diff --git "a/checkpoint-400/trainer_state.json" "b/checkpoint-400/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-400/trainer_state.json"
@@ -0,0 +1,5633 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7862407862407862,
+  "eval_steps": 500,
+  "global_step": 400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 562.125,
+      "epoch": 0.0019656019656019656,
+      "grad_norm": 0.0,
+      "kl": 0.0,
+      "learning_rate": 5e-08,
+      "loss": 0.0,
+      "num_tokens": 22458.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 1
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 382.84375,
+      "epoch": 0.003931203931203931,
+      "grad_norm": 0.0,
+      "kl": 0.0,
+      "learning_rate": 1e-07,
+      "loss": 0.0,
+      "num_tokens": 39539.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 2
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 296.84375,
+      "epoch": 0.005896805896805897,
+      "grad_norm": 0.0,
+      "kl": 0.0,
+      "learning_rate": 1.5e-07,
+      "loss": 0.0,
+      "num_tokens": 52892.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 3
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 562.84375,
+      "epoch": 0.007862407862407862,
+      "grad_norm": 0.0,
+      "kl": 0.0,
+      "learning_rate": 2e-07,
+      "loss": 0.0,
+      "num_tokens": 73799.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 4
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 454.9375,
+      "epoch": 0.009828009828009828,
+      "grad_norm": 0.0,
+      "kl": 0.0,
+      "learning_rate": 2.5e-07,
+      "loss": 0.0,
+      "num_tokens": 92389.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 5
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 401.5625,
+      "epoch": 0.011793611793611793,
+      "grad_norm": 0.8271026359027706,
+      "kl": 0.0,
+      "learning_rate": 3e-07,
+      "loss": 0.0459,
+      "num_tokens": 108305.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 6
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 390.625,
+      "epoch": 0.013759213759213759,
+      "grad_norm": 0.0003446225536382493,
+      "kl": 0.0002499818801879883,
+      "learning_rate": 3.5e-07,
+      "loss": 0.0,
+      "num_tokens": 125607.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 7
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 448.75,
+      "epoch": 0.015724815724815724,
+      "grad_norm": 0.0013026212429223525,
+      "kl": 0.0004665851593017578,
+      "learning_rate": 4e-07,
+      "loss": 0.0,
+      "num_tokens": 142955.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 8
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 461.90625,
+      "epoch": 0.01769041769041769,
+      "grad_norm": 0.0021171088963702245,
+      "kl": 0.0006817877292633057,
+      "learning_rate": 4.5e-07,
+      "loss": 0.0,
+      "num_tokens": 162956.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 9
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 330.6875,
+      "epoch": 0.019656019656019656,
+      "grad_norm": 1.3023849518044601,
+      "kl": 0.0006229877471923828,
+      "learning_rate": 5e-07,
+      "loss": 0.0444,
+      "num_tokens": 177758.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 10
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 513.1875,
+      "epoch": 0.021621621621621623,
+      "grad_norm": 0.00027502974416413834,
+      "kl": 0.00018683075904846191,
+      "learning_rate": 5.5e-07,
+      "loss": 0.0,
+      "num_tokens": 197364.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 11
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 424.625,
+      "epoch": 0.023587223587223587,
+      "grad_norm": 0.6170315097653724,
+      "kl": 0.00029702484607696533,
+      "learning_rate": 6e-07,
+      "loss": -0.0489,
+      "num_tokens": 214156.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 12
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 532.1875,
+      "epoch": 0.025552825552825554,
+      "grad_norm": 0.0002590285678817397,
+      "kl": 0.00019553303718566895,
+      "learning_rate": 6.5e-07,
+      "loss": 0.0,
+      "num_tokens": 234938.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 13
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 391.6875,
+      "epoch": 0.027518427518427518,
+      "grad_norm": 0.2468481109530296,
+      "kl": 0.0002957582473754883,
+      "learning_rate": 7e-07,
+      "loss": -0.042,
+      "num_tokens": 250970.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 14
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 526.75,
+      "epoch": 0.029484029484029485,
+      "grad_norm": 0.6658783590054981,
+      "kl": 0.00033867359161376953,
+      "learning_rate": 7.5e-07,
+      "loss": -0.0001,
+      "num_tokens": 271842.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 15
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 357.5,
+      "epoch": 0.03144963144963145,
+      "grad_norm": 0.001449294133896928,
+      "kl": 0.00034862756729125977,
+      "learning_rate": 8e-07,
+      "loss": 0.0,
+      "num_tokens": 288920.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 16
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 466.9375,
+      "epoch": 0.03341523341523341,
+      "grad_norm": 0.000550624611359764,
+      "kl": 0.00036454200744628906,
+      "learning_rate": 8.499999999999999e-07,
+      "loss": 0.0,
+      "num_tokens": 306990.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 17
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 425.15625,
+      "epoch": 0.03538083538083538,
+      "grad_norm": 0.4966139299418229,
+      "kl": 0.0004194974899291992,
+      "learning_rate": 9e-07,
+      "loss": -0.059,
+      "num_tokens": 324483.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 18
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 558.84375,
+      "epoch": 0.03734643734643735,
+      "grad_norm": 0.0005133453694281673,
+      "kl": 0.000286102294921875,
+      "learning_rate": 9.499999999999999e-07,
+      "loss": 0.0,
+      "num_tokens": 346726.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 19
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 462.96875,
+      "epoch": 0.03931203931203931,
+      "grad_norm": 0.7524081119290347,
+      "kl": 0.0004723072052001953,
+      "learning_rate": 1e-06,
+      "loss": -0.051,
+      "num_tokens": 364915.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 20
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 442.84375,
+      "epoch": 0.041277641277641275,
+      "grad_norm": 0.002276209851987083,
+      "kl": 0.00047028064727783203,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 382538.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 21
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 487.8125,
+      "epoch": 0.043243243243243246,
+      "grad_norm": 0.0009159776748314612,
+      "kl": 0.0003075599670410156,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 402660.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 22
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 521.09375,
+      "epoch": 0.04520884520884521,
+      "grad_norm": 0.4431344945845042,
+      "kl": 0.0008206367492675781,
+      "learning_rate": 1e-06,
+      "loss": 0.0654,
+      "num_tokens": 423169.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 23
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 424.9375,
+      "epoch": 0.04717444717444717,
+      "grad_norm": 0.00201223494239014,
+      "kl": 0.0007357597351074219,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 439839.0,
+      "reward": 0.0625,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0625,
+      "step": 24
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 593.625,
+      "epoch": 0.04914004914004914,
+      "grad_norm": 0.18561111784701523,
+      "kl": 0.0019412040710449219,
+      "learning_rate": 1e-06,
+      "loss": -0.0044,
+      "num_tokens": 463265.0,
+      "reward": 0.09375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.09375,
+      "step": 25
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 280.375,
+      "epoch": 0.05110565110565111,
+      "grad_norm": 5.884242558131956,
+      "kl": 0.0025135278701782227,
+      "learning_rate": 1e-06,
+      "loss": -0.0569,
+      "num_tokens": 475453.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 26
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 513.90625,
+      "epoch": 0.05307125307125307,
+      "grad_norm": 0.0019098622203968013,
+      "kl": 0.0006109476089477539,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 495570.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 27
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 537.8125,
+      "epoch": 0.055036855036855035,
+      "grad_norm": 0.22075214354309136,
+      "kl": 0.0009493827819824219,
+      "learning_rate": 1e-06,
+      "loss": -0.0333,
+      "num_tokens": 516232.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 28
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 485.6875,
+      "epoch": 0.057002457002457,
+      "grad_norm": 0.0027609914881789005,
+      "kl": 0.0013211965560913086,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 534828.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 29
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 394.8125,
+      "epoch": 0.05896805896805897,
+      "grad_norm": 1035336.9697758019,
+      "kl": 7488.001363635063,
+      "learning_rate": 1e-06,
+      "loss": 7.5002,
+      "num_tokens": 550682.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 30
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 472.96875,
+      "epoch": 0.060933660933660934,
+      "grad_norm": 0.005787911590971138,
+      "kl": 0.002418994903564453,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 568901.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 31
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 532.1875,
+      "epoch": 0.0628992628992629,
+      "grad_norm": 0.22606153101175416,
+      "kl": 0.0019965171813964844,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 589439.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 32
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 443.78125,
+      "epoch": 0.06486486486486487,
+      "grad_norm": 0.35741687124698773,
+      "kl": 0.002090930938720703,
+      "learning_rate": 1e-06,
+      "loss": 0.007,
+      "num_tokens": 606976.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 33
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 582.25,
+      "epoch": 0.06683046683046683,
+      "grad_norm": 0.0012562903075668153,
+      "kl": 0.0010848045349121094,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 629146.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 34
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 445.96875,
+      "epoch": 0.0687960687960688,
+      "grad_norm": 0.9035998758815753,
+      "kl": 0.0037384033203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0705,
+      "num_tokens": 646643.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 35
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 444.6875,
+      "epoch": 0.07076167076167077,
+      "grad_norm": 0.3422524583969719,
+      "kl": 0.0064411163330078125,
+      "learning_rate": 1e-06,
+      "loss": -0.0749,
+      "num_tokens": 666223.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 36
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 375.15625,
+      "epoch": 0.07272727272727272,
+      "grad_norm": 0.6543309959643355,
+      "kl": 0.025150299072265625,
+      "learning_rate": 1e-06,
+      "loss": -0.074,
+      "num_tokens": 681430.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 37
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 465.40625,
+      "epoch": 0.0746928746928747,
+      "grad_norm": 0.4629932120108371,
+      "kl": 0.006771087646484375,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 699985.0,
+      "reward": 0.09375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.09375,
+      "step": 38
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 386.46875,
+      "epoch": 0.07665847665847667,
+      "grad_norm": 0.2536795642942375,
+      "kl": 0.010662078857421875,
+      "learning_rate": 1e-06,
+      "loss": -0.0128,
+      "num_tokens": 715666.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 39
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 398.90625,
+      "epoch": 0.07862407862407862,
+      "grad_norm": 0.36055266191548657,
+      "kl": 0.014057159423828125,
+      "learning_rate": 1e-06,
+      "loss": -0.0222,
+      "num_tokens": 732739.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 40
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 497.78125,
+      "epoch": 0.08058968058968059,
+      "grad_norm": 0.5439508463238408,
+      "kl": 0.011962890625,
+      "learning_rate": 1e-06,
+      "loss": 0.0736,
+      "num_tokens": 752036.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 41
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 553.9375,
+      "epoch": 0.08255528255528255,
+      "grad_norm": 0.2940020586639228,
+      "kl": 0.01055145263671875,
+      "learning_rate": 1e-06,
+      "loss": -0.0098,
+      "num_tokens": 774304.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 42
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 506.78125,
+      "epoch": 0.08452088452088452,
+      "grad_norm": 0.34987523935436315,
+      "kl": 0.022228240966796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0136,
+      "num_tokens": 794573.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 43
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 534.71875,
+      "epoch": 0.08648648648648649,
+      "grad_norm": 0.28775109044810143,
+      "kl": 0.016704559326171875,
+      "learning_rate": 1e-06,
+      "loss": -0.022,
+      "num_tokens": 814952.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 44
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 551.125,
+      "epoch": 0.08845208845208845,
+      "grad_norm": 0.5294418172877363,
+      "kl": 0.01171875,
+      "learning_rate": 1e-06,
+      "loss": 0.0127,
+      "num_tokens": 837090.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 45
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 424.09375,
+      "epoch": 0.09041769041769042,
+      "grad_norm": 0.7442762340799661,
+      "kl": 0.0166015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0527,
+      "num_tokens": 855063.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 46
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 578.90625,
+      "epoch": 0.09238329238329239,
+      "grad_norm": 0.9964430347064654,
+      "kl": 0.01439666748046875,
+      "learning_rate": 1e-06,
+      "loss": -0.0347,
+      "num_tokens": 876960.0,
+      "reward": 0.1875,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.1875,
+      "step": 47
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 516.65625,
+      "epoch": 0.09434889434889435,
+      "grad_norm": 0.0029736329333483326,
+      "kl": 0.015178680419921875,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 897477.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 48
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 406.53125,
+      "epoch": 0.09631449631449632,
+      "grad_norm": 0.42366305164369583,
+      "kl": 0.0214385986328125,
+      "learning_rate": 1e-06,
+      "loss": 0.023,
+      "num_tokens": 913848.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 49
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 615.46875,
+      "epoch": 0.09828009828009827,
+      "grad_norm": 0.2894821606425995,
+      "kl": 0.010334014892578125,
+      "learning_rate": 1e-06,
+      "loss": -0.0008,
+      "num_tokens": 936513.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 50
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 484.40625,
+      "epoch": 0.10024570024570024,
+      "grad_norm": 0.8466055231288873,
+      "kl": 0.01735687255859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0053,
+      "num_tokens": 955662.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 51
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 603.40625,
+      "epoch": 0.10221130221130222,
+      "grad_norm": 0.4594988801570104,
+      "kl": 0.01914215087890625,
+      "learning_rate": 1e-06,
+      "loss": 0.0157,
+      "num_tokens": 978603.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 52
