (our mapping, constant external drive) the real brain turned the fly away in 4 of 4 runs and the scrambled brains in 0 of 3 (controls ledger study byo-flygym-loom). The 'away' direction rests on our mapping's DNa02 sign and on fly67's claim; we did not verify it in real flies. A primary source (Card and Dickinson 2008) shows flies jumping away from a looming threat at takeoff, not walking flies turning.", "try_url": "https://znatgost.github.io/fly67/", "try_status": "runs in headless Chromium without a GPU (checked 2026-10-03; not a play test)", "code_url": "https://github.com/znatgost/fly67", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 45f62fd9b8; GitHub API spdx_id=NOASSERTION", "data_licence": "FlyWire FAFB: CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "33.3 MB transferred in our browser check (connectome.bin.gz in two parts, 15.7 + 15.5 MB, and neurons.bin.gz 1.8 MB; 2026-10-03), not the ~100 MB we estimated before", "last_commit": { "date": "2026-09-29T23:22:04+04:00", "hash": "45f62fd9b867632141f84cc6daa77d871a2259bb", "branch": "main" }, "pushed_at": "2026-09-29T19:22:05Z", "last_release": "none", "created_at": "2026-09-29T18:13:28Z", "stars": 0, "stars_date": "2026-10-06", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://znatgost.github.io/fly67/", "status": "ok", "http_code": 200, "final_url": "https://znatgost.github.io/fly67/", "checked_at": "2026-10-06T09:40:17.048Z", "other_links": [ { "url": "https://github.com/znatgost/fly67", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "runs", "checked_at": "2026-10-03", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 33.28, "compute": "in-browser-worker", "first_activity_s": 11.57, "console_errors": 0, "page_errors": 0, "matches_catalogue": "no", "note": "Two connectome parts (15.7 + 15.5 MB) load into a Web Worker; the fly moves and the counters change; the sugar button worked. Our catalogue's '~100 MB' estimate was wrong: 33.3 MB.", "source": "Digital Fly Lab browser check of 2026-10-03 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-03-run7/#run-heading), result file fly67.json" }, "last_verified": "2026-10-06", "added_in_run": "run:721af05d-86ce-46b6-9c1d-0038d2469d32", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/znatgost/fly67", "https://znatgost.github.io/fly67/" ], "notes_limitations": "A faithful browser port of the Shiu model with hand-made inputs and outputs. The body is animated from a few descending-neuron rates; there is no ventral nerve cord or physics. No scrambled-wiring control. One gesture is scripted and labelled. Browser check 2026-10-03: the brain runs in a Web Worker as recorded; the download is 33.3 MB, not the ~100 MB in our earlier record (corrected). 5 Oct 2026: our own Brian2 run of the Shiu model reproduced the looming 'turn away' rule on both sides, and degree-preserving shuffles abolished it (controls ledger study byo-flygym-loom; this is our test, not a control by the author)." }, { "id": "flybrain-app-snedea", "name": "FlyBrain", "type": "browser-demo", "author_or_org": "snedea", "summary": "A 2D browser pet fly you can feed, touch, blow air at, or expose to light and temperature. A Web Worker runs spikes over the full FlyWire brain graph, and a panel shows them. What the fly does on screen is chosen by a threshold state machine and hand-coded steering, with the network's group activity as one input.", "claim": { "text": "139,255 neurons and 2.7M connections from the FlyWire FAFB v783 connectome run in real time via a leaky integrate-and-fire model", "url": "https://github.com/snedea/flybrain" }, "dataset": "FlyWire FAFB", "release": "v783", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "FlyWire FAFB v783, 139,255 neurons / 2,698,236 edges (syn_count x neurotransmitter sign), normalised; neurons sorted into 63 hand-made functional groups", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none in the brain; the page says a language-model agent ('Claude') reviews requests filed by a second agent that reads the fly's drive meters, and its approvals change the fly's world (food, light) (page text, 5 Oct 2026)", "body": "custom", "scripted_parts": "The data has 0 neurons in the leg and wing motor groups, so a hand-weighted 'virtual VNC' makes up the motor outputs from group activity. A threshold state machine picks the behaviour (walk, explore, feed, startle, fly, groom, rest, phototaxis). Food-seeking steers toward the food with atan2, and phototaxis steers toward the canvas centre. Hunger, fear, fatigue and curiosity are hand-coded counters. If the binary fails to load, a 59-group legacy model is used." }, "trained_class": "policy-or-llm", "trained_class_note": "a language-model agent reviews requests and changes the fly's world (food, light); the brain itself has no trained part", "grade_basis": [ "js/brain-worker-bridge.js:93 loads data/connectome.bin.gz into the worker [direct]", "data/neuron_meta.json: neuron_count 139255, edge_count 2698236; all MN_LEG_*, MN_WING_*, DN_WALK/FLIGHT/TURN groups have 0 neurons [direct]", "js/sim-worker.js:3,25-28,318 LIF (leak 0.95, threshold 1.0, weight scale 0.15) over CSR; groups with no recent activity are skipped (neuropil gating), 10 ticks/s [direct]", "js/brain-worker-bridge.js:252 'virtual VNC motor layer' builds leg/wing outputs from group activity with hand-set weights; :321 says steering is done by the behavioural layer, not by the network [direct]", "js/fly-logic.js:18,54 BEHAVIOR_THRESHOLDS and evaluateBehaviorEntry choose the behaviour state [direct]", "js/main.js:1045-1049 food-seeking steers by atan2 toward the food; js/main.js:1064 phototaxis steers toward the canvas centre [direct]", "js/connectome.js:175,182 drives are hand-coded increments (hunger +0.005 per tick, fear +0.3 on touch) [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://flybrain.app", "try_status": "runs in headless Chromium without a GPU (checked 2026-10-03; not a play test)", "code_url": "https://github.com/snedea/flybrain", "code_licence": "MIT", "code_licence_source": "LICENSE file 'license.md' read at commit 9191824d17; GitHub API spdx_id=MIT", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "~12.4 MB (connectome.bin.gz) plus small metadata", "last_commit": { "date": "2026-03-27T22:17:18-05:00", "hash": "9191824d17871b7851645782d53d23f213ddb938", "branch": "main" }, "pushed_at": "2026-08-13T21:13:05Z", "last_release": "none", "created_at": "2026-03-27T23:49:12Z", "stars": 161, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://flybrain.app", "status": "ok", "http_code": 200, "final_url": "https://flybrain.app", "checked_at": "2026-10-06T09:40:17.149Z", "other_links": [ { "url": "https://github.com/snedea/flybrain", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "runs", "checked_at": "2026-10-03", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 12.72, "compute": "unclear", "first_activity_s": 10.87, "console_errors": 2, "page_errors": 0, "matches_catalogue": "yes", "note": "The 12.5 MB connectome loads and a Web Worker runs; the view keeps changing. The page also tries a chat WebSocket and a chat-history server (both failed here), so by our fixed rule where the brain runs is 'unclear'.", "source": "Digital Fly Lab browser check of 2026-10-03 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-03-run7/#run-heading), result file flybrain-app-snedea.json" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/snedea/flybrain (clone, files read raw) [direct]", "https://flybrain.app (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. The whole-brain LIF does run on the CPU in a Web Worker, but only groups with recent activity are updated each tick. WebGL2 is used only for the neuron display. The README says the fly is not scripted and that behaviour 'emerges', but the code picks behaviours with thresholds and steers toward food and light with hand-written geometry. The README names data/neuron_meta.bin.gz, but the code loads data/connectome.bin.gz. The project is forked from a worm-sim demo. No control or benchmark. Browser check 2026-10-03: besides the brain worker, the page tries a chat WebSocket (wss://caretaker.flybrain.app/) and a chat-history endpoint on port 7600; both failed here. 5 Oct 2026 (plain GETs of index.html and js/caretaker-*.js): wss://caretaker.flybrain.app is a public tunnel to a caretaker server that receives the fly's drives, behaviour and position once per second; port 7600 is the same server run locally (chat and analytics, hidden on the public site). The page states that a language model ('Claude') acts as caretaker and approves or denies requests, which drops food or dims the lights. The server code is not public, so which model runs there was not verified." }, { "id": "flybrain-info-lab", "name": "flybrain.info Lab", "type": "browser-demo", "author_or_org": "not stated (flybrain.info, English/Russian site)", "summary": "A section of the flybrain.info site where you pick one of 8 stimuli (sugar, bitter, water, sound, looming, geosmin, light, 'runaway') and watch a 3D cloud of FlyWire neurons flash. The browser plays back spike runs computed offline with the site's own numpy re-implementation of the Shiu et al. model; nothing is simulated live. The same site is also a project directory (competitor listing).", "claim": { "text": "Every flash is a spike in a run of the Shiu et al. model (Nature, 2024) on the full FlyWire v783 connectome", "url": "https://flybrain.info/#lab" }, "dataset": "FlyWire FAFB", "release": "v783 (stated)", "evidence_grade": "U", "grade_note": null, "mechanism": { "wiring": "FlyWire v783 as stated (138,639 neurons, 15,091,983 connections) for the offline runs; the browser loads only positions, classes and types for 139,248 neurons", "neuron_model": "LIF (stated Shiu et al. re-implementation in numpy, offline; not run in the browser)", "input_mapping": "hand-made", "output_mapping": "none", "trained_parts": "none stated", "body": "none", "scripted_parts": "The whole Lab is playback: 8 precomputed spike files (/data/p/.bin, first of 10 trials) and precomputed mean rates from /data/presets.json. No neuron integration happens in the page." }, "trained_class": "not-assessed", "grade_basis": [ "https://flybrain.info/main.js loadPreset() fetches /data/p/${id}.bin and replays it with Replay.advance(); the hero view cycles 5 stored runs [direct]", "https://flybrain.info/brain.js Replay class (about lines 323-348) reads stored spike indices and times plus per-neuron mean rates and steps through them; there is no integration step [direct]", "https://flybrain.info/data/presets.json: 8 presets with stimulated-neuron counts, rates (40-220 Hz), trials 10, active-neuron counts and top responders [direct]", "https://flybrain.info/data/brain.json: count 139248 (positions/classes only) [direct]", "Page text states a numpy re-implementation with 0.94 firing-rate correlation vs Brian2 on the sugar experiment; no code, repository or data for this model is linked [direct]" ], "grade_date": "2026-09-28", "measured_result": "The site states that its numpy version of the Shiu model activates the same neurons as the Brian2 reference in the sugar experiment, with a 0.94 correlation of firing rates (https://flybrain.info/#lab). No code or data is given, so this cannot be checked.", "try_url": "https://flybrain.info/#lab", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": null, "code_licence": "n/a (no code repository)", "code_licence_source": "n/a", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "not stated (brain.bin plus 8 small replay files)", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "n/a", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://flybrain.info/#lab", "status": "ok", "http_code": 200, "final_url": "https://flybrain.info/#lab", "checked_at": "2026-10-06T09:40:17.649Z" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://flybrain.info/ (HTML fetched with node) [direct]", "https://flybrain.info/main.js and https://flybrain.info/brain.js [direct]", "https://flybrain.info/data/presets.json and https://flybrain.info/data/brain.json [direct]" ], "notes_limitations": "Page responds (HTTP 200); not played. Needs WebGL (three.js) only. Graded U because the model that made the runs is not published: the browser shows stored results, not a live simulation. The page itself says the 3D view replays the first of 10 runs and the list shows averages. It is open that this fly has no body and that smell stimuli had to be given at low rates, because the model 'runs away' into a shared ~10,000-neuron state. Also a competitor directory: it lists projects (/projects/) and hosts copies of cyber-larva and fly-worker under /play/. No author or licence is stated for the site or its data derivatives. Directory text was not copied." }, { "id": "flydrones", "name": "FlyDrones", "type": "browser-demo", "author_or_org": "SpikeCalls", "summary": "A Python package and a browser demo. Camera optic flow drives chosen fly visual neuron types in a LIF network, and six descending-neuron groups are read out as throttle and yaw for a simulated or real drone. The browser demo runs 'MiniFly', an 850-neuron hand-wired stand-in; MaleCNS v1.0 is an optional path in the Python package.", "claim": { "text": "A fruit fly's wiring diagram is flying this drone.", "url": "https://github.com/SpikeCalls/FlyDrones" }, "dataset": "MaleCNS", "release": "v1.0 (optional; default brain is synthetic MiniFly)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "Default and browser: MiniFly, a synthetic 850-neuron / 4,928-edge network given real cell-type names (not connectome data). Optional: MaleCNS v1.0 flat tables, >=3 synapses, transmitter signs", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "Optional ridge-regression readout from DN rates to throttle/yaw (flydrones calibrate); weights fixed", "body": "custom", "scripted_parts": "Tonic flying bias on DNg02. A giant-fibre rate above a threshold triggers a fixed 0.7 s climb/drop program. A safety governor can override the brain, and the drone's own flight controller keeps it level. The demo gestures are scripted optic-flow patterns." }, "trained_class": "none", "trained_class_note": "optional ridge readout via calibrate", "grade_basis": [ "docs/live/app.js:15 the browser demo fetches ./minifly.json; docs/live/minifly.json name 'minifly-synthetic', n=850, nnz=4928 [direct]", "docs/index.html:123 banner: 'MiniFly: 850-neuron synthetic stand-in with real fly cell types, running in your browser' [direct]", "src/flydrones/defaults.yaml:7 brain source defaults to 'minifly'; src/flydrones/brain/synthetic.py:1-3 says MiniFly is not real fly data [direct]", "src/flydrones/brain/connectome.py:214-219,230-329 optional MaleCNS v1.0 builder (min 3 synapses, NT signs); src/flydrones/brain/lif.py:1-40 