{ "machine": "NVIDIA DGX Spark GB10, 128 GB unified memory", "image": "radixark/miles:latest", "model": "Qwen/Qwen3-0.6B unless the run name says otherwise", "dataset": "zhuzilin/gsm8k", "timing_rule": "median_unpaused keeps only steps whose window overlaps no heat-guard pause. The rollout window ends at the rollout perf line and lasts rollout_time_s. The step window ends at the trainer perf line and lasts step_time_s.", "pauses": 1005, "excluded": { "true-on-policy": 1, "true-on-policy-fa3": 137, "e4-4b": 1, "e4-4b-lowmem": 1 }, "partial": [ "e2-n16", "e2b-dapo-filter" ], "runs": { "plain": { "steps": 12, "partial": false, "median": { "reward": 0.660156, "length": 693.33789, "cut_share": 0.296875, "reward_all_samples": 0.660156, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015376, "train_rollout_kl": 0.000705, "entropy": 0.0, "grad_norm": 0.434395, "clip_fraction": 0.0, "step_time_s": 167.258879, "rollout_time_s": 57.189112, "log_probs_time_s": 25.131595, "train_time_s": 105.495381, "update_weights_time_s": 0.647092, "sleep_time_s": 0.760047, "wake_up_time_s": 0.852631 }, "median_unpaused": { "step_time_s": { "median": 167.258879, "steps": 12 }, "rollout_time_s": { "median": 57.189112, "steps": 12 }, "log_probs_time_s": { "median": 25.131595, "steps": 12 }, "train_time_s": { "median": 105.495381, "steps": 12 }, "update_weights_time_s": { "median": 0.647092, "steps": 12 }, "sleep_time_s": { "median": 0.760047, "steps": 12 }, "wake_up_time_s": { "median": 0.852631, "steps": 12 } }, "per_step": [ { "step": 0, "length": 739.925781, "reward_all_samples": 0.628906, "groups_all_wrong": 3.0, "groups_all_right": 7.0, "rollout_time_s": 61.951834, "rollout_end": "2026-10-17 15:56:15", "reward": 0.628906, "cut_share": 0.320312, "logprob_abs_diff": 0.015336, "train_rollout_kl": 0.000684, "entropy": 0.0, "grad_norm": 0.521657, "clip_fraction": 0.0, "step_time_s": 233.294943, "log_probs_time_s": 31.453812, "train_time_s": 118.988866, "update_weights_time_s": 0.664753, "sleep_time_s": 2.111608, "wake_up_time_s": 0.46198, "step_end": "2026-10-17 15:58:15", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 1, "length": 768.90625, "reward_all_samples": 0.59375, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 65.475583, "rollout_end": "2026-10-17 15:59:27", "reward": 0.59375, "cut_share": 0.402344, "logprob_abs_diff": 0.016102, "train_rollout_kl": 0.000736, "entropy": 0.0, "grad_norm": 0.537533, "clip_fraction": 0.0, "step_time_s": 183.154475, "log_probs_time_s": 26.691365, "train_time_s": 110.04636, "update_weights_time_s": 0.641284, "sleep_time_s": 4.024037, "wake_up_time_s": 0.851914, "step_end": "2026-10-17 16:01:18", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 2, "length": 707.6875, "reward_all_samples": 0.757812, "groups_all_wrong": 2.0, "groups_all_right": 17.0, "rollout_time_s": 57.127128, "rollout_end": "2026-10-17 16:02:19", "reward": 0.757812, "cut_share": 0.234375, "logprob_abs_diff": 0.014313, "train_rollout_kl": 0.000648, "entropy": 0.0, "grad_norm": 0.365482, "clip_fraction": 0.0, "step_time_s": 167.868097, "log_probs_time_s": 25.260227, "train_time_s": 106.018326, "update_weights_time_s": 0.627722, "sleep_time_s": 1.126044, "wake_up_time_s": 0.759332, "step_end": "2026-10-17 16:04:06", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 3, "length": 683.726562, "reward_all_samples": 0.691406, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 56.082473, "rollout_end": "2026-10-17 16:05:06", "reward": 0.691406, "cut_share": 0.289062, "logprob_abs_diff": 0.015629, "train_rollout_kl": 0.000708, "entropy": 0.0, "grad_norm": 0.426275, "clip_fraction": 0.0, "step_time_s": 165.518385, "log_probs_time_s": 24.836356, "train_time_s": 104.625565, "update_weights_time_s": 0.624444, "sleep_time_s": 1.180077, "wake_up_time_s": 0.867438, "step_end": "2026-10-17 16:06:52", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 4, "length": 710.757812, "reward_all_samples": 0.589844, "groups_all_wrong": 7.0, "groups_all_right": 11.0, "rollout_time_s": 58.594705, "rollout_end": "2026-10-17 16:07:53", "reward": 0.589844, "cut_share": 0.304688, "logprob_abs_diff": 0.015349, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.420429, "clip_fraction": 0.0, "step_time_s": 169.089762, "log_probs_time_s": 25.277187, "train_time_s": 106.132133, "update_weights_time_s": 0.624263, "sleep_time_s": 0.730979, "wake_up_time_s": 0.7482, "step_end": "2026-10-17 16:09:41", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 5, "length": 711.59375, "reward_all_samples": 0.585938, "groups_all_wrong": 5.0, "groups_all_right": 11.0, "rollout_time_s": 59.594722, "rollout_end": "2026-10-17 16:10:43", "reward": 0.585938, "cut_share": 0.355469, "logprob_abs_diff": 0.016285, "train_rollout_kl": 0.000735, "entropy": 0.0, "grad_norm": 0.439617, "clip_fraction": 0.0, "step_time_s": 171.717234, "log_probs_time_s": 25.616835, "train_time_s": 107.587679, "update_weights_time_s": 0.6529, "sleep_time_s": 0.754161, "wake_up_time_s": 0.853348, "step_end": "2026-10-17 16:12:33", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 6, "length": 646.304688, "reward_all_samples": 0.6875, "groups_all_wrong": 4.0, "groups_all_right": 13.0, "rollout_time_s": 51.693421, "rollout_end": "2026-10-17 16:13:27", "reward": 0.6875, "cut_share": 0.21875, "logprob_abs_diff": 0.014716, "train_rollout_kl": 0.000674, "entropy": 0.0, "grad_norm": 0.385415, "clip_fraction": 0.0, "step_time_s": 158.198263, "log_probs_time_s": 23.921661, "train_time_s": 102.17084, "update_weights_time_s": 0.634809, "sleep_time_s": 0.725214, "wake_up_time_s": 0.819497, "step_end": "2026-10-17 16:15:11", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 7, "length": 615.285156, "reward_all_samples": 0.675781, "groups_all_wrong": 6.0, "groups_all_right": 14.0, "rollout_time_s": 49.270471, "rollout_end": "2026-10-17 16:16:03", "reward": 0.675781, "cut_share": 0.238281, "logprob_abs_diff": 0.015263, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.429172, "clip_fraction": 0.0, "step_time_s": 154.115779, "log_probs_time_s": 23.305815, "train_time_s": 100.443306, "update_weights_time_s": 0.654226, "sleep_time_s": 0.761873, "wake_up_time_s": 0.872319, "step_end": "2026-10-17 16:17:45", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 8, "length": 595.910156, "reward_all_samples": 0.746094, "groups_all_wrong": 4.0, "groups_all_right": 15.0, "rollout_time_s": 46.150371, "rollout_end": "2026-10-17 16:18:34", "reward": 0.746094, "cut_share": 0.171875, "logprob_abs_diff": 0.015133, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.40925, "clip_fraction": 0.0, "step_time_s": 148.907465, "log_probs_time_s": 22.685077, "train_time_s": 98.365629, "update_weights_time_s": 0.663219, "sleep_time_s": 0.758222, "wake_up_time_s": 0.857923, "step_end": "2026-10-17 16:20:14", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 9, "length": 695.355469, "reward_all_samples": 0.648438, "groups_all_wrong": 2.0, "groups_all_right": 6.0, "rollout_time_s": 57.436637, "rollout_end": "2026-10-17 16:21:14", "reward": 0.648438, "cut_share": 0.316406, "logprob_abs_diff": 0.016122, "train_rollout_kl": 0.000755, "entropy": 0.0, "grad_norm": 0.538225, "clip_fraction": 0.0, "step_time_s": 168.172353, "log_probs_time_s": 25.182564, "train_time_s": 106.328313, "update_weights_time_s": 0.658605, "sleep_time_s": 0.768291, "wake_up_time_s": 0.856969, "step_end": "2026-10-17 16:23:02", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 10, "length": 691.320312, "reward_all_samples": 0.65625, "groups_all_wrong": 5.0, "groups_all_right": 10.0, "rollout_time_s": 57.251096, "rollout_end": "2026-10-17 16:24:02", "reward": 0.65625, "cut_share": 0.316406, "logprob_abs_diff": 0.015403, "train_rollout_kl": 0.000707, "entropy": 0.0, "grad_norm": 0.508392, "clip_fraction": 0.0, "step_time_s": 166.64966, "log_probs_time_s": 25.080626, "train_time_s": 104.972435, "update_weights_time_s": 0.671099, "sleep_time_s": 0.75449, "wake_up_time_s": 0.869218, "step_end": "2026-10-17 16:25:49", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 11, "length": 632.492188, "reward_all_samples": 0.664062, "groups_all_wrong": 2.0, "groups_all_right": 10.0, "rollout_time_s": 51.114784, "rollout_end": "2026-10-17 16:26:43", "reward": 0.664062, "cut_share": 0.210938, "logprob_abs_diff": 0.01557, "train_rollout_kl": 0.000723, "entropy": 0.0, "grad_norm": 0.527859, "clip_fraction": 0.0, "step_time_s": 157.212977, "log_probs_time_s": 23.603658, "train_time_s": 101.718505, "update_weights_time_s": 0.632146, "sleep_time_s": 0.746159, "wake_up_time_s": 0.773254, "step_end": "2026-10-17 16:28:26", "rollout_paused_s": 0.0, "step_paused_s": 0.0 } ] }, "plain-48": { "steps": 48, "partial": false, "median": { "reward": 0.701172, "length": 599.765625, "cut_share": 0.183594, "reward_all_samples": 0.701172, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015093, "train_rollout_kl": 0.000698, "entropy": 0.0, "grad_norm": 0.494835, "clip_fraction": 0.0, "step_time_s": 152.643569, "rollout_time_s": 47.495245, "log_probs_time_s": 23.378027, "train_time_s": 100.290844, "update_weights_time_s": 0.638256, "sleep_time_s": 0.755802, "wake_up_time_s": 0.866168 }, "median_unpaused": { "step_time_s": { "median": 152.643569, "steps": 48 }, "rollout_time_s": { "median": 47.495245, "steps": 48 }, "log_probs_time_s": { "median": 23.378027, "steps": 48 }, "train_time_s": { "median": 100.290844, "steps": 48 }, "update_weights_time_s": { "median": 0.638256, "steps": 48 }, "sleep_time_s": { "median": 0.755802, "steps": 48 }, "wake_up_time_s": { "median": 0.866168, "steps": 48 } }, "per_step": [ { "step": 0, "length": 759.089844, "reward_all_samples": 0.582031, "groups_all_wrong": 4.0, "groups_all_right": 5.0, "rollout_time_s": 63.933147, "rollout_end": "2026-10-17 22:12:27", "reward": 0.582031, "cut_share": 0.347656, "logprob_abs_diff": 0.015669, "train_rollout_kl": 0.000709, "entropy": 0.0, "grad_norm": 0.521105, "clip_fraction": 0.0, "step_time_s": 241.163447, "log_probs_time_s": 31.416359, "train_time_s": 119.834962, "update_weights_time_s": 0.654977, "sleep_time_s": 1.919089, "wake_up_time_s": 0.415931, "step_end": "2026-10-17 22:14:28", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 1, "length": 771.066406, "reward_all_samples": 0.605469, "groups_all_wrong": 6.0, "groups_all_right": 11.0, "rollout_time_s": 65.763588, "rollout_end": "2026-10-17 22:15:40", "reward": 0.605469, "cut_share": 0.433594, "logprob_abs_diff": 0.015931, "train_rollout_kl": 0.000723, "entropy": 0.0, "grad_norm": 0.435982, "clip_fraction": 0.0, "step_time_s": 182.634498, "log_probs_time_s": 26.613997, "train_time_s": 109.538562, "update_weights_time_s": 0.609583, "sleep_time_s": 3.678752, "wake_up_time_s": 0.747428, "step_end": "2026-10-17 22:17:31", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 2, "length": 727.777344, "reward_all_samples": 0.726562, "groups_all_wrong": 4.0, "groups_all_right": 17.0, "rollout_time_s": 59.240092, "rollout_end": "2026-10-17 22:18:33", "reward": 0.726562, "cut_share": 0.238281, "logprob_abs_diff": 0.014127, "train_rollout_kl": 0.000631, "entropy": 0.0, "grad_norm": 0.346428, "clip_fraction": 0.0, "step_time_s": 171.734572, "log_probs_time_s": 25.874666, "train_time_s": 107.778728, "update_weights_time_s": 0.616892, "sleep_time_s": 1.186008, "wake_up_time_s": 0.779257, "step_end": "2026-10-17 22:20:22", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 3, "length": 709.699219, "reward_all_samples": 0.65625, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "rollout_time_s": 59.173676, "rollout_end": "2026-10-17 22:21:25", "reward": 0.65625, "cut_share": 0.363281, "logprob_abs_diff": 0.015862, "train_rollout_kl": 0.000723, "entropy": 0.0, "grad_norm": 0.416383, "clip_fraction": 0.0, "step_time_s": 171.227611, "log_probs_time_s": 25.577043, "train_time_s": 107.255066, "update_weights_time_s": 0.653515, "sleep_time_s": 1.163634, "wake_up_time_s": 0.86175, "step_end": "2026-10-17 22:23:14", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 4, "length": 718.226562, "reward_all_samples": 0.585938, "groups_all_wrong": 5.0, "groups_all_right": 10.0, "rollout_time_s": 58.979632, "rollout_end": "2026-10-17 22:24:16", "reward": 0.585938, "cut_share": 0.308594, "logprob_abs_diff": 0.014987, "train_rollout_kl": 0.000683, "entropy": 0.0, "grad_norm": 0.49931, "clip_fraction": 0.0, "step_time_s": 170.798829, "log_probs_time_s": 25.7429, "train_time_s": 107.47229, "update_weights_time_s": 0.618515, "sleep_time_s": 0.74057, "wake_up_time_s": 0.86466, "step_end": "2026-10-17 22:26:05", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 5, "length": 770.828125, "reward_all_samples": 0.558594, "groups_all_wrong": 6.0, "groups_all_right": 11.0, "rollout_time_s": 65.651539, "rollout_end": "2026-10-17 22:27:13", "reward": 0.558594, "cut_share": 0.410156, "logprob_abs_diff": 0.016161, "train_rollout_kl": 0.000717, "entropy": 0.0, "grad_norm": 0.401443, "clip_fraction": 0.0, "step_time_s": 182.30982, "log_probs_time_s": 27.279684, "train_time_s": 112.173221, "update_weights_time_s": 0.618493, "sleep_time_s": 0.736472, "wake_up_time_s": 0.86803, "step_end": "2026-10-17 22:29:07", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 6, "length": 673.179688, "reward_all_samples": 0.699219, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 54.313323, "rollout_end": "2026-10-17 22:30:04", "reward": 0.699219, "cut_share": 0.242188, "logprob_abs_diff": 0.014528, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.428595, "clip_fraction": 0.0, "step_time_s": 162.728847, "log_probs_time_s": 24.480756, "train_time_s": 103.934836, "update_weights_time_s": 0.613611, "sleep_time_s": 0.711622, "wake_up_time_s": 0.875444, "step_end": "2026-10-17 22:31:50", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 7, "length": 663.15625, "reward_all_samples": 0.675781, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 54.359556, "rollout_end": "2026-10-17 22:32:47", "reward": 0.675781, "cut_share": 0.265625, "logprob_abs_diff": 0.015301, "train_rollout_kl": 0.000708, "entropy": 0.0, "grad_norm": 0.513866, "clip_fraction": 0.0, "step_time_s": 162.968687, "log_probs_time_s": 24.43449, "train_time_s": 104.252596, "update_weights_time_s": 0.621659, "sleep_time_s": 0.730635, "wake_up_time_s": 0.755906, "step_end": "2026-10-17 22:34:33", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 8, "length": 626.171875, "reward_all_samples": 0.734375, "groups_all_wrong": 3.0, "groups_all_right": 15.0, "rollout_time_s": 49.095505, "rollout_end": "2026-10-17 22:35:24", "reward": 0.734375, "cut_share": 