{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.5, "eval_steps": 50, "global_step": 234, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4003.6, "completions/mean_length": 605.40107421875, "completions/mean_terminated_length": 529.9010864257813, "completions/min_length": 3.2, "completions/min_terminated_length": 3.2, "entropy": 0.25465373788028955, "epoch": 0.010683760683760684, "frac_reward_zero_std": 0.0, "grad_norm": 0.3200480043888092, "learning_rate": 4.2553191489361704e-07, "loss": 0.0007, "num_tokens": 8977579.0, "reward": 2.2984264850616456, "reward_std": 1.4261654615402222, "rewards/accuracy_reward/mean": 0.24833984375, "rewards/accuracy_reward/std": 0.4318830966949463, "rewards/brier_reward/mean": 0.2991298258304596, "rewards/brier_reward/std": 0.4151440501213074, "rewards/confidence_one_or_zero/mean": 0.3400390625, "rewards/confidence_one_or_zero/std": 0.4732877194881439, "rewards/format_reward/mean": 0.58623046875, "rewards/format_reward/std": 0.4925698161125183, "rewards/mean_confidence_reward/mean": 0.824687385559082, "rewards/mean_confidence_reward/std": 0.295770400762558, "step": 5, "step_time": 326.21757105822616 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 3745.8, "completions/mean_length": 604.62119140625, "completions/mean_terminated_length": 529.1286987304687, "completions/min_length": 6.6, "completions/min_terminated_length": 6.6, "entropy": 0.25508519501890986, "epoch": 0.021367521367521368, "frac_reward_zero_std": 0.0, "grad_norm": 2.269850492477417, "learning_rate": 9.574468085106384e-07, "loss": 0.0023, "num_tokens": 17931364.0, "reward": 2.3046632766723634, "reward_std": 1.4140528202056886, "rewards/accuracy_reward/mean": 0.24921875, "rewards/accuracy_reward/std": 0.4321252405643463, "rewards/brier_reward/mean": 0.3017696917057037, "rewards/brier_reward/std": 0.4140190303325653, "rewards/confidence_one_or_zero/mean": 0.32880859375, "rewards/confidence_one_or_zero/std": 0.46972647309303284, "rewards/format_reward/mean": 0.596875, "rewards/format_reward/std": 0.49034589529037476, "rewards/mean_confidence_reward/mean": 0.8279911875724792, "rewards/mean_confidence_reward/std": 0.2880759954452515, "step": 10, "step_time": 287.56116676276145 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02236328125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3914.0, "completions/mean_length": 602.30673828125, "completions/mean_terminated_length": 522.4651611328125, "completions/min_length": 8.4, "completions/min_terminated_length": 8.4, "entropy": 0.25588744818232956, "epoch": 0.03205128205128205, "frac_reward_zero_std": 0.0, "grad_norm": 0.9064725041389465, "learning_rate": 1.4893617021276596e-06, "loss": 0.0013, "num_tokens": 26816265.0, "reward": 2.288093090057373, "reward_std": 1.4102837562561035, "rewards/accuracy_reward/mean": 0.2458984375, "rewards/accuracy_reward/std": 0.4295665919780731, "rewards/brier_reward/mean": 0.2970101833343506, "rewards/brier_reward/std": 0.4126181125640869, "rewards/confidence_one_or_zero/mean": 0.3265625, "rewards/confidence_one_or_zero/std": 0.46868589520454407, "rewards/format_reward/mean": 0.58837890625, "rewards/format_reward/std": 0.49179903268814085, "rewards/mean_confidence_reward/mean": 0.8302430748939514, "rewards/mean_confidence_reward/std": 0.2874841451644897, "step": 15, "step_time": 323.37215538856105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.023828125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4035.6, "completions/mean_length": 626.5341796875, "completions/mean_terminated_length": 541.8522521972657, "completions/min_length": 3.4, "completions/min_terminated_length": 3.4, "entropy": 0.25333831016905606, "epoch": 0.042735042735042736, "frac_reward_zero_std": 0.0, "grad_norm": 1.3314036130905151, "learning_rate": 2.021276595744681e-06, "loss": 0.0032, "num_tokens": 36011015.0, "reward": 2.182456207275391, "reward_std": 1.3631898880004882, "rewards/accuracy_reward/mean": 0.20654296875, "rewards/accuracy_reward/std": 0.403618711233139, "rewards/brier_reward/mean": 0.2605383038520813, "rewards/brier_reward/std": 0.3904817759990692, "rewards/confidence_one_or_zero/mean": 0.32109375, "rewards/confidence_one_or_zero/std": 0.4667462706565857, "rewards/format_reward/mean": 0.5755859375, "rewards/format_reward/std": 0.49400630593299866, "rewards/mean_confidence_reward/mean": 0.818695342540741, "rewards/mean_confidence_reward/std": 0.30246469378471375, "step": 20, "step_time": 305.47666478096215 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02119140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4012.8, "completions/mean_length": 586.49697265625, "completions/mean_terminated_length": 510.53497314453125, "completions/min_length": 3.2, "completions/min_terminated_length": 3.2, "entropy": 0.2608537188032642, "epoch": 0.053418803418803416, "frac_reward_zero_std": 0.0, "grad_norm": 3.898252248764038, "learning_rate": 2.553191489361702e-06, "loss": 0.0022, "num_tokens": 44752328.0, "reward": 2.267803955078125, "reward_std": 1.41282000541687, "rewards/accuracy_reward/mean": 0.23798828125, "rewards/accuracy_reward/std": 0.42437363862991334, "rewards/brier_reward/mean": 0.2911687433719635, "rewards/brier_reward/std": 0.40819350481033323, "rewards/confidence_one_or_zero/mean": 0.32392578125, "rewards/confidence_one_or_zero/std": 0.46798876523971555, "rewards/format_reward/mean": 0.58642578125, "rewards/format_reward/std": 0.4922482430934906, "rewards/mean_confidence_reward/mean": 0.8282954454421997, "rewards/mean_confidence_reward/std": 0.2912151575088501, "step": 25, "step_time": 290.0282085505925 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02275390625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4032.8, "completions/mean_length": 610.38974609375, "completions/mean_terminated_length": 529.2322326660156, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.2174284148029983, "epoch": 0.0641025641025641, "frac_reward_zero_std": 0.0, "grad_norm": 17.150157928466797, "learning_rate": 3.0851063829787237e-06, "loss": 0.0009, "num_tokens": 53816895.0, "reward": 2.2753404140472413, "reward_std": 1.4286715030670165, "rewards/accuracy_reward/mean": 0.243359375, "rewards/accuracy_reward/std": 0.4284192621707916, "rewards/brier_reward/mean": 0.29526680111885073, "rewards/brier_reward/std": 0.4115871608257294, "rewards/confidence_one_or_zero/mean": 0.3330078125, "rewards/confidence_one_or_zero/std": 0.47137221693992615, "rewards/format_reward/mean": 0.58828125, "rewards/format_reward/std": 0.49213966727256775, "rewards/mean_confidence_reward/mean": 0.8154251933097839, "rewards/mean_confidence_reward/std": 0.3074728071689606, "step": 30, "step_time": 297.7445523235277 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0265625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4005.8, "completions/mean_length": 619.5669921875, "completions/mean_terminated_length": 524.9031799316406, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.15976690110983327, "epoch": 0.07478632478632478, "frac_reward_zero_std": 0.0, "grad_norm": 39.64427185058594, "learning_rate": 3.6170212765957453e-06, "loss": 0.0009, "num_tokens": 62908781.0, "reward": 2.244691181182861, "reward_std": 1.3894657611846923, "rewards/accuracy_reward/mean": 0.23349609375, "rewards/accuracy_reward/std": 0.4228110373020172, "rewards/brier_reward/mean": 