LIF with Shiu et al. parameters [direct]", "src/flydrones/motor/decoder.py:85-99 escape threshold triggers a fixed throttle burst [direct]", "ROADMAP.md:11 'MaleCNS group presets verified against neuPrint' still unchecked [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://spikecalls.github.io/FlyDrones/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/SpikeCalls/FlyDrones", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit 3e269346b3; GitHub API spdx_id=MIT", "data_licence": "CC-BY-4.0", "platform": [ "browser", "macOS", "Windows", "Linux" ], "gpu": "none", "download_size": "~1.2 GB MaleCNS data (optional; MiniFly needs none)", "last_commit": { "date": "2026-09-16T06:18:23+00:00", "hash": "3e269346b3882c291d2a977bc2c2c6a9c9213c21", "branch": "main" }, "pushed_at": "2026-09-16T06:18:25Z", "last_release": "none", "created_at": "2026-09-15T20:52:47Z", "stars": 232, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://spikecalls.github.io/FlyDrones/", "status": "ok", "http_code": 200, "final_url": "https://spikecalls.github.io/FlyDrones/", "checked_at": "2026-10-06T09:40:17.639Z", "other_links": [ { "url": "https://github.com/SpikeCalls/FlyDrones", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/SpikeCalls/FlyDrones [direct: local clone, HEAD 3e26934]", "https://spikecalls.github.io/FlyDrones/ [direct]" ], "notes_limitations": "CONFLICT resolved: yes, the browser demo at spikecalls.github.io/FlyDrones runs MiniFly, an 850-neuron hand-wired synthetic network, not the connectome (docs/live/app.js:15, minifly.json, docs/index.html:123). The page itself says so. The non-browser Python package can build a MaleCNS v1.0 graph and run it with Shiu-style LIF, feeding optic flow in and giving throttle/yaw out to a simulator or drone adapter. That path alone would be B-level, but it is not the default. The repo shows no MaleCNS run output: all GIFs, the swarm run and the dashboard are MiniFly, and the full-size figure is only a speed benchmark on a random graph. MaleCNS neuron-group presets are marked unverified. Hardware adapters are not flight-tested (README). There is no control or real-fly comparison. CITATION.cff names 'Claude' as author, which suggests AI-written code. Licence MIT." }, { "id": "flyleno", "name": "FlyLeno (TUURD Talk)", "type": "browser-demo", "author_or_org": "AgitationSkeleton", "summary": "A browser talk show in which the whole FlyWire v783 brain, run live as the Shiu et al. spiking model in a Web Worker, drives a ragdoll Grey Leno. Show events, music and a thrown sugar cube stimulate chosen sensory neurons; descending-neuron firing rates are mapped by hand to walking, turning, startle, grooming, feeding, voice and body muscles. Gait, balance, food seeking, homing, sleep and the choice of spontaneous actions are engineered layers that the README labels as such.", "claim": { "text": "Tonight's host: Grey Leno, piloted by Drosophila melanogaster", "url": "https://flyleno.viosarcade.xyz/" }, "dataset": "FlyWire FAFB", "release": "v783 via Shiu et al. Completeness_783.csv / Connectivity_783.parquet (138,639 neurons, 15.1 M connections, all synapse counts kept)", "evidence_grade": "B", "grade_note": "Borderline C: the brain really runs and its reflexes move the host, but walking rhythm, homing and the choice of what to do next are engineered layers, and there is no control. A scrambled-wiring or no-brain run of the same show could raise it to A; evidence that the engineered layers drive most of what is on screen would lower it to C.", "mechanism": { "wiring": "Whole FlyWire v783 brain from the Shiu et al. files, signed synapse counts; dopamine-gated plasticity (on by default) changes the weights of synapses onto the voice (articulator) motor neurons while the show runs", "neuron_model": "spiking: Shiu et al. leaky integrate-and-fire (event-driven JavaScript port in a Web Worker); optional spike-frequency adaptation, off by default", "input_mapping": "hand-set: show events, thrown objects, music loudness bands and food touch stimulate chosen sensory groups (sugar GRNs, Johnston's organ, looming, taste) as Poisson inputs; stimulus neuron IDs from erojasoficial-byte/fly-brain", "output_mapping": "hand-set: descending-neuron rates (P9/oDN1 forward, MDN backward, DNa01/02 turn, giant fiber startle, aDN1 groom, MN9 feed) divided by fixed saturation gains (40 Hz; 15 Hz for startle); ~1,250 other DNs pooled onto 46 ragdoll muscles; voice from song/flight DNs and mouthpart motor neurons", "trained_parts": "none offline; online dopamine-gated plasticity onto voice motor neurons, and an engineered action selector (softmax over values learned from the model's own dopamine) that starts spontaneous actions by stimulating descending neurons", "body": "custom (Rapier physics ragdoll of Grey Leno or a giant fly, three.js)", "scripted_parts": "Gait pattern generator ('VNC', engineered), balance 'puppet strings', saccadic turns and walking bouts, food taxis, homing (engineered input onto DNa01/02 and P9), sleep, the show director and all stage events; the README table marks each as engineered" }, "trained_class": "other", "trained_class_note": "online plasticity and an engineered value-based action selector", "grade_basis": [ "tools/build_connectome.py:3-22,41 builds data/connectome.bin.gz from philshiu Completeness_783.csv and Connectivity_783.parquet, min_syn 1 as Shiu [direct]", "js/brain-worker.js:5-14 Shiu LIF equations (dv/dt, dg/dt, v_th reset, Poisson stimulus kicks); :462 loads data/connectome.bin.gz; :42 dopamine plasticity PL.enabled true; :47 adaptation off by default [direct]", "js/motor.js:3-40 decodeMotor: rates of forward/backward/turnL-R/startle/groom/feed divided by motorGains (40 Hz, startle 15 Hz), EMA 250 ms [direct]", "js/brain-worker.js:39,205-207,253 dopamine-gated plasticity only on synapses onto the plastic readout (articulator) neurons [direct]", "js/instincts.js:1-25 neural vs engineered list (sugar GRNs -> MN9 decides eating; food taxis, homing, saccades, bouts engineered) [direct]", "js/mind.js:1-24 action selector outside the connectome: softmax(Q/T) over talk, stroll, turn, back up, fart, retch, rest; Q learned from DA = tanh((PAM - PPL1)/20 Hz); a chosen action stimulates the matching descending/motor neurons ('fictive drives') [direct]", "README.md 'What drives what' table: each behaviour marked neural or engineered [direct]", "No wiring control or no-brain baseline anywhere in the repository (grep for shuffle/scramble/rewire finds only playlist and animation code) [direct]" ], "grade_date": "2026-09-30", "measured_result": "No measured result or control in the repository.", "try_url": "https://flyleno.viosarcade.xyz/", "try_status": "loads and fetches its brain data, then froze in headless Chromium without a GPU (checked 2026-10-03; not a play test; a GPU machine was not tested)", "code_url": "https://github.com/AgitationSkeleton/flyleno", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit 078d8fa748); GitHub API spdx_id=None", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "platform_note": "static site; needs a desktop browser with Web Workers; music via an embedded YouTube playlist", "gpu": "none", "gpu_note": "runs in a desktop browser; the 31 MB compressed connectome is downloaded on start", "download_size": "about 32 MB of connectome and position data plus 3-D assets, fetched by the