sim_eval_phase_b/README.md

Phase B confirm (50 ep/task, num_exp=3)

• checkpoint: ep-0700
• HF folder: sim_eval_phase_b/
• Phase A screen artifacts (if present) live under sim_eval/ and are not overwritten.

File	Description
Phase B confirm (50 ep/task, num_exp=3)_eval_log.json	Full metrics JSON
Phase B confirm (50 ep/task, num_exp=3)_dashboard.png	Per-task SR plot
Phase B confirm (50 ep/task, num_exp=3)_summary.md	Human-readable table
Phase B confirm (50 ep/task, num_exp=3)_run.log	Eval stdout