diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3660 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.970666666666667,
+  "eval_steps": 10,
+  "global_step": 279,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "completion_length": 521.9583435058594,
+      "epoch": 0.010666666666666666,
+      "grad_norm": 0.1310514211654663,
+      "kl": 0.0,
+      "learning_rate": 1.0714285714285714e-07,
+      "loss": 0.0931,
+      "reward": 0.2083333358168602,
+      "reward_std": 0.30354244634509087,
+      "rewards/accuracy_reward": 0.2083333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 1
+    },
+    {
+      "completion_length": 489.4791793823242,
+      "epoch": 0.021333333333333333,
+      "grad_norm": 0.10328225791454315,
+      "kl": 0.0,
+      "learning_rate": 2.1428571428571428e-07,
+      "loss": -0.0373,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.20412414893507957,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 2
+    },
+    {
+      "completion_length": 544.6250228881836,
+      "epoch": 0.032,
+      "grad_norm": 0.11837511509656906,
+      "kl": 0.0002536177635192871,
+      "learning_rate": 3.214285714285714e-07,
+      "loss": -0.023,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.20412414520978928,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 3
+    },
+    {
+      "completion_length": 575.2500114440918,
+      "epoch": 0.042666666666666665,
+      "grad_norm": 0.08072590827941895,
+      "kl": 0.00017571449279785156,
+      "learning_rate": 4.2857142857142857e-07,
+      "loss": -0.0283,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.16661180555820465,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 4
+    },
+    {
+      "completion_length": 610.2083587646484,
+      "epoch": 0.05333333333333334,
+      "grad_norm": 0.08585350215435028,
+      "kl": 0.00016301870346069336,
+      "learning_rate": 5.357142857142858e-07,
+      "loss": 0.0194,
+      "reward": 0.22916666977107525,
+      "reward_std": 0.3170611374080181,
+      "rewards/accuracy_reward": 0.22916666977107525,
+      "rewards/format_reward": 0.0,
+      "step": 5
+    },
+    {
+      "completion_length": 558.0208511352539,
+      "epoch": 0.064,
+      "grad_norm": 0.07279166579246521,
+      "kl": 0.00014892220497131348,
+      "learning_rate": 6.428571428571428e-07,
+      "loss": -0.034,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.12909945845603943,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 6
+    },
+    {
+      "completion_length": 465.64584732055664,
+      "epoch": 0.07466666666666667,
+      "grad_norm": 0.09253880381584167,
+      "kl": 0.00017279386520385742,
+      "learning_rate": 7.5e-07,
+      "loss": -0.0474,
+      "reward": 0.1250000037252903,
+      "reward_std": 0.268673874437809,
+      "rewards/accuracy_reward": 0.1250000037252903,
+      "rewards/format_reward": 0.0,
+      "step": 7
+    },
+    {
+      "completion_length": 517.7083511352539,
+      "epoch": 0.08533333333333333,
+      "grad_norm": 0.07610676437616348,
+      "kl": 0.00014507770538330078,
+      "learning_rate": 8.571428571428571e-07,
+      "loss": 0.0472,
+      "reward": 0.10416666977107525,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.10416666977107525,
+      "rewards/format_reward": 0.0,
+      "step": 8
+    },
+    {
+      "completion_length": 458.43750762939453,
+      "epoch": 0.096,
+      "grad_norm": 0.08454978466033936,
+      "kl": 0.00015234947204589844,
+      "learning_rate": 9.642857142857145e-07,
+      "loss": 0.0193,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.16661179810762405,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 9
+    },
+    {
+      "completion_length": 548.1458549499512,
+      "epoch": 0.10666666666666667,
+      "grad_norm": 0.05857018381357193,
+      "kl": 0.00019532442092895508,
+      "learning_rate": 1.0714285714285716e-06,
+      "loss": 0.1103,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.16661179810762405,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 10
+    },
+    {
+      "completion_length": 445.7708511352539,
+      "epoch": 0.11733333333333333,
+      "grad_norm": 0.12454483658075333,
+      "kl": 0.00032699108123779297,
+      "learning_rate": 1.1785714285714285e-06,
+      "loss": -0.0483,
+      "reward": 0.06250000186264515,
+      "reward_std": 0.1530931107699871,
+      "rewards/accuracy_reward": 0.06250000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 11
+    },
+    {
+      "completion_length": 553.083345413208,
+      "epoch": 0.128,
+      "grad_norm": 0.07340668886899948,
+      "kl": 0.00017911195755004883,
+      "learning_rate": 1.2857142857142856e-06,
+      "loss": -0.0527,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.16661179810762405,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 12
+    },
+    {
+      "completion_length": 465.39584732055664,
+      "epoch": 0.13866666666666666,
+      "grad_norm": 0.16328425705432892,
+      "kl": 0.0002442598342895508,
+      "learning_rate": 1.392857142857143e-06,
+      "loss": 0.0427,
+      "reward": 0.1250000037252903,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.1250000037252903,
+      "rewards/format_reward": 0.0,
+      "step": 13
+    },
+    {
+      "completion_length": 396.0208435058594,
+      "epoch": 0.14933333333333335,
+      "grad_norm": 0.12855905294418335,
+      "kl": 0.00034868717193603516,
+      "learning_rate": 1.5e-06,
+      "loss": 0.032,
+      "reward": 0.18750000558793545,
+      "reward_std": 0.34674229845404625,
+      "rewards/accuracy_reward": 0.18750000558793545,
+      "rewards/format_reward": 0.0,
+      "step": 14
+    },
+    {
+      "completion_length": 544.8541793823242,
+      "epoch": 0.16,
+      "grad_norm": 0.07833456248044968,
+      "kl": 0.0003299713134765625,
+      "learning_rate": 1.6071428571428572e-06,
+      "loss": 0.0093,
+      "reward": 0.0833333358168602,
+      "reward_std": 0.16661179810762405,
+      "rewards/accuracy_reward": 0.0833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 15
+    },
+    {
+      "completion_length": 491.2291831970215,
+      "epoch": 0.17066666666666666,
+      "grad_norm": 0.0904361829161644,
+      "kl": 0.0005221366882324219,
+      "learning_rate": 1.7142857142857143e-06,
+      "loss": 0.0572,
+      "reward": 0.12500000186264515,
+      "reward_std": 0.22155842557549477,
+      "rewards/accuracy_reward": 0.12500000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 16
+    },
+    {
+      "completion_length": 474.6458435058594,
+      "epoch": 0.18133333333333335,
+      "grad_norm": 0.09762278199195862,
+      "kl": 0.0004999637603759766,
+      "learning_rate": 1.8214285714285714e-06,
+      "loss": -0.0009,
+      "reward": 0.12500000186264515,
+      "reward_std": 0.18404608592391014,
+      "rewards/accuracy_reward": 0.12500000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 17
+    },
+    {
+      "completion_length": 480.14584732055664,
+      "epoch": 0.192,
+      "grad_norm": 0.10621696710586548,
+      "kl": 0.0006742477416992188,
+      "learning_rate": 1.928571428571429e-06,
+      "loss": -0.0441,
+      "reward": 0.1458333358168602,
+      "reward_std": 0.2350771278142929,
+      "rewards/accuracy_reward": 0.1458333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 18
+    },
+    {
+      "completion_length": 480.04167556762695,
+      "epoch": 0.20266666666666666,
+      "grad_norm": 0.11984441429376602,
+      "kl": 0.0007910728454589844,
+      "learning_rate": 2.035714285714286e-06,
+      "loss": 0.0073,
+      "reward": 0.14583333767950535,
+      "reward_std": 0.31970490887761116,
+      "rewards/accuracy_reward": 0.14583333767950535,
+      "rewards/format_reward": 0.0,
+      "step": 19
+    },
+    {
+      "completion_length": 599.0625190734863,
+      "epoch": 0.21333333333333335,
+      "grad_norm": 0.1133371964097023,
+      "kl": 0.001157999038696289,
+      "learning_rate": 2.142857142857143e-06,
+      "loss": -0.0503,
+      "reward": 0.10416666977107525,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.10416666977107525,
+      "rewards/format_reward": 0.0,
+      "step": 20
+    },
+    {
+      "completion_length": 429.25001335144043,
+      "epoch": 0.224,
+      "grad_norm": 0.1195325180888176,
+      "kl": 0.0021119117736816406,
+      "learning_rate": 2.25e-06,
+      "loss": 0.0239,
+      "reward": 0.14583333767950535,
+      "reward_std": 0.28219256550073624,
+      "rewards/accuracy_reward": 0.10416666977107525,
+      "rewards/format_reward": 0.0416666679084301,
+      "step": 21
+    },
+    {
+      "completion_length": 540.0416793823242,
+      "epoch": 0.23466666666666666,
+      "grad_norm": 0.0632760301232338,
+      "kl": 0.0018351078033447266,
+      "learning_rate": 2.357142857142857e-06,
+      "loss": -0.0332,
+      "reward": 0.12500000186264515,
+      "reward_std": 0.18404608592391014,
+      "rewards/accuracy_reward": 0.12500000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 22
+    },
+    {
+      "completion_length": 515.2708511352539,
+      "epoch": 0.24533333333333332,
+      "grad_norm": 0.13157933950424194,
+      "kl": 0.002986907958984375,
+      "learning_rate": 2.4642857142857143e-06,
+      "loss": 0.0165,
+      "reward": 0.1875000037252903,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.1875000037252903,
+      "rewards/format_reward": 0.0,
+      "step": 23
+    },
+    {
+      "completion_length": 522.7916831970215,
+      "epoch": 0.256,
+      "grad_norm": 0.1456395834684372,
+      "kl": 0.0023260116577148438,
+      "learning_rate": 2.571428571428571e-06,
+      "loss": -0.0205,
+      "reward": 0.27083333767950535,
+      "reward_std": 0.41912321373820305,
+      "rewards/accuracy_reward": 0.2500000037252903,
+      "rewards/format_reward": 0.02083333395421505,
+      "step": 24
+    },
+    {
+      "completion_length": 490.93751525878906,
+      "epoch": 0.26666666666666666,
+      "grad_norm": 0.06600736081600189,
+      "kl": 0.00377655029296875,
+      "learning_rate": 2.678571428571429e-06,
+      "loss": 0.0319,
+      "reward": 0.16666666977107525,
+      "reward_std": 0.18404608592391014,
+      "rewards/accuracy_reward": 0.16666666977107525,
+      "rewards/format_reward": 0.0,
+      "step": 25
+    },
+    {
+      "completion_length": 528.3333511352539,
+      "epoch": 0.2773333333333333,
+      "grad_norm": 0.06686762720346451,
+      "kl": 0.0043849945068359375,
+      "learning_rate": 2.785714285714286e-06,
+      "loss": 0.011,
+      "reward": 0.12500000186264515,
+      "reward_std": 0.18404608592391014,
+      "rewards/accuracy_reward": 0.1041666679084301,
+      "rewards/format_reward": 0.02083333395421505,
+      "step": 26
+    },
+    {
+      "completion_length": 536.4166831970215,
+      "epoch": 0.288,
+      "grad_norm": 0.08273553848266602,
+      "kl": 0.005977630615234375,
+      "learning_rate": 2.892857142857143e-06,
+      "loss": -0.0035,
+      "reward": 0.12500000186264515,
+      "reward_std": 0.18404608592391014,
+      "rewards/accuracy_reward": 0.12500000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 27
+    },
+    {
+      "completion_length": 517.6875228881836,
+      "epoch": 0.2986666666666667,
+      "grad_norm": 0.11967507004737854,
+      "kl": 0.005385875701904297,
+      "learning_rate": 3e-06,
+      "loss": 0.0183,
+      "reward": 0.2500000037252903,
+      "reward_std": 0.3680921718478203,
+      "rewards/accuracy_reward": 0.2500000037252903,
+      "rewards/format_reward": 0.0,
+      "step": 28
+    },
+    {
+      "completion_length": 402.4583435058594,
+      "epoch": 0.30933333333333335,
+      "grad_norm": 0.1181207150220871,
+      "kl": 0.011768341064453125,
+      "learning_rate": 2.999882508108334e-06,
+      "loss": -0.0503,
+      "reward": 0.1875000037252903,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.1875000037252903,
+      "rewards/format_reward": 0.0,
+      "step": 29
+    },
+    {
+      "completion_length": 570.0833435058594,
+      "epoch": 0.32,
+      "grad_norm": 0.08682581782341003,
+      "kl": 0.008464813232421875,
+      "learning_rate": 2.9995300508391286e-06,
+      "loss": -0.0233,
+      "reward": 0.16666666977107525,
+      "reward_std": 0.2861081622540951,
+      "rewards/accuracy_reward": 0.16666666977107525,
+      "rewards/format_reward": 0.0,
+      "step": 30
+    },
+    {
+      "completion_length": 553.0208415985107,
+      "epoch": 0.33066666666666666,
+      "grad_norm": 0.10582250356674194,
+      "kl": 0.013153076171875,
+      "learning_rate": 2.998942683406879e-06,
+      "loss": -0.0808,
+      "reward": 0.25000000558793545,
+      "reward_std": 0.3506578914821148,
+      "rewards/accuracy_reward": 0.25000000558793545,
+      "rewards/format_reward": 0.0,
+      "step": 31
+    },
+    {
+      "completion_length": 461.916690826416,
+      "epoch": 0.3413333333333333,
+      "grad_norm": 0.09418044239282608,
+      "kl": 0.014652252197265625,
+      "learning_rate": 2.998120497826133e-06,
+      "loss": -0.0571,
+      "reward": 0.2083333395421505,
+      "reward_std": 0.25819891691207886,
+      "rewards/accuracy_reward": 0.2083333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 32
+    },
+    {
+      "completion_length": 518.8958435058594,
+      "epoch": 0.352,
+      "grad_norm": 0.10594867169857025,
+      "kl": 0.0132598876953125,
+      "learning_rate": 2.9970636228970757e-06,
+      "loss": 0.1187,
+      "reward": 0.5208333544433117,
+      "reward_std": 0.3842546381056309,
+      "rewards/accuracy_reward": 0.5208333544433117,
+      "rewards/format_reward": 0.0,
+      "step": 33
+    },
+    {
+      "completion_length": 418.77084159851074,
+      "epoch": 0.3626666666666667,
+      "grad_norm": 0.13837799429893494,
+      "kl": 0.02474212646484375,
+      "learning_rate": 2.995772224185354e-06,
+      "loss": -0.0023,
+      "reward": 0.3958333432674408,
+      "reward_std": 0.3816108666360378,
+      "rewards/accuracy_reward": 0.3958333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 34
+    },
+    {
+      "completion_length": 506.43751525878906,
+      "epoch": 0.37333333333333335,
+      "grad_norm": 0.10607906430959702,
+      "kl": 0.059417724609375,
+      "learning_rate": 2.9942465039961373e-06,
+      "loss": 0.042,
+      "reward": 0.25000000558793545,
+      "reward_std": 0.31314554437994957,
+      "rewards/accuracy_reward": 0.25000000558793545,
+      "rewards/format_reward": 0.0,
+      "step": 35
+    },
+    {