0.28838067054748534, "rewards/brier_reward/std": 0.4061505198478699, "rewards/confidence_one_or_zero/mean": 0.311328125, "rewards/confidence_one_or_zero/std": 0.4631028950214386, "rewards/format_reward/mean": 0.58876953125, "rewards/format_reward/std": 0.4920034110546112, "rewards/mean_confidence_reward/mean": 0.822716748714447, "rewards/mean_confidence_reward/std": 0.2938410401344299, "step": 35, "step_time": 319.35700375636804 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02001953125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3920.4, "completions/mean_length": 589.07705078125, "completions/mean_terminated_length": 517.4498657226562, "completions/min_length": 4.6, "completions/min_terminated_length": 4.6, "entropy": 0.11105124278110452, "epoch": 0.08547008547008547, "frac_reward_zero_std": 0.0, "grad_norm": 184.02346801757812, "learning_rate": 4.148936170212766e-06, "loss": 0.0007, "num_tokens": 71661602.0, "reward": 2.29984655380249, "reward_std": 1.3992369890213012, "rewards/accuracy_reward/mean": 0.23974609375, "rewards/accuracy_reward/std": 0.42584521174430845, "rewards/brier_reward/mean": 0.2921775460243225, "rewards/brier_reward/std": 0.4097095191478729, "rewards/confidence_one_or_zero/mean": 0.33916015625, "rewards/confidence_one_or_zero/std": 0.47322710752487185, "rewards/format_reward/mean": 0.58896484375, "rewards/format_reward/std": 0.4918659806251526, "rewards/mean_confidence_reward/mean": 0.8397979259490966, "rewards/mean_confidence_reward/std": 0.2777191400527954, "step": 40, "step_time": 307.38189966618376 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.025, "completions/max_length": 4096.0, "completions/max_terminated_length": 3979.6, "completions/mean_length": 616.15703125, "completions/mean_terminated_length": 526.949755859375, "completions/min_length": 4.2, "completions/min_terminated_length": 4.2, "entropy": 0.12535708377254196, "epoch": 0.09615384615384616, "frac_reward_zero_std": 0.0, "grad_norm": 273.0317077636719, "learning_rate": 4.680851063829788e-06, "loss": 0.0032, "num_tokens": 80759978.0, "reward": 2.2776510238647463, "reward_std": 1.4421053409576416, "rewards/accuracy_reward/mean": 0.24677734375, "rewards/accuracy_reward/std": 0.4308508217334747, "rewards/brier_reward/mean": 0.2974667429924011, "rewards/brier_reward/std": 0.4150512158870697, "rewards/confidence_one_or_zero/mean": 0.33642578125, "rewards/confidence_one_or_zero/std": 0.47240622639656066, "rewards/format_reward/mean": 0.57841796875, "rewards/format_reward/std": 0.49370808005332945, "rewards/mean_confidence_reward/mean": 0.8185631394386291, "rewards/mean_confidence_reward/std": 0.30447315573692324, "step": 45, "step_time": 311.8303961681813 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0234375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3974.2, "completions/mean_length": 612.52216796875, "completions/mean_terminated_length": 528.9533386230469, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.18446943076560274, "epoch": 0.10683760683760683, "frac_reward_zero_std": 0.0, "grad_norm": 134.59556579589844, "learning_rate": 4.9465240641711236e-06, "loss": 0.0015, "num_tokens": 89818829.0, "reward": 2.2920437812805177, "reward_std": 1.4526799678802491, "rewards/accuracy_reward/mean": 0.24951171875, "rewards/accuracy_reward/std": 0.4312021195888519, "rewards/brier_reward/mean": 0.29880380630493164, "rewards/brier_reward/std": 0.4153770744800568, "rewards/confidence_one_or_zero/mean": 0.3498046875, "rewards/confidence_one_or_zero/std": 0.47673011422157285, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.4936447739601135, "rewards/mean_confidence_reward/mean": 0.8157985091209412, "rewards/mean_confidence_reward/std": 0.3079785585403442, "step": 50, "step_time": 290.073405100184 }, { "epoch": 0.10683760683760683, "eval_clip_ratio/high_max": 0.0, "eval_clip_ratio/high_mean": 0.0, "eval_clip_ratio/low_mean": 0.0, "eval_clip_ratio/low_min": 0.0, "eval_clip_ratio/region_mean": 0.0, "eval_completions/clipped_ratio": 0.01953125, "eval_completions/max_length": 2901.71875, "eval_completions/max_terminated_length": 1713.96875, "eval_completions/mean_length": 577.7978515625, "eval_completions/mean_terminated_length": 507.27635383605957, "eval_completions/min_length": 119.90625, "eval_completions/min_terminated_length": 119.90625, "eval_entropy": 0.18825100967660546, "eval_frac_reward_zero_std": 1.0, "eval_loss": 0.0, "eval_num_tokens": 89818829.0, "eval_reward": 2.3230195492506027, "eval_reward_std": 1.399896901100874, "eval_rewards/accuracy_reward/mean": 0.2431640625, "eval_rewards/accuracy_reward/std": 0.4234647727571428, "eval_rewards/brier_reward/mean": 0.29576857411302626, "eval_rewards/brier_reward/std": 0.40745917009189725, "eval_rewards/confidence_one_or_zero/mean": 0.3486328125, "eval_rewards/confidence_one_or_zero/std": 0.47755033802241087, "eval_rewards/format_reward/mean": 0.6103515625, "eval_rewards/format_reward/std": 0.4917288227006793, "eval_rewards/mean_confidence_reward/mean": 0.8251025322824717, "eval_rewards/mean_confidence_reward/std": 0.29678282886743546, "eval_runtime": 1046.3126, "eval_samples_per_second": 0.956, "eval_steps_per_second": 0.031, "step": 50 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02412109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4027.8, "completions/mean_length": 625.873046875, "completions/mean_terminated_length": 540.0704345703125, "completions/min_length": 4.6, "completions/min_terminated_length": 4.6, "entropy": 0.17808166444301604, "epoch": 0.11752136752136752, "frac_reward_zero_std": 0.0, "grad_norm": 253.82821655273438, "learning_rate": 4.812834224598931e-06, "loss": 0.0053, "num_tokens": 98963737.0, "reward": 2.269825267791748, "reward_std": 1.405194306373596, "rewards/accuracy_reward/mean": 0.23798828125, "rewards/accuracy_reward/std": 0.4250611007213593, "rewards/brier_reward/mean": 0.2912949800491333, "rewards/brier_reward/std": 0.40859017968177797, "rewards/confidence_one_or_zero/mean": 0.32822265625, "rewards/confidence_one_or_zero/std": 0.46954097151756286, "rewards/format_reward/mean": 0.58798828125, "rewards/format_reward/std": 0.49162101149559023, "rewards/mean_confidence_reward/mean": 0.8243311047554016, "rewards/mean_confidence_reward/std": 0.29433915615081785, "step": 55, "step_time": 292.7474593430117 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.023828125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3962.6, "completions/mean_length": 619.4044921875, "completions/mean_terminated_length": 534.6329223632813, "completions/min_length": 14.2, "completions/min_terminated_length": 14.2, "entropy": 0.18673709755530582, "epoch": 0.1282051282051282, "frac_reward_zero_std": 0.0, "grad_norm": 168.89572143554688, "learning_rate": 4.6791443850267385e-06, "loss": -0.0028, "num_tokens": 108062663.0, "reward": 2.220626974105835, "reward_std": 1.403502869606018, "rewards/accuracy_reward/mean": 0.2279296875, "rewards/accuracy_reward/std": 0.41913889050483705, "rewards/brier_reward/mean": 0.27917802333831787, "rewards/brier_reward/std": 0.40315498113632203, "rewards/confidence_one_or_zero/mean": 0.32275390625, "rewards/confidence_one_or_zero/std": 0.46722970604896547, "rewards/format_reward/mean": 0.5724609375, "rewards/format_reward/std": 0.4941402614116669, "rewards/mean_confidence_reward/mean": 