page", "last_commit": { "date": "2026-09-27T16:52:30-07:00", "hash": "078d8fa748e8995e4d67051b377e9eb0bfa40bc7", "branch": "main" }, "pushed_at": "2026-09-27T23:52:08Z", "last_release": "none", "created_at": "2026-09-22T23:51:41Z", "stars": 0, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://flyleno.viosarcade.xyz/", "status": "ok", "http_code": 200, "final_url": "https://flyleno.viosarcade.xyz/", "checked_at": "2026-10-06T09:40:17.728Z", "other_links": [ { "url": "https://github.com/AgitationSkeleton/flyleno", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "page-unresponsive", "checked_at": "2026-10-03", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 33.95, "compute": "in-browser-worker", "first_activity_s": 10.44, "console_errors": 0, "page_errors": 0, "matches_catalogue": "unclear", "note": "The page fetched the 31.4 MB connectome into a Web Worker and its counters started, then the 3-D stage (software WebGL, no GPU) stopped answering after about 10 s; no server calls.", "source": "Digital Fly Lab browser check of 2026-10-03 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-03-run7/#run-heading), result file flyleno.json" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/AgitationSkeleton/flyleno [direct: clone HEAD 078d8fa, 2026-09-30]", "https://flyleno.viosarcade.xyz/ [direct: index.html and js/main.js, 2026-09-30]", "https://www.youtube.com/watch?v=BOSg6HYVX1o [video: 36,763 views, 2026-09-30 09:41Z]" ], "notes_limitations": "The whole v783 brain really runs, with no offline training, and the README labels every engineered layer. But the show is built around it: gait, balance, food seeking, homing and sleep are ordinary code, spontaneous actions are chosen by an engineered selector that then stimulates descending neurons, input neurons and output gains are chosen by hand, and plasticity rewrites voice synapses during the show. Reactions (startle, eating, grooming) do come from the model's own neurons. There is no scrambled-wiring or no-brain control, so how much of Leno's behaviour needs the real wiring is not shown. Borderline B/C. The viral video is a streamer's playthrough, not the author's." }, { "id": "flywire-pong", "name": "FlyWire Pong", "type": "browser-demo", "author_or_org": "zonzujiro", "summary": "A browser page in which a 1,230-neuron network with FlyWire FAFB v783 topology plays Atari Pong (the official ALE emulator in WebAssembly) against the built-in computer player. Selected visual neurons receive a brightness and motion encoding on an 11x11 grid; two descending neurons (DNp15, DNp01) feed a small action decoder. The weights were trained (supervised start, then PPO) in a separate project; the page runs inference only and includes an inspector for every decision.", "claim": { "text": "\"The selected checkpoint won 43 of 60 completed clean-Pong games in Python ... These results do not establish a causal advantage of fly topology and are not a measured browser win rate.\" (README)", "url": "https://github.com/zonzujiro/flywire-pong" }, "dataset": "FlyWire FAFB", "release": "v783 (FAFB783 neuron IDs in model.json): 1,230 selected neurons and 8,515 directed edge rows with trained weights", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "FlyWire FAFB v783 subgraph of 1,230 selected neurons (visual inputs to DNp15 and DNp01); topology from FlyWire, weights trained", "neuron_model": "rate: 12 synchronous recurrent updates per decision with held input and reset between decisions (network.mjs)", "input_mapping": "hand-made: brightness and temporal encoding of the game frame on 114 mapped columns of an 11x11 grid into 355 visual neurons; placement inferred from connectivity and column annotations", "output_mapping": "learned: a small action decoder on two descending readouts (DNp15, DNp01); actions are sampled", "trained_parts": "connection weights, biases, encoder arrays and the action decoder (supervised initialisation, then PPO, in the author's separate training project)", "body": "none", "scripted_parts": "none in the policy (no ball-following fallback, per the README); the opponent is the Atari game's own player" }, "trained_class": "whole-network-or-per-synapse", "trained_class_note": "supervised start then PPO on the subgraph's weights, encoder and action decoder; the page runs inference only", "grade_basis": [ "README.md at 769597a: network description, training lineage and the author's statement that results do not establish a topology advantage [direct]", "verification/python-checkpoint.json: checkpoint sha256 17c27074..., validation 23/30 wins and confirmation 20/30 wins in Python [direct]", "verification/port.json and tests/parity.mjs: browser inference matches 30 recorded Python frame pairs (README: max probability difference 0.000007871) [direct, file list and README]", "No shuffled-wiring or non-connectome baseline in the repository [direct]" ], "grade_date": "2026-10-06", "measured_result": "Author's committed receipt: the selected checkpoint won 43 of 60 Python games (23/30 and 20/30). No control.", "try_url": "https://zonzujiro.github.io/flywire-pong/", "try_status": "static GitHub Pages site linked from the README (emulator and network run in a browser worker); not loaded by us", "code_url": "https://github.com/zonzujiro/flywire-pong", "code_licence": "GPL-2.0", "code_licence_source": "LICENSE file at 769597a (GNU GPL version 2); vendor/ holds the unmodified ALE WebAssembly build with its upstream licence and a source archive", "data_licence": "FlyWire v783 derived (neuron IDs and trained weights in model.json); no data licence stated in the repository", "platform": [ "browser" ], "platform_note": "static site; tests need Node.js 22 or newer", "gpu": "none", "download_size": "repository about 13.5 MB, including an 11 MB ALE source archive (vendor/ale-source-v0.12.0.zip, not opened)", "last_commit": "n/a", "pushed_at": "n/a", "last_release": "none", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://zonzujiro.github.io/flywire-pong/", "status": "ok", "http_code": 200, "final_url": "https://zonzujiro.github.io/flywire-pong/", "checked_at": "2026-10-06T09:56:56.954Z", "other_links": [ { "url": "https://github.com/zonzujiro/flywire-pong", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "added_in_run": "run:c0ecce13-d7a1-40c8-b30e-661b847df700", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/zonzujiro/flywire-pong [direct]" ], "notes_limitations": "A small trained network that borrows FlyWire topology, not a simulation of a fly brain; the author says so. The training project and data are not in this repository, and no control compares the fly topology with random wiring. The bundled zip is the emulator's upstream source archive (by name and the README notice)." }, { "id": "help-the-fly-escape", "name": "Help the Fly Escape", "type": "browser-demo", "author_or_org": "dzhng", "summary": "A 3D browser puzzle game: you place household objects, then a swarm of flies tries to find the exit. Each fly runs its own noisy LIF simulation of a 70,000-neuron MaleCNS subgraph in Rust/WASM. Readouts from descending and motor neurons set each fly's walking thrust, turning and takeoff.", "claim": { "text": "A little house. A swarm of flies. A real connectome behind every wrong turn.", "url": "https://github.com/dzhng/fly-escape" }, "dataset": "MaleCNS", "release": "v1.0", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "subset: 