test_metrics.json

{
  "v3": {
    "type_macro_f1": 0.8944689733747935,
    "type_micro_f1": 0.9071487263763353,
    "top_level_f1": 0.9281437125748503,
    "ethics_f1": 1.0,
    "n_test": 1217,
    "test_loss": 0.4665677845478058,
    "template_version": 3,
    "ckpt": "checkpoints/section_v3"
  },
  "v2_baseline_same_split": {
    "type_macro_f1": 0.8257423979660845,
    "type_micro_f1": 0.8989317995069844,
    "top_level_f1": 0.9173553719008265,
    "ethics_f1": 0.0,
    "n_test": 1217,
    "template_version": 2,
    "ckpt": "thesanogoeffect/bibr-section-classifier-v2"
  },
  "gate": {
    "macro_ge_v2": true,
    "ethics_f1_ge_0_5": true,
    "micro_ge_v2_minus_0_005": true,
    "top_level_ge_0_90": true,
    "all_pass": true
  },
  "eval_script": "scripts/eval_section_checkpoint.py",
  "corpus": "data/section_corpus_v3",
  "date": "2026-07-02"
}