diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4638 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9980456026058632,
+  "eval_steps": 100,
+  "global_step": 383,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "completion_length": 870.0854415893555,
+      "epoch": 0.0026058631921824105,
+      "grad_norm": 0.16709676384925842,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": -0.0813,
+      "reward": 0.49043366871774197,
+      "reward_std": 0.34182496182620525,
+      "rewards/accuracy_reward": 0.30165815725922585,
+      "rewards/semantic_entropy_math_reward": 0.18877551332116127,
+      "step": 1
+    },
+    {
+      "completion_length": 879.3475608825684,
+      "epoch": 0.005211726384364821,
+      "grad_norm": 0.22682946920394897,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": -0.0873,
+      "reward": 0.4734876109287143,
+      "reward_std": 0.31084196362644434,
+      "rewards/accuracy_reward": 0.2780612204223871,
+      "rewards/semantic_entropy_math_reward": 0.19542638212442398,
+      "step": 2
+    },
+    {
+      "completion_length": 943.9164390563965,
+      "epoch": 0.007817589576547232,
+      "grad_norm": 0.23663833737373352,
+      "learning_rate": 1.5e-06,
+      "loss": -0.0952,
+      "reward": 0.4689322207123041,
+      "reward_std": 0.31457919254899025,
+      "rewards/accuracy_reward": 0.27678570710122585,
+      "rewards/semantic_entropy_math_reward": 0.1921465010382235,
+      "step": 3
+    },
+    {
+      "completion_length": 878.9853134155273,
+      "epoch": 0.010423452768729642,
+      "grad_norm": 0.3010266125202179,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": -0.0915,
+      "reward": 0.48915816470980644,
+      "reward_std": 0.3204598156735301,
+      "rewards/accuracy_reward": 0.29783162754029036,
+      "rewards/semantic_entropy_math_reward": 0.1913265334442258,
+      "step": 4
+    },
+    {
+      "completion_length": 861.0484504699707,
+      "epoch": 0.013029315960912053,
+      "grad_norm": 0.1940070539712906,
+      "learning_rate": 2.5e-06,
+      "loss": -0.0721,
+      "reward": 0.45945699140429497,
+      "reward_std": 0.31874877866357565,
+      "rewards/accuracy_reward": 0.2710459101945162,
+      "rewards/semantic_entropy_math_reward": 0.18841107934713364,
+      "step": 5
+    },
+    {
+      "completion_length": 937.8188629150391,
+      "epoch": 0.015635179153094463,
+      "grad_norm": 0.06981602311134338,
+      "learning_rate": 3e-06,
+      "loss": -0.096,
+      "reward": 0.5111151468008757,
+      "reward_std": 0.32009167689830065,
+      "rewards/accuracy_reward": 0.30102040339261293,
+      "rewards/semantic_entropy_math_reward": 0.21009475644677877,
+      "step": 6
+    },
+    {
+      "completion_length": 891.3507537841797,
+      "epoch": 0.018241042345276872,
+      "grad_norm": 0.1923670619726181,
+      "learning_rate": 3.5e-06,
+      "loss": -0.0935,
+      "reward": 0.5479227416217327,
+      "reward_std": 0.324985328130424,
+      "rewards/accuracy_reward": 0.34183673094958067,
+      "rewards/semantic_entropy_math_reward": 0.2060860088095069,
+      "step": 7
+    },
+    {
+      "completion_length": 889.3501167297363,
+      "epoch": 0.020846905537459284,
+      "grad_norm": 0.04959714412689209,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": -0.1075,
+      "reward": 0.5332543794065714,
+      "reward_std": 0.31880190316587687,
+      "rewards/accuracy_reward": 0.3227040730416775,
+      "rewards/semantic_entropy_math_reward": 0.21055030450224876,
+      "step": 8
+    },
+    {
+      "completion_length": 916.3003616333008,
+      "epoch": 0.023452768729641693,
+      "grad_norm": 0.06921309232711792,
+      "learning_rate": 4.5e-06,
+      "loss": -0.0841,
+      "reward": 0.5948433019220829,
+      "reward_std": 0.2975190822035074,
+      "rewards/accuracy_reward": 0.3590561132878065,
+      "rewards/semantic_entropy_math_reward": 0.23578718956559896,
+      "step": 9
+    },
+    {
+      "completion_length": 943.690673828125,
+      "epoch": 0.026058631921824105,
+      "grad_norm": 0.0358135886490345,
+      "learning_rate": 5e-06,
+      "loss": -0.0922,
+      "reward": 0.6337463427335024,
+      "reward_std": 0.2962046544998884,
+      "rewards/accuracy_reward": 0.38711734022945166,
+      "rewards/semantic_entropy_math_reward": 0.24662901367992163,
+      "step": 10
+    },
+    {
+      "completion_length": 931.7365913391113,
+      "epoch": 0.028664495114006514,
+      "grad_norm": 0.041781261563301086,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": -0.0894,
+      "reward": 0.6365706976503134,
+      "reward_std": 0.2825650768354535,
+      "rewards/accuracy_reward": 0.38647958263754845,
+      "rewards/semantic_entropy_math_reward": 0.25009111408144236,
+      "step": 11
+    },
+    {
+      "completion_length": 911.2397766113281,
+      "epoch": 0.031270358306188926,
+      "grad_norm": 0.06622398644685745,
+      "learning_rate": 6e-06,
+      "loss": -0.1081,
+      "reward": 0.7182033378630877,
+      "reward_std": 0.28334413934499025,
+      "rewards/accuracy_reward": 0.43239795323461294,
+      "rewards/semantic_entropy_math_reward": 0.28580539766699076,
+      "step": 12
+    },
+    {
+      "completion_length": 905.0082626342773,
+      "epoch": 0.033876221498371335,
+      "grad_norm": 0.02759523130953312,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": -0.0973,
+      "reward": 0.6452259533107281,
+      "reward_std": 0.27447106316685677,
+      "rewards/accuracy_reward": 0.38201529905200005,
+      "rewards/semantic_entropy_math_reward": 0.26321063842624426,
+      "step": 13
+    },
+    {
+      "completion_length": 881.1122283935547,
+      "epoch": 0.036482084690553744,
+      "grad_norm": 0.02610129304230213,
+      "learning_rate": 7e-06,
+      "loss": -0.0942,
+      "reward": 0.6983418371528387,
+      "reward_std": 0.27343872655183077,
+      "rewards/accuracy_reward": 0.4234693720936775,
+      "rewards/semantic_entropy_math_reward": 0.2748724529519677,
+      "step": 14
+    },
+    {
+      "completion_length": 883.7735748291016,
+      "epoch": 0.03908794788273615,
+      "grad_norm": 0.017909087240695953,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": -0.0933,
+      "reward": 0.6560677662491798,
+      "reward_std": 0.25880660582333803,
+      "rewards/accuracy_reward": 0.38265305012464523,
+      "rewards/semantic_entropy_math_reward": 0.2734147273004055,
+      "step": 15
+    },
+    {
+      "completion_length": 899.4240875244141,
+      "epoch": 0.04169381107491857,
+      "grad_norm": 0.0163668654859066,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": -0.0898,
+      "reward": 0.669825067743659,
+      "reward_std": 0.2818459044210613,
+      "rewards/accuracy_reward": 0.39732142351567745,
+      "rewards/semantic_entropy_math_reward": 0.27250364422798157,
+      "step": 16
+    },
+    {
+      "completion_length": 898.7939796447754,
+      "epoch": 0.04429967426710098,
+      "grad_norm": 0.009694581851363182,
+      "learning_rate": 8.5e-06,
+      "loss": -0.0763,
+      "reward": 0.6652696784585714,
+      "reward_std": 0.22073616785928607,
+      "rewards/accuracy_reward": 0.3781887646764517,
+      "rewards/semantic_entropy_math_reward": 0.2870809203013778,
+      "step": 17
+    },
+    {
+      "completion_length": 878.1428489685059,
+      "epoch": 0.046905537459283386,
+      "grad_norm": 0.010209346190094948,
+      "learning_rate": 9e-06,
+      "loss": -0.0777,
+      "reward": 0.7228498347103596,
+      "reward_std": 0.24326845351606607,
+      "rewards/accuracy_reward": 0.4311224389821291,
+      "rewards/semantic_entropy_math_reward": 0.29172741062939167,
+      "step": 18
+    },
+    {
+      "completion_length": 896.1135025024414,
+      "epoch": 0.049511400651465795,
+      "grad_norm": 0.013440839014947414,
+      "learning_rate": 9.5e-06,
+      "loss": -0.0752,
+      "reward": 0.6634475197643042,
+      "reward_std": 0.1989616984501481,
+      "rewards/accuracy_reward": 0.35905611515045166,
+      "rewards/semantic_entropy_math_reward": 0.3043913906440139,
+      "step": 19
+    },
+    {
+      "completion_length": 839.2117195129395,
+      "epoch": 0.05211726384364821,
+      "grad_norm": 0.017011163756251335,
+      "learning_rate": 1e-05,
+      "loss": -0.0586,
+      "reward": 0.7821610607206821,
+      "reward_std": 0.21515017794445157,
+      "rewards/accuracy_reward": 0.46938773803412914,
+      "rewards/semantic_entropy_math_reward": 0.3127733152359724,
+      "step": 20
+    },
+    {
+      "completion_length": 887.4808578491211,
+      "epoch": 0.05472312703583062,
+      "grad_norm": 0.009101642295718193,
+      "learning_rate": 9.999812749151968e-06,
+      "loss": -0.0595,
+      "reward": 0.6434037834405899,
+      "reward_std": 0.1854215101338923,
+      "rewards/accuracy_reward": 0.35331631638109684,
+      "rewards/semantic_entropy_math_reward": 0.2900874586775899,
+      "step": 21
+    },
+    {
+      "completion_length": 842.1600532531738,
+      "epoch": 0.05732899022801303,
+      "grad_norm": 0.011061098426580429,
+      "learning_rate": 9.99925101063302e-06,
+      "loss": -0.0568,
+      "reward": 0.722485426813364,
+      "reward_std": 0.21386212622746825,
+      "rewards/accuracy_reward": 0.40688774874433875,
+      "rewards/semantic_entropy_math_reward": 0.31559765338897705,
+      "step": 22
+    },
+    {
+      "completion_length": 872.0918083190918,
+      "epoch": 0.05993485342019544,
+      "grad_norm": 0.009069804102182388,
+      "learning_rate": 9.998314826517564e-06,
+      "loss": -0.0593,
+      "reward": 0.7820699661970139,
+      "reward_std": 0.21369126765057445,
+      "rewards/accuracy_reward": 0.4738520346581936,
+      "rewards/semantic_entropy_math_reward": 0.30821793153882027,
+      "step": 23
+    },
+    {
+      "completion_length": 790.3061065673828,
+      "epoch": 0.06254071661237785,
+      "grad_norm": 0.007478907238692045,
+      "learning_rate": 9.997004266926105e-06,
+      "loss": -0.0458,
+      "reward": 0.7919096164405346,
+      "reward_std": 0.1941184471361339,
+      "rewards/accuracy_reward": 0.45918366592377424,
+      "rewards/semantic_entropy_math_reward": 0.3327259514480829,
+      "step": 24
+    },
+    {
+      "completion_length": 849.5184745788574,
+      "epoch": 0.06514657980456026,
+      "grad_norm": 0.012659430503845215,
+      "learning_rate": 9.995319430020004e-06,
+      "loss": -0.0484,
+      "reward": 0.7839832194149494,
+      "reward_std": 0.1770358441863209,
+      "rewards/accuracy_reward": 0.46301019191741943,
+      "rewards/semantic_entropy_math_reward": 0.32097303587943316,
+      "step": 25
+    },
+    {
+      "completion_length": 840.4846801757812,
+      "epoch": 0.06775244299674267,
+      "grad_norm": 0.010360866785049438,
+      "learning_rate": 9.993260441994116e-06,
+      "loss": -0.0613,
+      "reward": 0.8100400976836681,
+      "reward_std": 0.1907408689148724,
+      "rewards/accuracy_reward": 0.484693868085742,
+      "rewards/semantic_entropy_math_reward": 0.32534620352089405,
+      "step": 26
+    },
+    {
+      "completion_length": 820.5082778930664,
+      "epoch": 0.07035830618892508,
+      "grad_norm": 0.01019768975675106,
+      "learning_rate": 9.990827457067342e-06,
+      "loss": -0.0412,
+      "reward": 0.8268039189279079,
+      "reward_std": 0.15273668291047215,
+      "rewards/accuracy_reward": 0.49362243339419365,
+      "rewards/semantic_entropy_math_reward": 0.33318148739635944,
+      "step": 27
+    },
+    {
+      "completion_length": 836.6026649475098,
+      "epoch": 0.07296416938110749,
+      "grad_norm": 0.009926711209118366,
+      "learning_rate": 9.988020657471078e-06,
+      "loss": -0.0391,
+      "reward": 0.728134099394083,
+      "reward_std": 0.15909703564830124,
+      "rewards/accuracy_reward": 0.40688774175941944,
+      "rewards/semantic_entropy_math_reward": 0.32124634832143784,
+      "step": 28
+    },
+    {
+      "completion_length": 793.8048286437988,
+      "epoch": 0.0755700325732899,
+      "grad_norm": 0.020279383286833763,
+      "learning_rate": 9.984840253435569e-06,
+      "loss": -0.0405,
+      "reward": 0.8198797274380922,
+      "reward_std": 0.18700192403048277,
+      "rewards/accuracy_reward": 0.4929846879094839,
+      "rewards/semantic_entropy_math_reward": 0.32689503859728575,
+      "step": 29
+    },
+    {
+      "completion_length": 744.598201751709,
+      "epoch": 0.0781758957654723,
+      "grad_norm": 0.011854313313961029,
+      "learning_rate": 9.98128648317415e-06,
+      "loss": -0.0396,
+      "reward": 0.850127536803484,
+      "reward_std": 0.16003184486180544,
+      "rewards/accuracy_reward": 0.48915815353393555,
+      "rewards/semantic_entropy_math_reward": 0.3609693832695484,
+      "step": 30
+    },
+    {
+      "completion_length": 764.6517677307129,
+      "epoch": 0.08078175895765473,
+      "grad_norm": 0.007330858614295721,
+      "learning_rate": 9.977359612865424e-06,
+      "loss": -0.0272,
+      "reward": 0.9031523056328297,
+      "reward_std": 0.1402361944783479,
+      "rewards/accuracy_reward": 0.5363520253449678,
+      "rewards/semantic_entropy_math_reward": 0.366800295189023,
+      "step": 31
+    },
+    {
+      "completion_length": 734.7104434967041,
+      "epoch": 0.08338762214983714,
+      "grad_norm": 0.016964299604296684,
+      "learning_rate": 9.973059936633308e-06,
+      "loss": -0.0311,
+      "reward": 0.8176020421087742,
+      "reward_std": 0.1600553193129599,
+      "rewards/accuracy_reward": 0.46619896963238716,
+      "rewards/semantic_entropy_math_reward": 0.35140305012464523,
+      "step": 32
+    },
+    {
+      "completion_length": 766.8622283935547,
+      "epoch": 0.08599348534201955,
+      "grad_norm": 0.013955877162516117,
+      "learning_rate": 9.968387776525009e-06,
+      "loss": -0.0309,
+      "reward": 0.8125910945236683,
+      "reward_std": 0.1700099449371919,
+      "rewards/accuracy_reward": 0.46874998696148396,
+      "rewards/semantic_entropy_math_reward": 0.34384110383689404,
+      "step": 33
+    },
+    {
+      "completion_length": 828.6734504699707,
+      "epoch": 0.08859934853420195,
+      "grad_norm": 0.015905598178505898,
+      "learning_rate": 9.963343482486907e-06,
+      "loss": -0.0339,
+      "reward": 0.7439868859946728,
+      "reward_std": 0.1670152919832617,
+      "rewards/accuracy_reward": 0.4119897894561291,
+      "rewards/semantic_entropy_math_reward": 0.33199707977473736,
+      "step": 34
+    },
+    {
+      "completion_length": 804.4808464050293,
+      "epoch": 0.09120521172638436,
+      "grad_norm": 0.011662989854812622,
+      "learning_rate": 9.957927432338332e-06,
+      "loss": -0.0364,
+      "reward": 0.7976494319736958,
+      "reward_std": 0.16072418447583914,
+      "rewards/accuracy_reward": 0.4547193758189678,
+      "rewards/semantic_entropy_math_reward": 0.34293002262711525,
+      "step": 35
+    },
+    {
+      "completion_length": 816.8692359924316,
+      "epoch": 0.09381107491856677,
+      "grad_norm": 0.014741488732397556,
+      "learning_rate": 9.952140031743282e-06,
+      "loss": -0.0329,
+      "reward": 0.8408345356583595,
+      "reward_std": 0.15276295854710042,
+      "rewards/accuracy_reward": 0.48915815353393555,
+      "rewards/semantic_entropy_math_reward": 0.3516763783991337,
+      "step": 36
+    },
+    {
+      "completion_length": 757.6179695129395,
+      "epoch": 0.09641693811074918,
+      "grad_norm": 0.021201658993959427,
+      "learning_rate": 9.945981714180021e-06,
+      "loss": -0.0365,
+      "reward": 0.8421100489795208,
+      "reward_std": 0.1715679601766169,
+      "rewards/accuracy_reward": 0.5038265194743872,
+      "rewards/semantic_entropy_math_reward": 0.33828352577984333,
+      "step": 37
+    },
+    {
+      "completion_length": 770.365421295166,
+      "epoch": 0.09902280130293159,
+      "grad_norm": 0.03967665135860443,
+      "learning_rate": 9.939452940908627e-06,
+      "loss": -0.0316,
+      "reward": 0.8511297404766083,
+      "reward_std": 0.14679769705981016,
+      "rewards/accuracy_reward": 0.498724477365613,
+      "rewards/semantic_entropy_math_reward": 0.35240524634718895,
+      "step": 38
+    },
+    {
+      "completion_length": 753.9119720458984,
+      "epoch": 0.10162866449511401,
+      "grad_norm": 0.02615044265985489,
+      "learning_rate": 9.932554200936428e-06,
+      "loss": -0.0365,
+      "reward": 0.8091289959847927,
+      "reward_std": 0.17394380597397685,
+      "rewards/accuracy_reward": 0.46173468325287104,
+      "rewards/semantic_entropy_math_reward": 0.3473943155258894,
+      "step": 39
+    },
+    {
+      "completion_length": 727.7244834899902,
+      "epoch": 0.10423452768729642,
+      "grad_norm": 0.20424392819404602,
+      "learning_rate": 9.925286010981394e-06,
+      "loss": -0.0368,
+      "reward": 0.8333636894822121,
+      "reward_std": 0.1578838722780347,
+      "rewards/accuracy_reward": 0.48214284889400005,
+      "rewards/semantic_entropy_math_reward": 0.3512208443135023,
+      "step": 40
+    },
+    {
+      "completion_length": 773.675365447998,
+      "epoch": 0.10684039087947883,
+      "grad_norm": 0.004368220455944538,
+      "learning_rate": 9.917648915433413e-06,
+      "loss": -0.0389,
+      "reward": 0.8713556751608849,
+      "reward_std": 0.18799859285354614,
+      "rewards/accuracy_reward": 0.5153061114251614,
+      "rewards/semantic_entropy_math_reward": 0.35604955814778805,
+      "step": 41
+    },
+    {
+      "completion_length": 756.2442359924316,
+      "epoch": 0.10944625407166124,
+      "grad_norm": 0.004377900157123804,
+      "learning_rate": 9.909643486313533e-06,
+      "loss": -0.0419,
+      "reward": 0.8909438773989677,
+      "reward_std": 0.15308711864054203,
+      "rewards/accuracy_reward": 0.5248724389821291,
+      "rewards/semantic_entropy_math_reward": 0.36607141979038715,
+      "step": 42
+    },
+    {
+      "completion_length": 740.527400970459,
+      "epoch": 0.11205211726384365,
+      "grad_norm": 0.004278510343283415,
+      "learning_rate": 9.901270323231114e-06,
+      "loss": -0.0376,
+      "reward": 0.8292638286948204,
+      "reward_std": 0.16438758978620172,
+      "rewards/accuracy_reward": 0.483418358489871,
+      "rewards/semantic_entropy_math_reward": 0.3458454851061106,
+      "step": 43
+    },
+    {
+      "completion_length": 665.3762626647949,
+      "epoch": 0.11465798045602606,
+      "grad_norm": 0.004815321881324053,
+      "learning_rate": 9.892530053338909e-06,
+      "loss": -0.0286,
+      "reward": 0.9122630879282951,
+      "reward_std": 0.16950253094546497,
+      "rewards/accuracy_reward": 0.5433673392981291,
+      "rewards/semantic_entropy_math_reward": 0.36889576353132725,
+      "step": 44
+    },
+    {
+      "completion_length": 692.8494720458984,
+      "epoch": 0.11726384364820847,
+      "grad_norm": 0.010744983330368996,
+      "learning_rate": 9.883423331286096e-06,
+      "loss": -0.025,
+      "reward": 0.875364426523447,
+      "reward_std": 0.15541742404457182,
+      "rewards/accuracy_reward": 0.5051020309329033,
+      "rewards/semantic_entropy_math_reward": 0.370262386277318,
+      "step": 45
+    },
+    {
+      "completion_length": 702.8003749847412,
+      "epoch": 0.11986970684039087,
+      "grad_norm": 0.006386768072843552,
+      "learning_rate": 9.873950839169248e-06,
+      "loss": -0.0375,
+      "reward": 0.8963192403316498,
+      "reward_std": 0.17361515946686268,
+      "rewards/accuracy_reward": 0.5344387628138065,
+      "rewards/semantic_entropy_math_reward": 0.3618804607540369,
+      "step": 46
+    },
+    {
+      "completion_length": 747.270393371582,
+      "epoch": 0.12247557003257328,
+      "grad_norm": 0.006052403245121241,
+      "learning_rate": 9.864113286481237e-06,
+      "loss": -0.0302,
+      "reward": 0.8596027493476868,
+      "reward_std": 0.16451329877600074,
+      "rewards/accuracy_reward": 0.5146683547645807,
+      "rewards/semantic_entropy_math_reward": 0.3449344038963318,
+      "step": 47
+    },
+    {
+      "completion_length": 677.6504974365234,
+      "epoch": 0.1250814332247557,
+      "grad_norm": 0.010900772176682949,
+      "learning_rate": 9.853911410058097e-06,
+      "loss": -0.026,
+      "reward": 0.8704445790499449,
+      "reward_std": 0.14085088530555367,
+      "rewards/accuracy_reward": 0.5165816182270646,
+      "rewards/semantic_entropy_math_reward": 0.3538629673421383,
+      "step": 48
+    },
+    {
