post-train evals: humaneval pass@1 final

stage2_paper_faithful/eval_results/paper_faithful_humaneval_t1.json

@@ -0,0 +1,12 @@
+{
+  "pass_at_1": 0.06707317073170732,
+  "n_runs_pass": 11,
+  "n_runs_total": 164,
+  "n_tasks": 164,
+  "ckpt": null,
+  "model": "Qwen/Qwen3.5-0.8B-Base",
+  "surgery": "rtpurbo",
+  "surgery_state_dict": "checkpoints/rtpurbo_qwen3_5_0_8b_base/stage2_paper_faithful/state_dict.pt",
+  "manifest": null,
+  "source": "eval_results/paper_faithful_humaneval_t1_samples_eval_results.json"
+}