Ethanfel e4dfaac63b Correct 4B 'partial' bias on identical values; harden verdict rule; note model-capability limits ...

The 4B over-uses 'partial' (mislabels identical ref/gen and clear opposites) and
also mis-identifies fine-grained content (e.g. names a position 'doggy'/'cowgirl'
when it is neither). Deterministic fix: force verdict=match when normalized
ref==gen. Prompt hardened to not default to 'partial' (opposites=mismatch). Docs:
the 4B is only reliable for coarse attributes — use the 30B for fine-grained
recognition; prefer grounded geometry axes over named-position labels.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-26 23:43:34 +02:00

..

AGENT_LOOP.md

describe emits one canonical reference; compare can anchor on it

2026-06-26 23:22:57 +02:00

CALIBRATION_POLICY.md

Correct 4B 'partial' bias on identical values; harden verdict rule; note model-capability limits

2026-06-26 23:43:34 +02:00

METHODOLOGY.md

Switch compare to discrete verdicts + granular pose axes + per-axis definitions

2026-06-26 23:15:51 +02:00