0.1875, "logprob_abs_diff": 0.014684, "train_rollout_kl": 0.000679, "entropy": 0.0, "grad_norm": 0.423416, "clip_fraction": 0.0, "step_time_s": 154.884985, "log_probs_time_s": 23.717284, "train_time_s": 101.412971, "update_weights_time_s": 0.647311, "sleep_time_s": 0.74533, "wake_up_time_s": 0.860148, "step_end": "2026-10-17 22:37:07", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 9, "length": 750.3125, "reward_all_samples": 0.636719, "groups_all_wrong": 3.0, "groups_all_right": 10.0, "rollout_time_s": 62.934368, "rollout_end": "2026-10-17 22:38:13", "reward": 0.636719, "cut_share": 0.359375, "logprob_abs_diff": 0.015498, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.428834, "clip_fraction": 0.0, "step_time_s": 177.996054, "log_probs_time_s": 26.75367, "train_time_s": 110.666249, "update_weights_time_s": 0.618735, "sleep_time_s": 0.723179, "wake_up_time_s": 0.763644, "step_end": "2026-10-17 22:40:05", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 10, "length": 742.851562, "reward_all_samples": 0.589844, "groups_all_wrong": 5.0, "groups_all_right": 8.0, "rollout_time_s": 62.600293, "rollout_end": "2026-10-17 22:41:11", "reward": 0.589844, "cut_share": 0.394531, "logprob_abs_diff": 0.015224, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.46757, "clip_fraction": 0.0, "step_time_s": 176.089982, "log_probs_time_s": 26.14628, "train_time_s": 109.024632, "update_weights_time_s": 0.657887, "sleep_time_s": 0.781956, "wake_up_time_s": 0.767804, "step_end": "2026-10-17 22:43:02", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 11, "length": 685.140625, "reward_all_samples": 0.652344, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 56.547826, "rollout_end": "2026-10-17 22:44:01", "reward": 0.652344, "cut_share": 0.277344, "logprob_abs_diff": 0.015469, "train_rollout_kl": 0.000717, "entropy": 0.0, "grad_norm": 0.452638, "clip_fraction": 0.0, "step_time_s": 165.982822, "log_probs_time_s": 24.952832, "train_time_s": 105.056897, "update_weights_time_s": 0.662412, "sleep_time_s": 0.780191, "wake_up_time_s": 0.787591, "step_end": "2026-10-17 22:45:47", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 12, "length": 661.117188, "reward_all_samples": 0.703125, "groups_all_wrong": 5.0, "groups_all_right": 13.0, "rollout_time_s": 52.828497, "rollout_end": "2026-10-17 22:46:43", "reward": 0.703125, "cut_share": 0.199219, "logprob_abs_diff": 0.015438, "train_rollout_kl": 0.000729, "entropy": 0.0, "grad_norm": 0.414709, "clip_fraction": 0.0, "step_time_s": 161.337155, "log_probs_time_s": 24.449404, "train_time_s": 104.055533, "update_weights_time_s": 0.64104, "sleep_time_s": 0.748105, "wake_up_time_s": 0.763542, "step_end": "2026-10-17 22:48:29", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 13, "length": 660.941406, "reward_all_samples": 0.726562, "groups_all_wrong": 2.0, "groups_all_right": 12.0, "rollout_time_s": 53.621578, "rollout_end": "2026-10-17 22:49:25", "reward": 0.726562, "cut_share": 0.222656, "logprob_abs_diff": 0.01475, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.452414, "clip_fraction": 0.0, "step_time_s": 161.85894, "log_probs_time_s": 24.388616, "train_time_s": 103.863585, "update_weights_time_s": 0.619815, "sleep_time_s": 0.735838, "wake_up_time_s": 0.768099, "step_end": "2026-10-17 22:51:11", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 14, "length": 726.292969, "reward_all_samples": 0.574219, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 61.094125, "rollout_end": "2026-10-17 22:52:15", "reward": 0.574219, "cut_share": 0.347656, "logprob_abs_diff": 0.016039, "train_rollout_kl": 0.000726, "entropy": 0.0, "grad_norm": 0.468297, "clip_fraction": 0.0, "step_time_s": 175.235298, "log_probs_time_s": 26.266796, "train_time_s": 109.57235, "update_weights_time_s": 0.653435, "sleep_time_s": 0.741664, "wake_up_time_s": 0.876988, "step_end": "2026-10-17 22:54:06", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 15, "length": 703.613281, "reward_all_samples": 0.609375, "groups_all_wrong": 4.0, "groups_all_right": 7.0, "rollout_time_s": 57.864722, "rollout_end": "2026-10-17 22:55:07", "reward": 0.609375, "cut_share": 0.273438, "logprob_abs_diff": 0.015107, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.519926, "clip_fraction": 0.0, "step_time_s": 169.372082, "log_probs_time_s": 25.417497, "train_time_s": 107.07354, "update_weights_time_s": 0.64528, "sleep_time_s": 0.744636, "wake_up_time_s": 0.77484, "step_end": "2026-10-17 22:56:55", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 16, "length": 734.929688, "reward_all_samples": 0.683594, "groups_all_right": 9.0, "rollout_time_s": 61.466163, "rollout_end": "2026-10-17 22:58:00", "reward": 0.683594, "cut_share": 0.300781, "logprob_abs_diff": 0.014879, "train_rollout_kl": 0.000692, "entropy": 0.0, "grad_norm": 0.487082, "clip_fraction": 0.0, "step_time_s": 174.900411, "log_probs_time_s": 26.276538, "train_time_s": 109.072645, "update_weights_time_s": 0.635465, "sleep_time_s": 0.718632, "wake_up_time_s": 0.860724, "step_end": "2026-10-17 22:59:50", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 17, "length": 576.945312, "reward_all_samples": 0.753906, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 45.043581, "rollout_end": "2026-10-17 23:00:38", "reward": 0.753906, "cut_share": 0.152344, "logprob_abs_diff": 0.015164, "train_rollout_kl": 0.00072, "entropy": 0.0, "grad_norm": 4.95085, "clip_fraction": 0.0, "step_time_s": 149.637168, "log_probs_time_s": 23.535019, "train_time_s": 100.149402, "update_weights_time_s": 0.655431, "sleep_time_s": 0.777732, "wake_up_time_s": 0.867676, "step_end": "2026-10-17 23:02:20", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 18, "length": 637.914062, "reward_all_samples": 0.671875, "groups_all_wrong": 2.0, "groups_all_right": 13.0, "rollout_time_s": 51.048412, "rollout_end": "2026-10-17 23:03:14", "reward": 0.671875, "cut_share": 0.226562, "logprob_abs_diff": 0.01515, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.501309, "clip_fraction": 0.0, "step_time_s": 157.130587, "log_probs_time_s": 23.927302, "train_time_s": 101.737664, "update_weights_time_s": 0.639389, "sleep_time_s": 0.775304, "wake_up_time_s": 0.792448, "step_end": "2026-10-17 23:04:57", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 19, "length": 652.972656, "reward_all_samples": 0.648438, "groups_all_wrong": 5.0, "groups_all_right": 13.0, "rollout_time_s": 53.803748, "rollout_end": "2026-10-17 23:05:54", "reward": 0.648438, "cut_share": 0.304688, "logprob_abs_diff": 0.015863, "train_rollout_kl": 0.000722, "entropy": 0.0, "grad_norm": 0.423897, "clip_fraction": 0.0, "step_time_s": 160.651578, "log_probs_time_s": 24.141524, "train_time_s": 102.378346, "update_weights_time_s": 0.650424, "sleep_time_s": 0.771141, "wake_up_time_s": 0.783426, "step_end": "2026-10-17 23:07:38", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 20, "length": 632.652344, "reward_all_samples": 0.710938, "groups_all_wrong": 4.0, "groups_all_right": 14.0, "rollout_time_s": 50.648913, "rollout_end": "2026-10-17 23:08:31", "reward": 0.710938, "cut_share": 0.207031, "logprob_abs_diff": 0.015775, "train_rollout_kl": 0.000748, "entropy": 0.0, "grad_norm": 0.479666, "clip_fraction": 0.0, "step_time_s": 157.433007, "log_probs_time_s": 23.85542, "train_time_s": 102.345721, "update_weights_time_s": 0.629764, "sleep_time_s": 0.756885, "wake_up_time_s": 0.792949, "step_end": "2026-10-17 23:10:15", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 21, "length": 625.300781, "reward_all_samples": 0.714844, "groups_all_wrong": 4.0, "groups_all_right": 14.0, "rollout_time_s": 50.376753, "rollout_end": "2026-10-17 23:11:08", "reward": 0.714844, "cut_share": 0.242188, "logprob_abs_diff": 0.015375, "train_rollout_kl": 0.000709, "entropy": 0.0, "grad_norm": 0.434129, "clip_fraction": 0.0, "step_time_s": 156.437052, "log_probs_time_s": 23.626298, "train_time_s": 101.600018, "update_weights_time_s": 0.67342, "sleep_time_s": 0.765025, "wake_up_time_s": 0.780865, "step_end": "2026-10-17 23:12:51", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 22, "length": 619.246094, "reward_all_samples": 0.65625, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 49.624756, "rollout_end": "2026-10-17 23:13:44", "reward": 0.65625, "cut_share": 0.21875, "logprob_abs_diff": 0.015396, "train_rollout_kl": 0.000707, "entropy": 0.0, "grad_norm": 0.571647, "clip_fraction": 0.0, "step_time_s": 154.459084, "log_probs_time_s": 23.336255, "train_time_s": 100.472648, "update_weights_time_s": 0.66852, "sleep_time_s": 0.788619, "wake_up_time_s": 0.778213, "step_end": "2026-10-17 23:15:26", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 23, "length": 618.351562, "reward_all_samples": 0.648438, "groups_all_wrong": 5.0, "groups_all_right": 15.0, "rollout_time_s": 49.8512, "rollout_end": "2026-10-17 23:16:19", "reward": 0.648438, "cut_share": 0.238281, "logprob_abs_diff": 0.015208, "train_rollout_kl": 0.000694, "entropy": 0.0, "grad_norm": 0.379865, "clip_fraction": 0.0, "step_time_s": 154.461905, "log_probs_time_s": 23.419798, "train_time_s": 100.208475, "update_weights_time_s": 0.626758, "sleep_time_s": 0.752285, "wake_up_time_s": 0.7828, "step_end": "2026-10-17 23:18:00", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 24, "length": 563.601562, "reward_all_samples": 0.753906, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 43.232453, "rollout_end": "2026-10-17 23:18:46", "reward": 0.753906, "cut_share": 0.15625, "logprob_abs_diff": 0.014739, "train_rollout_kl": 0.000693, "entropy": 0.0, "grad_norm": 0.502584, "clip_fraction": 0.0, "step_time_s": 144.459837, "log_probs_time_s": 22.226727, "train_time_s": 96.873981, "update_weights_time_s": 0.627553, "sleep_time_s": 0.727639, "wake_up_time_s": 0.870261, "step_end": "2026-10-17 23:20:25", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 25, "length": 545.945312, "reward_all_samples": 0.726562, "groups_all_wrong": 2.0, "groups_all_right": 13.0, "rollout_time_s": 41.207089, "rollout_end": "2026-10-17 23:21:09", "reward": 0.726562, "cut_share": 0.136719, "logprob_abs_diff": 0.01529, "train_rollout_kl": 0.000709, "entropy": 0.0, "grad_norm": 0.556809, "clip_fraction": 0.0, "step_time_s": 141.120321, "log_probs_time_s": 21.7008, "train_time_s": 95.523877, "update_weights_time_s": 0.637123, "sleep_time_s": 0.732249, "wake_up_time_s": 0.894515, "step_end": "2026-10-17 23:22:46", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 26, "length": 569.09375, "reward_all_samples": 0.664062, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 44.253983, "rollout_end": "2026-10-17 23:23:33", "reward": 0.664062, "cut_share": 0.171875, "logprob_abs_diff": 0.0153, "train_rollout_kl": 0.000722, "entropy": 0.0, "grad_norm": 0.525462, "clip_fraction": 0.0, "step_time_s": 145.946806, "log_probs_time_s": 22.340999, "train_time_s": 97.349053, "update_weights_time_s": 0.634633, "sleep_time_s": 0.726883, "wake_up_time_s": 0.878984, "step_end": "2026-10-17 23:25:12", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 27, "length": 500.863281, "reward_all_samples": 0.722656, "groups_all_wrong": 3.0, "groups_all_right": 15.0, "rollout_time_s": 37.749536, "rollout_end": "2026-10-17 23:25:53", "reward": 0.722656, "cut_share": 0.113281, "logprob_abs_diff": 0.014995, "train_rollout_kl": 0.000698, "entropy": 0.0, "grad_norm": 0.467593, "clip_fraction": 0.0, "step_time_s": 134.801117, "log_probs_time_s": 20.668748, "train_time_s": 92.581847, "update_weights_time_s": 0.665272, "sleep_time_s": 0.78301, "wake_up_time_s": 0.881068, "step_end": "2026-10-17 23:27:27", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 28, "length": 574.519531, "reward_all_samples": 0.707031, "groups_all_wrong": 5.0, "groups_all_right": 12.0, "rollout_time_s": 44.872831, "rollout_end": "2026-10-17 23:28:14", "reward": 0.707031, "cut_share": 0.175781, "logprob_abs_diff": 0.015649, "train_rollout_kl": 0.000718, "entropy": 0.0, "grad_norm": 0.508063, "clip_fraction": 0.0, "step_time_s": 147.458905, "log_probs_time_s": 22.455785, "train_time_s": 98.153023, "update_weights_time_s": 0.669487, "sleep_time_s": 0.763916, "wake_up_time_s": 0.88771, "step_end": "2026-10-17 23:29:54", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 29, "length": 471.332031, "reward_all_samples": 0.765625, "groups_all_wrong": 3.0, "groups_all_right": 16.0, "rollout_time_s": 35.005819, "rollout_end": "2026-10-17 23:30:32", "reward": 0.765625, "cut_share": 0.070312, "logprob_abs_diff": 0.013833, "train_rollout_kl": 0.000671, "entropy": 0.0, "grad_norm": 0.462457, "clip_fraction": 0.0, "step_time_s": 130.213958, "log_probs_time_s": 20.090797, "train_time_s": 90.86109, "update_weights_time_s": 0.629316, "sleep_time_s": 0.745049, "wake_up_time_s": 0.877676, "step_end": "2026-10-17 23:32:04", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 30, "length": 526.019531, "reward_all_samples": 0.800781, "groups_all_wrong": 1.0, "groups_all_right": 18.0, "rollout_time_s": 40.341773, "rollout_end": "2026-10-17 23:32:48", "reward": 0.800781, "cut_share": 0.128906, "logprob_abs_diff": 0.014549, "train_rollout_kl": 0.000702, "entropy": 0.0, "grad_norm": 0.47933, "clip_fraction": 0.0, "step_time_s": 139.143008, "log_probs_time_s": 21.363278, "train_time_s": 94.285925, "update_weights_time_s": 0.646129, "sleep_time_s": 0.754718, "wake_up_time_s": 0.860566, "step_end": "2026-10-17 23:34:24", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 31, "length": 628.351562, "reward_all_samples": 0.59375, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 50.337991, "rollout_end": "2026-10-17 23:35:17", "reward": 0.59375, "cut_share": 0.21875, "logprob_abs_diff": 0.015587, "train_rollout_kl": 0.000733, "entropy": 0.0, "grad_norm": 0.490361, "clip_fraction": 0.0, "step_time_s": 156.402337, "log_probs_time_s": 23.710797, "train_time_s": 101.637285, "update_weights_time_s": 0.625421, "sleep_time_s": 0.743645, "wake_up_time_s": 0.775833, "step_end": "2026-10-17 23:37:00", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 32, "length": 581.179688, "reward_all_samples": 0.714844, "groups_all_wrong": 3.0, "groups_all_right": 11.0, "rollout_time_s": 45.894985, "rollout_end": "2026-10-17 23:37:49", "reward": 0.714844, "cut_share": 0.179688, "logprob_abs_diff": 0.014496, "train_rollout_kl": 0.000658, "entropy": 0.0, "grad_norm": 0.535862, "clip_fraction": 0.0, "step_time_s": 150.828053, "log_probs_time_s": 23.022187, "train_time_s": 100.373213, "update_weights_time_s": 0.667431, "sleep_time_s": 0.762484, "wake_up_time_s": 0.882698, "step_end": "2026-10-17 