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 571.53125,
+      "epoch": 0.10417690417690417,
+      "grad_norm": 0.6301487761037352,
+      "kl": 0.01680755615234375,
+      "learning_rate": 1e-06,
+      "loss": -0.0518,
+      "num_tokens": 1000414.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 53
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.625,
+      "epoch": 0.10614250614250614,
+      "grad_norm": 0.1705061403375656,
+      "kl": 0.019412994384765625,
+      "learning_rate": 1e-06,
+      "loss": -0.0151,
+      "num_tokens": 1019984.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 54
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 586.5625,
+      "epoch": 0.10810810810810811,
+      "grad_norm": 0.3563070736118851,
+      "kl": 0.01416778564453125,
+      "learning_rate": 1e-06,
+      "loss": -0.0113,
+      "num_tokens": 1043134.0,
+      "reward": 0.03125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.03125,
+      "step": 55
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 573.78125,
+      "epoch": 0.11007371007371007,
+      "grad_norm": 0.8170380900607339,
+      "kl": 0.0185394287109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0107,
+      "num_tokens": 1065473.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 56
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 596.25,
+      "epoch": 0.11203931203931204,
+      "grad_norm": 0.6138181155074318,
+      "kl": 0.01763916015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0173,
+      "num_tokens": 1088233.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 57
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 579.96875,
+      "epoch": 0.114004914004914,
+      "grad_norm": 0.5592170560306685,
+      "kl": 0.015392303466796875,
+      "learning_rate": 1e-06,
+      "loss": -0.042,
+      "num_tokens": 1111490.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 58
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 590.3125,
+      "epoch": 0.11597051597051597,
+      "grad_norm": 0.004534498568535779,
+      "kl": 0.017578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 1134000.0,
+      "reward": 0.125,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.125,
+      "step": 59
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 552.75,
+      "epoch": 0.11793611793611794,
+      "grad_norm": 0.26122008088795556,
+      "kl": 0.019062042236328125,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 1155236.0,
+      "reward": 0.09375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.09375,
+      "step": 60
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 533.46875,
+      "epoch": 0.1199017199017199,
+      "grad_norm": 0.6246443042334695,
+      "kl": 0.01920318603515625,
+      "learning_rate": 1e-06,
+      "loss": -0.0014,
+      "num_tokens": 1175617.0,
+      "reward": 0.09375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.09375,
+      "step": 61
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 484.34375,
+      "epoch": 0.12186732186732187,
+      "grad_norm": 1.0553043786365885,
+      "kl": 0.0269622802734375,
+      "learning_rate": 1e-06,
+      "loss": 0.0461,
+      "num_tokens": 1194332.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 62
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 604.84375,
+      "epoch": 0.12383292383292384,
+      "grad_norm": 0.5614007392931322,
+      "kl": 0.0162506103515625,
+      "learning_rate": 1e-06,
+      "loss": 0.0098,
+      "num_tokens": 1217363.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 63
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 579.59375,
+      "epoch": 0.1257985257985258,
+      "grad_norm": 0.001324981221462767,
+      "kl": 0.0164794921875,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 1239884.0,
+      "reward": 0.0625,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0625,
+      "step": 64
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 643.0,
+      "epoch": 0.12776412776412777,
+      "grad_norm": 0.6061011991202276,
+      "kl": 0.01563262939453125,
+      "learning_rate": 1e-06,
+      "loss": -0.065,
+      "num_tokens": 1263644.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 65
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 487.875,
+      "epoch": 0.12972972972972974,
+      "grad_norm": 0.8366609324695296,
+      "kl": 0.02077484130859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0478,
+      "num_tokens": 1282820.0,
+      "reward": 0.15625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.15625,
+      "step": 66
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 662.59375,
+      "epoch": 0.1316953316953317,
+      "grad_norm": 0.17490281761649612,
+      "kl": 0.0147705078125,
+      "learning_rate": 1e-06,
+      "loss": 0.0204,
+      "num_tokens": 1307457.0,
+      "reward": 0.09375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.09375,
+      "step": 67
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 525.03125,
+      "epoch": 0.13366093366093365,
+      "grad_norm": 0.8215530702995869,
+      "kl": 0.020721435546875,
+      "learning_rate": 1e-06,
+      "loss": 0.064,
+      "num_tokens": 1327798.0,
+      "reward": 0.25,
+      "reward_std": 0.3535533845424652,
+      "rewards/reward_func": 0.25,
+      "step": 68
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 567.1875,
+      "epoch": 0.13562653562653562,
+      "grad_norm": 0.65089823105939,
+      "kl": 0.019927978515625,
+      "learning_rate": 1e-06,
+      "loss": -0.0026,
+      "num_tokens": 1350334.0,
+      "reward": 0.15625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.15625,
+      "step": 69
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 584.78125,
+      "epoch": 0.1375921375921376,
+      "grad_norm": 1.008422919649498,
+      "kl": 0.01912689208984375,
+      "learning_rate": 1e-06,
+      "loss": 0.0025,
+      "num_tokens": 1372311.0,
+      "reward": 0.25,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.25,
+      "step": 70
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 545.125,
+      "epoch": 0.13955773955773956,
+      "grad_norm": 0.6752475010590075,
+      "kl": 0.01819610595703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0177,
+      "num_tokens": 1393645.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 71
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 526.40625,
+      "epoch": 0.14152334152334153,
+      "grad_norm": 0.7868187210048034,
+      "kl": 0.0257568359375,
+      "learning_rate": 1e-06,
+      "loss": -0.0076,
+      "num_tokens": 1414260.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 72
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 465.46875,
+      "epoch": 0.14348894348894348,
+      "grad_norm": 0.40424658361331584,
+      "kl": 0.02496337890625,
+      "learning_rate": 1e-06,
+      "loss": 0.0034,
+      "num_tokens": 1432617.0,
+      "reward": 0.09375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.09375,
+      "step": 73
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 523.0,
+      "epoch": 0.14545454545454545,
+      "grad_norm": 0.7217516854340651,
+      "kl": 0.0191497802734375,
+      "learning_rate": 1e-06,
+      "loss": -0.0438,
+      "num_tokens": 1452631.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 74
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 564.59375,
+      "epoch": 0.14742014742014742,
+      "grad_norm": 1.020886020177138,
+      "kl": 0.02172088623046875,
+      "learning_rate": 1e-06,
+      "loss": 0.0443,
+      "num_tokens": 1474180.0,
+      "reward": 0.25,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.25,
+      "step": 75
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 599.03125,
+      "epoch": 0.1493857493857494,
+      "grad_norm": 0.38779629766219476,
+      "kl": 0.021942138671875,
+      "learning_rate": 1e-06,
+      "loss": 0.0253,
+      "num_tokens": 1496597.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 76
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 535.65625,
+      "epoch": 0.15135135135135136,
+      "grad_norm": 0.5231239100329028,
+      "kl": 0.0213470458984375,
+      "learning_rate": 1e-06,
+      "loss": -0.0072,
+      "num_tokens": 1518002.0,
+      "reward": 0.09375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.09375,
+      "step": 77
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 483.28125,
+      "epoch": 0.15331695331695333,
+      "grad_norm": 1.01964836669152,
+      "kl": 0.02301025390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0253,
+      "num_tokens": 1538251.0,
+      "reward": 0.1875,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.1875,
+      "step": 78
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 597.21875,
+      "epoch": 0.15528255528255527,
+      "grad_norm": 0.3894010167090417,
+      "kl": 0.01868438720703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0049,
+      "num_tokens": 1561066.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 79
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 494.21875,
+      "epoch": 0.15724815724815724,
+      "grad_norm": 0.4559857762776204,
+      "kl": 0.02178955078125,
+      "learning_rate": 1e-06,
+      "loss": -0.0144,
+      "num_tokens": 1579851.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 80
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 459.4375,
+      "epoch": 0.15921375921375922,
+      "grad_norm": 0.7408328801166559,
+      "kl": 0.02642822265625,
+      "learning_rate": 1e-06,
+      "loss": 0.0034,
+      "num_tokens": 1597675.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 81
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 601.90625,
+      "epoch": 0.16117936117936119,
+      "grad_norm": 0.4202464041477536,
+      "kl": 0.01740264892578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0084,
+      "num_tokens": 1621632.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 82
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 517.21875,
+      "epoch": 0.16314496314496316,
+      "grad_norm": 0.7785902853786559,
+      "kl": 0.02386474609375,
+      "learning_rate": 1e-06,
+      "loss": 0.0036,
+      "num_tokens": 1641873.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 83
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 390.6875,
+      "epoch": 0.1651105651105651,
+      "grad_norm": 0.6791841008394204,
+      "kl": 0.034637451171875,
+      "learning_rate": 1e-06,
+      "loss": -0.0184,
+      "num_tokens": 1658453.0,
+      "reward": 0.1875,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.1875,
+      "step": 84
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 485.78125,
+      "epoch": 0.16707616707616707,
+      "grad_norm": 0.4367719984766766,
+      "kl": 0.0251617431640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0156,
+      "num_tokens": 1676992.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 85
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 528.875,
+      "epoch": 0.16904176904176904,
+      "grad_norm": 0.7222029349504244,
+      "kl": 0.02223968505859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0459,
+      "num_tokens": 1700584.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 86
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 658.125,
+      "epoch": 0.171007371007371,
+      "grad_norm": 0.34834776108087606,
+      "kl": 0.01824951171875,
+      "learning_rate": 1e-06,
+      "loss": 0.0326,
+      "num_tokens": 1725524.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 87
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 644.5625,
+      "epoch": 0.17297297297297298,
+      "grad_norm": 0.6750518830704131,
+      "kl": 0.01828765869140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0412,
+      "num_tokens": 1749556.0,
+      "reward": 0.15625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.15625,
+      "step": 88
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 547.15625,
+      "epoch": 0.17493857493857493,
+      "grad_norm": 0.31981853904702845,
+      "kl": 0.01940155029296875,
+      "learning_rate": 1e-06,
+      "loss": 0.012,
+      "num_tokens": 1770175.0,
+      "reward": 0.1875,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.1875,
+      "step": 89
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 475.75,
+      "epoch": 0.1769041769041769,
+      "grad_norm": 1.176119812297408,
+      "kl": 0.0265045166015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0174,
+      "num_tokens": 1789467.0,
+      "reward": 0.1875,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.1875,
+      "step": 90
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 448.625,
+      "epoch": 0.17886977886977887,
+      "grad_norm": 0.7247013137515755,
+      "kl": 0.02973175048828125,
+      "learning_rate": 1e-06,
+      "loss": 0.0025,
+      "num_tokens": 1807295.0,
+      "reward": 0.09375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.09375,
+      "step": 91
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 461.84375,
+      "epoch": 0.18083538083538084,
+      "grad_norm": 0.9576563539486035,
+      "kl": 0.027984619140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0572,
+      "num_tokens": 1827104.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 92
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 549.375,
+      "epoch": 0.1828009828009828,
+      "grad_norm": 0.8080609769164848,
+      "kl": 0.02375030517578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0641,
+      "num_tokens": 1848272.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 93
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 673.53125,
+      "epoch": 0.18476658476658478,
+      "grad_norm": 0.6070813087390035,
+      "kl": 0.0193634033203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0232,
+      "num_tokens": 1873533.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 94
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 582.75,
+      "epoch": 0.18673218673218672,
+      "grad_norm": 0.6479257231531951,
+      "kl": 0.0185394287109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0017,
+      "num_tokens": 1895601.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 95
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 443.5,
+      "epoch": 0.1886977886977887,
+      "grad_norm": 0.7375270346991556,
+      "kl": 0.0291290283203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0343,
+      "num_tokens": 1912993.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 96
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 499.125,
+      "epoch": 0.19066339066339066,
+      "grad_norm": 0.6426005454499798,
+      "kl": 0.0281524658203125,
+      "learning_rate": 1e-06,
+      "loss": -0.0165,
+      "num_tokens": 1933631.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 97
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 583.21875,
+      "epoch": 0.19262899262899263,
+      "grad_norm": 0.656305823795339,
+      "kl": 0.0197601318359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0098,
+      "num_tokens": 1955588.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 98
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 604.5,
+      "epoch": 0.1945945945945946,
+      "grad_norm": 0.7159925904721693,
+      "kl": 0.0202484130859375,
+      "learning_rate": 1e-06,
+      "loss": -0.0024,
+      "num_tokens": 1978466.0,
+      "reward": 0.15625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.15625,
+      "step": 99
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 623.8125,
+      "epoch": 0.19656019656019655,
+      "grad_norm": 0.6011532165345448,
+      "kl": 0.0197296142578125,
+      "learning_rate": 1e-06,
+      "loss": 0.015,
+      "num_tokens": 2001554.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 100
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 487.84375,
+      "epoch": 0.19852579852579852,
+      "grad_norm": 0.6696267166180018,
+      "kl": 0.02446746826171875,
+      "learning_rate": 1e-06,
+      "loss": 0.0323,
+      "num_tokens": 2020681.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 101
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 512.90625,
+      "epoch": 0.2004914004914005,
+      "grad_norm": 0.5144645053944626,
+      "kl": 0.0222320556640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0308,
+      "num_tokens": 2040058.0,