0.8183044075965882, "rewards/mean_confidence_reward/std": 0.30232587456703186, "step": 60, "step_time": 299.5969597796153 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02529296875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4007.6, "completions/mean_length": 636.066015625, "completions/mean_terminated_length": 546.2889099121094, "completions/min_length": 2.6, "completions/min_terminated_length": 2.6, "entropy": 0.18173209796659648, "epoch": 0.1388888888888889, "frac_reward_zero_std": 0.0, "grad_norm": 226.22181701660156, "learning_rate": 4.5454545454545455e-06, "loss": 0.0017, "num_tokens": 117312555.0, "reward": 2.2279109001159667, "reward_std": 1.4209685325622559, "rewards/accuracy_reward/mean": 0.23447265625, "rewards/accuracy_reward/std": 0.4231623291969299, "rewards/brier_reward/mean": 0.28496766090393066, "rewards/brier_reward/std": 0.4072230398654938, "rewards/confidence_one_or_zero/mean": 0.31845703125, "rewards/confidence_one_or_zero/std": 0.46585850715637206, "rewards/format_reward/mean": 0.5765625, "rewards/format_reward/std": 0.49394460320472716, "rewards/mean_confidence_reward/mean": 0.8134510278701782, "rewards/mean_confidence_reward/std": 0.30829231142997743, "step": 65, "step_time": 350.8520451569901 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.023046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4050.2, "completions/mean_length": 623.25732421875, "completions/mean_terminated_length": 541.5183837890625, "completions/min_length": 2.2, "completions/min_terminated_length": 2.2, "entropy": 0.19747755720745772, "epoch": 0.14957264957264957, "frac_reward_zero_std": 0.0, "grad_norm": 190.5092010498047, "learning_rate": 4.411764705882353e-06, "loss": 0.0018, "num_tokens": 126463734.0, "reward": 2.2359805583953856, "reward_std": 1.4184848308563232, "rewards/accuracy_reward/mean": 0.230859375, "rewards/accuracy_reward/std": 0.41946170330047605, "rewards/brier_reward/mean": 0.283882212638855, "rewards/brier_reward/std": 0.40468925833702085, "rewards/confidence_one_or_zero/mean": 0.33095703125, "rewards/confidence_one_or_zero/std": 0.4701420605182648, "rewards/format_reward/mean": 0.57802734375, "rewards/format_reward/std": 0.4933392107486725, "rewards/mean_confidence_reward/mean": 0.8122546076774597, "rewards/mean_confidence_reward/std": 0.30792644023895266, "step": 70, "step_time": 302.926274932205 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.023046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4024.2, "completions/mean_length": 624.4099609375, "completions/mean_terminated_length": 542.5408142089843, "completions/min_length": 2.4, "completions/min_terminated_length": 2.4, "entropy": 0.22855277470080182, "epoch": 0.16025641025641027, "frac_reward_zero_std": 0.0, "grad_norm": 146.3217010498047, "learning_rate": 4.2780748663101604e-06, "loss": 0.001, "num_tokens": 135590524.0, "reward": 2.239028310775757, "reward_std": 1.4247822284698486, "rewards/accuracy_reward/mean": 0.2310546875, "rewards/accuracy_reward/std": 0.4212393045425415, "rewards/brier_reward/mean": 0.2818598568439484, "rewards/brier_reward/std": 0.4061828851699829, "rewards/confidence_one_or_zero/mean": 0.33271484375, "rewards/confidence_one_or_zero/std": 0.47072147727012636, "rewards/format_reward/mean": 0.57685546875, "rewards/format_reward/std": 0.4937345623970032, "rewards/mean_confidence_reward/mean": 0.8165434002876282, "rewards/mean_confidence_reward/std": 0.3071639001369476, "step": 75, "step_time": 290.6053208070225 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0203125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4012.6, "completions/mean_length": 585.38388671875, "completions/mean_terminated_length": 512.5918395996093, "completions/min_length": 2.2, "completions/min_terminated_length": 2.2, "entropy": 0.19709544841898605, "epoch": 0.17094017094017094, "frac_reward_zero_std": 0.0, "grad_norm": 123.4002456665039, "learning_rate": 4.144385026737968e-06, "loss": 0.0023, "num_tokens": 144329271.0, "reward": 2.226662302017212, "reward_std": 1.3714903354644776, "rewards/accuracy_reward/mean": 0.21884765625, "rewards/accuracy_reward/std": 0.4126656711101532, "rewards/brier_reward/mean": 0.27575421929359434, "rewards/brier_reward/std": 0.3990472376346588, "rewards/confidence_one_or_zero/mean": 0.3255859375, "rewards/confidence_one_or_zero/std": 0.46841105818748474, "rewards/format_reward/mean": 0.58466796875, "rewards/format_reward/std": 0.49257957339286806, "rewards/mean_confidence_reward/mean": 0.8218065023422241, "rewards/mean_confidence_reward/std": 0.2959056913852692, "step": 80, "step_time": 294.6295507659932 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.023828125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4035.6, "completions/mean_length": 609.34521484375, "completions/mean_terminated_length": 524.2319213867188, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.18681974695064127, "epoch": 0.18162393162393162, "frac_reward_zero_std": 0.0, "grad_norm": 125.80805206298828, "learning_rate": 4.010695187165775e-06, "loss": 0.0036, "num_tokens": 153292038.0, "reward": 2.259104061126709, "reward_std": 1.4132461547851562, "rewards/accuracy_reward/mean": 0.23759765625, "rewards/accuracy_reward/std": 0.4247436225414276, "rewards/brier_reward/mean": 0.2869098663330078, "rewards/brier_reward/std": 0.4095331788063049, "rewards/confidence_one_or_zero/mean": 0.33935546875, "rewards/confidence_one_or_zero/std": 0.47348698377609255, "rewards/format_reward/mean": 0.574609375, "rewards/format_reward/std": 0.4944670259952545, "rewards/mean_confidence_reward/mean": 0.8206317663192749, "rewards/mean_confidence_reward/std": 0.30208303928375246, "step": 85, "step_time": 312.40206820861783 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02314453125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3999.6, "completions/mean_length": 613.43818359375, "completions/mean_terminated_length": 530.9699645996094, "completions/min_length": 2.2, "completions/min_terminated_length": 2.2, "entropy": 0.18488237902056426, "epoch": 0.19230769230769232, "frac_reward_zero_std": 0.0, "grad_norm": 285.5928039550781, "learning_rate": 3.877005347593583e-06, "loss": 0.0063, "num_tokens": 162289293.0, "reward": 2.240404558181763, "reward_std": 1.3770250797271728, "rewards/accuracy_reward/mean": 0.2240234375, "rewards/accuracy_reward/std": 0.41667423844337464, "rewards/brier_reward/mean": 0.2779018759727478, "rewards/brier_reward/std": 0.40121889114379883, "rewards/confidence_one_or_zero/mean": 0.32548828125, "rewards/confidence_one_or_zero/std": 0.46859588027000426, "rewards/format_reward/mean": 0.58828125, "rewards/format_reward/std": 0.49194942116737367, "rewards/mean_confidence_reward/mean": 0.8247097253799438, "rewards/mean_confidence_reward/std": 0.29405966997146604, "step": 90, "step_time": 292.76033823157485 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02255859375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4060.0, "completions/mean_length": 612.75439453125, "completions/mean_terminated_length": 532.3500061035156, "completions/min_length": 2.8, "completions/min_terminated_length": 2.8, "entropy": 0.18186008472694085, "epoch": 0.202991452991453, "frac_reward_zero_std": 0.0, "grad_norm": 211.09112548828125, "learning_rate": 3.7433155080213907e-06, "loss": 0.0003, "num_tokens": 