20260306_075806/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/20260306_075806,outputs/task1_binary_runs/20260306_075806/best_model.pt,data_for_deberta/dedup_split/risk_test.csv,2879,2879,0.8968391802709275,0.8354570637119113,0.7600806451612904,0.9274292742927429,0.28238971865230983,0.7176102813476901