+      "completion_length": 713.0108261108398,
+      "epoch": 0.1276872964169381,
+      "grad_norm": 0.004999789409339428,
+      "learning_rate": 9.843345974023833e-06,
+      "loss": -0.0185,
+      "reward": 0.8516763634979725,
+      "reward_std": 0.14363743993453681,
+      "rewards/accuracy_reward": 0.4885203968733549,
+      "rewards/semantic_entropy_math_reward": 0.3631559703499079,
+      "step": 49
+    },
+    {
+      "completion_length": 696.6517677307129,
+      "epoch": 0.13029315960912052,
+      "grad_norm": 0.011122237890958786,
+      "learning_rate": 9.832417769733185e-06,
+      "loss": -0.0379,
+      "reward": 0.8595116399228573,
+      "reward_std": 0.16439938032999635,
+      "rewards/accuracy_reward": 0.506377536803484,
+      "rewards/semantic_entropy_math_reward": 0.35313410498201847,
+      "step": 50
+    },
+    {
+      "completion_length": 672.8233242034912,
+      "epoch": 0.13289902280130292,
+      "grad_norm": 0.006625417619943619,
+      "learning_rate": 9.821127615712365e-06,
+      "loss": -0.0212,
+      "reward": 0.9153607748448849,
+      "reward_std": 0.17063219111878425,
+      "rewards/accuracy_reward": 0.5542091727256775,
+      "rewards/semantic_entropy_math_reward": 0.3611515983939171,
+      "step": 51
+    },
+    {
+      "completion_length": 678.4897918701172,
+      "epoch": 0.13550488599348534,
+      "grad_norm": 0.006391481496393681,
+      "learning_rate": 9.809476357597738e-06,
+      "loss": -0.0301,
+      "reward": 0.8757288567721844,
+      "reward_std": 0.14586452580988407,
+      "rewards/accuracy_reward": 0.5012754984200001,
+      "rewards/semantic_entropy_math_reward": 0.37445334158837795,
+      "step": 52
+    },
+    {
+      "completion_length": 637.1996021270752,
+      "epoch": 0.13811074918566776,
+      "grad_norm": 0.004039392340928316,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": -0.0229,
+      "reward": 0.939504336565733,
+      "reward_std": 0.12869648274499923,
+      "rewards/accuracy_reward": 0.5529336631298065,
+      "rewards/semantic_entropy_math_reward": 0.38657068461179733,
+      "step": 53
+    },
+    {
+      "completion_length": 669.5726890563965,
+      "epoch": 0.14071661237785016,
+      "grad_norm": 0.004615962039679289,
+      "learning_rate": 9.785094046801256e-06,
+      "loss": -0.0263,
+      "reward": 0.9114431440830231,
+      "reward_std": 0.14767074608244002,
+      "rewards/accuracy_reward": 0.5293367225676775,
+      "rewards/semantic_entropy_math_reward": 0.3821064028888941,
+      "step": 54
+    },
+    {
+      "completion_length": 694.8207740783691,
+      "epoch": 0.14332247557003258,
+      "grad_norm": 0.0039047168102115393,
+      "learning_rate": 9.77236482036275e-06,
+      "loss": -0.0294,
+      "reward": 0.8075801469385624,
+      "reward_std": 0.11974696815013885,
+      "rewards/accuracy_reward": 0.4477040711790323,
+      "rewards/semantic_entropy_math_reward": 0.35987609066069126,
+      "step": 55
+    },
+    {
+      "completion_length": 667.4215412139893,
+      "epoch": 0.14592833876221498,
+      "grad_norm": 0.0031674006022512913,
+      "learning_rate": 9.759278142180348e-06,
+      "loss": -0.034,
+      "reward": 0.9129008483141661,
+      "reward_std": 0.1429015859030187,
+      "rewards/accuracy_reward": 0.5325254974886775,
+      "rewards/semantic_entropy_math_reward": 0.3803753647953272,
+      "step": 56
+    },
+    {
+      "completion_length": 667.6166915893555,
+      "epoch": 0.1485342019543974,
+      "grad_norm": 0.00376137369312346,
+      "learning_rate": 9.745834992450688e-06,
+      "loss": -0.027,
+      "reward": 0.8073068428784609,
+      "reward_std": 0.1307279309257865,
+      "rewards/accuracy_reward": 0.4521683547645807,
+      "rewards/semantic_entropy_math_reward": 0.35513848531991243,
+      "step": 57
+    },
+    {
+      "completion_length": 629.4394016265869,
+      "epoch": 0.1511400651465798,
+      "grad_norm": 0.00376802496612072,
+      "learning_rate": 9.732036378070243e-06,
+      "loss": -0.0146,
+      "reward": 0.8919460512697697,
+      "reward_std": 0.14173119352199137,
+      "rewards/accuracy_reward": 0.5223214142024517,
+      "rewards/semantic_entropy_math_reward": 0.36962463706731796,
+      "step": 58
+    },
+    {
+      "completion_length": 679.9802112579346,
+      "epoch": 0.15374592833876222,
+      "grad_norm": 0.00410770159214735,
+      "learning_rate": 9.717883332559911e-06,
+      "loss": -0.0235,
+      "reward": 0.8637937270104885,
+      "reward_std": 0.13675629626959562,
+      "rewards/accuracy_reward": 0.49298468604683876,
+      "rewards/semantic_entropy_math_reward": 0.3708090353757143,
+      "step": 59
+    },
+    {
+      "completion_length": 613.1179676055908,
+      "epoch": 0.1563517915309446,
+      "grad_norm": 0.0037613979075104,
+      "learning_rate": 9.703376915987601e-06,
+      "loss": -0.0233,
+      "reward": 0.8473032042384148,
+      "reward_std": 0.1510353204794228,
+      "rewards/accuracy_reward": 0.47130101174116135,
+      "rewards/semantic_entropy_math_reward": 0.3760021850466728,
+      "step": 60
+    },
+    {
+      "completion_length": 659.0210304260254,
+      "epoch": 0.15895765472312703,
+      "grad_norm": 0.0041074734181165695,
+      "learning_rate": 9.688518214888836e-06,
+      "loss": -0.0397,
+      "reward": 0.8512208238244057,
+      "reward_std": 0.13073399825952947,
+      "rewards/accuracy_reward": 0.4853316228836775,
+      "rewards/semantic_entropy_math_reward": 0.3658892083913088,
+      "step": 61
+    },
+    {
+      "completion_length": 647.3169574737549,
+      "epoch": 0.16156351791530946,
+      "grad_norm": 0.004084247630089521,
+      "learning_rate": 9.673308342185366e-06,
+      "loss": -0.0256,
+      "reward": 0.8939504288136959,
+      "reward_std": 0.1518772984854877,
+      "rewards/accuracy_reward": 0.5274234591051936,
+      "rewards/semantic_entropy_math_reward": 0.36652695946395397,
+      "step": 62
+    },
+    {
+      "completion_length": 617.5650367736816,
+      "epoch": 0.16416938110749185,
+      "grad_norm": 0.0045938920229673386,
+      "learning_rate": 9.657748437101819e-06,
+      "loss": -0.0281,
+      "reward": 0.9265670254826546,
+      "reward_std": 0.14439171738922596,
+      "rewards/accuracy_reward": 0.5573979429900646,
+      "rewards/semantic_entropy_math_reward": 0.36916909366846085,
+      "step": 63
+    },
+    {
+      "completion_length": 662.421537399292,
+      "epoch": 0.16677524429967427,
+      "grad_norm": 0.004075649194419384,
+      "learning_rate": 9.641839665080363e-06,
+      "loss": -0.0301,
+      "reward": 0.9174562618136406,
+      "reward_std": 0.14708451775368303,
+      "rewards/accuracy_reward": 0.5350765231996775,
+      "rewards/semantic_entropy_math_reward": 0.3823797293007374,
+      "step": 64
+    },
+    {
+      "completion_length": 679.7850646972656,
+      "epoch": 0.16938110749185667,
+      "grad_norm": 0.003912299871444702,
+      "learning_rate": 9.625583217693419e-06,
+      "loss": -0.0303,
+      "reward": 0.9068877436220646,
+      "reward_std": 0.13392312987707555,
+      "rewards/accuracy_reward": 0.5363520327955484,
+      "rewards/semantic_entropy_math_reward": 0.37053569965064526,
+      "step": 65
+    },
+    {
+      "completion_length": 648.4598083496094,
+      "epoch": 0.1719869706840391,
+      "grad_norm": 0.00416301004588604,
+      "learning_rate": 9.60898031255441e-06,
+      "loss": -0.0225,
+      "reward": 0.9016034826636314,
+      "reward_std": 0.14559987792745233,
+      "rewards/accuracy_reward": 0.5089285578578711,
+      "rewards/semantic_entropy_math_reward": 0.3926749173551798,
+      "step": 66
+    },
+    {
+      "completion_length": 573.0261383056641,
+      "epoch": 0.1745928338762215,
+      "grad_norm": 0.006119817961007357,
+      "learning_rate": 9.592032193226564e-06,
+      "loss": -0.015,
+      "reward": 0.9476129561662674,
+      "reward_std": 0.12818793952465057,
+      "rewards/accuracy_reward": 0.5656887628138065,
+      "rewards/semantic_entropy_math_reward": 0.38192419335246086,
+      "step": 67
+    },
+    {
+      "completion_length": 651.288890838623,
+      "epoch": 0.1771986970684039,
+      "grad_norm": 0.00714410375803709,
+      "learning_rate": 9.574740129129767e-06,
+      "loss": -0.0264,
+      "reward": 0.8074890486896038,
+      "reward_std": 0.12236620881594718,
+      "rewards/accuracy_reward": 0.4457908058539033,
+      "rewards/semantic_entropy_math_reward": 0.36169825680553913,
+      "step": 68
+    },
+    {
+      "completion_length": 643.5050830841064,
+      "epoch": 0.17980456026058633,
+      "grad_norm": 0.006653693970292807,
+      "learning_rate": 9.557105415445485e-06,
+      "loss": -0.0202,
+      "reward": 0.9240160100162029,
+      "reward_std": 0.1365425349213183,
+      "rewards/accuracy_reward": 0.5446428470313549,
+      "rewards/semantic_entropy_math_reward": 0.3793731667101383,
+      "step": 69
+    },
+    {
+      "completion_length": 582.908145904541,
+      "epoch": 0.18241042345276873,
+      "grad_norm": 0.01532233040779829,
+      "learning_rate": 9.539129373019755e-06,
+      "loss": -0.018,
+      "reward": 0.937226664274931,
+      "reward_std": 0.13966482668183744,
+      "rewards/accuracy_reward": 0.5548469256609678,
+      "rewards/semantic_entropy_math_reward": 0.382379736751318,
+      "step": 70
+    },
+    {
+      "completion_length": 583.0146541595459,
+      "epoch": 0.18501628664495115,
+      "grad_norm": 0.004541024565696716,
+      "learning_rate": 9.520813348264252e-06,
+      "loss": -0.0259,
+      "reward": 0.872904509305954,
+      "reward_std": 0.10791157651692629,
+      "rewards/accuracy_reward": 0.47704081051051617,
+      "rewards/semantic_entropy_math_reward": 0.3958636950701475,
+      "step": 71
+    },
+    {
+      "completion_length": 658.2251129150391,
+      "epoch": 0.18762214983713354,
+      "grad_norm": 0.00619466882199049,
+      "learning_rate": 9.502158713055444e-06,
+      "loss": -0.0395,
+      "reward": 0.8880284130573273,
+      "reward_std": 0.1113471242133528,
+      "rewards/accuracy_reward": 0.5089285634458065,
+      "rewards/semantic_entropy_math_reward": 0.37909984961152077,
+      "step": 72
+    },
+    {
+      "completion_length": 624.5025367736816,
+      "epoch": 0.19022801302931597,
+      "grad_norm": 0.0047431401908397675,
+      "learning_rate": 9.483166864631837e-06,
+      "loss": -0.0331,
+      "reward": 0.8842018693685532,
+      "reward_std": 0.10373708885163069,
+      "rewards/accuracy_reward": 0.5019132532179356,
+      "rewards/semantic_entropy_math_reward": 0.38228863291442394,
+      "step": 73
+    },
+    {
+      "completion_length": 589.4891395568848,
+      "epoch": 0.19283387622149836,
+      "grad_norm": 0.004772627260535955,
+      "learning_rate": 9.46383922548932e-06,
+      "loss": -0.0249,
+      "reward": 0.9389577209949493,
+      "reward_std": 0.11472503608092666,
+      "rewards/accuracy_reward": 0.545280609279871,
+      "rewards/semantic_entropy_math_reward": 0.39367711916565895,
+      "step": 74
+    },
+    {
+      "completion_length": 620.705982208252,
+      "epoch": 0.19543973941368079,
+      "grad_norm": 0.0062498371116817,
+      "learning_rate": 9.444177243274619e-06,
+      "loss": -0.0257,
+      "reward": 0.9043367132544518,
+      "reward_std": 0.14091078890487552,
+      "rewards/accuracy_reward": 0.5248724389821291,
+      "rewards/semantic_entropy_math_reward": 0.37946427799761295,
+      "step": 75
+    },
+    {
+      "completion_length": 648.9623603820801,
+      "epoch": 0.19804560260586318,
+      "grad_norm": 0.004575234837830067,
+      "learning_rate": 9.424182390676872e-06,
+      "loss": -0.0282,
+      "reward": 0.8821063973009586,
+      "reward_std": 0.11956909950822592,
+      "rewards/accuracy_reward": 0.5012755002826452,
+      "rewards/semantic_entropy_math_reward": 0.3808308970183134,
+      "step": 76
+    },
+    {
+      "completion_length": 591.5235843658447,
+      "epoch": 0.2006514657980456,
+      "grad_norm": 0.005398905370384455,
+      "learning_rate": 9.403856165317322e-06,
+      "loss": -0.0297,
+      "reward": 0.949161771684885,
+      "reward_std": 0.12031427887268364,
+      "rewards/accuracy_reward": 0.5548469219356775,
+      "rewards/semantic_entropy_math_reward": 0.3943148721009493,
+      "step": 77
+    },
+    {
+      "completion_length": 634.9636363983154,
+      "epoch": 0.20325732899022803,
+      "grad_norm": 0.0076393503695726395,
+      "learning_rate": 9.383200089637143e-06,
+      "loss": -0.0267,
+      "reward": 0.8719934187829494,
+      "reward_std": 0.12535155925434083,
+      "rewards/accuracy_reward": 0.49489794485270977,
+      "rewards/semantic_entropy_math_reward": 0.3770954702049494,
+      "step": 78
+    },
+    {
+      "completion_length": 594.6001148223877,
+      "epoch": 0.20586319218241042,
+      "grad_norm": 0.0050405478104949,
+      "learning_rate": 9.362215710783411e-06,
+      "loss": -0.0178,
+      "reward": 0.9342201091349125,
+      "reward_std": 0.1078009803313762,
+      "rewards/accuracy_reward": 0.5452806055545807,
+      "rewards/semantic_entropy_math_reward": 0.38893949054181576,
+      "step": 79
+    },
+    {
+      "completion_length": 616.5573806762695,
+      "epoch": 0.20846905537459284,
+      "grad_norm": 0.00523959519341588,
+      "learning_rate": 9.34090460049322e-06,
+      "loss": -0.0283,
+      "reward": 0.9657434336841106,
+      "reward_std": 0.1119958006311208,
+      "rewards/accuracy_reward": 0.5758928414434195,
+      "rewards/semantic_entropy_math_reward": 0.38985058665275574,
+      "step": 80
+    },
+    {
+      "completion_length": 597.3839282989502,
+      "epoch": 0.21107491856677524,
+      "grad_norm": 0.007267440669238567,
+      "learning_rate": 9.319268354975958e-06,
+      "loss": -0.0246,
+      "reward": 0.9277514442801476,
+      "reward_std": 0.1274285425315611,
+      "rewards/accuracy_reward": 0.5382652953267097,
+      "rewards/semantic_entropy_math_reward": 0.38948614709079266,
+      "step": 81
+    },
+    {
+      "completion_length": 635.056755065918,
+      "epoch": 0.21368078175895766,
+      "grad_norm": 0.009285375475883484,
+      "learning_rate": 9.297308594793757e-06,
+      "loss": -0.0302,
+      "reward": 0.9357689209282398,
+      "reward_std": 0.11352400272153318,
+      "rewards/accuracy_reward": 0.5414540711790323,
+      "rewards/semantic_entropy_math_reward": 0.394314868375659,
+      "step": 82
+    },
+    {
+      "completion_length": 599.1862125396729,
+      "epoch": 0.21628664495114006,
+      "grad_norm": 0.007068796548992395,
+      "learning_rate": 9.275026964740101e-06,
+      "loss": -0.0221,
+      "reward": 0.8811042085289955,
+      "reward_std": 0.14417032268829644,
+      "rewards/accuracy_reward": 0.4999999925494194,
+      "rewards/semantic_entropy_math_reward": 0.3811042346060276,
+      "step": 83
+    },
+    {
+      "completion_length": 617.5911846160889,
+      "epoch": 0.21889250814332248,
+      "grad_norm": 0.006198438350111246,
+      "learning_rate": 9.252425133716639e-06,
+      "loss": -0.0388,
+      "reward": 0.8672558218240738,
+      "reward_std": 0.12377029308117926,
+      "rewards/accuracy_reward": 0.4948979541659355,
+      "rewards/semantic_entropy_math_reward": 0.3723578620702028,
+      "step": 84
+    },
+    {
+      "completion_length": 561.2780570983887,
+      "epoch": 0.22149837133550487,
+      "grad_norm": 0.006817261688411236,
+      "learning_rate": 9.229504794608182e-06,
+      "loss": -0.0129,
+      "reward": 0.9309402145445347,
+      "reward_std": 0.1257460000924766,
+      "rewards/accuracy_reward": 0.5331632606685162,
+      "rewards/semantic_entropy_math_reward": 0.3977769538760185,
+      "step": 85
+    },
+    {
+      "completion_length": 547.4725685119629,
+      "epoch": 0.2241042345276873,
+      "grad_norm": 0.005772161763161421,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": -0.0404,
+      "reward": 0.9456996843218803,
+      "reward_std": 0.1207300984824542,
+      "rewards/accuracy_reward": 0.5529336631298065,
+      "rewards/semantic_entropy_math_reward": 0.3927660472691059,
+      "step": 86
+    },
+    {
+      "completion_length": 583.5822620391846,
+      "epoch": 0.22671009771986972,
+      "grad_norm": 0.020753948017954826,
+      "learning_rate": 9.182715482828764e-06,
+      "loss": -0.0265,
+      "reward": 0.9639212638139725,
+      "reward_std": 0.09742264985106885,
+      "rewards/accuracy_reward": 0.570790808647871,
+      "rewards/semantic_entropy_math_reward": 0.39313045144081116,
+      "step": 87
+    },
+    {
+      "completion_length": 584.519121170044,
+      "epoch": 0.2293159609120521,
+      "grad_norm": 0.007225371431559324,
+      "learning_rate": 9.158850014693123e-06,
+      "loss": -0.0268,
+      "reward": 0.922284971922636,
+      "reward_std": 0.1539946385892108,
+      "rewards/accuracy_reward": 0.5414540749043226,
+      "rewards/semantic_entropy_math_reward": 0.3808309081941843,
+      "step": 88
+    },
+    {
+      "completion_length": 588.8143997192383,
+      "epoch": 0.23192182410423454,
+      "grad_norm": 0.006209821905940771,
+      "learning_rate": 9.134673047280644e-06,
+      "loss": -0.0311,
+      "reward": 0.8960458897054195,
+      "reward_std": 0.09522696980275214,
+      "rewards/accuracy_reward": 0.5153061179444194,
+      "rewards/semantic_entropy_math_reward": 0.3807397987693548,
+      "step": 89
+    },
+    {
+      "completion_length": 563.0012645721436,
+      "epoch": 0.23452768729641693,
+      "grad_norm": 0.007620004937052727,
+      "learning_rate": 9.110186391454389e-06,
+      "loss": -0.0224,
+      "reward": 0.9369533210992813,
+      "reward_std": 0.10029464948456734,
+      "rewards/accuracy_reward": 0.5427295845001936,
+      "rewards/semantic_entropy_math_reward": 0.39422375708818436,
+      "step": 90
+    },
+    {
+      "completion_length": 562.1103248596191,
+      "epoch": 0.23713355048859935,
+      "grad_norm": 0.008318237029016018,
+      "learning_rate": 9.085391881273182e-06,
+      "loss": -0.0276,
+      "reward": 0.9031523168087006,
+      "reward_std": 0.11616136459633708,
+      "rewards/accuracy_reward": 0.5197704005986452,
+      "rewards/semantic_entropy_math_reward": 0.3833819255232811,
+      "step": 91
+    },
+    {
+      "completion_length": 566.0152969360352,
+      "epoch": 0.23973941368078175,
+      "grad_norm": 0.013799573294818401,
+      "learning_rate": 9.060291373854252e-06,
+      "loss": -0.0309,
+      "reward": 0.9221938513219357,
+      "reward_std": 0.1298237289302051,
+      "rewards/accuracy_reward": 0.5471938699483871,
+      "rewards/semantic_entropy_math_reward": 0.3749999888241291,
+      "step": 92
+    },
+    {
+      "completion_length": 478.0140209197998,
+      "epoch": 0.24234527687296417,
+      "grad_norm": 0.010932566598057747,
+      "learning_rate": 9.034886749234112e-06,
+      "loss": -0.0229,
+      "reward": 0.9631923995912075,
+      "reward_std": 0.08580980636179447,
+      "rewards/accuracy_reward": 0.5599489714950323,
+      "rewards/semantic_entropy_math_reward": 0.4032434243708849,
+      "step": 93
+    },
+    {
+      "completion_length": 528.445779800415,
+      "epoch": 0.24495114006514657,
+      "grad_norm": 0.03433404862880707,
+      "learning_rate": 9.009179910227767e-06,
+      "loss": -0.0259,
+      "reward": 0.8987791389226913,
+      "reward_std": 0.1273018317297101,
+      "rewards/accuracy_reward": 0.5191326383501291,
+      "rewards/semantic_entropy_math_reward": 0.3796464875340462,
+      "step": 94
+    },
+    {
+      "completion_length": 566.7614727020264,
+      "epoch": 0.247557003257329,
+      "grad_norm": 0.00829783733934164,
+      "learning_rate": 8.98317278228618e-06,
+      "loss": -0.0304,
+      "reward": 0.8782798796892166,
+      "reward_std": 0.13542052335105836,
+      "rewards/accuracy_reward": 0.4936224343255162,
+      "rewards/semantic_entropy_math_reward": 0.3846574295312166,
+      "step": 95
+    },
+    {
+      "completion_length": 506.3705234527588,
+      "epoch": 0.2501628664495114,
+      "grad_norm": 0.00881106685847044,
+      "learning_rate": 8.956867313352055e-06,
+      "loss": -0.0221,
+      "reward": 0.9182762429118156,
+      "reward_std": 0.08380106370896101,
+      "rewards/accuracy_reward": 0.5191326439380646,
+      "rewards/semantic_entropy_math_reward": 0.399143585935235,
+      "step": 96
+    },
+    {
+      "completion_length": 570.8501167297363,