171310682.0, "reward": 2.274180221557617, "reward_std": 1.4152658700942993, "rewards/accuracy_reward/mean": 0.2388671875, "rewards/accuracy_reward/std": 0.425787216424942, "rewards/brier_reward/mean": 0.2903932392597198, "rewards/brier_reward/std": 0.4095348954200745, "rewards/confidence_one_or_zero/mean": 0.33671875, "rewards/confidence_one_or_zero/std": 0.47252548336982725, "rewards/format_reward/mean": 0.58359375, "rewards/format_reward/std": 0.49275891184806825, "rewards/mean_confidence_reward/mean": 0.8246072769165039, "rewards/mean_confidence_reward/std": 0.2949615716934204, "step": 95, "step_time": 304.3870223330421 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.021875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3940.8, "completions/mean_length": 586.11689453125, "completions/mean_terminated_length": 507.6039672851563, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.19529976148623973, "epoch": 0.21367521367521367, "frac_reward_zero_std": 0.0, "grad_norm": 138.3889617919922, "learning_rate": 3.609625668449198e-06, "loss": 0.0035, "num_tokens": 180046023.0, "reward": 2.296976900100708, "reward_std": 1.4095505952835083, "rewards/accuracy_reward/mean": 0.25126953125, "rewards/accuracy_reward/std": 0.4335032939910889, "rewards/brier_reward/mean": 0.30398558974266054, "rewards/brier_reward/std": 0.4151215970516205, "rewards/confidence_one_or_zero/mean": 0.31865234375, "rewards/confidence_one_or_zero/std": 0.4660069942474365, "rewards/format_reward/mean": 0.59482421875, "rewards/format_reward/std": 0.4907715082168579, "rewards/mean_confidence_reward/mean": 0.8282452464103699, "rewards/mean_confidence_reward/std": 0.2885895729064941, "step": 100, "step_time": 293.22695188358483 }, { "epoch": 0.21367521367521367, "eval_clip_ratio/high_max": 0.0, "eval_clip_ratio/high_mean": 0.0, "eval_clip_ratio/low_mean": 0.0, "eval_clip_ratio/low_min": 0.0, "eval_clip_ratio/region_mean": 0.0, "eval_completions/clipped_ratio": 0.0205078125, "eval_completions/max_length": 2982.65625, "eval_completions/max_terminated_length": 1988.53125, "eval_completions/mean_length": 603.236328125, "eval_completions/mean_terminated_length": 529.7934732437134, "eval_completions/min_length": 121.59375, "eval_completions/min_terminated_length": 121.59375, "eval_entropy": 0.16897299326956272, "eval_frac_reward_zero_std": 1.0, "eval_loss": 0.0, "eval_num_tokens": 180046023.0, "eval_reward": 2.2179745994508266, "eval_reward_std": 1.402235945686698, "eval_rewards/accuracy_reward/mean": 0.2197265625, "eval_rewards/accuracy_reward/std": 0.40796902496367693, "eval_rewards/brier_reward/mean": 0.27754979068413377, "eval_rewards/brier_reward/std": 0.3955977242439985, "eval_rewards/confidence_one_or_zero/mean": 0.3310546875, "eval_rewards/confidence_one_or_zero/std": 0.4708655485883355, "eval_rewards/format_reward/mean": 0.58203125, "eval_rewards/format_reward/std": 0.49433132354170084, "eval_rewards/mean_confidence_reward/mean": 0.807612294331193, "eval_rewards/mean_confidence_reward/std": 0.3074530893936753, "eval_runtime": 1000.2467, "eval_samples_per_second": 1.0, "eval_steps_per_second": 0.032, "step": 100 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02314453125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4022.6, "completions/mean_length": 603.033984375, "completions/mean_terminated_length": 520.2775390625, "completions/min_length": 3.6, "completions/min_terminated_length": 3.6, "entropy": 0.19702966129407287, "epoch": 0.22435897435897437, "frac_reward_zero_std": 0.0, "grad_norm": 260.79986572265625, "learning_rate": 3.4759358288770056e-06, "loss": -0.0003, "num_tokens": 188956099.0, "reward": 2.2500012874603272, "reward_std": 1.4009554624557494, "rewards/accuracy_reward/mean": 0.22607421875, "rewards/accuracy_reward/std": 0.41807770133018496, "rewards/brier_reward/mean": 0.27851803302764894, "rewards/brier_reward/std": 0.4028440773487091, "rewards/confidence_one_or_zero/mean": 0.33525390625, "rewards/confidence_one_or_zero/std": 0.47164865732192995, "rewards/format_reward/mean": 0.5900390625, "rewards/format_reward/std": 0.4916222393512726, "rewards/mean_confidence_reward/mean": 0.8201160430908203, "rewards/mean_confidence_reward/std": 0.3016093969345093, "step": 105, "step_time": 318.2355111879093 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02001953125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3848.6, "completions/mean_length": 588.2689453125, "completions/mean_terminated_length": 516.6307312011719, "completions/min_length": 2.2, "completions/min_terminated_length": 2.2, "entropy": 0.2327494657598436, "epoch": 0.23504273504273504, "frac_reward_zero_std": 0.0, "grad_norm": 160.7519073486328, "learning_rate": 3.342245989304813e-06, "loss": 0.0025, "num_tokens": 197730821.0, "reward": 2.30214204788208, "reward_std": 1.395893669128418, "rewards/accuracy_reward/mean": 0.2439453125, "rewards/accuracy_reward/std": 0.42903687953948977, "rewards/brier_reward/mean": 0.29708718657493594, "rewards/brier_reward/std": 0.4112724483013153, "rewards/confidence_one_or_zero/mean": 0.325, "rewards/confidence_one_or_zero/std": 0.46816784143447876, "rewards/format_reward/mean": 0.59697265625, "rewards/format_reward/std": 0.4904723346233368, "rewards/mean_confidence_reward/mean": 0.8391369104385376, "rewards/mean_confidence_reward/std": 0.2748624235391617, "step": 110, "step_time": 303.38657617358257 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0232421875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3963.4, "completions/mean_length": 610.75927734375, "completions/mean_terminated_length": 527.7982788085938, "completions/min_length": 3.6, "completions/min_terminated_length": 3.6, "entropy": 0.23419087056536228, "epoch": 0.24572649572649571, "frac_reward_zero_std": 0.0, "grad_norm": 231.57655334472656, "learning_rate": 3.2085561497326205e-06, "loss": -0.0043, "num_tokens": 206727172.0, "reward": 2.2810462951660155, "reward_std": 1.417106819152832, "rewards/accuracy_reward/mean": 0.24169921875, "rewards/accuracy_reward/std": 0.4279089987277985, "rewards/brier_reward/mean": 0.2938065826892853, "rewards/brier_reward/std": 0.4112499177455902, "rewards/confidence_one_or_zero/mean": 0.32958984375, "rewards/confidence_one_or_zero/std": 0.4697917103767395, "rewards/format_reward/mean": 0.58798828125, "rewards/format_reward/std": 0.49208595752716067, "rewards/mean_confidence_reward/mean": 0.8279623985290527, "rewards/mean_confidence_reward/std": 0.2909824550151825, "step": 115, "step_time": 295.1056752460019 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0275390625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4063.2, "completions/mean_length": 633.58544921875, "completions/mean_terminated_length": 535.7213073730469, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.22616069127107039, "epoch": 0.2564102564102564, "frac_reward_zero_std": 0.0, "grad_norm": 236.416015625, "learning_rate": 3.074866310160428e-06, "loss": 0.0006, "num_tokens": 215966319.0, "reward": 2.204629898071289, "reward_std": 1.402574348449707, "rewards/accuracy_reward/mean": 0.2203125, "rewards/accuracy_reward/std": 0.41426042914390565, "rewards/brier_reward/mean": 0.27585427165031434, "rewards/brier_reward/std": 0.40066431164741517, "rewards/confidence_one_or_zero/mean": 0.32587890625, "rewards/confidence_one_or_zero/std": 