diff options
| author | YurenHao0426 <Blackhao0426@gmail.com> | 2026-04-08 05:39:39 -0500 |
|---|---|---|
| committer | YurenHao0426 <Blackhao0426@gmail.com> | 2026-04-08 05:39:39 -0500 |
| commit | 8dd65b2ec3df32749adabbf62c55101d5b00ae7b (patch) | |
| tree | 3a329bfdf9867ae13889dfcecd65ef216734947b /experiments/toy_lq.py | |
| parent | 68cfa13af2f026b7ff388aae4420eba0f0db804a (diff) | |
Round 32+33 H2 ablation: add no_residual_add flag; falsify residual-as-cause hypothesis
- models/residual_mlp.py: add residual_add and w2_std flags (default unchanged)
- experiments/snapshot_evolution_residual_explosion.py: add --no_residual_add and --w2_std CLI flags
- paper/main.tex §3 ¶3: add 1-sentence reference to no-residual control showing Mode 1 still fires
- paper/main.tex Appendix I: full smoke-test table + interpretation
- v2.2 main content stays at 8 pages (within 9-page E&D budget); 13 pages total
Smoke test (3 ep, w2_std=0.5, seed 42):
- DFA no-residual: ||h_L|| 4.69 -> 22050, ||g|| 1.6e-7 (Mode 1 (a) fires; (b) at floor)
- BP no-residual: acc only 0.16 at ep 3 (architecture is partially degenerate)
- Conclusion: residual skip is NOT necessary for Mode 1; the proximate trigger is more general
- Codex round 33 verdict: WALK BACK H2; demote 100ep run to confirmatory
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Diffstat (limited to 'experiments/toy_lq.py')
0 files changed, 0 insertions, 0 deletions