{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.14, "eval_steps": 50, "global_step": 350, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0004, "frac_reward_zero_std": 0.75, "grad_norm": 31.350610733032227, "kl": 0.0, "learning_rate": 0.0, "loss": -0.0, "num_tokens": 8880.0, "reward": 0.8125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.8125, "rewards/compute_reward/std": 0.40311288833618164, "step": 1 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0008, "frac_reward_zero_std": 0.75, "grad_norm": 40.88920593261719, "kl": 0.0, "learning_rate": 4e-08, "loss": 0.0, "num_tokens": 17760.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 2 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0012, "frac_reward_zero_std": 0.75, "grad_norm": 41.24170684814453, "kl": 0.0013443923089653254, "learning_rate": 8e-08, "loss": 0.0, "num_tokens": 26640.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 3 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0016, "frac_reward_zero_std": 1.0, "grad_norm": 0.025073383003473282, "kl": 0.0012363345449557528, "learning_rate": 1.2e-07, "loss": 0.0, "num_tokens": 35520.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 4 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.002, "frac_reward_zero_std": 0.5, "grad_norm": 45.65277862548828, "kl": 0.0009983214986277744, "learning_rate": 1.6e-07, "loss": 0.0, "num_tokens": 44400.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 5 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0024, "frac_reward_zero_std": 0.5, "grad_norm": 40.60847854614258, "kl": 0.0010500029457034543, "learning_rate": 2e-07, "loss": 0.0, "num_tokens": 53280.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 6 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0028, "frac_reward_zero_std": 0.5, "grad_norm": 37.455440521240234, "kl": 0.0013627598236780614, "learning_rate": 2.4e-07, "loss": 0.0, "num_tokens": 62160.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 7 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0032, "frac_reward_zero_std": 0.5, "grad_norm": 49.363807678222656, "kl": 0.0019289999618194997, "learning_rate": 2.8e-07, "loss": 0.0, "num_tokens": 71040.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 8 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0036, "frac_reward_zero_std": 1.0, "grad_norm": 0.055168308317661285, "kl": 0.0023864044342190027, "learning_rate": 3.2e-07, "loss": 0.0, "num_tokens": 79920.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 9 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.004, "frac_reward_zero_std": 0.75, "grad_norm": 27.4397029876709, "kl": 0.0029534806380979717, "learning_rate": 3.6e-07, "loss": 0.0, "num_tokens": 88800.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 10 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0044, "frac_reward_zero_std": 0.5, "grad_norm": 34.00825881958008, "kl": 0.004601971711963415, "learning_rate": 4e-07, "loss": 0.0, "num_tokens": 97680.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 11 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0048, "frac_reward_zero_std": 0.5, "grad_norm": 41.115177154541016, "kl": 0.005590100074186921, "learning_rate": 4.3999999999999997e-07, "loss": 0.0, "num_tokens": 106560.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 12 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0052, "frac_reward_zero_std": 0.5, "grad_norm": 37.690364837646484, "kl": 0.008083905442617834, "learning_rate": 4.8e-07, "loss": 0.0, "num_tokens": 115440.0, "reward": 0.25, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 13 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0056, "frac_reward_zero_std": 0.5, "grad_norm": 44.36626052856445, "kl": 0.01064743846654892, "learning_rate": 5.2e-07, "loss": 0.0, "num_tokens": 124320.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 14 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.006, "frac_reward_zero_std": 0.25, "grad_norm": 35.651023864746094, "kl": 0.01316027413122356, "learning_rate": 5.6e-07, "loss": 0.0001, "num_tokens": 133200.0, "reward": 0.75, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 15 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0064, "frac_reward_zero_std": 0.25, "grad_norm": 52.94963836669922, "kl": 0.020378376357257366, "learning_rate": 6e-07, "loss": 0.0001, "num_tokens": 142080.0, "reward": 0.4375, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 16 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0068, "frac_reward_zero_std": 0.75, "grad_norm": 36.93665313720703, "kl": 0.025172261521220207, "learning_rate": 6.4e-07, "loss": 0.0001, "num_tokens": 150960.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 17 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0072, "frac_reward_zero_std": 1.0, "grad_norm": 0.5544130206108093, "kl": 0.032024398911744356, "learning_rate": 6.800000000000001e-07, "loss": 0.0001, "num_tokens": 159840.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 18 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0076, "frac_reward_zero_std": 0.25, "grad_norm": 44.66011047363281, "kl": 0.039961141999810934, "learning_rate": 7.2e-07, "loss": 0.0002, "num_tokens": 168720.0, "reward": 0.25, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 19 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.008, "frac_reward_zero_std": 0.75, "grad_norm": 28.504152297973633, "kl": 0.036296515725553036, "learning_rate": 7.599999999999999e-07, "loss": 0.0001, "num_tokens": 177600.0, "reward": 0.9375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.9375, "rewards/compute_reward/std": 0.25, "step": 20 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0084, "frac_reward_zero_std": 0.25, "grad_norm": 50.410888671875, "kl": 0.04341549798846245, "learning_rate": 8e-07, "loss": 0.0002, "num_tokens": 186480.0, "reward": 0.375, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 21 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0088, "frac_reward_zero_std": 0.5, "grad_norm": 51.3606071472168, "kl": 0.050860715098679066, "learning_rate": 8.399999999999999e-07, "loss": 0.0002, "num_tokens": 195360.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 22 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0092, "frac_reward_zero_std": 0.75, "grad_norm": 24.91278076171875, "kl": 0.06213559303432703, "learning_rate": 8.799999999999999e-07, "loss": 0.0002, "num_tokens": 204240.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 23 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0096, "frac_reward_zero_std": 0.5, "grad_norm": 39.62419891357422, "kl": 0.0647846395149827, "learning_rate": 9.2e-07, "loss": 0.0003, "num_tokens": 213120.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 24 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.01, "frac_reward_zero_std": 0.25, "grad_norm": 63.561767578125, "kl": 0.0892756637185812, "learning_rate": 9.6e-07, "loss": 0.0004, "num_tokens": 222000.0, "reward": 0.5625, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 25 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0104, "frac_reward_zero_std": 0.5, "grad_norm": 42.179954528808594, "kl": 0.10400468856096268, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 230880.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 26 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0108, "frac_reward_zero_std": 0.75, "grad_norm": 25.707307815551758, "kl": 0.1136582437902689, "learning_rate": 1.04e-06, "loss": 0.0005, "num_tokens": 239760.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 27 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0112, "frac_reward_zero_std": 0.75, "grad_norm": 26.41197967529297, "kl": 0.13468273170292377, "learning_rate": 1.08e-06, "loss": 0.0005, "num_tokens": 248640.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 28 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0116, "frac_reward_zero_std": 0.75, "grad_norm": 35.851348876953125, "kl": 0.1477714516222477, "learning_rate": 1.12e-06, "loss": 0.0006, "num_tokens": 257520.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 29 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.012, "frac_reward_zero_std": 0.75, "grad_norm": 29.354488372802734, "kl": 0.14891785010695457, "learning_rate": 1.16e-06, "loss": 0.0006, "num_tokens": 266400.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 30 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0124, "frac_reward_zero_std": 0.5, "grad_norm": 39.174476623535156, "kl": 0.15899883769452572, "learning_rate": 1.2e-06, "loss": 0.0006, "num_tokens": 275280.0, "reward": 0.25, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 31 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0128, "frac_reward_zero_std": 0.75, "grad_norm": 44.24154281616211, "kl": 0.22540221363306046, "learning_rate": 1.24e-06, "loss": 0.0009, "num_tokens": 284160.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 32 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0132, "frac_reward_zero_std": 0.5, "grad_norm": 32.92380142211914, "kl": 0.19463030248880386, "learning_rate": 1.28e-06, "loss": 0.0008, "num_tokens": 293040.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 33 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0136, "frac_reward_zero_std": 0.75, "grad_norm": 32.65846633911133, "kl": 0.2380228154361248, "learning_rate": 1.32e-06, "loss": 0.001, "num_tokens": 301920.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 34 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.014, "frac_reward_zero_std": 0.5, "grad_norm": 77.5787582397461, "kl": 0.21238164231181145, "learning_rate": 1.3600000000000001e-06, "loss": 0.0008, "num_tokens": 310800.0, "reward": 0.125, "reward_std": 0.25, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 35 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0144, "frac_reward_zero_std": 0.75, "grad_norm": 29.662382125854492, "kl": 0.29322075098752975, "learning_rate": 1.4e-06, "loss": 0.0012, "num_tokens": 319680.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 36 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0148, "frac_reward_zero_std": 0.5, "grad_norm": 48.11992263793945, "kl": 0.25715991109609604, "learning_rate": 1.44e-06, "loss": 0.001, "num_tokens": 328560.0, "reward": 0.125, "reward_std": 0.25, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 37 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0152, "frac_reward_zero_std": 0.75, "grad_norm": 38.42544937133789, "kl": 0.2538900189101696, "learning_rate": 1.48e-06, "loss": 0.001, "num_tokens": 337440.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 38 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0156, "frac_reward_zero_std": 0.0, "grad_norm": 76.46257019042969, "kl": 0.29158805310726166, "learning_rate": 1.5199999999999998e-06, "loss": 0.0012, "num_tokens": 346320.0, "reward": 0.4375, "reward_std": 0.5580127239227295, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 39 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.016, "frac_reward_zero_std": 0.5, "grad_norm": 50.54710388183594, "kl": 0.42541511356830597, "learning_rate": 1.5599999999999999e-06, "loss": 0.0017, "num_tokens": 355200.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 40 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0164, "frac_reward_zero_std": 0.5, "grad_norm": 47.023494720458984, "kl": 0.3670000210404396, "learning_rate": 1.6e-06, "loss": 0.0015, "num_tokens": 364080.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 41 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0168, "frac_reward_zero_std": 0.75, "grad_norm": 32.30452346801758, "kl": 0.38483476638793945, "learning_rate": 1.6399999999999998e-06, "loss": 0.0015, "num_tokens": 372960.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 42 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0172, "frac_reward_zero_std": 0.5, "grad_norm": 42.685611724853516, "kl": 0.3910580351948738, "learning_rate": 1.6799999999999998e-06, "loss": 0.0016, "num_tokens": 381840.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 43 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0176, "frac_reward_zero_std": 0.75, "grad_norm": 39.6999626159668, "kl": 0.4346514791250229, "learning_rate": 1.7199999999999998e-06, "loss": 0.0017, "num_tokens": 390720.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 44 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.018, "frac_reward_zero_std": 0.75, "grad_norm": 32.37580490112305, "kl": 0.38541124016046524, "learning_rate": 1.7599999999999999e-06, "loss": 0.0015, "num_tokens": 399600.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 45 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0184, "frac_reward_zero_std": 0.5, "grad_norm": 37.88615798950195, "kl": 0.3760378360748291, "learning_rate": 1.8e-06, "loss": 0.0015, "num_tokens": 408480.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 46 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0188, "frac_reward_zero_std": 0.5, "grad_norm": 50.25699996948242, "kl": 0.4850776717066765, "learning_rate": 1.84e-06, "loss": 0.0019, "num_tokens": 417360.