0.46854068636894225, "rewards/format_reward/mean": 0.56953125, "rewards/format_reward/std": 0.49505711197853086, "rewards/mean_confidence_reward/mean": 0.8130529999732972, "rewards/mean_confidence_reward/std": 0.3048666715621948, "step": 120, "step_time": 309.6269192967884 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.023828125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4073.6, "completions/mean_length": 616.9107421875, "completions/mean_terminated_length": 531.9693237304688, "completions/min_length": 7.6, "completions/min_terminated_length": 7.6, "entropy": 0.22710780663182958, "epoch": 0.2670940170940171, "frac_reward_zero_std": 0.0, "grad_norm": 272.6275939941406, "learning_rate": 2.9411764705882355e-06, "loss": 0.0047, "num_tokens": 225034301.0, "reward": 2.245972490310669, "reward_std": 1.401333737373352, "rewards/accuracy_reward/mean": 0.235546875, "rewards/accuracy_reward/std": 0.42410356998443605, "rewards/brier_reward/mean": 0.2896265983581543, "rewards/brier_reward/std": 0.4079780995845795, "rewards/confidence_one_or_zero/mean": 0.3171875, "rewards/confidence_one_or_zero/std": 0.4653353154659271, "rewards/format_reward/mean": 0.58251953125, "rewards/format_reward/std": 0.4932447850704193, "rewards/mean_confidence_reward/mean": 0.8210919857025146, "rewards/mean_confidence_reward/std": 0.296706360578537, "step": 125, "step_time": 312.56211414858114 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02431640625, "completions/max_length": 4096.0, "completions/max_terminated_length": 3942.6, "completions/mean_length": 608.156640625, "completions/mean_terminated_length": 521.2976989746094, "completions/min_length": 4.4, "completions/min_terminated_length": 4.4, "entropy": 0.23180258893407882, "epoch": 0.2777777777777778, "frac_reward_zero_std": 0.0, "grad_norm": 228.5140838623047, "learning_rate": 2.807486631016043e-06, "loss": -0.0055, "num_tokens": 234007841.0, "reward": 2.3002397537231447, "reward_std": 1.4011871337890625, "rewards/accuracy_reward/mean": 0.24541015625, "rewards/accuracy_reward/std": 0.42985286116600036, "rewards/brier_reward/mean": 0.2962680757045746, "rewards/brier_reward/std": 0.41340049505233767, "rewards/confidence_one_or_zero/mean": 0.33349609375, "rewards/confidence_one_or_zero/std": 0.4714070737361908, "rewards/format_reward/mean": 0.5947265625, "rewards/format_reward/std": 0.49059703946113586, "rewards/mean_confidence_reward/mean": 0.8303389072418212, "rewards/mean_confidence_reward/std": 0.28693812489509585, "step": 130, "step_time": 309.1821164358465 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02021484375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3899.6, "completions/mean_length": 583.5189453125, "completions/mean_terminated_length": 511.06353149414065, "completions/min_length": 4.0, "completions/min_terminated_length": 4.0, "entropy": 0.22628546367632224, "epoch": 0.28846153846153844, "frac_reward_zero_std": 0.0, "grad_norm": 223.32774353027344, "learning_rate": 2.673796791443851e-06, "loss": 0.0061, "num_tokens": 242719683.0, "reward": 2.232788610458374, "reward_std": 1.3766098499298096, "rewards/accuracy_reward/mean": 0.229296875, "rewards/accuracy_reward/std": 0.41989706754684447, "rewards/brier_reward/mean": 0.2834498047828674, "rewards/brier_reward/std": 0.4037898898124695, "rewards/confidence_one_or_zero/mean": 0.3126953125, "rewards/confidence_one_or_zero/std": 0.4631091594696045, "rewards/format_reward/mean": 0.57978515625, "rewards/format_reward/std": 0.4935327351093292, "rewards/mean_confidence_reward/mean": 0.8275614142417907, "rewards/mean_confidence_reward/std": 0.2873734176158905, "step": 135, "step_time": 293.102621970803 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.01982421875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4007.4, "completions/mean_length": 594.617578125, "completions/mean_terminated_length": 523.8919799804687, "completions/min_length": 3.8, "completions/min_terminated_length": 3.8, "entropy": 0.21956374138826504, "epoch": 0.29914529914529914, "frac_reward_zero_std": 0.0, "grad_norm": 255.9053497314453, "learning_rate": 2.5401069518716583e-06, "loss": 0.0018, "num_tokens": 251593687.0, "reward": 2.292392873764038, "reward_std": 1.4128401756286622, "rewards/accuracy_reward/mean": 0.2466796875, "rewards/accuracy_reward/std": 0.4307744026184082, "rewards/brier_reward/mean": 0.2993623435497284, "rewards/brier_reward/std": 0.41323343515396116, "rewards/confidence_one_or_zero/mean": 0.32763671875, "rewards/confidence_one_or_zero/std": 0.4693611204624176, "rewards/format_reward/mean": 0.58935546875, "rewards/format_reward/std": 0.4918531060218811, "rewards/mean_confidence_reward/mean": 0.829358720779419, "rewards/mean_confidence_reward/std": 0.2864940404891968, "step": 140, "step_time": 315.13789040882256 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0220703125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3988.0, "completions/mean_length": 620.67421875, "completions/mean_terminated_length": 542.4297973632813, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.21909176429035143, "epoch": 0.30982905982905984, "frac_reward_zero_std": 0.0, "grad_norm": 261.5726013183594, "learning_rate": 2.4064171122994653e-06, "loss": 0.0009, "num_tokens": 260723887.0, "reward": 2.264259195327759, "reward_std": 1.42226722240448, "rewards/accuracy_reward/mean": 0.24033203125, "rewards/accuracy_reward/std": 0.42714357376098633, "rewards/brier_reward/mean": 0.2899028182029724, "rewards/brier_reward/std": 0.410666686296463, "rewards/confidence_one_or_zero/mean": 0.3353515625, "rewards/confidence_one_or_zero/std": 0.4718469202518463, "rewards/format_reward/mean": 0.578125, "rewards/format_reward/std": 0.49331154227256774, "rewards/mean_confidence_reward/mean": 0.8205477595329285, "rewards/mean_confidence_reward/std": 0.30001514554023745, "step": 145, "step_time": 327.641197694774 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0201171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4027.4, "completions/mean_length": 586.6529296875, "completions/mean_terminated_length": 514.5625793457032, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.21774800701532512, "epoch": 0.32051282051282054, "frac_reward_zero_std": 0.0, "grad_norm": 214.8507843017578, "learning_rate": 2.2727272727272728e-06, "loss": 0.0014, "num_tokens": 269479213.0, "reward": 2.238347053527832, "reward_std": 1.3909776210784912, "rewards/accuracy_reward/mean": 0.22890625, "rewards/accuracy_reward/std": 0.41975467205047606, "rewards/brier_reward/mean": 0.2842195689678192, "rewards/brier_reward/std": 0.4028289198875427, "rewards/confidence_one_or_zero/mean": 0.311328125, "rewards/confidence_one_or_zero/std": 0.4628018081188202, "rewards/format_reward/mean": 0.5876953125, "rewards/format_reward/std": 0.4922152101993561, "rewards/mean_confidence_reward/mean": 0.8261977553367614, "rewards/mean_confidence_reward/std": 0.2895132124423981, "step": 150, "step_time": 321.3506028565462 }, { "epoch": 0.32051282051282054, "eval_clip_ratio/high_max": 0.0, "eval_clip_ratio/high_mean": 0.0, "eval_clip_ratio/low_mean": 0.0, "eval_clip_ratio/low_min": 0.0, "eval_clip_ratio/region_mean": 0.0, "eval_completions/clipped_ratio": 0.0224609375, "eval_completions/max_length": 3129.46875, "eval_completions/max_terminated_length": 1870.03125, "eval_completions/mean_length": 