+      "completion_length": 475.06251525878906,
+      "epoch": 0.384,
+      "grad_norm": 0.11719807982444763,
+      "kl": 0.020172119140625,
+      "learning_rate": 2.992486701342427e-06,
+      "loss": -0.008,
+      "reward": 0.5625000111758709,
+      "reward_std": 0.41912321001291275,
+      "rewards/accuracy_reward": 0.5625000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 36
+    },
+    {
+      "completion_length": 460.35417556762695,
+      "epoch": 0.39466666666666667,
+      "grad_norm": 0.11237220466136932,
+      "kl": 0.0255279541015625,
+      "learning_rate": 2.9904930919076145e-06,
+      "loss": 0.0259,
+      "reward": 0.5000000111758709,
+      "reward_std": 0.4152076169848442,
+      "rewards/accuracy_reward": 0.5000000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 37
+    },
+    {
+      "completion_length": 488.08334732055664,
+      "epoch": 0.4053333333333333,
+      "grad_norm": 0.09729558229446411,
+      "kl": 0.02240753173828125,
+      "learning_rate": 2.9882659880022905e-06,
+      "loss": -0.0223,
+      "reward": 0.4583333469927311,
+      "reward_std": 0.442245002835989,
+      "rewards/accuracy_reward": 0.4583333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 38
+    },
+    {
+      "completion_length": 496.4166851043701,
+      "epoch": 0.416,
+      "grad_norm": 0.09237450361251831,
+      "kl": 0.02236175537109375,
+      "learning_rate": 2.985805738515323e-06,
+      "loss": 0.0356,
+      "reward": 0.7291666772216558,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.7291666772216558,
+      "rewards/format_reward": 0.0,
+      "step": 39
+    },
+    {
+      "completion_length": 480.3541793823242,
+      "epoch": 0.4266666666666667,
+      "grad_norm": 0.10983026027679443,
+      "kl": 0.03179931640625,
+      "learning_rate": 2.9831127288592004e-06,
+      "loss": -0.0175,
+      "reward": 0.5000000093132257,
+      "reward_std": 0.5050228200852871,
+      "rewards/accuracy_reward": 0.5000000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 40
+    },
+    {
+      "completion_length": 474.93751335144043,
+      "epoch": 0.43733333333333335,
+      "grad_norm": 0.07909613102674484,
+      "kl": 0.05350494384765625,
+      "learning_rate": 2.9801873809096543e-06,
+      "loss": 0.0335,
+      "reward": 0.6875000074505806,
+      "reward_std": 0.21764283999800682,
+      "rewards/accuracy_reward": 0.6875000074505806,
+      "rewards/format_reward": 0.0,
+      "step": 41
+    },
+    {
+      "completion_length": 416.0208396911621,
+      "epoch": 0.448,
+      "grad_norm": 0.10907845944166183,
+      "kl": 0.03131103515625,
+      "learning_rate": 2.97703015293957e-06,
+      "loss": 0.0711,
+      "reward": 0.7083333544433117,
+      "reward_std": 0.4152076207101345,
+      "rewards/accuracy_reward": 0.7083333544433117,
+      "rewards/format_reward": 0.0,
+      "step": 42
+    },
+    {
+      "completion_length": 440.1458435058594,
+      "epoch": 0.45866666666666667,
+      "grad_norm": 0.6414058804512024,
+      "kl": 0.1112213134765625,
+      "learning_rate": 2.9736415395471975e-06,
+      "loss": 0.0949,
+      "reward": 0.6041666753590107,
+      "reward_std": 0.3170611374080181,
+      "rewards/accuracy_reward": 0.6041666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 43
+    },
+    {
+      "completion_length": 381.37500953674316,
+      "epoch": 0.4693333333333333,
+      "grad_norm": 0.8017810583114624,
+      "kl": 0.08701324462890625,
+      "learning_rate": 2.970022071578666e-06,
+      "loss": -0.0049,
+      "reward": 0.8750000149011612,
+      "reward_std": 0.22155842930078506,
+      "rewards/accuracy_reward": 0.8750000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 44
+    },
+    {
+      "completion_length": 460.2916793823242,
+      "epoch": 0.48,
+      "grad_norm": 0.11192864924669266,
+      "kl": 0.0340576171875,
+      "learning_rate": 2.9661723160448267e-06,
+      "loss": -0.015,
+      "reward": 0.645833345130086,
+      "reward_std": 0.35457348078489304,
+      "rewards/accuracy_reward": 0.645833345130086,
+      "rewards/format_reward": 0.0,
+      "step": 45
+    },
+    {
+      "completion_length": 488.85417556762695,
+      "epoch": 0.49066666666666664,
+      "grad_norm": 0.08567113429307938,
+      "kl": 0.020843505859375,
+      "learning_rate": 2.962092876032427e-06,
+      "loss": 0.0172,
+      "reward": 0.729166679084301,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.729166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 46
+    },
+    {
+      "completion_length": 444.39584732055664,
+      "epoch": 0.5013333333333333,
+      "grad_norm": 0.045451682060956955,
+      "kl": 0.025360107421875,
+      "learning_rate": 2.957784390609632e-06,
+      "loss": 0.0185,
+      "reward": 0.6666666716337204,
+      "reward_std": 0.11949635669589043,
+      "rewards/accuracy_reward": 0.6666666716337204,
+      "rewards/format_reward": 0.0,
+      "step": 47
+    },
+    {
+      "completion_length": 493.2500114440918,
+      "epoch": 0.512,
+      "grad_norm": 0.12778422236442566,
+      "kl": 0.024078369140625,
+      "learning_rate": 2.9532475347259113e-06,
+      "loss": -0.0127,
+      "reward": 0.6458333544433117,
+      "reward_std": 0.49327604100108147,
+      "rewards/accuracy_reward": 0.6458333544433117,
+      "rewards/format_reward": 0.0,
+      "step": 48
+    },
+    {
+      "completion_length": 441.20834732055664,
+      "epoch": 0.5226666666666666,
+      "grad_norm": 0.09534750878810883,
+      "kl": 0.034832000732421875,
+      "learning_rate": 2.9484830191063057e-06,
+      "loss": 0.0846,
+      "reward": 0.6458333507180214,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.6458333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 49
+    },
+    {
+      "completion_length": 534.2708473205566,
+      "epoch": 0.5333333333333333,
+      "grad_norm": 0.13379552960395813,
+      "kl": 0.06119537353515625,
+      "learning_rate": 2.9434915901400857e-06,
+      "loss": -0.0142,
+      "reward": 0.5000000093132257,
+      "reward_std": 0.43264190107584,
+      "rewards/accuracy_reward": 0.5000000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 50
+    },
+    {
+      "completion_length": 424.2916793823242,
+      "epoch": 0.544,
+      "grad_norm": 0.14644326269626617,
+      "kl": 0.0355224609375,
+      "learning_rate": 2.938274029763826e-06,
+      "loss": 0.0701,
+      "reward": 0.5416666846722364,
+      "reward_std": 0.4152076169848442,
+      "rewards/accuracy_reward": 0.5416666846722364,
+      "rewards/format_reward": 0.0,
+      "step": 51
+    },
+    {
+      "completion_length": 483.93751525878906,
+      "epoch": 0.5546666666666666,
+      "grad_norm": 0.0993141159415245,
+      "kl": 0.02002716064453125,
+      "learning_rate": 2.9328311553389116e-06,
+      "loss": 0.0488,
+      "reward": 0.833333358168602,
+      "reward_std": 0.3332236036658287,
+      "rewards/accuracy_reward": 0.833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 52
+    },
+    {
+      "completion_length": 534.3333435058594,
+      "epoch": 0.5653333333333334,
+      "grad_norm": 0.07569706439971924,
+      "kl": 0.056056976318359375,
+      "learning_rate": 2.927163819523492e-06,
+      "loss": -0.0105,
+      "reward": 0.4166666716337204,
+      "reward_std": 0.19364918768405914,
+      "rewards/accuracy_reward": 0.4166666716337204,
+      "rewards/format_reward": 0.0,
+      "step": 53
+    },
+    {
+      "completion_length": 558.9375152587891,
+      "epoch": 0.576,
+      "grad_norm": 0.10566005110740662,
+      "kl": 0.06243133544921875,
+      "learning_rate": 2.9212729101389084e-06,
+      "loss": 0.1233,
+      "reward": 0.666666679084301,
+      "reward_std": 0.30354244261980057,
+      "rewards/accuracy_reward": 0.666666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 54
+    },
+    {
+      "completion_length": 735.3750228881836,
+      "epoch": 0.5866666666666667,
+      "grad_norm": 0.09483093023300171,
+      "kl": 0.033416748046875,
+      "learning_rate": 2.9151593500306103e-06,
+      "loss": 0.0045,
+      "reward": 0.31250000931322575,
+      "reward_std": 0.33713919296860695,
+      "rewards/accuracy_reward": 0.31250000931322575,
+      "rewards/format_reward": 0.0,
+      "step": 55
+    },
+    {
+      "completion_length": 524.4166831970215,
+      "epoch": 0.5973333333333334,
+      "grad_norm": 0.13574951887130737,
+      "kl": 0.0305328369140625,
+      "learning_rate": 2.9088240969235863e-06,
+      "loss": 0.0684,
+      "reward": 0.708333358168602,
+      "reward_std": 0.4797573499381542,
+      "rewards/accuracy_reward": 0.708333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 56
+    },
+    {
+      "completion_length": 500.56251525878906,
+      "epoch": 0.608,
+      "grad_norm": 0.1411762237548828,
+      "kl": 0.02851104736328125,
+      "learning_rate": 2.902268143272333e-06,
+      "loss": 0.076,
+      "reward": 0.7708333488553762,
+      "reward_std": 0.28219256922602654,
+      "rewards/accuracy_reward": 0.7708333488553762,
+      "rewards/format_reward": 0.0,
+      "step": 57
+    },
+    {
+      "completion_length": 541.3333473205566,
+      "epoch": 0.6186666666666667,
+      "grad_norm": 0.09211263805627823,
+      "kl": 0.02207183837890625,
+      "learning_rate": 2.895492516105378e-06,
+      "loss": 0.105,
+      "reward": 0.562500013038516,
+      "reward_std": 0.42872630804777145,
+      "rewards/accuracy_reward": 0.562500013038516,
+      "rewards/format_reward": 0.0,
+      "step": 58
+    },
+    {
+      "completion_length": 708.1458511352539,
+      "epoch": 0.6293333333333333,
+      "grad_norm": 0.08216460794210434,
+      "kl": 0.0366058349609375,
+      "learning_rate": 2.888498276864393e-06,
+      "loss": 0.1429,
+      "reward": 0.2916666753590107,
+      "reward_std": 0.2861081585288048,
+      "rewards/accuracy_reward": 0.2916666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 59
+    },
+    {
+      "completion_length": 629.0208511352539,
+      "epoch": 0.64,
+      "grad_norm": 0.10210894048213959,
+      "kl": 0.0289154052734375,
+      "learning_rate": 2.8812865212379087e-06,
+      "loss": 0.0857,
+      "reward": 0.5416666828095913,
+      "reward_std": 0.4326419048011303,
+      "rewards/accuracy_reward": 0.5416666828095913,
+      "rewards/format_reward": 0.0,
+      "step": 60
+    },
+    {
+      "completion_length": 588.7291831970215,
+      "epoch": 0.6506666666666666,
+      "grad_norm": 0.1291423887014389,
+      "kl": 0.0692291259765625,
+      "learning_rate": 2.8738583789896743e-06,
+      "loss": 0.0292,
+      "reward": 0.6458333488553762,
+      "reward_std": 0.42872631177306175,
+      "rewards/accuracy_reward": 0.6458333488553762,
+      "rewards/format_reward": 0.0,
+      "step": 61
+    },
+    {
+      "completion_length": 643.0833435058594,
+      "epoch": 0.6613333333333333,
+      "grad_norm": 0.08674075454473495,
+      "kl": 0.02400970458984375,
+      "learning_rate": 2.866215013781668e-06,
+      "loss": 0.069,
+      "reward": 0.5208333488553762,
+      "reward_std": 0.49327604472637177,
+      "rewards/accuracy_reward": 0.5208333488553762,
+      "rewards/format_reward": 0.0,
+      "step": 62
+    },
+    {
+      "completion_length": 517.1666793823242,
+      "epoch": 0.672,
+      "grad_norm": 0.12224513292312622,
+      "kl": 0.02893829345703125,
+      "learning_rate": 2.858357622991806e-06,
+      "loss": 0.0604,
+      "reward": 0.7916666865348816,
+      "reward_std": 0.3410547859966755,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 63
+    },
+    {
+      "completion_length": 517.8750152587891,
+      "epoch": 0.6826666666666666,
+      "grad_norm": 0.131320983171463,
+      "kl": 0.04400634765625,
+      "learning_rate": 2.8502874375263646e-06,
+      "loss": 0.0173,
+      "reward": 0.5625000149011612,
+      "reward_std": 0.40168891847133636,
+      "rewards/accuracy_reward": 0.5625000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 64
+    },
+    {
+      "completion_length": 494.06251525878906,
+      "epoch": 0.6933333333333334,
+      "grad_norm": 0.16409559547901154,
+      "kl": 0.04022216796875,
+      "learning_rate": 2.8420057216271533e-06,
+      "loss": 0.1869,
+      "reward": 0.6458333544433117,
+      "reward_std": 0.48367293551564217,
+      "rewards/accuracy_reward": 0.6458333544433117,
+      "rewards/format_reward": 0.0,
+      "step": 65
+    },
+    {
+      "completion_length": 545.6041870117188,
+      "epoch": 0.704,
+      "grad_norm": 0.12068071216344833,
+      "kl": 0.0428619384765625,
+      "learning_rate": 2.8335137726734607e-06,
+      "loss": 0.1349,
+      "reward": 0.7083333563059568,
+      "reward_std": 0.3881702348589897,
+      "rewards/accuracy_reward": 0.7083333563059568,
+      "rewards/format_reward": 0.0,
+      "step": 66
+    },
+    {
+      "completion_length": 523.1875133514404,
+      "epoch": 0.7146666666666667,
+      "grad_norm": 0.15233297646045685,
+      "kl": 0.085906982421875,
+      "learning_rate": 2.8248129209788164e-06,
+      "loss": 0.0918,
+      "reward": 0.5625000204890966,
+      "reward_std": 0.4758417569100857,
+      "rewards/accuracy_reward": 0.5625000204890966,
+      "rewards/format_reward": 0.0,
+      "step": 67
+    },
+    {
+      "completion_length": 547.4375190734863,
+      "epoch": 0.7253333333333334,
+      "grad_norm": 0.05940861254930496,
+      "kl": 0.0589141845703125,
+      "learning_rate": 2.8159045295825857e-06,
+      "loss": 0.0431,
+      "reward": 0.5625000074505806,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.5625000074505806,
+      "rewards/format_reward": 0.0,
+      "step": 68
+    },
+    {
+      "completion_length": 531.6458473205566,
+      "epoch": 0.736,
+      "grad_norm": 0.15939682722091675,
+      "kl": 0.063690185546875,
+      "learning_rate": 2.806789994036445e-06,
+      "loss": 0.1327,
+      "reward": 0.6666666865348816,
+      "reward_std": 0.49719163402915,
+      "rewards/accuracy_reward": 0.6666666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 69
+    },
+    {
+      "completion_length": 585.1875152587891,
+      "epoch": 0.7466666666666667,
+      "grad_norm": 0.11161062866449356,
+      "kl": 0.07476806640625,
+      "learning_rate": 2.7974707421857584e-06,
+      "loss": 0.0339,
+      "reward": 0.5000000111758709,
+      "reward_std": 0.31314554437994957,
+      "rewards/accuracy_reward": 0.5000000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 70
+    },
+    {