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 47 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0192, "frac_reward_zero_std": 1.0, "grad_norm": 0.5161222815513611, "kl": 0.3658723160624504, "learning_rate": 1.8799999999999998e-06, "loss": 0.0015, "num_tokens": 426240.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 48 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0196, "frac_reward_zero_std": 0.5, "grad_norm": 37.257171630859375, "kl": 0.3248850479722023, "learning_rate": 1.92e-06, "loss": 0.0013, "num_tokens": 435120.0, "reward": 0.875, "reward_std": 0.25, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3415650427341461, "step": 49 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.02, "frac_reward_zero_std": 0.5, "grad_norm": 40.35070037841797, "kl": 0.3114013075828552, "learning_rate": 1.96e-06, "loss": 0.0012, "num_tokens": 444000.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 50 }, { "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.02, "eval_loss": 0.0003695653285831213, "eval_runtime": 12968.8828, "eval_samples_per_second": 0.015, "eval_steps_per_second": 0.004, "frac_reward_zero_std": 0.54, "num_tokens": 555000.0, "reward": 0.445, "reward_std": 0.23928203105926513, "rewards/compute_reward/mean": 0.445, "rewards/compute_reward/std": 0.23928203105926513, "step": 50 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0204, "frac_reward_zero_std": 0.75, "grad_norm": 35.21977233886719, "kl": 0.40479009598493576, "learning_rate": 2e-06, "loss": -0.1234, "num_tokens": 563880.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 51 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0208, "frac_reward_zero_std": 0.75, "grad_norm": 17.721899032592773, "kl": 0.3238118216395378, "learning_rate": 1.9999756307053944e-06, "loss": 0.1263, "num_tokens": 572760.0, "reward": 0.875, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3415650427341461, "step": 52 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0212, "frac_reward_zero_std": 0.5, "grad_norm": 48.17813491821289, "kl": 0.4712698385119438, "learning_rate": 1.999902524009304e-06, "loss": -0.1231, "num_tokens": 581640.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 53 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0216, "frac_reward_zero_std": 1.0, "grad_norm": 21.3551025390625, "kl": 0.4696747958660126, "learning_rate": 1.999780683474845e-06, "loss": 0.1269, "num_tokens": 590520.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 54 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.022, "frac_reward_zero_std": 0.75, "grad_norm": 42.38340377807617, "kl": 0.5665004104375839, "learning_rate": 1.999610115040354e-06, "loss": 0.1272, "num_tokens": 599400.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 55 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0224, "frac_reward_zero_std": 0.5, "grad_norm": 40.7006721496582, "kl": 0.4810464084148407, "learning_rate": 1.9993908270190957e-06, "loss": -0.248, "num_tokens": 608280.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 56 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0228, "frac_reward_zero_std": 0.25, "grad_norm": 31.320539474487305, "kl": 0.44268205761909485, "learning_rate": 1.999122830098858e-06, "loss": 0.7516, "num_tokens": 617160.0, "reward": 0.3125, "reward_std": 0.375, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 57 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0232, "frac_reward_zero_std": 0.5, "grad_norm": 46.34135818481445, "kl": 0.4858321249485016, "learning_rate": 1.998806137341434e-06, "loss": -0.8729, "num_tokens": 626040.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 58 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0236, "frac_reward_zero_std": 0.5, "grad_norm": 50.77784729003906, "kl": 0.4124316945672035, "learning_rate": 1.998440764181981e-06, "loss": 0.4681, "num_tokens": 634920.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 59 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.024, "frac_reward_zero_std": 0.25, "grad_norm": 62.64232635498047, "kl": 0.4631865471601486, "learning_rate": 1.9980267284282714e-06, "loss": -0.0896, "num_tokens": 643800.0, "reward": 0.1875, "reward_std": 0.375, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 60 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0244, "frac_reward_zero_std": 0.5, "grad_norm": 55.10979080200195, "kl": 0.3911132216453552, "learning_rate": 1.997564050259824e-06, "loss": 0.1265, "num_tokens": 652680.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 61 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0248, "frac_reward_zero_std": 0.25, "grad_norm": 75.01067352294922, "kl": 0.4750595986843109, "learning_rate": 1.99705275222692e-06, "loss": -0.8729, "num_tokens": 661560.0, "reward": 0.5625, "reward_std": 0.375, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 62 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0252, "frac_reward_zero_std": 0.0, "grad_norm": 69.7686996459961, "kl": 0.4622643366456032, "learning_rate": 1.9964928592495045e-06, "loss": -0.3061, "num_tokens": 670440.0, "reward": 0.4375, "reward_std": 0.51933753490448, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 63 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0256, "frac_reward_zero_std": 0.0, "grad_norm": 78.1102294921875, "kl": 0.4316819906234741, "learning_rate": 1.99588439861597e-06, "loss": 0.8431, "num_tokens": 679320.0, "reward": 0.5625, "reward_std": 0.51933753490448, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 64 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.026, "frac_reward_zero_std": 0.25, "grad_norm": 70.63719940185547, "kl": 0.4158880114555359, "learning_rate": 1.995227399981831e-06, "loss": 0.0597, "num_tokens": 688200.0, "reward": 0.5625, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 65 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0264, "frac_reward_zero_std": 0.25, "grad_norm": 56.348411560058594, "kl": 0.30301499366760254, "learning_rate": 1.994521895368273e-06, "loss": -0.2153, "num_tokens": 697080.0, "reward": 0.5625, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 66 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0268, "frac_reward_zero_std": 0.5, "grad_norm": 25.096176147460938, "kl": 0.27180512621998787, "learning_rate": 1.9937679191605962e-06, "loss": 0.7175, "num_tokens": 705960.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 67 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0272, "frac_reward_zero_std": 0.25, "grad_norm": 43.11947250366211, "kl": 0.27942386642098427, "learning_rate": 1.992965508106537e-06, "loss": -0.3403, "num_tokens": 714840.0, "reward": 0.6875, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 68 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0276, "frac_reward_zero_std": 0.5, "grad_norm": 38.26469802856445, "kl": 0.24121180921792984, "learning_rate": 1.9921147013144777e-06, "loss": -0.124, "num_tokens": 723720.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 69 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.028, "frac_reward_zero_std": 1.0, "grad_norm": 16.814804077148438, "kl": 0.2414722703397274, "learning_rate": 1.9912155402515414e-06, "loss": 0.001, "num_tokens": 732600.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 70 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0284, "frac_reward_zero_std": 0.25, "grad_norm": 58.598567962646484, "kl": 0.4162477105855942, "learning_rate": 1.99026806874157e-06, "loss": -0.2483, "num_tokens": 741480.0, "reward": 0.4375, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 71 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0288, "frac_reward_zero_std": 0.5, "grad_norm": 56.5173454284668, "kl": 0.30467043071985245, "learning_rate": 1.9892723329629885e-06, "loss": 0.2512, "num_tokens": 750360.0, "reward": 0.25, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 72 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0292, "frac_reward_zero_std": 0.5, "grad_norm": 88.95624542236328, "kl": 0.35381778329610825, "learning_rate": 1.9882283814465526e-06, "loss": 0.5593, "num_tokens": 759240.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 73 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0296, "frac_reward_zero_std": 0.5, "grad_norm": 90.53428649902344, "kl": 0.3509898856282234, "learning_rate": 1.987136265072988e-06, "loss": -0.6815, "num_tokens": 768120.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 74 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.03, "frac_reward_zero_std": 0.5, "grad_norm": 40.467559814453125, "kl": 0.27240653708577156, "learning_rate": 1.985996037070505e-06, "loss": 0.501, "num_tokens": 777000.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 75 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0304, "frac_reward_zero_std": 0.75, "grad_norm": 42.016510009765625, "kl": 0.3176127299666405, "learning_rate": 1.984807753012208e-06, "loss": -0.3737, "num_tokens": 785880.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 76 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0308, "frac_reward_zero_std": 0.5, "grad_norm": 42.70841598510742, "kl": 0.40913257747888565, "learning_rate": 1.983571470813386e-06, "loss": 0.2181, "num_tokens": 794760.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 77 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0312, "frac_reward_zero_std": 0.5, "grad_norm": 48.8704833984375, "kl": 0.30612240731716156, "learning_rate": 1.9822872507286887e-06, "loss": -0.4652, "num_tokens": 803640.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 78 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0316, "frac_reward_zero_std": 1.0, "grad_norm": 10.744608879089355, "kl": 0.3573462963104248, "learning_rate": 1.9809551553491913e-06, "loss": 0.2514, "num_tokens": 812520.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 79 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.032, "frac_reward_zero_std": 0.25, "grad_norm": 44.66023254394531, "kl": 0.2872606888413429, "learning_rate": 1.979575249599344e-06, "loss": 0.1596, "num_tokens": 821400.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 80 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0324, "frac_reward_zero_std": 0.5, "grad_norm": 48.97246170043945, "kl": 0.41179851442575455, "learning_rate": 1.9781476007338054e-06, "loss": -0.2818, "num_tokens": 830280.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 81 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0328, "frac_reward_zero_std": 0.75, "grad_norm": 47.648284912109375, "kl": 0.361647367477417, "learning_rate": 1.9766722783341677e-06, "loss": -0.09, "num_tokens": 839160.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 82 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0332, "frac_reward_zero_std": 0.5, "grad_norm": 68.1531982421875, "kl": 0.35677963495254517, "learning_rate": 1.975149354305563e-06, "loss": 0.0929, "num_tokens": 848040.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 83 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0336, "frac_reward_zero_std": 1.0, "grad_norm": 25.728612899780273, "kl": 0.299656730145216, "learning_rate": 1.97357890287316e-06, "loss": 0.1262, "num_tokens": 856920.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 84 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.034, "frac_reward_zero_std": 0.25, "grad_norm": 43.74058151245117, "kl": 0.3780274763703346, "learning_rate": 1.9719610005785463e-06, "loss": -0.215, "num_tokens": 865800.0, "reward": 0.375, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 85 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0344, "frac_reward_zero_std": 1.0, "grad_norm": 22.789318084716797, "kl": 0.361059308052063, "learning_rate": 1.9702957262759963e-06, "loss": 0.2179, "num_tokens": 874680.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 86 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0348, "frac_reward_zero_std": 0.75, "grad_norm": 35.32533264160156, "kl": 0.44058068096637726, "learning_rate": 1.968583161128631e-06, "loss": 0.0018, "num_tokens": 883560.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 87 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0352, "frac_reward_zero_std": 0.5, "grad_norm": 48.396873474121094, "kl": 0.40034161880612373, "learning_rate": 1.9668233886044593e-06, "loss": 0.3765, "num_tokens": 892440.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 88 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0356, "frac_reward_zero_std": 0.5, "grad_norm": 51.92245101928711, "kl": 0.582250103354454, "learning_rate": 1.9650164944723112e-06, "loss": -0.2811, "num_tokens": 901320.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 89 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.036, "frac_reward_zero_std": 0.75, "grad_norm": 39.04520797729492, "kl": 0.46102364361286163, "learning_rate": 1.963162566797658e-06, "loss": 0.1603, "num_tokens": 910200.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 90 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0364, "frac_reward_zero_std": 0.25, "grad_norm": 64.48860168457031, "kl": 0.7562894448637962, "learning_rate": 1.9612616959383188e-06, "loss": 0.2195, "num_tokens": 919080.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 91 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0368, "frac_reward_zero_std": 0.5, "grad_norm": 71.13645935058594, "kl": 0.584452673792839, "learning_rate": 1.9593139745400573e-06, "loss": -0.2141, "num_tokens": 927960.