23:39:31", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 33, "length": 519.867188, "reward_all_samples": 0.707031, "groups_all_wrong": 3.0, "groups_all_right": 10.0, "rollout_time_s": 39.562394, "rollout_end": "2026-10-17 23:40:13", "reward": 0.707031, "cut_share": 0.128906, "logprob_abs_diff": 0.015362, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.58984, "clip_fraction": 0.0, "step_time_s": 138.307, "log_probs_time_s": 21.249908, "train_time_s": 94.241075, "update_weights_time_s": 0.629448, "sleep_time_s": 0.744975, "wake_up_time_s": 0.871281, "step_end": "2026-10-17 23:41:49", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 34, "length": 500.304688, "reward_all_samples": 0.675781, "groups_all_wrong": 4.0, "groups_all_right": 13.0, "rollout_time_s": 37.790307, "rollout_end": "2026-10-17 23:42:30", "reward": 0.675781, "cut_share": 0.117188, "logprob_abs_diff": 0.014537, "train_rollout_kl": 0.000674, "entropy": 0.0, "grad_norm": 0.599737, "clip_fraction": 0.0, "step_time_s": 135.683273, "log_probs_time_s": 20.860986, "train_time_s": 93.364834, "update_weights_time_s": 0.63374, "sleep_time_s": 0.743398, "wake_up_time_s": 0.883447, "step_end": "2026-10-17 23:44:05", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 35, "length": 525.792969, "reward_all_samples": 0.757812, "groups_all_wrong": 2.0, "groups_all_right": 13.0, "rollout_time_s": 39.786314, "rollout_end": "2026-10-17 23:44:47", "reward": 0.757812, "cut_share": 0.117188, "logprob_abs_diff": 0.015032, "train_rollout_kl": 0.000712, "entropy": 0.0, "grad_norm": 0.60635, "clip_fraction": 0.0, "step_time_s": 138.750453, "log_probs_time_s": 21.365617, "train_time_s": 94.556133, "update_weights_time_s": 0.626684, "sleep_time_s": 0.743434, "wake_up_time_s": 0.789839, "step_end": "2026-10-17 23:46:24", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 36, "length": 396.3125, "reward_all_samples": 0.871094, "groups_all_wrong": 1.0, "groups_all_right": 21.0, "rollout_time_s": 27.544213, "rollout_end": "2026-10-17 23:46:54", "reward": 0.871094, "cut_share": 0.015625, "logprob_abs_diff": 0.01448, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.501586, "clip_fraction": 0.0, "step_time_s": 116.747998, "log_probs_time_s": 18.257986, "train_time_s": 84.669865, "update_weights_time_s": 0.668365, "sleep_time_s": 0.761145, "wake_up_time_s": 0.869777, "step_end": "2026-10-17 23:48:20", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 37, "length": 539.265625, "reward_all_samples": 0.675781, "groups_all_wrong": 2.0, "groups_all_right": 14.0, "rollout_time_s": 40.821311, "rollout_end": "2026-10-17 23:49:04", "reward": 0.675781, "cut_share": 0.125, "logprob_abs_diff": 0.015078, "train_rollout_kl": 0.000693, "entropy": 0.0, "grad_norm": 0.486207, "clip_fraction": 0.0, "step_time_s": 140.5056, "log_probs_time_s": 21.493837, "train_time_s": 95.214937, "update_weights_time_s": 0.657029, "sleep_time_s": 0.768914, "wake_up_time_s": 0.776089, "step_end": "2026-10-17 23:50:41", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 38, "length": 478.722656, "reward_all_samples": 0.679688, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 36.484996, "rollout_end": "2026-10-17 23:51:20", "reward": 0.679688, "cut_share": 0.128906, "logprob_abs_diff": 0.014454, "train_rollout_kl": 0.000677, "entropy": 0.0, "grad_norm": 0.547757, "clip_fraction": 0.0, "step_time_s": 131.911865, "log_probs_time_s": 20.203343, "train_time_s": 90.924267, "update_weights_time_s": 0.636035, "sleep_time_s": 0.768382, "wake_up_time_s": 0.88824, "step_end": "2026-10-17 23:52:53", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 39, "length": 499.277344, "reward_all_samples": 0.820312, "groups_all_wrong": 1.0, "groups_all_right": 14.0, "rollout_time_s": 37.675957, "rollout_end": "2026-10-17 23:53:33", "reward": 0.820312, "cut_share": 0.09375, "logprob_abs_diff": 0.014327, "train_rollout_kl": 0.000687, "entropy": 0.0, "grad_norm": 0.648815, "clip_fraction": 0.0, "step_time_s": 134.689584, "log_probs_time_s": 20.677304, "train_time_s": 92.578923, "update_weights_time_s": 0.644006, "sleep_time_s": 0.761796, "wake_up_time_s": 0.895904, "step_end": "2026-10-17 23:55:07", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 40, "length": 436.679688, "reward_all_samples": 0.769531, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 31.195338, "rollout_end": "2026-10-17 23:55:42", "reward": 0.769531, "cut_share": 0.070312, "logprob_abs_diff": 0.014349, "train_rollout_kl": 0.000685, "entropy": 0.0, "grad_norm": 0.538948, "clip_fraction": 0.0, "step_time_s": 124.066315, "log_probs_time_s": 19.259552, "train_time_s": 88.380528, "update_weights_time_s": 0.628844, "sleep_time_s": 0.741202, "wake_up_time_s": 0.877157, "step_end": "2026-10-17 23:57:11", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 41, "length": 450.601562, "reward_all_samples": 0.777344, "groups_all_wrong": 1.0, "groups_all_right": 13.0, "rollout_time_s": 32.975102, "rollout_end": "2026-10-17 23:57:47", "reward": 0.777344, "cut_share": 0.0625, "logprob_abs_diff": 0.014435, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.650232, "clip_fraction": 0.0, "step_time_s": 127.420794, "log_probs_time_s": 19.735301, "train_time_s": 90.0794, "update_weights_time_s": 0.621142, "sleep_time_s": 0.770449, "wake_up_time_s": 0.900036, "step_end": "2026-10-17 23:59:19", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 42, "length": 472.621094, "reward_all_samples": 0.792969, "groups_all_wrong": 1.0, "groups_all_right": 16.0, "rollout_time_s": 35.311249, "rollout_end": "2026-10-17 23:59:57", "reward": 0.792969, "cut_share": 0.089844, "logprob_abs_diff": 0.014534, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.483062, "clip_fraction": 0.0, "step_time_s": 130.585529, "log_probs_time_s": 20.17387, "train_time_s": 90.818505, "update_weights_time_s": 0.65593, "sleep_time_s": 0.783026, "wake_up_time_s": 0.901579, "step_end": "2026-10-18 00:01:29", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 43, "length": 483.699219, "reward_all_samples": 0.675781, "groups_all_wrong": 2.0, "groups_all_right": 14.0, "rollout_time_s": 35.148744, "rollout_end": "2026-10-18 00:02:07", "reward": 0.675781, "cut_share": 0.058594, "logprob_abs_diff": 0.014407, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.540939, "clip_fraction": 0.0, "step_time_s": 131.107481, "log_probs_time_s": 20.182669, "train_time_s": 91.540745, "update_weights_time_s": 0.660267, "sleep_time_s": 0.764986, "wake_up_time_s": 0.87618, "step_end": "2026-10-18 00:03:41", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 44, "length": 425.105469, "reward_all_samples": 0.8125, "groups_all_wrong": 3.0, "groups_all_right": 15.0, "rollout_time_s": 31.510848, "rollout_end": "2026-10-18 00:04:15", "reward": 0.8125, "cut_share": 0.089844, "logprob_abs_diff": 0.015362, "train_rollout_kl": 0.000701, "entropy": 0.0, "grad_norm": 0.603242, "clip_fraction": 0.0, "step_time_s": 123.024417, "log_probs_time_s": 18.899076, "train_time_s": 87.030386, "update_weights_time_s": 0.633875, "sleep_time_s": 0.737595, "wake_up_time_s": 0.869562, "step_end": "2026-10-18 00:05:44", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 45, "length": 416.96875, "reward_all_samples": 0.746094, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 30.356738, "rollout_end": "2026-10-18 00:06:17", "reward": 0.746094, "cut_share": 0.082031, "logprob_abs_diff": 0.014662, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.60041, "clip_fraction": 0.0, "step_time_s": 121.57248, "log_probs_time_s": 18.753447, "train_time_s": 86.663565, "update_weights_time_s": 0.672493, "sleep_time_s": 0.786245, "wake_up_time_s": 0.880632, "step_end": "2026-10-18 00:07:45", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 46, "length": 470.039062, "reward_all_samples": 0.695312, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 35.380285, "rollout_end": "2026-10-18 00:08:23", "reward": 0.695312, "cut_share": 0.113281, "logprob_abs_diff": 0.015031, "train_rollout_kl": 0.000698, "entropy": 0.0, "grad_norm": 0.667854, "clip_fraction": 0.0, "step_time_s": 129.548462, "log_probs_time_s": 19.898874, "train_time_s": 89.653435, "update_weights_time_s": 0.651094, "sleep_time_s": 0.756892, "wake_up_time_s": 0.89712, "step_end": "2026-10-18 00:09:55", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 47, "length": 421.730469, "reward_all_samples": 0.792969, "groups_all_wrong": 2.0, "groups_all_right": 17.0, "rollout_time_s": 29.969306, "rollout_end": "2026-10-18 00:10:28", "reward": 0.792969, "cut_share": 0.035156, "logprob_abs_diff": 0.014127, "train_rollout_kl": 0.000679, "entropy": 0.0, "grad_norm": 0.474944, "clip_fraction": 0.0, "step_time_s": 121.499414, "log_probs_time_s": 18.915833, "train_time_s": 87.008881, "update_weights_time_s": 0.631867, "sleep_time_s": 0.748761, "wake_up_time_s": 0.882771, "step_end": "2026-10-18 00:11:56", "rollout_paused_s": 0.0, "step_paused_s": 0.0 } ] }, "plain-48-s2": { "steps": 48, "partial": false, "median": { "reward": 0.708984, "length": 547.648438, "cut_share": 0.15625, "reward_all_samples": 0.708984, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.01507, "train_rollout_kl": 0.000694, "entropy": 0.0, "grad_norm": 0.501535, "clip_fraction": 0.0, "step_time_s": 322.642254, "rollout_time_s": 86.239148, "log_probs_time_s": 66.084947, "train_time_s": 229.782396, "update_weights_time_s": 0.617402, "sleep_time_s": 0.719886, "wake_up_time_s": 0.817025 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 52.042886, "steps": 4 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 752.558594, "reward_all_samples": 0.582031, "groups_all_wrong": 7.0, "groups_all_right": 11.0, "rollout_time_s": 61.408626, "rollout_end": "2026-10-18 10:57:40", "reward": 0.582031, "cut_share": 0.386719, "logprob_abs_diff": 0.015252, "train_rollout_kl": 0.000679, "entropy": 0.0, "grad_norm": 0.347594, "clip_fraction": 0.0, "step_time_s": 322.166535, "log_probs_time_s": 45.211107, "train_time_s": 208.75754, "update_weights_time_s": 0.617355, "sleep_time_s": 2.118528, "wake_up_time_s": 0.385719, "step_end": "2026-10-18 11:01:10", "rollout_paused_s": 0.0, "step_paused_s": 108.0 }, { "step": 1, "length": 742.777344, "reward_all_samples": 0.660156, "groups_all_wrong": 3.0, "groups_all_right": 10.0, "rollout_time_s": 1606.364795, "rollout_end": "2026-10-18 11:28:03", "reward": 0.660156, "cut_share": 0.308594, "logprob_abs_diff": 0.015046, "train_rollout_kl": 0.000685, "entropy": 0.0, "grad_norm": 0.476477, "clip_fraction": 0.0, "step_time_s": 1901.745272, "log_probs_time_s": 70.737361, "train_time_s": 286.916212, "update_weights_time_s": 0.615267, "sleep_time_s": 4.944794, "wake_up_time_s": 0.813332, "step_end": "2026-10-18 11:32:51", "rollout_paused_s": 1546.0, "step_paused_s": 1728.0 }, { "step": 2, "length": 702.234375, "reward_all_samples": 0.644531, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 101.797609, "rollout_end": "2026-10-18 11:34:36", "reward": 0.644531, "cut_share": 0.277344, "logprob_abs_diff": 0.015226, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.4744, "clip_fraction": 0.0, "step_time_s": 391.111329, "log_probs_time_s": 69.842983, "train_time_s": 284.668676, "update_weights_time_s": 0.616496, "sleep_time_s": 1.107611, "wake_up_time_s": 0.82704, "step_end": "2026-10-18 11:39:23", "rollout_paused_s": 45.0, "step_paused_s": 229.0 }, { "step": 3, "length": 697.039062, "reward_all_samples": 0.738281, "groups_all_wrong": 2.0, "groups_all_right": 12.0, "rollout_time_s": 101.531486, "rollout_end": "2026-10-18 11:41:07", "reward": 0.738281, "cut_share": 0.25, "logprob_abs_diff": 0.015218, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.458058, "clip_fraction": 0.0, "step_time_s": 390.201879, "log_probs_time_s": 115.173476, "train_time_s": 284.097182, "update_weights_time_s": 0.636806, "sleep_time_s": 1.119223, "wake_up_time_s": 0.829764, "step_end": "2026-10-18 11:45:53", "rollout_paused_s": 46.0, "step_paused_s": 228.0 }, { "step": 4, "length": 666.070312, "reward_all_samples": 0.609375, "groups_all_wrong": 8.0, "groups_all_right": 11.0, "rollout_time_s": 144.501633, "rollout_end": "2026-10-18 11:48:20", "reward": 0.609375, "cut_share": 0.28125, "logprob_abs_diff": 0.015808, "train_rollout_kl": 0.000726, "entropy": 0.0, "grad_norm": 0.386409, "clip_fraction": 0.0, "step_time_s": 431.466638, "log_probs_time_s": 69.259217, "train_time_s": 282.720507, "update_weights_time_s": 0.615801, "sleep_time_s": 0.707751, "wake_up_time_s": 0.823274, "step_end": "2026-10-18 11:53:04", "rollout_paused_s": 90.0, "step_paused_s": 270.0 }, { "step": 5, "length": 724.230469, "reward_all_samples": 0.683594, "groups_all_wrong": 5.0, "groups_all_right": 10.0, "rollout_time_s": 104.906889, "rollout_end": "2026-10-18 11:54:52", "reward": 0.683594, "cut_share": 0.289062, "logprob_abs_diff": 0.015242, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.460536, "clip_fraction": 0.0, "step_time_s": 350.057507, "log_probs_time_s": 70.530826, "train_time_s": 241.048189, "update_weights_time_s": 0.590565, "sleep_time_s": 0.721519, "wake_up_time_s": 0.822689, "step_end": "2026-10-18 11:58:54", "rollout_paused_s": 45.0, "step_paused_s": 181.1 }, { "step": 6, "length": 700.234375, "reward_all_samples": 0.679688, "groups_all_wrong": 2.0, "groups_all_right": 8.0, "rollout_time_s": 147.478034, "rollout_end": "2026-10-18 12:01:24", "reward": 0.679688, "cut_share": 0.253906, "logprob_abs_diff": 0.015092, "train_rollout_kl": 0.000682, "entropy": 0.0, "grad_norm": 0.487071, "clip_fraction": 0.0, "step_time_s": 436.029222, "log_probs_time_s": 69.828648, "train_time_s": 284.427208, "update_weights_time_s": 0.601705, "sleep_time_s": 0.67285, "wake_up_time_s": 0.726147, "step_end": "2026-10-18 12:06:10", "rollout_paused_s": 91.0, "step_paused_s": 271.0 }, { "step": 7, "length": 712.496094, "reward_all_samples": 0.628906, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 59.118793, "rollout_end": "2026-10-18 12:07:12", "reward": 0.628906, "cut_share": 0.332031, "logprob_abs_diff": 0.015734, "train_rollout_kl": 0.00074, "entropy": 0.0, "grad_norm": 0.510967, "clip_fraction": 0.0, "step_time_s": 348.909349, "log_probs_time_s": 70.188628, "train_time_s": 285.761003, "update_weights_time_s": 0.605427, "sleep_time_s": 0.695774, "wake_up_time_s": 0.72846, "step_end": "2026-10-18 12:11:59", "rollout_paused_s": 0.0, "step_paused_s": 184.0 }, { "step": 8, "length": 686.101562, "reward_all_samples": 0.652344, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 100.636346, "rollout_end": "2026-10-18 12:13:42", "reward": 0.652344, "cut_share": 0.25, "logprob_abs_diff": 