20260222_175136/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/20260222_175136,outputs/task1_binary_runs/20260222_175136/best_model.pt,data_for_deberta/dedup_split/risk_test.csv,3389,3389,0.9321333726763057,0.8214285714285714,0.8252730109204368,0.8176197836166924,0.1909117733844792,0.8090882266155208