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 92 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0372, "frac_reward_zero_std": 0.25, "grad_norm": 44.049346923828125, "kl": 0.5378477424383163, "learning_rate": 1.957319497532067e-06, "loss": -0.2143, "num_tokens": 936840.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 93 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0376, "frac_reward_zero_std": 0.25, "grad_norm": 75.5696792602539, "kl": 0.5544816628098488, "learning_rate": 1.9552783621223435e-06, "loss": 0.5601, "num_tokens": 945720.0, "reward": 0.75, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 94 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.038, "frac_reward_zero_std": 1.0, "grad_norm": 17.455116271972656, "kl": 0.3371001221239567, "learning_rate": 1.953190667792947e-06, "loss": -0.59, "num_tokens": 954600.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 95 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0384, "frac_reward_zero_std": 0.25, "grad_norm": 59.06554412841797, "kl": 0.3672589622437954, "learning_rate": 1.9510565162951534e-06, "loss": 0.2179, "num_tokens": 963480.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 96 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0388, "frac_reward_zero_std": 0.5, "grad_norm": 22.509157180786133, "kl": 0.18590721115469933, "learning_rate": 1.9488760116444964e-06, "loss": -0.3407, "num_tokens": 972360.0, "reward": 0.6875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 97 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0392, "frac_reward_zero_std": 0.5, "grad_norm": 36.51370620727539, "kl": 0.25807878002524376, "learning_rate": 1.9466492601156963e-06, "loss": 0.126, "num_tokens": 981240.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 98 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0396, "frac_reward_zero_std": 0.75, "grad_norm": 13.788187026977539, "kl": 0.23087293654680252, "learning_rate": 1.944376370237481e-06, "loss": 0.2509, "num_tokens": 990120.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 99 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.04, "frac_reward_zero_std": 0.5, "grad_norm": 8.3355073928833, "kl": 0.1271139644086361, "learning_rate": 1.9420574527872966e-06, "loss": -0.1245, "num_tokens": 999000.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 100 }, { "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.04, "eval_loss": -0.00047406076919287443, "eval_runtime": 10044.1197, "eval_samples_per_second": 0.02, "eval_steps_per_second": 0.005, "frac_reward_zero_std": 0.66, "num_tokens": 1110000.0, "reward": 0.415, "reward_std": 0.17928203105926513, "rewards/compute_reward/mean": 0.415, "rewards/compute_reward/std": 0.17928203105926513, "step": 100 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0404, "frac_reward_zero_std": 0.75, "grad_norm": 1.7402211427688599, "kl": 0.12196930311620235, "learning_rate": 1.9396926207859082e-06, "loss": 0.217, "num_tokens": 1118880.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 101 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0408, "frac_reward_zero_std": 0.5, "grad_norm": 7.048949241638184, "kl": 0.19601425155997276, "learning_rate": 1.9372819894918914e-06, "loss": 0.0008, "num_tokens": 1127760.0, "reward": 0.75, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 102 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0412, "frac_reward_zero_std": 0.5, "grad_norm": 14.87850284576416, "kl": 0.20362638682127, "learning_rate": 1.9348256763960142e-06, "loss": -0.2157, "num_tokens": 1136640.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 103 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0416, "frac_reward_zero_std": 1.0, "grad_norm": 19.110157012939453, "kl": 0.17501263320446014, "learning_rate": 1.9323238012155122e-06, "loss": 0.0007, "num_tokens": 1145520.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 104 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.042, "frac_reward_zero_std": 0.5, "grad_norm": 16.843177795410156, "kl": 0.2675781920552254, "learning_rate": 1.929776485888251e-06, "loss": 0.376, "num_tokens": 1154400.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 105 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0424, "frac_reward_zero_std": 0.75, "grad_norm": 24.672325134277344, "kl": 0.2085116021335125, "learning_rate": 1.9271838545667875e-06, "loss": -0.3741, "num_tokens": 1163280.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 106 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0428, "frac_reward_zero_std": 0.75, "grad_norm": 3.470179319381714, "kl": 0.3066542334854603, "learning_rate": 1.9245460336123133e-06, "loss": 0.1262, "num_tokens": 1172160.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 107 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0432, "frac_reward_zero_std": 0.5, "grad_norm": 33.54941940307617, "kl": 0.36576371639966965, "learning_rate": 1.9218631515885003e-06, "loss": 0.1264, "num_tokens": 1181040.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 108 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0436, "frac_reward_zero_std": 0.75, "grad_norm": 92.37312316894531, "kl": 1.9885881915688515, "learning_rate": 1.9191353392552343e-06, "loss": 0.008, "num_tokens": 1189920.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 109 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.044, "frac_reward_zero_std": 0.75, "grad_norm": 33.686317443847656, "kl": 0.45819035917520523, "learning_rate": 1.9163627295622395e-06, "loss": -0.3731, "num_tokens": 1198800.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 110 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0444, "frac_reward_zero_std": 0.5, "grad_norm": 25.443355560302734, "kl": 0.4073995351791382, "learning_rate": 1.9135454576426007e-06, "loss": -0.4983, "num_tokens": 1207680.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 111 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0448, "frac_reward_zero_std": 0.75, "grad_norm": 48.13676834106445, "kl": 0.4856240525841713, "learning_rate": 1.910683660806177e-06, "loss": 0.4103, "num_tokens": 1216560.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 112 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0452, "frac_reward_zero_std": 0.5, "grad_norm": 26.295862197875977, "kl": 0.3765817731618881, "learning_rate": 1.9077774785329087e-06, "loss": 0.5929, "num_tokens": 1225440.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 113 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0456, "frac_reward_zero_std": 0.5, "grad_norm": 60.09946823120117, "kl": 0.5815477445721626, "learning_rate": 1.9048270524660196e-06, "loss": -0.5891, "num_tokens": 1234320.0, "reward": 0.25, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 114 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.046, "frac_reward_zero_std": 0.75, "grad_norm": 43.9576530456543, "kl": 0.48454301804304123, "learning_rate": 1.9018325264051138e-06, "loss": 0.4349, "num_tokens": 1243200.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 115 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0464, "frac_reward_zero_std": 0.5, "grad_norm": 34.4670295715332, "kl": 0.6097010597586632, "learning_rate": 1.8987940462991669e-06, "loss": 0.4689, "num_tokens": 1252080.0, "reward": 0.6875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 116 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0468, "frac_reward_zero_std": 0.75, "grad_norm": 44.33097839355469, "kl": 0.6179208308458328, "learning_rate": 1.8957117602394128e-06, "loss": -0.8969, "num_tokens": 1260960.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 117 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0472, "frac_reward_zero_std": 0.5, "grad_norm": 36.61655044555664, "kl": 0.6962894722819328, "learning_rate": 1.8925858184521255e-06, "loss": -0.3721, "num_tokens": 1269840.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 118 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0476, "frac_reward_zero_std": 0.25, "grad_norm": 80.13101959228516, "kl": 0.6878734081983566, "learning_rate": 1.8894163732912974e-06, "loss": 0.3442, "num_tokens": 1278720.0, "reward": 0.3125, "reward_std": 0.375, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 119 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.048, "frac_reward_zero_std": 0.75, "grad_norm": 90.39923858642578, "kl": 0.7930669263005257, "learning_rate": 1.8862035792312146e-06, "loss": 0.1281, "num_tokens": 1287600.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 120 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0484, "frac_reward_zero_std": 1.0, "grad_norm": 37.05628204345703, "kl": 0.7748868465423584, "learning_rate": 1.8829475928589268e-06, "loss": 0.1281, "num_tokens": 1296480.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 121 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0488, "frac_reward_zero_std": 0.0, "grad_norm": 85.06033325195312, "kl": 0.9438418298959732, "learning_rate": 1.8796485728666165e-06, "loss": -0.6791, "num_tokens": 1305360.0, "reward": 0.4375, "reward_std": 0.51933753490448, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 122 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0492, "frac_reward_zero_std": 0.5, "grad_norm": 116.63643646240234, "kl": 1.1643976122140884, "learning_rate": 1.8763066800438634e-06, "loss": 0.9375, "num_tokens": 1314240.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 123 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0496, "frac_reward_zero_std": 0.75, "grad_norm": 87.37930297851562, "kl": 1.1998874247074127, "learning_rate": 1.8729220772698095e-06, "loss": -0.1202, "num_tokens": 1323120.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 124 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.05, "frac_reward_zero_std": 0.75, "grad_norm": 78.85758972167969, "kl": 1.4393538534641266, "learning_rate": 1.869494929505219e-06, "loss": 0.0058, "num_tokens": 1332000.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 125 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0504, "frac_reward_zero_std": 0.25, "grad_norm": 103.32394409179688, "kl": 2.1740673780441284, "learning_rate": 1.8660254037844386e-06, "loss": 0.1337, "num_tokens": 1340880.0, "reward": 0.5625, "reward_std": 0.375, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 126 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0508, "frac_reward_zero_std": 0.25, "grad_norm": 121.61659240722656, "kl": 1.3770731836557388, "learning_rate": 1.8625136692072574e-06, "loss": 0.097, "num_tokens": 1349760.0, "reward": 0.25, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 127 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0512, "frac_reward_zero_std": 0.75, "grad_norm": 87.8947525024414, "kl": 1.433599203824997, "learning_rate": 1.8589598969306644e-06, "loss": 0.0972, "num_tokens": 1358640.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 128 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0516, "frac_reward_zero_std": 0.5, "grad_norm": 99.49671936035156, "kl": 1.3952637314796448, "learning_rate": 1.8553642601605066e-06, "loss": -0.0524, "num_tokens": 1367520.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 129 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.052, "frac_reward_zero_std": 0.75, "grad_norm": 44.72431564331055, "kl": 1.3837022185325623, "learning_rate": 1.8517269341430474e-06, "loss": -0.6193, "num_tokens": 1376400.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 130 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0524, "frac_reward_zero_std": 0.5, "grad_norm": 72.39948272705078, "kl": 1.6086200773715973, "learning_rate": 1.8480480961564257e-06, "loss": 0.3479, "num_tokens": 1385280.0, "reward": 0.6875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 131 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0528, "frac_reward_zero_std": 0.75, "grad_norm": 75.30799102783203, "kl": 1.511486679315567, "learning_rate": 1.844327925502015e-06, "loss": 0.0395, "num_tokens": 1394160.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 132 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0532, "frac_reward_zero_std": 0.75, "grad_norm": 67.3117446899414, "kl": 1.4110450446605682, "learning_rate": 1.8405666034956842e-06, "loss": -0.5523, "num_tokens": 1403040.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 133 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0536, "frac_reward_zero_std": 0.25, "grad_norm": 91.60472869873047, "kl": 1.495105355978012, "learning_rate": 1.8367643134589616e-06, "loss": 0.2224, "num_tokens": 1411920.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 134 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.054, "frac_reward_zero_std": 0.75, "grad_norm": 89.93647766113281, "kl": 1.5621980130672455, "learning_rate": 1.8329212407100993e-06, "loss": 0.2227, "num_tokens": 1420800.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 135 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0544, "frac_reward_zero_std": 1.0, "grad_norm": 38.31406784057617, "kl": 1.4404120743274689, "learning_rate": 1.8290375725550415e-06, "loss": 0.0058, "num_tokens": 1429680.