20260305_143247/analysis_bundle/summaries/final_eval_summary_evalonly.csv

run_dir,model_path,dataset,rows_raw,rows_evaluated,acc,f1,precision,recall,always_pick_acc,always_decline_acc
outputs/task1_binary_runs/20260305_143247,outputs/task1_binary_runs/20260305_143247/best_model.pt,data_for_deberta/dedup_split/risk_test.csv,2879,2879,0.8978812087530392,0.8414239482200647,0.8016443987667009,0.8853575482406356,0.3060090309135116,0.6939909690864884