20260305_160843/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/20260305_160843,outputs/task1_binary_runs/20260305_160843/best_model.pt,data_for_deberta/dedup_split/risk_test.csv,2879,2879,0.877387981938173,0.820538891713269,0.74860853432282,0.9077615298087739,0.30878777353247655,0.6912122264675234