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 136 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0548, "frac_reward_zero_std": 0.75, "grad_norm": 67.23009490966797, "kl": 1.8493570387363434, "learning_rate": 1.825113498278295e-06, "loss": 0.2239, "num_tokens": 1438560.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 137 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0552, "frac_reward_zero_std": 1.0, "grad_norm": 26.11929702758789, "kl": 1.655295968055725, "learning_rate": 1.821149209133704e-06, "loss": -0.2098, "num_tokens": 1447440.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 138 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0556, "frac_reward_zero_std": 0.75, "grad_norm": 68.21998596191406, "kl": 1.6474502980709076, "learning_rate": 1.8171448983351283e-06, "loss": -0.1184, "num_tokens": 1456320.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 139 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.056, "frac_reward_zero_std": 0.5, "grad_norm": 74.41326141357422, "kl": 1.287375420331955, "learning_rate": 1.8131007610470274e-06, "loss": 0.0051, "num_tokens": 1465200.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 140 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0564, "frac_reward_zero_std": 0.25, "grad_norm": 69.33372497558594, "kl": 1.3127607554197311, "learning_rate": 1.8090169943749474e-06, "loss": 0.8801, "num_tokens": 1474080.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 141 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0568, "frac_reward_zero_std": 0.25, "grad_norm": 76.3515625, "kl": 1.296535700559616, "learning_rate": 1.804893797355914e-06, "loss": -0.8027, "num_tokens": 1482960.0, "reward": 0.375, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 142 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0572, "frac_reward_zero_std": 0.75, "grad_norm": 87.84119415283203, "kl": 1.1495089530944824, "learning_rate": 1.8007313709487333e-06, "loss": -0.1873, "num_tokens": 1491840.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 143 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0576, "frac_reward_zero_std": 0.25, "grad_norm": 45.25433349609375, "kl": 1.1036857962608337, "learning_rate": 1.7965299180241961e-06, "loss": 0.2209, "num_tokens": 1500720.0, "reward": 0.5625, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 144 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.058, "frac_reward_zero_std": 0.5, "grad_norm": 48.654361724853516, "kl": 1.1962954103946686, "learning_rate": 1.7922896433551906e-06, "loss": 0.5382, "num_tokens": 1509600.0, "reward": 0.125, "reward_std": 0.25, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 145 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0584, "frac_reward_zero_std": 0.75, "grad_norm": 56.95619201660156, "kl": 1.3726016283035278, "learning_rate": 1.7880107536067217e-06, "loss": -0.2445, "num_tokens": 1518480.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 146 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0588, "frac_reward_zero_std": 0.75, "grad_norm": 81.80111694335938, "kl": 1.270168662071228, "learning_rate": 1.7836934573258397e-06, "loss": 0.0051, "num_tokens": 1527360.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 147 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0592, "frac_reward_zero_std": 0.75, "grad_norm": 55.87933349609375, "kl": 1.5507261157035828, "learning_rate": 1.7793379649314742e-06, "loss": -0.4602, "num_tokens": 1536240.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 148 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0596, "frac_reward_zero_std": 0.5, "grad_norm": 40.03895950317383, "kl": 1.3443613648414612, "learning_rate": 1.7749444887041795e-06, "loss": -0.2446, "num_tokens": 1545120.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 149 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.06, "frac_reward_zero_std": 0.0, "grad_norm": 60.483856201171875, "kl": 1.244696855545044, "learning_rate": 1.7705132427757892e-06, "loss": 0.2884, "num_tokens": 1554000.0, "reward": 0.4375, "reward_std": 0.51933753490448, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 150 }, { "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.06, "eval_loss": -0.0004801125905942172, "eval_runtime": 5038.6351, "eval_samples_per_second": 0.04, "eval_steps_per_second": 0.01, "frac_reward_zero_std": 0.64, "num_tokens": 1665000.0, "reward": 0.37, "reward_std": 0.18928203105926514, "rewards/compute_reward/mean": 0.37, "rewards/compute_reward/std": 0.18928203105926514, "step": 150 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0604, "frac_reward_zero_std": 1.0, "grad_norm": 102.57103729248047, "kl": 0.8290184140205383, "learning_rate": 1.766044443118978e-06, "loss": 0.5032, "num_tokens": 1673880.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 151 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0608, "frac_reward_zero_std": 0.75, "grad_norm": 10.56777286529541, "kl": 1.1542262136936188, "learning_rate": 1.7615383075367368e-06, "loss": 0.2211, "num_tokens": 1682760.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 152 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0612, "frac_reward_zero_std": 0.5, "grad_norm": 37.89947509765625, "kl": 0.9573570787906647, "learning_rate": 1.7569950556517563e-06, "loss": 0.1623, "num_tokens": 1691640.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 153 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0616, "frac_reward_zero_std": 1.0, "grad_norm": 30.090473175048828, "kl": 0.8319032788276672, "learning_rate": 1.7524149088957242e-06, "loss": -0.3716, "num_tokens": 1700520.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 154 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.062, "frac_reward_zero_std": 0.5, "grad_norm": 37.990379333496094, "kl": 0.7478515356779099, "learning_rate": 1.747798090498532e-06, "loss": 0.0365, "num_tokens": 1709400.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 155 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0624, "frac_reward_zero_std": 0.5, "grad_norm": 27.682296752929688, "kl": 0.902895525097847, "learning_rate": 1.743144825477394e-06, "loss": 0.528, "num_tokens": 1718280.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 156 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0628, "frac_reward_zero_std": 0.75, "grad_norm": 50.9190788269043, "kl": 0.8830957859754562, "learning_rate": 1.738455340625884e-06, "loss": -0.4294, "num_tokens": 1727160.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 157 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0632, "frac_reward_zero_std": 0.5, "grad_norm": 20.148906707763672, "kl": 0.7896845638751984, "learning_rate": 1.7337298645028762e-06, "loss": -0.4967, "num_tokens": 1736040.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 158 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0636, "frac_reward_zero_std": 0.5, "grad_norm": 32.239356994628906, "kl": 0.804187685251236, "learning_rate": 1.7289686274214115e-06, "loss": 0.7196, "num_tokens": 1744920.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 159 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.064, "frac_reward_zero_std": 0.5, "grad_norm": 24.03925895690918, "kl": 0.8029276132583618, "learning_rate": 1.7241718614374676e-06, "loss": -0.5882, "num_tokens": 1753800.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 160 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0644, "frac_reward_zero_std": 0.75, "grad_norm": 30.868242263793945, "kl": 0.8987027704715729, "learning_rate": 1.719339800338651e-06, "loss": 0.2535, "num_tokens": 1762680.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 161 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0648, "frac_reward_zero_std": 0.5, "grad_norm": 19.09716033935547, "kl": 0.9851841777563095, "learning_rate": 1.7144726796328032e-06, "loss": -0.2125, "num_tokens": 1771560.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 162 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0652, "frac_reward_zero_std": 0.5, "grad_norm": 32.937808990478516, "kl": 0.7938939332962036, "learning_rate": 1.7095707365365209e-06, "loss": 0.4696, "num_tokens": 1780440.0, "reward": 0.125, "reward_std": 0.25, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 163 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0656, "frac_reward_zero_std": 0.25, "grad_norm": 37.37956619262695, "kl": 0.7686180472373962, "learning_rate": 1.7046342099635947e-06, "loss": 0.311, "num_tokens": 1789320.0, "reward": 0.3125, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 164 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.066, "frac_reward_zero_std": 0.75, "grad_norm": 29.587915420532227, "kl": 0.6731294244527817, "learning_rate": 1.6996633405133653e-06, "loss": -0.5552, "num_tokens": 1798200.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 165 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0664, "frac_reward_zero_std": 1.0, "grad_norm": 28.84545135498047, "kl": 0.874516099691391, "learning_rate": 1.6946583704589972e-06, "loss": 0.0035, "num_tokens": 1807080.0, "reward": 0.75, "reward_std": 0.0, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 166 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0668, "frac_reward_zero_std": 1.0, "grad_norm": 0.2955371141433716, "kl": 0.7283145040273666, "learning_rate": 1.6896195437356697e-06, "loss": 0.0029, "num_tokens": 1815960.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 167 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0672, "frac_reward_zero_std": 0.5, "grad_norm": 17.424720764160156, "kl": 0.9037010669708252, "learning_rate": 1.6845471059286886e-06, "loss": 0.2536, "num_tokens": 1824840.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 168 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0676, "frac_reward_zero_std": 0.75, "grad_norm": 26.717079162597656, "kl": 0.6133365780115128, "learning_rate": 1.6794413042615166e-06, "loss": -0.031, "num_tokens": 1833720.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 169 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.068, "frac_reward_zero_std": 0.75, "grad_norm": 24.280805587768555, "kl": 0.7366704791784286, "learning_rate": 1.6743023875837233e-06, "loss": -0.0885, "num_tokens": 1842600.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 170 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0684, "frac_reward_zero_std": 0.5, "grad_norm": 23.280994415283203, "kl": 0.752781942486763, "learning_rate": 1.669130606358858e-06, "loss": -0.3719, "num_tokens": 1851480.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 171 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0688, "frac_reward_zero_std": 1.0, "grad_norm": 28.28022575378418, "kl": 0.7437139600515366, "learning_rate": 1.6639262126522415e-06, "loss": 0.2529, "num_tokens": 1860360.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 172 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0692, "frac_reward_zero_std": 0.25, "grad_norm": 18.955217361450195, "kl": 0.51456718146801, "learning_rate": 1.6586894601186803e-06, "loss": 0.0355, "num_tokens": 1869240.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 173 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0696, "frac_reward_zero_std": 0.75, "grad_norm": 27.11832618713379, "kl": 0.6736243739724159, "learning_rate": 1.6534206039901055e-06, "loss": -0.0308, "num_tokens": 1878120.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 174 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.07, "frac_reward_zero_std": 0.25, "grad_norm": 31.381895065307617, "kl": 0.6929000839591026, "learning_rate": 1.6481199010631309e-06, "loss": 0.0028, "num_tokens": 1887000.0, "reward": 0.4375, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 175 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0704, "frac_reward_zero_std": 1.0, "grad_norm": 48.79582977294922, "kl": 0.562452644109726, "learning_rate": 1.6427876096865393e-06, "loss": 0.0022, "num_tokens": 1895880.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 176 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0708, "frac_reward_zero_std": 0.25, "grad_norm": 14.692187309265137, "kl": 0.4838453531265259, "learning_rate": 1.6374239897486897e-06, "loss": -0.123, "num_tokens": 1904760.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 177 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0712, "frac_reward_zero_std": 0.5, "grad_norm": 32.34980773925781, "kl": 0.5593864694237709, "learning_rate": 1.6320293026648508e-06, "loss": 0.0357, "num_tokens": 1913640.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 178 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0716, "frac_reward_zero_std": 0.25, "grad_norm": 47.210731506347656, "kl": 1.054399996995926, "learning_rate": 1.6266038113644605e-06, "loss": -0.3372, "num_tokens": 1922520.0, "reward": 0.3125, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 179 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.072, "frac_reward_zero_std": 1.0, "grad_norm": 25.16176986694336, "kl": 0.6305755823850632, "learning_rate": 1.6211477802783102e-06, "loss": 0.4355, "num_tokens": 1931400.