+      "epoch": 0.25276872964169383,
+      "grad_norm": 0.007192031946033239,
+      "learning_rate": 8.930265473713939e-06,
+      "loss": -0.0305,
+      "reward": 0.9258381798863411,
+      "reward_std": 0.12165293795987964,
+      "rewards/accuracy_reward": 0.5440050940960646,
+      "rewards/semantic_entropy_math_reward": 0.3818330932408571,
+      "step": 97
+    },
+    {
+      "completion_length": 508.7321243286133,
+      "epoch": 0.2553745928338762,
+      "grad_norm": 0.007854027673602104,
+      "learning_rate": 8.90336925585864e-06,
+      "loss": -0.0173,
+      "reward": 0.9938957393169403,
+      "reward_std": 0.10928461601724848,
+      "rewards/accuracy_reward": 0.5911989696323872,
+      "rewards/semantic_entropy_math_reward": 0.40269677340984344,
+      "step": 98
+    },
+    {
+      "completion_length": 524.1619777679443,
+      "epoch": 0.2579804560260586,
+      "grad_norm": 0.01574082300066948,
+      "learning_rate": 8.876180674322006e-06,
+      "loss": -0.024,
+      "reward": 0.941326517611742,
+      "reward_std": 0.1133777701179497,
+      "rewards/accuracy_reward": 0.5459183529019356,
+      "rewards/semantic_entropy_math_reward": 0.39540815725922585,
+      "step": 99
+    },
+    {
+      "completion_length": 512.2608318328857,
+      "epoch": 0.26058631921824105,
+      "grad_norm": 0.00712746474891901,
+      "learning_rate": 8.84870176553801e-06,
+      "loss": -0.0223,
+      "reward": 0.9894314855337143,
+      "reward_std": 0.10396600887179375,
+      "rewards/accuracy_reward": 0.5778061039745808,
+      "rewards/semantic_entropy_math_reward": 0.4116253647953272,
+      "step": 100
+    },
+    {
+      "completion_length": 576.9661922454834,
+      "epoch": 0.26319218241042347,
+      "grad_norm": 0.007007178850471973,
+      "learning_rate": 8.820934587686247e-06,
+      "loss": -0.0296,
+      "reward": 0.8965925425291061,
+      "reward_std": 0.12761994334869087,
+      "rewards/accuracy_reward": 0.49936224333941936,
+      "rewards/semantic_entropy_math_reward": 0.39723031409084797,
+      "step": 101
+    },
+    {
+      "completion_length": 513.0254955291748,
+      "epoch": 0.26579804560260584,
+      "grad_norm": 0.007423168048262596,
+      "learning_rate": 8.792881220537752e-06,
+      "loss": -0.0214,
+      "reward": 0.9517128206789494,
+      "reward_std": 0.08766992692835629,
+      "rewards/accuracy_reward": 0.5465561114251614,
+      "rewards/semantic_entropy_math_reward": 0.4051567129790783,
+      "step": 102
+    },
+    {
+      "completion_length": 548.6473083496094,
+      "epoch": 0.26840390879478826,
+      "grad_norm": 0.008394024334847927,
+      "learning_rate": 8.764543765299245e-06,
+      "loss": -0.0183,
+      "reward": 0.9393221363425255,
+      "reward_std": 0.1155447622295469,
+      "rewards/accuracy_reward": 0.541454067453742,
+      "rewards/semantic_entropy_math_reward": 0.39786806143820286,
+      "step": 103
+    },
+    {
+      "completion_length": 543.691312789917,
+      "epoch": 0.2710097719869707,
+      "grad_norm": 0.006529523991048336,
+      "learning_rate": 8.735924344455732e-06,
+      "loss": -0.0215,
+      "reward": 0.9579081423580647,
+      "reward_std": 0.11347127868793905,
+      "rewards/accuracy_reward": 0.5554846785962582,
+      "rewards/semantic_entropy_math_reward": 0.40242345817387104,
+      "step": 104
+    },
+    {
+      "completion_length": 507.02295112609863,
+      "epoch": 0.2736156351791531,
+      "grad_norm": 0.007422362454235554,
+      "learning_rate": 8.707025101611546e-06,
+      "loss": -0.0244,
+      "reward": 0.9213738925755024,
+      "reward_std": 0.10976181924343109,
+      "rewards/accuracy_reward": 0.5172193758189678,
+      "rewards/semantic_entropy_math_reward": 0.40415450744330883,
+      "step": 105
+    },
+    {
+      "completion_length": 452.20470428466797,
+      "epoch": 0.2762214983713355,
+      "grad_norm": 0.007135653402656317,
+      "learning_rate": 8.677848201329775e-06,
+      "loss": -0.0076,
+      "reward": 0.9969934076070786,
+      "reward_std": 0.08652752835769206,
+      "rewards/accuracy_reward": 0.5752550922334194,
+      "rewards/semantic_entropy_math_reward": 0.4217383284121752,
+      "step": 106
+    },
+    {
+      "completion_length": 510.8156795501709,
+      "epoch": 0.2788273615635179,
+      "grad_norm": 0.00994784850627184,
+      "learning_rate": 8.64839582897015e-06,
+      "loss": -0.021,
+      "reward": 0.8670735992491245,
+      "reward_std": 0.11074995016679168,
+      "rewards/accuracy_reward": 0.4942601975053549,
+      "rewards/semantic_entropy_math_reward": 0.3728134222328663,
+      "step": 107
+    },
+    {
+      "completion_length": 523.8329010009766,
+      "epoch": 0.2814332247557003,
+      "grad_norm": 0.00812004879117012,
+      "learning_rate": 8.61867019052535e-06,
+      "loss": -0.0254,
+      "reward": 0.9470662958920002,
+      "reward_std": 0.10799844446592033,
+      "rewards/accuracy_reward": 0.5350765231996775,
+      "rewards/semantic_entropy_math_reward": 0.4119897708296776,
+      "step": 108
+    },
+    {
+      "completion_length": 484.8297176361084,
+      "epoch": 0.28403908794788274,
+      "grad_norm": 0.00825856626033783,
+      "learning_rate": 8.588673512455781e-06,
+      "loss": -0.0122,
+      "reward": 0.9991800077259541,
+      "reward_std": 0.10875873418990523,
+      "rewards/accuracy_reward": 0.5924744792282581,
+      "rewards/semantic_entropy_math_reward": 0.40670554153621197,
+      "step": 109
+    },
+    {
+      "completion_length": 465.98851013183594,
+      "epoch": 0.28664495114006516,
+      "grad_norm": 0.011785555630922318,
+      "learning_rate": 8.558408041522801e-06,
+      "loss": -0.011,
+      "reward": 0.9412353932857513,
+      "reward_std": 0.11806983663700521,
+      "rewards/accuracy_reward": 0.5382652934640646,
+      "rewards/semantic_entropy_math_reward": 0.4029701240360737,
+      "step": 110
+    },
+    {
+      "completion_length": 464.45534896850586,
+      "epoch": 0.28925081433224753,
+      "grad_norm": 0.008536071516573429,
+      "learning_rate": 8.527876044620453e-06,
+      "loss": -0.011,
+      "reward": 0.9099853932857513,
+      "reward_std": 0.12709335889667273,
+      "rewards/accuracy_reward": 0.5216836612671614,
+      "rewards/semantic_entropy_math_reward": 0.3883017376065254,
+      "step": 111
+    },
+    {
+      "completion_length": 508.1205291748047,
+      "epoch": 0.29185667752442995,
+      "grad_norm": 0.006830580998212099,
+      "learning_rate": 8.497079808605659e-06,
+      "loss": -0.0125,
+      "reward": 0.9495262391865253,
+      "reward_std": 0.08507522073341534,
+      "rewards/accuracy_reward": 0.5389030510559678,
+      "rewards/semantic_entropy_math_reward": 0.41062317602336407,
+      "step": 112
+    },
+    {
+      "completion_length": 468.2008819580078,
+      "epoch": 0.2944625407166124,
+      "grad_norm": 0.008947983384132385,
+      "learning_rate": 8.466021640126946e-06,
+      "loss": -0.0144,
+      "reward": 0.9574526138603687,
+      "reward_std": 0.0999740477418527,
+      "rewards/accuracy_reward": 0.5503826402127743,
+      "rewards/semantic_entropy_math_reward": 0.4070699568837881,
+      "step": 113
+    },
+    {
+      "completion_length": 535.3322677612305,
+      "epoch": 0.2970684039087948,
+      "grad_norm": 0.011211535893380642,
+      "learning_rate": 8.434703865451666e-06,
+      "loss": -0.0204,
+      "reward": 0.8916727267205715,
+      "reward_std": 0.13218142627738416,
+      "rewards/accuracy_reward": 0.5057397801429033,
+      "rewards/semantic_entropy_math_reward": 0.385932931676507,
+      "step": 114
+    },
+    {
+      "completion_length": 543.2493572235107,
+      "epoch": 0.2996742671009772,
+      "grad_norm": 0.010953478515148163,
+      "learning_rate": 8.403128830291767e-06,
+      "loss": -0.017,
+      "reward": 0.8920371569693089,
+      "reward_std": 0.10963007633108646,
+      "rewards/accuracy_reward": 0.5076530482620001,
+      "rewards/semantic_entropy_math_reward": 0.3843841031193733,
+      "step": 115
+    },
+    {
+      "completion_length": 527.7404232025146,
+      "epoch": 0.3022801302931596,
+      "grad_norm": 0.012620029971003532,
+      "learning_rate": 8.371298899628091e-06,
+      "loss": -0.024,
+      "reward": 0.9088921137154102,
+      "reward_std": 0.12522040773183107,
+      "rewards/accuracy_reward": 0.5242346785962582,
+      "rewards/semantic_entropy_math_reward": 0.3846574239432812,
+      "step": 116
+    },
+    {
+      "completion_length": 519.883918762207,
+      "epoch": 0.304885993485342,
+      "grad_norm": 0.013618368655443192,
+      "learning_rate": 8.339216457533244e-06,
+      "loss": -0.0187,
+      "reward": 0.919278409332037,
+      "reward_std": 0.1413876386359334,
+      "rewards/accuracy_reward": 0.5280612204223871,
+      "rewards/semantic_entropy_math_reward": 0.39121718890964985,
+      "step": 117
+    },
+    {
+      "completion_length": 569.508279800415,
+      "epoch": 0.30749185667752443,
+      "grad_norm": 0.009268795140087605,
+      "learning_rate": 8.306883906993022e-06,
+      "loss": -0.0223,
+      "reward": 0.8455721400678158,
+      "reward_std": 0.14180739130824804,
+      "rewards/accuracy_reward": 0.4693877473473549,
+      "rewards/semantic_entropy_math_reward": 0.3761843889951706,
+      "step": 118
+    },
+    {
+      "completion_length": 548.354585647583,
+      "epoch": 0.31009771986970686,
+      "grad_norm": 0.008740663528442383,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": -0.0185,
+      "reward": 0.904518935829401,
+      "reward_std": 0.11498924950137734,
+      "rewards/accuracy_reward": 0.5191326458007097,
+      "rewards/semantic_entropy_math_reward": 0.3853862937539816,
+      "step": 119
+    },
+    {
+      "completion_length": 504.8743495941162,
+      "epoch": 0.3127035830618892,
+      "grad_norm": 0.008732241578400135,
+      "learning_rate": 8.24147818600428e-06,
+      "loss": -0.0162,
+      "reward": 0.9195517338812351,
+      "reward_std": 0.10652195988222957,
+      "rewards/accuracy_reward": 0.5184948872774839,
+      "rewards/semantic_entropy_math_reward": 0.4010568540543318,
+      "step": 120
+    },
+    {
+      "completion_length": 552.7168216705322,
+      "epoch": 0.31530944625407165,
+      "grad_norm": 0.00948347244411707,
+      "learning_rate": 8.20840991446645e-06,
+      "loss": -0.0168,
+      "reward": 0.8840196654200554,
+      "reward_std": 0.11878083180636168,
+      "rewards/accuracy_reward": 0.498724477365613,
+      "rewards/semantic_entropy_math_reward": 0.3852951806038618,
+      "step": 121
+    },
+    {
+      "completion_length": 536.5752468109131,
+      "epoch": 0.31791530944625407,
+      "grad_norm": 0.010393758304417133,
+      "learning_rate": 8.175101331937692e-06,
+      "loss": -0.0261,
+      "reward": 0.9547193609178066,
+      "reward_std": 0.1325584959704429,
+      "rewards/accuracy_reward": 0.5618622377514839,
+      "rewards/semantic_entropy_math_reward": 0.3928571343421936,
+      "step": 122
+    },
+    {
+      "completion_length": 514.7365875244141,
+      "epoch": 0.3205211726384365,
+      "grad_norm": 0.011021023616194725,
+      "learning_rate": 8.141554933242135e-06,
+      "loss": -0.0215,
+      "reward": 0.839012373238802,
+      "reward_std": 0.11658453289419413,
+      "rewards/accuracy_reward": 0.4521683529019356,
+      "rewards/semantic_entropy_math_reward": 0.3868440166115761,
+      "step": 123
+    },
+    {
+      "completion_length": 546.1747303009033,
+      "epoch": 0.3231270358306189,
+      "grad_norm": 0.012457598000764847,
+      "learning_rate": 8.10777323101642e-06,
+      "loss": -0.0149,
+      "reward": 0.8359147049486637,
+      "reward_std": 0.13787020510062575,
+      "rewards/accuracy_reward": 0.4540816228836775,
+      "rewards/semantic_entropy_math_reward": 0.38183307833969593,
+      "step": 124
+    },
+    {
+      "completion_length": 540.0956497192383,
+      "epoch": 0.3257328990228013,
+      "grad_norm": 0.00853867456316948,
+      "learning_rate": 8.073758755521506e-06,
+      "loss": -0.0105,
+      "reward": 0.8753644078969955,
+      "reward_std": 0.10535615705884993,
+      "rewards/accuracy_reward": 0.4891581516712904,
+      "rewards/semantic_entropy_math_reward": 0.3862062580883503,
+      "step": 125
+    },
+    {
+      "completion_length": 528.5184841156006,
+      "epoch": 0.3283387622149837,
+      "grad_norm": 0.010948549024760723,
+      "learning_rate": 8.03951405445314e-06,
+      "loss": -0.0143,
+      "reward": 0.9124453142285347,
+      "reward_std": 0.12273865600582212,
+      "rewards/accuracy_reward": 0.5114795807749033,
+      "rewards/semantic_entropy_math_reward": 0.40096573159098625,
+      "step": 126
+    },
+    {
+      "completion_length": 544.5784301757812,
+      "epoch": 0.3309446254071661,
+      "grad_norm": 0.013710816390812397,
+      "learning_rate": 8.005041692751055e-06,
+      "loss": -0.0167,
+      "reward": 0.9290269501507282,
+      "reward_std": 0.13117512222379446,
+      "rewards/accuracy_reward": 0.5286989714950323,
+      "rewards/semantic_entropy_math_reward": 0.4003279786556959,
+      "step": 127
+    },
+    {
+      "completion_length": 577.0956478118896,
+      "epoch": 0.33355048859934855,
+      "grad_norm": 0.03651218116283417,
+      "learning_rate": 7.970344252406832e-06,
+      "loss": -0.0183,
+      "reward": 0.9039722643792629,
+      "reward_std": 0.1263378494186327,
+      "rewards/accuracy_reward": 0.5178571362048388,
+      "rewards/semantic_entropy_math_reward": 0.38611516170203686,
+      "step": 128
+    },
+    {
+      "completion_length": 541.4247360229492,
+      "epoch": 0.33615635179153097,
+      "grad_norm": 0.013164433650672436,
+      "learning_rate": 7.935424332270523e-06,
+      "loss": -0.0233,
+      "reward": 0.9340378865599632,
+      "reward_std": 0.1010712229181081,
+      "rewards/accuracy_reward": 0.5306122377514839,
+      "rewards/semantic_entropy_math_reward": 0.4034256376326084,
+      "step": 129
+    },
+    {
+      "completion_length": 581.0784339904785,
+      "epoch": 0.33876221498371334,
+      "grad_norm": 0.012624763883650303,
+      "learning_rate": 7.900284547855992e-06,
+      "loss": -0.0054,
+      "reward": 0.8548651225864887,
+      "reward_std": 0.11518756812438369,
+      "rewards/accuracy_reward": 0.467474477365613,
+      "rewards/semantic_entropy_math_reward": 0.3873906545341015,
+      "step": 130
+    },
+    {
+      "completion_length": 534.0159378051758,
+      "epoch": 0.34136807817589576,
+      "grad_norm": 0.011506155133247375,
+      "learning_rate": 7.864927531145012e-06,
+      "loss": -0.0124,
+      "reward": 0.9199161604046822,
+      "reward_std": 0.10405887564411387,
+      "rewards/accuracy_reward": 0.5204081516712904,
+      "rewards/semantic_entropy_math_reward": 0.39950801618397236,
+      "step": 131
+    },
+    {
+      "completion_length": 548.1957817077637,
+      "epoch": 0.3439739413680782,
+      "grad_norm": 0.013849125243723392,
+      "learning_rate": 7.829355930390126e-06,
+      "loss": -0.0222,
+      "reward": 0.8961370177567005,
+      "reward_std": 0.11209307186072692,
+      "rewards/accuracy_reward": 0.487882643006742,
+      "rewards/semantic_entropy_math_reward": 0.4082543719559908,
+      "step": 132
+    },
+    {
+      "completion_length": 533.2085285186768,
+      "epoch": 0.3465798045602606,
+      "grad_norm": 0.007529080845415592,
+      "learning_rate": 7.7935724099163e-06,
+      "loss": -0.011,
+      "reward": 0.9659256264567375,
+      "reward_std": 0.10825655586086214,
+      "rewards/accuracy_reward": 0.5688775414600968,
+      "rewards/semantic_entropy_math_reward": 0.3970481026917696,
+      "step": 133
+    },
+    {
+      "completion_length": 530.067590713501,
+      "epoch": 0.349185667752443,
+      "grad_norm": 0.013162490911781788,
+      "learning_rate": 7.757579649921354e-06,
+      "loss": -0.0155,
+      "reward": 0.9479773864150047,
+      "reward_std": 0.14899065819918178,
+      "rewards/accuracy_reward": 0.5433673392981291,
+      "rewards/semantic_entropy_math_reward": 0.4046100489795208,
+      "step": 134
+    },
+    {
+      "completion_length": 564.3252391815186,
+      "epoch": 0.3517915309446254,
+      "grad_norm": 0.014966626651585102,
+      "learning_rate": 7.721380346275221e-06,
+      "loss": -0.0217,
+      "reward": 0.9267492741346359,
+      "reward_std": 0.1269418167648837,
+      "rewards/accuracy_reward": 0.5306122275069356,
+      "rewards/semantic_entropy_math_reward": 0.39613701961934566,
+      "step": 135
+    },
+    {
+      "completion_length": 551.4642715454102,
+      "epoch": 0.3543973941368078,
+      "grad_norm": 0.007552569266408682,
+      "learning_rate": 7.684977210318024e-06,
+      "loss": -0.0128,
+      "reward": 0.9007835239171982,
+      "reward_std": 0.10546381218591705,
+      "rewards/accuracy_reward": 0.5031887609511614,
+      "rewards/semantic_entropy_math_reward": 0.39759473875164986,
+      "step": 136
+    },
+    {
+      "completion_length": 532.7742252349854,
+      "epoch": 0.35700325732899024,
+      "grad_norm": 0.008308288641273975,
+      "learning_rate": 7.648372968656995e-06,
+      "loss": -0.01,
+      "reward": 0.9131742008030415,
+      "reward_std": 0.11054620456707198,
+      "rewards/accuracy_reward": 0.5057397875934839,
+      "rewards/semantic_entropy_math_reward": 0.40743439458310604,
+      "step": 137
+    },
+    {
+      "completion_length": 537.5076389312744,
+      "epoch": 0.35960912052117266,
+      "grad_norm": 0.007219830993562937,
+      "learning_rate": 7.611570362962247e-06,
+      "loss": -0.0164,
+      "reward": 0.9657434225082397,
+      "reward_std": 0.12022296455688775,
+      "rewards/accuracy_reward": 0.5491071343421936,
+      "rewards/semantic_entropy_math_reward": 0.41663629189133644,
+      "step": 138
+    },
+    {
+      "completion_length": 516.6587944030762,
+      "epoch": 0.36221498371335503,
+      "grad_norm": 0.010173982009291649,
+      "learning_rate": 7.574572149761437e-06,
+      "loss": -0.0154,
+      "reward": 0.9734876044094563,
+      "reward_std": 0.11138522112742066,
+      "rewards/accuracy_reward": 0.5695152971893549,
+      "rewards/semantic_entropy_math_reward": 0.40397229604423046,
+      "step": 139
+    },
+    {
+      "completion_length": 567.7136325836182,
+      "epoch": 0.36482084690553745,
+      "grad_norm": 0.008159446530044079,
+      "learning_rate": 7.5373811002332785e-06,
+      "loss": -0.0118,
+      "reward": 0.910987600684166,
+      "reward_std": 0.09473348397295922,
+      "rewards/accuracy_reward": 0.5095663145184517,
+      "rewards/semantic_entropy_math_reward": 0.4014212768524885,
+      "step": 140
+    },
+    {
+      "completion_length": 566.8067493438721,
+      "epoch": 0.3674267100977199,
+      "grad_norm": 0.006627434398978949,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": -0.0138,
+      "reward": 0.9380466379225254,
+      "reward_std": 0.10335312166716903,
+      "rewards/accuracy_reward": 0.5382652897387743,
+      "rewards/semantic_entropy_math_reward": 0.39978133141994476,
+      "step": 141
+    },
+    {
+      "completion_length": 527.3093032836914,
+      "epoch": 0.3700325732899023,
+      "grad_norm": 0.007407406345009804,
+      "learning_rate": 7.462431648918689e-06,
+      "loss": -0.0173,
+      "reward": 0.904336716979742,
+      "reward_std": 0.11712831069598906,
+      "rewards/accuracy_reward": 0.5070152916014194,
+      "rewards/semantic_entropy_math_reward": 0.3973214253783226,
+      "step": 142
+    },
+    {
+      "completion_length": 515.315034866333,
+      "epoch": 0.37263843648208467,
+      "grad_norm": 0.008976518176496029,
+      "learning_rate": 7.424678860871584e-06,
+      "loss": -0.0096,
+      "reward": 0.9782252088189125,
+      "reward_std": 0.11401024123188108,
+      "rewards/accuracy_reward": 0.5752550922334194,
+      "rewards/semantic_entropy_math_reward": 0.4029701091349125,
+      "step": 143
+    },
+    {
+      "completion_length": 565.4763851165771,
+      "epoch": 0.3752442996742671,
+      "grad_norm": 0.008781256154179573,
+      "learning_rate": 7.3867444635553165e-06,
+      "loss": -0.0196,
+      "reward": 0.9317601844668388,
+      "reward_std": 0.12317358143627644,
+      "rewards/accuracy_reward": 0.5331632494926453,
+      "rewards/semantic_entropy_math_reward": 0.398596940562129,
+      "step": 144
+    },
+    {
+      "completion_length": 585.6358242034912,