612.17578125, "eval_completions/mean_terminated_length": 532.1511011123657, "eval_completions/min_length": 133.6875, "eval_completions/min_terminated_length": 133.6875, "eval_entropy": 0.21968279918655753, "eval_frac_reward_zero_std": 1.0, "eval_loss": 0.0, "eval_num_tokens": 269479213.0, "eval_reward": 2.3402221016585827, "eval_reward_std": 1.4399862512946129, "eval_rewards/accuracy_reward/mean": 0.2587890625, "eval_rewards/accuracy_reward/std": 0.4342051614075899, "eval_rewards/brier_reward/mean": 0.31138324039056897, "eval_rewards/brier_reward/std": 0.4186254357919097, "eval_rewards/confidence_one_or_zero/mean": 0.3466796875, "eval_rewards/confidence_one_or_zero/std": 0.4757600771263242, "eval_rewards/format_reward/mean": 0.6005859375, "eval_rewards/format_reward/std": 0.48949996195733547, "eval_rewards/mean_confidence_reward/mean": 0.822784174233675, "eval_rewards/mean_confidence_reward/std": 0.2945121091324836, "eval_runtime": 1148.2714, "eval_samples_per_second": 0.871, "eval_steps_per_second": 0.028, "step": 150 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0205078125, "completions/max_length": 4096.0, "completions/max_terminated_length": 3980.8, "completions/mean_length": 573.84677734375, "completions/mean_terminated_length": 500.0541076660156, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.24126820733072235, "epoch": 0.3311965811965812, "frac_reward_zero_std": 0.0, "grad_norm": 188.24107360839844, "learning_rate": 2.1390374331550802e-06, "loss": 0.0021, "num_tokens": 278106956.0, "reward": 2.270985221862793, "reward_std": 1.3821493864059449, "rewards/accuracy_reward/mean": 0.232421875, "rewards/accuracy_reward/std": 0.4222895622253418, "rewards/brier_reward/mean": 0.2869627296924591, "rewards/brier_reward/std": 0.4042757868766785, "rewards/confidence_one_or_zero/mean": 0.31904296875, "rewards/confidence_one_or_zero/std": 0.4660694539546967, "rewards/format_reward/mean": 0.60419921875, "rewards/format_reward/std": 0.48895737528800964, "rewards/mean_confidence_reward/mean": 0.8283583998680115, "rewards/mean_confidence_reward/std": 0.2896300792694092, "step": 155, "step_time": 319.6077715688851 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0208984375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3940.6, "completions/mean_length": 575.42421875, "completions/mean_terminated_length": 500.2835205078125, "completions/min_length": 2.6, "completions/min_terminated_length": 2.6, "entropy": 0.2091839660773985, "epoch": 0.3418803418803419, "frac_reward_zero_std": 0.0, "grad_norm": 286.1556396484375, "learning_rate": 2.0053475935828877e-06, "loss": 0.0009, "num_tokens": 286732740.0, "reward": 2.309709072113037, "reward_std": 1.4149959325790404, "rewards/accuracy_reward/mean": 0.25302734375, "rewards/accuracy_reward/std": 0.43359568119049074, "rewards/brier_reward/mean": 0.3039921760559082, "rewards/brier_reward/std": 0.4155700445175171, "rewards/confidence_one_or_zero/mean": 0.32490234375, "rewards/confidence_one_or_zero/std": 0.46794300675392153, "rewards/format_reward/mean": 0.59580078125, "rewards/format_reward/std": 0.4904149293899536, "rewards/mean_confidence_reward/mean": 0.831986379623413, "rewards/mean_confidence_reward/std": 0.28579835295677186, "step": 160, "step_time": 291.7342998010048 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.019140625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4007.6, "completions/mean_length": 575.7037109375, "completions/mean_terminated_length": 507.00523071289064, "completions/min_length": 2.4, "completions/min_terminated_length": 2.4, "entropy": 0.22379019369836897, "epoch": 0.3525641025641026, "frac_reward_zero_std": 0.0, "grad_norm": 250.2082061767578, "learning_rate": 1.8716577540106954e-06, "loss": -0.0016, "num_tokens": 295378954.0, "reward": 2.311230993270874, "reward_std": 1.4038845539093017, "rewards/accuracy_reward/mean": 0.24814453125, "rewards/accuracy_reward/std": 0.43102404475212097, "rewards/brier_reward/mean": 0.30173795819282534, "rewards/brier_reward/std": 0.41356743574142457, "rewards/confidence_one_or_zero/mean": 0.32275390625, "rewards/confidence_one_or_zero/std": 0.4671346664428711, "rewards/format_reward/mean": 0.60634765625, "rewards/format_reward/std": 0.48829332590103147, "rewards/mean_confidence_reward/mean": 0.832246994972229, "rewards/mean_confidence_reward/std": 0.285083270072937, "step": 165, "step_time": 309.93234630029764 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02451171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4026.4, "completions/mean_length": 630.9279296875, "completions/mean_terminated_length": 544.0782592773437, "completions/min_length": 6.0, "completions/min_terminated_length": 6.0, "entropy": 0.2206975874491036, "epoch": 0.36324786324786323, "frac_reward_zero_std": 0.0, "grad_norm": 259.3530578613281, "learning_rate": 1.7379679144385028e-06, "loss": 0.0006, "num_tokens": 304594728.0, "reward": 2.326840305328369, "reward_std": 1.4189069271087646, "rewards/accuracy_reward/mean": 0.253125, "rewards/accuracy_reward/std": 0.4322037935256958, "rewards/brier_reward/mean": 0.3068468630313873, "rewards/brier_reward/std": 0.4160425066947937, "rewards/confidence_one_or_zero/mean": 0.34296875, "rewards/confidence_one_or_zero/std": 0.4745549440383911, "rewards/format_reward/mean": 0.59599609375, "rewards/format_reward/std": 0.4901327550411224, "rewards/mean_confidence_reward/mean": 0.8279036164283753, "rewards/mean_confidence_reward/std": 0.292535662651062, "step": 170, "step_time": 293.40511639249746 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0228515625, "completions/max_length": 4096.0, "completions/max_terminated_length": 3966.4, "completions/mean_length": 589.12998046875, "completions/mean_terminated_length": 507.1199157714844, "completions/min_length": 3.8, "completions/min_terminated_length": 3.8, "entropy": 0.2160580841358751, "epoch": 0.37393162393162394, "frac_reward_zero_std": 0.0, "grad_norm": 253.39134216308594, "learning_rate": 1.6042780748663103e-06, "loss": -0.0012, "num_tokens": 313389403.0, "reward": 2.2744149684906008, "reward_std": 1.4243911743164062, "rewards/accuracy_reward/mean": 0.24775390625, "rewards/accuracy_reward/std": 0.43125388622283933, "rewards/brier_reward/mean": 0.2988166928291321, "rewards/brier_reward/std": 0.41332738995552065, "rewards/confidence_one_or_zero/mean": 0.3177734375, "rewards/confidence_one_or_zero/std": 0.46545001268386843, "rewards/format_reward/mean": 0.58818359375, "rewards/format_reward/std": 0.49211637377738954, "rewards/mean_confidence_reward/mean": 0.8218873739242554, "rewards/mean_confidence_reward/std": 0.29787588119506836, "step": 175, "step_time": 308.38890937356047 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02158203125, "completions/max_length": 4096.0, "completions/max_terminated_length": 4028.2, "completions/mean_length": 595.2263671875, "completions/mean_terminated_length": 517.9749084472656, "completions/min_length": 3.8, "completions/min_terminated_length": 3.8, "entropy": 0.20982481236569583, "epoch": 0.38461538461538464, "frac_reward_zero_std": 0.0, "grad_norm": 160.34878540039062, "learning_rate": 1.4705882352941177e-06, "loss": 0.0042, "num_tokens": 322233513.0, "reward": 2.2769351482391356, "reward_std": 1.4204734802246093, "rewards/accuracy_reward/mean": 