+      "reward": 0.0625,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.0625,
+      "step": 102
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 492.59375,
+      "epoch": 0.20245700245700246,
+      "grad_norm": 0.9792203746023931,
+      "kl": 0.02777099609375,
+      "learning_rate": 1e-06,
+      "loss": 0.0185,
+      "num_tokens": 2059481.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 103
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 540.5625,
+      "epoch": 0.20442260442260443,
+      "grad_norm": 0.6773524274596018,
+      "kl": 0.0245513916015625,
+      "learning_rate": 1e-06,
+      "loss": 0.0115,
+      "num_tokens": 2079795.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 104
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 629.75,
+      "epoch": 0.20638820638820637,
+      "grad_norm": 0.0014408090990997875,
+      "kl": 0.020843505859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 2104281.0,
+      "reward": 0.0,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.0,
+      "step": 105
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 489.625,
+      "epoch": 0.20835380835380835,
+      "grad_norm": 0.43577632912535613,
+      "kl": 0.02980804443359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0265,
+      "num_tokens": 2124275.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 106
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.90625,
+      "epoch": 0.21031941031941032,
+      "grad_norm": 0.5136803725458018,
+      "kl": 0.02611541748046875,
+      "learning_rate": 1e-06,
+      "loss": 0.0184,
+      "num_tokens": 2143080.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 107
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 659.9375,
+      "epoch": 0.2122850122850123,
+      "grad_norm": 0.4897037723304078,
+      "kl": 0.0196380615234375,
+      "learning_rate": 1e-06,
+      "loss": -0.0062,
+      "num_tokens": 2167550.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 108
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 549.53125,
+      "epoch": 0.21425061425061426,
+      "grad_norm": 0.6823729897684875,
+      "kl": 0.020660400390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0027,
+      "num_tokens": 2188815.0,
+      "reward": 0.1875,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.1875,
+      "step": 109
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 492.4375,
+      "epoch": 0.21621621621621623,
+      "grad_norm": 0.8520053469926316,
+      "kl": 0.02719879150390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0077,
+      "num_tokens": 2207977.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 110
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 487.34375,
+      "epoch": 0.21818181818181817,
+      "grad_norm": 0.8983378201666078,
+      "kl": 0.024871826171875,
+      "learning_rate": 1e-06,
+      "loss": 0.0207,
+      "num_tokens": 2226818.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 111
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 534.71875,
+      "epoch": 0.22014742014742014,
+      "grad_norm": 0.6341326642846564,
+      "kl": 0.022857666015625,
+      "learning_rate": 1e-06,
+      "loss": 0.005,
+      "num_tokens": 2247351.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 112
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 509.96875,
+      "epoch": 0.2221130221130221,
+      "grad_norm": 0.5084014496191064,
+      "kl": 0.02394866943359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0121,
+      "num_tokens": 2267646.0,
+      "reward": 0.1875,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.1875,
+      "step": 113
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 493.46875,
+      "epoch": 0.22407862407862408,
+      "grad_norm": 0.47081172119878306,
+      "kl": 0.023681640625,
+      "learning_rate": 1e-06,
+      "loss": 0.003,
+      "num_tokens": 2287461.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 114
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 593.46875,
+      "epoch": 0.22604422604422605,
+      "grad_norm": 0.5402745390200868,
+      "kl": 0.0226593017578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0254,
+      "num_tokens": 2310300.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 115
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 659.34375,
+      "epoch": 0.228009828009828,
+      "grad_norm": 0.7384552819583179,
+      "kl": 0.01863861083984375,
+      "learning_rate": 1e-06,
+      "loss": -0.011,
+      "num_tokens": 2336289.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 116
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 426.0625,
+      "epoch": 0.22997542997542997,
+      "grad_norm": 0.9856793371703316,
+      "kl": 0.029998779296875,
+      "learning_rate": 1e-06,
+      "loss": 0.0107,
+      "num_tokens": 2353069.0,
+      "reward": 0.1875,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.1875,
+      "step": 117
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 553.15625,
+      "epoch": 0.23194103194103194,
+      "grad_norm": 0.6524109632549928,
+      "kl": 0.0211334228515625,
+      "learning_rate": 1e-06,
+      "loss": 0.0166,
+      "num_tokens": 2374500.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 118
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 495.96875,
+      "epoch": 0.2339066339066339,
+      "grad_norm": 0.750164710998597,
+      "kl": 0.0268707275390625,
+      "learning_rate": 1e-06,
+      "loss": -0.0291,
+      "num_tokens": 2393455.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 119
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 519.375,
+      "epoch": 0.23587223587223588,
+      "grad_norm": 0.8136338780518924,
+      "kl": 0.0233917236328125,
+      "learning_rate": 1e-06,
+      "loss": 0.0213,
+      "num_tokens": 2413821.0,
+      "reward": 0.25,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.25,
+      "step": 120
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 492.5,
+      "epoch": 0.23783783783783785,
+      "grad_norm": 0.9472454617054751,
+      "kl": 0.0245513916015625,
+      "learning_rate": 1e-06,
+      "loss": 0.0434,
+      "num_tokens": 2432727.0,
+      "reward": 0.3125,
+      "reward_std": 0.3535533845424652,
+      "rewards/reward_func": 0.3125,
+      "step": 121
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 650.125,
+      "epoch": 0.2398034398034398,
+      "grad_norm": 0.7941644508393335,
+      "kl": 0.0188446044921875,
+      "learning_rate": 1e-06,
+      "loss": 0.0161,
+      "num_tokens": 2457059.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 122
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 505.40625,
+      "epoch": 0.24176904176904176,
+      "grad_norm": 0.9756007077675318,
+      "kl": 0.02396392822265625,
+      "learning_rate": 1e-06,
+      "loss": 0.0303,
+      "num_tokens": 2477004.0,
+      "reward": 0.53125,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.53125,
+      "step": 123
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 483.96875,
+      "epoch": 0.24373464373464374,
+      "grad_norm": 0.6941208484996456,
+      "kl": 0.027313232421875,
+      "learning_rate": 1e-06,
+      "loss": 0.0341,
+      "num_tokens": 2495959.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 124
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 561.3125,
+      "epoch": 0.2457002457002457,
+      "grad_norm": 0.5265733548592475,
+      "kl": 0.01922607421875,
+      "learning_rate": 1e-06,
+      "loss": 0.0423,
+      "num_tokens": 2517637.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 125
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 492.96875,
+      "epoch": 0.24766584766584768,
+      "grad_norm": 0.8350574297158757,
+      "kl": 0.023712158203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0469,
+      "num_tokens": 2538338.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 126
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 573.5,
+      "epoch": 0.24963144963144962,
+      "grad_norm": 0.8216738883653201,
+      "kl": 0.02204132080078125,
+      "learning_rate": 1e-06,
+      "loss": 0.0218,
+      "num_tokens": 2560112.0,
+      "reward": 0.28125,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.28125,
+      "step": 127
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 529.3125,
+      "epoch": 0.2515970515970516,
+      "grad_norm": 0.78248285660239,
+      "kl": 0.02326202392578125,
+      "learning_rate": 1e-06,
+      "loss": -0.044,
+      "num_tokens": 2580546.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 128
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 602.1875,
+      "epoch": 0.25356265356265356,
+      "grad_norm": 0.2772418236153871,
+      "kl": 0.019317626953125,
+      "learning_rate": 1e-06,
+      "loss": 0.0027,
+      "num_tokens": 2604460.0,
+      "reward": 0.15625,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.15625,
+      "step": 129
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 548.0625,
+      "epoch": 0.25552825552825553,
+      "grad_norm": 0.6625525706905956,
+      "kl": 0.02095794677734375,
+      "learning_rate": 1e-06,
+      "loss": 0.0115,
+      "num_tokens": 2624972.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 130
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 585.9375,
+      "epoch": 0.2574938574938575,
+      "grad_norm": 0.752169210277677,
+      "kl": 0.021514892578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0359,
+      "num_tokens": 2648150.0,
+      "reward": 0.375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.375,
+      "step": 131
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 574.0625,
+      "epoch": 0.2594594594594595,
+      "grad_norm": 0.5002363204112915,
+      "kl": 0.0212249755859375,
+      "learning_rate": 1e-06,
+      "loss": 0.023,
+      "num_tokens": 2670338.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 132
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 526.625,
+      "epoch": 0.26142506142506144,
+      "grad_norm": 0.9674758127355131,
+      "kl": 0.0261077880859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0159,
+      "num_tokens": 2691642.0,
+      "reward": 0.34375,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.34375,
+      "step": 133
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 412.59375,
+      "epoch": 0.2633906633906634,
+      "grad_norm": 1.0324181511597994,
+      "kl": 0.0365447998046875,
+      "learning_rate": 1e-06,
+      "loss": 0.0109,
+      "num_tokens": 2708471.0,
+      "reward": 0.53125,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.53125,
+      "step": 134
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 501.1875,
+      "epoch": 0.26535626535626533,
+      "grad_norm": 0.646085784140391,
+      "kl": 0.02829742431640625,
+      "learning_rate": 1e-06,
+      "loss": -0.0245,
+      "num_tokens": 2728631.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 135
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 524.59375,
+      "epoch": 0.2673218673218673,
+      "grad_norm": 0.5827914647557386,
+      "kl": 0.0233001708984375,
+      "learning_rate": 1e-06,
+      "loss": -0.0133,
+      "num_tokens": 2748682.0,
+      "reward": 0.375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.375,
+      "step": 136
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 493.625,
+      "epoch": 0.26928746928746927,
+      "grad_norm": 0.5383283228577356,
+      "kl": 0.02911376953125,
+      "learning_rate": 1e-06,
+      "loss": 0.006,
+      "num_tokens": 2768384.0,
+      "reward": 0.15625,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.15625,
+      "step": 137
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 563.34375,
+      "epoch": 0.27125307125307124,
+      "grad_norm": 0.44514691751088103,
+      "kl": 0.02154541015625,
+      "learning_rate": 1e-06,
+      "loss": -0.045,
+      "num_tokens": 2790263.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 138
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 520.8125,
+      "epoch": 0.2732186732186732,
+      "grad_norm": 0.7190292298516152,
+      "kl": 0.0236663818359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0132,
+      "num_tokens": 2810351.0,
+      "reward": 0.375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.375,
+      "step": 139
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 527.34375,
+      "epoch": 0.2751842751842752,
+      "grad_norm": 0.7242965068485637,
+      "kl": 0.0230255126953125,
+      "learning_rate": 1e-06,
+      "loss": 0.001,
+      "num_tokens": 2830716.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 140
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 565.8125,
+      "epoch": 0.27714987714987716,
+      "grad_norm": 0.6228089750555335,
+      "kl": 0.02398681640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0041,
+      "num_tokens": 2853810.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 141
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 555.1875,
+      "epoch": 0.2791154791154791,
+      "grad_norm": 0.37035042072698693,
+      "kl": 0.02410888671875,
+      "learning_rate": 1e-06,
+      "loss": 0.0047,
+      "num_tokens": 2875056.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 142
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 589.65625,
+      "epoch": 0.2810810810810811,
+      "grad_norm": 0.658035150283911,
+      "kl": 0.02394866943359375,
+      "learning_rate": 1e-06,
+      "loss": 0.025,
+      "num_tokens": 2897357.0,
+      "reward": 0.4375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.4375,
+      "step": 143
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 543.8125,
+      "epoch": 0.28304668304668307,
+      "grad_norm": 0.7501331474001123,
+      "kl": 0.025238037109375,
+      "learning_rate": 1e-06,
+      "loss": -0.0123,
+      "num_tokens": 2917753.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 144
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 612.28125,
+      "epoch": 0.28501228501228504,
+      "grad_norm": 0.5446469883013957,
+      "kl": 0.0233612060546875,
+      "learning_rate": 1e-06,
+      "loss": 0.0054,
+      "num_tokens": 2940774.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 145
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 539.78125,
+      "epoch": 0.28697788697788695,
+      "grad_norm": 0.6092756185830317,
+      "kl": 0.02280426025390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0263,
+      "num_tokens": 2961247.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 146
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 466.28125,
+      "epoch": 0.2889434889434889,
+      "grad_norm": 0.6941889736363939,
+      "kl": 0.0313873291015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0046,
+      "num_tokens": 2979622.0,
+      "reward": 0.09375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.09375,
+      "step": 147
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 567.53125,
+      "epoch": 0.2909090909090909,
+      "grad_norm": 0.3736888500181184,
+      "kl": 0.023223876953125,
+      "learning_rate": 1e-06,
+      "loss": 0.0099,
+      "num_tokens": 3002039.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 148
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 467.59375,
+      "epoch": 0.29287469287469287,
+      "grad_norm": 0.8319282300423898,
+      "kl": 0.02935791015625,
+      "learning_rate": 1e-06,
+      "loss": 0.043,
+      "num_tokens": 3019816.0,
+      "reward": 0.375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.375,
+      "step": 149
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 492.25,
+      "epoch": 0.29484029484029484,
+      "grad_norm": 0.7325276211486558,
+      "kl": 0.0263671875,
+      "learning_rate": 1e-06,
+      "loss": 0.0293,
+      "num_tokens": 3039146.0,
+      "reward": 0.46875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.46875,
+      "step": 150
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 592.1875,
+      "epoch": 0.2968058968058968,
+      "grad_norm": 0.43633918201867017,
+      "kl": 0.0228424072265625,
+      "learning_rate": 1e-06,
+      "loss": -0.0122,
+      "num_tokens": 3062102.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 151
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 563.1875,
+      "epoch": 0.2987714987714988,
+      "grad_norm": 0.38992921040393713,