+      "epoch": 0.3778501628664495,
+      "grad_norm": 0.006888892501592636,
+      "learning_rate": 7.3486312982691134e-06,
+      "loss": -0.0132,
+      "reward": 0.8779154568910599,
+      "reward_std": 0.10448942880611867,
+      "rewards/accuracy_reward": 0.48278059996664524,
+      "rewards/semantic_entropy_math_reward": 0.39513483084738255,
+      "step": 145
+    },
+    {
+      "completion_length": 547.9228191375732,
+      "epoch": 0.38045602605863194,
+      "grad_norm": 0.009227856993675232,
+      "learning_rate": 7.310342219701981e-06,
+      "loss": -0.0185,
+      "reward": 0.8972303010523319,
+      "reward_std": 0.13705383567139506,
+      "rewards/accuracy_reward": 0.49808672443032265,
+      "rewards/semantic_entropy_math_reward": 0.3991435803472996,
+      "step": 146
+    },
+    {
+      "completion_length": 527.695779800415,
+      "epoch": 0.38306188925081436,
+      "grad_norm": 0.010845191776752472,
+      "learning_rate": 7.271880095718895e-06,
+      "loss": -0.0189,
+      "reward": 0.9622813388705254,
+      "reward_std": 0.10206798242870718,
+      "rewards/accuracy_reward": 0.5612244810909033,
+      "rewards/semantic_entropy_math_reward": 0.40105685591697693,
+      "step": 147
+    },
+    {
+      "completion_length": 479.8839225769043,
+      "epoch": 0.3856677524429967,
+      "grad_norm": 0.012252271175384521,
+      "learning_rate": 7.233247807145989e-06,
+      "loss": -0.0217,
+      "reward": 0.8895772397518158,
+      "reward_std": 0.11137077608145773,
+      "rewards/accuracy_reward": 0.48022957891225815,
+      "rewards/semantic_entropy_math_reward": 0.4093476627022028,
+      "step": 148
+    },
+    {
+      "completion_length": 499.83226013183594,
+      "epoch": 0.38827361563517915,
+      "grad_norm": 0.01276180800050497,
+      "learning_rate": 7.19444824755478e-06,
+      "loss": -0.0136,
+      "reward": 0.9279336594045162,
+      "reward_std": 0.09600863244850188,
+      "rewards/accuracy_reward": 0.5312499888241291,
+      "rewards/semantic_entropy_math_reward": 0.3966836631298065,
+      "step": 149
+    },
+    {
+      "completion_length": 505.0191249847412,
+      "epoch": 0.39087947882736157,
+      "grad_norm": 0.009752064011991024,
+      "learning_rate": 7.155484323045442e-06,
+      "loss": -0.025,
+      "reward": 0.8462098762392998,
+      "reward_std": 0.13227392174303532,
+      "rewards/accuracy_reward": 0.4540816266089678,
+      "rewards/semantic_entropy_math_reward": 0.39212826639413834,
+      "step": 150
+    },
+    {
+      "completion_length": 486.4183597564697,
+      "epoch": 0.393485342019544,
+      "grad_norm": 0.011875058524310589,
+      "learning_rate": 7.11635895202914e-06,
+      "loss": -0.019,
+      "reward": 0.8894861489534378,
+      "reward_std": 0.09644467895850539,
+      "rewards/accuracy_reward": 0.4917091755196452,
+      "rewards/semantic_entropy_math_reward": 0.39777696318924427,
+      "step": 151
+    },
+    {
+      "completion_length": 516.7933540344238,
+      "epoch": 0.39609120521172636,
+      "grad_norm": 0.00881208572536707,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": -0.0209,
+      "reward": 0.9480685032904148,
+      "reward_std": 0.10766224935650826,
+      "rewards/accuracy_reward": 0.5471938634291291,
+      "rewards/semantic_entropy_math_reward": 0.40087462589144707,
+      "step": 152
+    },
+    {
+      "completion_length": 506.79143714904785,
+      "epoch": 0.3986970684039088,
+      "grad_norm": 0.012560566887259483,
+      "learning_rate": 7.037635604362786e-06,
+      "loss": -0.0162,
+      "reward": 0.9439686462283134,
+      "reward_std": 0.10271801124326885,
+      "rewards/accuracy_reward": 0.5318877454847097,
+      "rewards/semantic_entropy_math_reward": 0.41208088397979736,
+      "step": 153
+    },
+    {
+      "completion_length": 438.49106216430664,
+      "epoch": 0.4013029315960912,
+      "grad_norm": 0.011885056272149086,
+      "learning_rate": 6.9980435241181785e-06,
+      "loss": -0.0173,
+      "reward": 0.9576348140835762,
+      "reward_std": 0.09978608437813818,
+      "rewards/accuracy_reward": 0.541454067453742,
+      "rewards/semantic_entropy_math_reward": 0.4161807522177696,
+      "step": 154
+    },
+    {
+      "completion_length": 513.610954284668,
+      "epoch": 0.40390879478827363,
+      "grad_norm": 0.009011725895106792,
+      "learning_rate": 6.958301789735853e-06,
+      "loss": -0.0224,
+      "reward": 0.8590561151504517,
+      "reward_std": 0.10832268767990172,
+      "rewards/accuracy_reward": 0.46811223216354847,
+      "rewards/semantic_entropy_math_reward": 0.390943868085742,
+      "step": 155
+    },
+    {
+      "completion_length": 480.7538185119629,
+      "epoch": 0.40651465798045605,
+      "grad_norm": 0.013393928296864033,
+      "learning_rate": 6.918413377885193e-06,
+      "loss": -0.026,
+      "reward": 0.9114431329071522,
+      "reward_std": 0.09208173397928476,
+      "rewards/accuracy_reward": 0.49489794485270977,
+      "rewards/semantic_entropy_math_reward": 0.4165451843291521,
+      "step": 156
+    },
+    {
+      "completion_length": 491.4189987182617,
+      "epoch": 0.4091205211726384,
+      "grad_norm": 0.009988815523684025,
+      "learning_rate": 6.878381276221777e-06,
+      "loss": -0.0103,
+      "reward": 0.8825619332492352,
+      "reward_std": 0.10942790098488331,
+      "rewards/accuracy_reward": 0.4846938634291291,
+      "rewards/semantic_entropy_math_reward": 0.3978680744767189,
+      "step": 157
+    },
+    {
+      "completion_length": 496.1402931213379,
+      "epoch": 0.41172638436482084,
+      "grad_norm": 0.008957677520811558,
+      "learning_rate": 6.838208483163601e-06,
+      "loss": -0.0255,
+      "reward": 0.8830174691975117,
+      "reward_std": 0.10132573859300464,
+      "rewards/accuracy_reward": 0.49426019191741943,
+      "rewards/semantic_entropy_math_reward": 0.3887572903186083,
+      "step": 158
+    },
+    {
+      "completion_length": 493.60011863708496,
+      "epoch": 0.41433224755700326,
+      "grad_norm": 0.007134555373340845,
+      "learning_rate": 6.797898007666493e-06,
+      "loss": -0.0279,
+      "reward": 0.8826530575752258,
+      "reward_std": 0.0909608873189427,
+      "rewards/accuracy_reward": 0.4853316191583872,
+      "rewards/semantic_entropy_math_reward": 0.3973214253783226,
+      "step": 159
+    },
+    {
+      "completion_length": 483.09437370300293,
+      "epoch": 0.4169381107491857,
+      "grad_norm": 0.009405710734426975,
+      "learning_rate": 6.757452868998737e-06,
+      "loss": -0.016,
+      "reward": 0.9526238851249218,
+      "reward_std": 0.09115565207321197,
+      "rewards/accuracy_reward": 0.5363520253449678,
+      "rewards/semantic_entropy_math_reward": 0.4162718504667282,
+      "step": 160
+    },
+    {
+      "completion_length": 475.0184898376465,
+      "epoch": 0.41954397394136805,
+      "grad_norm": 0.008445804007351398,
+      "learning_rate": 6.716876096514944e-06,
+      "loss": -0.0231,
+      "reward": 0.9812317453324795,
+      "reward_std": 0.09919259638991207,
+      "rewards/accuracy_reward": 0.5631377417594194,
+      "rewards/semantic_entropy_math_reward": 0.4180940203368664,
+      "step": 161
+    },
+    {
+      "completion_length": 448.816951751709,
+      "epoch": 0.4221498371335505,
+      "grad_norm": 0.006540235131978989,
+      "learning_rate": 6.676170729429132e-06,
+      "loss": -0.0164,
+      "reward": 0.9816872999072075,
+      "reward_std": 0.07881008239928633,
+      "rewards/accuracy_reward": 0.5535714123398066,
+      "rewards/semantic_entropy_math_reward": 0.4281158857047558,
+      "step": 162
+    },
+    {
+      "completion_length": 501.0682258605957,
+      "epoch": 0.4247557003257329,
+      "grad_norm": 0.006146623287349939,
+      "learning_rate": 6.635339816587109e-06,
+      "loss": -0.0158,
+      "reward": 0.9382288418710232,
+      "reward_std": 0.09887587034609169,
+      "rewards/accuracy_reward": 0.5229591736570001,
+      "rewards/semantic_entropy_math_reward": 0.41526967100799084,
+      "step": 163
+    },
+    {
+      "completion_length": 506.3195095062256,
+      "epoch": 0.4273615635179153,
+      "grad_norm": 0.007581164129078388,
+      "learning_rate": 6.594386416238095e-06,
+      "loss": -0.0192,
+      "reward": 0.908618800342083,
+      "reward_std": 0.12335213355254382,
+      "rewards/accuracy_reward": 0.5153061114251614,
+      "rewards/semantic_entropy_math_reward": 0.39331267960369587,
+      "step": 164
+    },
+    {
+      "completion_length": 503.4725742340088,
+      "epoch": 0.42996742671009774,
+      "grad_norm": 0.006690696347504854,
+      "learning_rate": 6.553313595805666e-06,
+      "loss": -0.018,
+      "reward": 0.9704810380935669,
+      "reward_std": 0.08365448092808947,
+      "rewards/accuracy_reward": 0.5497448891401291,
+      "rewards/semantic_entropy_math_reward": 0.42073615454137325,
+      "step": 165
+    },
+    {
+      "completion_length": 469.7703971862793,
+      "epoch": 0.4325732899022801,
+      "grad_norm": 0.006663058884441853,
+      "learning_rate": 6.512124431658006e-06,
+      "loss": -0.0153,
+      "reward": 0.9451530575752258,
+      "reward_std": 0.09975362673867494,
+      "rewards/accuracy_reward": 0.5338010098785162,
+      "rewards/semantic_entropy_math_reward": 0.411352027207613,
+      "step": 166
+    },
+    {
+      "completion_length": 488.9763946533203,
+      "epoch": 0.43517915309446253,
+      "grad_norm": 0.010632487945258617,
+      "learning_rate": 6.470822008877482e-06,
+      "loss": -0.013,
+      "reward": 0.968567755073309,
+      "reward_std": 0.09277999610640109,
+      "rewards/accuracy_reward": 0.5682397782802582,
+      "rewards/semantic_entropy_math_reward": 0.40032798051834106,
+      "step": 167
+    },
+    {
+      "completion_length": 489.58289909362793,
+      "epoch": 0.43778501628664496,
+      "grad_norm": 0.006177806295454502,
+      "learning_rate": 6.4294094210295725e-06,
+      "loss": -0.0196,
+      "reward": 0.9651056863367558,
+      "reward_std": 0.08780966483755037,
+      "rewards/accuracy_reward": 0.5478316228836775,
+      "rewards/semantic_entropy_math_reward": 0.4172740466892719,
+      "step": 168
+    },
+    {
+      "completion_length": 475.0695056915283,
+      "epoch": 0.4403908794788274,
+      "grad_norm": 0.008250612765550613,
+      "learning_rate": 6.3878897699311525e-06,
+      "loss": -0.0085,
+      "reward": 0.9698432683944702,
+      "reward_std": 0.09511806583032012,
+      "rewards/accuracy_reward": 0.5452805999666452,
+      "rewards/semantic_entropy_math_reward": 0.424562668427825,
+      "step": 169
+    },
+    {
+      "completion_length": 485.93302726745605,
+      "epoch": 0.44299674267100975,
+      "grad_norm": 0.00776874041184783,
+      "learning_rate": 6.346266165418173e-06,
+      "loss": -0.0268,
+      "reward": 0.9740342572331429,
+      "reward_std": 0.09444212296511978,
+      "rewards/accuracy_reward": 0.554846927523613,
+      "rewards/semantic_entropy_math_reward": 0.41918730549514294,
+      "step": 170
+    },
+    {
+      "completion_length": 497.7283020019531,
+      "epoch": 0.44560260586319217,
+      "grad_norm": 0.007973974570631981,
+      "learning_rate": 6.304541725112734e-06,
+      "loss": -0.0158,
+      "reward": 0.9323068186640739,
+      "reward_std": 0.09420971409417689,
+      "rewards/accuracy_reward": 0.5280612092465162,
+      "rewards/semantic_entropy_math_reward": 0.4042456317692995,
+      "step": 171
+    },
+    {
+      "completion_length": 498.0956573486328,
+      "epoch": 0.4482084690553746,
+      "grad_norm": 0.007753515616059303,
+      "learning_rate": 6.262719574189564e-06,
+      "loss": -0.0223,
+      "reward": 0.9188228957355022,
+      "reward_std": 0.09362561779562384,
+      "rewards/accuracy_reward": 0.5197703912854195,
+      "rewards/semantic_entropy_math_reward": 0.39905248768627644,
+      "step": 172
+    },
+    {
+      "completion_length": 470.75508880615234,
+      "epoch": 0.450814332247557,
+      "grad_norm": 0.007259304169565439,
+      "learning_rate": 6.2208028451419575e-06,
+      "loss": -0.01,
+      "reward": 0.9154518619179726,
+      "reward_std": 0.08905957284150645,
+      "rewards/accuracy_reward": 0.5076530463993549,
+      "rewards/semantic_entropy_math_reward": 0.4077988266944885,
+      "step": 173
+    },
+    {
+      "completion_length": 454.5631294250488,
+      "epoch": 0.45342019543973944,
+      "grad_norm": 0.0075342534109950066,
+      "learning_rate": 6.178794677547138e-06,
+      "loss": -0.0105,
+      "reward": 1.002368789166212,
+      "reward_std": 0.0902844921220094,
+      "rewards/accuracy_reward": 0.5733418203890324,
+      "rewards/semantic_entropy_math_reward": 0.4290269613265991,
+      "step": 174
+    },
+    {
+      "completion_length": 456.7385063171387,
+      "epoch": 0.4560260586319218,
+      "grad_norm": 0.008164818398654461,
+      "learning_rate": 6.136698217831106e-06,
+      "loss": -0.0083,
+      "reward": 0.9758564047515392,
+      "reward_std": 0.08968472620472312,
+      "rewards/accuracy_reward": 0.5471938662230968,
+      "rewards/semantic_entropy_math_reward": 0.42866252548992634,
+      "step": 175
+    },
+    {
+      "completion_length": 461.8769073486328,
+      "epoch": 0.4586319218241042,
+      "grad_norm": 0.018572386354207993,
+      "learning_rate": 6.094516619032975e-06,
+      "loss": -0.0141,
+      "reward": 0.9573614969849586,
+      "reward_std": 0.09445905161555856,
+      "rewards/accuracy_reward": 0.5452805962413549,
+      "rewards/semantic_entropy_math_reward": 0.41208089515566826,
+      "step": 176
+    },
+    {
+      "completion_length": 460.4668254852295,
+      "epoch": 0.46123778501628665,
+      "grad_norm": 0.01110108196735382,
+      "learning_rate": 6.052253040568804e-06,
+      "loss": -0.0138,
+      "reward": 0.9251093156635761,
+      "reward_std": 0.1255237814038992,
+      "rewards/accuracy_reward": 0.5191326439380646,
+      "rewards/semantic_entropy_math_reward": 0.4059766735881567,
+      "step": 177
+    },
+    {
+      "completion_length": 450.43493843078613,
+      "epoch": 0.4638436482084691,
+      "grad_norm": 0.010779962874948978,
+      "learning_rate": 6.009910647994956e-06,
+      "loss": -0.0185,
+      "reward": 0.9765852577984333,
+      "reward_std": 0.10055442503653467,
+      "rewards/accuracy_reward": 0.560586716979742,
+      "rewards/semantic_entropy_math_reward": 0.4159985240548849,
+      "step": 178
+    },
+    {
+      "completion_length": 507.37243843078613,
+      "epoch": 0.46644951140065144,
+      "grad_norm": 0.009717591106891632,
+      "learning_rate": 5.967492612770999e-06,
+      "loss": -0.0185,
+      "reward": 0.947704054415226,
+      "reward_std": 0.09296219941461459,
+      "rewards/accuracy_reward": 0.5401785597205162,
+      "rewards/semantic_entropy_math_reward": 0.4075255021452904,
+      "step": 179
+    },
+    {
+      "completion_length": 454.82906913757324,
+      "epoch": 0.46905537459283386,
+      "grad_norm": 0.007440619170665741,
+      "learning_rate": 5.925002112022158e-06,
+      "loss": -0.0158,
+      "reward": 0.9303935505449772,
+      "reward_std": 0.10131418029777706,
+      "rewards/accuracy_reward": 0.5210459046065807,
+      "rewards/semantic_entropy_math_reward": 0.40934766083955765,
+      "step": 180
+    },
+    {
+      "completion_length": 485.0899181365967,
+      "epoch": 0.4716612377850163,
+      "grad_norm": 0.011163070797920227,
+      "learning_rate": 5.882442328301356e-06,
+      "loss": -0.025,
+      "reward": 0.958454791456461,
+      "reward_std": 0.10061666322872043,
+      "rewards/accuracy_reward": 0.554209167137742,
+      "rewards/semantic_entropy_math_reward": 0.4042456094175577,
+      "step": 181
+    },
+    {
+      "completion_length": 465.15815353393555,
+      "epoch": 0.4742671009771987,
+      "grad_norm": 0.008569203317165375,
+      "learning_rate": 5.839816449350824e-06,
+      "loss": -0.0195,
+      "reward": 0.9420553632080555,
+      "reward_std": 0.06814239080995321,
+      "rewards/accuracy_reward": 0.529336728155613,
+      "rewards/semantic_entropy_math_reward": 0.41271864995360374,
+      "step": 182
+    },
+    {
+      "completion_length": 514.3443832397461,
+      "epoch": 0.47687296416938113,
+      "grad_norm": 0.0077494359575212,
+      "learning_rate": 5.7971276678633625e-06,
+      "loss": -0.0187,
+      "reward": 1.012208428233862,
+      "reward_std": 0.07250937505159527,
+      "rewards/accuracy_reward": 0.600127536803484,
+      "rewards/semantic_entropy_math_reward": 0.4120808970183134,
+      "step": 183
+    },
+    {
+      "completion_length": 487.92983055114746,
+      "epoch": 0.4794788273615635,
+      "grad_norm": 0.007494843099266291,
+      "learning_rate": 5.754379181243179e-06,
+      "loss": -0.0225,
+      "reward": 0.9169095903635025,
+      "reward_std": 0.09801324107684195,
+      "rewards/accuracy_reward": 0.5159438718110323,
+      "rewards/semantic_entropy_math_reward": 0.40096573159098625,
+      "step": 184
+    },
+    {
+      "completion_length": 480.8692512512207,
+      "epoch": 0.4820846905537459,
+      "grad_norm": 0.008148902095854282,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": -0.0216,
+      "reward": 1.0052842535078526,
+      "reward_std": 0.09645737172104418,
+      "rewards/accuracy_reward": 0.5880101881921291,
+      "rewards/semantic_entropy_math_reward": 0.41727403923869133,
+      "step": 185
+    },
+    {
+      "completion_length": 472.0790710449219,
+      "epoch": 0.48469055374592834,
+      "grad_norm": 0.00966109149158001,
+      "learning_rate": 5.668715904341365e-06,
+      "loss": -0.0201,
+      "reward": 0.9480684921145439,
+      "reward_std": 0.09934344311477616,
+      "rewards/accuracy_reward": 0.5299744755029678,
+      "rewards/semantic_entropy_math_reward": 0.4180940203368664,
+      "step": 186
+    },
+    {
+      "completion_length": 475.868616104126,
+      "epoch": 0.48729641693811077,
+      "grad_norm": 0.010433156974613667,
+      "learning_rate": 5.62580753026823e-06,
+      "loss": -0.0251,
+      "reward": 0.8839285559952259,
+      "reward_std": 0.0782707966864109,
+      "rewards/accuracy_reward": 0.4687499953433871,
+      "rewards/semantic_entropy_math_reward": 0.4151785559952259,
+      "step": 187
+    },
+    {
+      "completion_length": 476.2665729522705,
+      "epoch": 0.48990228013029313,
+      "grad_norm": 0.007323419209569693,
+      "learning_rate": 5.5828522829987965e-06,
+      "loss": -0.0156,
+      "reward": 0.9895225577056408,
+      "reward_std": 0.08914521127007902,
+      "rewards/accuracy_reward": 0.5669642705470324,
+      "rewards/semantic_entropy_math_reward": 0.42255830578505993,
+      "step": 188
+    },
+    {
+      "completion_length": 479.14284896850586,
+      "epoch": 0.49250814332247556,
+      "grad_norm": 0.0054732197895646095,
+      "learning_rate": 5.539853379895656e-06,
+      "loss": -0.0189,
+      "reward": 1.0017310306429863,
+      "reward_std": 0.06712543140747584,
+      "rewards/accuracy_reward": 0.5733418166637421,
+      "rewards/semantic_entropy_math_reward": 0.42838921397924423,
+      "step": 189
+    },
+    {
+      "completion_length": 462.00828552246094,
+      "epoch": 0.495114006514658,
+      "grad_norm": 0.0109707685187459,
+      "learning_rate": 5.496814041591234e-06,
+      "loss": -0.0202,
+      "reward": 0.9079810194671154,
+      "reward_std": 0.10293658822774887,
+      "rewards/accuracy_reward": 0.5012755002826452,
+      "rewards/semantic_entropy_math_reward": 0.4067055266350508,
+      "step": 190
+    },
+    {
+      "completion_length": 497.70853996276855,
+      "epoch": 0.4977198697068404,
+      "grad_norm": 0.008094045333564281,
+      "learning_rate": 5.453737491746572e-06,
+      "loss": -0.0197,
+      "reward": 0.9156340956687927,
+      "reward_std": 0.0916732897167094,
+      "rewards/accuracy_reward": 0.5191326476633549,
+      "rewards/semantic_entropy_math_reward": 0.39650145545601845,
+      "step": 191
+    },
+    {
+      "completion_length": 478.77613639831543,
+      "epoch": 0.5003257328990228,
+      "grad_norm": 0.014000942930579185,
+      "learning_rate": 5.410626956809864e-06,
+      "loss": -0.0141,
+      "reward": 0.9786807335913181,
+      "reward_std": 0.09398427966516465,
+      "rewards/accuracy_reward": 0.5637754946947098,
+      "rewards/semantic_entropy_math_reward": 0.4149052444845438,
+      "step": 192
+    },
+    {