0.2404296875, "rewards/accuracy_reward/std": 0.42643400430679324, "rewards/brier_reward/mean": 0.29003825783729553, "rewards/brier_reward/std": 0.40969622135162354, "rewards/confidence_one_or_zero/mean": 0.33857421875, "rewards/confidence_one_or_zero/std": 0.47265806794166565, "rewards/format_reward/mean": 0.5818359375, "rewards/format_reward/std": 0.49277991652488706, "rewards/mean_confidence_reward/mean": 0.8260570168495178, "rewards/mean_confidence_reward/std": 0.2940966486930847, "step": 180, "step_time": 306.76037236634875 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0232421875, "completions/max_length": 4096.0, "completions/max_terminated_length": 4022.4, "completions/mean_length": 619.701953125, "completions/mean_terminated_length": 537.0830200195312, "completions/min_length": 2.8, "completions/min_terminated_length": 2.8, "entropy": 0.21132461737142877, "epoch": 0.3952991452991453, "frac_reward_zero_std": 0.0, "grad_norm": 249.3165283203125, "learning_rate": 1.3368983957219254e-06, "loss": 0.0012, "num_tokens": 331378973.0, "reward": 2.223415565490723, "reward_std": 1.4154028654098512, "rewards/accuracy_reward/mean": 0.23271484375, "rewards/accuracy_reward/std": 0.41807675957679746, "rewards/brier_reward/mean": 0.28476929664611816, "rewards/brier_reward/std": 0.40382109880447387, "rewards/confidence_one_or_zero/mean": 0.3205078125, "rewards/confidence_one_or_zero/std": 0.4667299807071686, "rewards/format_reward/mean": 0.57353515625, "rewards/format_reward/std": 0.49417153000831604, "rewards/mean_confidence_reward/mean": 0.8118884563446045, "rewards/mean_confidence_reward/std": 0.3092004060745239, "step": 185, "step_time": 351.6109183413908 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02451171875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3990.6, "completions/mean_length": 622.701953125, "completions/mean_terminated_length": 535.5741455078125, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.20809936788864433, "epoch": 0.405982905982906, "frac_reward_zero_std": 0.0, "grad_norm": 284.8410339355469, "learning_rate": 1.2032085561497326e-06, "loss": 0.0008, "num_tokens": 340494929.0, "reward": 2.241306734085083, "reward_std": 1.4264638900756836, "rewards/accuracy_reward/mean": 0.232421875, "rewards/accuracy_reward/std": 0.421414977312088, "rewards/brier_reward/mean": 0.28441300988197327, "rewards/brier_reward/std": 0.4062588274478912, "rewards/confidence_one_or_zero/mean": 0.3353515625, "rewards/confidence_one_or_zero/std": 0.47193573117256166, "rewards/format_reward/mean": 0.57431640625, "rewards/format_reward/std": 0.4940219223499298, "rewards/mean_confidence_reward/mean": 0.8148040056228638, "rewards/mean_confidence_reward/std": 0.3065529942512512, "step": 190, "step_time": 308.9337789218989 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.02431640625, "completions/max_length": 4096.0, "completions/max_terminated_length": 4030.0, "completions/mean_length": 620.21796875, "completions/mean_terminated_length": 533.6152893066406, "completions/min_length": 13.4, "completions/min_terminated_length": 13.4, "entropy": 0.2087443238706328, "epoch": 0.4166666666666667, "frac_reward_zero_std": 0.0, "grad_norm": 201.34759521484375, "learning_rate": 1.0695187165775401e-06, "loss": 0.0045, "num_tokens": 349613161.0, "reward": 2.273044395446777, "reward_std": 1.4159342765808105, "rewards/accuracy_reward/mean": 0.24111328125, "rewards/accuracy_reward/std": 0.4254138171672821, "rewards/brier_reward/mean": 0.291348472237587, "rewards/brier_reward/std": 0.40924651026725767, "rewards/confidence_one_or_zero/mean": 0.33984375, "rewards/confidence_one_or_zero/std": 0.47324607968330384, "rewards/format_reward/mean": 0.5798828125, "rewards/format_reward/std": 0.4933952629566193, "rewards/mean_confidence_reward/mean": 0.8208560585975647, "rewards/mean_confidence_reward/std": 0.30015876293182375, "step": 195, "step_time": 304.7307929251809 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0216796875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3961.8, "completions/mean_length": 607.18359375, "completions/mean_terminated_length": 529.8887329101562, "completions/min_length": 3.8, "completions/min_terminated_length": 3.8, "entropy": 0.20513830555137247, "epoch": 0.42735042735042733, "frac_reward_zero_std": 0.0, "grad_norm": 258.32965087890625, "learning_rate": 9.358288770053477e-07, "loss": 0.0044, "num_tokens": 358576929.0, "reward": 2.271790313720703, "reward_std": 1.4089332818984985, "rewards/accuracy_reward/mean": 0.2375, "rewards/accuracy_reward/std": 0.425096195936203, "rewards/brier_reward/mean": 0.28932103514671326, "rewards/brier_reward/std": 0.40852566957473757, "rewards/confidence_one_or_zero/mean": 0.33759765625, "rewards/confidence_one_or_zero/std": 0.47278434634208677, "rewards/format_reward/mean": 0.58271484375, "rewards/format_reward/std": 0.4926397502422333, "rewards/mean_confidence_reward/mean": 0.8246568441390991, "rewards/mean_confidence_reward/std": 0.294880747795105, "step": 200, "step_time": 310.373915496649 }, { "epoch": 0.42735042735042733, "eval_clip_ratio/high_max": 0.0, "eval_clip_ratio/high_mean": 0.0, "eval_clip_ratio/low_mean": 0.0, "eval_clip_ratio/low_min": 0.0, "eval_clip_ratio/region_mean": 0.0, "eval_completions/clipped_ratio": 0.015625, "eval_completions/max_length": 2836.8125, "eval_completions/max_terminated_length": 2001.96875, "eval_completions/mean_length": 576.6494140625, "eval_completions/mean_terminated_length": 520.7525768280029, "eval_completions/min_length": 123.6875, "eval_completions/min_terminated_length": 123.6875, "eval_entropy": 0.19752578996121883, "eval_frac_reward_zero_std": 1.0, "eval_loss": 0.0, "eval_num_tokens": 358576929.0, "eval_reward": 2.328748621046543, "eval_reward_std": 1.3927721418440342, "eval_rewards/accuracy_reward/mean": 0.2412109375, "eval_rewards/accuracy_reward/std": 0.42269768565893173, "eval_rewards/brier_reward/mean": 0.29818232264369726, "eval_rewards/brier_reward/std": 0.40743969194591045, "eval_rewards/confidence_one_or_zero/mean": 0.3623046875, "eval_rewards/confidence_one_or_zero/std": 0.48153325729072094, "eval_rewards/format_reward/mean": 0.5986328125, "eval_rewards/format_reward/std": 0.49181301984936, "eval_rewards/mean_confidence_reward/mean": 0.8284178636968136, "eval_rewards/mean_confidence_reward/std": 0.28921784972772, "eval_runtime": 1067.6255, "eval_samples_per_second": 0.937, "eval_steps_per_second": 0.03, "step": 200 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.01787109375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3980.4, "completions/mean_length": 573.26318359375, "completions/mean_terminated_length": 509.2129638671875, "completions/min_length": 4.4, "completions/min_terminated_length": 4.4, "entropy": 0.2126567040453665, "epoch": 0.43803418803418803, "frac_reward_zero_std": 0.0, "grad_norm": 218.33091735839844, "learning_rate": 8.021390374331551e-07, "loss": 0.0038, "num_tokens": 367194600.0, "reward": 2.3272546768188476, "reward_std": 1.3900749444961549, "rewards/accuracy_reward/mean": 0.2466796875, "rewards/accuracy_reward/std": 0.4290748596191406, "rewards/brier_reward/mean": 0.30053516030311583, "rewards/brier_reward/std": 0.41140108108520507, "rewards/confidence_one_or_zero/mean": 0.33564453125, "rewards/confidence_one_or_zero/std": 0.4720482587814331, "rewards/format_reward/mean": 