From 94fc8471c515fb0baacc0aa79d4a47f539858e57 Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Wed, 22 Jul 2026 06:25:06 -0500 Subject: experiments: freeze oral A confirmation gates --- ORAL_A.md | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'ORAL_A.md') diff --git a/ORAL_A.md b/ORAL_A.md index e3427c4..1a07708 100644 --- a/ORAL_A.md +++ b/ORAL_A.md @@ -118,7 +118,9 @@ No recovery branch follows an A3 failure. Only after A3 passes, run exact BP, selected DFA, and selected SDIL at depths 20, 32, and 56 for model/data-loader seeds 10--14. Hyperparameters and epoch -budgets are copied without depth-specific tuning. The CIFAR-10 test set is +budgets are copied without depth-specific tuning. Confirmation refits on all +50,000 original training examples (`val_examples=0`) after development has +closed. The CIFAR-10 test set is evaluated exactly once at the end of every run; no intermediate test metric is logged or used for selection. -- cgit v1.2.3