20260306_145453/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_145453,outputs/task1_binary_runs/20260306_145453/best_model.pt,data_for_deberta/dedup_split/risk_test.csv,2879,2879,0.8753039249739493,0.8077129084092126,0.7119924457034938,0.9331683168316832,0.28065300451545677,0.7193469954845433