CPU-only, pure-Rust implementation of LiquidAI's LFM2.5-8B-A1B LLM
Terminal score: 0–1000 raw, weighted across 4 dimensions. Public score: 0–10 normalized (shown in the 30-day stars chart above).