+      "completion_length": 456.85266494750977,
+      "epoch": 0.5029315960912052,
+      "grad_norm": 0.00816336553543806,
+      "learning_rate": 5.367485665774802e-06,
+      "loss": -0.0095,
+      "reward": 0.9785896465182304,
+      "reward_std": 0.08977296199009288,
+      "rewards/accuracy_reward": 0.565051008015871,
+      "rewards/semantic_entropy_math_reward": 0.4135386198759079,
+      "step": 193
+    },
+    {
+      "completion_length": 449.39922523498535,
+      "epoch": 0.5055374592833877,
+      "grad_norm": 0.007704311050474644,
+      "learning_rate": 5.324316849938715e-06,
+      "loss": -0.0191,
+      "reward": 0.9876093044877052,
+      "reward_std": 0.10061339178355411,
+      "rewards/accuracy_reward": 0.581632636487484,
+      "rewards/semantic_entropy_math_reward": 0.40597667545080185,
+      "step": 194
+    },
+    {
+      "completion_length": 450.8992233276367,
+      "epoch": 0.50814332247557,
+      "grad_norm": 0.008355666883289814,
+      "learning_rate": 5.281123742660558e-06,
+      "loss": -0.0215,
+      "reward": 0.9452441520988941,
+      "reward_std": 0.09853553911671042,
+      "rewards/accuracy_reward": 0.5312499823048711,
+      "rewards/semantic_entropy_math_reward": 0.41399415768682957,
+      "step": 195
+    },
+    {
+      "completion_length": 475.535701751709,
+      "epoch": 0.5107491856677524,
+      "grad_norm": 0.00795848947018385,
+      "learning_rate": 5.237909579118713e-06,
+      "loss": -0.0228,
+      "reward": 0.9396865740418434,
+      "reward_std": 0.08185443480033427,
+      "rewards/accuracy_reward": 0.5293367225676775,
+      "rewards/semantic_entropy_math_reward": 0.41034984588623047,
+      "step": 196
+    },
+    {
+      "completion_length": 444.8494815826416,
+      "epoch": 0.5133550488599349,
+      "grad_norm": 0.011833498254418373,
+      "learning_rate": 5.194677596068689e-06,
+      "loss": -0.0114,
+      "reward": 0.9045189321041107,
+      "reward_std": 0.086035909014754,
+      "rewards/accuracy_reward": 0.4910714216530323,
+      "rewards/semantic_entropy_math_reward": 0.413447517901659,
+      "step": 197
+    },
+    {
+      "completion_length": 444.4464168548584,
+      "epoch": 0.5159609120521172,
+      "grad_norm": 0.00945763848721981,
+      "learning_rate": 5.1514310316006835e-06,
+      "loss": -0.0135,
+      "reward": 0.9529883228242397,
+      "reward_std": 0.08475825749337673,
+      "rewards/accuracy_reward": 0.5350765138864517,
+      "rewards/semantic_entropy_math_reward": 0.4179117977619171,
+      "step": 198
+    },
+    {
+      "completion_length": 471.7232036590576,
+      "epoch": 0.5185667752442997,
+      "grad_norm": 0.009042466059327126,
+      "learning_rate": 5.1081731248970435e-06,
+      "loss": -0.0201,
+      "reward": 0.953717190772295,
+      "reward_std": 0.09949397918535396,
+      "rewards/accuracy_reward": 0.5446428433060646,
+      "rewards/semantic_entropy_math_reward": 0.4090743362903595,
+      "step": 199
+    },
+    {
+      "completion_length": 445.26083183288574,
+      "epoch": 0.5211726384364821,
+      "grad_norm": 0.012422044761478901,
+      "learning_rate": 5.064907115989655e-06,
+      "loss": -0.0276,
+      "reward": 0.9502550885081291,
+      "reward_std": 0.11640744589385577,
+      "rewards/accuracy_reward": 0.5471938643604517,
+      "rewards/semantic_entropy_math_reward": 0.4030612148344517,
+      "step": 200
+    },
+    {
+      "completion_length": 475.43494415283203,
+      "epoch": 0.5237785016286645,
+      "grad_norm": 0.013881414197385311,
+      "learning_rate": 5.021636245517261e-06,
+      "loss": -0.019,
+      "reward": 0.9939868673682213,
+      "reward_std": 0.1037318678572774,
+      "rewards/accuracy_reward": 0.5790816228836775,
+      "rewards/semantic_entropy_math_reward": 0.4149052333086729,
+      "step": 201
+    },
+    {
+      "completion_length": 469.54272079467773,
+      "epoch": 0.5263843648208469,
+      "grad_norm": 0.013678316958248615,
+      "learning_rate": 4.978363754482741e-06,
+      "loss": -0.0194,
+      "reward": 0.8790087252855301,
+      "reward_std": 0.1069271593587473,
+      "rewards/accuracy_reward": 0.48915815725922585,
+      "rewards/semantic_entropy_math_reward": 0.38985056802630424,
+      "step": 202
+    },
+    {
+      "completion_length": 471.13008880615234,
+      "epoch": 0.5289902280130293,
+      "grad_norm": 0.012411783449351788,
+      "learning_rate": 4.935092884010347e-06,
+      "loss": -0.0129,
+      "reward": 0.9805028922855854,
+      "reward_std": 0.09484815062023699,
+      "rewards/accuracy_reward": 0.5695152897387743,
+      "rewards/semantic_entropy_math_reward": 0.41098759695887566,
+      "step": 203
+    },
+    {
+      "completion_length": 494.3297119140625,
+      "epoch": 0.5315960912052117,
+      "grad_norm": 0.009701291099190712,
+      "learning_rate": 4.891826875102958e-06,
+      "loss": -0.0205,
+      "reward": 0.9369533248245716,
+      "reward_std": 0.11289398185908794,
+      "rewards/accuracy_reward": 0.5414540693163872,
+      "rewards/semantic_entropy_math_reward": 0.3954992648214102,
+      "step": 204
+    },
+    {
+      "completion_length": 433.42409896850586,
+      "epoch": 0.5342019543973942,
+      "grad_norm": 0.011294050142168999,
+      "learning_rate": 4.848568968399317e-06,
+      "loss": -0.0136,
+      "reward": 0.9180940166115761,
+      "reward_std": 0.10581530025228858,
+      "rewards/accuracy_reward": 0.5121173374354839,
+      "rewards/semantic_entropy_math_reward": 0.4059766698628664,
+      "step": 205
+    },
+    {
+      "completion_length": 444.698335647583,
+      "epoch": 0.5368078175895765,
+      "grad_norm": 0.008201202377676964,
+      "learning_rate": 4.805322403931312e-06,
+      "loss": -0.0129,
+      "reward": 1.0149416662752628,
+      "reward_std": 0.08048964510089718,
+      "rewards/accuracy_reward": 0.5956632532179356,
+      "rewards/semantic_entropy_math_reward": 0.419278422370553,
+      "step": 206
+    },
+    {
+      "completion_length": 476.51465797424316,
+      "epoch": 0.539413680781759,
+      "grad_norm": 0.011591507121920586,
+      "learning_rate": 4.762090420881289e-06,
+      "loss": -0.0122,
+      "reward": 0.9323979467153549,
+      "reward_std": 0.11307998397387564,
+      "rewards/accuracy_reward": 0.5216836631298065,
+      "rewards/semantic_entropy_math_reward": 0.41071428172290325,
+      "step": 207
+    },
+    {
+      "completion_length": 476.31631660461426,
+      "epoch": 0.5420195439739414,
+      "grad_norm": 0.008181254379451275,
+      "learning_rate": 4.718876257339444e-06,
+      "loss": -0.0196,
+      "reward": 0.9384110532701015,
+      "reward_std": 0.08237648417707533,
+      "rewards/accuracy_reward": 0.5223214160650969,
+      "rewards/semantic_entropy_math_reward": 0.41608964651823044,
+      "step": 208
+    },
+    {
+      "completion_length": 490.5790729522705,
+      "epoch": 0.5446254071661237,
+      "grad_norm": 0.009356926195323467,
+      "learning_rate": 4.6756831500612846e-06,
+      "loss": -0.0217,
+      "reward": 0.9779518954455853,
+      "reward_std": 0.10836848570033908,
+      "rewards/accuracy_reward": 0.5561224352568388,
+      "rewards/semantic_entropy_math_reward": 0.4218294396996498,
+      "step": 209
+    },
+    {
+      "completion_length": 516.4917030334473,
+      "epoch": 0.5472312703583062,
+      "grad_norm": 0.011323334649205208,
+      "learning_rate": 4.632514334225201e-06,
+      "loss": -0.0184,
+      "reward": 0.9516217112541199,
+      "reward_std": 0.09244993596803397,
+      "rewards/accuracy_reward": 0.5248724445700645,
+      "rewards/semantic_entropy_math_reward": 0.42674924805760384,
+      "step": 210
+    },
+    {
+      "completion_length": 492.48659896850586,
+      "epoch": 0.5498371335504886,
+      "grad_norm": 0.009480936452746391,
+      "learning_rate": 4.589373043190137e-06,
+      "loss": -0.0143,
+      "reward": 0.9234693683683872,
+      "reward_std": 0.10728166525950655,
+      "rewards/accuracy_reward": 0.5140305981040001,
+      "rewards/semantic_entropy_math_reward": 0.4094387833029032,
+      "step": 211
+    },
+    {
+      "completion_length": 491.74297523498535,
+      "epoch": 0.552442996742671,
+      "grad_norm": 0.010586866177618504,
+      "learning_rate": 4.546262508253429e-06,
+      "loss": -0.0127,
+      "reward": 0.9772230423986912,
+      "reward_std": 0.10844850935973227,
+      "rewards/accuracy_reward": 0.5612244736403227,
+      "rewards/semantic_entropy_math_reward": 0.4159985277801752,
+      "step": 212
+    },
+    {
+      "completion_length": 458.1702690124512,
+      "epoch": 0.5550488599348534,
+      "grad_norm": 0.00833533238619566,
+      "learning_rate": 4.503185958408767e-06,
+      "loss": -0.021,
+      "reward": 1.0099307373166084,
+      "reward_std": 0.10288633638992906,
+      "rewards/accuracy_reward": 0.5854591727256775,
+      "rewards/semantic_entropy_math_reward": 0.4244715701788664,
+      "step": 213
+    },
+    {
+      "completion_length": 514.4853172302246,
+      "epoch": 0.5576547231270358,
+      "grad_norm": 0.007394696585834026,
+      "learning_rate": 4.460146620104347e-06,
+      "loss": -0.0224,
+      "reward": 0.9210094511508942,
+      "reward_std": 0.11905729840509593,
+      "rewards/accuracy_reward": 0.5178571306169033,
+      "rewards/semantic_entropy_math_reward": 0.4031523112207651,
+      "step": 214
+    },
+    {
+      "completion_length": 476.78442573547363,
+      "epoch": 0.5602605863192183,
+      "grad_norm": 0.009832123294472694,
+      "learning_rate": 4.417147717001205e-06,
+      "loss": -0.0229,
+      "reward": 0.8485787101089954,
+      "reward_std": 0.09484727546805516,
+      "rewards/accuracy_reward": 0.445790808647871,
+      "rewards/semantic_entropy_math_reward": 0.40278790704905987,
+      "step": 215
+    },
+    {
+      "completion_length": 459.61988639831543,
+      "epoch": 0.5628664495114006,
+      "grad_norm": 0.012813005596399307,
+      "learning_rate": 4.374192469731771e-06,
+      "loss": -0.0134,
+      "reward": 0.9386843740940094,
+      "reward_std": 0.095578909793403,
+      "rewards/accuracy_reward": 0.5223214216530323,
+      "rewards/semantic_entropy_math_reward": 0.41636295430362225,
+      "step": 216
+    },
+    {
+      "completion_length": 434.995530128479,
+      "epoch": 0.5654723127035831,
+      "grad_norm": 0.0082651786506176,
+      "learning_rate": 4.331284095658637e-06,
+      "loss": -0.0164,
+      "reward": 0.981778409332037,
+      "reward_std": 0.09468310995725915,
+      "rewards/accuracy_reward": 0.5618622284382582,
+      "rewards/semantic_entropy_math_reward": 0.41991617158055305,
+      "step": 217
+    },
+    {
+      "completion_length": 474.4394016265869,
+      "epoch": 0.5680781758957655,
+      "grad_norm": 0.010903375223279,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": -0.007,
+      "reward": 0.9291180595755577,
+      "reward_std": 0.11282830289565027,
+      "rewards/accuracy_reward": 0.5184948835521936,
+      "rewards/semantic_entropy_math_reward": 0.4106231704354286,
+      "step": 218
+    },
+    {
+      "completion_length": 482.2563667297363,
+      "epoch": 0.5706840390879478,
+      "grad_norm": 0.0167391300201416,
+      "learning_rate": 4.245620818756822e-06,
+      "loss": -0.0106,
+      "reward": 0.931213553994894,
+      "reward_std": 0.08836283767595887,
+      "rewards/accuracy_reward": 0.5153061132878065,
+      "rewards/semantic_entropy_math_reward": 0.41590742766857147,
+      "step": 219
+    },
+    {
+      "completion_length": 544.2040729522705,
+      "epoch": 0.5732899022801303,
+      "grad_norm": 0.009337971918284893,
+      "learning_rate": 4.202872332136639e-06,
+      "loss": -0.0152,
+      "reward": 0.9517128244042397,
+      "reward_std": 0.09688535577151924,
+      "rewards/accuracy_reward": 0.533163245767355,
+      "rewards/semantic_entropy_math_reward": 0.4185495525598526,
+      "step": 220
+    },
+    {
+      "completion_length": 641.5816173553467,
+      "epoch": 0.5758957654723127,
+      "grad_norm": 0.008485456928610802,
+      "learning_rate": 4.160183550649176e-06,
+      "loss": -0.0199,
+      "reward": 0.9877004250884056,
+      "reward_std": 0.0827121082256781,
+      "rewards/accuracy_reward": 0.5637754965573549,
+      "rewards/semantic_entropy_math_reward": 0.4239249173551798,
+      "step": 221
+    },
+    {
+      "completion_length": 570.7506313323975,
+      "epoch": 0.5785016286644951,
+      "grad_norm": 0.010655037127435207,
+      "learning_rate": 4.117557671698648e-06,
+      "loss": -0.0159,
+      "reward": 0.9324890524148941,
+      "reward_std": 0.11734752007760108,
+      "rewards/accuracy_reward": 0.5325254965573549,
+      "rewards/semantic_entropy_math_reward": 0.399963553994894,
+      "step": 222
+    },
+    {
+      "completion_length": 640.9151592254639,
+      "epoch": 0.5811074918566775,
+      "grad_norm": 0.018679160624742508,
+      "learning_rate": 4.074997887977843e-06,
+      "loss": -0.0209,
+      "reward": 0.9269314669072628,
+      "reward_std": 0.09998149517923594,
+      "rewards/accuracy_reward": 0.5242346804589033,
+      "rewards/semantic_entropy_math_reward": 0.4026967938989401,
+      "step": 223
+    },
+    {
+      "completion_length": 558.047176361084,
+      "epoch": 0.5837133550488599,
+      "grad_norm": 0.009510348550975323,
+      "learning_rate": 4.032507387229002e-06,
+      "loss": -0.0118,
+      "reward": 0.9808673299849033,
+      "reward_std": 0.12139773485250771,
+      "rewards/accuracy_reward": 0.5605867244303226,
+      "rewards/semantic_entropy_math_reward": 0.4202805943787098,
+      "step": 224
+    },
+    {
+      "completion_length": 568.3686084747314,
+      "epoch": 0.5863192182410424,
+      "grad_norm": 0.009974831715226173,
+      "learning_rate": 3.9900893520050446e-06,
+      "loss": -0.0176,
+      "reward": 0.9728498421609402,
+      "reward_std": 0.1163054957287386,
+      "rewards/accuracy_reward": 0.561224477365613,
+      "rewards/semantic_entropy_math_reward": 0.4116253536194563,
+      "step": 225
+    },
+    {
+      "completion_length": 528.923454284668,
+      "epoch": 0.5889250814332248,
+      "grad_norm": 0.012413197197020054,
+      "learning_rate": 3.9477469594311975e-06,
+      "loss": -0.0071,
+      "reward": 0.9459730163216591,
+      "reward_std": 0.13538508489727974,
+      "rewards/accuracy_reward": 0.5491071380674839,
+      "rewards/semantic_entropy_math_reward": 0.3968658857047558,
+      "step": 226
+    },
+    {
+      "completion_length": 507.6760063171387,
+      "epoch": 0.5915309446254071,
+      "grad_norm": 0.009683658368885517,
+      "learning_rate": 3.905483380967027e-06,
+      "loss": -0.0079,
+      "reward": 0.9868804439902306,
+      "reward_std": 0.09878062365169171,
+      "rewards/accuracy_reward": 0.5644132550805807,
+      "rewards/semantic_entropy_math_reward": 0.42246719636023045,
+      "step": 227
+    },
+    {
+      "completion_length": 537.5465469360352,
+      "epoch": 0.5941368078175896,
+      "grad_norm": 0.018190421164035797,
+      "learning_rate": 3.863301782168896e-06,
+      "loss": -0.0161,
+      "reward": 0.8993257842957973,
+      "reward_std": 0.10135983489453793,
+      "rewards/accuracy_reward": 0.4910714142024517,
+      "rewards/semantic_entropy_math_reward": 0.40825437754392624,
+      "step": 228
+    },
+    {
+      "completion_length": 543.8405494689941,
+      "epoch": 0.596742671009772,
+      "grad_norm": 0.011154105886816978,
+      "learning_rate": 3.821205322452863e-06,
+      "loss": -0.0188,
+      "reward": 0.8654336482286453,
+      "reward_std": 0.1210846480098553,
+      "rewards/accuracy_reward": 0.47576529532670975,
+      "rewards/semantic_entropy_math_reward": 0.3896683640778065,
+      "step": 229
+    },
+    {
+      "completion_length": 499.15688133239746,
+      "epoch": 0.5993485342019544,
+      "grad_norm": 0.010389345698058605,
+      "learning_rate": 3.779197154858044e-06,
+      "loss": -0.0076,
+      "reward": 0.8955903425812721,
+      "reward_std": 0.0861530471011065,
+      "rewards/accuracy_reward": 0.4980867300182581,
+      "rewards/semantic_entropy_math_reward": 0.39750363677740097,
+      "step": 230
+    },
+    {
+      "completion_length": 535.5503730773926,
+      "epoch": 0.6019543973941368,
+      "grad_norm": 0.011222809553146362,
+      "learning_rate": 3.7372804258104367e-06,
+      "loss": -0.0143,
+      "reward": 1.011935107409954,
+      "reward_std": 0.08841318869963288,
+      "rewards/accuracy_reward": 0.5835459046065807,
+      "rewards/semantic_entropy_math_reward": 0.4283891972154379,
+      "step": 231
+    },
+    {
+      "completion_length": 420.7837963104248,
+      "epoch": 0.6045602605863192,
+      "grad_norm": 0.01220863126218319,
+      "learning_rate": 3.695458274887268e-06,
+      "loss": -0.0126,
+      "reward": 0.9781341105699539,
+      "reward_std": 0.09525648917770013,
+      "rewards/accuracy_reward": 0.5720663126558065,
+      "rewards/semantic_entropy_math_reward": 0.40606776997447014,
+      "step": 232
+    },
+    {
+      "completion_length": 444.399866104126,
+      "epoch": 0.6071661237785017,
+      "grad_norm": 0.008971706964075565,
+      "learning_rate": 3.6537338345818273e-06,
+      "loss": -0.0234,
+      "reward": 0.9351311847567558,
+      "reward_std": 0.10607956547755748,
+      "rewards/accuracy_reward": 0.5248724352568388,
+      "rewards/semantic_entropy_math_reward": 0.41025873832404613,
+      "step": 233
+    },
+    {
+      "completion_length": 408.4878730773926,
+      "epoch": 0.609771986970684,
+      "grad_norm": 0.009326077997684479,
+      "learning_rate": 3.6121102300688504e-06,
+      "loss": -0.0136,
+      "reward": 1.0050109177827835,
+      "reward_std": 0.08207120059523731,
+      "rewards/accuracy_reward": 0.5695152934640646,
+      "rewards/semantic_entropy_math_reward": 0.43549559637904167,
+      "step": 234
+    },
+    {
+      "completion_length": 430.00828742980957,
+      "epoch": 0.6123778501628665,
+      "grad_norm": 0.01133815012872219,
+      "learning_rate": 3.5705905789704296e-06,
+      "loss": -0.0174,
+      "reward": 0.9181851297616959,
+      "reward_std": 0.10594801499973983,
+      "rewards/accuracy_reward": 0.5210459036752582,
+      "rewards/semantic_entropy_math_reward": 0.39713920652866364,
+      "step": 235
+    },
+    {
+      "completion_length": 408.0299701690674,
+      "epoch": 0.6149837133550489,
+      "grad_norm": 0.008568109013140202,
+      "learning_rate": 3.529177991122519e-06,
+      "loss": -0.0131,
+      "reward": 0.958545908331871,
+      "reward_std": 0.08321992680430412,
+      "rewards/accuracy_reward": 0.5535714142024517,
+      "rewards/semantic_entropy_math_reward": 0.4049744922667742,
+      "step": 236
+    },
+    {
+      "completion_length": 436.0191307067871,
+      "epoch": 0.6175895765472312,
+      "grad_norm": 0.015509643591940403,
+      "learning_rate": 3.487875568341995e-06,
+      "loss": -0.0115,
+      "reward": 0.8883017376065254,
+      "reward_std": 0.09416370574035682,
+      "rewards/accuracy_reward": 0.4859693758189678,
+      "rewards/semantic_entropy_math_reward": 0.4023323506116867,
+      "step": 237
+    },
+    {
+      "completion_length": 350.04272270202637,
+      "epoch": 0.6201954397394137,
+      "grad_norm": 0.012752141803503036,
+      "learning_rate": 3.446686404194337e-06,
+      "loss": -0.014,
+      "reward": 0.9876093231141567,
+      "reward_std": 0.08275497186696157,
+      "rewards/accuracy_reward": 0.5644132569432259,
+      "rewards/semantic_entropy_math_reward": 0.4231960419565439,
+      "step": 238
+    },
+    {
+      "completion_length": 392.4043254852295,
+      "epoch": 0.6228013029315961,
+      "grad_norm": 0.014255838468670845,
+      "learning_rate": 3.4056135837619077e-06,
+      "loss": -0.0111,
+      "reward": 1.0281523205339909,
+      "reward_std": 0.06774875608971342,
+      "rewards/accuracy_reward": 0.5835459064692259,
+      "rewards/semantic_entropy_math_reward": 0.4446064066141844,
+      "step": 239
+    },
+    {
+      "completion_length": 403.4055976867676,
+      "epoch": 0.6254071661237784,
+      "grad_norm": 0.013257958926260471,
+      "learning_rate": 3.3646601834128924e-06,
+      "loss": -0.0138,
+      "reward": 0.9554482288658619,
+      "reward_std": 0.09038078342564404,
+      "rewards/accuracy_reward": 0.5414540711790323,
+      "rewards/semantic_entropy_math_reward": 0.41399415768682957,
+      "step": 240
+    },
+    {