0.016124, "train_rollout_kl": 0.000746, "entropy": 0.0, "grad_norm": 0.501578, "clip_fraction": 0.0, "step_time_s": 388.849529, "log_probs_time_s": 115.173647, "train_time_s": 284.210706, "update_weights_time_s": 0.620227, "sleep_time_s": 0.674571, "wake_up_time_s": 0.715202, "step_end": "2026-10-18 12:18:28", "rollout_paused_s": 46.0, "step_paused_s": 229.0 }, { "step": 9, "length": 787.382812, "reward_all_samples": 0.535156, "groups_all_wrong": 4.0, "groups_all_right": 7.0, "rollout_time_s": 157.289128, "rollout_end": "2026-10-18 12:21:08", "reward": 0.535156, "cut_share": 0.367188, "logprob_abs_diff": 0.015281, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.484503, "clip_fraction": 0.0, "step_time_s": 498.934398, "log_probs_time_s": 72.097301, "train_time_s": 337.567236, "update_weights_time_s": 0.623603, "sleep_time_s": 0.724018, "wake_up_time_s": 0.724846, "step_end": "2026-10-18 12:26:47", "rollout_paused_s": 91.0, "step_paused_s": 316.0 }, { "step": 10, "length": 704.570312, "reward_all_samples": 0.695312, "groups_all_wrong": 2.0, "groups_all_right": 9.0, "rollout_time_s": 103.274477, "rollout_end": "2026-10-18 12:28:33", "reward": 0.695312, "cut_share": 0.308594, "logprob_abs_diff": 0.015055, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.50491, "clip_fraction": 0.0, "step_time_s": 393.302577, "log_probs_time_s": 70.277387, "train_time_s": 285.897742, "update_weights_time_s": 0.623891, "sleep_time_s": 0.697458, "wake_up_time_s": 0.705862, "step_end": "2026-10-18 12:33:20", "rollout_paused_s": 45.0, "step_paused_s": 226.0 }, { "step": 11, "length": 661.285156, "reward_all_samples": 0.636719, "groups_all_wrong": 2.0, "groups_all_right": 11.0, "rollout_time_s": 98.024474, "rollout_end": "2026-10-18 12:35:01", "reward": 0.636719, "cut_share": 0.246094, "logprob_abs_diff": 0.015128, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.510521, "clip_fraction": 0.0, "step_time_s": 338.872948, "log_probs_time_s": 69.09177, "train_time_s": 236.662446, "update_weights_time_s": 0.615023, "sleep_time_s": 0.716409, "wake_up_time_s": 0.707572, "step_end": "2026-10-18 12:38:59", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 12, "length": 649.5625, "reward_all_samples": 0.617188, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 142.444631, "rollout_end": "2026-10-18 12:41:24", "reward": 0.617188, "cut_share": 0.238281, "logprob_abs_diff": 0.015791, "train_rollout_kl": 0.00075, "entropy": 0.0, "grad_norm": 0.474243, "clip_fraction": 0.0, "step_time_s": 383.522168, "log_probs_time_s": 68.969289, "train_time_s": 236.80075, "update_weights_time_s": 0.62925, "sleep_time_s": 0.720548, "wake_up_time_s": 0.830579, "step_end": "2026-10-18 12:45:23", "rollout_paused_s": 90.0, "step_paused_s": 225.0 }, { "step": 13, "length": 655.085938, "reward_all_samples": 0.613281, "groups_all_wrong": 6.0, "groups_all_right": 12.0, "rollout_time_s": 143.48573, "rollout_end": "2026-10-18 12:47:49", "reward": 0.613281, "cut_share": 0.28125, "logprob_abs_diff": 0.01498, "train_rollout_kl": 0.000665, "entropy": 0.0, "grad_norm": 0.442606, "clip_fraction": 0.0, "step_time_s": 429.479123, "log_probs_time_s": 68.994851, "train_time_s": 281.852688, "update_weights_time_s": 0.634583, "sleep_time_s": 0.714621, "wake_up_time_s": 0.807025, "step_end": "2026-10-18 12:52:32", "rollout_paused_s": 90.0, "step_paused_s": 271.0 }, { "step": 14, "length": 588.621094, "reward_all_samples": 0.742188, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 44.96698, "rollout_end": "2026-10-18 12:53:20", "reward": 0.742188, "cut_share": 0.136719, "logprob_abs_diff": 0.014973, "train_rollout_kl": 0.000714, "entropy": 0.0, "grad_norm": 0.525593, "clip_fraction": 0.0, "step_time_s": 281.203978, "log_probs_time_s": 67.506419, "train_time_s": 231.948795, "update_weights_time_s": 0.618365, "sleep_time_s": 0.698567, "wake_up_time_s": 0.84471, "step_end": "2026-10-18 12:57:13", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 15, "length": 591.855469, "reward_all_samples": 0.71875, "groups_all_wrong": 2.0, "groups_all_right": 14.0, "rollout_time_s": 137.115887, "rollout_end": "2026-10-18 12:59:33", "reward": 0.71875, "cut_share": 0.207031, "logprob_abs_diff": 0.014857, "train_rollout_kl": 0.00069, "entropy": 0.0, "grad_norm": 0.471427, "clip_fraction": 0.0, "step_time_s": 374.133741, "log_probs_time_s": 67.682408, "train_time_s": 232.921076, "update_weights_time_s": 0.607599, "sleep_time_s": 0.707062, "wake_up_time_s": 0.810887, "step_end": "2026-10-18 13:03:27", "rollout_paused_s": 90.0, "step_paused_s": 227.0 }, { "step": 16, "length": 657.214844, "reward_all_samples": 0.570312, "groups_all_wrong": 6.0, "groups_all_right": 7.0, "rollout_time_s": 144.015922, "rollout_end": "2026-10-18 13:05:54", "reward": 0.570312, "cut_share": 0.292969, "logprob_abs_diff": 0.015483, "train_rollout_kl": 0.000706, "entropy": 0.0, "grad_norm": 0.537013, "clip_fraction": 0.0, "step_time_s": 338.987573, "log_probs_time_s": 23.580889, "train_time_s": 190.933353, "update_weights_time_s": 0.612331, "sleep_time_s": 0.70518, "wake_up_time_s": 0.736861, "step_end": "2026-10-18 13:09:06", "rollout_paused_s": 91.0, "step_paused_s": 182.0 }, { "step": 17, "length": 674.394531, "reward_all_samples": 0.621094, "groups_all_wrong": 3.0, "groups_all_right": 8.0, "rollout_time_s": 146.497035, "rollout_end": "2026-10-18 13:11:36", "reward": 0.621094, "cut_share": 0.304688, "logprob_abs_diff": 0.015473, "train_rollout_kl": 0.000719, "entropy": 0.0, "grad_norm": 0.559751, "clip_fraction": 0.0, "step_time_s": 434.355815, "log_probs_time_s": 69.544481, "train_time_s": 283.690891, "update_weights_time_s": 0.624475, "sleep_time_s": 0.697467, "wake_up_time_s": 0.730756, "step_end": "2026-10-18 13:16:21", "rollout_paused_s": 90.0, "step_paused_s": 270.0 }, { "step": 18, "length": 612.0625, "reward_all_samples": 0.707031, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 94.284055, "rollout_end": "2026-10-18 13:17:58", "reward": 0.707031, "cut_share": 0.25, "logprob_abs_diff": 0.015469, "train_rollout_kl": 0.000718, "entropy": 0.0, "grad_norm": 0.460865, "clip_fraction": 0.0, "step_time_s": 332.029795, "log_probs_time_s": 68.020912, "train_time_s": 233.691536, "update_weights_time_s": 0.600797, "sleep_time_s": 0.719625, "wake_up_time_s": 0.74116, "step_end": "2026-10-18 13:21:53", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 19, "length": 646.644531, "reward_all_samples": 0.632812, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 143.334563, "rollout_end": "2026-10-18 13:24:19", "reward": 0.632812, "cut_share": 0.292969, "logprob_abs_diff": 0.016102, "train_rollout_kl": 0.000744, "entropy": 0.0, "grad_norm": 0.496125, "clip_fraction": 0.0, "step_time_s": 383.328754, "log_probs_time_s": 68.853581, "train_time_s": 235.813482, "update_weights_time_s": 0.63744, "sleep_time_s": 0.701787, "wake_up_time_s": 0.828318, "step_end": "2026-10-18 13:28:16", "rollout_paused_s": 90.0, "step_paused_s": 225.0 }, { "step": 20, "length": 573.222656, "reward_all_samples": 0.730469, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "rollout_time_s": 90.123078, "rollout_end": "2026-10-18 13:29:49", "reward": 0.730469, "cut_share": 0.175781, "logprob_abs_diff": 0.014415, "train_rollout_kl": 0.000662, "entropy": 0.0, "grad_norm": 0.52391, "clip_fraction": 0.0, "step_time_s": 325.934649, "log_probs_time_s": 67.4102, "train_time_s": 231.643832, "update_weights_time_s": 0.611104, "sleep_time_s": 0.72545, "wake_up_time_s": 0.727866, "step_end": "2026-10-18 13:33:42", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 21, "length": 566.851562, "reward_all_samples": 0.757812, "groups_all_wrong": 1.0, "groups_all_right": 13.0, "rollout_time_s": 88.315944, "rollout_end": "2026-10-18 13:35:13", "reward": 0.757812, "cut_share": 0.105469, "logprob_abs_diff": 0.014028, "train_rollout_kl": 0.000649, "entropy": 0.0, "grad_norm": 0.495194, "clip_fraction": 0.0, "step_time_s": 324.18034, "log_probs_time_s": 67.168335, "train_time_s": 231.592756, "update_weights_time_s": 0.601313, "sleep_time_s": 0.721943, "wake_up_time_s": 0.834935, "step_end": "2026-10-18 13:39:06", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 22, "length": 518.421875, "reward_all_samples": 0.746094, "groups_all_wrong": 4.0, "groups_all_right": 14.0, "rollout_time_s": 83.963543, "rollout_end": "2026-10-18 13:40:33", "reward": 0.746094, "cut_share": 0.109375, "logprob_abs_diff": 0.015165, "train_rollout_kl": 0.000716, "entropy": 0.0, "grad_norm": 0.501493, "clip_fraction": 0.0, "step_time_s": 315.398214, "log_probs_time_s": 65.993478, "train_time_s": 227.327396, "update_weights_time_s": 0.598749, "sleep_time_s": 0.680577, "wake_up_time_s": 0.837358, "step_end": "2026-10-18 13:44:22", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 23, "length": 568.769531, "reward_all_samples": 0.761719, "groups_all_wrong": 2.0, "groups_all_right": 14.0, "rollout_time_s": 89.448185, "rollout_end": "2026-10-18 13:45:54", "reward": 0.761719, "cut_share": 0.160156, "logprob_abs_diff": 0.014625, "train_rollout_kl": 0.000683, "entropy": 0.0, "grad_norm": 0.481305, "clip_fraction": 0.0, "step_time_s": 325.410797, "log_probs_time_s": 67.145284, "train_time_s": 231.833704, "update_weights_time_s": 0.607934, "sleep_time_s": 0.694531, "wake_up_time_s": 0.833421, "step_end": "2026-10-18 13:49:47", "rollout_paused_s": 45.0, "step_paused_s": 182.0 }, { "step": 24, "length": 527.332031, "reward_all_samples": 0.691406, "groups_all_wrong": 5.0, "groups_all_right": 19.0, "rollout_time_s": 131.066209, "rollout_end": "2026-10-18 13:52:01", "reward": 0.691406, "cut_share": 0.164062, "logprob_abs_diff": 0.014917, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.405193, "clip_fraction": 0.0, "step_time_s": 364.714386, "log_probs_time_s": 20.95308, "train_time_s": 229.554716, "update_weights_time_s": 0.652692, "sleep_time_s": 0.739103, "wake_up_time_s": 0.726534, "step_end": "2026-10-18 13:55:52", "rollout_paused_s": 91.0, "step_paused_s": 226.0 }, { "step": 25, "length": 550.96875, "reward_all_samples": 0.679688, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 88.390883, "rollout_end": "2026-10-18 13:57:23", "reward": 0.679688, "cut_share": 0.164062, "logprob_abs_diff": 0.01506, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.5627, "clip_fraction": 0.0, "step_time_s": 323.117972, "log_probs_time_s": 66.849289, "train_time_s": 230.449343, "update_weights_time_s": 0.596766, "sleep_time_s": 0.721384, "wake_up_time_s": 0.840523, "step_end": "2026-10-18 14:01:15", "rollout_paused_s": 46.0, "step_paused_s": 181.0 }, { "step": 26, "length": 514.984375, "reward_all_samples": 0.695312, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "rollout_time_s": 83.743882, "rollout_end": "2026-10-18 14:02:41", "reward": 0.695312, "cut_share": 0.101562, "logprob_abs_diff": 0.014845, "train_rollout_kl": 0.000692, "entropy": 0.0, "grad_norm": 0.551856, "clip_fraction": 0.0, "step_time_s": 360.907057, "log_probs_time_s": 65.99542, "train_time_s": 273.044495, "update_weights_time_s": 0.649024, "sleep_time_s": 0.724668, "wake_up_time_s": 0.734231, "step_end": "2026-10-18 14:07:16", "rollout_paused_s": 45.0, "step_paused_s": 225.0 }, { "step": 27, "length": 544.328125, "reward_all_samples": 0.75, "groups_all_wrong": 1.0, "groups_all_right": 15.0, "rollout_time_s": 86.57473, "rollout_end": "2026-10-18 14:08:45", "reward": 0.75, "cut_share": 0.132812, "logprob_abs_diff": 0.013762, "train_rollout_kl": 0.000627, "entropy": 0.0, "grad_norm": 0.440029, "clip_fraction": 0.0, "step_time_s": 320.713871, "log_probs_time_s": 21.192841, "train_time_s": 230.010075, "update_weights_time_s": 0.611801, "sleep_time_s": 0.723462, "wake_up_time_s": 0.830456, "step_end": "2026-10-18 14:12:37", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 28, "length": 512.226562, "reward_all_samples": 0.800781, "groups_all_wrong": 2.0, "groups_all_right": 17.0, "rollout_time_s": 37.824491, "rollout_end": "2026-10-18 14:13:17", "reward": 0.800781, "cut_share": 0.09375, "logprob_abs_diff": 0.015247, "train_rollout_kl": 0.000724, "entropy": 0.0, "grad_norm": 0.472875, "clip_fraction": 0.0, "step_time_s": 269.108574, "log_probs_time_s": 65.809833, "train_time_s": 226.97203, "update_weights_time_s": 0.641483, "sleep_time_s": 0.728716, "wake_up_time_s": 0.828482, "step_end": "2026-10-18 14:17:06", "rollout_paused_s": 0.0, "step_paused_s": 138.0 }, { "step": 29, "length": 496.246094, "reward_all_samples": 0.746094, "groups_all_wrong": 2.0, "groups_all_right": 12.0, "rollout_time_s": 82.369938, "rollout_end": "2026-10-18 14:18:31", "reward": 0.746094, "cut_share": 0.136719, "logprob_abs_diff": 0.015082, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.60279, "clip_fraction": 0.0, "step_time_s": 312.69519, "log_probs_time_s": 65.477605, "train_time_s": 226.107644, "update_weights_time_s": 0.632646, "sleep_time_s": 0.730395, "wake_up_time_s": 0.74407, "step_end": "2026-10-18 14:22:18", "rollout_paused_s": 46.0, "step_paused_s": 184.0 }, { "step": 30, "length": 494.734375, "reward_all_samples": 0.746094, "groups_all_wrong": 5.0, "groups_all_right": 17.0, "rollout_time_s": 82.993939, "rollout_end": "2026-10-18 14:23:44", "reward": 0.746094, "cut_share": 0.125, "logprob_abs_diff": 0.014501, "train_rollout_kl": 0.000667, "entropy": 0.0, "grad_norm": 0.461325, "clip_fraction": 0.0, "step_time_s": 267.654295, "log_probs_time_s": 65.370531, "train_time_s": 180.469308, "update_weights_time_s": 0.641436, "sleep_time_s": 0.720147, "wake_up_time_s": 0.835856, "step_end": "2026-10-18 14:26:46", "rollout_paused_s": 46.0, "step_paused_s": 138.0 }, { "step": 31, "length": 495.085938, "reward_all_samples": 0.71875, "groups_all_wrong": 5.0, "groups_all_right": 16.0, "rollout_time_s": 82.364456, "rollout_end": "2026-10-18 14:28:11", "reward": 0.71875, "cut_share": 0.109375, "logprob_abs_diff": 0.014625, "train_rollout_kl": 0.000681, "entropy": 0.0, "grad_norm": 0.49958, "clip_fraction": 0.0, "step_time_s": 312.40562, "log_probs_time_s": 65.542132, "train_time_s": 225.849372, "update_weights_time_s": 0.650426, "sleep_time_s": 0.730518, "wake_up_time_s": 0.840304, "step_end": "2026-10-18 14:31:58", "rollout_paused_s": 46.0, "step_paused_s": 183.0 }, { "step": 32, "length": 442.71875, "reward_all_samples": 0.785156, "groups_all_wrong": 1.0, "groups_all_right": 15.0, "rollout_time_s": 77.153895, "rollout_end": "2026-10-18 14:33:18", "reward": 0.785156, "cut_share": 0.085938, "logprob_abs_diff": 0.014608, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.593121, "clip_fraction": 0.0, "step_time_s": 303.442862, "log_probs_time_s": 