+      "kl": 0.025421142578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0327,
+      "num_tokens": 3084284.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 152
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 634.9375,
+      "epoch": 0.30073710073710075,
+      "grad_norm": 0.41943188112426816,
+      "kl": 0.0197906494140625,
+      "learning_rate": 1e-06,
+      "loss": 0.001,
+      "num_tokens": 3108216.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 153
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 636.9375,
+      "epoch": 0.3027027027027027,
+      "grad_norm": 0.19968498604663434,
+      "kl": 0.02194976806640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0001,
+      "num_tokens": 3133100.0,
+      "reward": 0.09375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.09375,
+      "step": 154
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 568.21875,
+      "epoch": 0.3046683046683047,
+      "grad_norm": 0.7545074739064522,
+      "kl": 0.02685546875,
+      "learning_rate": 1e-06,
+      "loss": 0.04,
+      "num_tokens": 3154579.0,
+      "reward": 0.46875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.46875,
+      "step": 155
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 529.8125,
+      "epoch": 0.30663390663390666,
+      "grad_norm": 0.6884386708248261,
+      "kl": 0.025360107421875,
+      "learning_rate": 1e-06,
+      "loss": -0.0075,
+      "num_tokens": 3174845.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 156
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 513.0625,
+      "epoch": 0.3085995085995086,
+      "grad_norm": 0.5265944726858988,
+      "kl": 0.0264892578125,
+      "learning_rate": 1e-06,
+      "loss": -0.0245,
+      "num_tokens": 3195631.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 157
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 520.0,
+      "epoch": 0.31056511056511055,
+      "grad_norm": 0.45924555811174794,
+      "kl": 0.026092529296875,
+      "learning_rate": 1e-06,
+      "loss": -0.0189,
+      "num_tokens": 3216143.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 158
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 568.15625,
+      "epoch": 0.3125307125307125,
+      "grad_norm": 0.5301940769841443,
+      "kl": 0.025177001953125,
+      "learning_rate": 1e-06,
+      "loss": -0.0008,
+      "num_tokens": 3238264.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 159
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 508.59375,
+      "epoch": 0.3144963144963145,
+      "grad_norm": 0.8755698562076372,
+      "kl": 0.0290069580078125,
+      "learning_rate": 1e-06,
+      "loss": 0.0536,
+      "num_tokens": 3257837.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 160
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 481.15625,
+      "epoch": 0.31646191646191646,
+      "grad_norm": 0.6081291396581243,
+      "kl": 0.0274810791015625,
+      "learning_rate": 1e-06,
+      "loss": 0.0018,
+      "num_tokens": 3276820.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 161
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 598.28125,
+      "epoch": 0.31842751842751843,
+      "grad_norm": 0.5215378127068343,
+      "kl": 0.0220794677734375,
+      "learning_rate": 1e-06,
+      "loss": 0.0012,
+      "num_tokens": 3299357.0,
+      "reward": 0.375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.375,
+      "step": 162
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 487.09375,
+      "epoch": 0.3203931203931204,
+      "grad_norm": 0.5432850798950747,
+      "kl": 0.03083038330078125,
+      "learning_rate": 1e-06,
+      "loss": -0.0034,
+      "num_tokens": 3318084.0,
+      "reward": 0.5,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.5,
+      "step": 163
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 531.125,
+      "epoch": 0.32235872235872237,
+      "grad_norm": 0.46027668679847555,
+      "kl": 0.026092529296875,
+      "learning_rate": 1e-06,
+      "loss": 0.0154,
+      "num_tokens": 3338328.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 164
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 624.8125,
+      "epoch": 0.32432432432432434,
+      "grad_norm": 0.18474075838750667,
+      "kl": 0.02459716796875,
+      "learning_rate": 1e-06,
+      "loss": -0.0011,
+      "num_tokens": 3361788.0,
+      "reward": 0.21875,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.21875,
+      "step": 165
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 549.78125,
+      "epoch": 0.3262899262899263,
+      "grad_norm": 0.7718886882448309,
+      "kl": 0.027069091796875,
+      "learning_rate": 1e-06,
+      "loss": -0.005,
+      "num_tokens": 3382981.0,
+      "reward": 0.46875,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.46875,
+      "step": 166
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 575.78125,
+      "epoch": 0.32825552825552823,
+      "grad_norm": 0.5895346584970902,
+      "kl": 0.02323150634765625,
+      "learning_rate": 1e-06,
+      "loss": -0.0178,
+      "num_tokens": 3405274.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 167
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 576.875,
+      "epoch": 0.3302211302211302,
+      "grad_norm": 0.4959807619539926,
+      "kl": 0.0241851806640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0022,
+      "num_tokens": 3427392.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 168
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 561.78125,
+      "epoch": 0.33218673218673217,
+      "grad_norm": 0.8681759456851222,
+      "kl": 0.0249176025390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0274,
+      "num_tokens": 3448681.0,
+      "reward": 0.34375,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.34375,
+      "step": 169
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 483.53125,
+      "epoch": 0.33415233415233414,
+      "grad_norm": 0.5297566985301704,
+      "kl": 0.0336761474609375,
+      "learning_rate": 1e-06,
+      "loss": -0.0019,
+      "num_tokens": 3467212.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 170
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 581.40625,
+      "epoch": 0.3361179361179361,
+      "grad_norm": 0.6192719333892511,
+      "kl": 0.025787353515625,
+      "learning_rate": 1e-06,
+      "loss": 0.0075,
+      "num_tokens": 3489355.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 171
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 709.09375,
+      "epoch": 0.3380835380835381,
+      "grad_norm": 0.47423370876978443,
+      "kl": 0.0204620361328125,
+      "learning_rate": 1e-06,
+      "loss": -0.0251,
+      "num_tokens": 3515834.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 172
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 614.21875,
+      "epoch": 0.34004914004914005,
+      "grad_norm": 0.6479887044288859,
+      "kl": 0.0229034423828125,
+      "learning_rate": 1e-06,
+      "loss": -0.0212,
+      "num_tokens": 3539067.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 173
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 552.6875,
+      "epoch": 0.342014742014742,
+      "grad_norm": 0.4712074892004097,
+      "kl": 0.0272216796875,
+      "learning_rate": 1e-06,
+      "loss": -0.0209,
+      "num_tokens": 3560435.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 174
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 599.75,
+      "epoch": 0.343980343980344,
+      "grad_norm": 0.7325904251977917,
+      "kl": 0.0259552001953125,
+      "learning_rate": 1e-06,
+      "loss": 0.0087,
+      "num_tokens": 3582617.0,
+      "reward": 0.25,
+      "reward_std": 0.3535533845424652,
+      "rewards/reward_func": 0.25,
+      "step": 175
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 589.46875,
+      "epoch": 0.34594594594594597,
+      "grad_norm": 0.718404368627785,
+      "kl": 0.024017333984375,
+      "learning_rate": 1e-06,
+      "loss": 0.0069,
+      "num_tokens": 3606486.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 176
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 525.0625,
+      "epoch": 0.34791154791154794,
+      "grad_norm": 0.5807447821668908,
+      "kl": 0.0262451171875,
+      "learning_rate": 1e-06,
+      "loss": 0.006,
+      "num_tokens": 3626528.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 177
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 539.875,
+      "epoch": 0.34987714987714985,
+      "grad_norm": 0.3836295461022733,
+      "kl": 0.025665283203125,
+      "learning_rate": 1e-06,
+      "loss": -0.0233,
+      "num_tokens": 3647890.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 178
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 641.28125,
+      "epoch": 0.3518427518427518,
+      "grad_norm": 0.3325268331032711,
+      "kl": 0.02361297607421875,
+      "learning_rate": 1e-06,
+      "loss": 0.0117,
+      "num_tokens": 3672977.0,
+      "reward": 0.1875,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.1875,
+      "step": 179
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 611.1875,
+      "epoch": 0.3538083538083538,
+      "grad_norm": 0.5612204711943589,
+      "kl": 0.025909423828125,
+      "learning_rate": 1e-06,
+      "loss": 0.014,
+      "num_tokens": 3696201.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 180
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 587.125,
+      "epoch": 0.35577395577395576,
+      "grad_norm": 0.48905557864261767,
+      "kl": 0.0284881591796875,
+      "learning_rate": 1e-06,
+      "loss": -0.0135,
+      "num_tokens": 3718645.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 181
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 532.8125,
+      "epoch": 0.35773955773955773,
+      "grad_norm": 0.6196793685491477,
+      "kl": 0.0283050537109375,
+      "learning_rate": 1e-06,
+      "loss": -0.021,
+      "num_tokens": 3739419.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 182
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 491.25,
+      "epoch": 0.3597051597051597,
+      "grad_norm": 0.7943049071356907,
+      "kl": 0.0270843505859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0343,
+      "num_tokens": 3758175.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 183
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 629.3125,
+      "epoch": 0.3616707616707617,
+      "grad_norm": 0.4358875914582024,
+      "kl": 0.02367401123046875,
+      "learning_rate": 1e-06,
+      "loss": 0.0133,
+      "num_tokens": 3782789.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 184
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 540.59375,
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.7097269605040931,
+      "kl": 0.028656005859375,
+      "learning_rate": 1e-06,
+      "loss": 0.023,
+      "num_tokens": 3804202.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 185
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 517.40625,
+      "epoch": 0.3656019656019656,
+      "grad_norm": 0.8226505283368968,
+      "kl": 0.0278778076171875,
+      "learning_rate": 1e-06,
+      "loss": -0.0374,
+      "num_tokens": 3824813.0,
+      "reward": 0.3125,
+      "reward_std": 0.3535533845424652,
+      "rewards/reward_func": 0.3125,
+      "step": 186
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 499.3125,
+      "epoch": 0.3675675675675676,
+      "grad_norm": 0.8593511870826509,
+      "kl": 0.026092529296875,
+      "learning_rate": 1e-06,
+      "loss": 0.0443,
+      "num_tokens": 3844419.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 187
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 497.5,
+      "epoch": 0.36953316953316956,
+      "grad_norm": 0.6026343875240253,
+      "kl": 0.0312042236328125,
+      "learning_rate": 1e-06,
+      "loss": -0.0226,
+      "num_tokens": 3863979.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 188
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 547.0625,
+      "epoch": 0.3714987714987715,
+      "grad_norm": 0.6577878892956752,
+      "kl": 0.030914306640625,
+      "learning_rate": 1e-06,
+      "loss": -0.002,
+      "num_tokens": 3885407.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 189
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 605.0625,
+      "epoch": 0.37346437346437344,
+      "grad_norm": 0.45127850046328344,
+      "kl": 0.024169921875,
+      "learning_rate": 1e-06,
+      "loss": -0.003,
+      "num_tokens": 3908713.0,
+      "reward": 0.40625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.40625,
+      "step": 190
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 529.46875,
+      "epoch": 0.3754299754299754,
+      "grad_norm": 0.5469774623716551,
+      "kl": 0.025115966796875,
+      "learning_rate": 1e-06,
+      "loss": -0.0056,
+      "num_tokens": 3928856.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 191
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 475.15625,
+      "epoch": 0.3773955773955774,
+      "grad_norm": 0.9374126682104605,
+      "kl": 0.0326385498046875,
+      "learning_rate": 1e-06,
+      "loss": 0.0543,
+      "num_tokens": 3947241.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 192
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 547.1875,
+      "epoch": 0.37936117936117936,
+      "grad_norm": 0.5527175997380437,
+      "kl": 0.0286407470703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0571,
+      "num_tokens": 3968765.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 193
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 581.53125,
+      "epoch": 0.38132678132678133,
+      "grad_norm": 0.5091715028558698,
+      "kl": 0.02716064453125,
+      "learning_rate": 1e-06,
+      "loss": -0.0032,
+      "num_tokens": 3990744.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 194
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 545.3125,
+      "epoch": 0.3832923832923833,
+      "grad_norm": 0.6906699369220881,
+      "kl": 0.026611328125,
+      "learning_rate": 1e-06,
+      "loss": 0.0317,
+      "num_tokens": 4011242.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 195
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 553.0625,
+      "epoch": 0.38525798525798527,
+      "grad_norm": 0.803923186478945,
+      "kl": 0.026153564453125,
+      "learning_rate": 1e-06,
+      "loss": 0.0114,
+      "num_tokens": 4032454.0,
+      "reward": 0.34375,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.34375,
+      "step": 196
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 514.875,
+      "epoch": 0.38722358722358724,
+      "grad_norm": 0.5425045400963313,
+      "kl": 0.029937744140625,
+      "learning_rate": 1e-06,
+      "loss": -0.0032,
+      "num_tokens": 4053082.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 197
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 677.78125,
+      "epoch": 0.3891891891891892,
+      "grad_norm": 0.6672952970651821,
+      "kl": 0.0213623046875,
+      "learning_rate": 1e-06,
+      "loss": 0.0023,
+      "num_tokens": 4078691.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 198
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 512.03125,
+      "epoch": 0.3911547911547912,
+      "grad_norm": 0.476359304779961,
+      "kl": 0.035552978515625,
+      "learning_rate": 1e-06,
+      "loss": 0.0066,
+      "num_tokens": 4099720.0,
+      "reward": 0.21875,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.21875,
+      "step": 199
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 743.375,
+      "epoch": 0.3931203931203931,
+      "grad_norm": 0.4679126082574762,
+      "kl": 0.0198974609375,
+      "learning_rate": 1e-06,
+      "loss": -0.0029,
+      "num_tokens": 4127002.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 200
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 557.0,
+      "epoch": 0.39508599508599507,
+      "grad_norm": 0.6988574973859861,
+      "kl": 0.028167724609375,
+      "learning_rate": 1e-06,
+      "loss": -0.0327,
+      "num_tokens": 4148548.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 201
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 566.28125,
+      "epoch": 0.39705159705159704,
+      "grad_norm": 0.7768469504643993,
+      "kl": 0.0265960693359375,
+      "learning_rate": 1e-06,
+      "loss": -0.0108,
+      "num_tokens": 4170063.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 202
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 610.5,