+      "completion_length": 608.8333549499512,
+      "epoch": 0.7573333333333333,
+      "grad_norm": 0.17030133306980133,
+      "kl": 0.106109619140625,
+      "learning_rate": 2.7879482339458973e-06,
+      "loss": 0.1139,
+      "reward": 0.4791666865348816,
+      "reward_std": 0.34674229472875595,
+      "rewards/accuracy_reward": 0.4791666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 71
+    },
+    {
+      "completion_length": 505.1875114440918,
+      "epoch": 0.768,
+      "grad_norm": 0.10563177615404129,
+      "kl": 0.07091522216796875,
+      "learning_rate": 2.778223961073537e-06,
+      "loss": 0.0784,
+      "reward": 0.5625000186264515,
+      "reward_std": 0.31970490515232086,
+      "rewards/accuracy_reward": 0.5625000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 72
+    },
+    {
+      "completion_length": 511.8750114440918,
+      "epoch": 0.7786666666666666,
+      "grad_norm": 0.25055283308029175,
+      "kl": 0.1245269775390625,
+      "learning_rate": 2.7682994469329646e-06,
+      "loss": 0.1622,
+      "reward": 0.6458333563059568,
+      "reward_std": 0.4662386514246464,
+      "rewards/accuracy_reward": 0.6458333563059568,
+      "rewards/format_reward": 0.0,
+      "step": 73
+    },
+    {
+      "completion_length": 529.8750190734863,
+      "epoch": 0.7893333333333333,
+      "grad_norm": 0.1791120171546936,
+      "kl": 0.0995941162109375,
+      "learning_rate": 2.7581762462574334e-06,
+      "loss": 0.1168,
+      "reward": 0.6041666753590107,
+      "reward_std": 0.3266642391681671,
+      "rewards/accuracy_reward": 0.6041666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 74
+    },
+    {
+      "completion_length": 647.3333511352539,
+      "epoch": 0.8,
+      "grad_norm": 0.22069552540779114,
+      "kl": 0.11065673828125,
+      "learning_rate": 2.747855944905606e-06,
+      "loss": 0.1082,
+      "reward": 0.5000000093132257,
+      "reward_std": 0.4326419048011303,
+      "rewards/accuracy_reward": 0.5000000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 75
+    },
+    {
+      "completion_length": 607.041690826416,
+      "epoch": 0.8106666666666666,
+      "grad_norm": 0.21902619302272797,
+      "kl": 0.2556610107421875,
+      "learning_rate": 2.73734015961312e-06,
+      "loss": 0.196,
+      "reward": 0.6041666809469461,
+      "reward_std": 0.42872630804777145,
+      "rewards/accuracy_reward": 0.6041666809469461,
+      "rewards/format_reward": 0.0,
+      "step": 76
+    },
+    {
+      "completion_length": 516.2708511352539,
+      "epoch": 0.8213333333333334,
+      "grad_norm": 0.3198850750923157,
+      "kl": 0.22869873046875,
+      "learning_rate": 2.7266305377393183e-06,
+      "loss": 0.168,
+      "reward": 0.6250000204890966,
+      "reward_std": 0.4701542481780052,
+      "rewards/accuracy_reward": 0.6250000204890966,
+      "rewards/format_reward": 0.0,
+      "step": 77
+    },
+    {
+      "completion_length": 681.2708511352539,
+      "epoch": 0.832,
+      "grad_norm": 0.5761048197746277,
+      "kl": 0.41461181640625,
+      "learning_rate": 2.7157287570091765e-06,
+      "loss": 0.1671,
+      "reward": 0.5833333469927311,
+      "reward_std": 0.5050228163599968,
+      "rewards/accuracy_reward": 0.5833333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 78
+    },
+    {
+      "completion_length": 615.8541793823242,
+      "epoch": 0.8426666666666667,
+      "grad_norm": 0.2917896509170532,
+      "kl": 0.3033447265625,
+      "learning_rate": 2.7046365252504835e-06,
+      "loss": 0.0716,
+      "reward": 0.7083333507180214,
+      "reward_std": 0.2957112602889538,
+      "rewards/accuracy_reward": 0.7083333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 79
+    },
+    {
+      "completion_length": 645.1458511352539,
+      "epoch": 0.8533333333333334,
+      "grad_norm": 6.9831318855285645,
+      "kl": 1.189453125,
+      "learning_rate": 2.693355580126295e-06,
+      "loss": 0.2756,
+      "reward": 0.4791666753590107,
+      "reward_std": 0.41912321373820305,
+      "rewards/accuracy_reward": 0.4791666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 80
+    },
+    {
+      "completion_length": 621.3541793823242,
+      "epoch": 0.864,
+      "grad_norm": 0.24997614324092865,
+      "kl": 0.3677978515625,
+      "learning_rate": 2.6818876888627203e-06,
+      "loss": 0.1308,
+      "reward": 0.5000000111758709,
+      "reward_std": 0.30354244261980057,
+      "rewards/accuracy_reward": 0.5000000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 81
+    },
+    {
+      "completion_length": 585.8958549499512,
+      "epoch": 0.8746666666666667,
+      "grad_norm": 0.35314127802848816,
+      "kl": 0.3326416015625,
+      "learning_rate": 2.6702346479720777e-06,
+      "loss": 0.1922,
+      "reward": 0.4583333469927311,
+      "reward_std": 0.4152076132595539,
+      "rewards/accuracy_reward": 0.4583333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 82
+    },
+    {
+      "completion_length": 589.2291793823242,
+      "epoch": 0.8853333333333333,
+      "grad_norm": 0.6790549755096436,
+      "kl": 0.385284423828125,
+      "learning_rate": 2.6583982829714576e-06,
+      "loss": 0.2359,
+      "reward": 0.5833333488553762,
+      "reward_std": 0.3506578877568245,
+      "rewards/accuracy_reward": 0.5833333488553762,
+      "rewards/format_reward": 0.0,
+      "step": 83
+    },
+    {
+      "completion_length": 530.7916870117188,
+      "epoch": 0.896,
+      "grad_norm": 0.22984516620635986,
+      "kl": 0.20062255859375,
+      "learning_rate": 2.6463804480967447e-06,
+      "loss": 0.1143,
+      "reward": 0.708333345130086,
+      "reward_std": 0.2957112602889538,
+      "rewards/accuracy_reward": 0.708333345130086,
+      "rewards/format_reward": 0.0,
+      "step": 84
+    },
+    {
+      "completion_length": 501.1875190734863,
+      "epoch": 0.9066666666666666,
+      "grad_norm": 0.10622619092464447,
+      "kl": 0.05927276611328125,
+      "learning_rate": 2.6341830260121445e-06,
+      "loss": 0.0508,
+      "reward": 0.5208333469927311,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.5208333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 85
+    },
+    {
+      "completion_length": 560.9166793823242,
+      "epoch": 0.9173333333333333,
+      "grad_norm": 0.8767772912979126,
+      "kl": 0.2178192138671875,
+      "learning_rate": 2.6218079275152485e-06,
+      "loss": 0.1668,
+      "reward": 0.5208333432674408,
+      "reward_std": 0.3816108703613281,
+      "rewards/accuracy_reward": 0.5208333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 86
+    },
+    {
+      "completion_length": 587.3750305175781,
+      "epoch": 0.928,
+      "grad_norm": 0.28781571984291077,
+      "kl": 0.1433258056640625,
+      "learning_rate": 2.609257091237699e-06,
+      "loss": 0.0095,
+      "reward": 0.6041666809469461,
+      "reward_std": 0.41912321001291275,
+      "rewards/accuracy_reward": 0.6041666809469461,
+      "rewards/format_reward": 0.0,
+      "step": 87
+    },
+    {
+      "completion_length": 468.1666793823242,
+      "epoch": 0.9386666666666666,
+      "grad_norm": 0.7696180939674377,
+      "kl": 0.12293243408203125,
+      "learning_rate": 2.5965324833414926e-06,
+      "loss": 0.0333,
+      "reward": 0.562500013038516,
+      "reward_std": 0.35457348451018333,
+      "rewards/accuracy_reward": 0.562500013038516,
+      "rewards/format_reward": 0.0,
+      "step": 88
+    },
+    {
+      "completion_length": 603.3750190734863,
+      "epoch": 0.9493333333333334,
+      "grad_norm": 0.32361355423927307,
+      "kl": 0.2462005615234375,
+      "learning_rate": 2.5836360972109642e-06,
+      "loss": 0.0375,
+      "reward": 0.45833334513008595,
+      "reward_std": 0.3332235962152481,
+      "rewards/accuracy_reward": 0.45833334513008595,
+      "rewards/format_reward": 0.0,
+      "step": 89
+    },
+    {
+      "completion_length": 446.7916793823242,
+      "epoch": 0.96,
+      "grad_norm": 0.12218780070543289,
+      "kl": 0.180206298828125,
+      "learning_rate": 2.5705699531405174e-06,
+      "loss": 0.0221,
+      "reward": 0.7500000111758709,
+      "reward_std": 0.30354244261980057,
+      "rewards/accuracy_reward": 0.7500000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 90
+    },
+    {
+      "completion_length": 499.9791831970215,
+      "epoch": 0.9706666666666667,
+      "grad_norm": 0.25978320837020874,
+      "kl": 0.4139404296875,
+      "learning_rate": 2.55733609801813e-06,
+      "loss": -0.011,
+      "reward": 0.47916667349636555,
+      "reward_std": 0.2446802221238613,
+      "rewards/accuracy_reward": 0.47916667349636555,
+      "rewards/format_reward": 0.0,
+      "step": 91
+    },
+    {
+      "completion_length": 474.83334732055664,
+      "epoch": 0.9813333333333333,
+      "grad_norm": 0.34271982312202454,
+      "kl": 0.241973876953125,
+      "learning_rate": 2.5439366050046985e-06,
+      "loss": 0.0207,
+      "reward": 0.6458333414047956,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.6458333414047956,
+      "rewards/format_reward": 0.0,
+      "step": 92
+    },
+    {
+      "completion_length": 528.0208435058594,
+      "epoch": 0.992,
+      "grad_norm": 0.4584012031555176,
+      "kl": 0.1243743896484375,
+      "learning_rate": 2.5303735732092655e-06,
+      "loss": 0.0295,
+      "reward": 0.6041666753590107,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.6041666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 93
+    },
+    {
+      "completion_length": 578.3055699666342,
+      "epoch": 1.0,
+      "grad_norm": 0.4584012031555176,
+      "kl": 0.04486083984375,
+      "learning_rate": 2.51664912736018e-06,
+      "loss": -0.0211,
+      "reward": 0.7222222313284874,
+      "reward_std": 0.22214906911055246,
+      "rewards/accuracy_reward": 0.7222222313284874,
+      "rewards/format_reward": 0.0,
+      "step": 94
+    },
+    {
+      "completion_length": 447.9583396911621,
+      "epoch": 1.0106666666666666,
+      "grad_norm": 0.4173281192779541,
+      "kl": 0.398345947265625,
+      "learning_rate": 2.5027654174722495e-06,
+      "loss": 0.0542,
+      "reward": 0.7500000149011612,
+      "reward_std": 0.2957112565636635,
+      "rewards/accuracy_reward": 0.7500000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 95
+    },
+    {
+      "completion_length": 395.4583377838135,
+      "epoch": 1.0213333333333334,
+      "grad_norm": 0.09239889681339264,
+      "kl": 0.19317626953125,
+      "learning_rate": 2.4887246185099237e-06,
+      "loss": 0.0251,
+      "reward": 0.8125000149011612,
+      "reward_std": 0.1530931107699871,
+      "rewards/accuracy_reward": 0.8125000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 96
+    },
+    {
+      "completion_length": 426.6875228881836,
+      "epoch": 1.032,
+      "grad_norm": 0.6046031713485718,
+      "kl": 0.07399749755859375,
+      "learning_rate": 2.474528930046578e-06,
+      "loss": 0.0611,
+      "reward": 0.8541666865348816,
+      "reward_std": 0.27258946001529694,
+      "rewards/accuracy_reward": 0.8541666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 97
+    },
+    {
+      "completion_length": 423.70834732055664,
+      "epoch": 1.0426666666666666,
+      "grad_norm": 0.11195939779281616,
+      "kl": 0.19476318359375,
+      "learning_rate": 2.4601805759199317e-06,
+      "loss": 0.0256,
+      "reward": 0.604166679084301,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.604166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 98
+    },
+    {
+      "completion_length": 432.6041889190674,
+      "epoch": 1.0533333333333332,
+      "grad_norm": 0.1116723120212555,
+      "kl": 0.0628662109375,
+      "learning_rate": 2.445681803883678e-06,
+      "loss": 0.0363,
+      "reward": 0.7916666828095913,
+      "reward_std": 0.2861081585288048,
+      "rewards/accuracy_reward": 0.7916666828095913,
+      "rewards/format_reward": 0.0,
+      "step": 99
+    },
+    {
+      "completion_length": 583.3125190734863,
+      "epoch": 1.064,
+      "grad_norm": 0.49316033720970154,
+      "kl": 0.0525665283203125,
+      "learning_rate": 2.4310348852553554e-06,
+      "loss": 0.0765,
+      "reward": 0.6041666809469461,
+      "reward_std": 0.31970490887761116,
+      "rewards/accuracy_reward": 0.6041666809469461,
+      "rewards/format_reward": 0.0,
+      "step": 100
+    },
+    {
+      "completion_length": 448.39584159851074,
+      "epoch": 1.0746666666666667,
+      "grad_norm": 0.08873621374368668,
+      "kl": 0.0771636962890625,
+      "learning_rate": 2.4162421145605308e-06,
+      "loss": 0.0606,
+      "reward": 0.6666666772216558,
+      "reward_std": 0.3584890738129616,
+      "rewards/accuracy_reward": 0.6666666772216558,
+      "rewards/format_reward": 0.0,
+      "step": 101
+    },
+    {
+      "completion_length": 459.9166831970215,
+      "epoch": 1.0853333333333333,
+      "grad_norm": 0.09432709962129593,
+      "kl": 0.14544677734375,
+      "learning_rate": 2.4013058091733546e-06,
+      "loss": 0.0741,
+      "reward": 0.7500000167638063,
+      "reward_std": 0.2686738781630993,
+      "rewards/accuracy_reward": 0.7500000167638063,
+      "rewards/format_reward": 0.0,
+      "step": 102
+    },
+    {
+      "completion_length": 457.6041774749756,
+      "epoch": 1.096,
+      "grad_norm": 0.588718831539154,
+      "kl": 0.1486968994140625,
+      "learning_rate": 2.3862283089535265e-06,
+      "loss": -0.0277,
+      "reward": 0.6458333414047956,
+      "reward_std": 0.299626849591732,
+      "rewards/accuracy_reward": 0.6458333414047956,
+      "rewards/format_reward": 0.0,
+      "step": 103
+    },
+    {
+      "completion_length": 462.25001525878906,
+      "epoch": 1.1066666666666667,
+      "grad_norm": 0.15233053267002106,
+      "kl": 0.395355224609375,
+      "learning_rate": 2.371011975879742e-06,
+      "loss": 0.0436,
+      "reward": 0.7916666865348816,
+      "reward_std": 0.3506578840315342,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 104
+    },
+    {
+      "completion_length": 437.1458511352539,
+      "epoch": 1.1173333333333333,
+      "grad_norm": 0.61964350938797,
+      "kl": 0.1255950927734375,
+      "learning_rate": 2.3556591936796777e-06,
+      "loss": 0.0454,
+      "reward": 0.7916666716337204,
+      "reward_std": 0.11949635669589043,
+      "rewards/accuracy_reward": 0.7916666716337204,
+      "rewards/format_reward": 0.0,
+      "step": 105
+    },
+    {
+      "completion_length": 422.04167556762695,
+      "epoch": 1.1280000000000001,