64.151894, "train_time_s": 222.055324, "update_weights_time_s": 0.633072, "sleep_time_s": 0.731262, "wake_up_time_s": 0.745108, "step_end": "2026-10-18 14:37:02", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 33, "length": 463.496094, "reward_all_samples": 0.816406, "groups_all_right": 14.0, "rollout_time_s": 78.748006, "rollout_end": "2026-10-18 14:38:23", "reward": 0.816406, "cut_share": 0.050781, "logprob_abs_diff": 0.014321, "train_rollout_kl": 0.000677, "entropy": 0.0, "grad_norm": 0.634749, "clip_fraction": 0.0, "step_time_s": 306.045562, "log_probs_time_s": 19.239156, "train_time_s": 223.124633, "update_weights_time_s": 0.637515, "sleep_time_s": 0.708421, "wake_up_time_s": 0.732725, "step_end": "2026-10-18 14:42:08", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 34, "length": 503.972656, "reward_all_samples": 0.710938, "groups_all_wrong": 3.0, "groups_all_right": 11.0, "rollout_time_s": 83.648728, "rollout_end": "2026-10-18 14:43:34", "reward": 0.710938, "cut_share": 0.117188, "logprob_abs_diff": 0.01578, "train_rollout_kl": 0.000725, "entropy": 0.0, "grad_norm": 0.576522, "clip_fraction": 0.0, "step_time_s": 269.253371, "log_probs_time_s": 20.261316, "train_time_s": 181.506833, "update_weights_time_s": 0.602773, "sleep_time_s": 0.710995, "wake_up_time_s": 0.820718, "step_end": "2026-10-18 14:46:37", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 35, "length": 518.351562, "reward_all_samples": 0.71875, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 130.588551, "rollout_end": "2026-10-18 14:48:50", "reward": 0.71875, "cut_share": 0.144531, "logprob_abs_diff": 0.015291, "train_rollout_kl": 0.000709, "entropy": 0.0, "grad_norm": 0.494253, "clip_fraction": 0.0, "step_time_s": 317.715663, "log_probs_time_s": 20.767683, "train_time_s": 182.973197, "update_weights_time_s": 0.639216, "sleep_time_s": 0.712486, "wake_up_time_s": 0.825302, "step_end": "2026-10-18 14:51:55", "rollout_paused_s": 90.0, "step_paused_s": 180.0 }, { "step": 36, "length": 443.574219, "reward_all_samples": 0.804688, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 77.2396, "rollout_end": "2026-10-18 14:53:15", "reward": 0.804688, "cut_share": 0.066406, "logprob_abs_diff": 0.014098, "train_rollout_kl": 0.000651, "entropy": 0.0, "grad_norm": 0.64746, "clip_fraction": 0.0, "step_time_s": 303.985387, "log_probs_time_s": 64.259554, "train_time_s": 222.496111, "update_weights_time_s": 0.614512, "sleep_time_s": 0.755498, "wake_up_time_s": 0.726886, "step_end": "2026-10-18 14:56:59", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 37, "length": 532.769531, "reward_all_samples": 0.777344, "groups_all_wrong": 1.0, "groups_all_right": 9.0, "rollout_time_s": 85.463383, "rollout_end": "2026-10-18 14:58:27", "reward": 0.777344, "cut_share": 0.109375, "logprob_abs_diff": 0.015058, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.575229, "clip_fraction": 0.0, "step_time_s": 318.172677, "log_probs_time_s": 66.174475, "train_time_s": 228.573884, "update_weights_time_s": 0.614766, "sleep_time_s": 0.699137, "wake_up_time_s": 0.828356, "step_end": "2026-10-18 15:02:17", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 38, "length": 441.675781, "reward_all_samples": 0.777344, "groups_all_wrong": 2.0, "groups_all_right": 18.0, "rollout_time_s": 76.491319, "rollout_end": "2026-10-18 15:03:36", "reward": 0.777344, "cut_share": 0.035156, "logprob_abs_diff": 0.013844, "train_rollout_kl": 0.000643, "entropy": 0.0, "grad_norm": 0.527118, "clip_fraction": 0.0, "step_time_s": 303.103846, "log_probs_time_s": 64.267028, "train_time_s": 222.468734, "update_weights_time_s": 0.618843, "sleep_time_s": 0.711818, "wake_up_time_s": 0.847835, "step_end": "2026-10-18 15:07:20", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 39, "length": 483.289062, "reward_all_samples": 0.71875, "groups_all_wrong": 4.0, "groups_all_right": 16.0, "rollout_time_s": 81.33857, "rollout_end": "2026-10-18 15:08:44", "reward": 0.71875, "cut_share": 0.09375, "logprob_abs_diff": 0.01476, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.459736, "clip_fraction": 0.0, "step_time_s": 310.294958, "log_probs_time_s": 65.080355, "train_time_s": 224.865642, "update_weights_time_s": 0.613744, "sleep_time_s": 0.693547, "wake_up_time_s": 0.826262, "step_end": "2026-10-18 15:12:30", "rollout_paused_s": 46.0, "step_paused_s": 183.0 }, { "step": 40, "length": 489.320312, "reward_all_samples": 0.699219, "groups_all_wrong": 2.0, "groups_all_right": 15.0, "rollout_time_s": 82.891813, "rollout_end": "2026-10-18 15:13:56", "reward": 0.699219, "cut_share": 0.152344, "logprob_abs_diff": 0.015281, "train_rollout_kl": 0.000718, "entropy": 0.0, "grad_norm": 0.575656, "clip_fraction": 0.0, "step_time_s": 312.975697, "log_probs_time_s": 65.307335, "train_time_s": 226.031494, "update_weights_time_s": 0.616653, "sleep_time_s": 0.696185, "wake_up_time_s": 0.744733, "step_end": "2026-10-18 15:17:43", "rollout_paused_s": 46.0, "step_paused_s": 181.0 }, { "step": 41, "length": 438.003906, "reward_all_samples": 0.785156, "groups_all_wrong": 4.0, "groups_all_right": 13.0, "rollout_time_s": 77.374043, "rollout_end": "2026-10-18 15:19:04", "reward": 0.785156, "cut_share": 0.078125, "logprob_abs_diff": 0.014813, "train_rollout_kl": 0.000688, "entropy": 0.0, "grad_norm": 0.58699, "clip_fraction": 0.0, "step_time_s": 303.879963, "log_probs_time_s": 64.333129, "train_time_s": 222.237169, "update_weights_time_s": 0.618767, "sleep_time_s": 0.706169, "wake_up_time_s": 0.841916, "step_end": "2026-10-18 15:22:47", "rollout_paused_s": 46.0, "step_paused_s": 184.0 }, { "step": 42, "length": 480.53125, "reward_all_samples": 0.761719, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 81.963411, "rollout_end": "2026-10-18 15:24:12", "reward": 0.761719, "cut_share": 0.136719, "logprob_abs_diff": 0.014493, "train_rollout_kl": 0.000668, "entropy": 0.0, "grad_norm": 0.492368, "clip_fraction": 0.0, "step_time_s": 264.931327, "log_probs_time_s": 65.319021, "train_time_s": 178.819364, "update_weights_time_s": 0.600363, "sleep_time_s": 0.704342, "wake_up_time_s": 0.750154, "step_end": "2026-10-18 15:27:12", "rollout_paused_s": 46.0, "step_paused_s": 137.0 }, { "step": 43, "length": 429.277344, "reward_all_samples": 0.839844, "groups_all_right": 17.0, "rollout_time_s": 75.920946, "rollout_end": "2026-10-18 15:28:31", "reward": 0.839844, "cut_share": 0.046875, "logprob_abs_diff": 0.013986, "train_rollout_kl": 0.000644, "entropy": 0.0, "grad_norm": 0.525474, "clip_fraction": 0.0, "step_time_s": 302.716054, "log_probs_time_s": 64.084906, "train_time_s": 222.485706, "update_weights_time_s": 0.617449, "sleep_time_s": 0.725516, "wake_up_time_s": 0.836966, "step_end": "2026-10-18 15:32:15", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 44, "length": 521.574219, "reward_all_samples": 0.6875, "groups_all_wrong": 4.0, "groups_all_right": 14.0, "rollout_time_s": 85.903566, "rollout_end": "2026-10-18 15:33:44", "reward": 0.6875, "cut_share": 0.183594, "logprob_abs_diff": 0.015992, "train_rollout_kl": 0.000757, "entropy": 0.0, "grad_norm": 0.505298, "clip_fraction": 0.0, "step_time_s": 318.977254, "log_probs_time_s": 66.211496, "train_time_s": 228.994133, "update_weights_time_s": 0.638632, "sleep_time_s": 0.72308, "wake_up_time_s": 0.736335, "step_end": "2026-10-18 15:37:34", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 45, "length": 476.285156, "reward_all_samples": 0.773438, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 80.490551, "rollout_end": "2026-10-18 15:38:57", "reward": 0.773438, "cut_share": 0.105469, "logprob_abs_diff": 0.015417, "train_rollout_kl": 0.00073, "entropy": 0.0, "grad_norm": 0.509567, "clip_fraction": 0.0, "step_time_s": 309.938618, "log_probs_time_s": 65.150472, "train_time_s": 225.338297, "update_weights_time_s": 0.649852, "sleep_time_s": 0.729578, "wake_up_time_s": 0.741856, "step_end": "2026-10-18 15:42:44", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 46, "length": 494.171875, "reward_all_samples": 0.695312, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 82.781482, "rollout_end": "2026-10-18 15:44:10", "reward": 0.695312, "cut_share": 0.121094, "logprob_abs_diff": 0.01508, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.579494, "clip_fraction": 0.0, "step_time_s": 313.324795, "log_probs_time_s": 65.550115, "train_time_s": 226.203359, "update_weights_time_s": 0.658409, "sleep_time_s": 0.732908, "wake_up_time_s": 0.823567, "step_end": "2026-10-18 15:47:57", "rollout_paused_s": 45.0, "step_paused_s": 182.0 }, { "step": 47, "length": 514.039062, "reward_all_samples": 0.648438, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 83.99452, "rollout_end": "2026-10-18 15:49:24", "reward": 0.648438, "cut_share": 0.121094, "logprob_abs_diff": 0.014964, "train_rollout_kl": 0.000676, "entropy": 0.0, "grad_norm": 0.582691, "clip_fraction": 0.0, "step_time_s": 315.437781, "log_probs_time_s": 65.90031, "train_time_s": 227.139489, "update_weights_time_s": 0.60277, "sleep_time_s": 0.724207, "wake_up_time_s": 0.835459, "step_end": "2026-10-18 15:53:13", "rollout_paused_s": 46.0, "step_paused_s": 183.0 } ] }, "plain-48-s3": { "steps": 48, "partial": false, "median": { "reward": 0.71875, "length": 559.935547, "cut_share": 0.175781, "reward_all_samples": 0.71875, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015138, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.505451, "clip_fraction": 0.0, "step_time_s": 318.724924, "rollout_time_s": 90.906474, "log_probs_time_s": 66.293575, "train_time_s": 227.965522, "update_weights_time_s": 0.611949, "sleep_time_s": 0.718178, "wake_up_time_s": 0.814259 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 36.089931, "steps": 7 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 707.78125, "reward_all_samples": 0.632812, "groups_all_wrong": 5.0, "groups_all_right": 11.0, "rollout_time_s": 104.439087, "rollout_end": "2026-10-18 15:57:07", "reward": 0.632812, "cut_share": 0.320312, "logprob_abs_diff": 0.014946, "train_rollout_kl": 0.000661, "entropy": 0.0, "grad_norm": 0.429164, "clip_fraction": 0.0, "step_time_s": 411.203166, "log_probs_time_s": 74.839585, "train_time_s": 249.4332, "update_weights_time_s": 0.610188, "sleep_time_s": 2.506842, "wake_up_time_s": 0.3863, "step_end": "2026-10-18 16:01:18", "rollout_paused_s": 45.0, "step_paused_s": 183.0 }, { "step": 1, "length": 733.195312, "reward_all_samples": 0.636719, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 151.986619, "rollout_end": "2026-10-18 16:03:56", "reward": 0.636719, "cut_share": 0.304688, "logprob_abs_diff": 0.015562, "train_rollout_kl": 0.000718, "entropy": 0.0, "grad_norm": 0.475114, "clip_fraction": 0.0, "step_time_s": 446.980515, "log_probs_time_s": 116.406189, "train_time_s": 287.38011, "update_weights_time_s": 0.582423, "sleep_time_s": 4.086954, "wake_up_time_s": 0.828163, "step_end": "2026-10-18 16:08:45", "rollout_paused_s": 90.0, "step_paused_s": 273.0 }, { "step": 2, "length": 758.941406, "reward_all_samples": 0.554688, "groups_all_wrong": 5.0, "groups_all_right": 8.0, "rollout_time_s": 155.033947, "rollout_end": "2026-10-18 16:11:23", "reward": 0.554688, "cut_share": 0.417969, "logprob_abs_diff": 0.015992, "train_rollout_kl": 0.000735, "entropy": 0.0, "grad_norm": 0.444735, "clip_fraction": 0.0, "step_time_s": 448.733365, "log_probs_time_s": 71.38709, "train_time_s": 288.994096, "update_weights_time_s": 0.584157, "sleep_time_s": 1.201412, "wake_up_time_s": 0.808908, "step_end": "2026-10-18 16:16:13", "rollout_paused_s": 91.0, "step_paused_s": 271.0 }, { "step": 3, "length": 750.867188, "reward_all_samples": 0.609375, "groups_all_wrong": 4.0, "groups_all_right": 8.0, "rollout_time_s": 107.860711, "rollout_end": "2026-10-18 16:18:04", "reward": 0.609375, "cut_share": 0.34375, "logprob_abs_diff": 0.015438, "train_rollout_kl": 0.000684, "entropy": 0.0, "grad_norm": 0.551633, "clip_fraction": 0.0, "step_time_s": 401.364995, "log_probs_time_s": 116.753217, "train_time_s": 289.016748, "update_weights_time_s": 0.615789, "sleep_time_s": 1.139914, "wake_up_time_s": 0.73346, "step_end": "2026-10-18 16:22:55", "rollout_paused_s": 45.0, "step_paused_s": 226.0 }, { "step": 4, "length": 780.464844, "reward_all_samples": 0.527344, "groups_all_wrong": 4.0, "groups_all_right": 6.0, "rollout_time_s": 157.139435, "rollout_end": "2026-10-18 16:25:35", "reward": 0.527344, "cut_share": 0.425781, "logprob_abs_diff": 0.016354, "train_rollout_kl": 0.00074, "entropy": 0.0, "grad_norm": 0.510983, "clip_fraction": 0.0, "step_time_s": 452.124256, "log_probs_time_s": 71.90754, "train_time_s": 290.851522, "update_weights_time_s": 0.622688, "sleep_time_s": 0.691342, "wake_up_time_s": 0.814259, "step_end": "2026-10-18 16:30:27", "rollout_paused_s": 90.0, "step_paused_s": 271.0 }, { "step": 5, "length": 767.554688, "reward_all_samples": 0.59375, "groups_all_wrong": 3.0, "groups_all_right": 8.0, "rollout_time_s": 156.186483, "rollout_end": "2026-10-18 16:33:06", "reward": 0.59375, "cut_share": 0.40625, "logprob_abs_diff": 0.015741, "train_rollout_kl": 0.000741, "entropy": 0.0, "grad_norm": 0.504753, "clip_fraction": 0.0, "step_time_s": 450.82017, "log_probs_time_s": 71.718196, "train_time_s": 290.375954, "update_weights_time_s": 0.59172, "sleep_time_s": 0.684593, "wake_up_time_s": 0.83998, "step_end": "2026-10-18 16:37:58", "rollout_paused_s": 91.0, "step_paused_s": 273.0 }, { "step": 6, "length": 688.675781, "reward_all_samples": 0.652344, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 147.227955, "rollout_end": "2026-10-18 16:40:28", "reward": 0.652344, "cut_share": 0.261719, "logprob_abs_diff": 0.015383, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.519008, "clip_fraction": 0.0, "step_time_s": 390.543494, "log_probs_time_s": 69.887308, "train_time_s": 239.104362, "update_weights_time_s": 0.589328, "sleep_time_s": 0.697385, "wake_up_time_s": 0.824077, "step_end": "2026-10-18 16:44:28", "rollout_paused_s": 92.0, "step_paused_s": 230.0 }, { "step": 7, "length": 681.417969, "reward_all_samples": 0.707031, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 145.750559, "rollout_end": "2026-10-18 16:46:57", "reward": 0.707031, "cut_share": 0.261719, "logprob_abs_diff": 0.01576, "train_rollout_kl": 0.000718, "entropy": 0.0, "grad_norm": 0.497401, "clip_fraction": 0.0, "step_time_s": 388.809899, "log_probs_time_s": 69.70487, "train_time_s": 238.951403, "update_weights_time_s": 0.630311, "sleep_time_s": 0.743899, "wake_up_time_s": 0.727499, "step_end": "2026-10-18 16:50:57", "rollout_paused_s": 92.0, "step_paused_s": 229.0 }, { "step": 8, "length": 633.761719, "reward_all_samples": 