wassname fc30514b23 feat: T5 eval-time ablation for route + fix route deployment invariant ...

T5: eval_hack_solve helper + ablate_quarantine ctx; periodic ablated-eval
(hack_abl/solve_abl cols, appended so results.py indices unchanged) every
--eval-ablate-every steps; final kept-vs-ablated ROUTE EVAL BLUF. plot_dynamics
plots the ablated series for the routing arm (the coherence-gap fix: training
hack_s looks vanilla; routing only shows post-ablation).

External-review fixes (docs/spec/20260530_code_review.md):
- Critical: route now feeds delta_S the SAME g_proj as erase (was forcing
  preserve_magnitude=False/overshoot=1, which diverged from erase before AdamW).
  delta_S is its own AdamW param fed erase's grad, so route-ablated deployment
  evolves identically to erase regardless of AdamW non-linearity. Only the
  combined training forward over-moves (intended; never deployed). Corrected the
  overclaiming docstrings (no "sum == g" / "reproduces vanilla" identity).
- Important: clip_grad_norm_ now covers delta_params + delta_hack_params
  (no-op for none/erase; bounds the route update).
- Important: results.py paired-delta table includes routing (keyed on arm).

smoke route/erase/vanilla green: dsh route=0.0105 erase/none=0, span=2.9e-7,
ROUTE EVAL BLUF prints.

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

2026-05-30 00:50:53 +00:00

..

build_combined_pool.py

wip

2026-05-29 06:29:46 +00:00

plot_dynamics.py

feat: T5 eval-time ablation for route + fix route deployment invariant

2026-05-30 00:50:53 +00:00

results.py

feat: T5 eval-time ablation for route + fix route deployment invariant

2026-05-30 00:50:53 +00:00