Fly Language Model Integrates Complete Fruit Fly Connectome into a 1.2B LLM

The Fly Language Model (FLM) represents a significant advancement in artificial intelligence by integrating the full MaleCNS v1.0 fruit fly connectome with a frozen LiquidAI LFM2.5-1.2B architecture. Developed as a public chatbot, FLM claims to be the first of its kind, although it does not officially label itself as such. Interestingly, tests reveal that a control model, which lacks the fly connectome, performs slightly better than FLM. The system operates as a reservoir computer attached to a language model, utilizing all 166,700 nodes and 25,582,938 directed edges of the MaleCNS graph. Only a small portion of the parameters are trained, leaving the majority fixed. Results from experiments show that while the fly readout improves performance marginally, a direct-input control consistently outperforms it. Further analysis indicates that the connectome does not contribute to long-term memory retention, as state differences diminish rapidly. The findings challenge the assumption that complex biological wiring enhances language processing capabilities in AI.