+      "grad_norm": 0.24629823863506317,
+      "kl": 0.149658203125,
+      "learning_rate": 2.340172367456564e-06,
+      "loss": 0.0524,
+      "reward": 0.8125000074505806,
+      "reward_std": 0.11558076366782188,
+      "rewards/accuracy_reward": 0.8125000074505806,
+      "rewards/format_reward": 0.0,
+      "step": 106
+    },
+    {
+      "completion_length": 405.25000762939453,
+      "epoch": 1.1386666666666667,
+      "grad_norm": 0.21166791021823883,
+      "kl": 0.0451202392578125,
+      "learning_rate": 2.3245539233124133e-06,
+      "loss": 0.0262,
+      "reward": 0.7916666939854622,
+      "reward_std": 0.3977733291685581,
+      "rewards/accuracy_reward": 0.7916666939854622,
+      "rewards/format_reward": 0.0,
+      "step": 107
+    },
+    {
+      "completion_length": 471.68751525878906,
+      "epoch": 1.1493333333333333,
+      "grad_norm": 0.07898660004138947,
+      "kl": 0.4488983154296875,
+      "learning_rate": 2.308806307967955e-06,
+      "loss": 0.0136,
+      "reward": 0.8333333432674408,
+      "reward_std": 0.10206207260489464,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 108
+    },
+    {
+      "completion_length": 382.3333435058594,
+      "epoch": 1.16,
+      "grad_norm": 1.6361370086669922,
+      "kl": 0.073944091796875,
+      "learning_rate": 2.2929319883793442e-06,
+      "loss": -0.0109,
+      "reward": 0.7500000055879354,
+      "reward_std": 0.23899271339178085,
+      "rewards/accuracy_reward": 0.7500000055879354,
+      "rewards/format_reward": 0.0,
+      "step": 109
+    },
+    {
+      "completion_length": 532.5208511352539,
+      "epoch": 1.1706666666666667,
+      "grad_norm": 0.06067619472742081,
+      "kl": 0.10120391845703125,
+      "learning_rate": 2.276933451351696e-06,
+      "loss": 0.0476,
+      "reward": 0.729166679084301,
+      "reward_std": 0.2350771240890026,
+      "rewards/accuracy_reward": 0.729166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 110
+    },
+    {
+      "completion_length": 466.6666793823242,
+      "epoch": 1.1813333333333333,
+      "grad_norm": 0.11847035586833954,
+      "kl": 0.12054443359375,
+      "learning_rate": 2.2608132031495184e-06,
+      "loss": 0.0345,
+      "reward": 0.625000013038516,
+      "reward_std": 0.2861081622540951,
+      "rewards/accuracy_reward": 0.625000013038516,
+      "rewards/format_reward": 0.0,
+      "step": 111
+    },
+    {
+      "completion_length": 531.9583473205566,
+      "epoch": 1.192,
+      "grad_norm": 0.15303348004817963,
+      "kl": 0.3079833984375,
+      "learning_rate": 2.244573769104084e-06,
+      "loss": 0.0383,
+      "reward": 0.7083333414047956,
+      "reward_std": 0.16661180183291435,
+      "rewards/accuracy_reward": 0.7083333414047956,
+      "rewards/format_reward": 0.0,
+      "step": 112
+    },
+    {
+      "completion_length": 480.166690826416,
+      "epoch": 1.2026666666666666,
+      "grad_norm": 0.1532401591539383,
+      "kl": 0.077362060546875,
+      "learning_rate": 2.228217693217826e-06,
+      "loss": 0.0285,
+      "reward": 0.729166692122817,
+      "reward_std": 0.3842546306550503,
+      "rewards/accuracy_reward": 0.729166692122817,
+      "rewards/format_reward": 0.0,
+      "step": 113
+    },
+    {
+      "completion_length": 494.3125114440918,
+      "epoch": 1.2133333333333334,
+      "grad_norm": 0.14180000126361847,
+      "kl": 0.2322540283203125,
+      "learning_rate": 2.211747537765808e-06,
+      "loss": -0.0281,
+      "reward": 0.6875000186264515,
+      "reward_std": 0.28219256550073624,
+      "rewards/accuracy_reward": 0.6875000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 114
+    },
+    {
+      "completion_length": 498.6875114440918,
+      "epoch": 1.224,
+      "grad_norm": 0.31408238410949707,
+      "kl": 0.0731964111328125,
+      "learning_rate": 2.195165882894322e-06,
+      "loss": 0.0403,
+      "reward": 0.5416666809469461,
+      "reward_std": 0.3506578914821148,
+      "rewards/accuracy_reward": 0.5416666809469461,
+      "rewards/format_reward": 0.0,
+      "step": 115
+    },
+    {
+      "completion_length": 456.3333396911621,
+      "epoch": 1.2346666666666666,
+      "grad_norm": 0.13221539556980133,
+      "kl": 0.1497650146484375,
+      "learning_rate": 2.1784753262166986e-06,
+      "loss": 0.0148,
+      "reward": 0.7916666865348816,
+      "reward_std": 0.3506578840315342,
+      "rewards/accuracy_reward": 0.7916666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 116
+    },
+    {
+      "completion_length": 580.9166793823242,
+      "epoch": 1.2453333333333334,
+      "grad_norm": 0.15423516929149628,
+      "kl": 0.1390533447265625,
+      "learning_rate": 2.161678482406374e-06,
+      "loss": 0.0291,
+      "reward": 0.41666668094694614,
+      "reward_std": 0.3602609857916832,
+      "rewards/accuracy_reward": 0.41666668094694614,
+      "rewards/format_reward": 0.0,
+      "step": 117
+    },
+    {
+      "completion_length": 430.7083435058594,
+      "epoch": 1.256,
+      "grad_norm": 0.17135007679462433,
+      "kl": 0.1475372314453125,
+      "learning_rate": 2.144777982787286e-06,
+      "loss": 0.0145,
+      "reward": 0.645833345130086,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.645833345130086,
+      "rewards/format_reward": 0.0,
+      "step": 118
+    },
+    {
+      "completion_length": 591.9166736602783,
+      "epoch": 1.2666666666666666,
+      "grad_norm": 0.4398789703845978,
+      "kl": 0.23828125,
+      "learning_rate": 2.127776474921661e-06,
+      "loss": 0.0436,
+      "reward": 0.5416666846722364,
+      "reward_std": 0.3332235999405384,
+      "rewards/accuracy_reward": 0.5416666846722364,
+      "rewards/format_reward": 0.0,
+      "step": 119
+    },
+    {
+      "completion_length": 375.1875171661377,
+      "epoch": 1.2773333333333334,
+      "grad_norm": 0.19881169497966766,
+      "kl": 0.38733673095703125,
+      "learning_rate": 2.1106766221952607e-06,
+      "loss": -0.0301,
+      "reward": 0.916666679084301,
+      "reward_std": 0.16661180183291435,
+      "rewards/accuracy_reward": 0.916666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 120
+    },
+    {
+      "completion_length": 527.6250133514404,
+      "epoch": 1.288,
+      "grad_norm": 0.3269664943218231,
+      "kl": 0.12579345703125,
+      "learning_rate": 2.0934811034001427e-06,
+      "loss": 0.0109,
+      "reward": 0.750000013038516,
+      "reward_std": 0.2861081548035145,
+      "rewards/accuracy_reward": 0.750000013038516,
+      "rewards/format_reward": 0.0,
+      "step": 121
+    },
+    {
+      "completion_length": 440.5833435058594,
+      "epoch": 1.2986666666666666,
+      "grad_norm": 0.3481252193450928,
+      "kl": 1.69122314453125,
+      "learning_rate": 2.0761926123150166e-06,
+      "loss": 0.1111,
+      "reward": 0.8750000223517418,
+      "reward_std": 0.2686738669872284,
+      "rewards/accuracy_reward": 0.8750000223517418,
+      "rewards/format_reward": 0.0,
+      "step": 122
+    },
+    {
+      "completion_length": 473.87500381469727,
+      "epoch": 1.3093333333333335,
+      "grad_norm": 3.613269090652466,
+      "kl": 0.87823486328125,
+      "learning_rate": 2.058813857283244e-06,
+      "loss": 0.0814,
+      "reward": 0.6458333432674408,
+      "reward_std": 0.1801304928958416,
+      "rewards/accuracy_reward": 0.6458333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 123
+    },
+    {
+      "completion_length": 527.291675567627,
+      "epoch": 1.32,
+      "grad_norm": 0.14033041894435883,
+      "kl": 0.0919647216796875,
+      "learning_rate": 2.041347560788562e-06,
+      "loss": 0.0705,
+      "reward": 0.6041666846722364,
+      "reward_std": 0.40168892592191696,
+      "rewards/accuracy_reward": 0.6041666846722364,
+      "rewards/format_reward": 0.0,
+      "step": 124
+    },
+    {
+      "completion_length": 509.95834159851074,
+      "epoch": 1.3306666666666667,
+      "grad_norm": 0.19475488364696503,
+      "kl": 0.170867919921875,
+      "learning_rate": 2.0237964590285924e-06,
+      "loss": 0.079,
+      "reward": 0.6875000186264515,
+      "reward_std": 0.31970490515232086,
+      "rewards/accuracy_reward": 0.6875000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 125
+    },
+    {
+      "completion_length": 521.8958435058594,
+      "epoch": 1.3413333333333333,
+      "grad_norm": 0.30634742975234985,
+      "kl": 0.1868133544921875,
+      "learning_rate": 2.0061633014861975e-06,
+      "loss": 0.058,
+      "reward": 0.7083333432674408,
+      "reward_std": 0.37592336162924767,
+      "rewards/accuracy_reward": 0.7083333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 126
+    },
+    {
+      "completion_length": 388.7083511352539,
+      "epoch": 1.3519999999999999,
+      "grad_norm": 0.23578259348869324,
+      "kl": 0.802276611328125,
+      "learning_rate": 1.988450850498758e-06,
+      "loss": 0.0803,
+      "reward": 0.5625000167638063,
+      "reward_std": 0.34674229845404625,
+      "rewards/accuracy_reward": 0.5625000167638063,
+      "rewards/format_reward": 0.0,
+      "step": 127
+    },
+    {
+      "completion_length": 473.77084732055664,
+      "epoch": 1.3626666666666667,
+      "grad_norm": 0.7195316553115845,
+      "kl": 0.4127960205078125,
+      "learning_rate": 1.970661880825437e-06,
+      "loss": 0.0466,
+      "reward": 0.6458333432674408,
+      "reward_std": 0.28219256550073624,
+      "rewards/accuracy_reward": 0.6458333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 128
+    },
+    {
+      "completion_length": 606.7500190734863,
+      "epoch": 1.3733333333333333,
+      "grad_norm": 0.3938013017177582,
+      "kl": 0.5943145751953125,
+      "learning_rate": 1.952799179212498e-06,
+      "loss": 0.0522,
+      "reward": 0.39583334513008595,
+      "reward_std": 0.37377968430519104,
+      "rewards/accuracy_reward": 0.39583334513008595,
+      "rewards/format_reward": 0.0,
+      "step": 129
+    },
+    {
+      "completion_length": 446.3333435058594,
+      "epoch": 1.384,
+      "grad_norm": 1.3521168231964111,
+      "kl": 0.2550506591796875,
+      "learning_rate": 1.934865543956745e-06,
+      "loss": 0.0415,
+      "reward": 0.729166679084301,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.729166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 130
+    },
+    {
+      "completion_length": 402.5833511352539,
+      "epoch": 1.3946666666666667,
+      "grad_norm": 0.35991618037223816,
+      "kl": 0.294891357421875,
+      "learning_rate": 1.916863784467152e-06,
+      "loss": -0.0116,
+      "reward": 0.708333358168602,
+      "reward_std": 0.3236205019056797,
+      "rewards/accuracy_reward": 0.708333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 131
+    },
+    {
+      "completion_length": 428.6041793823242,
+      "epoch": 1.4053333333333333,
+      "grad_norm": 0.151389479637146,
+      "kl": 0.09417724609375,
+      "learning_rate": 1.8987967208247542e-06,
+      "loss": 0.055,
+      "reward": 0.7291666828095913,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.7291666828095913,
+      "rewards/format_reward": 0.0,
+      "step": 132
+    },
+    {
+      "completion_length": 538.6666831970215,
+      "epoch": 1.416,
+      "grad_norm": 0.0664629116654396,
+      "kl": 0.16326904296875,
+      "learning_rate": 1.8806671833408633e-06,
+      "loss": 0.0104,
+      "reward": 0.729166679084301,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.729166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 133
+    },
+    {
+      "completion_length": 406.70834732055664,
+      "epoch": 1.4266666666666667,
+      "grad_norm": 0.21904991567134857,
+      "kl": 0.29296875,
+      "learning_rate": 1.8624780121136837e-06,
+      "loss": 0.0634,
+      "reward": 0.8333333432674408,
+      "reward_std": 0.10206206887960434,
+      "rewards/accuracy_reward": 0.8333333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 134
+    },
+    {
+      "completion_length": 441.9791831970215,
+      "epoch": 1.4373333333333334,
+      "grad_norm": 1.1249885559082031,
+      "kl": 0.94659423828125,
+      "learning_rate": 1.8442320565833972e-06,
+      "loss": 0.0591,
+      "reward": 0.7291666865348816,
+      "reward_std": 0.41912321001291275,
+      "rewards/accuracy_reward": 0.7291666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 135
+    },
+    {
+      "completion_length": 506.5625114440918,
+      "epoch": 1.448,
+      "grad_norm": 0.2663201093673706,
+      "kl": 0.28574371337890625,
+      "learning_rate": 1.8259321750857772e-06,
+      "loss": 0.0732,
+      "reward": 0.6250000111758709,
+      "reward_std": 0.3506578877568245,
+      "rewards/accuracy_reward": 0.6250000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 136
+    },
+    {
+      "completion_length": 429.52084732055664,
+      "epoch": 1.4586666666666668,
+      "grad_norm": 0.4094298481941223,
+      "kl": 0.2675933837890625,
+      "learning_rate": 1.8075812344044163e-06,
+      "loss": 0.0398,
+      "reward": 0.6875000111758709,
+      "reward_std": 0.28219256550073624,
+      "rewards/accuracy_reward": 0.6875000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 137
+    },
+    {
+      "completion_length": 540.2708435058594,
+      "epoch": 1.4693333333333334,
+      "grad_norm": 0.5432186126708984,
+      "kl": 0.3033294677734375,
+      "learning_rate": 1.7891821093216275e-06,
+      "loss": 0.1026,
+      "reward": 0.6666666902601719,
+      "reward_std": 0.4701542519032955,
+      "rewards/accuracy_reward": 0.6666666902601719,
+      "rewards/format_reward": 0.0,
+      "step": 138
+    },
+    {
+      "completion_length": 478.37501525878906,
+      "epoch": 1.48,
+      "grad_norm": 0.1327861100435257,
+      "kl": 0.496612548828125,
+      "learning_rate": 1.770737682168091e-06,
+      "loss": 0.0499,
+      "reward": 0.770833345130086,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.770833345130086,
+      "rewards/format_reward": 0.0,
+      "step": 139
+    },
+    {
+      "completion_length": 560.7083511352539,
+      "epoch": 1.4906666666666666,
+      "grad_norm": 0.4717060625553131,
+      "kl": 0.289215087890625,
+      "learning_rate": 1.7522508423713243e-06,
+      "loss": 0.058,
+      "reward": 0.6250000186264515,
+      "reward_std": 0.3881702274084091,
+      "rewards/accuracy_reward": 0.6250000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 140
+    },
+    {
+      "completion_length": 388.1666717529297,
+      "epoch": 1.5013333333333332,
