FlyGPT lands on Hugging Face: trainable LM wired as a fruit-fly connectome
QuixiAI · x · 2026-09-14
QuixiAI published FlyGPT weights on Hugging Face: a character-level language model whose recurrent architecture is a real subgraph of the fruit-fly brain connectome (MaleCNS v1.0).
Key points:
- Unlike the earlier frozen-reservoir approach (ngxson/fly-llm-hf), FlyGPT trains one value per real synaptic connection via gradient descent while keeping the fly's edge topology fixed
- Compares against degree-preserving scrambled-connection controls across paired seeds to test whether real brain wiring learns better
- Built on QuixiAI/MaleCNS, a lossless packaging of the connectivity tables; graph extraction is deterministic with pinned revisions, each edge traceable to the original tables
- safetensors checkpoint, cc-by-4.0 license
Related event: QuixiAI Open-Sources FlyGPT, a Language Model Built on Fly Brain Connectome(2 posts)→
More from Research
- New research: standard SGD matches AdamW for LLM RL training, with far less memory overhead — zhaoran_wang · 2026-09-14
- RSI work separates practical harness self-improvement from unproven intelligence explosion — arthurcolle · 2026-09-14
- Amazon Proposes Query-Aware Index Pruning to Optimize Retrieval Under Budget Constraints — _reachsumit · 2026-09-14
- New Paper Finds Retrieval Signals Give No Reliable Routing Gain in Adaptive Multimodal RAG — _reachsumit · 2026-09-14
- Google: Graph RAG Cuts API Hallucination Rate from 56.4% to 16.2% in Java-to-Python Migration — _reachsumit · 2026-09-14
- Position Paper: Recommender Systems Should Shift to Personal Agent-Mediated Recommendation — _reachsumit · 2026-09-14