faeval.git/experiments/minimal_aux_compression.py, branch master Unnamed repository; edit this file 'description' to name the repository. Add Phase 10A.7: minimal aux compression — continuous trainability is essential 2026-03-27T19:25:00+00:00 YurenHao0426 Blackhao0426@gmail.com 2026-03-27T19:25:00+00:00 65d97ad1ef4b552103420e6501655df192c98d57 8-branch dissection: - zero_target + normmatched both crash: non-zero direction necessary, not norm - perlayer_vector: +0.7% (per-block trainable vector works, network not required) - freeze_after_{1,5,10}: ALL crash to ~13-14% (continuous trainability essential) - random_trainable: +1.0% (reference) Minimal mechanism: continuously trainable, non-zero, depth-aware auxiliary perturbation. Freezing at ANY point destroys the benefit entirely. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> 
8-branch dissection:
- zero_target + normmatched both crash: non-zero direction necessary, not norm
- perlayer_vector: +0.7% (per-block trainable vector works, network not required)
- freeze_after_{1,5,10}: ALL crash to ~13-14% (continuous trainability essential)
- random_trainable: +1.0% (reference)

Minimal mechanism: continuously trainable, non-zero, depth-aware auxiliary perturbation.
Freezing at ANY point destroys the benefit entirely.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>