+      "epoch": 0.399017199017199,
+      "grad_norm": 0.3681456638360366,
+      "kl": 0.0287933349609375,
+      "learning_rate": 1e-06,
+      "loss": -0.0088,
+      "num_tokens": 4194559.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 203
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 639.34375,
+      "epoch": 0.400982800982801,
+      "grad_norm": 0.6149062975054262,
+      "kl": 0.0237884521484375,
+      "learning_rate": 1e-06,
+      "loss": 0.0481,
+      "num_tokens": 4218340.0,
+      "reward": 0.1875,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.1875,
+      "step": 204
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 524.15625,
+      "epoch": 0.40294840294840295,
+      "grad_norm": 0.7831991086164788,
+      "kl": 0.0277557373046875,
+      "learning_rate": 1e-06,
+      "loss": 0.0474,
+      "num_tokens": 4239031.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 205
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 553.5625,
+      "epoch": 0.4049140049140049,
+      "grad_norm": 0.7849973591038859,
+      "kl": 0.0251617431640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0378,
+      "num_tokens": 4260199.0,
+      "reward": 0.40625,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.40625,
+      "step": 206
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.75,
+      "epoch": 0.4068796068796069,
+      "grad_norm": 0.6839800139372438,
+      "kl": 0.03155517578125,
+      "learning_rate": 1e-06,
+      "loss": 0.016,
+      "num_tokens": 4278919.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 207
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 606.34375,
+      "epoch": 0.40884520884520886,
+      "grad_norm": 0.3896782835395279,
+      "kl": 0.026397705078125,
+      "learning_rate": 1e-06,
+      "loss": 0.0216,
+      "num_tokens": 4302006.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 208
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 629.21875,
+      "epoch": 0.41081081081081083,
+      "grad_norm": 0.5046764418165803,
+      "kl": 0.0250244140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0093,
+      "num_tokens": 4325779.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 209
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 639.59375,
+      "epoch": 0.41277641277641275,
+      "grad_norm": 0.5479053649859522,
+      "kl": 0.0309600830078125,
+      "learning_rate": 1e-06,
+      "loss": -0.008,
+      "num_tokens": 4349352.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 210
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 608.875,
+      "epoch": 0.4147420147420147,
+      "grad_norm": 0.27291739633494205,
+      "kl": 0.025482177734375,
+      "learning_rate": 1e-06,
+      "loss": 0.008,
+      "num_tokens": 4372614.0,
+      "reward": 0.28125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.28125,
+      "step": 211
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 700.3125,
+      "epoch": 0.4167076167076167,
+      "grad_norm": 0.6410316114692441,
+      "kl": 0.0247802734375,
+      "learning_rate": 1e-06,
+      "loss": 0.0235,
+      "num_tokens": 4398294.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 212
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 558.625,
+      "epoch": 0.41867321867321866,
+      "grad_norm": 0.30488486047826086,
+      "kl": 0.028900146484375,
+      "learning_rate": 1e-06,
+      "loss": 0.0073,
+      "num_tokens": 4419278.0,
+      "reward": 0.28125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.28125,
+      "step": 213
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 454.71875,
+      "epoch": 0.42063882063882063,
+      "grad_norm": 0.6876906732910688,
+      "kl": 0.0352935791015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0201,
+      "num_tokens": 4437525.0,
+      "reward": 0.4375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.4375,
+      "step": 214
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 586.34375,
+      "epoch": 0.4226044226044226,
+      "grad_norm": 0.7070061651726116,
+      "kl": 0.0324859619140625,
+      "learning_rate": 1e-06,
+      "loss": -0.0072,
+      "num_tokens": 4459576.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 215
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 598.25,
+      "epoch": 0.4245700245700246,
+      "grad_norm": 0.6961718312404855,
+      "kl": 0.031036376953125,
+      "learning_rate": 1e-06,
+      "loss": 0.025,
+      "num_tokens": 4482428.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 216
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 550.5,
+      "epoch": 0.42653562653562654,
+      "grad_norm": 0.6664650861834673,
+      "kl": 0.030609130859375,
+      "learning_rate": 1e-06,
+      "loss": -0.0165,
+      "num_tokens": 4503774.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 217
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 492.875,
+      "epoch": 0.4285012285012285,
+      "grad_norm": 0.6916569742061264,
+      "kl": 0.0322113037109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0245,
+      "num_tokens": 4523210.0,
+      "reward": 0.53125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.53125,
+      "step": 218
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 556.125,
+      "epoch": 0.4304668304668305,
+      "grad_norm": 0.6079132124740041,
+      "kl": 0.0360260009765625,
+      "learning_rate": 1e-06,
+      "loss": 0.0504,
+      "num_tokens": 4544250.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 219
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 640.625,
+      "epoch": 0.43243243243243246,
+      "grad_norm": 0.49303925630539,
+      "kl": 0.02655029296875,
+      "learning_rate": 1e-06,
+      "loss": -0.0042,
+      "num_tokens": 4568206.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 220
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 591.15625,
+      "epoch": 0.43439803439803437,
+      "grad_norm": 0.3790789264426733,
+      "kl": 0.026641845703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0029,
+      "num_tokens": 4592127.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 221
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 559.5,
+      "epoch": 0.43636363636363634,
+      "grad_norm": 0.7205730486755546,
+      "kl": 0.0282440185546875,
+      "learning_rate": 1e-06,
+      "loss": 0.0047,
+      "num_tokens": 4613469.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 222
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 616.78125,
+      "epoch": 0.4383292383292383,
+      "grad_norm": 0.2994169582864983,
+      "kl": 0.0284271240234375,
+      "learning_rate": 1e-06,
+      "loss": 0.0124,
+      "num_tokens": 4636780.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 223
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 546.96875,
+      "epoch": 0.4402948402948403,
+      "grad_norm": 0.6443329354884543,
+      "kl": 0.03546142578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0477,
+      "num_tokens": 4658003.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 224
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 691.59375,
+      "epoch": 0.44226044226044225,
+      "grad_norm": 0.5810139994156356,
+      "kl": 0.0257415771484375,
+      "learning_rate": 1e-06,
+      "loss": 0.0137,
+      "num_tokens": 4683758.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 225
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 654.46875,
+      "epoch": 0.4442260442260442,
+      "grad_norm": 0.4628887828349593,
+      "kl": 0.0256500244140625,
+      "learning_rate": 1e-06,
+      "loss": -0.0144,
+      "num_tokens": 4708083.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 226
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.71875,
+      "epoch": 0.4461916461916462,
+      "grad_norm": 0.5979661864229516,
+      "kl": 0.0352325439453125,
+      "learning_rate": 1e-06,
+      "loss": 0.0019,
+      "num_tokens": 4727130.0,
+      "reward": 0.5625,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.5625,
+      "step": 227
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 541.5,
+      "epoch": 0.44815724815724817,
+      "grad_norm": 0.5705252239874116,
+      "kl": 0.0313720703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0057,
+      "num_tokens": 4747992.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 228
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 612.46875,
+      "epoch": 0.45012285012285014,
+      "grad_norm": 0.5836545557559704,
+      "kl": 0.027099609375,
+      "learning_rate": 1e-06,
+      "loss": -0.0091,
+      "num_tokens": 4771251.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 229
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 637.8125,
+      "epoch": 0.4520884520884521,
+      "grad_norm": 0.4498303433751672,
+      "kl": 0.0286102294921875,
+      "learning_rate": 1e-06,
+      "loss": -0.0034,
+      "num_tokens": 4796355.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 230
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 610.625,
+      "epoch": 0.4540540540540541,
+      "grad_norm": 0.6013423970019107,
+      "kl": 0.0283966064453125,
+      "learning_rate": 1e-06,
+      "loss": 0.0639,
+      "num_tokens": 4819261.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 231
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 584.0625,
+      "epoch": 0.456019656019656,
+      "grad_norm": 0.7699293584446985,
+      "kl": 0.03173828125,
+      "learning_rate": 1e-06,
+      "loss": 0.0023,
+      "num_tokens": 4841987.0,
+      "reward": 0.1875,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.1875,
+      "step": 232
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 617.21875,
+      "epoch": 0.45798525798525797,
+      "grad_norm": 0.5200942231005387,
+      "kl": 0.078887939453125,
+      "learning_rate": 1e-06,
+      "loss": 0.0358,
+      "num_tokens": 4865308.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 233
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 549.21875,
+      "epoch": 0.45995085995085994,
+      "grad_norm": 0.6584051899429707,
+      "kl": 0.0289764404296875,
+      "learning_rate": 1e-06,
+      "loss": 0.0274,
+      "num_tokens": 4886777.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 234
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 671.78125,
+      "epoch": 0.4619164619164619,
+      "grad_norm": 0.299051256935548,
+      "kl": 0.0264892578125,
+      "learning_rate": 1e-06,
+      "loss": -0.0096,
+      "num_tokens": 4911454.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 235
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 571.28125,
+      "epoch": 0.4638820638820639,
+      "grad_norm": 0.6402318572716406,
+      "kl": 0.03070068359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0066,
+      "num_tokens": 4933175.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 236
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 561.59375,
+      "epoch": 0.46584766584766585,
+      "grad_norm": 0.6512014891525182,
+      "kl": 0.033782958984375,
+      "learning_rate": 1e-06,
+      "loss": -0.0072,
+      "num_tokens": 4955004.0,
+      "reward": 0.5,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.5,
+      "step": 237
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 532.5625,
+      "epoch": 0.4678132678132678,
+      "grad_norm": 0.4328249392514584,
+      "kl": 0.03594970703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0331,
+      "num_tokens": 4976386.0,
+      "reward": 0.46875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.46875,
+      "step": 238
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 584.03125,
+      "epoch": 0.4697788697788698,
+      "grad_norm": 0.4634554918945285,
+      "kl": 0.032501220703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0107,
+      "num_tokens": 4999461.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 239
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 618.0,
+      "epoch": 0.47174447174447176,
+      "grad_norm": 0.00408085066316832,
+      "kl": 0.03265380859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 5023321.0,
+      "reward": 0.1875,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.1875,
+      "step": 240
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 562.40625,
+      "epoch": 0.47371007371007373,
+      "grad_norm": 0.7043144078754758,
+      "kl": 0.03973388671875,
+      "learning_rate": 1e-06,
+      "loss": 0.0176,
+      "num_tokens": 5044300.0,
+      "reward": 0.53125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.53125,
+      "step": 241
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 534.5,
+      "epoch": 0.4756756756756757,
+      "grad_norm": 1.0386591332218544,
+      "kl": 0.0329742431640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0069,
+      "num_tokens": 5064534.0,
+      "reward": 0.34375,
+      "reward_std": 0.39774755761027336,
+      "rewards/reward_func": 0.34375,
+      "step": 242
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 579.625,
+      "epoch": 0.4776412776412776,
+      "grad_norm": 0.5640869175323494,
+      "kl": 0.030670166015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0096,
+      "num_tokens": 5086222.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 243
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 494.0625,
+      "epoch": 0.4796068796068796,
+      "grad_norm": 0.6343600501349801,
+      "kl": 0.033843994140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0089,
+      "num_tokens": 5105218.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 244
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 631.71875,
+      "epoch": 0.48157248157248156,
+      "grad_norm": 0.5374583150390896,
+      "kl": 0.02874755859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0166,
+      "num_tokens": 5130717.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 245
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 533.84375,
+      "epoch": 0.48353808353808353,
+      "grad_norm": 0.833366654624395,
+      "kl": 0.0404052734375,
+      "learning_rate": 1e-06,
+      "loss": 0.0297,
+      "num_tokens": 5151728.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 246
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 718.875,
+      "epoch": 0.4855036855036855,
+      "grad_norm": 0.34812015193013685,
+      "kl": 0.0258331298828125,
+      "learning_rate": 1e-06,
+      "loss": 0.006,
+      "num_tokens": 5178038.0,
+      "reward": 0.09375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.09375,
+      "step": 247
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 537.5,
+      "epoch": 0.48746928746928747,
+      "grad_norm": 0.646907137960505,
+      "kl": 0.031463623046875,
+      "learning_rate": 1e-06,
+      "loss": 0.018,
+      "num_tokens": 5198692.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 248
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 582.90625,
+      "epoch": 0.48943488943488944,
+      "grad_norm": 0.46174894142283385,
+      "kl": 0.030975341796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0172,
+      "num_tokens": 5220241.0,
+      "reward": 0.40625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.40625,
+      "step": 249
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 639.1875,
+      "epoch": 0.4914004914004914,
+      "grad_norm": 0.45612193916391053,
+      "kl": 0.031524658203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0042,
+      "num_tokens": 5246589.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 250
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 543.0625,
+      "epoch": 0.4933660933660934,
+      "grad_norm": 0.5938060228305441,
+      "kl": 0.031219482421875,
+      "learning_rate": 1e-06,
+      "loss": -0.0037,
+      "num_tokens": 5267427.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 251
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 541.0625,
+      "epoch": 0.49533169533169535,
+      "grad_norm": 0.6182124930877956,
+      "kl": 0.032958984375,
+      "learning_rate": 1e-06,
+      "loss": 0.0064,
+      "num_tokens": 5288851.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 252
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 603.5,
+      "epoch": 0.4972972972972973,
+      "grad_norm": 0.5011035982710935,
+      "kl": 0.0300750732421875,
+      "learning_rate": 1e-06,
+      "loss": 0.0113,
