raw rescored eval logs, 6 open-weight monitors

add significance/power analysis

open-weight monitor spectrum: per-PR gradual vs positional discrimination

open-weight monitor spectrum: per-PR gradual vs positional discrimination

Replace analysis_llama31-8b.json: complete 283-eval run + relabel qwen->llama31_8b_runmax

Relabel columns in analysis_qwen25-7b.json (qwen->qwen25_7b_runmax)

Relabel columns in analysis_qwen25-14b.json (qwen->qwen25_14b_runmax)

open-weight monitor spectrum: per-PR gradual vs positional discrimination

Add open-monitor rescore backup: analysis_llama31-8b.json

Add open-monitor rescore backup: analysis_qwen25-7b.json

Add open-monitor rescore backup: scores_llama31-8b.tar.zst

Add open-monitor rescore backup: scores_qwen25-7b.tar.zst

Add open-monitor rescore backup: scores_qwen25-14b.tar.zst

Fix mislabeled columns in analysis_gpt4o.json (qwen->gpt4o_runmax, glm->base_glm_runmax)

Add GPT-4o rescore: analysis_gpt4o.json

Add GPT-4o rescore: scores_gpt4o.tar.zst

add Japan logs+pipeline

add F logs+pipeline

add E logs+pipeline

add D logs+pipeline

add manifest

initial commit