faeval.git/results/confirmatory, branch master

Add new experiment scripts, figures, and paper assets; untrack pyc/build artifacts

2026-06-14T09:06:32+00:00

Co-Authored-By: Claude Opus 4.8 (1M context)

Add 5 extra seeds to synthetic cross-state distance (now 10 seeds for all methods)

2026-04-04T00:10:08+00:00

BP/DFA/SB/CB: added seeds 2048,3000,4000,5000,6000 (L=4 only, all 3 alphas)
Total: 1290 rows (was 990)

Co-Authored-By: Claude Opus 4.6 (1M context)

Complete EP data: 10-seed synthetic + 6-seed CIFAR persample + cross-state

2026-04-03T18:51:15+00:00

EP synthetic: 30 JSONs + 30 checkpoints (10 seeds × 3α)
EP CIFAR persample: 6 seeds × 4 layers × 256 samples = 6144 rows added
Synth cross-state: 150 EP rows added (990 total)
cifar_persample_all.csv: 30720 rows (was 24576, +6144 EP)

Co-Authored-By: Claude Opus 4.6 (1M context)

EP synthetic 10 seeds complete: 30 JSONs + 30 checkpoints + cross-state distance

2026-04-03T18:36:54+00:00

Updated synth_cross_state_distance.csv with 150 EP rows (990 total).

Co-Authored-By: Claude Opus 4.6 (1M context)

Add EP cross-state distance for CIFAR + verify CNN summary

2026-04-03T15:58:22+00:00

EP CIFAR d_BP: L0=2.0×, L4=26.7× (much closer to BP than DFA=162×/2.5M×)
EP synthetic: no checkpoints saved (ep_synthetic.py didn't save .pt)
CNN summary: 20 rows confirmed correct

Co-Authored-By: Claude Opus 4.6 (1M context)

Add EP synthetic per-seed CSV + synthetic cross-state distance

2026-04-03T15:46:24+00:00

EP synthetic: 15 rows (3α × 5 seeds)
Synth cross-state: 840 rows (3α × 2L × 4 methods × 5 seeds × (L+1) layers)

Co-Authored-By: Claude Opus 4.6 (1M context)

Add EP to BP gradient sparsity analysis

2026-04-03T15:37:49+00:00

EP CIFAR d=256: s(1e-6)=100%, mean_norm=1.41e-04
EP produces networks where ALL samples have non-zero BP gradients,
unlike DFA (0.4%), SB (21%), CB (3%). EP is closer to BP (98.7%).

Updated clean_sparsity_summary.csv (980 rows, now includes EP).

Co-Authored-By: Claude Opus 4.6 (1M context)

Add cross-method hidden state distance vs BP

2026-04-03T15:34:33+00:00

Non-BP methods produce radically different representations:
DFA L0: 162×, L4: 2.5M× relative to BP hidden norms
SB L0: 3.2×, L4: 1.1M×
CB L0: 59×, L4: 1.4M×
(BP vs itself = 0)

Co-Authored-By: Claude Opus 4.6 (1M context)

Add d=512 support sparsity: 20 JSONs + summary CSV

2026-04-02T02:09:20+00:00

BP: s(1e-6)=92.7%, norm=2.70e-04, r_inf=0.159, PR=0.300
DFA: s(1e-6)=0.1%, norm=5.31e-08
SB: s(1e-6)=20.3%, norm=2.33e-06
CB: s(1e-6)=1.2%, norm=9.88e-08

Same pattern as d=256, confirming width-independence of the sparsity gap.

Co-Authored-By: Claude Opus 4.6 (1M context)

Add missing bp_s456.json for CIFAR d=512 (rerun after SIGTERM)

2026-04-02T01:54:58+00:00

bp s=456: acc=0.5999, rho=0.9881, nse=0.4764

Co-Authored-By: Claude Opus 4.6 (1M context)