+      "grad_norm": 0.16122129559516907,
+      "kl": 0.19562530517578125,
+      "learning_rate": 1.7337244860030312e-06,
+      "loss": -0.0199,
+      "reward": 0.8541666772216558,
+      "reward_std": 0.1530931107699871,
+      "rewards/accuracy_reward": 0.8541666772216558,
+      "rewards/format_reward": 0.0,
+      "step": 141
+    },
+    {
+      "completion_length": 542.2291793823242,
+      "epoch": 1.512,
+      "grad_norm": 0.15329065918922424,
+      "kl": 0.45532989501953125,
+      "learning_rate": 1.7151615153254192e-06,
+      "loss": -0.0001,
+      "reward": 0.6458333432674408,
+      "reward_std": 0.1705273911356926,
+      "rewards/accuracy_reward": 0.6458333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 142
+    },
+    {
+      "completion_length": 500.3541793823242,
+      "epoch": 1.5226666666666666,
+      "grad_norm": 0.30960169434547424,
+      "kl": 0.4269866943359375,
+      "learning_rate": 1.6965648383365416e-06,
+      "loss": 0.1314,
+      "reward": 0.6458333488553762,
+      "reward_std": 0.41912321001291275,
+      "rewards/accuracy_reward": 0.6458333488553762,
+      "rewards/format_reward": 0.0,
+      "step": 143
+    },
+    {
+      "completion_length": 519.7916812896729,
+      "epoch": 1.5333333333333332,
+      "grad_norm": 0.2385687530040741,
+      "kl": 0.47222900390625,
+      "learning_rate": 1.6779373683147423e-06,
+      "loss": 0.0957,
+      "reward": 0.645833345130086,
+      "reward_std": 0.35457347705960274,
+      "rewards/accuracy_reward": 0.645833345130086,
+      "rewards/format_reward": 0.0,
+      "step": 144
+    },
+    {
+      "completion_length": 523.5833549499512,
+      "epoch": 1.544,
+      "grad_norm": 0.18868695199489594,
+      "kl": 0.495849609375,
+      "learning_rate": 1.6592820233622743e-06,
+      "loss": 0.0457,
+      "reward": 0.6666666772216558,
+      "reward_std": 0.3680921718478203,
+      "rewards/accuracy_reward": 0.6666666772216558,
+      "rewards/format_reward": 0.0,
+      "step": 145
+    },
+    {
+      "completion_length": 535.979190826416,
+      "epoch": 1.5546666666666666,
+      "grad_norm": 0.2131403535604477,
+      "kl": 0.47271728515625,
+      "learning_rate": 1.6406017259481624e-06,
+      "loss": -0.003,
+      "reward": 0.45833334140479565,
+      "reward_std": 0.30354244261980057,
+      "rewards/accuracy_reward": 0.45833334140479565,
+      "rewards/format_reward": 0.0,
+      "step": 146
+    },
+    {
+      "completion_length": 525.7708435058594,
+      "epoch": 1.5653333333333332,
+      "grad_norm": 0.3221761882305145,
+      "kl": 0.982421875,
+      "learning_rate": 1.621899402450379e-06,
+      "loss": 0.0629,
+      "reward": 0.6875000316649675,
+      "reward_std": 0.4217669740319252,
+      "rewards/accuracy_reward": 0.6875000316649675,
+      "rewards/format_reward": 0.0,
+      "step": 147
+    },
+    {
+      "completion_length": 558.7916793823242,
+      "epoch": 1.576,
+      "grad_norm": 0.7135828733444214,
+      "kl": 0.4044647216796875,
+      "learning_rate": 1.6031779826974138e-06,
+      "loss": 0.047,
+      "reward": 0.6041666809469461,
+      "reward_std": 0.31970490887761116,
+      "rewards/accuracy_reward": 0.6041666809469461,
+      "rewards/format_reward": 0.0,
+      "step": 148
+    },
+    {
+      "completion_length": 449.75000381469727,
+      "epoch": 1.5866666666666667,
+      "grad_norm": 0.20385055243968964,
+      "kl": 0.481414794921875,
+      "learning_rate": 1.5844403995092944e-06,
+      "loss": 0.0377,
+      "reward": 0.6250000018626451,
+      "reward_std": 0.32097673043608665,
+      "rewards/accuracy_reward": 0.6250000018626451,
+      "rewards/format_reward": 0.0,
+      "step": 149
+    },
+    {
+      "completion_length": 523.3125190734863,
+      "epoch": 1.5973333333333333,
+      "grad_norm": 0.1701987087726593,
+      "kl": 0.2771759033203125,
+      "learning_rate": 1.5656895882381465e-06,
+      "loss": 0.058,
+      "reward": 0.5416666753590107,
+      "reward_std": 0.3506578840315342,
+      "rewards/accuracy_reward": 0.5416666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 150
+    },
+    {
+      "completion_length": 493.4166831970215,
+      "epoch": 1.608,
+      "grad_norm": 0.1551842838525772,
+      "kl": 0.755859375,
+      "learning_rate": 1.5469284863083523e-06,
+      "loss": 0.0446,
+      "reward": 0.5625000093132257,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.5625000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 151
+    },
+    {
+      "completion_length": 567.7291793823242,
+      "epoch": 1.6186666666666667,
+      "grad_norm": 1.0995123386383057,
+      "kl": 0.3485260009765625,
+      "learning_rate": 1.5281600327563853e-06,
+      "loss": 0.0618,
+      "reward": 0.5000000093132257,
+      "reward_std": 0.3680921755731106,
+      "rewards/accuracy_reward": 0.5000000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 152
+    },
+    {
+      "completion_length": 467.18750762939453,
+      "epoch": 1.6293333333333333,
+      "grad_norm": 0.18746134638786316,
+      "kl": 0.257080078125,
+      "learning_rate": 1.5093871677703947e-06,
+      "loss": 0.0503,
+      "reward": 0.7083333507180214,
+      "reward_std": 0.2957112602889538,
+      "rewards/accuracy_reward": 0.7083333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 153
+    },
+    {
+      "completion_length": 583.5416793823242,
+      "epoch": 1.6400000000000001,
+      "grad_norm": 0.12889082729816437,
+      "kl": 0.1910552978515625,
+      "learning_rate": 1.4906128322296056e-06,
+      "loss": 0.0376,
+      "reward": 0.5625000204890966,
+      "reward_std": 0.4662386514246464,
+      "rewards/accuracy_reward": 0.5625000204890966,
+      "rewards/format_reward": 0.0,
+      "step": 154
+    },
+    {
+      "completion_length": 585.5625190734863,
+      "epoch": 1.6506666666666665,
+      "grad_norm": 0.22395263612270355,
+      "kl": 0.3731842041015625,
+      "learning_rate": 1.471839967243615e-06,
+      "loss": 0.0831,
+      "reward": 0.33333333767950535,
+      "reward_std": 0.31314554437994957,
+      "rewards/accuracy_reward": 0.33333333767950535,
+      "rewards/format_reward": 0.0,
+      "step": 155
+    },
+    {
+      "completion_length": 555.7500228881836,
+      "epoch": 1.6613333333333333,
+      "grad_norm": 0.1528111696243286,
+      "kl": 0.4114990234375,
+      "learning_rate": 1.453071513691648e-06,
+      "loss": 0.0309,
+      "reward": 0.3750000111758709,
+      "reward_std": 0.16661180555820465,
+      "rewards/accuracy_reward": 0.3750000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 156
+    },
+    {
+      "completion_length": 626.854190826416,
+      "epoch": 1.6720000000000002,
+      "grad_norm": 1.3844571113586426,
+      "kl": 0.439453125,
+      "learning_rate": 1.4343104117618536e-06,
+      "loss": 0.0231,
+      "reward": 0.4166666679084301,
+      "reward_std": 0.20148037374019623,
+      "rewards/accuracy_reward": 0.4166666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 157
+    },
+    {
+      "completion_length": 484.0416831970215,
+      "epoch": 1.6826666666666665,
+      "grad_norm": 0.21760034561157227,
+      "kl": 0.401092529296875,
+      "learning_rate": 1.415559600490706e-06,
+      "loss": 0.0216,
+      "reward": 0.500000013038516,
+      "reward_std": 0.2861081585288048,
+      "rewards/accuracy_reward": 0.500000013038516,
+      "rewards/format_reward": 0.0,
+      "step": 158
+    },
+    {
+      "completion_length": 418.87501525878906,
+      "epoch": 1.6933333333333334,
+      "grad_norm": 0.11006595194339752,
+      "kl": 0.1508026123046875,
+      "learning_rate": 1.3968220173025865e-06,
+      "loss": 0.0566,
+      "reward": 0.8125000111758709,
+      "reward_std": 0.2350771240890026,
+      "rewards/accuracy_reward": 0.8125000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 159
+    },
+    {
+      "completion_length": 448.8750114440918,
+      "epoch": 1.704,
+      "grad_norm": 0.6605373024940491,
+      "kl": 0.356903076171875,
+      "learning_rate": 1.3781005975496212e-06,
+      "loss": 0.095,
+      "reward": 0.6250000093132257,
+      "reward_std": 0.3776952736079693,
+      "rewards/accuracy_reward": 0.6250000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 160
+    },
+    {
+      "completion_length": 583.6875152587891,
+      "epoch": 1.7146666666666666,
+      "grad_norm": 0.1950928419828415,
+      "kl": 0.2814178466796875,
+      "learning_rate": 1.359398274051838e-06,
+      "loss": 0.041,
+      "reward": 0.4583333395421505,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.4583333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 161
+    },
+    {
+      "completion_length": 504.18751525878906,
+      "epoch": 1.7253333333333334,
+      "grad_norm": 0.23138457536697388,
+      "kl": 0.20184326171875,
+      "learning_rate": 1.340717976637726e-06,
+      "loss": 0.0413,
+      "reward": 0.7083333469927311,
+      "reward_std": 0.3776952736079693,
+      "rewards/accuracy_reward": 0.7083333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 162
+    },
+    {
+      "completion_length": 569.791675567627,
+      "epoch": 1.736,
+      "grad_norm": 0.09024322777986526,
+      "kl": 0.105865478515625,
+      "learning_rate": 1.3220626316852584e-06,
+      "loss": 0.0311,
+      "reward": 0.7083333525806665,
+      "reward_std": 0.3332235999405384,
+      "rewards/accuracy_reward": 0.7083333525806665,
+      "rewards/format_reward": 0.0,
+      "step": 163
+    },
+    {
+      "completion_length": 439.8125190734863,
+      "epoch": 1.7466666666666666,
+      "grad_norm": 0.23019039630889893,
+      "kl": 0.1641387939453125,
+      "learning_rate": 1.303435161663459e-06,
+      "loss": -0.0082,
+      "reward": 0.7708333507180214,
+      "reward_std": 0.28219256550073624,
+      "rewards/accuracy_reward": 0.7708333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 164
+    },
+    {
+      "completion_length": 569.1250114440918,
+      "epoch": 1.7573333333333334,
+      "grad_norm": 0.15718743205070496,
+      "kl": 0.53558349609375,
+      "learning_rate": 1.2848384846745813e-06,
+      "loss": 0.0633,
+      "reward": 0.4583333432674408,
+      "reward_std": 0.24859581887722015,
+      "rewards/accuracy_reward": 0.4583333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 165
+    },
+    {
+      "completion_length": 471.64585876464844,
+      "epoch": 1.768,
+      "grad_norm": 0.3743435740470886,
+      "kl": 0.29827880859375,
+      "learning_rate": 1.2662755139969693e-06,
+      "loss": 0.0573,
+      "reward": 0.6458333469927311,
+      "reward_std": 0.309229951351881,
+      "rewards/accuracy_reward": 0.6458333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 166
+    },
+    {
+      "completion_length": 656.2291870117188,
+      "epoch": 1.7786666666666666,
+      "grad_norm": 0.2532089650630951,
+      "kl": 0.27117919921875,
+      "learning_rate": 1.2477491576286764e-06,
+      "loss": 0.064,
+      "reward": 0.6875000149011612,
+      "reward_std": 0.44616059958934784,
+      "rewards/accuracy_reward": 0.6875000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 167
+    },
+    {
+      "completion_length": 563.5208473205566,
+      "epoch": 1.7893333333333334,
+      "grad_norm": 0.20364205539226532,
+      "kl": 0.1768341064453125,
+      "learning_rate": 1.2292623178319094e-06,
+      "loss": 0.0382,
+      "reward": 0.5625000111758709,
+      "reward_std": 0.2350771240890026,
+      "rewards/accuracy_reward": 0.5625000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 168
+    },
+    {
+      "completion_length": 496.68751525878906,
+      "epoch": 1.8,
+      "grad_norm": 0.10253948718309402,
+      "kl": 0.3402252197265625,
+      "learning_rate": 1.2108178906783732e-06,
+      "loss": 0.059,
+      "reward": 0.6250000093132257,
+      "reward_std": 0.2957112565636635,
+      "rewards/accuracy_reward": 0.6250000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 169
+    },
+    {
+      "completion_length": 479.14584732055664,
+      "epoch": 1.8106666666666666,
+      "grad_norm": 0.1836300790309906,
+      "kl": 0.35040283203125,
+      "learning_rate": 1.1924187655955838e-06,
+      "loss": 0.0119,
+      "reward": 0.645833345130086,
+      "reward_std": 0.21764283254742622,
+      "rewards/accuracy_reward": 0.645833345130086,
+      "rewards/format_reward": 0.0,
+      "step": 170
+    },
+    {
+      "completion_length": 470.6041774749756,
+      "epoch": 1.8213333333333335,
+      "grad_norm": 0.1450057029724121,
+      "kl": 0.4749908447265625,
+      "learning_rate": 1.1740678249142229e-06,
+      "loss": 0.108,
+      "reward": 0.7500000223517418,
+      "reward_std": 0.3332235962152481,
+      "rewards/accuracy_reward": 0.7500000223517418,
+      "rewards/format_reward": 0.0,
+      "step": 171
+    },
+    {
+      "completion_length": 513.5625152587891,
+      "epoch": 1.8319999999999999,
+      "grad_norm": 0.25482821464538574,
+      "kl": 0.2546844482421875,
+      "learning_rate": 1.1557679434166027e-06,
+      "loss": 0.0149,
+      "reward": 0.7500000149011612,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.7500000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 172
+    },
+    {
+      "completion_length": 472.0416831970215,
+      "epoch": 1.8426666666666667,
+      "grad_norm": 0.18026013672351837,
+      "kl": 0.26741790771484375,
+      "learning_rate": 1.1375219878863162e-06,
+      "loss": 0.0597,
+      "reward": 0.7708333507180214,
+      "reward_std": 0.21764283254742622,
+      "rewards/accuracy_reward": 0.7708333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 173
+    },
+    {
+      "completion_length": 439.47917556762695,
+      "epoch": 1.8533333333333335,
+      "grad_norm": 0.21356335282325745,
+      "kl": 0.47174072265625,
+      "learning_rate": 1.1193328166591372e-06,
+      "loss": 0.0628,
+      "reward": 0.6875000167638063,
+      "reward_std": 0.33713918924331665,
+      "rewards/accuracy_reward": 0.6875000167638063,
+      "rewards/format_reward": 0.0,
+      "step": 174
+    },
+    {
+      "completion_length": 518.7291793823242,
+      "epoch": 1.8639999999999999,
+      "grad_norm": 0.8327506184577942,
+      "kl": 0.773590087890625,
+      "learning_rate": 1.1012032791752457e-06,
+      "loss": 0.0461,
+      "reward": 0.604166679084301,
+      "reward_std": 0.299626849591732,
+      "rewards/accuracy_reward": 0.604166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 175
+    },
+    {
+      "completion_length": 528.9583473205566,
