20260306_061847/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_061847,outputs/task1_binary_runs/20260306_061847/best_model.pt,data_for_deberta/dedup_split/risk_test.csv,2879,2879,0.8749565821465787,0.8123044838373306,0.7049773755656109,0.958179581795818,0.28238971865230983,0.7176102813476901