+      "num_tokens": 5312549.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 253
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 571.375,
+      "epoch": 0.49926289926289924,
+      "grad_norm": 0.7964580067571678,
+      "kl": 0.0327606201171875,
+      "learning_rate": 1e-06,
+      "loss": 0.0314,
+      "num_tokens": 5334683.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 254
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 500.34375,
+      "epoch": 0.5012285012285013,
+      "grad_norm": 0.6228321163898966,
+      "kl": 0.030609130859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0119,
+      "num_tokens": 5353900.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 255
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 672.78125,
+      "epoch": 0.5031941031941032,
+      "grad_norm": 0.41977655143914255,
+      "kl": 0.0276641845703125,
+      "learning_rate": 1e-06,
+      "loss": 0.005,
+      "num_tokens": 5378597.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 256
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 603.875,
+      "epoch": 0.5051597051597052,
+      "grad_norm": 0.7823424230820542,
+      "kl": 0.0303192138671875,
+      "learning_rate": 1e-06,
+      "loss": -0.0065,
+      "num_tokens": 5402139.0,
+      "reward": 0.15625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.15625,
+      "step": 257
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 521.46875,
+      "epoch": 0.5071253071253071,
+      "grad_norm": 0.5439456208837793,
+      "kl": 0.0348052978515625,
+      "learning_rate": 1e-06,
+      "loss": 0.0027,
+      "num_tokens": 5422456.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 258
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 481.3125,
+      "epoch": 0.509090909090909,
+      "grad_norm": 0.7976544572687917,
+      "kl": 0.03790283203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0277,
+      "num_tokens": 5440868.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 259
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 482.65625,
+      "epoch": 0.5110565110565111,
+      "grad_norm": 0.6137513319405157,
+      "kl": 0.037567138671875,
+      "learning_rate": 1e-06,
+      "loss": 0.0212,
+      "num_tokens": 5459865.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 260
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 582.53125,
+      "epoch": 0.513022113022113,
+      "grad_norm": 0.561088121226348,
+      "kl": 0.03466796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0325,
+      "num_tokens": 5482278.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 261
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 610.9375,
+      "epoch": 0.514987714987715,
+      "grad_norm": 0.40737350155236235,
+      "kl": 0.0304107666015625,
+      "learning_rate": 1e-06,
+      "loss": 0.0164,
+      "num_tokens": 5505802.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 262
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 558.28125,
+      "epoch": 0.5169533169533169,
+      "grad_norm": 0.8967360376806364,
+      "kl": 0.0367889404296875,
+      "learning_rate": 1e-06,
+      "loss": 0.0129,
+      "num_tokens": 5526721.0,
+      "reward": 0.375,
+      "reward_std": 0.3535533845424652,
+      "rewards/reward_func": 0.375,
+      "step": 263
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 668.1875,
+      "epoch": 0.518918918918919,
+      "grad_norm": 0.5741886327516559,
+      "kl": 0.0272674560546875,
+      "learning_rate": 1e-06,
+      "loss": 0.0009,
+      "num_tokens": 5552191.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 264
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 592.75,
+      "epoch": 0.5208845208845209,
+      "grad_norm": 0.6032240213293185,
+      "kl": 0.031951904296875,
+      "learning_rate": 1e-06,
+      "loss": -0.005,
+      "num_tokens": 5574565.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 265
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 637.90625,
+      "epoch": 0.5228501228501229,
+      "grad_norm": 0.41254200551494014,
+      "kl": 0.03033447265625,
+      "learning_rate": 1e-06,
+      "loss": 0.0074,
+      "num_tokens": 5598552.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 266
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 567.40625,
+      "epoch": 0.5248157248157248,
+      "grad_norm": 0.4342634053395224,
+      "kl": 0.034881591796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0119,
+      "num_tokens": 5620875.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 267
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 572.46875,
+      "epoch": 0.5267813267813268,
+      "grad_norm": 0.737529118820131,
+      "kl": 0.028900146484375,
+      "learning_rate": 1e-06,
+      "loss": 0.0298,
+      "num_tokens": 5642648.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 268
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 685.625,
+      "epoch": 0.5287469287469287,
+      "grad_norm": 0.4500569807633216,
+      "kl": 0.026824951171875,
+      "learning_rate": 1e-06,
+      "loss": 0.0106,
+      "num_tokens": 5668694.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 269
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 679.28125,
+      "epoch": 0.5307125307125307,
+      "grad_norm": 0.4849462015076887,
+      "kl": 0.024566650390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0078,
+      "num_tokens": 5694389.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 270
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 528.0,
+      "epoch": 0.5326781326781327,
+      "grad_norm": 0.28404451611955495,
+      "kl": 0.0351409912109375,
+      "learning_rate": 1e-06,
+      "loss": -0.0026,
+      "num_tokens": 5715327.0,
+      "reward": 0.21875,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.21875,
+      "step": 271
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 568.78125,
+      "epoch": 0.5346437346437346,
+      "grad_norm": 0.6496110805941108,
+      "kl": 0.0332794189453125,
+      "learning_rate": 1e-06,
+      "loss": -0.007,
+      "num_tokens": 5737064.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 272
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 619.03125,
+      "epoch": 0.5366093366093366,
+      "grad_norm": 0.5926254229816691,
+      "kl": 0.034912109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0138,
+      "num_tokens": 5760589.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 273
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 536.1875,
+      "epoch": 0.5385749385749385,
+      "grad_norm": 0.8103580352644913,
+      "kl": 0.0359039306640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0085,
+      "num_tokens": 5780855.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 274
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 639.84375,
+      "epoch": 0.5405405405405406,
+      "grad_norm": 0.4708407621573234,
+      "kl": 0.0315093994140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0003,
+      "num_tokens": 5805138.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 275
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 499.40625,
+      "epoch": 0.5425061425061425,
+      "grad_norm": 0.7738022512202402,
+      "kl": 0.0649566650390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0035,
+      "num_tokens": 5824025.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 276
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 557.15625,
+      "epoch": 0.5444717444717445,
+      "grad_norm": 0.6506653997702296,
+      "kl": 0.035491943359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0106,
+      "num_tokens": 5845784.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 277
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.4375,
+      "epoch": 0.5464373464373464,
+      "grad_norm": 0.7038827199007561,
+      "kl": 0.0393524169921875,
+      "learning_rate": 1e-06,
+      "loss": -0.0202,
+      "num_tokens": 5864738.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 278
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 573.53125,
+      "epoch": 0.5484029484029485,
+      "grad_norm": 0.9617362599461011,
+      "kl": 0.0326080322265625,
+      "learning_rate": 1e-06,
+      "loss": -0.0444,
+      "num_tokens": 5887175.0,
+      "reward": 0.375,
+      "reward_std": 0.3535533845424652,
+      "rewards/reward_func": 0.375,
+      "step": 279
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 522.5625,
+      "epoch": 0.5503685503685504,
+      "grad_norm": 0.615849444934388,
+      "kl": 0.03692626953125,
+      "learning_rate": 1e-06,
+      "loss": 0.0023,
+      "num_tokens": 5907477.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 280
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 524.5,
+      "epoch": 0.5523341523341523,
+      "grad_norm": 0.3464132826493981,
+      "kl": 0.03509521484375,
+      "learning_rate": 1e-06,
+      "loss": 0.0217,
+      "num_tokens": 5927731.0,
+      "reward": 0.375,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.375,
+      "step": 281
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 461.8125,
+      "epoch": 0.5542997542997543,
+      "grad_norm": 0.6909097974436648,
+      "kl": 0.042205810546875,
+      "learning_rate": 1e-06,
+      "loss": -0.0181,
+      "num_tokens": 5945637.0,
+      "reward": 0.5625,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.5625,
+      "step": 282
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 587.28125,
+      "epoch": 0.5562653562653562,
+      "grad_norm": 0.7289737680839917,
+      "kl": 0.050811767578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0467,
+      "num_tokens": 5969752.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 283
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 495.5625,
+      "epoch": 0.5582309582309583,
+      "grad_norm": 0.5981873778704219,
+      "kl": 0.0392303466796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0157,
+      "num_tokens": 5989426.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 284
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 516.21875,
+      "epoch": 0.5601965601965602,
+      "grad_norm": 0.9106360418769466,
+      "kl": 0.033355712890625,
+      "learning_rate": 1e-06,
+      "loss": 0.0412,
+      "num_tokens": 6009763.0,
+      "reward": 0.40625,
+      "reward_std": 0.39774755761027336,
+      "rewards/reward_func": 0.40625,
+      "step": 285
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 557.3125,
+      "epoch": 0.5621621621621622,
+      "grad_norm": 0.602763648223161,
+      "kl": 0.0350799560546875,
+      "learning_rate": 1e-06,
+      "loss": 0.0376,
+      "num_tokens": 6031665.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 286
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 529.1875,
+      "epoch": 0.5641277641277641,
+      "grad_norm": 0.7103001938223117,
+      "kl": 0.0367889404296875,
+      "learning_rate": 1e-06,
+      "loss": 0.0208,
+      "num_tokens": 6052263.0,
+      "reward": 0.40625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.40625,
+      "step": 287
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 579.3125,
+      "epoch": 0.5660933660933661,
+      "grad_norm": 0.6000018822425005,
+      "kl": 0.0293121337890625,
+      "learning_rate": 1e-06,
+      "loss": -0.0064,
+      "num_tokens": 6074509.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 288
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 561.0625,
+      "epoch": 0.568058968058968,
+      "grad_norm": 0.6084807148588314,
+      "kl": 0.0366363525390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0132,
+      "num_tokens": 6096751.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 289
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 445.21875,
+      "epoch": 0.5700245700245701,
+      "grad_norm": 0.897808318980553,
+      "kl": 0.0430908203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0165,
+      "num_tokens": 6114208.0,
+      "reward": 0.25,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.25,
+      "step": 290
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 507.65625,
+      "epoch": 0.571990171990172,
+      "grad_norm": 0.7237211163121627,
+      "kl": 0.030975341796875,
+      "learning_rate": 1e-06,
+      "loss": -0.005,
+      "num_tokens": 6134967.0,
+      "reward": 0.40625,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.40625,
+      "step": 291
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 636.28125,
+      "epoch": 0.5739557739557739,
+      "grad_norm": 0.4667444055129069,
+      "kl": 0.03082275390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0189,
+      "num_tokens": 6158956.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 292
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 430.4375,
+      "epoch": 0.5759213759213759,
+      "grad_norm": 0.773152795620052,
+      "kl": 0.04296875,
+      "learning_rate": 1e-06,
+      "loss": 0.0254,
+      "num_tokens": 6176394.0,
+      "reward": 0.59375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.59375,
+      "step": 293
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 529.65625,
+      "epoch": 0.5778869778869778,
+      "grad_norm": 0.5498942657198764,
+      "kl": 0.031890869140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0031,
+      "num_tokens": 6198079.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 294
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 429.34375,
+      "epoch": 0.5798525798525799,
+      "grad_norm": 0.8165142043245753,
+      "kl": 0.0404205322265625,
+      "learning_rate": 1e-06,
+      "loss": -0.0292,
+      "num_tokens": 6214812.0,
+      "reward": 0.375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.375,
+      "step": 295
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 563.59375,
+      "epoch": 0.5818181818181818,
+      "grad_norm": 0.004534611785819621,
+      "kl": 0.036376953125,
+      "learning_rate": 1e-06,
+      "loss": 0.0,
+      "num_tokens": 6236273.0,
+      "reward": 0.5,
+      "reward_std": 0.0,
+      "rewards/reward_func": 0.5,
+      "step": 296
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 611.09375,
+      "epoch": 0.5837837837837838,
+      "grad_norm": 0.3953649614977329,
+      "kl": 0.0288238525390625,
+      "learning_rate": 1e-06,
+      "loss": -0.0081,
+      "num_tokens": 6259916.0,
+      "reward": 0.21875,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.21875,
+      "step": 297
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 502.5625,
+      "epoch": 0.5857493857493857,
+      "grad_norm": 0.8629063325607663,
+      "kl": 0.034912109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0142,
+      "num_tokens": 6279672.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 298
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 512.8125,
+      "epoch": 0.5877149877149878,
+      "grad_norm": 0.7312949259699398,
+      "kl": 0.0346527099609375,
+      "learning_rate": 1e-06,
+      "loss": 0.0231,
+      "num_tokens": 6299108.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 299
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 669.3125,
+      "epoch": 0.5896805896805897,
+      "grad_norm": 0.3844211231718793,
+      "kl": 0.0296630859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0046,
+      "num_tokens": 6324478.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 300
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 663.71875,
+      "epoch": 0.5916461916461917,
+      "grad_norm": 0.5222770323856585,
+      "kl": 0.0269317626953125,
+      "learning_rate": 1e-06,
+      "loss": -0.0194,
+      "num_tokens": 6349253.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 301
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 610.59375,
+      "epoch": 0.5936117936117936,
+      "grad_norm": 0.5352648071252188,
+      "kl": 0.0316619873046875,
+      "learning_rate": 1e-06,
+      "loss": 0.023,
+      "num_tokens": 6372508.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 302
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 523.4375,
+      "epoch": 0.5955773955773955,
+      "grad_norm": 0.4400789874053765,
+      "kl": 0.0358734130859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0053,
+      "num_tokens": 6392690.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 303
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 617.3125,