70,000 neurons / 798,715 edges of MaleCNS v1.0 (seed-touching extraction, minimum weight 5)", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none", "body": "custom", "scripted_parts": "Readouts are fixed formulas: the mean voltage of left/right DN and MN sets gives thrust and turn, and the olfactory spike difference is scaled by a calibrated 4.0. Walking, flying and landing switch at hand-set thresholds. An authored 'exit suction' airflow near doorways helps flies out. The kinematic body has no leg gait." }, "trained_class": "none", "grade_basis": [ "data/processed/brain/manifest.json: dataset 'MaleCNS v1.0', neuronCount 70000, edgeCount 798715, minimumWeight 5, synthetic false [direct]", "packages/sim-client/src/attempt-worker.ts:52 fetches /brain/graph.bin into the WASM AttemptSession [direct]", "crates/sim/src/lif.rs:21-27 LIF parameters (tau 20, threshold 1, noise 0.015); :296 membrane update with synaptic input, external current and noise [direct]", "crates/sim/src/lif.rs:351 motor_output maps DN/MN mean voltages and olfactory spike fractions to thrust, turn and flight [direct]", "crates/sim/src/body.rs:693 takeoff when flight_thrust > threshold 0.2; :733 adds authored exit_suction_velocity [direct]", "specs/done/neural-vision/README.md:23,29 controlled neural tests pass, but approach/avoidance behaviour is not established [direct]" ], "grade_date": "2026-09-28", "measured_result": "specs/done/neural-vision/README.md: 7 prespecified directional/intensity contrasts pass with Bonferroni correction over 847 comparisons. Input silencing reproduces the dark-input trajectory. Motor and behavioural (approach/avoidance) effects were not established. None of this is a wiring control.", "try_url": "https://fly-escape.vercel.app/", "try_status": "loads, then froze before fetching its brain data in headless Chromium without a GPU (checked 2026-10-03; not a play test; a GPU machine was not tested)", "code_url": "https://github.com/dzhng/fly-escape", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit bff49a376f); GitHub API spdx_id=None", "data_licence": "CC-BY-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "~11 MB brain data (graph.bin 9.9 MB, manifest 0.6 MB, retinal map 0.2 MB) plus WASM and 3D assets; total not measured", "last_commit": { "date": "2026-09-18T03:57:16+07:00", "hash": "bff49a376f0844c918eb7f2be83e95f2699b0d14", "branch": "main" }, "pushed_at": "2026-09-17T20:58:08Z", "last_release": "none", "created_at": "2026-09-06T03:09:44Z", "stars": 49, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://fly-escape.vercel.app/", "status": "ok", "http_code": 200, "final_url": "https://fly-escape.vercel.app/", "checked_at": "2026-10-06T09:40:17.758Z", "other_links": [ { "url": "https://github.com/dzhng/fly-escape", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "page-unresponsive", "checked_at": "2026-10-03", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 2.57, "compute": "unclear", "first_activity_s": null, "console_errors": 0, "page_errors": 0, "matches_catalogue": "unclear", "note": "Loaded its WebAssembly (1 MB) and a worker, but the page stopped answering from the first screenshot and the 9.9 MB brain graph was never fetched in 84 s.", "source": "Digital Fly Lab browser check of 2026-10-03 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-03-run7/#run-heading), result file help-the-fly-escape.json" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/dzhng/fly-escape (clone, files read raw) [direct]", "https://fly-escape.vercel.app/ (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. Needs WebAssembly and WebGL (three.js); the simulation runs on the CPU. The lab view has 20 flies and retinal mode at most 16, and each fly runs its own 70k-neuron brain (about 42% of MaleCNS, not the whole CNS). The README is careful and calls the input and movement rules models. The authored doorway 'exit suction' is not mentioned in the README. The project's controls test input propagation (silencing, sham ablation), not the wiring itself. No licence file." }, { "id": "neural-canvas", "name": "Neural Canvas (Fruit Fly Simulation)", "type": "browser-demo", "author_or_org": "Xenova (Hugging Face Space; commit author 'Joshua')", "summary": "You paint neuron positions on a map of the brain, or press walk/turn/fly buttons. The browser then runs a Shiu-style spiking model of the whole MaleCNS v1.0 brain and nerve cord, and descending-neuron firing rates move an animated NeuroMechFly body.", "claim": { "text": "Simulated neural activity drives crafted walking, turning, and flight animations. The movements are illustrative, not validated predictions of fly behavior.", "url": "https://huggingface.co/spaces/Xenova/fruit-fly-simulation" }, "dataset": "MaleCNS", "release": "v1.0", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "MaleCNS v1.0: 166,700 neurons (all annotated rows with a superclass, brain + nerve cord), 25,582,938 directed edges, no synapse threshold; histamine/unknown transmitters get 0 weight (~11.6k neurons have no fast output)", "neuron_model": "LIF (Shiu et al. 2024 parameters, event-driven, dt 0.1 ms)", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none", "body": "custom", "scripted_parts": "Body motion is kinematic: an authored tripod gait with joint-limited IK, and a bounded flight/landing controller that starts when the DNp01 rate passes 100 Hz. The walk/turn/fly buttons stimulate chosen cells (LC9, LC4). The turn buttons also stimulate the DNa02 readout neurons directly, so turning can appear without recurrent transmission. No contact physics and no sensory feedback from body to brain." }, "trained_class": "none", "grade_basis": [ "public/model.json 'data' block: dataset 'MaleCNS v1.0', neurons 166700, directedEdges 25582938, source male-cns.janelia.org, CC BY 4.0 [direct]", "public/data/manifest.json: source feather files from storage.googleapis.com/flyem-male-cns/v1.0/... with SHA-256 hashes; zeroFastCurrentNeurons 11609 [direct]", "src/brain.js:1-12 LIF constants (dt 0.1 ms, rest -52, threshold -45 mV, tau 20/5 ms, 0.275 mV per synapse); src/brain.js:66-129 event-driven LIF over the outgoing CSR with a 1.8 ms delay queue and Poisson drive [direct]", "src/worker.js:31,41 the WebGPU path is checked against the JavaScript reference and falls back to JS on failure [direct]", "src/stimulus.js:59-66 readout groups (DNp09/DNg100/DNg97 walk, DNa02/DNa11/DNg13 turn, MDN reverse, DNp01 escape) and stimulus targets (LC9 walk, LC4 fly, plus DNa02 direct for turns) [direct]", "src/controller.js:49-66 smoothed DN rates map to ground speed (8*tanh) and yaw rate (3.8*tanh); DNp01 > 100 Hz starts the authored escape flight [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://huggingface.co/spaces/Xenova/fruit-fly-simulation", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://huggingface.co/spaces/Xenova/fruit-fly-simulation", "code_licence": "MIT (app code per LICENSE file; MaleCNS data CC BY 4.0; body and kernels Apache-2.0). Space metadata says apache-2.0", "code_licence_source": "LICENSE file in the Hugging Face Space repo (shallow clone 2026-09-28) and Space README metadata", "data_licence": "CC-BY-4.0", "platform": [ "browser" ], "gpu": "optional", "download_size": "~77 MB compressed connectome arrays (sum of manifest parts 76,667,180 bytes) plus body meshes; cached after first