+      "completion_length": 422.57716178894043,
+      "epoch": 0.6280130293159609,
+      "grad_norm": 0.009931429289281368,
+      "learning_rate": 3.3238292705708675e-06,
+      "loss": -0.0042,
+      "reward": 0.9145408011972904,
+      "reward_std": 0.08421986317262053,
+      "rewards/accuracy_reward": 0.4961734600365162,
+      "rewards/semantic_entropy_math_reward": 0.4183673318475485,
+      "step": 241
+    },
+    {
+      "completion_length": 385.666446685791,
+      "epoch": 0.6306188925081433,
+      "grad_norm": 0.020119503140449524,
+      "learning_rate": 3.2831239034850593e-06,
+      "loss": -0.0077,
+      "reward": 0.960094753652811,
+      "reward_std": 0.10691097256494686,
+      "rewards/accuracy_reward": 0.5389030482620001,
+      "rewards/semantic_entropy_math_reward": 0.4211916830390692,
+      "step": 242
+    },
+    {
+      "completion_length": 401.9961643218994,
+      "epoch": 0.6332247557003258,
+      "grad_norm": 0.015258848667144775,
+      "learning_rate": 3.2425471310012645e-06,
+      "loss": -0.0073,
+      "reward": 0.921100564301014,
+      "reward_std": 0.09436794393695891,
+      "rewards/accuracy_reward": 0.5204081535339355,
+      "rewards/semantic_entropy_math_reward": 0.40069241262972355,
+      "step": 243
+    },
+    {
+      "completion_length": 425.9100704193115,
+      "epoch": 0.6358306188925081,
+      "grad_norm": 0.009504982270300388,
+      "learning_rate": 3.2021019923335093e-06,
+      "loss": -0.0087,
+      "reward": 0.8651603311300278,
+      "reward_std": 0.05741609656251967,
+      "rewards/accuracy_reward": 0.4534438643604517,
+      "rewards/semantic_entropy_math_reward": 0.4117164649069309,
+      "step": 244
+    },
+    {
+      "completion_length": 391.7308597564697,
+      "epoch": 0.6384364820846905,
+      "grad_norm": 0.008338342420756817,
+      "learning_rate": 3.1617915168363994e-06,
+      "loss": -0.0048,
+      "reward": 0.9585458934307098,
+      "reward_std": 0.09049107087776065,
+      "rewards/accuracy_reward": 0.5363520346581936,
+      "rewards/semantic_entropy_math_reward": 0.42219386994838715,
+      "step": 245
+    },
+    {
+      "completion_length": 426.6033020019531,
+      "epoch": 0.641042345276873,
+      "grad_norm": 0.018495287746191025,
+      "learning_rate": 3.121618723778225e-06,
+      "loss": -0.0197,
+      "reward": 0.9272047840058804,
+      "reward_std": 0.09722855128347874,
+      "rewards/accuracy_reward": 0.5248724352568388,
+      "rewards/semantic_entropy_math_reward": 0.4023323468863964,
+      "step": 246
+    },
+    {
+      "completion_length": 403.5931053161621,
+      "epoch": 0.6436482084690553,
+      "grad_norm": 0.011493879370391369,
+      "learning_rate": 3.081586622114809e-06,
+      "loss": -0.0171,
+      "reward": 0.9197339601814747,
+      "reward_std": 0.09083502274006605,
+      "rewards/accuracy_reward": 0.5076530463993549,
+      "rewards/semantic_entropy_math_reward": 0.41208089888095856,
+      "step": 247
+    },
+    {
+      "completion_length": 390.2971897125244,
+      "epoch": 0.6462540716612378,
+      "grad_norm": 0.012770140543580055,
+      "learning_rate": 3.041698210264149e-06,
+      "loss": -0.0035,
+      "reward": 0.9225582852959633,
+      "reward_std": 0.10072695807320997,
+      "rewards/accuracy_reward": 0.5133928395807743,
+      "rewards/semantic_entropy_math_reward": 0.40916544012725353,
+      "step": 248
+    },
+    {
+      "completion_length": 386.1511402130127,
+      "epoch": 0.6488599348534202,
+      "grad_norm": 0.011112012900412083,
+      "learning_rate": 3.001956475881822e-06,
+      "loss": -0.0101,
+      "reward": 0.9212827943265438,
+      "reward_std": 0.11616118880920112,
+      "rewards/accuracy_reward": 0.5019132569432259,
+      "rewards/semantic_entropy_math_reward": 0.4193695206195116,
+      "step": 249
+    },
+    {
+      "completion_length": 416.366060256958,
+      "epoch": 0.6514657980456026,
+      "grad_norm": 0.013438215479254723,
+      "learning_rate": 2.962364395637216e-06,
+      "loss": -0.0105,
+      "reward": 0.938411071896553,
+      "reward_std": 0.09871607949025929,
+      "rewards/accuracy_reward": 0.5344387628138065,
+      "rewards/semantic_entropy_math_reward": 0.4039722979068756,
+      "step": 250
+    },
+    {
+      "completion_length": 411.18238830566406,
+      "epoch": 0.654071661237785,
+      "grad_norm": 0.014477844350039959,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": -0.0185,
+      "reward": 0.9585458822548389,
+      "reward_std": 0.07620244607096538,
+      "rewards/accuracy_reward": 0.5350765157490969,
+      "rewards/semantic_entropy_math_reward": 0.4234693832695484,
+      "step": 251
+    },
+    {
+      "completion_length": 429.57652282714844,
+      "epoch": 0.6566775244299674,
+      "grad_norm": 0.009758449159562588,
+      "learning_rate": 2.8836410479708625e-06,
+      "loss": -0.0066,
+      "reward": 0.9825072661042213,
+      "reward_std": 0.0784962668112712,
+      "rewards/accuracy_reward": 0.5567601975053549,
+      "rewards/semantic_entropy_math_reward": 0.4257470816373825,
+      "step": 252
+    },
+    {
+      "completion_length": 410.80356216430664,
+      "epoch": 0.6592833876221499,
+      "grad_norm": 0.011529424227774143,
+      "learning_rate": 2.84451567695456e-06,
+      "loss": -0.0155,
+      "reward": 0.9254737384617329,
+      "reward_std": 0.08181263762526214,
+      "rewards/accuracy_reward": 0.5159438662230968,
+      "rewards/semantic_entropy_math_reward": 0.409529872238636,
+      "step": 253
+    },
+    {
+      "completion_length": 418.6396598815918,
+      "epoch": 0.6618892508143323,
+      "grad_norm": 0.012521790340542793,
+      "learning_rate": 2.805551752445222e-06,
+      "loss": -0.0072,
+      "reward": 0.9676566906273365,
+      "reward_std": 0.07710790610872209,
+      "rewards/accuracy_reward": 0.5484693758189678,
+      "rewards/semantic_entropy_math_reward": 0.41918731294572353,
+      "step": 254
+    },
+    {
+      "completion_length": 418.7059841156006,
+      "epoch": 0.6644951140065146,
+      "grad_norm": 0.010554789565503597,
+      "learning_rate": 2.766752192854012e-06,
+      "loss": -0.0079,
+      "reward": 0.9526238888502121,
+      "reward_std": 0.10265779460314661,
+      "rewards/accuracy_reward": 0.5369897820055485,
+      "rewards/semantic_entropy_math_reward": 0.41563410870730877,
+      "step": 255
+    },
+    {
+      "completion_length": 408.3431034088135,
+      "epoch": 0.6671009771986971,
+      "grad_norm": 0.009969337843358517,
+      "learning_rate": 2.728119904281105e-06,
+      "loss": -0.0084,
+      "reward": 0.986880462616682,
+      "reward_std": 0.09043557988479733,
+      "rewards/accuracy_reward": 0.5682397838681936,
+      "rewards/semantic_entropy_math_reward": 0.41864065267145634,
+      "step": 256
+    },
+    {
+      "completion_length": 420.5044631958008,
+      "epoch": 0.6697068403908795,
+      "grad_norm": 0.011664465069770813,
+      "learning_rate": 2.6896577802980184e-06,
+      "loss": -0.0099,
+      "reward": 0.9866982325911522,
+      "reward_std": 0.10724300192669034,
+      "rewards/accuracy_reward": 0.5752550922334194,
+      "rewards/semantic_entropy_math_reward": 0.4114431384950876,
+      "step": 257
+    },
+    {
+      "completion_length": 453.64093589782715,
+      "epoch": 0.6723127035830619,
+      "grad_norm": 0.00969005562365055,
+      "learning_rate": 2.651368701730889e-06,
+      "loss": -0.0053,
+      "reward": 0.9547193720936775,
+      "reward_std": 0.10146705091756303,
+      "rewards/accuracy_reward": 0.5331632532179356,
+      "rewards/semantic_entropy_math_reward": 0.4215561244636774,
+      "step": 258
+    },
+    {
+      "completion_length": 482.51912689208984,
+      "epoch": 0.6749185667752443,
+      "grad_norm": 0.009962331503629684,
+      "learning_rate": 2.6132555364446856e-06,
+      "loss": -0.0089,
+      "reward": 0.8847485221922398,
+      "reward_std": 0.10891495714895427,
+      "rewards/accuracy_reward": 0.4904336640611291,
+      "rewards/semantic_entropy_math_reward": 0.3943148609250784,
+      "step": 259
+    },
+    {
+      "completion_length": 472.77039527893066,
+      "epoch": 0.6775244299674267,
+      "grad_norm": 0.008120470680296421,
+      "learning_rate": 2.5753211391284172e-06,
+      "loss": -0.0056,
+      "reward": 0.9668367207050323,
+      "reward_std": 0.08560021687299013,
+      "rewards/accuracy_reward": 0.5433673411607742,
+      "rewards/semantic_entropy_math_reward": 0.42346938140690327,
+      "step": 260
+    },
+    {
+      "completion_length": 455.7142734527588,
+      "epoch": 0.6801302931596092,
+      "grad_norm": 0.008612104691565037,
+      "learning_rate": 2.537568351081311e-06,
+      "loss": -0.0055,
+      "reward": 0.9637390524148941,
+      "reward_std": 0.08404424658510834,
+      "rewards/accuracy_reward": 0.5580357015132904,
+      "rewards/semantic_entropy_math_reward": 0.4057033322751522,
+      "step": 261
+    },
+    {
+      "completion_length": 452.24807357788086,
+      "epoch": 0.6827361563517915,
+      "grad_norm": 0.007962207309901714,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": -0.0213,
+      "reward": 1.033163245767355,
+      "reward_std": 0.08445578732062131,
+      "rewards/accuracy_reward": 0.6065050885081291,
+      "rewards/semantic_entropy_math_reward": 0.4266581479460001,
+      "step": 262
+    },
+    {
+      "completion_length": 445.8513927459717,
+      "epoch": 0.6853420195439739,
+      "grad_norm": 0.00862111710011959,
+      "learning_rate": 2.4626188997667224e-06,
+      "loss": -0.0107,
+      "reward": 0.9684766530990601,
+      "reward_std": 0.106060303398408,
+      "rewards/accuracy_reward": 0.5497448798269033,
+      "rewards/semantic_entropy_math_reward": 0.41873176395893097,
+      "step": 263
+    },
+    {
+      "completion_length": 438.24169731140137,
+      "epoch": 0.6879478827361564,
+      "grad_norm": 0.007783900015056133,
+      "learning_rate": 2.425427850238565e-06,
+      "loss": -0.0095,
+      "reward": 0.9946246296167374,
+      "reward_std": 0.09433889319188893,
+      "rewards/accuracy_reward": 0.5784438736736774,
+      "rewards/semantic_entropy_math_reward": 0.4161807559430599,
+      "step": 264
+    },
+    {
+      "completion_length": 446.2295837402344,
+      "epoch": 0.6905537459283387,
+      "grad_norm": 0.008394320495426655,
+      "learning_rate": 2.388429637037753e-06,
+      "loss": 0.0002,
+      "reward": 1.0675109289586544,
+      "reward_std": 0.08982915757223964,
+      "rewards/accuracy_reward": 0.6301020346581936,
+      "rewards/semantic_entropy_math_reward": 0.43740889988839626,
+      "step": 265
+    },
+    {
+      "completion_length": 420.28698539733887,
+      "epoch": 0.6931596091205212,
+      "grad_norm": 0.011627813801169395,
+      "learning_rate": 2.3516270313430085e-06,
+      "loss": -0.0075,
+      "reward": 0.9636479578912258,
+      "reward_std": 0.09795083408243954,
+      "rewards/accuracy_reward": 0.5420918222516775,
+      "rewards/semantic_entropy_math_reward": 0.4215561132878065,
+      "step": 266
+    },
+    {
+      "completion_length": 504.16006660461426,
+      "epoch": 0.6957654723127036,
+      "grad_norm": 0.013352944515645504,
+      "learning_rate": 2.3150227896819782e-06,
+      "loss": 0.0002,
+      "reward": 0.9232871569693089,
+      "reward_std": 0.11924422497395426,
+      "rewards/accuracy_reward": 0.5172193739563227,
+      "rewards/semantic_entropy_math_reward": 0.4060677755624056,
+      "step": 267
+    },
+    {
+      "completion_length": 440.1288185119629,
+      "epoch": 0.698371335504886,
+      "grad_norm": 0.008399576880037785,
+      "learning_rate": 2.278619653724781e-06,
+      "loss": -0.005,
+      "reward": 0.9916180558502674,
+      "reward_std": 0.0687884486396797,
+      "rewards/accuracy_reward": 0.5637754965573549,
+      "rewards/semantic_entropy_math_reward": 0.4278425555676222,
+      "step": 268
+    },
+    {
+      "completion_length": 496.5880012512207,
+      "epoch": 0.7009771986970684,
+      "grad_norm": 0.01074276678264141,
+      "learning_rate": 2.2424203500786473e-06,
+      "loss": -0.0021,
+      "reward": 0.9390488304197788,
+      "reward_std": 0.10012663877569139,
+      "rewards/accuracy_reward": 0.5357142742723227,
+      "rewards/semantic_entropy_math_reward": 0.403334541246295,
+      "step": 269
+    },
+    {
+      "completion_length": 462.8348159790039,
+      "epoch": 0.7035830618892508,
+      "grad_norm": 0.012036530300974846,
+      "learning_rate": 2.206427590083703e-06,
+      "loss": -0.0068,
+      "reward": 0.9727587252855301,
+      "reward_std": 0.07587114511989057,
+      "rewards/accuracy_reward": 0.5516581516712904,
+      "rewards/semantic_entropy_math_reward": 0.42110058292746544,
+      "step": 270
+    },
+    {
+      "completion_length": 462.12371253967285,
+      "epoch": 0.7061889250814333,
+      "grad_norm": 0.00853002816438675,
+      "learning_rate": 2.170644069609876e-06,
+      "loss": -0.0111,
+      "reward": 0.9571792744100094,
+      "reward_std": 0.09096840798156336,
+      "rewards/accuracy_reward": 0.5503826402127743,
+      "rewards/semantic_entropy_math_reward": 0.4067966416478157,
+      "step": 271
+    },
+    {
+      "completion_length": 428.95662117004395,
+      "epoch": 0.7087947882736156,
+      "grad_norm": 0.009711945429444313,
+      "learning_rate": 2.1350724688549906e-06,
+      "loss": -0.0152,
+      "reward": 0.9361333549022675,
+      "reward_std": 0.08675176231190562,
+      "rewards/accuracy_reward": 0.515306118875742,
+      "rewards/semantic_entropy_math_reward": 0.4208272397518158,
+      "step": 272
+    },
+    {
+      "completion_length": 462.191312789917,
+      "epoch": 0.711400651465798,
+      "grad_norm": 0.013711754232645035,
+      "learning_rate": 2.09971545214401e-06,
+      "loss": -0.0046,
+      "reward": 0.9466107748448849,
+      "reward_std": 0.09364626772003248,
+      "rewards/accuracy_reward": 0.5338010117411613,
+      "rewards/semantic_entropy_math_reward": 0.4128097593784332,
+      "step": 273
+    },
+    {
+      "completion_length": 483.0580234527588,
+      "epoch": 0.7140065146579805,
+      "grad_norm": 0.008413317613303661,
+      "learning_rate": 2.0645756677294788e-06,
+      "loss": -0.0117,
+      "reward": 0.9661078602075577,
+      "reward_std": 0.07902665832079947,
+      "rewards/accuracy_reward": 0.5414540749043226,
+      "rewards/semantic_entropy_math_reward": 0.4246537797152996,
+      "step": 274
+    },
+    {
+      "completion_length": 485.255090713501,
+      "epoch": 0.7166123778501629,
+      "grad_norm": 0.01059303991496563,
+      "learning_rate": 2.029655747593169e-06,
+      "loss": -0.0064,
+      "reward": 0.9963556602597237,
+      "reward_std": 0.08435712786740623,
+      "rewards/accuracy_reward": 0.5797193804755807,
+      "rewards/semantic_entropy_math_reward": 0.41663630679249763,
+      "step": 275
+    },
+    {
+      "completion_length": 524.714916229248,
+      "epoch": 0.7192182410423453,
+      "grad_norm": 0.010354903526604176,
+      "learning_rate": 1.9949583072489455e-06,
+      "loss": -0.0095,
+      "reward": 0.9393221400678158,
+      "reward_std": 0.10403139912523329,
+      "rewards/accuracy_reward": 0.536352027207613,
+      "rewards/semantic_entropy_math_reward": 0.4029701128602028,
+      "step": 276
+    },
+    {
+      "completion_length": 493.7672061920166,
+      "epoch": 0.7218241042345277,
+      "grad_norm": 0.009775525890290737,
+      "learning_rate": 1.9604859455468587e-06,
+      "loss": -0.0051,
+      "reward": 0.9140852615237236,
+      "reward_std": 0.07866077253129333,
+      "rewards/accuracy_reward": 0.48405610769987106,
+      "rewards/semantic_entropy_math_reward": 0.4300291370600462,
+      "step": 277
+    },
+    {
+      "completion_length": 486.60011863708496,
+      "epoch": 0.7244299674267101,
+      "grad_norm": 0.010633684694766998,
+      "learning_rate": 1.926241244478496e-06,
+      "loss": -0.0114,
+      "reward": 0.9836916588246822,
+      "reward_std": 0.13731968333013356,
+      "rewards/accuracy_reward": 0.5714285597205162,
+      "rewards/semantic_entropy_math_reward": 0.4122631140053272,
+      "step": 278
+    },
+    {
+      "completion_length": 490.8520317077637,
+      "epoch": 0.7270358306188925,
+      "grad_norm": 0.01332115475088358,
+      "learning_rate": 1.8922267689835806e-06,
+      "loss": -0.0069,
+      "reward": 0.9645590335130692,
+      "reward_std": 0.08677118818741292,
+      "rewards/accuracy_reward": 0.5395408049225807,
+      "rewards/semantic_entropy_math_reward": 0.42501820996403694,
+      "step": 279
+    },
+    {
+      "completion_length": 510.359037399292,
+      "epoch": 0.7296416938110749,
+      "grad_norm": 0.010328337550163269,
+      "learning_rate": 1.8584450667578656e-06,
+      "loss": -0.0053,
+      "reward": 0.9415087439119816,
+      "reward_std": 0.1030702197458595,
+      "rewards/accuracy_reward": 0.5414540711790323,
+      "rewards/semantic_entropy_math_reward": 0.4000546522438526,
+      "step": 280
+    },
+    {
+      "completion_length": 488.9508819580078,
+      "epoch": 0.7322475570032573,
+      "grad_norm": 0.009401313029229641,
+      "learning_rate": 1.8248986680623077e-06,
+      "loss": -0.0144,
+      "reward": 1.0168549418449402,
+      "reward_std": 0.08474622888024896,
+      "rewards/accuracy_reward": 0.5988520290702581,
+      "rewards/semantic_entropy_math_reward": 0.41800292022526264,
+      "step": 281
+    },
+    {
+      "completion_length": 461.94578552246094,
+      "epoch": 0.7348534201954398,
+      "grad_norm": 0.01206615474075079,
+      "learning_rate": 1.7915900855335506e-06,
+      "loss": -0.0079,
+      "reward": 1.0083818957209587,
+      "reward_std": 0.06797894753981382,
+      "rewards/accuracy_reward": 0.5841836612671614,
+      "rewards/semantic_entropy_math_reward": 0.42419823817908764,
+      "step": 282
+    },
+    {
+      "completion_length": 493.0318717956543,
+      "epoch": 0.7374592833876221,
+      "grad_norm": 0.009084763005375862,
+      "learning_rate": 1.7585218139957205e-06,
+      "loss": -0.0093,
+      "reward": 1.0141217038035393,
+      "reward_std": 0.10102251661010087,
+      "rewards/accuracy_reward": 0.5899234563112259,
+      "rewards/semantic_entropy_math_reward": 0.42419824562966824,
+      "step": 283
+    },
+    {
+      "completion_length": 480.63264083862305,
+      "epoch": 0.7400651465798046,
+      "grad_norm": 0.01057640090584755,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": -0.0137,
+      "reward": 0.9901603423058987,
+      "reward_std": 0.11432356107980013,
+      "rewards/accuracy_reward": 0.5790816191583872,
+      "rewards/semantic_entropy_math_reward": 0.4110787082463503,
+      "step": 284
+    },
+    {
+      "completion_length": 534.7289409637451,
+      "epoch": 0.742671009771987,
+      "grad_norm": 0.011105289682745934,
+      "learning_rate": 1.6931160930069789e-06,
+      "loss": -0.0098,
+      "reward": 0.9258381761610508,
+      "reward_std": 0.0985641247825697,
+      "rewards/accuracy_reward": 0.5197703987360001,
+      "rewards/semantic_entropy_math_reward": 0.4060677830129862,
+      "step": 285
+    },
+    {
+      "completion_length": 492.93940353393555,
+      "epoch": 0.7452768729641693,
+      "grad_norm": 0.010538027621805668,
+      "learning_rate": 1.6607835424667578e-06,
+      "loss": -0.0052,
+      "reward": 0.9999999739229679,
+      "reward_std": 0.0725285234657349,
+      "rewards/accuracy_reward": 0.5707908049225807,
+      "rewards/semantic_entropy_math_reward": 0.4292091801762581,
+      "step": 286
+    },
+    {
+      "completion_length": 479.8284320831299,
+      "epoch": 0.7478827361563518,
+      "grad_norm": 0.012495830655097961,
+      "learning_rate": 1.6287011003719105e-06,
+      "loss": -0.0044,
+      "reward": 0.9080721437931061,
+      "reward_std": 0.08464997098781168,
+      "rewards/accuracy_reward": 0.498724477365613,
+      "rewards/semantic_entropy_math_reward": 0.40934765711426735,
+      "step": 287
+    },
+    {
+      "completion_length": 479.3641548156738,
+      "epoch": 0.7504885993485342,
+      "grad_norm": 0.014318572357296944,
+      "learning_rate": 1.596871169708235e-06,
+      "loss": -0.0114,
+      "reward": 0.9038812033832073,
+      "reward_std": 0.08901800989406183,
+      "rewards/accuracy_reward": 0.4942601975053549,
+      "rewards/semantic_entropy_math_reward": 0.4096209779381752,
+      "step": 288
