faeval.git/experiments/dfa_direction_quality_test.py, branch master Unnamed repository; edit this file 'description' to name the repository. Add DFA direction-quality direct test (codex round 13 option c) 2026-04-08T03:37:49+00:00 YurenHao0426 Blackhao0426@gmail.com 2026-04-08T03:37:49+00:00 0c245f5683cceba448d20d9dfc2090adb3503f14 Trains both vanilla DFA (lam=0) and penalized DFA (lam=1e-2) from the same seed, then directly measures the per-layer cosine between DFA's local credit signal e_T @ B_l^T and the BP gradient at hidden layers. Uses the training Bs (not fresh ones, per the Bs-specificity finding from earlier). The penalized run is the key measurement: in that condition the BP grad is ~10^-7 (well above the eps=1e-8 floor), so a near-zero cosine here would be the direct evidence of the second failure mode (direction-quality ceiling) that codex round 13 hypothesized. Pre-registered prediction: penalized cos(DFA, BP) ~ 0.01-0.05 -> direction quality is the second, separable failure mode. Saves the penalized checkpoint so the diagnostic protocol can be re-applied to it (where (a) and (b) should pass, (d) should still fail). 
Trains both vanilla DFA (lam=0) and penalized DFA (lam=1e-2) from the same
seed, then directly measures the per-layer cosine between DFA's local
credit signal e_T @ B_l^T and the BP gradient at hidden layers. Uses the
training Bs (not fresh ones, per the Bs-specificity finding from earlier).

The penalized run is the key measurement: in that condition the BP grad is
~10^-7 (well above the eps=1e-8 floor), so a near-zero cosine here would
be the direct evidence of the second failure mode (direction-quality
ceiling) that codex round 13 hypothesized.

Pre-registered prediction: penalized cos(DFA, BP) ~ 0.01-0.05 -> direction
quality is the second, separable failure mode. Saves the penalized
checkpoint so the diagnostic protocol can be re-applied to it (where (a)
and (b) should pass, (d) should still fail).