20260228_014435/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/20260228_014435,outputs/task1_binary_runs/20260228_014435/best_model.pt,data_for_deberta/dedup_split/risk_test.csv,2893,2893,0.908745247148289,0.8674698795180723,0.818957345971564,0.9220917822838848,0.3238852402350501,0.6761147597649498