diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-03-24 18:03:55 -0500 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-03-24 18:03:55 -0500 |
| commit | 5550e2cac45758e579810ae36bf716a0b819cebc (patch) | |
| tree | 28f263e4030d6d5144af5badcebd533b27f4da78 /report/toy_term_grad_effect.png | |
| parent | 3d17cbad98f320905c52509c7f18691eab8bf2a0 (diff) | |
Add Phase 5: vector field audit, frozen CIFAR transfer, online pilot
Phase 5A: Audit passes — shuffle control collapses, gains are real
Phase 5B: Transfer SUCCESS — vec_M4 beats scalar CB by +0.25 Gamma, +0.31 rho on frozen CIFAR
Phase 5C: Online FAILURE — vec does worse than scalar CB online despite better frozen credit
Core finding: bottleneck is in local surrogate / co-adaptation, not estimator quality
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Diffstat (limited to 'report/toy_term_grad_effect.png')
0 files changed, 0 insertions, 0 deletions