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 180 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0724, "frac_reward_zero_std": 1.0, "grad_norm": 2.085853099822998, "kl": 0.6664913669228554, "learning_rate": 1.615661475325658e-06, "loss": 0.0027, "num_tokens": 1940280.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 181 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0728, "frac_reward_zero_std": 0.75, "grad_norm": 0.25921252369880676, "kl": 0.5506083369255066, "learning_rate": 1.6101451639012677e-06, "loss": 0.0022, "num_tokens": 1949160.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 182 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0732, "frac_reward_zero_std": 0.75, "grad_norm": 19.31712532043457, "kl": 0.5009290054440498, "learning_rate": 1.604599114862375e-06, "loss": 0.4349, "num_tokens": 1958040.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 183 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0736, "frac_reward_zero_std": 1.0, "grad_norm": 14.426288604736328, "kl": 0.8416369184851646, "learning_rate": 1.5990235985155857e-06, "loss": -0.4296, "num_tokens": 1966920.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 184 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.074, "frac_reward_zero_std": 0.25, "grad_norm": 9.469975471496582, "kl": 0.6057900637388229, "learning_rate": 1.5934188866037015e-06, "loss": -0.0311, "num_tokens": 1975800.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 185 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0744, "frac_reward_zero_std": 0.75, "grad_norm": 28.550851821899414, "kl": 0.6094760373234749, "learning_rate": 1.587785252292473e-06, "loss": 0.0359, "num_tokens": 1984680.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 186 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0748, "frac_reward_zero_std": 0.0, "grad_norm": 14.759550094604492, "kl": 0.5186473429203033, "learning_rate": 1.5821229701572893e-06, "loss": -0.2479, "num_tokens": 1993560.0, "reward": 0.5625, "reward_std": 0.51933753490448, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 187 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0752, "frac_reward_zero_std": 0.75, "grad_norm": 24.977191925048828, "kl": 0.7406383752822876, "learning_rate": 1.5764323161697932e-06, "loss": 0.0364, "num_tokens": 2002440.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 188 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0756, "frac_reward_zero_std": 0.5, "grad_norm": 19.971651077270508, "kl": 0.578004002571106, "learning_rate": 1.5707135676844319e-06, "loss": 0.0938, "num_tokens": 2011320.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 189 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.076, "frac_reward_zero_std": 0.5, "grad_norm": 12.372550964355469, "kl": 0.4948463439941406, "learning_rate": 1.564967003424938e-06, "loss": 0.3099, "num_tokens": 2020200.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 190 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0764, "frac_reward_zero_std": 0.5, "grad_norm": 13.063384056091309, "kl": 0.4803493767976761, "learning_rate": 1.5591929034707466e-06, "loss": 0.0689, "num_tokens": 2029080.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 191 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0768, "frac_reward_zero_std": 0.25, "grad_norm": 20.98461151123047, "kl": 0.5075587034225464, "learning_rate": 1.553391549243344e-06, "loss": -0.4979, "num_tokens": 2037960.0, "reward": 0.5625, "reward_std": 0.375, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 192 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0772, "frac_reward_zero_std": 0.75, "grad_norm": 22.815135955810547, "kl": 0.5010422468185425, "learning_rate": 1.5475632234925502e-06, "loss": 0.3769, "num_tokens": 2046840.0, "reward": 0.8125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.8125, "rewards/compute_reward/std": 0.40311288833618164, "step": 193 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0776, "frac_reward_zero_std": 1.0, "grad_norm": 16.002519607543945, "kl": 0.4861938878893852, "learning_rate": 1.54170821028274e-06, "loss": -0.123, "num_tokens": 2055720.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 194 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.078, "frac_reward_zero_std": 0.5, "grad_norm": 12.706498146057129, "kl": 0.45841076225042343, "learning_rate": 1.5358267949789964e-06, "loss": 0.2518, "num_tokens": 2064600.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 195 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0784, "frac_reward_zero_std": 0.25, "grad_norm": 49.78022003173828, "kl": 0.5610195696353912, "learning_rate": 1.5299192642332049e-06, "loss": 0.0357, "num_tokens": 2073480.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 196 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0788, "frac_reward_zero_std": 0.25, "grad_norm": 26.672563552856445, "kl": 0.4388924762606621, "learning_rate": 1.5239859059700793e-06, "loss": -0.2817, "num_tokens": 2082360.0, "reward": 0.25, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 197 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0792, "frac_reward_zero_std": 1.0, "grad_norm": 22.82637596130371, "kl": 0.39291463047266006, "learning_rate": 1.5180270093731302e-06, "loss": 0.0016, "num_tokens": 2091240.0, "reward": 0.75, "reward_std": 0.0, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 198 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0796, "frac_reward_zero_std": 0.75, "grad_norm": 1.730869174003601, "kl": 0.4376737102866173, "learning_rate": 1.5120428648705715e-06, "loss": -0.1232, "num_tokens": 2100120.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 199 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.08, "frac_reward_zero_std": 1.0, "grad_norm": 6.7394537925720215, "kl": 0.4456604868173599, "learning_rate": 1.5060337641211636e-06, "loss": 0.1268, "num_tokens": 2109000.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 200 }, { "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.08, "eval_loss": 0.0004982490791007876, "eval_runtime": 7044.2588, "eval_samples_per_second": 0.028, "eval_steps_per_second": 0.007, "frac_reward_zero_std": 0.6122448979591837, "num_tokens": 2217780.0, "reward": 0.413265306122449, "reward_std": 0.19703470444192692, "rewards/compute_reward/mean": 0.413265306122449, "rewards/compute_reward/std": 0.19703470444192692, "step": 200 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0804, "frac_reward_zero_std": 0.5, "grad_norm": 16.79532814025879, "kl": 0.47658685594797134, "learning_rate": 1.5e-06, "loss": 0.0019, "num_tokens": 2226660.0, "reward": 0.6875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 201 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0808, "frac_reward_zero_std": 0.5, "grad_norm": 19.146230697631836, "kl": 0.47041744738817215, "learning_rate": 1.4939418665842309e-06, "loss": 0.0019, "num_tokens": 2235540.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 202 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0812, "frac_reward_zero_std": 0.5, "grad_norm": 10.93484115600586, "kl": 0.48019571602344513, "learning_rate": 1.4878596591387326e-06, "loss": 0.0019, "num_tokens": 2244420.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 203 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0816, "frac_reward_zero_std": 0.25, "grad_norm": 24.61666488647461, "kl": 0.6040880158543587, "learning_rate": 1.4817536741017151e-06, "loss": 0.0024, "num_tokens": 2253300.0, "reward": 0.6875, "reward_std": 0.375, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 204 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.082, "frac_reward_zero_std": 0.5, "grad_norm": 21.1997013092041, "kl": 0.7307578474283218, "learning_rate": 1.4756242090702753e-06, "loss": 0.0029, "num_tokens": 2262180.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 205 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0824, "frac_reward_zero_std": 0.5, "grad_norm": 21.705259323120117, "kl": 1.0300785303115845, "learning_rate": 1.4694715627858908e-06, "loss": 0.0041, "num_tokens": 2271060.0, "reward": 0.125, "reward_std": 0.25, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 206 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0828, "frac_reward_zero_std": 0.75, "grad_norm": 20.124446868896484, "kl": 0.7309058308601379, "learning_rate": 1.4632960351198617e-06, "loss": 0.0029, "num_tokens": 2279940.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 207 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0832, "frac_reward_zero_std": 0.75, "grad_norm": 24.016128540039062, "kl": 0.8935306221246719, "learning_rate": 1.4570979270586943e-06, "loss": 0.0036, "num_tokens": 2288820.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 208 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0836, "frac_reward_zero_std": 0.75, "grad_norm": 14.154572486877441, "kl": 0.9832312911748886, "learning_rate": 1.4508775406894306e-06, "loss": 0.0039, "num_tokens": 2297700.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 209 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.084, "frac_reward_zero_std": 0.5, "grad_norm": 27.507604598999023, "kl": 0.9729635268449783, "learning_rate": 1.4446351791849273e-06, "loss": 0.0039, "num_tokens": 2306580.0, "reward": 0.25, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 210 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0844, "frac_reward_zero_std": 1.0, "grad_norm": 0.7109271287918091, "kl": 1.0758344233036041, "learning_rate": 1.4383711467890773e-06, "loss": 0.0043, "num_tokens": 2315460.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 211 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0848, "frac_reward_zero_std": 0.25, "grad_norm": 38.93098449707031, "kl": 1.124314084649086, "learning_rate": 1.4320857488019824e-06, "loss": 0.0045, "num_tokens": 2324340.0, "reward": 0.4375, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 212 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0852, "frac_reward_zero_std": 0.0, "grad_norm": 41.04912567138672, "kl": 1.214218646287918, "learning_rate": 1.4257792915650725e-06, "loss": 0.0049, "num_tokens": 2333220.0, "reward": 0.375, "reward_std": 0.5386751294136047, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 213 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0856, "frac_reward_zero_std": 0.75, "grad_norm": 27.365114212036133, "kl": 1.4852878749370575, "learning_rate": 1.419452082446177e-06, "loss": 0.0059, "num_tokens": 2342100.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 214 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.086, "frac_reward_zero_std": 0.5, "grad_norm": 41.17222595214844, "kl": 1.54292231798172, "learning_rate": 1.4131044298245418e-06, "loss": 0.0062, "num_tokens": 2350980.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 215 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0864, "frac_reward_zero_std": 0.25, "grad_norm": 40.92033386230469, "kl": 1.4181734025478363, "learning_rate": 1.4067366430758004e-06, "loss": 0.0057, "num_tokens": 2359860.0, "reward": 0.5625, "reward_std": 0.375, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 216 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0868, "frac_reward_zero_std": 1.0, "grad_norm": 0.1861588954925537, "kl": 1.296187847852707, "learning_rate": 1.400349032556895e-06, "loss": 0.0052, "num_tokens": 2368740.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 217 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0872, "frac_reward_zero_std": 0.25, "grad_norm": 37.92814636230469, "kl": 1.2197140753269196, "learning_rate": 1.393941909590951e-06, "loss": 0.0049, "num_tokens": 2377620.0, "reward": 0.375, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 218 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0876, "frac_reward_zero_std": 0.25, "grad_norm": 46.5506706237793, "kl": 1.2918847501277924, "learning_rate": 1.3875155864521028e-06, "loss": 0.0052, "num_tokens": 2386500.0, "reward": 0.375, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 219 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.088, "frac_reward_zero_std": 0.75, "grad_norm": 29.01485252380371, "kl": 1.2826491594314575, "learning_rate": 1.3810703763502743e-06, "loss": 0.0051, "num_tokens": 2395380.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 220 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0884, "frac_reward_zero_std": 0.5, "grad_norm": 29.103654861450195, "kl": 1.2425887286663055, "learning_rate": 1.374606593415912e-06, "loss": 0.005, "num_tokens": 2404260.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 221 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0888, "frac_reward_zero_std": 0.75, "grad_norm": 21.997817993164062, "kl": 1.2478073239326477, "learning_rate": 1.3681245526846781e-06, "loss": 0.005, "num_tokens": 2413140.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 222 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0892, "frac_reward_zero_std": 1.0, "grad_norm": 1.3906086683273315, "kl": 1.2335122525691986, "learning_rate": 1.361624570082092e-06, "loss": 0.0049, "num_tokens": 2422020.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 223 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0896, "frac_reward_zero_std": 0.5, "grad_norm": 31.229379653930664, "kl": 1.1545628011226654, "learning_rate": 1.3551069624081371e-06, "loss": 0.0046, "num_tokens": 2430900.