+      "epoch": 1.8746666666666667,
+      "grad_norm": 0.17722496390342712,
+      "kl": 0.52960205078125,
+      "learning_rate": 1.0831362155328477e-06,
+      "loss": 0.0374,
+      "reward": 0.4375000111758709,
+      "reward_std": 0.3816108666360378,
+      "rewards/accuracy_reward": 0.4375000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 176
+    },
+    {
+      "completion_length": 421.2291717529297,
+      "epoch": 1.8853333333333333,
+      "grad_norm": 0.27561092376708984,
+      "kl": 0.6494598388671875,
+      "learning_rate": 1.0651344560432548e-06,
+      "loss": 0.0383,
+      "reward": 0.7500000149011612,
+      "reward_std": 0.2957112565636635,
+      "rewards/accuracy_reward": 0.7500000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 177
+    },
+    {
+      "completion_length": 562.2291831970215,
+      "epoch": 1.896,
+      "grad_norm": 0.7644630670547485,
+      "kl": 1.01025390625,
+      "learning_rate": 1.0472008207875023e-06,
+      "loss": 0.043,
+      "reward": 0.6250000204890966,
+      "reward_std": 0.3977733366191387,
+      "rewards/accuracy_reward": 0.6250000204890966,
+      "rewards/format_reward": 0.0,
+      "step": 178
+    },
+    {
+      "completion_length": 457.8333435058594,
+      "epoch": 1.9066666666666667,
+      "grad_norm": 0.22743047773838043,
+      "kl": 0.37873077392578125,
+      "learning_rate": 1.0293381191745637e-06,
+      "loss": 0.0305,
+      "reward": 0.6875000111758709,
+      "reward_std": 0.28219256550073624,
+      "rewards/accuracy_reward": 0.6875000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 179
+    },
+    {
+      "completion_length": 526.6458511352539,
+      "epoch": 1.9173333333333333,
+      "grad_norm": 0.15377774834632874,
+      "kl": 0.2771453857421875,
+      "learning_rate": 1.0115491495012424e-06,
+      "loss": 0.0542,
+      "reward": 0.6250000074505806,
+      "reward_std": 0.18404608964920044,
+      "rewards/accuracy_reward": 0.6250000074505806,
+      "rewards/format_reward": 0.0,
+      "step": 180
+    },
+    {
+      "completion_length": 596.8750190734863,
+      "epoch": 1.928,
+      "grad_norm": 0.12089336663484573,
+      "kl": 0.39202880859375,
+      "learning_rate": 9.938366985138028e-07,
+      "loss": 0.0009,
+      "reward": 0.520833333954215,
+      "reward_std": 0.18796167895197868,
+      "rewards/accuracy_reward": 0.520833333954215,
+      "rewards/format_reward": 0.0,
+      "step": 181
+    },
+    {
+      "completion_length": 491.6666736602783,
+      "epoch": 1.9386666666666668,
+      "grad_norm": 0.17418581247329712,
+      "kl": 0.27716064453125,
+      "learning_rate": 9.762035409714077e-07,
+      "loss": 0.0182,
+      "reward": 0.6666666734963655,
+      "reward_std": 0.16661179810762405,
+      "rewards/accuracy_reward": 0.6666666734963655,
+      "rewards/format_reward": 0.0,
+      "step": 182
+    },
+    {
+      "completion_length": 506.02084732055664,
+      "epoch": 1.9493333333333334,
+      "grad_norm": 0.16895778477191925,
+      "kl": 0.2916107177734375,
+      "learning_rate": 9.586524392114385e-07,
+      "loss": 0.0549,
+      "reward": 0.5625000111758709,
+      "reward_std": 0.2446802221238613,
+      "rewards/accuracy_reward": 0.5625000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 183
+    },
+    {
+      "completion_length": 377.27084159851074,
+      "epoch": 1.96,
+      "grad_norm": 0.10825838148593903,
+      "kl": 0.2097015380859375,
+      "learning_rate": 9.411861427167567e-07,
+      "loss": 0.0361,
+      "reward": 0.6875000186264515,
+      "reward_std": 0.31970491260290146,
+      "rewards/accuracy_reward": 0.6875000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 184
+    },
+    {
+      "completion_length": 432.77084732055664,
+      "epoch": 1.9706666666666668,
+      "grad_norm": 0.13714736700057983,
+      "kl": 0.476348876953125,
+      "learning_rate": 9.238073876849837e-07,
+      "loss": 0.0672,
+      "reward": 0.750000013038516,
+      "reward_std": 0.2861081585288048,
+      "rewards/accuracy_reward": 0.750000013038516,
+      "rewards/format_reward": 0.0,
+      "step": 185
+    },
+    {
+      "completion_length": 560.0208511352539,
+      "epoch": 1.9813333333333332,
+      "grad_norm": 0.5157011151313782,
+      "kl": 0.41839599609375,
+      "learning_rate": 9.065188965998574e-07,
+      "loss": 0.0186,
+      "reward": 0.5416666883975267,
+      "reward_std": 0.3881702348589897,
+      "rewards/accuracy_reward": 0.5416666883975267,
+      "rewards/format_reward": 0.0,
+      "step": 186
+    },
+    {
+      "completion_length": 494.9583435058594,
+      "epoch": 1.992,
+      "grad_norm": 0.08847740292549133,
+      "kl": 0.40655517578125,
+      "learning_rate": 8.893233778047394e-07,
+      "loss": 0.0679,
+      "reward": 0.625000013038516,
+      "reward_std": 0.2861081622540951,
+      "rewards/accuracy_reward": 0.625000013038516,
+      "rewards/format_reward": 0.0,
+      "step": 187
+    },
+    {
+      "completion_length": 493.91668192545575,
+      "epoch": 2.0,
+      "grad_norm": 0.4142110049724579,
+      "kl": 2.4609375,
+      "learning_rate": 8.722235250783389e-07,
+      "loss": 0.0632,
+      "reward": 0.6388889029622078,
+      "reward_std": 0.20412414272626242,
+      "rewards/accuracy_reward": 0.6388889029622078,
+      "rewards/format_reward": 0.0,
+      "step": 188
+    },
+    {
+      "completion_length": 534.291675567627,
+      "epoch": 2.010666666666667,
+      "grad_norm": 2.5997705459594727,
+      "kl": 0.265838623046875,
+      "learning_rate": 8.552220172127148e-07,
+      "loss": 0.0693,
+      "reward": 0.6250000149011612,
+      "reward_std": 0.2957112565636635,
+      "rewards/accuracy_reward": 0.6250000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 189
+    },
+    {
+      "completion_length": 440.52084732055664,
+      "epoch": 2.021333333333333,
+      "grad_norm": 0.143593892455101,
+      "kl": 0.670684814453125,
+      "learning_rate": 8.383215175936263e-07,
+      "loss": 0.0278,
+      "reward": 0.7708333432674408,
+      "reward_std": 0.1801304928958416,
+      "rewards/accuracy_reward": 0.7708333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 190
+    },
+    {
+      "completion_length": 423.6875114440918,
+      "epoch": 2.032,
+      "grad_norm": 1.9302064180374146,
+      "kl": 0.1420440673828125,
+      "learning_rate": 8.215246737833017e-07,
+      "loss": 0.0173,
+      "reward": 0.916666679084301,
+      "reward_std": 0.16661180555820465,
+      "rewards/accuracy_reward": 0.916666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 191
+    },
+    {
+      "completion_length": 516.7291831970215,
+      "epoch": 2.042666666666667,
+      "grad_norm": 0.6368203163146973,
+      "kl": 0.239837646484375,
+      "learning_rate": 8.048341171056786e-07,
+      "loss": 0.0214,
+      "reward": 0.791666679084301,
+      "reward_std": 0.31314554437994957,
+      "rewards/accuracy_reward": 0.791666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 192
+    },
+    {
+      "completion_length": 491.18751525878906,
+      "epoch": 2.0533333333333332,
+      "grad_norm": 0.16937457025051117,
+      "kl": 0.2440948486328125,
+      "learning_rate": 7.882524622341923e-07,
+      "loss": 0.0573,
+      "reward": 0.5833333525806665,
+      "reward_std": 0.333223607391119,
+      "rewards/accuracy_reward": 0.5833333525806665,
+      "rewards/format_reward": 0.0,
+      "step": 193
+    },
+    {
+      "completion_length": 395.1666717529297,
+      "epoch": 2.064,
+      "grad_norm": 0.06481139361858368,
+      "kl": 0.24591064453125,
+      "learning_rate": 7.717823067821738e-07,
+      "loss": 0.0204,
+      "reward": 0.7500000093132257,
+      "reward_std": 0.23116153106093407,
+      "rewards/accuracy_reward": 0.7500000093132257,
+      "rewards/format_reward": 0.0,
+      "step": 194
+    },
+    {
+      "completion_length": 398.9375114440918,
+      "epoch": 2.074666666666667,
+      "grad_norm": 0.3042905032634735,
+      "kl": 0.137603759765625,
+      "learning_rate": 7.554262308959165e-07,
+      "loss": 0.0244,
+      "reward": 0.7500000037252903,
+      "reward_std": 0.26603010296821594,
+      "rewards/accuracy_reward": 0.7500000037252903,
+      "rewards/format_reward": 0.0,
+      "step": 195
+    },
+    {
+      "completion_length": 550.3333473205566,
+      "epoch": 2.0853333333333333,
+      "grad_norm": 0.1332455575466156,
+      "kl": 0.43536376953125,
+      "learning_rate": 7.39186796850482e-07,
+      "loss": 0.0316,
+      "reward": 0.4375000149011612,
+      "reward_std": 0.1530931033194065,
+      "rewards/accuracy_reward": 0.4375000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 196
+    },
+    {
+      "completion_length": 527.1250152587891,
+      "epoch": 2.096,
+      "grad_norm": 0.8033120632171631,
+      "kl": 0.222808837890625,
+      "learning_rate": 7.23066548648304e-07,
+      "loss": 0.0639,
+      "reward": 0.5000000074505806,
+      "reward_std": 0.19364918768405914,
+      "rewards/accuracy_reward": 0.5000000074505806,
+      "rewards/format_reward": 0.0,
+      "step": 197
+    },
+    {
+      "completion_length": 464.81251525878906,
+      "epoch": 2.1066666666666665,
+      "grad_norm": 0.26147812604904175,
+      "kl": 0.40435791015625,
+      "learning_rate": 7.070680116206564e-07,
+      "loss": 0.062,
+      "reward": 0.5416666865348816,
+      "reward_std": 0.45271996408700943,
+      "rewards/accuracy_reward": 0.5416666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 198
+    },
+    {
+      "completion_length": 504.8125228881836,
+      "epoch": 2.1173333333333333,
+      "grad_norm": 0.1162574291229248,
+      "kl": 0.147796630859375,
+      "learning_rate": 6.911936920320457e-07,
+      "loss": -0.0433,
+      "reward": 0.6458333469927311,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.6458333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 199
+    },
+    {
+      "completion_length": 712.8958587646484,
+      "epoch": 2.128,
+      "grad_norm": 0.06791964173316956,
+      "kl": 0.1807403564453125,
+      "learning_rate": 6.754460766875871e-07,
+      "loss": 0.0092,
+      "reward": 0.4583333395421505,
+      "reward_std": 0.19364918768405914,
+      "rewards/accuracy_reward": 0.4583333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 200
+    },
+    {
+      "completion_length": 467.39585304260254,
+      "epoch": 2.1386666666666665,
+      "grad_norm": 0.17917577922344208,
+      "kl": 0.1912841796875,
+      "learning_rate": 6.598276325434357e-07,
+      "loss": 0.0206,
+      "reward": 0.7708333507180214,
+      "reward_std": 0.36417658627033234,
+      "rewards/accuracy_reward": 0.7708333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 201
+    },
+    {
+      "completion_length": 461.9583435058594,
+      "epoch": 2.1493333333333333,
+      "grad_norm": 0.11866424977779388,
+      "kl": 0.153533935546875,
+      "learning_rate": 6.443408063203226e-07,
+      "loss": 0.0722,
+      "reward": 0.6875000111758709,
+      "reward_std": 0.3720077686011791,
+      "rewards/accuracy_reward": 0.6875000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 202
+    },
+    {
+      "completion_length": 391.6041736602783,
+      "epoch": 2.16,
+      "grad_norm": 0.09952881932258606,
+      "kl": 0.1327056884765625,
+      "learning_rate": 6.289880241202581e-07,
+      "loss": 0.024,
+      "reward": 0.8125000149011612,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.8125000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 203
+    },
+    {
+      "completion_length": 542.3125190734863,
+      "epoch": 2.1706666666666665,
+      "grad_norm": 0.11927027255296707,
+      "kl": 0.2522430419921875,
+      "learning_rate": 6.137716910464735e-07,
+      "loss": 0.0212,
+      "reward": 0.7291666865348816,
+      "reward_std": 0.34674229100346565,
+      "rewards/accuracy_reward": 0.7291666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 204
+    },
+    {
+      "completion_length": 488.5208549499512,
+      "epoch": 2.1813333333333333,
+      "grad_norm": 0.11296381801366806,
+      "kl": 0.1716156005859375,
+      "learning_rate": 5.986941908266454e-07,
+      "loss": 0.0381,
+      "reward": 0.604166679084301,
+      "reward_std": 0.309229951351881,
+      "rewards/accuracy_reward": 0.604166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 205
+    },
+    {
+      "completion_length": 447.0000114440918,
+      "epoch": 2.192,
+      "grad_norm": 0.15477237105369568,
+      "kl": 0.107513427734375,
+      "learning_rate": 5.837578854394692e-07,
+      "loss": 0.0712,
+      "reward": 0.8333333507180214,
+      "reward_std": 0.2861081548035145,
+      "rewards/accuracy_reward": 0.8333333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 206
+    },
+    {
+      "completion_length": 448.5000114440918,
+      "epoch": 2.2026666666666666,
+      "grad_norm": 0.3046277165412903,
+      "kl": 0.2861480712890625,
+      "learning_rate": 5.689651147446448e-07,
+      "loss": 0.073,
+      "reward": 0.625000013038516,
+      "reward_std": 0.4230388104915619,
+      "rewards/accuracy_reward": 0.625000013038516,
+      "rewards/format_reward": 0.0,
+      "step": 207
+    },
+    {
+      "completion_length": 500.29167556762695,
+      "epoch": 2.2133333333333334,
+      "grad_norm": 0.08913067728281021,
+      "kl": 0.178436279296875,
+      "learning_rate": 5.543181961163219e-07,
+      "loss": 0.0098,
+      "reward": 0.6250000055879354,
+      "reward_std": 0.24859581887722015,
+      "rewards/accuracy_reward": 0.6250000055879354,
+      "rewards/format_reward": 0.0,
+      "step": 208
+    },
+    {
+      "completion_length": 438.8541831970215,
+      "epoch": 2.224,
+      "grad_norm": 0.1435471922159195,
+      "kl": 0.1600494384765625,
+      "learning_rate": 5.398194240800684e-07,
+      "loss": 0.0633,
+      "reward": 0.8125000298023224,
+      "reward_std": 0.3842546418309212,
+      "rewards/accuracy_reward": 0.8125000298023224,
+      "rewards/format_reward": 0.0,
+      "step": 209
+    },
+    {
+      "completion_length": 541.8125152587891,
+      "epoch": 2.2346666666666666,
+      "grad_norm": 0.20641887187957764,
+      "kl": 0.2934722900390625,
+      "learning_rate": 5.254710699534226e-07,
+      "loss": -0.0121,
+      "reward": 0.39583334513008595,
+      "reward_std": 0.36417658627033234,
+      "rewards/accuracy_reward": 0.39583334513008595,
+      "rewards/format_reward": 0.0,
+      "step": 210
+    },
+    {
