RicPini/fgrf

FGRF for LLM Optimization: a scale-dependent structural probe for weight matrices, validated across Qwen2.5, SmolLM2, and GPT-2

1 best points · 1 comments · 1 Hacker News post · 1 GitHub stars · View on GitHub

Hacker News discussions

• Show HN: FGRF v3.0 – Scale-dependent structural probing of LLM weight topologies (1 points, 1 comments, by RicPini, 2026-09-30)