+    },
+    {
+      "completion_length": 478.834171295166,
+      "epoch": 0.7530944625407167,
+      "grad_norm": 0.011369728483259678,
+      "learning_rate": 1.5652961345483353e-06,
+      "loss": -0.0118,
+      "reward": 0.9845116622745991,
+      "reward_std": 0.10669447341933846,
+      "rewards/accuracy_reward": 0.5752550875768065,
+      "rewards/semantic_entropy_math_reward": 0.4092565383762121,
+      "step": 289
+    },
+    {
+      "completion_length": 515.2219257354736,
+      "epoch": 0.755700325732899,
+      "grad_norm": 0.01009457278996706,
+      "learning_rate": 1.5339783598730568e-06,
+      "loss": -0.0119,
+      "reward": 0.933946780860424,
+      "reward_std": 0.09003342734649777,
+      "rewards/accuracy_reward": 0.5280612148344517,
+      "rewards/semantic_entropy_math_reward": 0.4058855511248112,
+      "step": 290
+    },
+    {
+      "completion_length": 524.5248641967773,
+      "epoch": 0.7583061889250814,
+      "grad_norm": 0.009736046195030212,
+      "learning_rate": 1.5029201913943425e-06,
+      "loss": -0.0148,
+      "reward": 0.9850582927465439,
+      "reward_std": 0.10781788837630302,
+      "rewards/accuracy_reward": 0.5733418259769678,
+      "rewards/semantic_entropy_math_reward": 0.4117164667695761,
+      "step": 291
+    },
+    {
+      "completion_length": 512.2193756103516,
+      "epoch": 0.7609120521172639,
+      "grad_norm": 0.007600969634950161,
+      "learning_rate": 1.4721239553795485e-06,
+      "loss": -0.0079,
+      "reward": 0.9823250733315945,
+      "reward_std": 0.09170323330909014,
+      "rewards/accuracy_reward": 0.5580357071012259,
+      "rewards/semantic_entropy_math_reward": 0.4242893476039171,
+      "step": 292
+    },
+    {
+      "completion_length": 450.7780475616455,
+      "epoch": 0.7635179153094462,
+      "grad_norm": 0.00899919867515564,
+      "learning_rate": 1.4415919584771999e-06,
+      "loss": -0.0121,
+      "reward": 1.015852764248848,
+      "reward_std": 0.10539426596369594,
+      "rewards/accuracy_reward": 0.5943877436220646,
+      "rewards/semantic_entropy_math_reward": 0.42146499268710613,
+      "step": 293
+    },
+    {
+      "completion_length": 484.11031913757324,
+      "epoch": 0.7661237785016287,
+      "grad_norm": 0.012872518040239811,
+      "learning_rate": 1.4113264875442201e-06,
+      "loss": -0.0119,
+      "reward": 0.949526209384203,
+      "reward_std": 0.08500090334564447,
+      "rewards/accuracy_reward": 0.5242346823215485,
+      "rewards/semantic_entropy_math_reward": 0.4252915382385254,
+      "step": 294
+    },
+    {
+      "completion_length": 507.8871078491211,
+      "epoch": 0.7687296416938111,
+      "grad_norm": 0.01095646247267723,
+      "learning_rate": 1.3813298094746491e-06,
+      "loss": -0.0146,
+      "reward": 0.9959912374615669,
+      "reward_std": 0.09144581796135753,
+      "rewards/accuracy_reward": 0.5682397857308388,
+      "rewards/semantic_entropy_math_reward": 0.4277514461427927,
+      "step": 295
+    },
+    {
+      "completion_length": 492.8201427459717,
+      "epoch": 0.7713355048859935,
+      "grad_norm": 0.007233647629618645,
+      "learning_rate": 1.35160417102985e-06,
+      "loss": -0.0137,
+      "reward": 0.96993438154459,
+      "reward_std": 0.0985939132515341,
+      "rewards/accuracy_reward": 0.5580357015132904,
+      "rewards/semantic_entropy_math_reward": 0.41189867816865444,
+      "step": 296
+    },
+    {
+      "completion_length": 479.56950759887695,
+      "epoch": 0.7739413680781759,
+      "grad_norm": 0.010265953838825226,
+      "learning_rate": 1.3221517986702249e-06,
+      "loss": -0.0049,
+      "reward": 0.9783162996172905,
+      "reward_std": 0.0974840281996876,
+      "rewards/accuracy_reward": 0.5605867244303226,
+      "rewards/semantic_entropy_math_reward": 0.41772958263754845,
+      "step": 297
+    },
+    {
+      "completion_length": 514.214277267456,
+      "epoch": 0.7765472312703583,
+      "grad_norm": 0.006785084027796984,
+      "learning_rate": 1.292974898388456e-06,
+      "loss": -0.015,
+      "reward": 0.9332179203629494,
+      "reward_std": 0.08228117297403514,
+      "rewards/accuracy_reward": 0.5102040730416775,
+      "rewards/semantic_entropy_math_reward": 0.423013836145401,
+      "step": 298
+    },
+    {
+      "completion_length": 485.87435150146484,
+      "epoch": 0.7791530944625407,
+      "grad_norm": 0.016027240082621574,
+      "learning_rate": 1.2640756555442684e-06,
+      "loss": -0.0101,
+      "reward": 0.8885750621557236,
+      "reward_std": 0.07733660298981704,
+      "rewards/accuracy_reward": 0.47895407397300005,
+      "rewards/semantic_entropy_math_reward": 0.40962098725140095,
+      "step": 299
+    },
+    {
+      "completion_length": 505.1874885559082,
+      "epoch": 0.7817589576547231,
+      "grad_norm": 0.009378643706440926,
+      "learning_rate": 1.235456234700756e-06,
+      "loss": -0.0063,
+      "reward": 0.9045189283788204,
+      "reward_std": 0.07878015120513737,
+      "rewards/accuracy_reward": 0.5012754946947098,
+      "rewards/semantic_entropy_math_reward": 0.4032434318214655,
+      "step": 300
+    },
+    {
+      "completion_length": 475.78697967529297,
+      "epoch": 0.7843648208469055,
+      "grad_norm": 0.008114474825561047,
+      "learning_rate": 1.207118779462248e-06,
+      "loss": -0.0127,
+      "reward": 1.0125728696584702,
+      "reward_std": 0.08841744478559121,
+      "rewards/accuracy_reward": 0.5873724296689034,
+      "rewards/semantic_entropy_math_reward": 0.42520042695105076,
+      "step": 301
+    },
+    {
+      "completion_length": 517.571418762207,
+      "epoch": 0.786970684039088,
+      "grad_norm": 0.0094959931448102,
+      "learning_rate": 1.1790654123137552e-06,
+      "loss": -0.0013,
+      "reward": 0.9774963334202766,
+      "reward_std": 0.08901865262305364,
+      "rewards/accuracy_reward": 0.5561224445700645,
+      "rewards/semantic_entropy_math_reward": 0.4213739037513733,
+      "step": 302
+    },
+    {
+      "completion_length": 443.41580295562744,
+      "epoch": 0.7895765472312704,
+      "grad_norm": 0.00883433036506176,
+      "learning_rate": 1.1512982344619904e-06,
+      "loss": -0.0046,
+      "reward": 1.045098390430212,
+      "reward_std": 0.06228679193009157,
+      "rewards/accuracy_reward": 0.593749986961484,
+      "rewards/semantic_entropy_math_reward": 0.4513483848422766,
+      "step": 303
+    },
+    {
+      "completion_length": 522.1760082244873,
+      "epoch": 0.7921824104234527,
+      "grad_norm": 0.008245239965617657,
+      "learning_rate": 1.1238193256779955e-06,
+      "loss": -0.0149,
+      "reward": 0.9394132420420647,
+      "reward_std": 0.09620352147612721,
+      "rewards/accuracy_reward": 0.5293367244303226,
+      "rewards/semantic_entropy_math_reward": 0.410076517611742,
+      "step": 304
+    },
+    {
+      "completion_length": 448.2991008758545,
+      "epoch": 0.7947882736156352,
+      "grad_norm": 0.012398505583405495,
+      "learning_rate": 1.0966307441413598e-06,
+      "loss": -0.0058,
+      "reward": 0.9621902219951153,
+      "reward_std": 0.08929981710389256,
+      "rewards/accuracy_reward": 0.5338010061532259,
+      "rewards/semantic_entropy_math_reward": 0.4283892009407282,
+      "step": 305
+    },
+    {
+      "completion_length": 491.30419731140137,
+      "epoch": 0.7973941368078176,
+      "grad_norm": 0.009740319103002548,
+      "learning_rate": 1.0697345262860638e-06,
+      "loss": -0.0068,
+      "reward": 0.9690233021974564,
+      "reward_std": 0.08115439751418307,
+      "rewards/accuracy_reward": 0.5459183566272259,
+      "rewards/semantic_entropy_math_reward": 0.4231049530208111,
+      "step": 306
+    },
+    {
+      "completion_length": 497.92601203918457,
+      "epoch": 0.8,
+      "grad_norm": 0.012608438730239868,
+      "learning_rate": 1.0431326866479457e-06,
+      "loss": -0.0167,
+      "reward": 0.8842018768191338,
+      "reward_std": 0.11789050558581948,
+      "rewards/accuracy_reward": 0.48724489100277424,
+      "rewards/semantic_entropy_math_reward": 0.39695698767900467,
+      "step": 307
+    },
+    {
+      "completion_length": 488.2034339904785,
+      "epoch": 0.8026058631921824,
+      "grad_norm": 0.008150148205459118,
+      "learning_rate": 1.01682721771382e-06,
+      "loss": -0.0213,
+      "reward": 1.0246902257204056,
+      "reward_std": 0.07845566415926442,
+      "rewards/accuracy_reward": 0.5778061132878065,
+      "rewards/semantic_entropy_math_reward": 0.446884099394083,
+      "step": 308
+    },
+    {
+      "completion_length": 501.41644287109375,
+      "epoch": 0.8052117263843648,
+      "grad_norm": 0.009125792421400547,
+      "learning_rate": 9.908200897722332e-07,
+      "loss": -0.0093,
+      "reward": 0.9240160211920738,
+      "reward_std": 0.09154359906096943,
+      "rewards/accuracy_reward": 0.5127550903707743,
+      "rewards/semantic_entropy_math_reward": 0.41126092337071896,
+      "step": 309
+    },
+    {
+      "completion_length": 485.86797523498535,
+      "epoch": 0.8078175895765473,
+      "grad_norm": 0.013204295188188553,
+      "learning_rate": 9.6511325076589e-07,
+      "loss": -0.0129,
+      "reward": 0.9015123657882214,
+      "reward_std": 0.08482572343200445,
+      "rewards/accuracy_reward": 0.4878826420754194,
+      "rewards/semantic_entropy_math_reward": 0.4136297255754471,
+      "step": 310
+    },
+    {
+      "completion_length": 472.4763946533203,
+      "epoch": 0.8104234527687296,
+      "grad_norm": 0.009827625937759876,
+      "learning_rate": 9.397086261457511e-07,
+      "loss": -0.0147,
+      "reward": 0.9812317602336407,
+      "reward_std": 0.11015114816837013,
+      "rewards/accuracy_reward": 0.5656887628138065,
+      "rewards/semantic_entropy_math_reward": 0.41554298996925354,
+      "step": 311
+    },
+    {
+      "completion_length": 487.07525062561035,
+      "epoch": 0.8130293159609121,
+      "grad_norm": 0.014018489047884941,
+      "learning_rate": 9.146081187268185e-07,
+      "loss": -0.0131,
+      "reward": 0.9449708163738251,
+      "reward_std": 0.10275353176984936,
+      "rewards/accuracy_reward": 0.5459183566272259,
+      "rewards/semantic_entropy_math_reward": 0.39905246533453465,
+      "step": 312
+    },
+    {
+      "completion_length": 478.88073348999023,
+      "epoch": 0.8156351791530945,
+      "grad_norm": 0.008230605162680149,
+      "learning_rate": 8.898136085456127e-07,
+      "loss": -0.0052,
+      "reward": 0.9649234525859356,
+      "reward_std": 0.08008495910326019,
+      "rewards/accuracy_reward": 0.5459183566272259,
+      "rewards/semantic_entropy_math_reward": 0.41900509409606457,
+      "step": 313
+    },
+    {
+      "completion_length": 477.53187561035156,
+      "epoch": 0.8182410423452768,
+      "grad_norm": 0.011597930453717709,
+      "learning_rate": 8.65326952719357e-07,
+      "loss": -0.0081,
+      "reward": 1.0210458859801292,
+      "reward_std": 0.08366672950796783,
+      "rewards/accuracy_reward": 0.5816326439380646,
+      "rewards/semantic_entropy_math_reward": 0.43941325321793556,
+      "step": 314
+    },
+    {
+      "completion_length": 481.01720428466797,
+      "epoch": 0.8208469055374593,
+      "grad_norm": 0.01356886513531208,
+      "learning_rate": 8.411499853068783e-07,
+      "loss": -0.0092,
+      "reward": 0.9314868524670601,
+      "reward_std": 0.09904140722937882,
+      "rewards/accuracy_reward": 0.5121173355728388,
+      "rewards/semantic_entropy_math_reward": 0.4193695206195116,
+      "step": 315
+    },
+    {
+      "completion_length": 447.973840713501,
+      "epoch": 0.8234527687296417,
+      "grad_norm": 0.0068409559316933155,
+      "learning_rate": 8.172845171712379e-07,
+      "loss": -0.0132,
+      "reward": 1.0037354119122028,
+      "reward_std": 0.07698465495195705,
+      "rewards/accuracy_reward": 0.5637755002826452,
+      "rewards/semantic_entropy_math_reward": 0.4399599079042673,
+      "step": 316
+    },
+    {
+      "completion_length": 483.59756660461426,
+      "epoch": 0.826058631921824,
+      "grad_norm": 0.011241559870541096,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": -0.017,
+      "reward": 0.9950801581144333,
+      "reward_std": 0.08729210350429639,
+      "rewards/accuracy_reward": 0.5637754984200001,
+      "rewards/semantic_entropy_math_reward": 0.4313046485185623,
+      "step": 317
+    },
+    {
+      "completion_length": 467.542724609375,
+      "epoch": 0.8286644951140065,
+      "grad_norm": 0.014086179435253143,
+      "learning_rate": 7.70495205391818e-07,
+      "loss": -0.014,
+      "reward": 0.9365889057517052,
+      "reward_std": 0.08978691761149094,
+      "rewards/accuracy_reward": 0.5280612111091614,
+      "rewards/semantic_entropy_math_reward": 0.4085276983678341,
+      "step": 318
+    },
+    {
+      "completion_length": 469.43749046325684,
+      "epoch": 0.8312703583061889,
+      "grad_norm": 0.009214820340275764,
+      "learning_rate": 7.475748662833615e-07,
+      "loss": -0.0077,
+      "reward": 0.9763119295239449,
+      "reward_std": 0.09077268361579627,
+      "rewards/accuracy_reward": 0.5542091699317098,
+      "rewards/semantic_entropy_math_reward": 0.42210277169942856,
+      "step": 319
+    },
+    {
+      "completion_length": 465.4904251098633,
+      "epoch": 0.8338762214983714,
+      "grad_norm": 0.007901620119810104,
+      "learning_rate": 7.249730352599e-07,
+      "loss": -0.0133,
+      "reward": 0.9463374502956867,
+      "reward_std": 0.08334645116701722,
+      "rewards/accuracy_reward": 0.5312499832361937,
+      "rewards/semantic_entropy_math_reward": 0.4150874614715576,
+      "step": 320
+    },
+    {
+      "completion_length": 461.4591751098633,
+      "epoch": 0.8364820846905537,
+      "grad_norm": 0.012061836197972298,
+      "learning_rate": 7.026914052062433e-07,
+      "loss": -0.006,
+      "reward": 1.0148505568504333,
+      "reward_std": 0.08093352918513119,
+      "rewards/accuracy_reward": 0.5816326476633549,
+      "rewards/semantic_entropy_math_reward": 0.4332179147750139,
+      "step": 321
+    },
+    {
+      "completion_length": 425.9540729522705,
+      "epoch": 0.8390879478827361,
+      "grad_norm": 0.010147208347916603,
+      "learning_rate": 6.807316450240425e-07,
+      "loss": -0.008,
+      "reward": 1.0222302973270416,
+      "reward_std": 0.055277756706345826,
+      "rewards/accuracy_reward": 0.5918367300182581,
+      "rewards/semantic_entropy_math_reward": 0.430393585935235,
+      "step": 322
+    },
+    {
+      "completion_length": 456.4866008758545,
+      "epoch": 0.8416938110749186,
+      "grad_norm": 0.0100060123950243,
+      "learning_rate": 6.590953995067812e-07,
+      "loss": -0.013,
+      "reward": 0.9208272397518158,
+      "reward_std": 0.07600355852628127,
+      "rewards/accuracy_reward": 0.49808672070503235,
+      "rewards/semantic_entropy_math_reward": 0.42274050787091255,
+      "step": 323
+    },
+    {
+      "completion_length": 449.3877468109131,
+      "epoch": 0.844299674267101,
+      "grad_norm": 0.011219106614589691,
+      "learning_rate": 6.377842892165892e-07,
+      "loss": -0.0098,
+      "reward": 0.9591836631298065,
+      "reward_std": 0.11074057850055397,
+      "rewards/accuracy_reward": 0.5427295789122581,
+      "rewards/semantic_entropy_math_reward": 0.4164540767669678,
+      "step": 324
+    },
+    {
+      "completion_length": 476.1007499694824,
+      "epoch": 0.8469055374592834,
+      "grad_norm": 0.02076311595737934,
+      "learning_rate": 6.167999103628569e-07,
+      "loss": -0.0083,
+      "reward": 0.9911625236272812,
+      "reward_std": 0.10244781541405246,
+      "rewards/accuracy_reward": 0.5714285615831614,
+      "rewards/semantic_entropy_math_reward": 0.419733963906765,
+      "step": 325
+    },
+    {
+      "completion_length": 446.6645317077637,
+      "epoch": 0.8495114006514658,
+      "grad_norm": 0.019433218985795975,
+      "learning_rate": 5.961438346826792e-07,
+      "loss": -0.0009,
+      "reward": 0.9219205342233181,
+      "reward_std": 0.07894981937715784,
+      "rewards/accuracy_reward": 0.5012755040079355,
+      "rewards/semantic_entropy_math_reward": 0.4206450283527374,
+      "step": 326
+    },
+    {
+      "completion_length": 459.9164447784424,
+      "epoch": 0.8521172638436482,
+      "grad_norm": 0.009447368793189526,
+      "learning_rate": 5.758176093231294e-07,
+      "loss": -0.0137,
+      "reward": 1.0237791389226913,
+      "reward_std": 0.08699435199378058,
+      "rewards/accuracy_reward": 0.5956632476300001,
+      "rewards/semantic_entropy_math_reward": 0.4281158745288849,
+      "step": 327
+    },
+    {
+      "completion_length": 486.7385063171387,
+      "epoch": 0.8547231270358306,
+      "grad_norm": 0.012096602469682693,
+      "learning_rate": 5.558227567253832e-07,
+      "loss": -0.0081,
+      "reward": 0.9570881724357605,
+      "reward_std": 0.12206664809491485,
+      "rewards/accuracy_reward": 0.531249986961484,
+      "rewards/semantic_entropy_math_reward": 0.4258381836116314,
+      "step": 328
+    },
+    {
+      "completion_length": 475.43876457214355,
+      "epoch": 0.857328990228013,
+      "grad_norm": 0.00933607667684555,
+      "learning_rate": 5.361607745106817e-07,
+      "loss": -0.0169,
+      "reward": 0.9730320535600185,
+      "reward_std": 0.08804468740709126,
+      "rewards/accuracy_reward": 0.5567601844668388,
+      "rewards/semantic_entropy_math_reward": 0.4162718579173088,
+      "step": 329
+    },
+    {
+      "completion_length": 431.8603210449219,
+      "epoch": 0.8599348534201955,
+      "grad_norm": 0.02015325427055359,
+      "learning_rate": 5.168331353681643e-07,
+      "loss": -0.0021,
+      "reward": 0.9969023317098618,
+      "reward_std": 0.09806176437996328,
+      "rewards/accuracy_reward": 0.5733418297022581,
+      "rewards/semantic_entropy_math_reward": 0.42356047965586185,
+      "step": 330
+    },
+    {
+      "completion_length": 427.3131294250488,
+      "epoch": 0.8625407166123779,
+      "grad_norm": 0.013669871725142002,
+      "learning_rate": 4.97841286944557e-07,
+      "loss": -0.0153,
+      "reward": 1.0245991051197052,
+      "reward_std": 0.07894148712512106,
+      "rewards/accuracy_reward": 0.578443868085742,
+      "rewards/semantic_entropy_math_reward": 0.4461552295833826,
+      "step": 331
+    },
+    {
+      "completion_length": 422.89731216430664,
+      "epoch": 0.8651465798045602,
+      "grad_norm": 0.010783183388411999,
+      "learning_rate": 4.791866517357491e-07,
+      "loss": -0.0006,
+      "reward": 1.0032798871397972,
+      "reward_std": 0.07655263572814874,
+      "rewards/accuracy_reward": 0.5758928395807743,
+      "rewards/semantic_entropy_math_reward": 0.42738700844347477,
+      "step": 332
+    },
+    {
+      "completion_length": 433.96299934387207,
+      "epoch": 0.8677524429967427,
+      "grad_norm": 0.010686203837394714,
+      "learning_rate": 4.608706269802471e-07,
+      "loss": -0.0031,
+      "reward": 1.0082908011972904,
+      "reward_std": 0.0961291438434273,
+      "rewards/accuracy_reward": 0.5848214142024517,
+      "rewards/semantic_entropy_math_reward": 0.42346937395632267,
+      "step": 333
+    },
+    {
+      "completion_length": 483.29973793029785,
+      "epoch": 0.8703583061889251,
+      "grad_norm": 0.013967434875667095,
+      "learning_rate": 4.428945845545168e-07,
+      "loss": -0.0167,
+      "reward": 0.9419642761349678,
+      "reward_std": 0.08668096817564219,
+      "rewards/accuracy_reward": 0.5433673374354839,
+      "rewards/semantic_entropy_math_reward": 0.3985969349741936,
+      "step": 334
+    },
+    {
+      "completion_length": 453.30356216430664,
+      "epoch": 0.8729641693811075,
+      "grad_norm": 0.009853254072368145,
+      "learning_rate": 4.2525987087023433e-07,
+      "loss": -0.0047,
+      "reward": 0.9753097556531429,
+      "reward_std": 0.09639269160106778,
+      "rewards/accuracy_reward": 0.5452805999666452,
+      "rewards/semantic_entropy_math_reward": 0.43002915009856224,
+      "step": 335
+    },
+    {
+      "completion_length": 431.1064987182617,
+      "epoch": 0.8755700325732899,
+      "grad_norm": 0.009506824426352978,
+      "learning_rate": 4.0796780677343606e-07,
+      "loss": -0.0053,
+      "reward": 0.9678389020264149,
+      "reward_std": 0.0869278380414471,
+      "rewards/accuracy_reward": 0.5497448872774839,
+      "rewards/semantic_entropy_math_reward": 0.41809402219951153,
+      "step": 336
