UnchaosLM / eval /eval.py

Commit History

  1. Benchmarks: 7-model comparison incl. Qwen3.5, Gemma-4, 2507; false-alarm metric; rescore tooling

    Verified signature
    0d43137

  1. Benchmarks: add Qwen3-8B and Llama-3.2-3B comparisons

    Verified signature
    8c92c1c
  2. Publish benchmarks: v4 vs base Qwen3-4B, eval script + test set

    Verified signature
    f2accd8