+      "completion_length": 461.3333511352539,
+      "epoch": 2.2453333333333334,
+      "grad_norm": 0.08080989122390747,
+      "kl": 0.3165435791015625,
+      "learning_rate": 5.112753814900761e-07,
+      "loss": 0.075,
+      "reward": 0.8125000111758709,
+      "reward_std": 0.2350771240890026,
+      "rewards/accuracy_reward": 0.8125000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 211
+    },
+    {
+      "completion_length": 491.79167556762695,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 0.2446739375591278,
+      "kl": 0.353851318359375,
+      "learning_rate": 4.97234582527751e-07,
+      "loss": 0.0811,
+      "reward": 0.5416666828095913,
+      "reward_std": 0.43264190107584,
+      "rewards/accuracy_reward": 0.5416666828095913,
+      "rewards/format_reward": 0.0,
+      "step": 212
+    },
+    {
+      "completion_length": 456.0833492279053,
+      "epoch": 2.2666666666666666,
+      "grad_norm": 0.14784441888332367,
+      "kl": 0.2136383056640625,
+      "learning_rate": 4.8335087263982e-07,
+      "loss": 0.0308,
+      "reward": 0.5208333414047956,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.5208333414047956,
+      "rewards/format_reward": 0.0,
+      "step": 213
+    },
+    {
+      "completion_length": 485.5625190734863,
+      "epoch": 2.2773333333333334,
+      "grad_norm": 0.10131195187568665,
+      "kl": 0.114349365234375,
+      "learning_rate": 4.6962642679073465e-07,
+      "loss": 0.0004,
+      "reward": 0.7500000260770321,
+      "reward_std": 0.3977733328938484,
+      "rewards/accuracy_reward": 0.7500000260770321,
+      "rewards/format_reward": 0.0,
+      "step": 214
+    },
+    {
+      "completion_length": 507.020845413208,
+      "epoch": 2.288,
+      "grad_norm": 0.17605960369110107,
+      "kl": 0.524017333984375,
+      "learning_rate": 4.560633949953017e-07,
+      "loss": 0.0169,
+      "reward": 0.6041666772216558,
+      "reward_std": 0.2996268533170223,
+      "rewards/accuracy_reward": 0.6041666772216558,
+      "rewards/format_reward": 0.0,
+      "step": 215
+    },
+    {
+      "completion_length": 582.1250228881836,
+      "epoch": 2.2986666666666666,
+      "grad_norm": 0.9037051796913147,
+      "kl": 0.363861083984375,
+      "learning_rate": 4.426639019818704e-07,
+      "loss": 0.0399,
+      "reward": 0.6041666865348816,
+      "reward_std": 0.3842546381056309,
+      "rewards/accuracy_reward": 0.6041666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 216
+    },
+    {
+      "completion_length": 526.4791831970215,
+      "epoch": 2.3093333333333335,
+      "grad_norm": 0.15531790256500244,
+      "kl": 0.1482086181640625,
+      "learning_rate": 4.2943004685948286e-07,
+      "loss": 0.0644,
+      "reward": 0.8125000111758709,
+      "reward_std": 0.2350771240890026,
+      "rewards/accuracy_reward": 0.8125000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 217
+    },
+    {
+      "completion_length": 489.8750190734863,
+      "epoch": 2.32,
+      "grad_norm": 0.2516200542449951,
+      "kl": 0.39373779296875,
+      "learning_rate": 4.163639027890363e-07,
+      "loss": 0.0626,
+      "reward": 0.6041666846722364,
+      "reward_std": 0.33713919296860695,
+      "rewards/accuracy_reward": 0.6041666846722364,
+      "rewards/format_reward": 0.0,
+      "step": 218
+    },
+    {
+      "completion_length": 463.02084732055664,
+      "epoch": 2.3306666666666667,
+      "grad_norm": 0.16724379360675812,
+      "kl": 0.218902587890625,
+      "learning_rate": 4.034675166585079e-07,
+      "loss": 0.0045,
+      "reward": 0.6458333469927311,
+      "reward_std": 0.309229951351881,
+      "rewards/accuracy_reward": 0.6458333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 219
+    },
+    {
+      "completion_length": 545.5000114440918,
+      "epoch": 2.3413333333333335,
+      "grad_norm": 0.11626483500003815,
+      "kl": 0.242828369140625,
+      "learning_rate": 3.907429087623007e-07,
+      "loss": 0.0289,
+      "reward": 0.39583334140479565,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.39583334140479565,
+      "rewards/format_reward": 0.0,
+      "step": 220
+    },
+    {
+      "completion_length": 505.4375228881836,
+      "epoch": 2.352,
+      "grad_norm": 0.6116917729377747,
+      "kl": 0.6152496337890625,
+      "learning_rate": 3.7819207248475174e-07,
+      "loss": 0.0516,
+      "reward": 0.47916667722165585,
+      "reward_std": 0.299626849591732,
+      "rewards/accuracy_reward": 0.47916667722165585,
+      "rewards/format_reward": 0.0,
+      "step": 221
+    },
+    {
+      "completion_length": 503.12501525878906,
+      "epoch": 2.3626666666666667,
+      "grad_norm": 0.17723074555397034,
+      "kl": 0.342559814453125,
+      "learning_rate": 3.6581697398785615e-07,
+      "loss": 0.0017,
+      "reward": 0.5833333414047956,
+      "reward_std": 0.16661180183291435,
+      "rewards/accuracy_reward": 0.5833333414047956,
+      "rewards/format_reward": 0.0,
+      "step": 222
+    },
+    {
+      "completion_length": 470.04168128967285,
+      "epoch": 2.3733333333333335,
+      "grad_norm": 0.1384473443031311,
+      "kl": 0.77313232421875,
+      "learning_rate": 3.5361955190325563e-07,
+      "loss": 0.0587,
+      "reward": 0.7500000223517418,
+      "reward_std": 0.4056045152246952,
+      "rewards/accuracy_reward": 0.7500000223517418,
+      "rewards/format_reward": 0.0,
+      "step": 223
+    },
+    {
+      "completion_length": 522.2500190734863,
+      "epoch": 2.384,
+      "grad_norm": 0.4199267625808716,
+      "kl": 0.51361083984375,
+      "learning_rate": 3.4160171702854286e-07,
+      "loss": 0.1006,
+      "reward": 0.5625000111758709,
+      "reward_std": 0.5185415111482143,
+      "rewards/accuracy_reward": 0.5625000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 224
+    },
+    {
+      "completion_length": 440.56250953674316,
+      "epoch": 2.3946666666666667,
+      "grad_norm": 0.23059825599193573,
+      "kl": 0.504302978515625,
+      "learning_rate": 3.297653520279226e-07,
+      "loss": 0.0634,
+      "reward": 0.7916666753590107,
+      "reward_std": 0.24859581515192986,
+      "rewards/accuracy_reward": 0.7916666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 225
+    },
+    {
+      "completion_length": 464.31251525878906,
+      "epoch": 2.405333333333333,
+      "grad_norm": 0.1300867199897766,
+      "kl": 0.534576416015625,
+      "learning_rate": 3.1811231113727974e-07,
+      "loss": 0.1312,
+      "reward": 0.7500000111758709,
+      "reward_std": 0.31314554437994957,
+      "rewards/accuracy_reward": 0.7500000111758709,
+      "rewards/format_reward": 0.0,
+      "step": 226
+    },
+    {
+      "completion_length": 450.50001525878906,
+      "epoch": 2.416,
+      "grad_norm": 0.30021610856056213,
+      "kl": 0.429931640625,
+      "learning_rate": 3.066444198737053e-07,
+      "loss": 0.0551,
+      "reward": 0.7083333507180214,
+      "reward_std": 0.2957112565636635,
+      "rewards/accuracy_reward": 0.7083333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 227
+    },
+    {
+      "completion_length": 461.3333435058594,
+      "epoch": 2.4266666666666667,
+      "grad_norm": 0.2617259919643402,
+      "kl": 0.419158935546875,
+      "learning_rate": 2.9536347474951645e-07,
+      "loss": 0.0583,
+      "reward": 0.6875000186264515,
+      "reward_std": 0.28219256922602654,
+      "rewards/accuracy_reward": 0.6875000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 228
+    },
+    {
+      "completion_length": 480.1666793823242,
+      "epoch": 2.437333333333333,
+      "grad_norm": 0.20455387234687805,
+      "kl": 0.4808349609375,
+      "learning_rate": 2.842712429908237e-07,
+      "loss": 0.0427,
+      "reward": 0.6875000279396772,
+      "reward_std": 0.4488043636083603,
+      "rewards/accuracy_reward": 0.6875000279396772,
+      "rewards/format_reward": 0.0,
+      "step": 229
+    },
+    {
+      "completion_length": 424.7083511352539,
+      "epoch": 2.448,
+      "grad_norm": 0.20351466536521912,
+      "kl": 0.1056365966796875,
+      "learning_rate": 2.733694622606824e-07,
+      "loss": 0.0177,
+      "reward": 0.7500000186264515,
+      "reward_std": 0.3506578840315342,
+      "rewards/accuracy_reward": 0.7500000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 230
+    },
+    {
+      "completion_length": 534.8541831970215,
+      "epoch": 2.458666666666667,
+      "grad_norm": 0.15089009702205658,
+      "kl": 0.5475311279296875,
+      "learning_rate": 2.626598403868801e-07,
+      "loss": 0.0696,
+      "reward": 0.541666679084301,
+      "reward_std": 0.32274864614009857,
+      "rewards/accuracy_reward": 0.541666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 231
+    },
+    {
+      "completion_length": 508.50001525878906,
+      "epoch": 2.469333333333333,
+      "grad_norm": 0.18934805691242218,
+      "kl": 0.216339111328125,
+      "learning_rate": 2.521440550943943e-07,
+      "loss": 0.0387,
+      "reward": 0.5416666772216558,
+      "reward_std": 0.3776952736079693,
+      "rewards/accuracy_reward": 0.5416666772216558,
+      "rewards/format_reward": 0.0,
+      "step": 232
+    },
+    {
+      "completion_length": 450.2291793823242,
+      "epoch": 2.48,
+      "grad_norm": 0.21853598952293396,
+      "kl": 1.242919921875,
+      "learning_rate": 2.418237537425666e-07,
+      "loss": 0.1013,
+      "reward": 0.6875000055879354,
+      "reward_std": 0.1801304966211319,
+      "rewards/accuracy_reward": 0.6875000055879354,
+      "rewards/format_reward": 0.0,
+      "step": 233
+    },
+    {
+      "completion_length": 359.4791831970215,
+      "epoch": 2.490666666666667,
+      "grad_norm": 0.8898715376853943,
+      "kl": 0.1141510009765625,
+      "learning_rate": 2.3170055306703512e-07,
+      "loss": -0.0016,
+      "reward": 0.8958333395421505,
+      "reward_std": 0.11558076739311218,
+      "rewards/accuracy_reward": 0.8958333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 234
+    },
+    {
+      "completion_length": 446.04168701171875,
+      "epoch": 2.501333333333333,
+      "grad_norm": 0.0969981923699379,
+      "kl": 0.379180908203125,
+      "learning_rate": 2.2177603892646277e-07,
+      "loss": 0.0337,
+      "reward": 0.8958333507180214,
+      "reward_std": 0.21764283627271652,
+      "rewards/accuracy_reward": 0.8958333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 235
+    },
+    {
+      "completion_length": 458.93750953674316,
+      "epoch": 2.512,
+      "grad_norm": 0.22483277320861816,
+      "kl": 0.6645660400390625,
+      "learning_rate": 2.1205176605410259e-07,
+      "loss": 0.1187,
+      "reward": 0.8125000204890966,
+      "reward_std": 0.25515517592430115,
+      "rewards/accuracy_reward": 0.8125000204890966,
+      "rewards/format_reward": 0.0,
+      "step": 236
+    },
+    {
+      "completion_length": 482.1666793823242,
+      "epoch": 2.522666666666667,
+      "grad_norm": 0.8316857218742371,
+      "kl": 0.3072509765625,
+      "learning_rate": 2.0252925781424153e-07,
+      "loss": -0.0087,
+      "reward": 0.6041666753590107,
+      "reward_std": 0.1801304966211319,
+      "rewards/accuracy_reward": 0.6041666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 237
+    },
+    {
+      "completion_length": 411.22917556762695,
+      "epoch": 2.533333333333333,
+      "grad_norm": 0.557529866695404,
+      "kl": 0.473541259765625,
+      "learning_rate": 1.932100059635551e-07,
+      "loss": 0.0201,
+      "reward": 0.7708333507180214,
+      "reward_std": 0.35457348078489304,
+      "rewards/accuracy_reward": 0.7708333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 238
+    },
+    {
+      "completion_length": 491.58334732055664,
+      "epoch": 2.544,
+      "grad_norm": 0.15882809460163116,
+      "kl": 0.38347625732421875,
+      "learning_rate": 1.8409547041741436e-07,
+      "loss": 0.0933,
+      "reward": 0.5416666716337204,
+      "reward_std": 0.33057983219623566,
+      "rewards/accuracy_reward": 0.5416666716337204,
+      "rewards/format_reward": 0.0,
+      "step": 239
+    },
+    {
+      "completion_length": 541.1875228881836,
+      "epoch": 2.554666666666667,
+      "grad_norm": 0.39457714557647705,
+      "kl": 0.536773681640625,
+      "learning_rate": 1.751870790211838e-07,
+      "loss": 0.0648,
+      "reward": 0.6250000055879354,
+      "reward_std": 0.24859581515192986,
+      "rewards/accuracy_reward": 0.6250000055879354,
+      "rewards/format_reward": 0.0,
+      "step": 240
+    },
+    {
+      "completion_length": 536.8333396911621,
+      "epoch": 2.5653333333333332,
+      "grad_norm": 0.09183813631534576,
+      "kl": 0.370025634765625,
+      "learning_rate": 1.6648622732653946e-07,
+      "loss": 0.0173,
+      "reward": 0.5833333507180214,
+      "reward_std": 0.2861081622540951,
+      "rewards/accuracy_reward": 0.5833333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 241
+    },
+    {
+      "completion_length": 519.7500228881836,
+      "epoch": 2.576,
+      "grad_norm": 0.18523578345775604,
+      "kl": 0.464324951171875,
+      "learning_rate": 1.5799427837284686e-07,
+      "loss": 0.0487,
+      "reward": 0.6666666753590107,
+      "reward_std": 0.2861081585288048,
+      "rewards/accuracy_reward": 0.6666666753590107,
+      "rewards/format_reward": 0.0,
+      "step": 242
+    },
+    {
+      "completion_length": 589.7083587646484,
+      "epoch": 2.586666666666667,
+      "grad_norm": 0.3257755935192108,
+      "kl": 3.33953857421875,
+      "learning_rate": 1.4971256247363534e-07,
+      "loss": 0.1904,
+      "reward": 0.6250000149011612,
+      "reward_std": 0.22155843302607536,
+      "rewards/accuracy_reward": 0.6250000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 243
+    },
+    {
+      "completion_length": 441.083345413208,
+      "epoch": 2.5973333333333333,
+      "grad_norm": 5.152331352233887,
+      "kl": 0.1742095947265625,
+      "learning_rate": 1.416423770081941e-07,
+      "loss": 0.0165,
+      "reward": 0.7916666716337204,
+      "reward_std": 0.11949635669589043,
+      "rewards/accuracy_reward": 0.7916666716337204,
+      "rewards/format_reward": 0.0,
+      "step": 244
+    },
+    {
+      "completion_length": 416.04167556762695,
+      "epoch": 2.608,
+      "grad_norm": 0.14691489934921265,
+      "kl": 0.259857177734375,
+      "learning_rate": 1.3378498621833236e-07,
+      "loss": 0.0833,
+      "reward": 0.8125000223517418,
+      "reward_std": 0.33713919669389725,
+      "rewards/accuracy_reward": 0.8125000223517418,