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 224 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.09, "frac_reward_zero_std": 0.25, "grad_norm": 37.78306198120117, "kl": 1.1417304575443268, "learning_rate": 1.3485720473218152e-06, "loss": 0.0046, "num_tokens": 2439780.0, "reward": 0.25, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 225 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0904, "frac_reward_zero_std": 1.0, "grad_norm": 0.7812005877494812, "kl": 1.2861113250255585, "learning_rate": 1.3420201433256689e-06, "loss": 0.0051, "num_tokens": 2448660.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 226 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0908, "frac_reward_zero_std": 0.5, "grad_norm": 37.55674743652344, "kl": 1.3373107016086578, "learning_rate": 1.3354515697502551e-06, "loss": 0.0053, "num_tokens": 2457540.0, "reward": 0.5, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 227 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0912, "frac_reward_zero_std": 1.0, "grad_norm": 1.6536084413528442, "kl": 1.4782708883285522, "learning_rate": 1.3288666467385831e-06, "loss": 0.0059, "num_tokens": 2466420.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 228 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0916, "frac_reward_zero_std": 0.75, "grad_norm": 21.825687408447266, "kl": 1.3577501475811005, "learning_rate": 1.3222656952305111e-06, "loss": 0.0054, "num_tokens": 2475300.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 229 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.092, "frac_reward_zero_std": 0.75, "grad_norm": 22.898921966552734, "kl": 1.3672588467597961, "learning_rate": 1.3156490369471024e-06, "loss": 0.0055, "num_tokens": 2484180.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 230 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0924, "frac_reward_zero_std": 0.5, "grad_norm": 36.318870544433594, "kl": 1.4862727522850037, "learning_rate": 1.3090169943749473e-06, "loss": 0.0059, "num_tokens": 2493060.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 231 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0928, "frac_reward_zero_std": 0.75, "grad_norm": 24.24749755859375, "kl": 1.3374225199222565, "learning_rate": 1.3023698907504446e-06, "loss": 0.0053, "num_tokens": 2501940.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 232 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0932, "frac_reward_zero_std": 0.25, "grad_norm": 43.74175262451172, "kl": 1.3122918009757996, "learning_rate": 1.2957080500440467e-06, "loss": 0.0052, "num_tokens": 2510820.0, "reward": 0.5625, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 233 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0936, "frac_reward_zero_std": 0.75, "grad_norm": 27.245943069458008, "kl": 1.3207648694515228, "learning_rate": 1.2890317969444716e-06, "loss": 0.0053, "num_tokens": 2519700.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 234 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.094, "frac_reward_zero_std": 0.5, "grad_norm": 39.74917984008789, "kl": 1.4372384995222092, "learning_rate": 1.2823414568428766e-06, "loss": 0.0057, "num_tokens": 2528580.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 235 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0944, "frac_reward_zero_std": 0.5, "grad_norm": 39.7496223449707, "kl": 1.1310810148715973, "learning_rate": 1.275637355816999e-06, "loss": 0.0045, "num_tokens": 2537460.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 236 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0948, "frac_reward_zero_std": 0.25, "grad_norm": 29.689462661743164, "kl": 1.105551227927208, "learning_rate": 1.2689198206152656e-06, "loss": 0.0044, "num_tokens": 2546340.0, "reward": 0.75, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 237 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0952, "frac_reward_zero_std": 0.75, "grad_norm": 19.529863357543945, "kl": 1.0155319422483444, "learning_rate": 1.2621891786408648e-06, "loss": 0.0041, "num_tokens": 2555220.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 238 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0956, "frac_reward_zero_std": 0.5, "grad_norm": 40.87265396118164, "kl": 1.1060357093811035, "learning_rate": 1.2554457579357905e-06, "loss": 0.0044, "num_tokens": 2564100.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 239 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.096, "frac_reward_zero_std": 0.5, "grad_norm": 43.7393798828125, "kl": 1.0452822744846344, "learning_rate": 1.2486898871648551e-06, "loss": 0.0042, "num_tokens": 2572980.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 240 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0964, "frac_reward_zero_std": 0.5, "grad_norm": 42.85210418701172, "kl": 1.2131805717945099, "learning_rate": 1.2419218955996676e-06, "loss": 0.0049, "num_tokens": 2581860.0, "reward": 0.25, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 241 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0968, "frac_reward_zero_std": 0.75, "grad_norm": 25.9547119140625, "kl": 1.0465219914913177, "learning_rate": 1.23514211310259e-06, "loss": 0.0042, "num_tokens": 2590740.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 242 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0972, "frac_reward_zero_std": 0.5, "grad_norm": 45.28058624267578, "kl": 1.167938232421875, "learning_rate": 1.2283508701106558e-06, "loss": 0.0047, "num_tokens": 2599620.0, "reward": 0.125, "reward_std": 0.25, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 243 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0976, "frac_reward_zero_std": 0.25, "grad_norm": 60.2187614440918, "kl": 0.9804764240980148, "learning_rate": 1.2215484976194675e-06, "loss": 0.0039, "num_tokens": 2608500.0, "reward": 0.375, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 244 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.098, "frac_reward_zero_std": 0.75, "grad_norm": 29.929412841796875, "kl": 1.051619529724121, "learning_rate": 1.2147353271670632e-06, "loss": 0.0042, "num_tokens": 2617380.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 245 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0984, "frac_reward_zero_std": 0.5, "grad_norm": 38.48051452636719, "kl": 1.0911252945661545, "learning_rate": 1.207911690817759e-06, "loss": 0.0044, "num_tokens": 2626260.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 246 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0988, "frac_reward_zero_std": 0.75, "grad_norm": 24.381433486938477, "kl": 1.092143639922142, "learning_rate": 1.2010779211459648e-06, "loss": 0.0044, "num_tokens": 2635140.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 247 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0992, "frac_reward_zero_std": 0.5, "grad_norm": 47.642539978027344, "kl": 0.852707639336586, "learning_rate": 1.194234351219972e-06, "loss": 0.0034, "num_tokens": 2644020.0, "reward": 0.75, "reward_std": 0.25, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 248 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.0996, "frac_reward_zero_std": 0.5, "grad_norm": 37.34217071533203, "kl": 1.0041948407888412, "learning_rate": 1.1873813145857248e-06, "loss": 0.004, "num_tokens": 2652900.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 249 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1, "frac_reward_zero_std": 0.5, "grad_norm": 27.347667694091797, "kl": 0.9091732352972031, "learning_rate": 1.18051914525056e-06, "loss": 0.0036, "num_tokens": 2661780.0, "reward": 0.8125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.8125, "rewards/compute_reward/std": 0.40311288833618164, "step": 250 }, { "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1, "eval_loss": 0.0010513612069189548, "eval_runtime": 5613.4422, "eval_samples_per_second": 0.036, "eval_steps_per_second": 0.009, "frac_reward_zero_std": 0.7, "num_tokens": 2772780.0, "reward": 0.39, "reward_std": 0.15773502588272095, "rewards/compute_reward/mean": 0.39, "rewards/compute_reward/std": 0.15773502588272095, "step": 250 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1004, "frac_reward_zero_std": 0.5, "grad_norm": 22.235748291015625, "kl": 1.82113316655159, "learning_rate": 1.1736481776669305e-06, "loss": 0.0073, "num_tokens": 2781660.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 251 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1008, "frac_reward_zero_std": 0.75, "grad_norm": 39.217552185058594, "kl": 1.126240998506546, "learning_rate": 1.1667687467161023e-06, "loss": -0.212, "num_tokens": 2790540.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 252 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1012, "frac_reward_zero_std": 0.5, "grad_norm": 34.31150436401367, "kl": 1.3839496076107025, "learning_rate": 1.1598811876918349e-06, "loss": 0.0055, "num_tokens": 2799420.0, "reward": 0.25, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 253 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1016, "frac_reward_zero_std": 0.25, "grad_norm": 39.29949188232422, "kl": 0.9901166260242462, "learning_rate": 1.1529858362840382e-06, "loss": 0.5619, "num_tokens": 2808300.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 254 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.102, "frac_reward_zero_std": 1.0, "grad_norm": 16.917476654052734, "kl": 0.8453292548656464, "learning_rate": 1.1460830285624116e-06, "loss": -0.3381, "num_tokens": 2817180.0, "reward": 0.75, "reward_std": 0.0, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 255 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1024, "frac_reward_zero_std": 0.5, "grad_norm": 49.95372772216797, "kl": 1.1023323237895966, "learning_rate": 1.1391731009600653e-06, "loss": -0.1206, "num_tokens": 2826060.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 256 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1028, "frac_reward_zero_std": 0.75, "grad_norm": 31.83892822265625, "kl": 1.1208092421293259, "learning_rate": 1.1322563902571225e-06, "loss": 0.1295, "num_tokens": 2834940.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 257 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1032, "frac_reward_zero_std": 0.5, "grad_norm": 19.977169036865234, "kl": 0.9258421361446381, "learning_rate": 1.1253332335643042e-06, "loss": -0.3712, "num_tokens": 2843820.0, "reward": 0.3125, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 258 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1036, "frac_reward_zero_std": 0.25, "grad_norm": 45.17646026611328, "kl": 1.021436646580696, "learning_rate": 1.1184039683065012e-06, "loss": 0.504, "num_tokens": 2852700.0, "reward": 0.3125, "reward_std": 0.375, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 259 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.104, "frac_reward_zero_std": 0.75, "grad_norm": 29.35318374633789, "kl": 0.7680477648973465, "learning_rate": 1.1114689322063254e-06, "loss": -0.1219, "num_tokens": 2861580.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 260 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1044, "frac_reward_zero_std": 0.25, "grad_norm": 28.826967239379883, "kl": 0.9056837409734726, "learning_rate": 1.1045284632676535e-06, "loss": 0.2201, "num_tokens": 2870460.0, "reward": 0.3125, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 261 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1048, "frac_reward_zero_std": 0.75, "grad_norm": 30.201587677001953, "kl": 1.171682357788086, "learning_rate": 1.0975828997591495e-06, "loss": -0.2118, "num_tokens": 2879340.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 262 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1052, "frac_reward_zero_std": 0.25, "grad_norm": 31.657508850097656, "kl": 0.9881609976291656, "learning_rate": 1.0906325801977803e-06, "loss": 0.1289, "num_tokens": 2888220.0, "reward": 0.4375, "reward_std": 0.375, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 263 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1056, "frac_reward_zero_std": 1.0, "grad_norm": 13.9951810836792, "kl": 1.0576434880495071, "learning_rate": 1.0836778433323157e-06, "loss": -0.1207, "num_tokens": 2897100.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 264 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.106, "frac_reward_zero_std": 0.25, "grad_norm": 24.13888931274414, "kl": 0.7200263738632202, "learning_rate": 1.0767190281268186e-06, "loss": 0.2193, "num_tokens": 2905980.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 265 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1064, "frac_reward_zero_std": 0.25, "grad_norm": 47.248741149902344, "kl": 0.9038263559341431, "learning_rate": 1.069756473744125e-06, "loss": -0.2463, "num_tokens": 2914860.0, "reward": 0.375, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 266 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1068, "frac_reward_zero_std": 0.5, "grad_norm": 30.434354782104492, "kl": 0.8619530349969864, "learning_rate": 1.0627905195293135e-06, "loss": 0.2869, "num_tokens": 2923740.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 267 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1072, "frac_reward_zero_std": 0.75, "grad_norm": 14.322568893432617, "kl": 0.8955692648887634, "learning_rate": 1.055821504993164e-06, "loss": -0.3713, "num_tokens": 2932620.