0.683594, "groups_all_wrong": 5.0, "groups_all_right": 15.0, "rollout_time_s": 141.325476, "rollout_end": "2026-10-18 16:53:21", "reward": 0.683594, "cut_share": 0.257812, "logprob_abs_diff": 0.015256, "train_rollout_kl": 0.000712, "entropy": 0.0, "grad_norm": 0.42741, "clip_fraction": 0.0, "step_time_s": 380.9388, "log_probs_time_s": 68.591889, "train_time_s": 235.363843, "update_weights_time_s": 0.613709, "sleep_time_s": 0.72166, "wake_up_time_s": 0.813626, "step_end": "2026-10-18 16:57:18", "rollout_paused_s": 91.0, "step_paused_s": 229.0 }, { "step": 9, "length": 697.070312, "reward_all_samples": 0.644531, "groups_all_wrong": 5.0, "groups_all_right": 14.0, "rollout_time_s": 103.228772, "rollout_end": "2026-10-18 16:59:04", "reward": 0.644531, "cut_share": 0.304688, "logprob_abs_diff": 0.015477, "train_rollout_kl": 0.000713, "entropy": 0.0, "grad_norm": 0.390738, "clip_fraction": 0.0, "step_time_s": 347.809578, "log_probs_time_s": 70.327513, "train_time_s": 240.482067, "update_weights_time_s": 0.58814, "sleep_time_s": 0.714502, "wake_up_time_s": 0.81426, "step_end": "2026-10-18 17:03:06", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 10, "length": 616.128906, "reward_all_samples": 0.753906, "groups_all_wrong": 2.0, "groups_all_right": 17.0, "rollout_time_s": 139.768973, "rollout_end": "2026-10-18 17:05:28", "reward": 0.753906, "cut_share": 0.222656, "logprob_abs_diff": 0.015178, "train_rollout_kl": 0.000715, "entropy": 0.0, "grad_norm": 0.422095, "clip_fraction": 0.0, "step_time_s": 378.33772, "log_probs_time_s": 68.3037, "train_time_s": 234.506451, "update_weights_time_s": 0.605986, "sleep_time_s": 0.72049, "wake_up_time_s": 0.727735, "step_end": "2026-10-18 17:09:24", "rollout_paused_s": 90.0, "step_paused_s": 225.0 }, { "step": 11, "length": 656.871094, "reward_all_samples": 0.664062, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 99.248542, "rollout_end": "2026-10-18 17:11:06", "reward": 0.664062, "cut_share": 0.269531, "logprob_abs_diff": 0.015491, "train_rollout_kl": 0.000708, "entropy": 0.0, "grad_norm": 0.443776, "clip_fraction": 0.0, "step_time_s": 387.081343, "log_probs_time_s": 69.413378, "train_time_s": 283.571994, "update_weights_time_s": 0.584115, "sleep_time_s": 0.714218, "wake_up_time_s": 0.831131, "step_end": "2026-10-18 17:15:51", "rollout_paused_s": 45.0, "step_paused_s": 225.0 }, { "step": 12, "length": 687.109375, "reward_all_samples": 0.558594, "groups_all_wrong": 7.0, "groups_all_right": 9.0, "rollout_time_s": 102.497401, "rollout_end": "2026-10-18 17:17:36", "reward": 0.558594, "cut_share": 0.339844, "logprob_abs_diff": 0.015561, "train_rollout_kl": 0.000719, "entropy": 0.0, "grad_norm": 0.467353, "clip_fraction": 0.0, "step_time_s": 391.80597, "log_probs_time_s": 115.420068, "train_time_s": 285.178494, "update_weights_time_s": 0.59211, "sleep_time_s": 0.708643, "wake_up_time_s": 0.832887, "step_end": "2026-10-18 17:22:23", "rollout_paused_s": 45.0, "step_paused_s": 226.0 }, { "step": 13, "length": 613.769531, "reward_all_samples": 0.636719, "groups_all_wrong": 3.0, "groups_all_right": 10.0, "rollout_time_s": 139.714796, "rollout_end": "2026-10-18 17:24:45", "reward": 0.636719, "cut_share": 0.199219, "logprob_abs_diff": 0.015889, "train_rollout_kl": 0.00073, "entropy": 0.0, "grad_norm": 0.515478, "clip_fraction": 0.0, "step_time_s": 379.127742, "log_probs_time_s": 68.487212, "train_time_s": 235.345222, "update_weights_time_s": 0.622188, "sleep_time_s": 0.721253, "wake_up_time_s": 0.742204, "step_end": "2026-10-18 17:28:42", "rollout_paused_s": 91.0, "step_paused_s": 227.0 }, { "step": 14, "length": 617.664062, "reward_all_samples": 0.726562, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "rollout_time_s": 94.027574, "rollout_end": "2026-10-18 17:30:19", "reward": 0.726562, "cut_share": 0.191406, "logprob_abs_diff": 0.015594, "train_rollout_kl": 0.000728, "entropy": 0.0, "grad_norm": 0.472441, "clip_fraction": 0.0, "step_time_s": 333.012976, "log_probs_time_s": 68.337605, "train_time_s": 234.832341, "update_weights_time_s": 0.586792, "sleep_time_s": 0.70207, "wake_up_time_s": 0.725804, "step_end": "2026-10-18 17:34:15", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 15, "length": 542.21875, "reward_all_samples": 0.816406, "groups_all_right": 15.0, "rollout_time_s": 131.605318, "rollout_end": "2026-10-18 17:36:30", "reward": 0.816406, "cut_share": 0.09375, "logprob_abs_diff": 0.014335, "train_rollout_kl": 0.000667, "entropy": 0.0, "grad_norm": 0.514845, "clip_fraction": 0.0, "step_time_s": 364.730598, "log_probs_time_s": 21.019519, "train_time_s": 228.923656, "update_weights_time_s": 0.620453, "sleep_time_s": 0.709443, "wake_up_time_s": 0.714706, "step_end": "2026-10-18 17:40:20", "rollout_paused_s": 91.0, "step_paused_s": 228.0 }, { "step": 16, "length": 557.921875, "reward_all_samples": 0.71875, "groups_all_wrong": 4.0, "groups_all_right": 15.0, "rollout_time_s": 43.056249, "rollout_end": "2026-10-18 17:41:06", "reward": 0.71875, "cut_share": 0.203125, "logprob_abs_diff": 0.014604, "train_rollout_kl": 0.000681, "entropy": 0.0, "grad_norm": 0.454955, "clip_fraction": 0.0, "step_time_s": 232.989696, "log_probs_time_s": 21.598283, "train_time_s": 185.919449, "update_weights_time_s": 0.594526, "sleep_time_s": 0.70484, "wake_up_time_s": 0.734773, "step_end": "2026-10-18 17:44:13", "rollout_paused_s": 0.0, "step_paused_s": 92.0 }, { "step": 17, "length": 593.5625, "reward_all_samples": 0.675781, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 92.033538, "rollout_end": "2026-10-18 17:45:48", "reward": 0.675781, "cut_share": 0.210938, "logprob_abs_diff": 0.015744, "train_rollout_kl": 0.00072, "entropy": 0.0, "grad_norm": 0.595523, "clip_fraction": 0.0, "step_time_s": 329.252387, "log_probs_time_s": 67.698671, "train_time_s": 233.101326, "update_weights_time_s": 0.588729, "sleep_time_s": 0.687692, "wake_up_time_s": 0.746266, "step_end": "2026-10-18 17:49:42", "rollout_paused_s": 46.0, "step_paused_s": 181.0 }, { "step": 18, "length": 606.480469, "reward_all_samples": 0.683594, "groups_all_wrong": 2.0, "groups_all_right": 11.0, "rollout_time_s": 93.036252, "rollout_end": "2026-10-18 17:51:18", "reward": 0.683594, "cut_share": 0.175781, "logprob_abs_diff": 0.014993, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.578667, "clip_fraction": 0.0, "step_time_s": 330.806337, "log_probs_time_s": 67.958863, "train_time_s": 233.741391, "update_weights_time_s": 0.59271, "sleep_time_s": 0.67666, "wake_up_time_s": 0.769958, "step_end": "2026-10-18 17:55:13", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 19, "length": 572.746094, "reward_all_samples": 0.738281, "groups_all_wrong": 3.0, "groups_all_right": 16.0, "rollout_time_s": 135.756645, "rollout_end": "2026-10-18 17:57:31", "reward": 0.738281, "cut_share": 0.160156, "logprob_abs_diff": 0.014841, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.479442, "clip_fraction": 0.0, "step_time_s": 326.225204, "log_probs_time_s": 21.880784, "train_time_s": 186.313931, "update_weights_time_s": 0.621721, "sleep_time_s": 0.709413, "wake_up_time_s": 0.831363, "step_end": "2026-10-18 18:00:39", "rollout_paused_s": 90.0, "step_paused_s": 180.0 }, { "step": 20, "length": 583.617188, "reward_all_samples": 0.742188, "groups_all_wrong": 2.0, "groups_all_right": 10.0, "rollout_time_s": 91.58616, "rollout_end": "2026-10-18 18:02:13", "reward": 0.742188, "cut_share": 0.203125, "logprob_abs_diff": 0.016004, "train_rollout_kl": 0.000738, "entropy": 0.0, "grad_norm": 0.557174, "clip_fraction": 0.0, "step_time_s": 328.625004, "log_probs_time_s": 67.65001, "train_time_s": 232.838418, "update_weights_time_s": 0.639482, "sleep_time_s": 0.731681, "wake_up_time_s": 0.83048, "step_end": "2026-10-18 18:06:08", "rollout_paused_s": 45.0, "step_paused_s": 182.0 }, { "step": 21, "length": 611.289062, "reward_all_samples": 0.730469, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 94.316462, "rollout_end": "2026-10-18 18:07:45", "reward": 0.730469, "cut_share": 0.234375, "logprob_abs_diff": 0.015639, "train_rollout_kl": 0.000707, "entropy": 0.0, "grad_norm": 0.478907, "clip_fraction": 0.0, "step_time_s": 378.761629, "log_probs_time_s": 68.328597, "train_time_s": 280.29012, "update_weights_time_s": 0.6267, "sleep_time_s": 0.739829, "wake_up_time_s": 0.76379, "step_end": "2026-10-18 18:12:27", "rollout_paused_s": 46.0, "step_paused_s": 229.0 }, { "step": 22, "length": 561.949219, "reward_all_samples": 0.726562, "groups_all_wrong": 1.0, "groups_all_right": 16.0, "rollout_time_s": 89.181048, "rollout_end": "2026-10-18 18:13:59", "reward": 0.726562, "cut_share": 0.191406, "logprob_abs_diff": 0.015313, "train_rollout_kl": 0.000703, "entropy": 0.0, "grad_norm": 0.46053, "clip_fraction": 0.0, "step_time_s": 278.377752, "log_probs_time_s": 66.896715, "train_time_s": 184.923646, "update_weights_time_s": 0.624764, "sleep_time_s": 0.71426, "wake_up_time_s": 0.821828, "step_end": "2026-10-18 18:17:05", "rollout_paused_s": 46.0, "step_paused_s": 136.0 }, { "step": 23, "length": 555.121094, "reward_all_samples": 0.683594, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 134.130854, "rollout_end": "2026-10-18 18:19:22", "reward": 0.683594, "cut_share": 0.199219, "logprob_abs_diff": 0.015244, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.524152, "clip_fraction": 0.0, "step_time_s": 277.166271, "log_probs_time_s": 21.311482, "train_time_s": 138.863114, "update_weights_time_s": 0.634087, "sleep_time_s": 0.730295, "wake_up_time_s": 0.824124, "step_end": "2026-10-18 18:21:42", "rollout_paused_s": 91.0, "step_paused_s": 136.0 }, { "step": 24, "length": 532.746094, "reward_all_samples": 0.71875, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 86.107026, "rollout_end": "2026-10-18 18:23:11", "reward": 0.71875, "cut_share": 0.152344, "logprob_abs_diff": 0.014961, "train_rollout_kl": 0.000681, "entropy": 0.0, "grad_norm": 0.579211, "clip_fraction": 0.0, "step_time_s": 319.385102, "log_probs_time_s": 111.941232, "train_time_s": 228.969029, "update_weights_time_s": 0.616071, "sleep_time_s": 0.72492, "wake_up_time_s": 0.825499, "step_end": "2026-10-18 18:27:01", "rollout_paused_s": 46.0, "step_paused_s": 182.0 }, { "step": 25, "length": 523.246094, "reward_all_samples": 0.75, "groups_all_wrong": 3.0, "groups_all_right": 16.0, "rollout_time_s": 85.279614, "rollout_end": "2026-10-18 18:28:29", "reward": 0.75, "cut_share": 0.109375, "logprob_abs_diff": 0.014428, "train_rollout_kl": 0.000681, "entropy": 0.0, "grad_norm": 0.448662, "clip_fraction": 0.0, "step_time_s": 272.44472, "log_probs_time_s": 66.205574, "train_time_s": 182.968114, "update_weights_time_s": 0.629703, "sleep_time_s": 0.731838, "wake_up_time_s": 0.834335, "step_end": "2026-10-18 18:31:34", "rollout_paused_s": 46.0, "step_paused_s": 138.0 }, { "step": 26, "length": 501.003906, "reward_all_samples": 0.769531, "groups_all_wrong": 2.0, "groups_all_right": 20.0, "rollout_time_s": 83.092677, "rollout_end": "2026-10-18 18:33:00", "reward": 0.769531, "cut_share": 0.117188, "logprob_abs_diff": 0.015202, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.350711, "clip_fraction": 0.0, "step_time_s": 268.934543, "log_probs_time_s": 65.808205, "train_time_s": 181.708735, "update_weights_time_s": 0.607471, "sleep_time_s": 0.699798, "wake_up_time_s": 0.826785, "step_end": "2026-10-18 18:36:03", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 27, "length": 498.753906, "reward_all_samples": 0.730469, "groups_all_wrong": 4.0, "groups_all_right": 18.0, "rollout_time_s": 82.699003, "rollout_end": "2026-10-18 18:37:28", "reward": 0.730469, "cut_share": 0.117188, "logprob_abs_diff": 0.014727, "train_rollout_kl": 0.000673, "entropy": 0.0, "grad_norm": 0.464033, "clip_fraction": 0.0, "step_time_s": 314.156489, "log_probs_time_s": 65.735548, "train_time_s": 227.183398, "update_weights_time_s": 0.607245, "sleep_time_s": 0.716437, "wake_up_time_s": 0.839137, "step_end": "2026-10-18 18:41:17", "rollout_paused_s": 46.0, "step_paused_s": 184.0 }, { "step": 28, "length": 512.695312, "reward_all_samples": 0.761719, "groups_all_wrong": 2.0, "groups_all_right": 13.0, "rollout_time_s": 84.839792, "rollout_end": "2026-10-18 18:42:45", "reward": 0.761719, "cut_share": 0.171875, "logprob_abs_diff": 0.014963, "train_rollout_kl": 0.000701, "entropy": 0.0, "grad_norm": 0.465453, "clip_fraction": 0.0, "step_time_s": 316.847837, "log_probs_time_s": 65.982103, "train_time_s": 227.904119, "update_weights_time_s": 0.596035, "sleep_time_s": 0.691798, "wake_up_time_s": 0.734197, "step_end": "2026-10-18 18:46:34", "rollout_paused_s": 46.0, "step_paused_s": 184.0 }, { "step": 29, "length": 573.710938, "reward_all_samples": 0.625, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 90.226788, "rollout_end": "2026-10-18 18:48:07", "reward": 0.625, "cut_share": 0.179688, "logprob_abs_diff": 0.015209, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.534689, "clip_fraction": 0.0, "step_time_s": 280.682621, "log_probs_time_s": 67.374668, "train_time_s": 186.354497, "update_weights_time_s": 0.638439, "sleep_time_s": 0.724964, "wake_up_time_s": 0.746046, "step_end": "2026-10-18 18:51:15", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 30, "length": 488.554688, "reward_all_samples": 0.828125, "groups_all_wrong": 1.0, "groups_all_right": 15.0, "rollout_time_s": 36.089931, "rollout_end": "2026-10-18 18:51:53", "reward": 0.828125, "cut_share": 0.085938, "logprob_abs_diff": 0.014287, "train_rollout_kl": 0.000654, "entropy": 0.0, "grad_norm": 0.515927, "clip_fraction": 0.0, "step_time_s": 221.661925, "log_probs_time_s": 65.462479, "train_time_s": 181.459023, "update_weights_time_s": 0.598315, "sleep_time_s": 0.704135, "wake_up_time_s": 0.853168, "step_end": "2026-10-18 18:54:56", "rollout_paused_s": 0.0, "step_paused_s": 91.0 }, { "step": 31, "length": 571.5, "reward_all_samples": 0.679688, "groups_all_wrong": 2.0, "groups_all_right": 9.0, "rollout_time_s": 136.025415, "rollout_end": "2026-10-18 18:57:15", "reward": 0.679688, "cut_share": 0.179688, "logprob_abs_diff": 0.014959, "train_rollout_kl": 0.000707, "entropy": 0.0, "grad_norm": 0.574525, "clip_fraction": 0.0, "step_time_s": 372.036102, "log_probs_time_s": 21.911744, "train_time_s": 231.931841, "update_weights_time_s": 0.60861, "sleep_time_s": 0.75032, "wake_up_time_s": 0.729985, "step_end": "2026-10-18 19:01:08", "rollout_paused_s": 90.0, "step_paused_s": 225.0 }, { "step": 32, "length": 437.546875, "reward_all_samples": 0.765625, "groups_all_wrong": 1.0, "groups_all_right": 15.0, "rollout_time_s": 30.931513, "rollout_end": "2026-10-18 19:01:42", "reward": 0.765625, "cut_share": 