+      "rewards/format_reward": 0.0,
+      "step": 245
+    },
+    {
+      "completion_length": 506.7916793823242,
+      "epoch": 2.618666666666667,
+      "grad_norm": 0.1701485514640808,
+      "kl": 0.4063262939453125,
+      "learning_rate": 1.2614162101032601e-07,
+      "loss": 0.0419,
+      "reward": 0.6458333432674408,
+      "reward_std": 0.2446802258491516,
+      "rewards/accuracy_reward": 0.6458333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 246
+    },
+    {
+      "completion_length": 557.1875190734863,
+      "epoch": 2.6293333333333333,
+      "grad_norm": 0.17964352667331696,
+      "kl": 0.74188232421875,
+      "learning_rate": 1.1871347876209121e-07,
+      "loss": 0.1261,
+      "reward": 0.5833333358168602,
+      "reward_std": 0.30354244261980057,
+      "rewards/accuracy_reward": 0.5833333358168602,
+      "rewards/format_reward": 0.0,
+      "step": 247
+    },
+    {
+      "completion_length": 548.1250190734863,
+      "epoch": 2.64,
+      "grad_norm": 0.18035776913166046,
+      "kl": 0.470611572265625,
+      "learning_rate": 1.1150172313560741e-07,
+      "loss": 0.0889,
+      "reward": 0.6458333432674408,
+      "reward_std": 0.3816108666360378,
+      "rewards/accuracy_reward": 0.6458333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 248
+    },
+    {
+      "completion_length": 467.6875114440918,
+      "epoch": 2.6506666666666665,
+      "grad_norm": 0.247036412358284,
+      "kl": 0.37184906005859375,
+      "learning_rate": 1.0450748389462184e-07,
+      "loss": 0.0436,
+      "reward": 0.7708333507180214,
+      "reward_std": 0.35457348078489304,
+      "rewards/accuracy_reward": 0.7708333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 249
+    },
+    {
+      "completion_length": 473.64584732055664,
+      "epoch": 2.6613333333333333,
+      "grad_norm": 0.14761115610599518,
+      "kl": 0.6578369140625,
+      "learning_rate": 9.773185672766705e-08,
+      "loss": 0.0342,
+      "reward": 0.5208333507180214,
+      "reward_std": 0.35457347705960274,
+      "rewards/accuracy_reward": 0.5208333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 250
+    },
+    {
+      "completion_length": 497.58334732055664,
+      "epoch": 2.672,
+      "grad_norm": 0.16981832683086395,
+      "kl": 0.1938018798828125,
+      "learning_rate": 9.117590307641361e-08,
+      "loss": 0.0362,
+      "reward": 0.4791666716337204,
+      "reward_std": 0.18796168267726898,
+      "rewards/accuracy_reward": 0.4791666716337204,
+      "rewards/format_reward": 0.0,
+      "step": 251
+    },
+    {
+      "completion_length": 499.8541831970215,
+      "epoch": 2.6826666666666665,
+      "grad_norm": 0.17235523462295532,
+      "kl": 0.473602294921875,
+      "learning_rate": 8.484064996939023e-08,
+      "loss": 0.06,
+      "reward": 0.6666666828095913,
+      "reward_std": 0.2957112565636635,
+      "rewards/accuracy_reward": 0.6666666828095913,
+      "rewards/format_reward": 0.0,
+      "step": 252
+    },
+    {
+      "completion_length": 716.5625152587891,
+      "epoch": 2.6933333333333334,
+      "grad_norm": 0.1538318246603012,
+      "kl": 0.3689422607421875,
+      "learning_rate": 7.872708986109183e-08,
+      "loss": 0.0931,
+      "reward": 0.3333333395421505,
+      "reward_std": 0.2957112602889538,
+      "rewards/accuracy_reward": 0.3333333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 253
+    },
+    {
+      "completion_length": 484.77084732055664,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 0.25252780318260193,
+      "kl": 0.298797607421875,
+      "learning_rate": 7.283618047650808e-08,
+      "loss": 0.0304,
+      "reward": 0.5833333395421505,
+      "reward_std": 0.18404608592391014,
+      "rewards/accuracy_reward": 0.5833333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 254
+    },
+    {
+      "completion_length": 530.8958473205566,
+      "epoch": 2.7146666666666666,
+      "grad_norm": 0.11056818813085556,
+      "kl": 0.280242919921875,
+      "learning_rate": 6.716884466108869e-08,
+      "loss": 0.0474,
+      "reward": 0.43750000186264515,
+      "reward_std": 0.2525114119052887,
+      "rewards/accuracy_reward": 0.43750000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 255
+    },
+    {
+      "completion_length": 568.7708473205566,
+      "epoch": 2.7253333333333334,
+      "grad_norm": 0.12770210206508636,
+      "kl": 0.272369384765625,
+      "learning_rate": 6.17259702361741e-08,
+      "loss": 0.069,
+      "reward": 0.5833333395421505,
+      "reward_std": 0.33057983219623566,
+      "rewards/accuracy_reward": 0.5833333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 256
+    },
+    {
+      "completion_length": 431.60417556762695,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 0.06056330353021622,
+      "kl": 5.344085693359375,
+      "learning_rate": 5.65084098599144e-08,
+      "loss": 0.317,
+      "reward": 0.7500000149011612,
+      "reward_std": 0.22155843302607536,
+      "rewards/accuracy_reward": 0.7500000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 257
+    },
+    {
+      "completion_length": 472.1250114440918,
+      "epoch": 2.7466666666666666,
+      "grad_norm": 17.56707763671875,
+      "kl": 0.407470703125,
+      "learning_rate": 5.151698089369433e-08,
+      "loss": 0.0359,
+      "reward": 0.6666666828095913,
+      "reward_std": 0.2957112565636635,
+      "rewards/accuracy_reward": 0.6666666828095913,
+      "rewards/format_reward": 0.0,
+      "step": 258
+    },
+    {
+      "completion_length": 580.9375228881836,
+      "epoch": 2.7573333333333334,
+      "grad_norm": 0.22059756517410278,
+      "kl": 0.3290863037109375,
+      "learning_rate": 4.675246527408872e-08,
+      "loss": 0.0287,
+      "reward": 0.583333333954215,
+      "reward_std": 0.25642700493335724,
+      "rewards/accuracy_reward": 0.583333333954215,
+      "rewards/format_reward": 0.0,
+      "step": 259
+    },
+    {
+      "completion_length": 448.5000114440918,
+      "epoch": 2.768,
+      "grad_norm": 0.1809317022562027,
+      "kl": 0.4580078125,
+      "learning_rate": 4.221560939036828e-08,
+      "loss": 0.0359,
+      "reward": 0.645833358168602,
+      "reward_std": 0.41129202023148537,
+      "rewards/accuracy_reward": 0.645833358168602,
+      "rewards/format_reward": 0.0,
+      "step": 260
+    },
+    {
+      "completion_length": 427.87501525878906,
+      "epoch": 2.7786666666666666,
+      "grad_norm": 0.18283458054065704,
+      "kl": 0.5572967529296875,
+      "learning_rate": 3.790712396757317e-08,
+      "loss": 0.0282,
+      "reward": 0.6875000149011612,
+      "reward_std": 0.290023747831583,
+      "rewards/accuracy_reward": 0.6875000149011612,
+      "rewards/format_reward": 0.0,
+      "step": 261
+    },
+    {
+      "completion_length": 493.6875190734863,
+      "epoch": 2.7893333333333334,
+      "grad_norm": 0.37019890546798706,
+      "kl": 0.863800048828125,
+      "learning_rate": 3.382768395517355e-08,
+      "loss": 0.0368,
+      "reward": 0.5208333395421505,
+      "reward_std": 0.36417657881975174,
+      "rewards/accuracy_reward": 0.5208333395421505,
+      "rewards/format_reward": 0.0,
+      "step": 262
+    },
+    {
+      "completion_length": 530.0000152587891,
+      "epoch": 2.8,
+      "grad_norm": 1.078745722770691,
+      "kl": 0.419952392578125,
+      "learning_rate": 2.997792842133445e-08,
+      "loss": 0.0122,
+      "reward": 0.6250000186264515,
+      "reward_std": 0.3602609895169735,
+      "rewards/accuracy_reward": 0.6250000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 263
+    },
+    {
+      "completion_length": 497.25000762939453,
+      "epoch": 2.8106666666666666,
+      "grad_norm": 0.26311469078063965,
+      "kl": 0.536895751953125,
+      "learning_rate": 2.6358460452802768e-08,
+      "loss": 0.0022,
+      "reward": 0.6458333469927311,
+      "reward_std": 0.299626849591732,
+      "rewards/accuracy_reward": 0.6458333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 264
+    },
+    {
+      "completion_length": 554.2083587646484,
+      "epoch": 2.8213333333333335,
+      "grad_norm": 0.24127168953418732,
+      "kl": 0.6036834716796875,
+      "learning_rate": 2.296984706042993e-08,
+      "loss": 0.0309,
+      "reward": 0.7708333414047956,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.7708333414047956,
+      "rewards/format_reward": 0.0,
+      "step": 265
+    },
+    {
+      "completion_length": 468.75002098083496,
+      "epoch": 2.832,
+      "grad_norm": 0.18969446420669556,
+      "kl": 0.3581695556640625,
+      "learning_rate": 1.9812619090345984e-08,
+      "loss": 0.035,
+      "reward": 0.729166679084301,
+      "reward_std": 0.2350771203637123,
+      "rewards/accuracy_reward": 0.729166679084301,
+      "rewards/format_reward": 0.0,
+      "step": 266
+    },
+    {
+      "completion_length": 448.2708435058594,
+      "epoch": 2.8426666666666667,
+      "grad_norm": 0.14069421589374542,
+      "kl": 0.3592987060546875,
+      "learning_rate": 1.688727114079963e-08,
+      "loss": 0.0095,
+      "reward": 0.7500000186264515,
+      "reward_std": 0.3506578877568245,
+      "rewards/accuracy_reward": 0.7500000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 267
+    },
+    {
+      "completion_length": 512.2708511352539,
+      "epoch": 2.8533333333333335,
+      "grad_norm": 0.7901196479797363,
+      "kl": 1.0736083984375,
+      "learning_rate": 1.4194261484677007e-08,
+      "loss": 0.1151,
+      "reward": 0.5833333507180214,
+      "reward_std": 0.3881702274084091,
+      "rewards/accuracy_reward": 0.5833333507180214,
+      "rewards/format_reward": 0.0,
+      "step": 268
+    },
+    {
+      "completion_length": 621.3125114440918,
+      "epoch": 2.864,
+      "grad_norm": 0.17526622116565704,
+      "kl": 0.5525665283203125,
+      "learning_rate": 1.1734011997709626e-08,
+      "loss": 0.0434,
+      "reward": 0.43750000558793545,
+      "reward_std": 0.1705273911356926,
+      "rewards/accuracy_reward": 0.43750000558793545,
+      "rewards/format_reward": 0.0,
+      "step": 269
+    },
+    {
+      "completion_length": 518.1875114440918,
+      "epoch": 2.8746666666666667,
+      "grad_norm": 0.25227832794189453,
+      "kl": 0.6986083984375,
+      "learning_rate": 9.50690809238558e-09,
+      "loss": 0.0322,
+      "reward": 0.6250000186264515,
+      "reward_std": 0.3881702274084091,
+      "rewards/accuracy_reward": 0.6250000186264515,
+      "rewards/format_reward": 0.0,
+      "step": 270
+    },
+    {
+      "completion_length": 355.6041774749756,
+      "epoch": 2.8853333333333335,
+      "grad_norm": 0.29838475584983826,
+      "kl": 0.3779754638671875,
+      "learning_rate": 7.513298657572831e-09,
+      "loss": 0.0323,
+      "reward": 0.7291666828095913,
+      "reward_std": 0.25515517219901085,
+      "rewards/accuracy_reward": 0.7291666828095913,
+      "rewards/format_reward": 0.0,
+      "step": 271
+    },
+    {
+      "completion_length": 468.1875114440918,
+      "epoch": 2.896,
+      "grad_norm": 0.17359645664691925,
+      "kl": 0.265350341796875,
+      "learning_rate": 5.753496003862935e-09,
+      "loss": 0.0155,
+      "reward": 0.8541666865348816,
+      "reward_std": 0.28219256177544594,
+      "rewards/accuracy_reward": 0.8541666865348816,
+      "rewards/format_reward": 0.0,
+      "step": 272
+    },
+    {
+      "completion_length": 498.3750190734863,
+      "epoch": 2.9066666666666667,
+      "grad_norm": 0.38747698068618774,
+      "kl": 0.51171875,
+      "learning_rate": 4.227775814646173e-09,
+      "loss": 0.0413,
+      "reward": 0.7708333432674408,
+      "reward_std": 0.1801304966211319,
+      "rewards/accuracy_reward": 0.7708333432674408,
+      "rewards/format_reward": 0.0,
+      "step": 273
+    },
+    {
+      "completion_length": 426.833345413208,
+      "epoch": 2.9173333333333336,
+      "grad_norm": 0.1415153592824936,
+      "kl": 0.4593963623046875,
+      "learning_rate": 2.9363771029242705e-09,
+      "loss": 0.0572,
+      "reward": 0.6458333469927311,
+      "reward_std": 0.34674229100346565,
+      "rewards/accuracy_reward": 0.6458333469927311,
+      "rewards/format_reward": 0.0,
+      "step": 274
+    },
+    {
+      "completion_length": 620.479190826416,
+      "epoch": 2.928,
+      "grad_norm": 0.15173114836215973,
+      "kl": 4.2080078125,
+      "learning_rate": 1.8795021738670093e-09,
+      "loss": 0.2004,
+      "reward": 0.25000000931322575,
+      "reward_std": 0.23116152733564377,
+      "rewards/accuracy_reward": 0.25000000931322575,
+      "rewards/format_reward": 0.0,
+      "step": 275
+    },
+    {
+      "completion_length": 433.81250762939453,
+      "epoch": 2.9386666666666668,
+      "grad_norm": 11.822624206542969,
+      "kl": 1.3686065673828125,
+      "learning_rate": 1.0573165931209162e-09,
+      "loss": 0.0631,
+      "reward": 0.791666679084301,
+      "reward_std": 0.30354244261980057,
+      "rewards/accuracy_reward": 0.791666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 276
+    },
+    {
+      "completion_length": 503.7708511352539,
+      "epoch": 2.9493333333333336,
+      "grad_norm": 0.19741491973400116,
+      "kl": 0.20467376708984375,
+      "learning_rate": 4.699491608713413e-10,
+      "loss": 0.0096,
+      "reward": 0.791666679084301,
+      "reward_std": 0.23116153106093407,
+      "rewards/accuracy_reward": 0.791666679084301,
+      "rewards/format_reward": 0.0,
+      "step": 277
+    },
+    {
+      "completion_length": 473.9375114440918,
+      "epoch": 2.96,
+      "grad_norm": 0.16124479472637177,
+      "kl": 0.34735870361328125,
+      "learning_rate": 1.1749189166609943e-10,
+      "loss": 0.0333,
+      "reward": 0.8125000074505806,
+      "reward_std": 0.11558076366782188,
+      "rewards/accuracy_reward": 0.8125000074505806,
+      "rewards/format_reward": 0.0,
+      "step": 278
+    },
+    {
+      "completion_length": 476.12500953674316,
+      "epoch": 2.970666666666667,
+      "grad_norm": 0.2687714993953705,
+      "kl": 0.3852691650390625,
+      "learning_rate": 0.0,
+      "loss": 0.0885,
+      "reward": 0.8125000204890966,
+      "reward_std": 0.25515517219901085,
+      "rewards/accuracy_reward": 0.8125000204890966,
+      "rewards/format_reward": 0.0,
+      "step": 279
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 279,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "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": 2,
+  "trial_name": null,
+  "trial_params": null
+}