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 268 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1076, "frac_reward_zero_std": 0.25, "grad_norm": 38.34691619873047, "kl": 0.8839395940303802, "learning_rate": 1.0488497697956134e-06, "loss": 0.0035, "num_tokens": 2941500.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 269 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.108, "frac_reward_zero_std": 0.5, "grad_norm": 34.70941925048828, "kl": 0.8146094232797623, "learning_rate": 1.0418756537291995e-06, "loss": 0.4697, "num_tokens": 2950380.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 270 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1084, "frac_reward_zero_std": 0.5, "grad_norm": 22.19832992553711, "kl": 0.7530200630426407, "learning_rate": 1.034899496702501e-06, "loss": -0.2135, "num_tokens": 2959260.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 271 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1088, "frac_reward_zero_std": 0.5, "grad_norm": 34.179874420166016, "kl": 0.8426708728075027, "learning_rate": 1.0279216387235689e-06, "loss": -0.1216, "num_tokens": 2968140.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 272 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1092, "frac_reward_zero_std": 0.25, "grad_norm": 20.24189567565918, "kl": 0.8162007182836533, "learning_rate": 1.020942419883357e-06, "loss": -0.0882, "num_tokens": 2977020.0, "reward": 0.4375, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 273 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1096, "frac_reward_zero_std": 0.25, "grad_norm": 28.002971649169922, "kl": 0.669597715139389, "learning_rate": 1.0139621803391454e-06, "loss": 0.0942, "num_tokens": 2985900.0, "reward": 0.3125, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 274 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.11, "frac_reward_zero_std": 0.75, "grad_norm": 19.8535213470459, "kl": 0.8049573004245758, "learning_rate": 1.0069812602979615e-06, "loss": 0.2197, "num_tokens": 2994780.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 275 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1104, "frac_reward_zero_std": 0.75, "grad_norm": 17.800783157348633, "kl": 0.8933698236942291, "learning_rate": 1e-06, "loss": -0.2129, "num_tokens": 3003660.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 276 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1108, "frac_reward_zero_std": 0.75, "grad_norm": 12.544022560119629, "kl": 0.7986416071653366, "learning_rate": 9.930187397020384e-07, "loss": 0.1282, "num_tokens": 3012540.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 277 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1112, "frac_reward_zero_std": 1.0, "grad_norm": 3.61922550201416, "kl": 0.7199362516403198, "learning_rate": 9.860378196608547e-07, "loss": -0.1221, "num_tokens": 3021420.0, "reward": 0.75, "reward_std": 0.0, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 278 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1116, "frac_reward_zero_std": 0.5, "grad_norm": 19.929853439331055, "kl": 0.7778548300266266, "learning_rate": 9.790575801166431e-07, "loss": 0.2531, "num_tokens": 3030300.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 279 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.112, "frac_reward_zero_std": 0.5, "grad_norm": 17.565427780151367, "kl": 0.8007985502481461, "learning_rate": 9.720783612764313e-07, "loss": -0.3717, "num_tokens": 3039180.0, "reward": 0.6875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 280 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1124, "frac_reward_zero_std": 0.75, "grad_norm": 8.918659210205078, "kl": 0.67110276222229, "learning_rate": 9.651005032974993e-07, "loss": 0.1277, "num_tokens": 3048060.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 281 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1128, "frac_reward_zero_std": 0.75, "grad_norm": 7.046802997589111, "kl": 0.6922563761472702, "learning_rate": 9.581243462708005e-07, "loss": 0.1277, "num_tokens": 3056940.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 282 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1132, "frac_reward_zero_std": 0.5, "grad_norm": 16.287311553955078, "kl": 0.7062659412622452, "learning_rate": 9.511502302043867e-07, "loss": -0.2471, "num_tokens": 3065820.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 283 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1136, "frac_reward_zero_std": 0.5, "grad_norm": 14.158629417419434, "kl": 0.8502112925052643, "learning_rate": 9.441784950068361e-07, "loss": 0.7198, "num_tokens": 3074700.0, "reward": 0.6875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 284 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.114, "frac_reward_zero_std": 1.0, "grad_norm": 11.075551986694336, "kl": 0.6977677792310715, "learning_rate": 9.372094804706866e-07, "loss": -0.5886, "num_tokens": 3083580.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 285 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1144, "frac_reward_zero_std": 0.75, "grad_norm": 11.198731422424316, "kl": 0.6550881266593933, "learning_rate": 9.302435262558747e-07, "loss": -0.1224, "num_tokens": 3092460.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 286 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1148, "frac_reward_zero_std": 0.75, "grad_norm": 14.440635681152344, "kl": 0.6763903424143791, "learning_rate": 9.232809718731813e-07, "loss": 0.1277, "num_tokens": 3101340.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 287 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1152, "frac_reward_zero_std": 0.5, "grad_norm": 14.166817665100098, "kl": 0.8325612992048264, "learning_rate": 9.163221566676847e-07, "loss": 0.2198, "num_tokens": 3110220.0, "reward": 0.75, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 288 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1156, "frac_reward_zero_std": 0.25, "grad_norm": 20.33856773376465, "kl": 0.6119056344032288, "learning_rate": 9.093674198022199e-07, "loss": -0.397, "num_tokens": 3119100.0, "reward": 0.5, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 289 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.116, "frac_reward_zero_std": 0.5, "grad_norm": 17.874990463256836, "kl": 0.8161603510379791, "learning_rate": 9.024171002408506e-07, "loss": -0.0302, "num_tokens": 3127980.0, "reward": 0.1875, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 290 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1164, "frac_reward_zero_std": 0.75, "grad_norm": 9.15800952911377, "kl": 0.6517329961061478, "learning_rate": 8.954715367323466e-07, "loss": 0.594, "num_tokens": 3136860.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 291 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1168, "frac_reward_zero_std": 1.0, "grad_norm": 15.840858459472656, "kl": 0.6931683421134949, "learning_rate": 8.885310677936746e-07, "loss": -0.3722, "num_tokens": 3145740.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 292 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1172, "frac_reward_zero_std": 0.25, "grad_norm": 15.857455253601074, "kl": 0.5569572299718857, "learning_rate": 8.81596031693499e-07, "loss": 0.2187, "num_tokens": 3154620.0, "reward": 0.4375, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 293 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1176, "frac_reward_zero_std": 0.5, "grad_norm": 14.625517845153809, "kl": 0.7273398488759995, "learning_rate": 8.746667664356955e-07, "loss": -0.6465, "num_tokens": 3163500.0, "reward": 0.5, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 294 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.118, "frac_reward_zero_std": 0.25, "grad_norm": 225.3708038330078, "kl": 0.7756715416908264, "learning_rate": 8.677436097428774e-07, "loss": 0.0031, "num_tokens": 3172380.0, "reward": 0.6875, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 295 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1184, "frac_reward_zero_std": 0.0, "grad_norm": 24.645357131958008, "kl": 0.7122261449694633, "learning_rate": 8.608268990399348e-07, "loss": 0.1858, "num_tokens": 3181260.0, "reward": 0.5, "reward_std": 0.5, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 296 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1188, "frac_reward_zero_std": 0.75, "grad_norm": 12.578080177307129, "kl": 0.8148588389158249, "learning_rate": 8.539169714375885e-07, "loss": 0.4697, "num_tokens": 3190140.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 297 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1192, "frac_reward_zero_std": 0.5, "grad_norm": 11.741436004638672, "kl": 0.7392874956130981, "learning_rate": 8.47014163715962e-07, "loss": -0.3385, "num_tokens": 3199020.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 298 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1196, "frac_reward_zero_std": 0.25, "grad_norm": 21.388893127441406, "kl": 0.8566679060459137, "learning_rate": 8.401188123081652e-07, "loss": 0.2199, "num_tokens": 3207900.0, "reward": 0.4375, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 299 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.12, "frac_reward_zero_std": 0.75, "grad_norm": 13.048949241638184, "kl": 0.8310280442237854, "learning_rate": 8.332312532838978e-07, "loss": -0.3381, "num_tokens": 3216780.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 300 }, { "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.12, "eval_loss": 0.002101506805047393, "eval_runtime": 7435.4156, "eval_samples_per_second": 0.027, "eval_steps_per_second": 0.007, "frac_reward_zero_std": 0.68, "num_tokens": 3327780.0, "reward": 0.42, "reward_std": 0.16618802070617675, "rewards/compute_reward/mean": 0.42, "rewards/compute_reward/std": 0.16618802070617675, "step": 300 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1204, "frac_reward_zero_std": 0.75, "grad_norm": 9.166318893432617, "kl": 0.6840480268001556, "learning_rate": 8.263518223330696e-07, "loss": 0.2527, "num_tokens": 3336660.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 301 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1208, "frac_reward_zero_std": 0.5, "grad_norm": 19.465709686279297, "kl": 0.743310958147049, "learning_rate": 8.194808547494401e-07, "loss": -0.0305, "num_tokens": 3345540.0, "reward": 0.75, "reward_std": 0.28867512941360474, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 302 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1212, "frac_reward_zero_std": 0.5, "grad_norm": 20.55376625061035, "kl": 1.060431182384491, "learning_rate": 8.126186854142751e-07, "loss": 0.2877, "num_tokens": 3354420.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 303 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1216, "frac_reward_zero_std": 1.0, "grad_norm": 7.348158359527588, "kl": 0.790182501077652, "learning_rate": 8.057656487800282e-07, "loss": -0.4967, "num_tokens": 3363300.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 304 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.122, "frac_reward_zero_std": 0.25, "grad_norm": 15.049077033996582, "kl": 0.6799326837062836, "learning_rate": 7.989220788540355e-07, "loss": -0.1223, "num_tokens": 3372180.0, "reward": 0.4375, "reward_std": 0.375, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 305 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1224, "frac_reward_zero_std": 0.25, "grad_norm": 22.41951560974121, "kl": 0.8887222111225128, "learning_rate": 7.920883091822408e-07, "loss": 0.7199, "num_tokens": 3381060.0, "reward": 0.5625, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 306 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1228, "frac_reward_zero_std": 0.75, "grad_norm": 19.698179244995117, "kl": 0.6762166917324066, "learning_rate": 7.852646728329367e-07, "loss": -0.5887, "num_tokens": 3389940.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 307 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1232, "frac_reward_zero_std": 0.75, "grad_norm": 16.77066993713379, "kl": 0.8449449837207794, "learning_rate": 7.784515023805327e-07, "loss": -0.1216, "num_tokens": 3398820.0, "reward": 0.6875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 308 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1236, "frac_reward_zero_std": 0.75, "grad_norm": 20.008758544921875, "kl": 0.9211454391479492, "learning_rate": 7.716491298893441e-07, "loss": 0.1287, "num_tokens": 3407700.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 309 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.124, "frac_reward_zero_std": 0.5, "grad_norm": 8.361716270446777, "kl": 0.8212382197380066, "learning_rate": 7.648578868974099e-07, "loss": -0.1217, "num_tokens": 3416580.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 310 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1244, "frac_reward_zero_std": 0.75, "grad_norm": 17.695589065551758, "kl": 0.9585025310516357, "learning_rate": 7.580781044003324e-07, "loss": 0.2538, "num_tokens": 3425460.