+      "epoch": 0.5975429975429976,
+      "grad_norm": 0.5569387236547813,
+      "kl": 0.038787841796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0386,
+      "num_tokens": 6416800.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 304
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 622.125,
+      "epoch": 0.5995085995085995,
+      "grad_norm": 0.3406849270712581,
+      "kl": 0.031158447265625,
+      "learning_rate": 1e-06,
+      "loss": 0.0027,
+      "num_tokens": 6442038.0,
+      "reward": 0.1875,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.1875,
+      "step": 305
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 492.75,
+      "epoch": 0.6014742014742015,
+      "grad_norm": 0.5946586449478722,
+      "kl": 0.0381927490234375,
+      "learning_rate": 1e-06,
+      "loss": 0.0293,
+      "num_tokens": 6461826.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 306
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 528.78125,
+      "epoch": 0.6034398034398034,
+      "grad_norm": 0.5089011679410208,
+      "kl": 0.0418548583984375,
+      "learning_rate": 1e-06,
+      "loss": 0.0248,
+      "num_tokens": 6482279.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 307
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 594.0,
+      "epoch": 0.6054054054054054,
+      "grad_norm": 0.6413836653957254,
+      "kl": 0.03863525390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0207,
+      "num_tokens": 6504697.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 308
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 634.28125,
+      "epoch": 0.6073710073710074,
+      "grad_norm": 0.3854408237264038,
+      "kl": 0.0313720703125,
+      "learning_rate": 1e-06,
+      "loss": 0.038,
+      "num_tokens": 6529342.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 309
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 525.65625,
+      "epoch": 0.6093366093366094,
+      "grad_norm": 0.4971690550435448,
+      "kl": 0.0362091064453125,
+      "learning_rate": 1e-06,
+      "loss": 0.0044,
+      "num_tokens": 6550177.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 310
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 533.0,
+      "epoch": 0.6113022113022113,
+      "grad_norm": 0.31552374439782604,
+      "kl": 0.0349884033203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0034,
+      "num_tokens": 6570467.0,
+      "reward": 0.5,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.5,
+      "step": 311
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 588.65625,
+      "epoch": 0.6132678132678133,
+      "grad_norm": 0.6494668114475654,
+      "kl": 0.032928466796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0078,
+      "num_tokens": 6592632.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 312
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 490.375,
+      "epoch": 0.6152334152334152,
+      "grad_norm": 0.6225762736741876,
+      "kl": 0.0364532470703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0188,
+      "num_tokens": 6612088.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 313
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 504.84375,
+      "epoch": 0.6171990171990172,
+      "grad_norm": 0.7409698772731983,
+      "kl": 0.0345001220703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0004,
+      "num_tokens": 6631959.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 314
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 520.0625,
+      "epoch": 0.6191646191646192,
+      "grad_norm": 0.9716296912093264,
+      "kl": 0.0350494384765625,
+      "learning_rate": 1e-06,
+      "loss": 0.0063,
+      "num_tokens": 6652429.0,
+      "reward": 0.28125,
+      "reward_std": 0.39774755761027336,
+      "rewards/reward_func": 0.28125,
+      "step": 315
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 463.5,
+      "epoch": 0.6211302211302211,
+      "grad_norm": 0.4764914574248498,
+      "kl": 0.0382232666015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0084,
+      "num_tokens": 6671487.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 316
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 578.15625,
+      "epoch": 0.6230958230958231,
+      "grad_norm": 0.5689759451374822,
+      "kl": 0.0312042236328125,
+      "learning_rate": 1e-06,
+      "loss": -0.0122,
+      "num_tokens": 6693304.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 317
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 450.125,
+      "epoch": 0.625061425061425,
+      "grad_norm": 0.5697978782946299,
+      "kl": 0.04327392578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0086,
+      "num_tokens": 6710562.0,
+      "reward": 0.46875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.46875,
+      "step": 318
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 506.46875,
+      "epoch": 0.6270270270270271,
+      "grad_norm": 0.5626725420705256,
+      "kl": 0.036956787109375,
+      "learning_rate": 1e-06,
+      "loss": -0.007,
+      "num_tokens": 6729951.0,
+      "reward": 0.4375,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.4375,
+      "step": 319
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 547.0625,
+      "epoch": 0.628992628992629,
+      "grad_norm": 0.5894973888067989,
+      "kl": 0.033966064453125,
+      "learning_rate": 1e-06,
+      "loss": -0.0202,
+      "num_tokens": 6750997.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 320
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 708.78125,
+      "epoch": 0.630958230958231,
+      "grad_norm": 0.37243224501970285,
+      "kl": 0.0262451171875,
+      "learning_rate": 1e-06,
+      "loss": -0.0023,
+      "num_tokens": 6777310.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 321
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 600.90625,
+      "epoch": 0.6329238329238329,
+      "grad_norm": 0.5442459669871826,
+      "kl": 0.0332794189453125,
+      "learning_rate": 1e-06,
+      "loss": 0.003,
+      "num_tokens": 6800313.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 322
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 509.4375,
+      "epoch": 0.634889434889435,
+      "grad_norm": 0.5413789628110073,
+      "kl": 0.0334320068359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0002,
+      "num_tokens": 6819943.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 323
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 499.71875,
+      "epoch": 0.6368550368550369,
+      "grad_norm": 0.3901208892455512,
+      "kl": 0.0421600341796875,
+      "learning_rate": 1e-06,
+      "loss": 0.004,
+      "num_tokens": 6839360.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 324
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 455.90625,
+      "epoch": 0.6388206388206388,
+      "grad_norm": 0.5926575612429559,
+      "kl": 0.038970947265625,
+      "learning_rate": 1e-06,
+      "loss": -0.0026,
+      "num_tokens": 6858437.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 325
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 546.625,
+      "epoch": 0.6407862407862408,
+      "grad_norm": 0.7062156547375247,
+      "kl": 0.0332183837890625,
+      "learning_rate": 1e-06,
+      "loss": 0.0253,
+      "num_tokens": 6879563.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 326
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 419.21875,
+      "epoch": 0.6427518427518427,
+      "grad_norm": 0.8379864895035511,
+      "kl": 0.046844482421875,
+      "learning_rate": 1e-06,
+      "loss": 0.011,
+      "num_tokens": 6895844.0,
+      "reward": 0.4375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.4375,
+      "step": 327
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 489.03125,
+      "epoch": 0.6447174447174447,
+      "grad_norm": 0.6425469992752795,
+      "kl": 0.03472900390625,
+      "learning_rate": 1e-06,
+      "loss": 0.0442,
+      "num_tokens": 6914839.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 328
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 490.21875,
+      "epoch": 0.6466830466830467,
+      "grad_norm": 0.5538877152424146,
+      "kl": 0.0370941162109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0017,
+      "num_tokens": 6933922.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 329
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 605.84375,
+      "epoch": 0.6486486486486487,
+      "grad_norm": 0.2911469195673005,
+      "kl": 0.02972412109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0052,
+      "num_tokens": 6956815.0,
+      "reward": 0.21875,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.21875,
+      "step": 330
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 524.96875,
+      "epoch": 0.6506142506142506,
+      "grad_norm": 0.6692638014108062,
+      "kl": 0.0333404541015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0261,
+      "num_tokens": 6976844.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 331
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 504.09375,
+      "epoch": 0.6525798525798526,
+      "grad_norm": 0.7535594730898131,
+      "kl": 0.0408782958984375,
+      "learning_rate": 1e-06,
+      "loss": -0.0078,
+      "num_tokens": 6996319.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 332
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 622.625,
+      "epoch": 0.6545454545454545,
+      "grad_norm": 0.37491504190078967,
+      "kl": 0.030029296875,
+      "learning_rate": 1e-06,
+      "loss": -0.0127,
+      "num_tokens": 7020507.0,
+      "reward": 0.34375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.34375,
+      "step": 333
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 528.96875,
+      "epoch": 0.6565110565110565,
+      "grad_norm": 0.6920450435680606,
+      "kl": 0.0441131591796875,
+      "learning_rate": 1e-06,
+      "loss": 0.0129,
+      "num_tokens": 7041354.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 334
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 449.46875,
+      "epoch": 0.6584766584766585,
+      "grad_norm": 0.6734712281138222,
+      "kl": 0.043548583984375,
+      "learning_rate": 1e-06,
+      "loss": -0.0095,
+      "num_tokens": 7060813.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 335
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 566.625,
+      "epoch": 0.6604422604422604,
+      "grad_norm": 0.7077598401138161,
+      "kl": 0.040802001953125,
+      "learning_rate": 1e-06,
+      "loss": -0.0183,
+      "num_tokens": 7081947.0,
+      "reward": 0.25,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.25,
+      "step": 336
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.09375,
+      "epoch": 0.6624078624078624,
+      "grad_norm": 0.5853457266532924,
+      "kl": 0.043914794921875,
+      "learning_rate": 1e-06,
+      "loss": -0.0052,
+      "num_tokens": 7102076.0,
+      "reward": 0.4375,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.4375,
+      "step": 337
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 567.65625,
+      "epoch": 0.6643734643734643,
+      "grad_norm": 0.5842956050999084,
+      "kl": 0.0388946533203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0072,
+      "num_tokens": 7124387.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 338
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 511.125,
+      "epoch": 0.6663390663390664,
+      "grad_norm": 0.4668262300707244,
+      "kl": 0.03558349609375,
+      "learning_rate": 1e-06,
+      "loss": -0.0015,
+      "num_tokens": 7144605.0,
+      "reward": 0.375,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.375,
+      "step": 339
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 508.59375,
+      "epoch": 0.6683046683046683,
+      "grad_norm": 0.8250382613151874,
+      "kl": 0.0343170166015625,
+      "learning_rate": 1e-06,
+      "loss": 0.0932,
+      "num_tokens": 7164278.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 340
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 489.96875,
+      "epoch": 0.6702702702702703,
+      "grad_norm": 0.7144131063171276,
+      "kl": 0.0345001220703125,
+      "learning_rate": 1e-06,
+      "loss": -0.0074,
+      "num_tokens": 7184667.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 341
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 544.21875,
+      "epoch": 0.6722358722358722,
+      "grad_norm": 0.8700749896413762,
+      "kl": 0.0315093994140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0151,
+      "num_tokens": 7205714.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 342
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 610.375,
+      "epoch": 0.6742014742014742,
+      "grad_norm": 0.45694633600819196,
+      "kl": 0.030517578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0112,
+      "num_tokens": 7229098.0,
+      "reward": 0.125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.125,
+      "step": 343
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 521.46875,
+      "epoch": 0.6761670761670762,
+      "grad_norm": 0.23257958215792865,
+      "kl": 0.0365447998046875,
+      "learning_rate": 1e-06,
+      "loss": -0.0096,
+      "num_tokens": 7249091.0,
+      "reward": 0.34375,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.34375,
+      "step": 344
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 577.875,
+      "epoch": 0.6781326781326781,
+      "grad_norm": 0.4152502823473006,
+      "kl": 0.0332489013671875,
+      "learning_rate": 1e-06,
+      "loss": -0.0014,
+      "num_tokens": 7271555.0,
+      "reward": 0.4375,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.4375,
+      "step": 345
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 446.3125,
+      "epoch": 0.6800982800982801,
+      "grad_norm": 0.6310780819784905,
+      "kl": 0.04241943359375,
+      "learning_rate": 1e-06,
+      "loss": 0.0235,
+      "num_tokens": 7288979.0,
+      "reward": 0.46875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.46875,
+      "step": 346
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 565.5625,
+      "epoch": 0.682063882063882,
+      "grad_norm": 0.6457400724427969,
+      "kl": 0.041015625,
+      "learning_rate": 1e-06,
+      "loss": 0.0055,
+      "num_tokens": 7311269.0,
+      "reward": 0.375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.375,
+      "step": 347
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 521.09375,
+      "epoch": 0.684029484029484,
+      "grad_norm": 0.5857246046657881,
+      "kl": 0.037384033203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0203,
+      "num_tokens": 7332160.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 348
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 558.28125,
+      "epoch": 0.685995085995086,
+      "grad_norm": 0.36864784062194783,
+      "kl": 0.035400390625,
+      "learning_rate": 1e-06,
+      "loss": -0.0255,
+      "num_tokens": 7354603.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 349
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 510.09375,
+      "epoch": 0.687960687960688,
+      "grad_norm": 0.37595426080670763,
+      "kl": 0.0396881103515625,
+      "learning_rate": 1e-06,
+      "loss": 0.0045,
+      "num_tokens": 7374876.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 350
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 606.25,
+      "epoch": 0.6899262899262899,
+      "grad_norm": 0.7050890671231397,
+      "kl": 0.032562255859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0187,
+      "num_tokens": 7397590.0,
+      "reward": 0.15625,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.15625,
+      "step": 351
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 639.0625,
+      "epoch": 0.6918918918918919,
+      "grad_norm": 0.623394039605989,
+      "kl": 0.029327392578125,
+      "learning_rate": 1e-06,
+      "loss": -0.0084,
+      "num_tokens": 7422452.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 352
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 514.75,
+      "epoch": 0.6938574938574938,
+      "grad_norm": 0.7054121163036189,
+      "kl": 0.03265380859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0129,
+      "num_tokens": 7442250.0,
+      "reward": 0.375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.375,
+      "step": 353
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 592.0625,