0.605859375, "rewards/format_reward/std": 0.48814900517463683, "rewards/mean_confidence_reward/mean": 0.8385359048843384, "rewards/mean_confidence_reward/std": 0.27658396363258364, "step": 205, "step_time": 312.38765997856973 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0224609375, "completions/max_length": 4096.0, "completions/max_terminated_length": 4008.4, "completions/mean_length": 613.34365234375, "completions/mean_terminated_length": 533.35732421875, "completions/min_length": 15.4, "completions/min_terminated_length": 15.4, "entropy": 0.20988439373904838, "epoch": 0.44871794871794873, "frac_reward_zero_std": 0.0, "grad_norm": 255.08921813964844, "learning_rate": 6.684491978609627e-07, "loss": -0.0028, "num_tokens": 376242663.0, "reward": 2.258999395370483, "reward_std": 1.381415557861328, "rewards/accuracy_reward/mean": 0.2353515625, "rewards/accuracy_reward/std": 0.4236945390701294, "rewards/brier_reward/mean": 0.28892136216163633, "rewards/brier_reward/std": 0.4066075265407562, "rewards/confidence_one_or_zero/mean": 0.31591796875, "rewards/confidence_one_or_zero/std": 0.4645472884178162, "rewards/format_reward/mean": 0.5935546875, "rewards/format_reward/std": 0.49092593789100647, "rewards/mean_confidence_reward/mean": 0.8252537369728088, "rewards/mean_confidence_reward/std": 0.2916837990283966, "step": 210, "step_time": 308.5079409700178 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.023046875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3993.8, "completions/mean_length": 604.281640625, "completions/mean_terminated_length": 521.9120666503907, "completions/min_length": 7.4, "completions/min_terminated_length": 7.4, "entropy": 0.21650093857897446, "epoch": 0.4594017094017094, "frac_reward_zero_std": 0.0, "grad_norm": 203.42086791992188, "learning_rate": 5.347593582887701e-07, "loss": 0.0017, "num_tokens": 385164139.0, "reward": 2.2653226375579836, "reward_std": 1.4082147121429442, "rewards/accuracy_reward/mean": 0.2435546875, "rewards/accuracy_reward/std": 0.4288760840892792, "rewards/brier_reward/mean": 0.2964475333690643, "rewards/brier_reward/std": 0.41219221949577334, "rewards/confidence_one_or_zero/mean": 0.3189453125, "rewards/confidence_one_or_zero/std": 0.4657912731170654, "rewards/format_reward/mean": 0.58359375, "rewards/format_reward/std": 0.4930082082748413, "rewards/mean_confidence_reward/mean": 0.8227813720703125, "rewards/mean_confidence_reward/std": 0.2937847971916199, "step": 215, "step_time": 301.61203362010417 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.024609375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3957.0, "completions/mean_length": 609.91484375, "completions/mean_terminated_length": 522.0267639160156, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "entropy": 0.21155671409796922, "epoch": 0.4700854700854701, "frac_reward_zero_std": 0.0, "grad_norm": 188.8226318359375, "learning_rate": 4.0106951871657757e-07, "loss": 0.0027, "num_tokens": 394143171.0, "reward": 2.2572365283966063, "reward_std": 1.3983298063278198, "rewards/accuracy_reward/mean": 0.23544921875, "rewards/accuracy_reward/std": 0.42343607544898987, "rewards/brier_reward/mean": 0.2879025101661682, "rewards/brier_reward/std": 0.40668407678604124, "rewards/confidence_one_or_zero/mean": 0.32060546875, "rewards/confidence_one_or_zero/std": 0.46630818843841554, "rewards/format_reward/mean": 0.58720703125, "rewards/format_reward/std": 0.49222953915596007, "rewards/mean_confidence_reward/mean": 0.8260722994804383, "rewards/mean_confidence_reward/std": 0.29320241808891295, "step": 220, "step_time": 291.255253496184 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.021484375, "completions/max_length": 4096.0, "completions/max_terminated_length": 3945.6, "completions/mean_length": 605.4865234375, "completions/mean_terminated_length": 528.8855224609375, "completions/min_length": 9.6, "completions/min_terminated_length": 9.6, "entropy": 0.21072454319219106, "epoch": 0.4807692307692308, "frac_reward_zero_std": 0.0, "grad_norm": 212.77940368652344, "learning_rate": 2.6737967914438503e-07, "loss": -0.0007, "num_tokens": 403066585.0, "reward": 2.2713153839111326, "reward_std": 1.408125376701355, "rewards/accuracy_reward/mean": 0.24130859375, "rewards/accuracy_reward/std": 0.42738835215568544, "rewards/brier_reward/mean": 0.29448471665382386, "rewards/brier_reward/std": 0.409639573097229, "rewards/confidence_one_or_zero/mean": 0.3185546875, "rewards/confidence_one_or_zero/std": 0.4659416854381561, "rewards/format_reward/mean": 0.58828125, "rewards/format_reward/std": 0.4917002320289612, "rewards/mean_confidence_reward/mean": 0.8286861181259155, "rewards/mean_confidence_reward/std": 0.28555492758750917, "step": 225, "step_time": 313.09385580177184 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0232421875, "completions/max_length": 4096.0, "completions/max_terminated_length": 3981.0, "completions/mean_length": 611.399609375, "completions/mean_terminated_length": 528.4930725097656, "completions/min_length": 15.4, "completions/min_terminated_length": 15.4, "entropy": 0.20962072565453127, "epoch": 0.49145299145299143, "frac_reward_zero_std": 0.0, "grad_norm": 193.35745239257812, "learning_rate": 1.3368983957219251e-07, "loss": 0.0001, "num_tokens": 412054069.0, "reward": 2.319254446029663, "reward_std": 1.4233505725860596, "rewards/accuracy_reward/mean": 0.256640625, "rewards/accuracy_reward/std": 0.4365252435207367, "rewards/brier_reward/mean": 0.3055537760257721, "rewards/brier_reward/std": 0.4187450110912323, "rewards/confidence_one_or_zero/mean": 0.33564453125, "rewards/confidence_one_or_zero/std": 0.4720422148704529, "rewards/format_reward/mean": 0.5904296875, "rewards/format_reward/std": 0.4914150834083557, "rewards/mean_confidence_reward/mean": 0.8309858441352844, "rewards/mean_confidence_reward/std": 0.28812336921691895, "step": 230, "step_time": 296.3028953729139 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0230712890625, "completions/max_length": 4096.0, "completions/max_terminated_length": 3856.5, "completions/mean_length": 608.772216796875, "completions/mean_terminated_length": 526.4675445556641, "completions/min_length": 3.0, "completions/min_terminated_length": 3.0, "entropy": 0.21378142246976495, "epoch": 0.5, "frac_reward_zero_std": 0.0, "num_tokens": 419246379.0, "reward": 2.3010610342025757, "reward_std": 1.4046761989593506, "rewards/accuracy_reward/mean": 0.254638671875, "rewards/accuracy_reward/std": 0.4350617974996567, "rewards/brier_reward/mean": 0.3073633164167404, "rewards/brier_reward/std": 0.4163747727870941, "rewards/confidence_one_or_zero/mean": 0.3199462890625, "rewards/confidence_one_or_zero/std": 0.46621015667915344, "rewards/format_reward/mean": 0.5926513671875, "rewards/format_reward/std": 0.49101030081510544, "rewards/mean_confidence_reward/mean": 0.8264614343643188, "rewards/mean_confidence_reward/std": 0.2889925092458725, "step": 234, "step_time": 311.87753583449376, "total_flos": 0.0, "train_loss": 0.0014963890027859781, "train_runtime": 79990.8765, "train_samples_per_second": 0.188, "train_steps_per_second": 0.003 } ], "logging_steps": 5, "max_steps": 234, "num_input_tokens_seen": 419246379, "num_train_epochs": 1, "save_steps": 60, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 4, "trial_name": null, "trial_params": null }