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 311 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1248, "frac_reward_zero_std": 0.75, "grad_norm": 16.259077072143555, "kl": 0.6005022376775742, "learning_rate": 7.513101128351453e-07, "loss": -0.339, "num_tokens": 3434340.0, "reward": 0.125, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3415650427341461, "step": 312 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1252, "frac_reward_zero_std": 0.25, "grad_norm": 15.013197898864746, "kl": 0.7894881889224052, "learning_rate": 7.445542420642096e-07, "loss": 0.3446, "num_tokens": 3443220.0, "reward": 0.5625, "reward_std": 0.375, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 313 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1256, "frac_reward_zero_std": 0.75, "grad_norm": 21.608163833618164, "kl": 0.6923844665288925, "learning_rate": 7.378108213591354e-07, "loss": -0.3722, "num_tokens": 3452100.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 314 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.126, "frac_reward_zero_std": 0.75, "grad_norm": 6.553980827331543, "kl": 0.6568052619695663, "learning_rate": 7.310801793847343e-07, "loss": 0.6855, "num_tokens": 3460980.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 315 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1264, "frac_reward_zero_std": 0.75, "grad_norm": 16.87816619873047, "kl": 0.6174427717924118, "learning_rate": 7.243626441830009e-07, "loss": -0.1805, "num_tokens": 3469860.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 316 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1268, "frac_reward_zero_std": 0.75, "grad_norm": 7.426641464233398, "kl": 0.783763200044632, "learning_rate": 7.176585431571233e-07, "loss": -0.4968, "num_tokens": 3478740.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 317 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1272, "frac_reward_zero_std": 0.75, "grad_norm": 12.207983016967773, "kl": 0.5200645849108696, "learning_rate": 7.109682030555282e-07, "loss": 0.252, "num_tokens": 3487620.0, "reward": 0.375, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 318 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1276, "frac_reward_zero_std": 0.25, "grad_norm": 12.933839797973633, "kl": 0.485185407102108, "learning_rate": 7.042919499559536e-07, "loss": 0.0934, "num_tokens": 3496500.0, "reward": 0.6875, "reward_std": 0.41367512941360474, "rewards/compute_reward/mean": 0.6875, "rewards/compute_reward/std": 0.4787135720252991, "step": 319 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.128, "frac_reward_zero_std": 1.0, "grad_norm": 19.10908317565918, "kl": 0.4806456193327904, "learning_rate": 6.976301092495556e-07, "loss": -0.0896, "num_tokens": 3505380.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 320 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1284, "frac_reward_zero_std": 0.25, "grad_norm": 15.49258804321289, "kl": 0.4671884775161743, "learning_rate": 6.909830056250526e-07, "loss": 0.5933, "num_tokens": 3514260.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 321 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1288, "frac_reward_zero_std": 0.75, "grad_norm": 16.881296157836914, "kl": 0.5447326600551605, "learning_rate": 6.843509630528976e-07, "loss": -0.3393, "num_tokens": 3523140.0, "reward": 0.0625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.0625, "rewards/compute_reward/std": 0.25, "step": 322 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1292, "frac_reward_zero_std": 0.5, "grad_norm": 15.101067543029785, "kl": 0.36705372855067253, "learning_rate": 6.77734304769489e-07, "loss": -0.2485, "num_tokens": 3532020.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 323 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1296, "frac_reward_zero_std": 0.5, "grad_norm": 16.100318908691406, "kl": 0.37945468723773956, "learning_rate": 6.711333532614167e-07, "loss": -0.3734, "num_tokens": 3540900.0, "reward": 0.25, "reward_std": 0.25, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 324 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.13, "frac_reward_zero_std": 0.75, "grad_norm": 10.508184432983398, "kl": 0.27911943942308426, "learning_rate": 6.645484302497451e-07, "loss": 0.751, "num_tokens": 3549780.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 325 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1304, "frac_reward_zero_std": 0.75, "grad_norm": 7.249373435974121, "kl": 0.3483077771961689, "learning_rate": 6.579798566743313e-07, "loss": -0.6235, "num_tokens": 3558660.0, "reward": 0.4375, "reward_std": 0.125, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 326 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1308, "frac_reward_zero_std": 0.25, "grad_norm": 19.470088958740234, "kl": 0.3275032192468643, "learning_rate": 6.51427952678185e-07, "loss": 0.2178, "num_tokens": 3567540.0, "reward": 0.625, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 327 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1312, "frac_reward_zero_std": 0.75, "grad_norm": 20.5013427734375, "kl": 0.42966996133327484, "learning_rate": 6.448930375918631e-07, "loss": 0.0352, "num_tokens": 3576420.0, "reward": 0.625, "reward_std": 0.14433756470680237, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 328 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1316, "frac_reward_zero_std": 0.5, "grad_norm": 20.570777893066406, "kl": 0.5069122761487961, "learning_rate": 6.383754299179078e-07, "loss": 0.252, "num_tokens": 3585300.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 329 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.132, "frac_reward_zero_std": 0.5, "grad_norm": 16.90059471130371, "kl": 0.3699001967906952, "learning_rate": 6.31875447315322e-07, "loss": -0.3734, "num_tokens": 3594180.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 330 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1324, "frac_reward_zero_std": 1.0, "grad_norm": 14.63191032409668, "kl": 0.41268204152584076, "learning_rate": 6.253934065840879e-07, "loss": 0.1266, "num_tokens": 3603060.0, "reward": 0.25, "reward_std": 0.0, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.44721361994743347, "step": 331 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1328, "frac_reward_zero_std": 0.5, "grad_norm": 12.684980392456055, "kl": 0.39383379369974136, "learning_rate": 6.189296236497259e-07, "loss": -0.1234, "num_tokens": 3611940.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 332 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1332, "frac_reward_zero_std": 0.5, "grad_norm": 19.5969181060791, "kl": 0.5324902758002281, "learning_rate": 6.124844135478971e-07, "loss": 0.0021, "num_tokens": 3620820.0, "reward": 0.375, "reward_std": 0.25, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5, "step": 333 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1336, "frac_reward_zero_std": 0.5, "grad_norm": 13.67305850982666, "kl": 0.3931396156549454, "learning_rate": 6.060580904090489e-07, "loss": 0.0016, "num_tokens": 3629700.0, "reward": 0.875, "reward_std": 0.25, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3415650427341461, "step": 334 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.134, "frac_reward_zero_std": 0.75, "grad_norm": 9.632564544677734, "kl": 0.4459044635295868, "learning_rate": 5.996509674431051e-07, "loss": 0.2517, "num_tokens": 3638580.0, "reward": 0.5625, "reward_std": 0.125, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 335 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1344, "frac_reward_zero_std": 0.5, "grad_norm": 25.69095230102539, "kl": 0.4609299376606941, "learning_rate": 5.932633569241999e-07, "loss": -0.3731, "num_tokens": 3647460.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 336 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1348, "frac_reward_zero_std": 0.25, "grad_norm": 19.697153091430664, "kl": 0.47081001847982407, "learning_rate": 5.868955701754583e-07, "loss": 0.4683, "num_tokens": 3656340.0, "reward": 0.75, "reward_std": 0.39433756470680237, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.44721361994743347, "step": 337 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1352, "frac_reward_zero_std": 0.75, "grad_norm": 26.496923446655273, "kl": 0.5724581703543663, "learning_rate": 5.805479175538228e-07, "loss": -0.0892, "num_tokens": 3665220.0, "reward": 0.1875, "reward_std": 0.125, "rewards/compute_reward/mean": 0.1875, "rewards/compute_reward/std": 0.40311288833618164, "step": 338 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1356, "frac_reward_zero_std": 0.5, "grad_norm": 9.314996719360352, "kl": 0.36011531203985214, "learning_rate": 5.742207084349273e-07, "loss": -0.3735, "num_tokens": 3674100.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 339 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.136, "frac_reward_zero_std": 0.25, "grad_norm": 23.731319427490234, "kl": 0.8630180954933167, "learning_rate": 5.679142511980175e-07, "loss": -0.3715, "num_tokens": 3682980.0, "reward": 0.3125, "reward_std": 0.375, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 340 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1364, "frac_reward_zero_std": 0.75, "grad_norm": 15.965517044067383, "kl": 0.5544756576418877, "learning_rate": 5.616288532109224e-07, "loss": 0.3771, "num_tokens": 3691860.0, "reward": 0.8125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.8125, "rewards/compute_reward/std": 0.40311288833618164, "step": 341 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1368, "frac_reward_zero_std": 0.5, "grad_norm": 8.894350051879883, "kl": 0.4514187201857567, "learning_rate": 5.553648208150728e-07, "loss": -0.2481, "num_tokens": 3700740.0, "reward": 0.625, "reward_std": 0.25, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5, "step": 342 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1372, "frac_reward_zero_std": 0.5, "grad_norm": 28.423559188842773, "kl": 0.4208730310201645, "learning_rate": 5.491224593105694e-07, "loss": 0.3766, "num_tokens": 3709620.0, "reward": 0.5, "reward_std": 0.25, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 343 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1376, "frac_reward_zero_std": 0.5, "grad_norm": 8.600762367248535, "kl": 0.5391575843095779, "learning_rate": 5.42902072941306e-07, "loss": 0.2856, "num_tokens": 3718500.0, "reward": 0.4375, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.4375, "rewards/compute_reward/std": 0.5123475790023804, "step": 344 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.138, "frac_reward_zero_std": 1.0, "grad_norm": 13.370346069335938, "kl": 0.6912270113825798, "learning_rate": 5.367039648801385e-07, "loss": -0.1557, "num_tokens": 3727380.0, "reward": 0.5, "reward_std": 0.0, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 345 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1384, "frac_reward_zero_std": 0.5, "grad_norm": 13.797508239746094, "kl": 0.5032974556088448, "learning_rate": 5.305284372141095e-07, "loss": 0.252, "num_tokens": 3736260.0, "reward": 0.5625, "reward_std": 0.26933756470680237, "rewards/compute_reward/mean": 0.5625, "rewards/compute_reward/std": 0.5123475790023804, "step": 346 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1388, "frac_reward_zero_std": 0.75, "grad_norm": 13.878506660461426, "kl": 0.4989786595106125, "learning_rate": 5.243757909297246e-07, "loss": 0.127, "num_tokens": 3745140.0, "reward": 0.3125, "reward_std": 0.125, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 347 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1392, "frac_reward_zero_std": 0.25, "grad_norm": 14.72158145904541, "kl": 0.5256261751055717, "learning_rate": 5.182463258982846e-07, "loss": 0.0021, "num_tokens": 3754020.0, "reward": 0.3125, "reward_std": 0.375, "rewards/compute_reward/mean": 0.3125, "rewards/compute_reward/std": 0.4787135720252991, "step": 348 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.1396, "frac_reward_zero_std": 0.0, "grad_norm": 19.358739852905273, "kl": 0.5609803348779678, "learning_rate": 5.121403408612671e-07, "loss": -0.1227, "num_tokens": 3762900.0, "reward": 0.5, "reward_std": 0.5, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5163977742195129, "step": 349 }, { "completion_length": 512.0, "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.14, "frac_reward_zero_std": 0.25, "grad_norm": 29.80731773376465, "kl": 0.6624534577131271, "learning_rate": 5.060581334157692e-07, "loss": 0.1276, "num_tokens": 3771780.0, "reward": 0.8125, "reward_std": 0.375, "rewards/compute_reward/mean": 0.8125, "rewards/compute_reward/std": 0.40311288833618164, "step": 350 }, { "completions/clipped_ratio": 1.0, "completions/max_length": 512.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 512.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 512.0, "completions/min_terminated_length": 0.0, "epoch": 0.14, "eval_loss": -0.008800813928246498, "eval_runtime": 7087.5567, "eval_samples_per_second": 0.028, "eval_steps_per_second": 0.007, "frac_reward_zero_std": 0.64, "num_tokens": 3882780.0, "reward": 0.445, "reward_std": 0.18773502588272095, "rewards/compute_reward/mean": 0.445, "rewards/compute_reward/std": 0.18773502588272095, "step": 350 } ], "logging_steps": 1, "max_steps": 500, "num_input_tokens_seen": 3882780, "num_train_epochs": 1, "save_steps": 50, "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 }