label,sigma,n_samples,rollouts,ckpt_root,ckpt_name,perturb,noise_distribution,mean_rollout_exact,mean_rollout_token_acc,pass_at_k,all_k,correct_count_mean,correct_count_std,correct_count_q10,correct_count_q50,correct_count_q90,zero_frac,full_frac trm_multi4_best,0.0,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8920000195503235,0.9585950970649719,0.8920000195503235,0.8920000195503235,7.136000156402588,2.4830431938171387,0.0,8.0,8.0,0.1080000028014183,0.8920000195503235 trm_multi4_best,3e-05,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8941249847412109,0.9596672654151917,0.9577999711036682,0.8095999956130981,7.1529998779296875,2.0940847396850586,4.0,8.0,8.0,0.0421999990940094,0.8095999956130981 trm_multi4_best,0.0001,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8919249773025513,0.9588295817375183,0.9574000239372253,0.8040000200271606,7.13539981842041,2.107004165649414,4.0,8.0,8.0,0.04259999841451645,0.8040000200271606 trm_multi4_best,0.0003,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8939999938011169,0.9595761895179749,0.9570000171661377,0.8118000030517578,7.1519999504089355,2.106584072113037,4.0,8.0,8.0,0.0430000014603138,0.8118000030517578 trm_multi4_best,0.001,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8934500217437744,0.9593700170516968,0.9535999894142151,0.8091999888420105,7.147600173950195,2.1107852458953857,4.0,8.0,8.0,0.04639999940991402,0.8091999888420105 trm_multi4_best,0.003,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8934749960899353,0.9593802690505981,0.9593999981880188,0.8101999759674072,7.147799968719482,2.1013221740722656,4.0,8.0,8.0,0.0406000018119812,0.8101999759674072 trm_multi4_best,0.01,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8938249945640564,0.9593154191970825,0.9599999785423279,0.7993999719619751,7.150599956512451,2.065894365310669,4.0,8.0,8.0,0.03999999910593033,0.7993999719619751 trm_multi4_best,0.03,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8929749727249146,0.9591527581214905,0.9606000185012817,0.7986000180244446,7.143799781799316,2.0669593811035156,4.0,8.0,8.0,0.039400000125169754,0.7986000180244446 trm_multi4_best,0.1,5000,8,/home/yurenh2/rrm/trm/checkpoints/Sudoku-extreme-1k-aug-1000-ACT-torch/pretrain_mlp_t_sudoku_official_gbs768_multi4_loguniform_repro,step_35805,both,gaussian,0.8917250037193298,0.9587355256080627,0.9639999866485596,0.79339998960495,7.133800029754639,2.062788724899292,4.0,8.0,8.0,0.035999998450279236,0.79339998960495