diff options
Diffstat (limited to 'protocol/EVIDENCE_SUMMARY.md')
| -rw-r--r-- | protocol/EVIDENCE_SUMMARY.md | 28 |
1 files changed, 15 insertions, 13 deletions
diff --git a/protocol/EVIDENCE_SUMMARY.md b/protocol/EVIDENCE_SUMMARY.md index 2591c97..bab8764 100644 --- a/protocol/EVIDENCE_SUMMARY.md +++ b/protocol/EVIDENCE_SUMMARY.md @@ -122,12 +122,12 @@ on deep layers. **Caught by direct per-layer cosine measurement.** | 456 | 0.364 | 4.1e4 | 9.0e-7 | +0.139 | | **mean** | **0.363 ± 0.001** | **4.0e4** | **9.0e-7** | **+0.151 ± 0.012** | -### BP+penalty 2×2 grid (raw acc, primary number per round 20) +### BP+penalty 2×2 grid (matched 30-epoch 3-seed values, paper v2.32) | | no penalty | with penalty | penalty effect | |---|---:|---:|---:| -| BP | 0.609 | **0.530** | −8 pp (capacity loss) | -| DFA | 0.308 | 0.363 | +5.5 pp (rescue) | +| BP | 0.585 ± 0.001 | **0.532 ± 0.006** | −5.3 pp (capacity loss) | +| DFA | 0.301 ± 0.005 | 0.360 ± 0.001 | +5.9 pp (rescue) | ### Round 20 phrasing for the gap @@ -170,19 +170,21 @@ All 6 bugs from `protocol/CHECKLIST.md` have a reproducer: | ViT-Mini (4-block d=128) | yes | 0.26 | 0.80 | (a)+(b) ep 1-3 | never | | StudentNet (4-block d=128) | **no** | 0.33 | 0.62 | (a) ep 18, **(b) NEVER** | never | -### Penalty rescue (3-seed for λ=1e-2, single-seed for others) +### Penalty rescue (matched 30-epoch 3-seed values, paper v2.32) | condition | acc | ‖h_L‖ | ‖g_L‖ | |---|---:|---:|---:| -| DFA-vanilla | 0.308 ± 0.014 | 4.4e8 | 5e-10 | -| DFA + λ=1e-3 ‖f‖² | 0.372 (1 seed) | 4.0e4 | 7e-7 | -| DFA + λ=1e-2 ‖f‖² | 0.363 ± 0.001 | 3.8e4 | 1e-6 | -| DFA + λ=1e-1 ‖f‖² | (running) | (running) | (running) | -| DFA-shallow baseline | 0.349 ± 0.002 | (n/a) | (n/a) | -| BP-trainable | 0.609 ± 0.004 | 2.0e2 | 5e-5 | - -The penalty rescues by +5.5 pp over vanilla DFA but only +1.4 pp over the -shallow baseline; mechanism is necessary but not sufficient. +| DFA-vanilla 30ep (3-seed) | 0.301 ± 0.005 | 4.4e8 (s42) / 5e8 (3-seed mean) | 4e-10 | +| DFA + λ=1e-4 ‖f‖² 30ep (3-seed) | 0.360 | 2.2e4 | 7e-7 | +| DFA + λ=1e-2 ‖f‖² 30ep (3-seed) | 0.360 ± 0.001 | 1.3e4 | 1.6e-6 | +| DFA + λ=1e-1 ‖f‖² 30ep (s42) | 0.349 | 1.2e4 | 1.6e-6 | +| DFA-shallow baseline (frozen) | 0.349 ± 0.002 | (n/a) | (n/a) | +| BP-trainable 30ep (3-seed) | 0.585 ± 0.001 | (n/a) | (n/a) | +| BP-trainable 100ep (3-seed) | 0.6147 ± 0.004 | 2.0e2 | 5e-5 | +| BP+pen λ=1e-2 30ep (3-seed) | 0.532 ± 0.006 | 4.0e4 | (matches DFA+pen) | + +The penalty rescues by +5.9 pp over vanilla DFA (matched 30-ep) but only ++1.1 pp over the shallow baseline; mechanism is necessary but not sufficient. ## Figures (paper-ready) |