0.066406, "logprob_abs_diff": 0.014506, "train_rollout_kl": 0.000658, "entropy": 0.0, "grad_norm": 0.532253, "clip_fraction": 0.0, "step_time_s": 212.279124, "log_probs_time_s": 64.25824, "train_time_s": 177.20445, "update_weights_time_s": 0.623958, "sleep_time_s": 0.716964, "wake_up_time_s": 0.842786, "step_end": "2026-10-18 19:04:40", "rollout_paused_s": 0.0, "step_paused_s": 91.0 }, { "step": 33, "length": 522.359375, "reward_all_samples": 0.660156, "groups_all_wrong": 7.0, "groups_all_right": 14.0, "rollout_time_s": 85.911782, "rollout_end": "2026-10-18 19:06:09", "reward": 0.660156, "cut_share": 0.140625, "logprob_abs_diff": 0.014716, "train_rollout_kl": 0.000665, "entropy": 0.0, "grad_norm": 0.495516, "clip_fraction": 0.0, "step_time_s": 273.668308, "log_probs_time_s": 66.381575, "train_time_s": 183.707815, "update_weights_time_s": 0.602254, "sleep_time_s": 0.691983, "wake_up_time_s": 0.776379, "step_end": "2026-10-18 19:09:14", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 34, "length": 463.34375, "reward_all_samples": 0.75, "groups_all_wrong": 2.0, "groups_all_right": 15.0, "rollout_time_s": 79.201988, "rollout_end": "2026-10-18 19:10:36", "reward": 0.75, "cut_share": 0.066406, "logprob_abs_diff": 0.014686, "train_rollout_kl": 0.000672, "entropy": 0.0, "grad_norm": 0.542131, "clip_fraction": 0.0, "step_time_s": 261.594111, "log_probs_time_s": 64.767057, "train_time_s": 178.278193, "update_weights_time_s": 0.597221, "sleep_time_s": 0.704102, "wake_up_time_s": 0.736434, "step_end": "2026-10-18 19:13:36", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 35, "length": 529.988281, "reward_all_samples": 0.625, "groups_all_wrong": 7.0, "groups_all_right": 13.0, "rollout_time_s": 131.1678, "rollout_end": "2026-10-18 19:15:50", "reward": 0.625, "cut_share": 0.152344, "logprob_abs_diff": 0.015098, "train_rollout_kl": 0.000689, "entropy": 0.0, "grad_norm": 0.448962, "clip_fraction": 0.0, "step_time_s": 318.064745, "log_probs_time_s": 20.761446, "train_time_s": 182.600056, "update_weights_time_s": 0.63048, "sleep_time_s": 0.734418, "wake_up_time_s": 0.844993, "step_end": "2026-10-18 19:18:54", "rollout_paused_s": 92.0, "step_paused_s": 184.0 }, { "step": 36, "length": 480.015625, "reward_all_samples": 0.730469, "groups_all_wrong": 3.0, "groups_all_right": 15.0, "rollout_time_s": 81.89394, "rollout_end": "2026-10-18 19:20:19", "reward": 0.730469, "cut_share": 0.128906, "logprob_abs_diff": 0.01504, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.480885, "clip_fraction": 0.0, "step_time_s": 311.3561, "log_probs_time_s": 65.231459, "train_time_s": 225.153261, "update_weights_time_s": 0.60544, "sleep_time_s": 0.73819, "wake_up_time_s": 0.82688, "step_end": "2026-10-18 19:24:05", "rollout_paused_s": 46.0, "step_paused_s": 184.0 }, { "step": 37, "length": 564.8125, "reward_all_samples": 0.640625, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 89.689503, "rollout_end": "2026-10-18 19:25:37", "reward": 0.640625, "cut_share": 0.175781, "logprob_abs_diff": 0.0152, "train_rollout_kl": 0.000701, "entropy": 0.0, "grad_norm": 0.616132, "clip_fraction": 0.0, "step_time_s": 325.287729, "log_probs_time_s": 67.177624, "train_time_s": 231.6009, "update_weights_time_s": 0.606325, "sleep_time_s": 0.681515, "wake_up_time_s": 0.726427, "step_end": "2026-10-18 19:29:30", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 38, "length": 479.054688, "reward_all_samples": 0.785156, "groups_all_wrong": 3.0, "groups_all_right": 17.0, "rollout_time_s": 36.191402, "rollout_end": "2026-10-18 19:30:09", "reward": 0.785156, "cut_share": 0.101562, "logprob_abs_diff": 0.014687, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.507801, "clip_fraction": 0.0, "step_time_s": 266.567071, "log_probs_time_s": 65.471877, "train_time_s": 226.238131, "update_weights_time_s": 0.634269, "sleep_time_s": 0.749971, "wake_up_time_s": 0.756235, "step_end": "2026-10-18 19:33:57", "rollout_paused_s": 0.0, "step_paused_s": 137.0 }, { "step": 39, "length": 486.734375, "reward_all_samples": 0.71875, "groups_all_wrong": 2.0, "groups_all_right": 15.0, "rollout_time_s": 36.359297, "rollout_end": "2026-10-18 19:34:36", "reward": 0.71875, "cut_share": 0.121094, "logprob_abs_diff": 0.015184, "train_rollout_kl": 0.000726, "entropy": 0.0, "grad_norm": 0.505879, "clip_fraction": 0.0, "step_time_s": 266.5179, "log_probs_time_s": 65.385888, "train_time_s": 225.95189, "update_weights_time_s": 0.62698, "sleep_time_s": 0.754202, "wake_up_time_s": 0.859574, "step_end": "2026-10-18 19:38:24", "rollout_paused_s": 0.0, "step_paused_s": 138.0 }, { "step": 40, "length": 509.023438, "reward_all_samples": 0.738281, "groups_all_wrong": 3.0, "groups_all_right": 15.0, "rollout_time_s": 83.726477, "rollout_end": "2026-10-18 19:39:50", "reward": 0.738281, "cut_share": 0.144531, "logprob_abs_diff": 0.014945, "train_rollout_kl": 0.000667, "entropy": 0.0, "grad_norm": 0.501801, "clip_fraction": 0.0, "step_time_s": 270.192335, "log_probs_time_s": 20.493126, "train_time_s": 182.249116, "update_weights_time_s": 0.628562, "sleep_time_s": 0.73757, "wake_up_time_s": 0.757213, "step_end": "2026-10-18 19:42:54", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 41, "length": 413.195312, "reward_all_samples": 0.800781, "groups_all_wrong": 1.0, "groups_all_right": 14.0, "rollout_time_s": 74.749628, "rollout_end": "2026-10-18 19:44:11", "reward": 0.800781, "cut_share": 0.046875, "logprob_abs_diff": 0.014267, "train_rollout_kl": 0.000663, "entropy": 0.0, "grad_norm": 0.606443, "clip_fraction": 0.0, "step_time_s": 254.923167, "log_probs_time_s": 63.888781, "train_time_s": 176.107411, "update_weights_time_s": 0.633526, "sleep_time_s": 0.723599, "wake_up_time_s": 0.726308, "step_end": "2026-10-18 19:47:09", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 42, "length": 408.113281, "reward_all_samples": 0.800781, "groups_all_right": 14.0, "rollout_time_s": 74.55257, "rollout_end": "2026-10-18 19:48:26", "reward": 0.800781, "cut_share": 0.046875, "logprob_abs_diff": 0.013609, "train_rollout_kl": 0.000627, "entropy": 0.0, "grad_norm": 0.596887, "clip_fraction": 0.0, "step_time_s": 254.584961, "log_probs_time_s": 63.815096, "train_time_s": 175.767255, "update_weights_time_s": 0.600414, "sleep_time_s": 0.719392, "wake_up_time_s": 0.849089, "step_end": "2026-10-18 19:51:23", "rollout_paused_s": 45.0, "step_paused_s": 137.0 }, { "step": 43, "length": 459.605469, "reward_all_samples": 0.730469, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 79.578585, "rollout_end": "2026-10-18 19:52:46", "reward": 0.730469, "cut_share": 0.109375, "logprob_abs_diff": 0.015311, "train_rollout_kl": 0.000723, "entropy": 0.0, "grad_norm": 0.571635, "clip_fraction": 0.0, "step_time_s": 262.580027, "log_probs_time_s": 64.74513, "train_time_s": 178.716075, "update_weights_time_s": 0.596885, "sleep_time_s": 0.704026, "wake_up_time_s": 0.861019, "step_end": "2026-10-18 19:55:46", "rollout_paused_s": 46.0, "step_paused_s": 137.0 }, { "step": 44, "length": 420.011719, "reward_all_samples": 0.808594, "groups_all_wrong": 2.0, "groups_all_right": 17.0, "rollout_time_s": 29.959044, "rollout_end": "2026-10-18 19:56:19", "reward": 0.808594, "cut_share": 0.046875, "logprob_abs_diff": 0.014355, "train_rollout_kl": 0.000679, "entropy": 0.0, "grad_norm": 0.505023, "clip_fraction": 0.0, "step_time_s": 255.694533, "log_probs_time_s": 63.939519, "train_time_s": 221.546, "update_weights_time_s": 0.625672, "sleep_time_s": 0.719558, "wake_up_time_s": 0.766565, "step_end": "2026-10-18 20:00:02", "rollout_paused_s": 0.0, "step_paused_s": 137.0 }, { "step": 45, "length": 447.398438, "reward_all_samples": 0.792969, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 32.994834, "rollout_end": "2026-10-18 20:00:37", "reward": 0.792969, "cut_share": 0.097656, "logprob_abs_diff": 0.014755, "train_rollout_kl": 0.000676, "entropy": 0.0, "grad_norm": 0.537729, "clip_fraction": 0.0, "step_time_s": 216.103152, "log_probs_time_s": 64.72533, "train_time_s": 178.793168, "update_weights_time_s": 0.619464, "sleep_time_s": 0.714642, "wake_up_time_s": 0.846226, "step_end": "2026-10-18 20:03:38", "rollout_paused_s": 0.0, "step_paused_s": 91.0 }, { "step": 46, "length": 474.742188, "reward_all_samples": 0.765625, "groups_all_wrong": 2.0, "groups_all_right": 16.0, "rollout_time_s": 80.393214, "rollout_end": "2026-10-18 20:05:01", "reward": 0.765625, "cut_share": 0.078125, "logprob_abs_diff": 0.014494, "train_rollout_kl": 0.000679, "entropy": 0.0, "grad_norm": 0.901652, "clip_fraction": 0.0, "step_time_s": 264.148441, "log_probs_time_s": 65.044658, "train_time_s": 179.571846, "update_weights_time_s": 0.635985, "sleep_time_s": 0.715638, "wake_up_time_s": 0.848844, "step_end": "2026-10-18 20:08:02", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 47, "length": 513.554688, "reward_all_samples": 0.6875, "groups_all_wrong": 4.0, "groups_all_right": 13.0, "rollout_time_s": 84.795099, "rollout_end": "2026-10-18 20:09:29", "reward": 0.6875, "cut_share": 0.132812, "logprob_abs_diff": 0.014793, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.596318, "clip_fraction": 0.0, "step_time_s": 317.155423, "log_probs_time_s": 66.080422, "train_time_s": 228.026924, "update_weights_time_s": 0.640234, "sleep_time_s": 0.726995, "wake_up_time_s": 0.833739, "step_end": "2026-10-18 20:13:19", "rollout_paused_s": 46.0, "step_paused_s": 184.0 } ] }, "e6a-base": { "steps": 12, "partial": false, "median": { "reward": 0.664063, "length": 711.138672, "cut_share": 0.3125, "reward_all_samples": 0.664063, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.01526, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.412269, "clip_fraction": 0.0, "step_time_s": 309.902648, "rollout_time_s": 83.8914, "log_probs_time_s": 70.705599, "train_time_s": 198.885573, "update_weights_time_s": 0.639915, "sleep_time_s": 0.732583, "wake_up_time_s": 0.834077 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 59.145517, "steps": 6 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 799.589844, "reward_all_samples": 0.527344, "groups_all_wrong": 8.0, "groups_all_right": 6.0, "rollout_time_s": 68.602707, "rollout_end": "2026-10-21 12:44:20", "reward": 0.527344, "cut_share": 0.414062, "logprob_abs_diff": 0.015271, "train_rollout_kl": 0.000689, "entropy": 0.0, "grad_norm": 0.445129, "clip_fraction": 0.0, "step_time_s": 333.342026, "log_probs_time_s": 77.535854, "train_time_s": 212.378844, "update_weights_time_s": 0.65543, "sleep_time_s": 2.162173, "wake_up_time_s": 0.407667, "step_end": "2026-10-21 12:47:54", "rollout_paused_s": 0.0, "step_paused_s": 91.0 }, { "step": 1, "length": 699.695312, "reward_all_samples": 0.695312, "groups_all_wrong": 1.0, "groups_all_right": 14.0, "rollout_time_s": 56.70353, "rollout_end": "2026-10-21 12:48:57", "reward": 0.695312, "cut_share": 0.273438, "logprob_abs_diff": 0.015624, "train_rollout_kl": 0.000716, "entropy": 0.0, "grad_norm": 0.465367, "clip_fraction": 0.0, "step_time_s": 214.037414, "log_probs_time_s": 70.238615, "train_time_s": 149.731666, "update_weights_time_s": 0.606036, "sleep_time_s": 4.086345, "wake_up_time_s": 0.739043, "step_end": "2026-10-21 12:51:28", "rollout_paused_s": 0.0, "step_paused_s": 46.0 }, { "step": 2, "length": 764.441406, "reward_all_samples": 0.628906, "groups_all_wrong": 5.0, "groups_all_right": 8.0, "rollout_time_s": 108.681644, "rollout_end": "2026-10-21 12:53:20", "reward": 0.628906, "cut_share": 0.3125, "logprob_abs_diff": 0.015236, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.434013, "clip_fraction": 0.0, "step_time_s": 313.400895, "log_probs_time_s": 71.813061, "train_time_s": 199.980013, "update_weights_time_s": 0.638911, "sleep_time_s": 1.179615, "wake_up_time_s": 0.841096, "step_end": "2026-10-21 12:56:41", "rollout_paused_s": 45.0, "step_paused_s": 137.0 }, { "step": 3, "length": 785.648438, "reward_all_samples": 0.648438, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 66.42962, "rollout_end": "2026-10-21 12:57:51", "reward": 0.648438, "cut_share": 0.375, "logprob_abs_diff": 0.014961, "train_rollout_kl": 0.000701, "entropy": 0.0, "grad_norm": 0.405877, "clip_fraction": 0.0, "step_time_s": 319.299013, "log_probs_time_s": 72.53325, "train_time_s": 248.160932, "update_weights_time_s": 0.62847, "sleep_time_s": 1.231346, "wake_up_time_s": 0.749585, "step_end": "2026-10-21 13:02:00", "rollout_paused_s": 0.0, "step_paused_s": 137.0 }, { "step": 4, "length": 685.25, "reward_all_samples": 0.734375, "groups_all_right": 16.0, "rollout_time_s": 55.791397, "rollout_end": "2026-10-21 13:02:59", "reward": 0.734375, "cut_share": 0.261719, "logprob_abs_diff": 0.015005, "train_rollout_kl": 0.000692, "entropy": 0.0, "grad_norm": 0.404694, "clip_fraction": 0.0, "step_time_s": 254.202335, "log_probs_time_s": 69.797034, "train_time_s": 194.102849, "update_weights_time_s": 0.643977, "sleep_time_s": 0.695079, "wake_up_time_s": 0.74961, "step_end": "2026-10-21 13:06:15", "rollout_paused_s": 0.0, "step_paused_s": 91.0 }, { "step": 5, "length": 711.746094, "reward_all_samples": 0.625, "groups_all_wrong": 7.0, "groups_all_right": 12.0, "rollout_time_s": 104.236274, "rollout_end": "2026-10-21 13:08:02", "reward": 0.625, "cut_share": 0.339844, "logprob_abs_diff": 0.015939, "train_rollout_kl": 0.000732, "entropy": 0.0, "grad_norm": 0.378897, "clip_fraction": 0.0, "step_time_s": 350.935813, "log_probs_time_s": 70.682469, "train_time_s": 242.275716, "update_weights_time_s": 0.643839, "sleep_time_s": 0.732948, "wake_up_time_s": 0.830999, "step_end": "2026-10-21 13:12:05", "rollout_paused_s": 46.0, "step_paused_s": 184.0 }, { "step": 6, "length": 719.476562, "reward_all_samples": 0.601562, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 105.066137, "rollout_end": "2026-10-21 13:13:53", "reward": 0.601562, "cut_share": 0.328125, "logprob_abs_diff": 0.015249, "train_rollout_kl": 0.00069, "entropy": 0.0, "grad_norm": 0.418661, "clip_fraction": 0.0, "step_time_s": 306.404401, "log_probs_time_s": 70.758341, "train_time_s": 196.926513, "update_weights_time_s": 0.64345, "sleep_time_s": 0.721664, "wake_up_time_s": 0.841015, "step_end": "2026-10-21 13:17:12", "rollout_paused_s": 45.0, "step_paused_s": 137.0 }, { "step": 7, "length": 739.429688, "reward_all_samples": 0.636719, "groups_all_wrong": 6.0, "groups_all_right": 13.0, "rollout_time_s": 61.587505, "rollout_end": "2026-10-21 13:18:16", "reward": 0.636719, "cut_share": 0.359375, "logprob_abs_diff": 0.015542, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.357419, "clip_fraction": 0.0, "step_time_s": 263.72269, "log_probs_time_s": 71.139157, "train_time_s": 197.791133, "update_weights_time_s": 