+      "epoch": 0.6958230958230959,
+      "grad_norm": 0.6548424627448317,
+      "kl": 0.0325927734375,
+      "learning_rate": 1e-06,
+      "loss": -0.0205,
+      "num_tokens": 7464522.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 354
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 526.125,
+      "epoch": 0.6977886977886978,
+      "grad_norm": 0.9159125424427512,
+      "kl": 0.0395050048828125,
+      "learning_rate": 1e-06,
+      "loss": 0.0266,
+      "num_tokens": 7486958.0,
+      "reward": 0.375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.375,
+      "step": 355
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 524.8125,
+      "epoch": 0.6997542997542997,
+      "grad_norm": 0.747990779466027,
+      "kl": 0.0347137451171875,
+      "learning_rate": 1e-06,
+      "loss": 0.037,
+      "num_tokens": 7508700.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 356
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 670.3125,
+      "epoch": 0.7017199017199017,
+      "grad_norm": 0.5139288731237621,
+      "kl": 0.0316314697265625,
+      "learning_rate": 1e-06,
+      "loss": 0.0135,
+      "num_tokens": 7533558.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 357
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 555.25,
+      "epoch": 0.7036855036855036,
+      "grad_norm": 0.7849105797708434,
+      "kl": 0.034576416015625,
+      "learning_rate": 1e-06,
+      "loss": 0.0218,
+      "num_tokens": 7554376.0,
+      "reward": 0.28125,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.28125,
+      "step": 358
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 517.125,
+      "epoch": 0.7056511056511057,
+      "grad_norm": 0.8194811532288894,
+      "kl": 0.0421905517578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0362,
+      "num_tokens": 7574152.0,
+      "reward": 0.46875,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.46875,
+      "step": 359
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 468.9375,
+      "epoch": 0.7076167076167076,
+      "grad_norm": 0.4941945178172363,
+      "kl": 0.041046142578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0048,
+      "num_tokens": 7592906.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 360
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 498.9375,
+      "epoch": 0.7095823095823096,
+      "grad_norm": 0.8121886110174811,
+      "kl": 0.04071044921875,
+      "learning_rate": 1e-06,
+      "loss": 0.0818,
+      "num_tokens": 7612886.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 361
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 601.15625,
+      "epoch": 0.7115479115479115,
+      "grad_norm": 0.45512137045157774,
+      "kl": 0.03326416015625,
+      "learning_rate": 1e-06,
+      "loss": -0.0033,
+      "num_tokens": 7635799.0,
+      "reward": 0.4375,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.4375,
+      "step": 362
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 545.40625,
+      "epoch": 0.7135135135135136,
+      "grad_norm": 0.830092454914979,
+      "kl": 0.0363311767578125,
+      "learning_rate": 1e-06,
+      "loss": -0.0188,
+      "num_tokens": 7656276.0,
+      "reward": 0.40625,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.40625,
+      "step": 363
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 451.9375,
+      "epoch": 0.7154791154791155,
+      "grad_norm": 0.3436418620406507,
+      "kl": 0.04986572265625,
+      "learning_rate": 1e-06,
+      "loss": -0.0065,
+      "num_tokens": 7674878.0,
+      "reward": 0.3125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.3125,
+      "step": 364
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 473.40625,
+      "epoch": 0.7174447174447175,
+      "grad_norm": 0.6642678609768281,
+      "kl": 0.0440521240234375,
+      "learning_rate": 1e-06,
+      "loss": 0.0141,
+      "num_tokens": 7694745.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 365
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.71875,
+      "epoch": 0.7194103194103194,
+      "grad_norm": 0.7446714368531804,
+      "kl": 0.046142578125,
+      "learning_rate": 1e-06,
+      "loss": -0.0123,
+      "num_tokens": 7713650.0,
+      "reward": 0.4375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.4375,
+      "step": 366
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 484.375,
+      "epoch": 0.7213759213759213,
+      "grad_norm": 0.8698002010484014,
+      "kl": 0.042236328125,
+      "learning_rate": 1e-06,
+      "loss": 0.0213,
+      "num_tokens": 7732482.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 367
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 461.875,
+      "epoch": 0.7233415233415234,
+      "grad_norm": 0.9898476778510431,
+      "kl": 0.03765869140625,
+      "learning_rate": 1e-06,
+      "loss": -0.0047,
+      "num_tokens": 7750484.0,
+      "reward": 0.46875,
+      "reward_std": 0.39774755761027336,
+      "rewards/reward_func": 0.46875,
+      "step": 368
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 457.6875,
+      "epoch": 0.7253071253071253,
+      "grad_norm": 0.8069454227623065,
+      "kl": 0.049163818359375,
+      "learning_rate": 1e-06,
+      "loss": -0.0093,
+      "num_tokens": 7767968.0,
+      "reward": 0.4375,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.4375,
+      "step": 369
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 439.3125,
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.8746030989783482,
+      "kl": 0.043609619140625,
+      "learning_rate": 1e-06,
+      "loss": 0.0183,
+      "num_tokens": 7786330.0,
+      "reward": 0.5625,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.5625,
+      "step": 370
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 446.4375,
+      "epoch": 0.7292383292383292,
+      "grad_norm": 0.4845476792447974,
+      "kl": 0.1286468505859375,
+      "learning_rate": 1e-06,
+      "loss": -0.0175,
+      "num_tokens": 7804720.0,
+      "reward": 0.21875,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.21875,
+      "step": 371
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 434.53125,
+      "epoch": 0.7312039312039312,
+      "grad_norm": 0.7019254770499221,
+      "kl": 0.0400238037109375,
+      "learning_rate": 1e-06,
+      "loss": -0.0082,
+      "num_tokens": 7821607.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 372
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 624.34375,
+      "epoch": 0.7331695331695331,
+      "grad_norm": 0.6696488313790595,
+      "kl": 0.03546142578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0218,
+      "num_tokens": 7844870.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 373
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 561.6875,
+      "epoch": 0.7351351351351352,
+      "grad_norm": 0.8600206462556853,
+      "kl": 0.0367431640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0546,
+      "num_tokens": 7866172.0,
+      "reward": 0.53125,
+      "reward_std": 0.30935921147465706,
+      "rewards/reward_func": 0.53125,
+      "step": 374
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 497.125,
+      "epoch": 0.7371007371007371,
+      "grad_norm": 0.7638586068500758,
+      "kl": 0.0342559814453125,
+      "learning_rate": 1e-06,
+      "loss": -0.0107,
+      "num_tokens": 7885490.0,
+      "reward": 0.46875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.46875,
+      "step": 375
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 566.53125,
+      "epoch": 0.7390663390663391,
+      "grad_norm": 0.4829267801913666,
+      "kl": 0.0334930419921875,
+      "learning_rate": 1e-06,
+      "loss": -0.0167,
+      "num_tokens": 7908661.0,
+      "reward": 0.28125,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.28125,
+      "step": 376
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 480.40625,
+      "epoch": 0.741031941031941,
+      "grad_norm": 0.740899432881196,
+      "kl": 0.0396575927734375,
+      "learning_rate": 1e-06,
+      "loss": -0.0194,
+      "num_tokens": 7927568.0,
+      "reward": 0.46875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.46875,
+      "step": 377
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 568.09375,
+      "epoch": 0.742997542997543,
+      "grad_norm": 0.5106656848937849,
+      "kl": 0.036956787109375,
+      "learning_rate": 1e-06,
+      "loss": 0.0017,
+      "num_tokens": 7950839.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 378
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 522.8125,
+      "epoch": 0.744963144963145,
+      "grad_norm": 0.6659830768357405,
+      "kl": 0.03570556640625,
+      "learning_rate": 1e-06,
+      "loss": 0.0149,
+      "num_tokens": 7971197.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 379
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 512.9375,
+      "epoch": 0.7469287469287469,
+      "grad_norm": 0.7524006374031159,
+      "kl": 0.0421142578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0224,
+      "num_tokens": 7991209.0,
+      "reward": 0.25,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.25,
+      "step": 380
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 545.75,
+      "epoch": 0.7488943488943489,
+      "grad_norm": 0.4839548397489806,
+      "kl": 0.0340118408203125,
+      "learning_rate": 1e-06,
+      "loss": -0.014,
+      "num_tokens": 8012103.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 381
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 532.15625,
+      "epoch": 0.7508599508599508,
+      "grad_norm": 0.7060415982161667,
+      "kl": 0.0348968505859375,
+      "learning_rate": 1e-06,
+      "loss": -0.003,
+      "num_tokens": 8032922.0,
+      "reward": 0.40625,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.40625,
+      "step": 382
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 593.0625,
+      "epoch": 0.7528255528255529,
+      "grad_norm": 0.8703459584049694,
+      "kl": 0.0374755859375,
+      "learning_rate": 1e-06,
+      "loss": 0.0024,
+      "num_tokens": 8055494.0,
+      "reward": 0.3125,
+      "reward_std": 0.2651650384068489,
+      "rewards/reward_func": 0.3125,
+      "step": 383
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 605.28125,
+      "epoch": 0.7547911547911548,
+      "grad_norm": 0.531040583745666,
+      "kl": 0.0342864990234375,
+      "learning_rate": 1e-06,
+      "loss": 0.004,
+      "num_tokens": 8078407.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 384
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 472.71875,
+      "epoch": 0.7567567567567568,
+      "grad_norm": 0.6413661425786513,
+      "kl": 0.0486602783203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0229,
+      "num_tokens": 8096908.0,
+      "reward": 0.5,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.5,
+      "step": 385
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 540.40625,
+      "epoch": 0.7587223587223587,
+      "grad_norm": 0.5887264630204668,
+      "kl": 0.0734100341796875,
+      "learning_rate": 1e-06,
+      "loss": 0.015,
+      "num_tokens": 8118141.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 386
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 552.3125,
+      "epoch": 0.7606879606879607,
+      "grad_norm": 0.35879317767107755,
+      "kl": 0.0392608642578125,
+      "learning_rate": 1e-06,
+      "loss": 0.0237,
+      "num_tokens": 8139923.0,
+      "reward": 0.15625,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.15625,
+      "step": 387
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 426.0,
+      "epoch": 0.7626535626535627,
+      "grad_norm": 0.6612874102908352,
+      "kl": 0.0440216064453125,
+      "learning_rate": 1e-06,
+      "loss": -0.0187,
+      "num_tokens": 8157183.0,
+      "reward": 0.3125,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.3125,
+      "step": 388
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 431.96875,
+      "epoch": 0.7646191646191646,
+      "grad_norm": 0.7214155280951005,
+      "kl": 0.0413360595703125,
+      "learning_rate": 1e-06,
+      "loss": 0.0028,
+      "num_tokens": 8174882.0,
+      "reward": 0.28125,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.28125,
+      "step": 389
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 464.0625,
+      "epoch": 0.7665847665847666,
+      "grad_norm": 0.6542555963916522,
+      "kl": 0.0447845458984375,
+      "learning_rate": 1e-06,
+      "loss": -0.0012,
+      "num_tokens": 8192790.0,
+      "reward": 0.4375,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.4375,
+      "step": 390
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 508.65625,
+      "epoch": 0.7685503685503685,
+      "grad_norm": 0.8263885848953906,
+      "kl": 0.03887939453125,
+      "learning_rate": 1e-06,
+      "loss": 0.0084,
+      "num_tokens": 8212111.0,
+      "reward": 0.46875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.46875,
+      "step": 391
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 615.46875,
+      "epoch": 0.7705159705159705,
+      "grad_norm": 0.5049205742260084,
+      "kl": 0.0379638671875,
+      "learning_rate": 1e-06,
+      "loss": -0.0336,
+      "num_tokens": 8236940.0,
+      "reward": 0.21875,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.21875,
+      "step": 392
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 600.15625,
+      "epoch": 0.7724815724815725,
+      "grad_norm": 0.6038728188859286,
+      "kl": 0.047821044921875,
+      "learning_rate": 1e-06,
+      "loss": 0.0255,
+      "num_tokens": 8260543.0,
+      "reward": 0.1875,
+      "reward_std": 0.1767766922712326,
+      "rewards/reward_func": 0.1875,
+      "step": 393
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 490.40625,
+      "epoch": 0.7744471744471745,
+      "grad_norm": 0.48639402941702775,
+      "kl": 0.0445709228515625,
+      "learning_rate": 1e-06,
+      "loss": 0.0125,
+      "num_tokens": 8279292.0,
+      "reward": 0.40625,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.40625,
+      "step": 394
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 526.28125,
+      "epoch": 0.7764127764127764,
+      "grad_norm": 0.37376433880625853,
+      "kl": 0.04119873046875,
+      "learning_rate": 1e-06,
+      "loss": -0.0067,
+      "num_tokens": 8300045.0,
+      "reward": 0.125,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.125,
+      "step": 395
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 587.78125,
+      "epoch": 0.7783783783783784,
+      "grad_norm": 0.5600748468776157,
+      "kl": 0.034210205078125,
+      "learning_rate": 1e-06,
+      "loss": -0.0145,
+      "num_tokens": 8323160.0,
+      "reward": 0.34375,
+      "reward_std": 0.13258251920342445,
+      "rewards/reward_func": 0.34375,
+      "step": 396
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 546.1875,
+      "epoch": 0.7803439803439803,
+      "grad_norm": 0.6575791496386095,
+      "kl": 0.044769287109375,
+      "learning_rate": 1e-06,
+      "loss": -0.0228,
+      "num_tokens": 8345354.0,
+      "reward": 0.21875,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.21875,
+      "step": 397
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 550.65625,
+      "epoch": 0.7823095823095824,
+      "grad_norm": 0.7332146409060798,
+      "kl": 0.0418243408203125,
+      "learning_rate": 1e-06,
+      "loss": 0.0071,
+      "num_tokens": 8367361.0,
+      "reward": 0.34375,
+      "reward_std": 0.22097086533904076,
+      "rewards/reward_func": 0.34375,
+      "step": 398
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 578.8125,
+      "epoch": 0.7842751842751843,
+      "grad_norm": 0.35869977857179974,
+      "kl": 0.03509521484375,
+      "learning_rate": 1e-06,
+      "loss": 0.0202,
+      "num_tokens": 8390055.0,
+      "reward": 0.25,
+      "reward_std": 0.0883883461356163,
+      "rewards/reward_func": 0.25,
+      "step": 399
+    },
+    {
+      "clip_ratio": 0.0,
+      "completion_length": 510.09375,
+      "epoch": 0.7862407862407862,
+      "grad_norm": 0.4437939683355927,
+      "kl": 0.0397796630859375,
+      "learning_rate": 1e-06,
+      "loss": 0.002,
+      "num_tokens": 8411168.0,
+      "reward": 0.28125,
+      "reward_std": 0.04419417306780815,
+      "rewards/reward_func": 0.28125,
+      "step": 400
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 508,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}