load", "last_commit": { "date": "2026-09-06T23:33:42+00:00", "hash": "776d115ee5aa934578a87fd6d260d138084f59c1", "branch": "main" }, "pushed_at": "n/a", "last_release": "none", "created_at": "n/a", "stars": null, "stars_date": null, "archived": null, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://huggingface.co/spaces/Xenova/fruit-fly-simulation", "status": "ok", "http_code": 200, "final_url": "https://huggingface.co/spaces/Xenova/fruit-fly-simulation", "checked_at": "2026-10-06T09:40:17.800Z" }, "browser_check": { "status": "runs", "checked_at": "2026-10-06", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 5.87, "compute": "unclear", "first_activity_s": 39.67, "console_errors": 0, "page_errors": 0, "matches_catalogue": "no", "note": "Ran by our rule (the view moved), but only 5.87 MB came down in the first 60-90 s and no Web Worker started; its ~77 MB brain download was not seen in our check window.", "source": "Digital Fly Lab browser check of 2026-10-06 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-06-run9/#browser-heading), result file neural-canvas.json" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://huggingface.co/spaces/Xenova/fruit-fly-simulation (shallow clone, LFS skipped; README, LICENSE, public/model.json, public/data/manifest.json, src/*.js read raw) [direct]", "https://xenova-fruit-fly-simulation.static.hf.space/index.html (HTTP 200) [direct]", "https://fruitfly.live/ (HTTP 200; HTML diffed against the Space's dist/index.html) [direct]", "https://fruitfly.live/model.json and https://fruitfly.live/data/manifest.json (byte-identical to the Space) [direct]", "https://fruitfly.live/assets/index-DyzlLaks.js (websocket trade handler mapping buy/sell to walk/fly stimulation) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. WebGPU (via @huggingface/kernels) is used when available, with a JavaScript fallback. WebGL2 is required for rendering. All 166,700 neurons are integrated, so the whole CNS really runs. The README is unusually honest: it calls the movement illustrative and says Shiu's validation does not carry over to MaleCNS. The model is a female-brain model applied to a male CNS, shown on a female NeuroMechFly body. Minor internal mismatch: model.json lists 11,068 omitted fast-output neurons, but the data manifest lists 11,609. Space licence metadata: apache-2.0. LICENSE file: MIT for app code, CC BY 4.0 for MaleCNS data, Apache-2.0 for body and kernels. fruitfly.live: serves the same app with a different build. /model.json and /data/manifest.json are byte-identical to the Space. But the page replaces the paint tools with a 'live Solana trade tape': buy/sell trades arriving over a websocket (/ws on the same host) trigger the walk/fly stimulation presets. That page names no author and shows no licence or link to the original Space; it only links the MaleCNS data and the Hugging Face kernels blog. List only the Hugging Face Space as try_url. 6 Oct 2026 browser check (headless, no GPU): after 'Download & start' the page moved but fetched only 5.87 MB in 60-90 s and started no Web Worker, so its ~77 MB brain download was not seen in our check window and the simulation was not confirmed; we will re-read the code." }, { "id": "open-fly", "name": "Open Fly", "type": "browser-demo", "author_or_org": "Pr1nted", "summary": "The whole Shiu et al. FlyWire v783 brain, ported to JavaScript and run in a browser worker, plays the grand-strategy game Open Doctrines (compiled to WebAssembly). Each turn, game events stimulate sugar, bitter, water and Johnston's-organ neurons; after 200 ms of simulated time, spikes of 1,299 descending neurons, dealt by a fixed seed into 39 action groups, choose the orders.", "claim": { "text": "A simulated fruit-fly brain plays Open Doctrines, live in your browser.", "url": "https://github.com/Pr1nted/Open-Fly" }, "dataset": "FlyWire FAFB", "release": "v783 via Shiu et al. Completeness_783.csv / Connectivity_783.parquet; annotations for descending neurons", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "Whole FlyWire v783 brain, Shiu weights unchanged", "neuron_model": "spiking: Shiu et al. LIF (Python port and a JavaScript worker, checked spike for spike against Brian2 on a scripted input, per README)", "input_mapping": "hand-set: land and surplus -> sugar GRNs, losses/deficit/new wars -> bitter, treasury -> water, number of wars -> Johnston's organ", "output_mapping": "hand-set: 1,299 descending neurons dealt into 39 balanced groups by a committed seed; spikes per group choose actions under the game AI's per-turn budget", "trained_parts": "none", "body": "none", "scripted_parts": "Game engine, action menu and per-turn budget are the game's own; the neuron-to-action assignment is arbitrary by design" }, "trained_class": "none", "grade_basis": [ "open_fly/brain.py:13,28-30 Shiu constants (v_th -45 mV, w_syn 0.275 mV, f_poi 250) [direct]", "open_fly/live.py:401-403 loads Completeness_783.csv and Connectivity_783.parquet; four input channels [direct]", "open_fly/encode.py:9,29-31 sugar/bitter/water/jon from land, deficit, wars and treasury [direct]", "open_fly/decode.py:1-8,33-43 descending neurons dealt into 39 groups by seed; the shuffled-connectome control uses the same groups [direct]", "notebooks/open_fly_colab.ipynb cell 5: shuffled control permutes the postsynaptic column; PREREGISTRATION.md fixes the comparison, but no completed brain-vs-shuffle result is committed [direct]" ], "grade_date": "2026-09-30", "measured_result": "No completed result: a pre-registered comparison with a shuffled-connectome control (postsynaptic column permuted, same action groups) is implemented, and the author discarded 96 early control runs as invalid (PREREGISTRATION.md), but no brain-vs-control result is committed.", "try_url": null, "try_status": "no no-install option", "code_url": "https://github.com/Pr1nted/Open-Fly", "code_licence": "Apache-2.0", "code_licence_source": "LICENSE file 'LICENSE' read at commit 1df3ff68db; GitHub API spdx_id=Apache-2.0", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser", "Colab" ], "platform_note": "browser page with a WebAssembly game; Colab notebook for the Brian2 path", "gpu": "none", "download_size": "Shiu v783 files (about 100 MB) fetched by the workflow or notebook", "last_commit": { "date": "2026-10-05T05:45:36+00:00", "hash": "1df3ff68db4caa747ed4a75469d8a9a558992c92", "branch": "main" }, "pushed_at": "2026-09-28T21:22:43Z", "last_release": "none", "created_at": "2026-09-13T17:21:03Z", "stars": 3, "stars_date": "2026-09-30", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-30", "archived": false, "paper_url": null, "peer_review": "none", "link_status": { "url": "https://github.com/Pr1nted/Open-Fly", "status": "ok", "http_code": 200, "final_url": "https://github.com/Pr1nted/Open-Fly", "checked_at": "2026-10-06T09:40:18.553Z" }, "last_verified": "2026-10-06", "added_in_run": "run:a46416d4-3706-4c00-ba7d-1371567178c4", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/Pr1nted/Open-Fly [direct: clone HEAD 47f5aa8, 2026-09-30]" ], "notes_limitations": "The whole brain runs untrained, but the mapping from 1,299 descending neurons to 39 game actions is arbitrary by design, and game events reach the brain as four taste/hearing channels. The author pre-registered a shuffled-wiring comparison and documents invalid runs openly, but no result is in the repository yet, so there is no evidence that the