+    },
+    {
+      "completion_length": 432.71044731140137,
+      "epoch": 0.8781758957654723,
+      "grad_norm": 0.012929832562804222,
+      "learning_rate": 3.910196874455896e-07,
+      "loss": -0.0077,
+      "reward": 0.9853316098451614,
+      "reward_std": 0.08627731067826971,
+      "rewards/accuracy_reward": 0.5535714197903872,
+      "rewards/semantic_entropy_math_reward": 0.43176020309329033,
+      "step": 337
+    },
+    {
+      "completion_length": 440.03825759887695,
+      "epoch": 0.8807817589576548,
+      "grad_norm": 0.01218812633305788,
+      "learning_rate": 3.744167823065814e-07,
+      "loss": -0.0122,
+      "reward": 0.9091654308140278,
+      "reward_std": 0.09821587102487683,
+      "rewards/accuracy_reward": 0.4968112148344517,
+      "rewards/semantic_entropy_math_reward": 0.4123542197048664,
+      "step": 338
+    },
+    {
+      "completion_length": 457.76976108551025,
+      "epoch": 0.8833876221498371,
+      "grad_norm": 0.008668128401041031,
+      "learning_rate": 3.581603349196372e-07,
+      "loss": -0.0077,
+      "reward": 0.930757999420166,
+      "reward_std": 0.07734870351850986,
+      "rewards/accuracy_reward": 0.5197703987360001,
+      "rewards/semantic_entropy_math_reward": 0.41098760068416595,
+      "step": 339
+    },
+    {
+      "completion_length": 442.6613426208496,
+      "epoch": 0.8859934853420195,
+      "grad_norm": 0.008681760169565678,
+      "learning_rate": 3.4225156289818096e-07,
+      "loss": -0.0059,
+      "reward": 1.0034620836377144,
+      "reward_std": 0.08158292178995907,
+      "rewards/accuracy_reward": 0.5816326476633549,
+      "rewards/semantic_entropy_math_reward": 0.42182943783700466,
+      "step": 340
+    },
+    {
+      "completion_length": 485.2876148223877,
+      "epoch": 0.888599348534202,
+      "grad_norm": 0.008547916077077389,
+      "learning_rate": 3.26691657814634e-07,
+      "loss": -0.0087,
+      "reward": 0.9770408011972904,
+      "reward_std": 0.07393853948451579,
+      "rewards/accuracy_reward": 0.5433673355728388,
+      "rewards/semantic_entropy_math_reward": 0.43367345817387104,
+      "step": 341
+    },
+    {
+      "completion_length": 452.9138927459717,
+      "epoch": 0.8912052117263843,
+      "grad_norm": 0.01058235950767994,
+      "learning_rate": 3.1148178511116624e-07,
+      "loss": -0.0017,
+      "reward": 1.0015488117933273,
+      "reward_std": 0.09451375540811568,
+      "rewards/accuracy_reward": 0.5701530463993549,
+      "rewards/semantic_entropy_math_reward": 0.4313957579433918,
+      "step": 342
+    },
+    {
+      "completion_length": 476.1976947784424,
+      "epoch": 0.8938110749185668,
+      "grad_norm": 0.015603607520461082,
+      "learning_rate": 2.966230840124007e-07,
+      "loss": -0.0084,
+      "reward": 0.9884292893111706,
+      "reward_std": 0.0848599491873756,
+      "rewards/accuracy_reward": 0.548469377681613,
+      "rewards/semantic_entropy_math_reward": 0.4399598930031061,
+      "step": 343
+    },
+    {
+      "completion_length": 474.1007595062256,
+      "epoch": 0.8964169381107492,
+      "grad_norm": 0.009830434806644917,
+      "learning_rate": 2.821166674400905e-07,
+      "loss": -0.0111,
+      "reward": 0.9547193609178066,
+      "reward_std": 0.08753847831394523,
+      "rewards/accuracy_reward": 0.5414540711790323,
+      "rewards/semantic_entropy_math_reward": 0.4132652934640646,
+      "step": 344
+    },
+    {
+      "completion_length": 462.9668254852295,
+      "epoch": 0.8990228013029316,
+      "grad_norm": 0.012285560369491577,
+      "learning_rate": 2.6796362192975766e-07,
+      "loss": -0.0061,
+      "reward": 0.9971756413578987,
+      "reward_std": 0.10323424229864031,
+      "rewards/accuracy_reward": 0.5733418203890324,
+      "rewards/semantic_entropy_math_reward": 0.4238338116556406,
+      "step": 345
+    },
+    {
+      "completion_length": 448.88455963134766,
+      "epoch": 0.901628664495114,
+      "grad_norm": 0.010707194916903973,
+      "learning_rate": 2.5416500754931294e-07,
+      "loss": -0.0088,
+      "reward": 1.0225036218762398,
+      "reward_std": 0.09066123812226579,
+      "rewards/accuracy_reward": 0.6001275330781937,
+      "rewards/semantic_entropy_math_reward": 0.4223760813474655,
+      "step": 346
+    },
+    {
+      "completion_length": 447.4311122894287,
+      "epoch": 0.9042345276872964,
+      "grad_norm": 0.009493795223534107,
+      "learning_rate": 2.407218578196524e-07,
+      "loss": -0.0071,
+      "reward": 0.9620080180466175,
+      "reward_std": 0.09404018375789747,
+      "rewards/accuracy_reward": 0.5382652971893549,
+      "rewards/semantic_entropy_math_reward": 0.4237427096813917,
+      "step": 347
+    },
+    {
+      "completion_length": 486.0082836151123,
+      "epoch": 0.9068403908794789,
+      "grad_norm": 0.010954687371850014,
+      "learning_rate": 2.2763517963725169e-07,
+      "loss": -0.0128,
+      "reward": 0.9949890375137329,
+      "reward_std": 0.095983712002635,
+      "rewards/accuracy_reward": 0.5829081572592258,
+      "rewards/semantic_entropy_math_reward": 0.4120809007436037,
+      "step": 348
+    },
+    {
+      "completion_length": 431.5510139465332,
+      "epoch": 0.9094462540716612,
+      "grad_norm": 0.012899513356387615,
+      "learning_rate": 2.1490595319874574e-07,
+      "loss": -0.0081,
+      "reward": 0.9666545018553734,
+      "reward_std": 0.09545677714049816,
+      "rewards/accuracy_reward": 0.5471938643604517,
+      "rewards/semantic_entropy_math_reward": 0.41946063563227654,
+      "step": 349
+    },
+    {
+      "completion_length": 485.63200759887695,
+      "epoch": 0.9120521172638436,
+      "grad_norm": 0.01148886326700449,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": -0.0082,
+      "reward": 0.9554482325911522,
+      "reward_std": 0.100743565504672,
+      "rewards/accuracy_reward": 0.5440050885081291,
+      "rewards/semantic_entropy_math_reward": 0.4114431384950876,
+      "step": 350
+    },
+    {
+      "completion_length": 460.51784324645996,
+      "epoch": 0.9146579804560261,
+      "grad_norm": 0.012849576771259308,
+      "learning_rate": 1.905236424022633e-07,
+      "loss": -0.012,
+      "reward": 0.8979591764509678,
+      "reward_std": 0.09470229386352003,
+      "rewards/accuracy_reward": 0.4872448891401291,
+      "rewards/semantic_entropy_math_reward": 0.41071427799761295,
+      "step": 351
+    },
+    {
+      "completion_length": 487.3086643218994,
+      "epoch": 0.9172638436482085,
+      "grad_norm": 0.01112964004278183,
+      "learning_rate": 1.7887238428763553e-07,
+      "loss": -0.0101,
+      "reward": 0.931213553994894,
+      "reward_std": 0.10900401265826076,
+      "rewards/accuracy_reward": 0.5197703968733549,
+      "rewards/semantic_entropy_math_reward": 0.4114431384950876,
+      "step": 352
+    },
+    {
+      "completion_length": 445.08800315856934,
+      "epoch": 0.9198697068403909,
+      "grad_norm": 0.009342041797935963,
+      "learning_rate": 1.6758223026681507e-07,
+      "loss": -0.0039,
+      "reward": 0.9932579882442951,
+      "reward_std": 0.06484666049072985,
+      "rewards/accuracy_reward": 0.5561224408447742,
+      "rewards/semantic_entropy_math_reward": 0.4371355604380369,
+      "step": 353
+    },
+    {
+      "completion_length": 441.0204029083252,
+      "epoch": 0.9224755700325733,
+      "grad_norm": 0.012000740505754948,
+      "learning_rate": 1.5665402597616842e-07,
+      "loss": -0.0088,
+      "reward": 0.9781340956687927,
+      "reward_std": 0.09719502518419176,
+      "rewards/accuracy_reward": 0.5682397782802582,
+      "rewards/semantic_entropy_math_reward": 0.40989430248737335,
+      "step": 354
+    },
+    {
+      "completion_length": 419.2601947784424,
+      "epoch": 0.9250814332247557,
+      "grad_norm": 0.013637058436870575,
+      "learning_rate": 1.4608858994190344e-07,
+      "loss": -0.0084,
+      "reward": 0.9946246184408665,
+      "reward_std": 0.09013567195506766,
+      "rewards/accuracy_reward": 0.5758928433060646,
+      "rewards/semantic_entropy_math_reward": 0.41873177886009216,
+      "step": 355
+    },
+    {
+      "completion_length": 424.69897270202637,
+      "epoch": 0.9276872964169381,
+      "grad_norm": 0.011202794499695301,
+      "learning_rate": 1.358867135187636e-07,
+      "loss": -0.0031,
+      "reward": 1.043093990534544,
+      "reward_std": 0.07236083660973236,
+      "rewards/accuracy_reward": 0.6052295807749033,
+      "rewards/semantic_entropy_math_reward": 0.43786441907286644,
+      "step": 356
+    },
+    {
+      "completion_length": 453.50764656066895,
+      "epoch": 0.9302931596091205,
+      "grad_norm": 0.011242231354117393,
+      "learning_rate": 1.2604916083075236e-07,
+      "loss": -0.0184,
+      "reward": 0.9417820461094379,
+      "reward_std": 0.11088896542787552,
+      "rewards/accuracy_reward": 0.5344387609511614,
+      "rewards/semantic_entropy_math_reward": 0.40734328515827656,
+      "step": 357
+    },
+    {
+      "completion_length": 401.16899394989014,
+      "epoch": 0.9328990228013029,
+      "grad_norm": 0.009380885399878025,
+      "learning_rate": 1.1657666871390471e-07,
+      "loss": 0.0012,
+      "reward": 1.0115706995129585,
+      "reward_std": 0.08255202136933804,
+      "rewards/accuracy_reward": 0.5841836612671614,
+      "rewards/semantic_entropy_math_reward": 0.42738702334463596,
+      "step": 358
+    },
+    {
+      "completion_length": 450.27869033813477,
+      "epoch": 0.9355048859934854,
+      "grad_norm": 0.012342661619186401,
+      "learning_rate": 1.0746994666109234e-07,
+      "loss": -0.0078,
+      "reward": 0.9944424033164978,
+      "reward_std": 0.08844125363975763,
+      "rewards/accuracy_reward": 0.5758928433060646,
+      "rewards/semantic_entropy_math_reward": 0.41854955442249775,
+      "step": 359
+    },
+    {
+      "completion_length": 439.46044921875,
+      "epoch": 0.9381107491856677,
+      "grad_norm": 0.01147682499140501,
+      "learning_rate": 9.872967676888611e-08,
+      "loss": -0.0142,
+      "reward": 0.9127186425030231,
+      "reward_std": 0.11078671208815649,
+      "rewards/accuracy_reward": 0.5121173411607742,
+      "rewards/semantic_entropy_math_reward": 0.4006012938916683,
+      "step": 360
+    },
+    {
+      "completion_length": 484.5235786437988,
+      "epoch": 0.9407166123778502,
+      "grad_norm": 0.023287979885935783,
+      "learning_rate": 9.035651368646647e-08,
+      "loss": -0.0154,
+      "reward": 0.8902150094509125,
+      "reward_std": 0.10496682248776779,
+      "rewards/accuracy_reward": 0.47257651947438717,
+      "rewards/semantic_entropy_math_reward": 0.41763847321271896,
+      "step": 361
+    },
+    {
+      "completion_length": 460.7825222015381,
+      "epoch": 0.9433224755700326,
+      "grad_norm": 0.010122385807335377,
+      "learning_rate": 8.235108456658814e-08,
+      "loss": -0.0086,
+      "reward": 0.9826894924044609,
+      "reward_std": 0.09733415581285954,
+      "rewards/accuracy_reward": 0.5644132476300001,
+      "rewards/semantic_entropy_math_reward": 0.4182762373238802,
+      "step": 362
+    },
+    {
+      "completion_length": 488.81886291503906,
+      "epoch": 0.9459283387622149,
+      "grad_norm": 0.010546687990427017,
+      "learning_rate": 7.471398901860772e-08,
+      "loss": -0.0247,
+      "reward": 0.9782251939177513,
+      "reward_std": 0.10399020987097174,
+      "rewards/accuracy_reward": 0.5605867225676775,
+      "rewards/semantic_entropy_math_reward": 0.41763848438858986,
+      "step": 363
+    },
+    {
+      "completion_length": 421.8590488433838,
+      "epoch": 0.9485342019543974,
+      "grad_norm": 0.011549089103937149,
+      "learning_rate": 6.744579906357185e-08,
+      "loss": -0.0123,
+      "reward": 0.9907980933785439,
+      "reward_std": 0.08595408496330492,
+      "rewards/accuracy_reward": 0.5567601937800646,
+      "rewards/semantic_entropy_math_reward": 0.434037895873189,
+      "step": 364
+    },
+    {
+      "completion_length": 454.81504821777344,
+      "epoch": 0.9511400651465798,
+      "grad_norm": 0.012615454383194447,
+      "learning_rate": 6.054705909137426e-08,
+      "loss": -0.011,
+      "reward": 0.9697521664202213,
+      "reward_std": 0.10905627987813205,
+      "rewards/accuracy_reward": 0.5516581553965807,
+      "rewards/semantic_entropy_math_reward": 0.4180940240621567,
+      "step": 365
+    },
+    {
+      "completion_length": 454.48596382141113,
+      "epoch": 0.9537459283387623,
+      "grad_norm": 0.010557249188423157,
+      "learning_rate": 5.401828581997948e-08,
+      "loss": -0.0029,
+      "reward": 0.9792273677885532,
+      "reward_std": 0.08946075174026191,
+      "rewards/accuracy_reward": 0.5497448872774839,
+      "rewards/semantic_entropy_math_reward": 0.42948249727487564,
+      "step": 366
+    },
+    {
+      "completion_length": 438.12499046325684,
+      "epoch": 0.9563517915309446,
+      "grad_norm": 0.008979610167443752,
+      "learning_rate": 4.7859968256719344e-08,
+      "loss": -0.0124,
+      "reward": 0.9787718281149864,
+      "reward_std": 0.08572447136975825,
+      "rewards/accuracy_reward": 0.5452805999666452,
+      "rewards/semantic_entropy_math_reward": 0.43349125795066357,
+      "step": 367
+    },
+    {
+      "completion_length": 449.46810150146484,
+      "epoch": 0.958957654723127,
+      "grad_norm": 0.01041451282799244,
+      "learning_rate": 4.207256766166845e-08,
+      "loss": -0.0071,
+      "reward": 0.9146319180727005,
+      "reward_std": 0.08887826814316213,
+      "rewards/accuracy_reward": 0.5063775405287743,
+      "rewards/semantic_entropy_math_reward": 0.40825437009334564,
+      "step": 368
+    },
+    {
+      "completion_length": 447.2391471862793,
+      "epoch": 0.9615635179153095,
+      "grad_norm": 0.008584603667259216,
+      "learning_rate": 3.665651751309451e-08,
+      "loss": -0.0096,
+      "reward": 0.9768585935235023,
+      "reward_std": 0.08100897446274757,
+      "rewards/accuracy_reward": 0.544005086645484,
+      "rewards/semantic_entropy_math_reward": 0.43285347521305084,
+      "step": 369
+    },
+    {
+      "completion_length": 423.11478424072266,
+      "epoch": 0.9641693811074918,
+      "grad_norm": 0.009456335566937923,
+      "learning_rate": 3.16122234749916e-08,
+      "loss": -0.0034,
+      "reward": 1.026512373238802,
+      "reward_std": 0.08707130316179246,
+      "rewards/accuracy_reward": 0.5943877454847097,
+      "rewards/semantic_entropy_math_reward": 0.4321246314793825,
+      "step": 370
+    },
+    {
+      "completion_length": 447.5299644470215,
+      "epoch": 0.9667752442996743,
+      "grad_norm": 0.015312188304960728,
+      "learning_rate": 2.6940063366693303e-08,
+      "loss": -0.0019,
+      "reward": 0.9723943062126637,
+      "reward_std": 0.08906970918178558,
+      "rewards/accuracy_reward": 0.5535714235156775,
+      "rewards/semantic_entropy_math_reward": 0.41882289201021194,
+      "step": 371
+    },
+    {
+      "completion_length": 449.52231216430664,
+      "epoch": 0.9693811074918567,
+      "grad_norm": 0.009211807511746883,
+      "learning_rate": 2.264038713457706e-08,
+      "loss": -0.0086,
+      "reward": 0.9643768034875393,
+      "reward_std": 0.09445711580337957,
+      "rewards/accuracy_reward": 0.542091827839613,
+      "rewards/semantic_entropy_math_reward": 0.42228496447205544,
+      "step": 372
+    },
+    {
+      "completion_length": 427.89858627319336,
+      "epoch": 0.971986970684039,
+      "grad_norm": 0.014192759990692139,
+      "learning_rate": 1.8713516825851207e-08,
+      "loss": -0.0046,
+      "reward": 0.975127536803484,
+      "reward_std": 0.08639455388765782,
+      "rewards/accuracy_reward": 0.5382652953267097,
+      "rewards/semantic_entropy_math_reward": 0.4368622377514839,
+      "step": 373
+    },
+    {
+      "completion_length": 458.57715606689453,
+      "epoch": 0.9745928338762215,
+      "grad_norm": 0.017490204423666,
+      "learning_rate": 1.51597465644332e-08,
+      "loss": -0.0092,
+      "reward": 0.9249271154403687,
+      "reward_std": 0.10873797861859202,
+      "rewards/accuracy_reward": 0.5191326383501291,
+      "rewards/semantic_entropy_math_reward": 0.40579446218907833,
+      "step": 374
+    },
+    {
+      "completion_length": 458.8858337402344,
+      "epoch": 0.9771986970684039,
+      "grad_norm": 0.01276995800435543,
+      "learning_rate": 1.1979342528922189e-08,
+      "loss": -0.0095,
+      "reward": 0.9507106244564056,
+      "reward_std": 0.1035163603228284,
+      "rewards/accuracy_reward": 0.5350765194743872,
+      "rewards/semantic_entropy_math_reward": 0.4156340938061476,
+      "step": 375
+    },
+    {
+      "completion_length": 443.0637664794922,
+      "epoch": 0.9798045602605863,
+      "grad_norm": 0.008754832670092583,
+      "learning_rate": 9.1725429326589e-09,
+      "loss": -0.0068,
+      "reward": 0.987335991114378,
+      "reward_std": 0.07803339039674029,
+      "rewards/accuracy_reward": 0.5542091745883226,
+      "rewards/semantic_entropy_math_reward": 0.4331268183887005,
+      "step": 376
+    },
+    {
+      "completion_length": 438.10521697998047,
+      "epoch": 0.9824104234527687,
+      "grad_norm": 0.01094447448849678,
+      "learning_rate": 6.739558005884883e-09,
+      "loss": -0.0073,
+      "reward": 0.945426382124424,
+      "reward_std": 0.07054417789913714,
+      "rewards/accuracy_reward": 0.5318877473473549,
+      "rewards/semantic_entropy_math_reward": 0.41353862918913364,
+      "step": 377
+    },
+    {
+      "completion_length": 422.69897079467773,
+      "epoch": 0.9850162866449511,
+      "grad_norm": 0.01369545143097639,
+      "learning_rate": 4.6805699799967744e-09,
+      "loss": -0.0054,
+      "reward": 0.9959912300109863,
+      "reward_std": 0.09532720362767577,
+      "rewards/accuracy_reward": 0.5682397801429033,
+      "rewards/semantic_entropy_math_reward": 0.42775145545601845,
+      "step": 378
+    },
+    {
+      "completion_length": 425.08226013183594,
+      "epoch": 0.9876221498371336,
+      "grad_norm": 0.009248544462025166,
+      "learning_rate": 2.995733073895557e-09,
+      "loss": -0.0082,
+      "reward": 1.0608600415289402,
+      "reward_std": 0.09763774264138192,
+      "rewards/accuracy_reward": 0.6352040730416775,
+      "rewards/semantic_entropy_math_reward": 0.425655959174037,
+      "step": 379
+    },
+    {
+      "completion_length": 419.81759452819824,
+      "epoch": 0.990228013029316,
+      "grad_norm": 0.008247760124504566,
+      "learning_rate": 1.6851734824380184e-09,
+      "loss": -0.0027,
+      "reward": 0.985878262668848,
+      "reward_std": 0.08342396467924118,
+      "rewards/accuracy_reward": 0.5720663201063871,
+      "rewards/semantic_entropy_math_reward": 0.41381194815039635,
+      "step": 380
+    },
+    {
+      "completion_length": 452.42090797424316,
+      "epoch": 0.9928338762214983,
+      "grad_norm": 0.011512123979628086,
+      "learning_rate": 7.48989366980979e-10,
+      "loss": -0.0088,
+      "reward": 0.9500728640705347,
+      "reward_std": 0.08807020282256417,
+      "rewards/accuracy_reward": 0.5459183580242097,
+      "rewards/semantic_entropy_math_reward": 0.4041545204818249,
+      "step": 381
+    },
+    {
+      "completion_length": 460.566951751709,
+      "epoch": 0.9954397394136808,
+      "grad_norm": 0.014522072859108448,
+      "learning_rate": 1.872508480332824e-10,
+      "loss": -0.0131,
+      "reward": 0.9883381687104702,
+      "reward_std": 0.09443258820101619,
+      "rewards/accuracy_reward": 0.5573979429900646,
+      "rewards/semantic_entropy_math_reward": 0.43094020895659924,
+      "step": 382
+    },
+    {
+      "completion_length": 440.04973220825195,
+      "epoch": 0.9980456026058632,
+      "grad_norm": 0.015413707122206688,
+      "learning_rate": 0.0,
+      "loss": -0.0133,
+      "reward": 0.9645590074360371,
+      "reward_std": 0.0993940774933435,
+      "rewards/accuracy_reward": 0.5497448854148388,
+      "rewards/semantic_entropy_math_reward": 0.4148141276091337,
+      "step": 383
+    },
+    {
+      "epoch": 0.9980456026058632,
+      "step": 383,
+      "total_flos": 0.0,
+      "train_loss": 0.0,
+      "train_runtime": 2.7488,
+      "train_samples_per_second": 31267.524,
+      "train_steps_per_second": 139.332
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 383,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 20,
+  "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
+}