summaryrefslogtreecommitdiff
path: root/ep_run/eval_relax_s3200.py
diff options
context:
space:
mode:
authorYuren Hao <yurenh2@illinois.edu>2026-08-04 04:13:43 -0500
committerYuren Hao <yurenh2@illinois.edu>2026-08-04 04:13:43 -0500
commit3962588a667d0d8945044b105dedda75e3fa3a41 (patch)
tree923698a7dacada2770c2c94860a3243b1496a04e /ep_run/eval_relax_s3200.py
parent1d2d0338ae25de7d60756545dad692c2f833c532 (diff)
RESULT 89: 135M判决 — 修复版3.2024 vs BP双seed均值3.2054(gap -0.0030=seed间距一半), 统计不可分
旧读出同点+0.3703=+44.8%ppl, 即"宽度墙"真实体量, 全部由一行读出改写消除。五尺寸(11M-135M) 修复后全部落入种子噪声。BP s2完赛best 3.0968。EP剩11.4k步出终局。 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014FAPDWQ49M5Ye3NpTndTpn
Diffstat (limited to 'ep_run/eval_relax_s3200.py')
0 files changed, 0 insertions, 0 deletions