real wiring plays differently from a scrambled one." }, { "id": "same-smell", "name": "Same Smell", "type": "browser-demo", "author_or_org": "anzal1", "summary": "Shows, side by side, subsets of about 600 neurons of the cVA pheromone pathway from a male (MaleCNS v1.0) and a female (FlyWire v783) brain, spiking in a slowed-down LIF model after a 'puff'. The first descending-neuron spike triggers a puppet animation, but the code fixes the kind of reaction (lunge or pause) by sex.", "claim": { "text": "Two real fruit fly brains. One pheromone. Opposite reactions.", "url": "https://github.com/anzal1/samesmell" }, "dataset": "several", "release": "MaleCNS v1.0; FlyWire FAFB v783", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "subset: 615 neurons / 5,837 edges of MaleCNS v1.0 and 599 neurons / 2,171 edges of FlyWire v783 (ORN_DA1, DA1 PNs, top lateral-horn targets, pC1/mAL or pC1/vpoEN/oviIN/mAL, top 30 DNs, plus ~298 'context' neurons picked for looks). Synapse counts are stored but not used; weights are hand-set by edge class.", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "none", "body": "custom", "scripted_parts": "Fly bodies are keyframed puppets. After the first descending-neuron spike, the male body plays 'lunge' and the female body plays 'pause' (chosen by a sex flag). Verdict text is fixed for each side. Weights are hand-set (strong 'spine' edges along a chosen pathway, weak lateral edges), and delays are stretched to 0.72 s per hop. In Pong mode, input goes to context neurons binned by soma height." }, "trained_class": "none", "grade_basis": [ "src/main.js:128-146 loads data/real/{male,female}.json first, then procedural fallbacks [direct]", "data/real/male.json meta.provenance: MaleCNS v1.0 via neuPrint; data/real/female.json: FlyWire FAFB v783, Connectivity_783.parquet [direct]", "src/sim.js:38-42,148-165 edge weights come from edge class (spine/feedforward/lateral), not synapse count; src/sim.js:20 DELAY_BASE 0.72 s per hop [direct]", "src/sim.js:303-305 the first descending-neuron spike calls onVerdict [direct]", "src/body.js:248,474 the body enters (male ? 'lunge' : 'pause') from the sex flag, not from network output [direct]", "src/modes.js:28-30 VERDICT_TEXT is fixed per side [direct]" ], "grade_date": "2026-09-28", "measured_result": null, "try_url": "https://anzal1.github.io/samesmell/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/anzal1/samesmell", "code_licence": "none found", "code_licence_source": "no LICENSE/COPYING file at repo root (commit f5891d9206); GitHub API spdx_id=None", "data_licence": "several (see dataset entries)", "platform": [ "browser" ], "gpu": "none", "download_size": "~11 MB (data/real JSON ~10 MB plus 0.6 MB bundle)", "last_commit": { "date": "2026-09-13T00:44:51+05:30", "hash": "f5891d9206802e172c619e2bcede9d25e2cb79f1", "branch": "main" }, "pushed_at": "2026-09-12T19:14:54Z", "last_release": "none", "created_at": "2026-09-11T13:03:51Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://anzal1.github.io/samesmell/", "status": "ok", "http_code": 200, "final_url": "https://anzal1.github.io/samesmell/", "checked_at": "2026-10-06T09:40:18.122Z", "other_links": [ { "url": "https://github.com/anzal1/samesmell", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/anzal1/samesmell (clone: README.md, src/*.js, scripts/fetchReal.py, data/real/*.json, dist/app.js) [direct]", "https://anzal1.github.io/samesmell/ (HTTP 200) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. Uses WebGL through three.js; no WASM or WebGPU. About 600 neurons per sex are simulated, not whole brains. The README says 'nobody keyframed their choices' and credits the wiring for the difference, but the code hard-codes lunge versus pause by sex. The connectome subsets only decide when a descending neuron first fires. The in-app 'what's real' panel admits the dynamics are staged and the bodies are puppets. The linked Cell paper is the MaleCNS data paper, not a paper about this project. No licence file." }, { "id": "vial", "name": "Vial (fly-vial)", "type": "browser-demo", "author_or_org": "Raphael Rocha (RaphaelSR)", "summary": "A virtual-pet game where you feed, clean and train a 3D fly. A web worker runs a LIF simulation of the whole FlyWire brain, pruned to connections of 5 or more synapses. Firing of chosen descending and motor neurons sets reflex animations, and a small REINFORCE policy learns 'tricks' from descending-neuron activity.", "claim": { "text": "a fruit fly you keep, running on her own connectome — 138k neurons in your browser", "url": "https://github.com/RaphaelSR/fly-vial" }, "dataset": "FlyWire FAFB", "release": "v783", "evidence_grade": "B", "grade_note": null, "mechanism": { "wiring": "FlyWire v783, all 138,639 neurons; only edges with >=5 synapses kept (2,700,513 of 15,091,983); sign per neuron", "neuron_model": "LIF (Shiu et al. 2024 parameters, event-driven active set, dt 0.1 ms)", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "A linear softmax policy (5 actions x 58 descending cell-type features), trained with REINFORCE from the user's reward clicks, chooses the 'taught' action. The connectome itself has no plasticity.", "body": "custom", "scripted_parts": "Idle wandering is sine-driven (app.js:177-181). Grooming comes from the pet's 'dirt' need, not from neurons (cues.js:47 sets groom 0). Sleep and stopping come from energy and light state. Trained actions play fixed drive patterns (cues.js:52-60). The body is procedural WebGL2 animation, not physics. Hunger, thirst and the 50-day lifespan are game timers." }, "trained_class": "policy-or-llm", "trained_class_note": "linear softmax policy trained by REINFORCE chooses the taught action", "grade_basis": [ "web/js/engine/sim.worker.js:1-10 LIF constants from Shiu et al. (V0 -52, Vth -45, tau 20/5 ms, wsyn 0.275) [direct]", "web/js/engine/sim.worker.js:48-95 spike delivery over the CSR connectome, Poisson drive on stimulated neurons, integrate and threshold [direct]", "web/data/meta.json.gz: version 'flywire-783', threshold 5, n_neurons 138639, n_edges 2700513 [direct]", "web/js/game/brain.js:28-41 the whole packed connectome is decoded and sent to the worker [direct]", "web/js/game/cues.js:10-22 inputs are annotated sensory types (LB3, BM_InOm, JO-B*, ORN_DM1/DM2, LPLC2, hygrosensory) [direct]", "web/js/game/cues.js:36-48 reflex output = mean firing rate of named DN/MN channels (DNp01, MN9/MN10, DNa02, ...) through hand-set saturation constants [direct]", "web/js/game/learn.js:13-70 5-action softmax policy with a REINFORCE update (trained readout) [direct]", "web/js/app.js:171-183 body drive comes from the policy action or instinct channels, plus scripted idle wandering and need-based grooming [direct]" ], "grade_date": "2026-09-28", "measured_result": "README only: after 28 contrastive trials, buzz->proboscis 87% and puff->jump 87% (chance 20%); an untaught odour gave proboscis 65%. Subtracting the pre-cue baseline cut the cosine similarity between two cues from 0.85 to 0.007. The pruned model keeps 92.5% of responding neurons (rate correlation 0.988 vs the full model). No shuffled-wiring or no-connectome control. The scripts need tools/lif.py and data files that are not in this repo.", "try_url": "https://raphaelsr.github.io/fly-vial/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/RaphaelSR/fly-vial", "code_licence": "MIT (code); LICENSE notes web/data/ (FlyWire v783-derived) keeps its data licence", "code_licence_source": "LICENSE file 'LICENSE' read at commit 5aa9780c9b; GitHub API spdx_id=NOASSERTION", "data_licence": "CC-BY-NC-4.0", "platform": [ "browser" ], "gpu": "none", "download_size": "7.43 MB transferred in our browser check (6 Oct 2026; largest file conn.bin.gz 6.84 MB). Figure from the code before the check: ~8 MB data (conn.bin.gz 6.8 MB, positions 0.8 MB, labels 0.4 MB)", "last_commit": { "date": "2026-09-19T00:08:05-03:00", "hash": "5aa9780c9bb93a12121432d9acba6e3eb83c5038", "branch": "main" }, "pushed_at": "2026-09-19T03:08:08Z", "last_release": "none", "created_at": "2026-09-13T00:18:23Z", "stars": 0, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://raphaelsr.github.io/fly-vial/", "status": "ok", "http_code": 200, "final_url": "https://raphaelsr.github.io/fly-vial/", "checked_at": "2026-10-06T09:40:18.164Z", "other_links": [ { "url": "https://github.com/RaphaelSR/fly-vial", "status": "ok", "http_code": 200 } ] }, "browser_check": { "status": "runs", "checked_at": "2026-10-06", "browser": "headless Chromium 154, Linux, no GPU", "transferred_mb": 7.43, "compute": "in-browser-worker", "first_activity_s": 10.62, "console_errors": 0, "page_errors": 0, "matches_catalogue": "yes", "note": "Ran; in your browser (Web Worker); 7.43 MB in the first 60-90 s (largest: conn.bin.gz 6.84 MB).", "source": "Digital Fly Lab browser check of 2026-10-06 (report: https://shaduf.ai/p/digital-fly-catalog/reports/2026-10-06-run9/#browser-heading), result file vial.json" }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/RaphaelSR/fly-vial (clone at HEAD 5aa9780) [direct]", "https://raphaelsr.github.io/fly-vial/ (HTTP 200; data/conn.bin.gz HTTP 200, 6,836,121 bytes) [direct]" ], "notes_limitations": "Page responds (HTTP), not played. The whole connectome really is simulated in a worker. Needs WebGL2, module workers and DecompressionStream; no WebGPU or WASM. The README says 'no scripted reactions', yet much of what you see is not neural: idle walking, grooming (brushing actually stops it, against the README table), sleep and the fixed trick animations. Trick-learning is a trained policy on top of the connectome, and the 87% result is compared only with chance. The repo is marked 'parked'. Licence: the LICENSE file is MIT for code and notes that the data in web/data/ (derived from FlyWire v783 and Schlegel et al. annotations) stays CC BY 4.0, which is why the metadata shows NOASSERTION." }, { "id": "webgpu-fly", "name": "webgpu-fly", "type": "browser-demo", "author_or_org": "abgnydn", "summary": "A WebGPU browser demo: keys stimulate named descending neurons in a whole-brain FlyWire v783 LIF simulation, and the spikes pass by cell-type name into a MANC nerve-cord LIF simulation. The averaged leg motor output scales a hand-written tripod gait on a flybody MuJoCo/WASM body, and by default it also sets the body's velocity directly.", "claim": { "text": "A whole-animal Drosophila nervous system — brain, spinal cord, and body — running end-to-end in a browser tab on WebGPU", "url": "https://github.com/abgnydn/webgpu-fly" }, "dataset": "several", "release": "FlyWire v783 + MANC (version not stated)", "evidence_grade": "C", "grade_note": null, "mechanism": { "wiring": "FlyWire FAFB v783 whole brain (139,255 neurons per README/build script) + MANC VNC (23,188 neurons per README), linked by DN cell-type name", "neuron_model": "LIF", "input_mapping": "hand-made", "output_mapping": "hand-made", "trained_parts": "An optional flybody RL walking policy (Vaxenburg et al.) runs on a separate path that bypasses brain and VNC. w_syn was tuned to reach Kenyon-cell sparsity.", "body": "flybody", "scripted_parts": "A hand-written sine tripod CPG (driveLegs) is scaled by two numbers from the VNC (walk magnitude, turn bias). A kinematic assist, on by default, writes body velocity and yaw straight from that command. A pitch/roll damper is always on. If vnc.bin is missing, the code falls back to a 200-neuron hand-designed spine." }, "trained_class": "none", "trained_class_note": "optional flybody RL walking policy bypasses the brain", "grade_basis": [ "tools/build_csr.py:139-140 builds brain CSR from proofread_connections_783 / root_ids_783 [direct]", "src/sim.ts:48 navigator.gpu.requestAdapter: WebGPU LIF (src/shaders/lif.wgsl) [direct]", "src/main.ts:627-646 loads MANC vnc.bin as a second FlySim, otherwise a synthetic 200-neuron fallback (src/vnc.ts:1-3) [direct]", "src/physics.ts:695-750 driveLegs: sine tripod gait scaled by walk/turn [direct]", "src/physics.ts:627-633 kinematic assist sets qvel translation and yaw from the command [direct]", "LIMITATIONS.md:319-330 assist on/off table: the assist does the locomotion whatever the controller is [direct]" ], "grade_date": "2026-09-28", "measured_result": "LIMITATIONS.md:319-330: with the assist on, the body moves 0.80-0.88 cm/sim s whatever the controller; with it off, 0.029-0.163 cm/sim s, and the CPG path mostly spins in place. README: the RL policy with the assist off covers 2.004-2.021 cm/sim s vs 0.032 with no policy (brain bypassed). The brain kernel runs at ~0.25 kHz biological time on an M2 Pro. These are body and throughput checks, not connectome controls.", "try_url": "https://webgpu-fly.pages.dev/", "try_status": "page responds (HTTP 200, checked 2026-10-06), not played by us", "code_url": "https://github.com/abgnydn/webgpu-fly", "code_licence": "MIT", "code_licence_source": "LICENSE file 'LICENSE' read at commit bb00419e87; GitHub API spdx_id=MIT", "data_licence": "several (see dataset entries)", "platform": [ "browser" ], "gpu": "required", "download_size": "~315 MB (brain.bin 126 MB, vnc.bin 43 MB, flybody bundle 140 MB, walking policy 5 MB; sizes from HTTP HEAD)", "last_commit": { "date": "2026-09-05T00:43:32+07:00", "hash": "bb00419e874eee9e878542dcc5fce289ede2e1b9", "branch": "main" }, "pushed_at": "2026-09-04T17:43:33Z", "last_release": "none", "created_at": "2026-05-04T10:22:20Z", "stars": 9, "stars_date": "2026-09-27", "api_fields_reused": [ "stars", "created_at", "pushed_at" ], "api_fields_reused_from": "2026-09-27", "archived": false, "archived_source": "GitHub repository page banner, checked 2026-09-30", "paper_url": null, "peer_review": "none", "link_status": { "url": "https://webgpu-fly.pages.dev/", "status": "ok", "http_code": 200, "final_url": "https://webgpu-fly.pages.dev/", "checked_at": "2026-10-06T09:40:18.285Z", "other_links": [ { "url": "https://github.com/abgnydn/webgpu-fly", "status": "ok", "http_code": 200 } ] }, "last_verified": "2026-10-06", "controls": "none", "wiring_effect": "not-tested", "sources": [ "https://github.com/abgnydn/webgpu-fly (clone, files read raw) [direct]", "https://webgpu-fly.pages.dev/ and /app (HTTP 200), assets.json and R2 HEAD sizes [direct]" ], "notes_limitations": "Page responds (HTTP), not played. Needs WebGPU (no fallback for the brain) and WASM (MuJoCo); data is served from an R2 bucket. The headline 'whole-animal nervous system' is more than the code delivers, but the README and LIMITATIONS.md say openly that the connectome only scales a hand-written gait and the assist does the moving. The brain-to-VNC link joins cell-type names across two different animals. The only dynamics check is KC sparsity, which was also the tuning target. The src/vnc.ts header comment ('MANC isn't loaded here yet') is out of date." } ] }