From 752dfb833b06a6fb974df892de560caf328ed1dd Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Wed, 8 Apr 2026 18:03:16 -0500 Subject: paper v2.31: matched 30-epoch BP/DFA controls (was unsourced 0.609/0.308) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The §5 ¶3 BP-no-penalty value of 0.609 ± 0.004 and DFA-no-penalty value of 0.308 ± 0.014 turned out to be unsourced — they were carried over from a hardcoded comment in experiments/bp_with_penalty_control.py ("BP-trainable (3-seed mean): 0.609") that nobody had actually measured with a matched 30-epoch run. Ran the missing matched controls under the same recipe as BP+pen (lam=0, 30 epochs, AdamW 1e-3, wd 0.01, cosine schedule, batch 128, 3 seeds 42/123/456): BP no-pen 30ep: per-seed 0.5851, 0.5845, 0.5863 → 0.585 ± 0.001 (paper said 0.609 ± 0.004, off by 0.024) DFA no-pen 30ep: per-seed 0.3070, 0.2985, 0.2966 → 0.301 ± 0.005 (paper said 0.308 ± 0.014) Also re-grounded DFA+penalty 30ep using the dfa_pen_short 3-seed run (0.3593, 0.3610, 0.3604 → 0.360 ± 0.001), which is what the deep-cosine +0.155 figure was computed on. The paper had 0.363 ± 0.001 — that came from the 100-epoch run, not the 30-epoch run, so it was an apples-to- oranges comparison with BP+pen 30-ep. Paper changes (§5 ¶3): BP penalty cost: -8 pp → -5.5 pp DFA pen rescue: +5.5 → +5.9 pp DFA+pen margin vs frozen: +1.4 → +1.1 pp BP-to-DFA gap: 17 → 17.0 pp (unchanged) BP-to-SB gap: 7.7 → 7.7 pp (unchanged) BP-to-DFA gap is still the lower-bound credit-quality cost claim; 17 pp gap is unchanged in magnitude. Also updated: - §5 ¶1 prose: 0.363 → 0.360, 0.308 → 0.301 - §4 ¶4 prose: DFA+pen 0.363 → 0.360 - Appendix J Table 9 caption: 0.363 → 0.360, +9.0 → +9.3 pp gap to SB - Appendix L paragraph: +5.5 → +5.9 pp DFA penalty rescue - Figure 3 panel C bar values + title pen-cost annotation - New results/matched_30ep_control_summary.json as auditable record Page layout preserved: 9 main pages + refs p10, 18 total, 0 overfull. Co-Authored-By: Claude Opus 4.6 (1M context) --- results/bp_nopen_30ep_3seed.log | 58 +++++++++++++++++++++++++++++++++++++++++ 1 file changed, 58 insertions(+) create mode 100644 results/bp_nopen_30ep_3seed.log (limited to 'results/bp_nopen_30ep_3seed.log') diff --git a/results/bp_nopen_30ep_3seed.log b/results/bp_nopen_30ep_3seed.log new file mode 100644 index 0000000..f88e725 --- /dev/null +++ b/results/bp_nopen_30ep_3seed.log @@ -0,0 +1,58 @@ +=== BP no-pen 30ep seed=42 === +BP + ‖f‖² penalty: seed=42, lam=0.0, epochs=30 + ep 1: test_acc=0.3920 + ep 5: test_acc=0.4950 + ep 10: test_acc=0.5296 + ep 15: test_acc=0.5539 + ep 20: test_acc=0.5721 + ep 25: test_acc=0.5813 + ep 30: test_acc=0.5851 + +FINAL test acc: 0.5851 +Compare to: + BP-trainable (3-seed mean): 0.609 + Penalized DFA lam=1e-2: 0.363 + DFA-shallow: 0.349 + +Margin vs DFA-shallow baseline: +23.61 pp + → BP+penalty intermediate; partial capacity loss + residual mode 2 +Saved results/bp_no_penalty_30ep/bp_pen_lam0.0_s42.json +=== BP no-pen 30ep seed=123 === +BP + ‖f‖² penalty: seed=123, lam=0.0, epochs=30 + ep 1: test_acc=0.3958 + ep 5: test_acc=0.4851 + ep 10: test_acc=0.5345 + ep 15: test_acc=0.5564 + ep 20: test_acc=0.5748 + ep 25: test_acc=0.5827 + ep 30: test_acc=0.5845 + +FINAL test acc: 0.5845 +Compare to: + BP-trainable (3-seed mean): 0.609 + Penalized DFA lam=1e-2: 0.363 + DFA-shallow: 0.349 + +Margin vs DFA-shallow baseline: +23.55 pp + → BP+penalty intermediate; partial capacity loss + residual mode 2 +Saved results/bp_no_penalty_30ep/bp_pen_lam0.0_s123.json +=== BP no-pen 30ep seed=456 === +BP + ‖f‖² penalty: seed=456, lam=0.0, epochs=30 + ep 1: test_acc=0.3969 + ep 5: test_acc=0.4957 + ep 10: test_acc=0.5344 + ep 15: test_acc=0.5612 + ep 20: test_acc=0.5716 + ep 25: test_acc=0.5845 + ep 30: test_acc=0.5863 + +FINAL test acc: 0.5863 +Compare to: + BP-trainable (3-seed mean): 0.609 + Penalized DFA lam=1e-2: 0.363 + DFA-shallow: 0.349 + +Margin vs DFA-shallow baseline: +23.73 pp + → BP+penalty intermediate; partial capacity loss + residual mode 2 +Saved results/bp_no_penalty_30ep/bp_pen_lam0.0_s456.json +=== ALL DONE === -- cgit v1.2.3