0.608306, "sleep_time_s": 0.693529, "wake_up_time_s": 0.853953, "step_end": "2026-10-21 13:21:36", "rollout_paused_s": 0.0, "step_paused_s": 92.0 }, { "step": 8, "length": 710.53125, "reward_all_samples": 0.683594, "groups_all_wrong": 4.0, "groups_all_right": 14.0, "rollout_time_s": 103.778202, "rollout_end": "2026-10-21 13:23:22", "reward": 0.683594, "cut_share": 0.3125, "logprob_abs_diff": 0.015565, "train_rollout_kl": 0.000705, "entropy": 0.0, "grad_norm": 0.399646, "clip_fraction": 0.0, "step_time_s": 350.169724, "log_probs_time_s": 70.728729, "train_time_s": 242.083571, "update_weights_time_s": 0.640919, "sleep_time_s": 0.720725, "wake_up_time_s": 0.769053, "step_end": "2026-10-21 13:27:26", "rollout_paused_s": 46.0, "step_paused_s": 184.0 }, { "step": 9, "length": 680.03125, "reward_all_samples": 0.722656, "groups_all_wrong": 1.0, "groups_all_right": 12.0, "rollout_time_s": 99.852728, "rollout_end": "2026-10-21 13:29:09", "reward": 0.722656, "cut_share": 0.246094, "logprob_abs_diff": 0.015288, "train_rollout_kl": 0.00071, "entropy": 0.0, "grad_norm": 0.464533, "clip_fraction": 0.0, "step_time_s": 297.911535, "log_probs_time_s": 69.778775, "train_time_s": 193.612554, "update_weights_time_s": 0.644573, "sleep_time_s": 0.750042, "wake_up_time_s": 0.854498, "step_end": "2026-10-21 13:32:24", "rollout_paused_s": 45.0, "step_paused_s": 137.0 }, { "step": 10, "length": 694.914062, "reward_all_samples": 0.679688, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 56.318288, "rollout_end": "2026-10-21 13:33:23", "reward": 0.679688, "cut_share": 0.234375, "logprob_abs_diff": 0.015213, "train_rollout_kl": 0.000718, "entropy": 0.0, "grad_norm": 0.618469, "clip_fraction": 0.0, "step_time_s": 256.547086, "log_probs_time_s": 70.343644, "train_time_s": 195.787758, "update_weights_time_s": 0.632294, "sleep_time_s": 0.732218, "wake_up_time_s": 0.850572, "step_end": "2026-10-21 13:36:40", "rollout_paused_s": 0.0, "step_paused_s": 92.0 }, { "step": 11, "length": 666.710938, "reward_all_samples": 0.703125, "groups_all_wrong": 5.0, "groups_all_right": 17.0, "rollout_time_s": 99.180093, "rollout_end": "2026-10-21 13:38:22", "reward": 0.703125, "cut_share": 0.253906, "logprob_abs_diff": 0.01501, "train_rollout_kl": 0.000689, "entropy": 0.0, "grad_norm": 0.399845, "clip_fraction": 0.0, "step_time_s": 341.677632, "log_probs_time_s": 69.574996, "train_time_s": 238.083864, "update_weights_time_s": 0.635, "sleep_time_s": 0.730788, "wake_up_time_s": 0.837155, "step_end": "2026-10-21 13:42:22", "rollout_paused_s": 46.0, "step_paused_s": 184.0 } ] }, "e6b-det-triton": { "steps": 12, "partial": false, "median": { "reward": 0.636718, "length": 717.550781, "cut_share": 0.324219, "reward_all_samples": 0.636718, "groups_all_wrong": 4.0, "groups_all_right": 10.5, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015182, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.440292, "clip_fraction": 0.0, "step_time_s": 332.642004, "rollout_time_s": 84.54196, "log_probs_time_s": 70.485603, "train_time_s": 221.837525, "update_weights_time_s": 0.621104, "sleep_time_s": 0.748587, "wake_up_time_s": 0.770688 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 81.95139, "steps": 8 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 797.996094, "reward_all_samples": 0.496094, "groups_all_wrong": 9.0, "groups_all_right": 5.0, "rollout_time_s": 97.45813, "rollout_end": "2026-10-21 13:48:12", "reward": 0.496094, "cut_share": 0.453125, "logprob_abs_diff": 0.015404, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.415838, "clip_fraction": 0.0, "step_time_s": 412.199546, "log_probs_time_s": 76.756903, "train_time_s": 256.667198, "update_weights_time_s": 0.628479, "sleep_time_s": 2.156437, "wake_up_time_s": 0.458705, "step_end": "2026-10-21 13:52:30", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 1, "length": 726.34375, "reward_all_samples": 0.644531, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 83.852392, "rollout_end": "2026-10-21 13:54:00", "reward": 0.644531, "cut_share": 0.316406, "logprob_abs_diff": 0.015387, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.440843, "clip_fraction": 0.0, "step_time_s": 334.266752, "log_probs_time_s": 71.034831, "train_time_s": 242.283514, "update_weights_time_s": 0.636824, "sleep_time_s": 4.519573, "wake_up_time_s": 0.851636, "step_end": "2026-10-21 13:58:04", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 2, "length": 769.074219, "reward_all_samples": 0.628906, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 134.826757, "rollout_end": "2026-10-21 14:00:22", "reward": 0.628906, "cut_share": 0.332031, "logprob_abs_diff": 0.015454, "train_rollout_kl": 0.000708, "entropy": 0.0, "grad_norm": 0.457364, "clip_fraction": 0.0, "step_time_s": 385.249926, "log_probs_time_s": 72.042618, "train_time_s": 245.622247, "update_weights_time_s": 0.603611, "sleep_time_s": 1.139963, "wake_up_time_s": 0.860167, "step_end": "2026-10-21 14:04:29", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 3, "length": 782.273438, "reward_all_samples": 0.597656, "groups_all_wrong": 4.0, "groups_all_right": 7.0, "rollout_time_s": 137.118907, "rollout_end": "2026-10-21 14:06:50", "reward": 0.597656, "cut_share": 0.375, "logprob_abs_diff": 0.014653, "train_rollout_kl": 0.000663, "entropy": 0.0, "grad_norm": 0.439741, "clip_fraction": 0.0, "step_time_s": 344.572796, "log_probs_time_s": 72.53891, "train_time_s": 202.739726, "update_weights_time_s": 0.605593, "sleep_time_s": 1.147929, "wake_up_time_s": 0.758125, "step_end": "2026-10-21 14:10:14", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 4, "length": 700.402344, "reward_all_samples": 0.726562, "groups_all_wrong": 1.0, "groups_all_right": 12.0, "rollout_time_s": 80.050388, "rollout_end": "2026-10-21 14:12:22", "reward": 0.726562, "cut_share": 0.242188, "logprob_abs_diff": 0.014927, "train_rollout_kl": 0.000675, "entropy": 0.0, "grad_norm": 0.446586, "clip_fraction": 0.0, "step_time_s": 370.532864, "log_probs_time_s": 70.361377, "train_time_s": 240.935324, "update_weights_time_s": 0.613729, "sleep_time_s": 46.017773, "wake_up_time_s": 0.853792, "step_end": "2026-10-21 14:16:24", "rollout_paused_s": 0.0, "step_paused_s": 181.0 }, { "step": 5, "length": 726.753906, "reward_all_samples": 0.59375, "groups_all_wrong": 8.0, "groups_all_right": 11.0, "rollout_time_s": 85.015617, "rollout_end": "2026-10-21 14:17:52", "reward": 0.59375, "cut_share": 0.375, "logprob_abs_diff": 0.015796, "train_rollout_kl": 0.000718, "entropy": 0.0, "grad_norm": 0.361157, "clip_fraction": 0.0, "step_time_s": 287.435263, "log_probs_time_s": 25.868375, "train_time_s": 198.201598, "update_weights_time_s": 0.607031, "sleep_time_s": 0.694053, "wake_up_time_s": 0.755654, "step_end": "2026-10-21 14:21:12", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 6, "length": 713.785156, "reward_all_samples": 0.617188, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 128.467611, "rollout_end": "2026-10-21 14:23:23", "reward": 0.617188, "cut_share": 0.335938, "logprob_abs_diff": 0.014886, "train_rollout_kl": 0.000672, "entropy": 0.0, "grad_norm": 0.461092, "clip_fraction": 0.0, "step_time_s": 374.6276, "log_probs_time_s": 25.287607, "train_time_s": 241.932986, "update_weights_time_s": 0.610269, "sleep_time_s": 0.740666, "wake_up_time_s": 0.824879, "step_end": "2026-10-21 14:27:26", "rollout_paused_s": 46.0, "step_paused_s": 183.0 }, { "step": 7, "length": 721.316406, "reward_all_samples": 0.621094, "groups_all_wrong": 7.0, "groups_all_right": 11.0, "rollout_time_s": 84.068303, "rollout_end": "2026-10-21 14:28:53", "reward": 0.621094, "cut_share": 0.355469, "logprob_abs_diff": 0.015627, "train_rollout_kl": 0.000711, "entropy": 0.0, "grad_norm": 0.428072, "clip_fraction": 0.0, "step_time_s": 331.017256, "log_probs_time_s": 70.871988, "train_time_s": 242.830801, "update_weights_time_s": 0.608578, "sleep_time_s": 0.696454, "wake_up_time_s": 0.76492, "step_end": "2026-10-21 14:32:57", "rollout_paused_s": 0.0, "step_paused_s": 138.0 }, { "step": 8, "length": 675.65625, "reward_all_samples": 0.695312, "groups_all_wrong": 4.0, "groups_all_right": 15.0, "rollout_time_s": 122.34904, "rollout_end": "2026-10-21 14:35:02", "reward": 0.695312, "cut_share": 0.269531, "logprob_abs_diff": 0.015522, "train_rollout_kl": 0.000721, "entropy": 0.0, "grad_norm": 0.404928, "clip_fraction": 0.0, "step_time_s": 321.152141, "log_probs_time_s": 69.91311, "train_time_s": 194.574049, "update_weights_time_s": 0.656323, "sleep_time_s": 0.732933, "wake_up_time_s": 0.776456, "step_end": "2026-10-21 14:38:18", "rollout_paused_s": 46.0, "step_paused_s": 138.0 }, { "step": 9, "length": 678.351562, "reward_all_samples": 0.734375, "groups_all_wrong": 2.0, "groups_all_right": 14.0, "rollout_time_s": 77.345819, "rollout_end": "2026-10-21 14:39:39", "reward": 0.734375, "cut_share": 0.210938, "logprob_abs_diff": 0.014958, "train_rollout_kl": 0.000683, "entropy": 0.0, "grad_norm": 0.4046, "clip_fraction": 0.0, "step_time_s": 275.772091, "log_probs_time_s": 69.72932, "train_time_s": 194.089224, "update_weights_time_s": 0.631259, "sleep_time_s": 0.737828, "wake_up_time_s": 0.734516, "step_end": "2026-10-21 14:42:54", "rollout_paused_s": 0.0, "step_paused_s": 92.0 }, { "step": 10, "length": 696.925781, "reward_all_samples": 0.671875, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 79.371989, "rollout_end": "2026-10-21 14:44:16", "reward": 0.671875, "cut_share": 0.226562, "logprob_abs_diff": 0.014976, "train_rollout_kl": 0.000698, "entropy": 0.0, "grad_norm": 0.510611, "clip_fraction": 0.0, "step_time_s": 280.308884, "log_probs_time_s": 70.609829, "train_time_s": 196.548924, "update_weights_time_s": 0.634972, "sleep_time_s": 0.708501, "wake_up_time_s": 0.854855, "step_end": "2026-10-21 14:47:35", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 11, "length": 645.628906, "reward_all_samples": 0.703125, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 73.211995, "rollout_end": "2026-10-21 14:48:51", "reward": 0.703125, "cut_share": 0.203125, "logprob_abs_diff": 0.014756, "train_rollout_kl": 0.000675, "entropy": 0.0, "grad_norm": 0.516273, "clip_fraction": 0.0, "step_time_s": 269.249516, "log_probs_time_s": 23.661317, "train_time_s": 191.784944, "update_weights_time_s": 0.6465, "sleep_time_s": 0.756508, "wake_up_time_s": 0.744761, "step_end": "2026-10-21 14:52:04", "rollout_paused_s": 0.0, "step_paused_s": 90.0 } ] }, "e6c-det-flashinfer": { "steps": 12, "partial": false, "median": { "reward": 0.646485, "length": 714.72461, "cut_share": 0.318359, "reward_all_samples": 0.646485, "groups_all_wrong": 6.0, "groups_all_right": 10.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015204, "train_rollout_kl": 0.00069, "entropy": 0.0, "grad_norm": 0.43493, "clip_fraction": 0.0, "step_time_s": 322.090524, "rollout_time_s": 87.301295, "log_probs_time_s": 70.409177, "train_time_s": 238.690878, "update_weights_time_s": 0.626143, "sleep_time_s": 0.726511, "wake_up_time_s": 0.841942 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 77.343458, "steps": 7 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 803.917969, "reward_all_samples": 0.519531, "groups_all_wrong": 7.0, "groups_all_right": 7.0, "rollout_time_s": 95.042311, "rollout_end": "2026-10-21 14:57:50", "reward": 0.519531, "cut_share": 0.457031, "logprob_abs_diff": 0.01532, "train_rollout_kl": 0.00069, "entropy": 0.0, "grad_norm": 0.402813, "clip_fraction": 0.0, "step_time_s": 363.532846, "log_probs_time_s": 76.707283, "train_time_s": 211.257826, "update_weights_time_s": 0.613508, "sleep_time_s": 2.040443, "wake_up_time_s": 0.443776, "step_end": "2026-10-21 15:01:23", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 1, "length": 718.84375, "reward_all_samples": 0.6875, "groups_all_wrong": 1.0, "groups_all_right": 14.0, "rollout_time_s": 79.560279, "rollout_end": "2026-10-21 15:02:49", "reward": 0.6875, "cut_share": 0.269531, "logprob_abs_diff": 0.015118, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.416472, "clip_fraction": 0.0, "step_time_s": 238.777737, "log_probs_time_s": 25.376099, "train_time_s": 151.658698, "update_weights_time_s": 0.65186, "sleep_time_s": 4.08722, "wake_up_time_s": 0.740325, "step_end": "2026-10-21 15:05:22", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 2, "length": 754.105469, "reward_all_samples": 0.621094, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 129.574253, "rollout_end": "2026-10-21 15:07:34", "reward": 0.621094, "cut_share": 0.316406, "logprob_abs_diff": 0.015322, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.451486, "clip_fraction": 0.0, "step_time_s": 378.861819, "log_probs_time_s": 71.633242, "train_time_s": 244.576471, "update_weights_time_s": 0.608202, "sleep_time_s": 1.072887, "wake_up_time_s": 0.853181, "step_end": "2026-10-21 15:11:40", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 3, "length": 785.152344, "reward_all_samples": 0.585938, "groups_all_wrong": 6.0, "groups_all_right": 10.0, "rollout_time_s": 135.154401, "rollout_end": "2026-10-21 15:13:59", "reward": 0.585938, "cut_share": 0.390625, "logprob_abs_diff": 0.014563, "train_rollout_kl": 0.000651, "entropy": 0.0, "grad_norm": 0.396257, "clip_fraction": 0.0, "step_time_s": 342.352985, "log_probs_time_s": 72.484656, "train_time_s": 202.354389, "update_weights_time_s": 0.634826, "sleep_time_s": 1.149464, "wake_up_time_s": 0.851487, "step_end": "2026-10-21 15:17:23", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 4, "length": 710.605469, "reward_all_samples": 0.703125, "groups_all_right": 10.0, "rollout_time_s": 78.661913, "rollout_end": "2026-10-21 15:18:44", "reward": 0.703125, "cut_share": 0.296875, "logprob_abs_diff": 0.015008, "train_rollout_kl": 0.000674, "entropy": 0.0, "grad_norm": 0.482674, "clip_fraction": 0.0, "step_time_s": 323.793881, "log_probs_time_s": 70.550449, "train_time_s": 240.842702, "update_weights_time_s": 0.646358, "sleep_time_s": 0.696243, "wake_up_time_s": 0.851974, "step_end": "2026-10-21 15:22:47", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 5, "length": 696.480469, "reward_all_samples": 0.636719, "groups_all_wrong": 6.0, "groups_all_right": 10.0, "rollout_time_s": 77.343458, "rollout_end": "2026-10-21 15:24:07", "reward": 0.636719, "cut_share": 0.324219, "logprob_abs_diff": 0.015693, "train_rollout_kl": 0.00072, "entropy": 0.0, "grad_norm": 0.442103, "clip_fraction": 0.0, "step_time_s": 276.643451, "log_probs_time_s": 70.267906, "train_time_s": 194.99784, "update_weights_time_s": 0.636862, "sleep_time_s": 0.715806, "wake_up_time_s": 0.850396, "step_end": "2026-10-21 15:27:23", "rollout_paused_s": 0.0, "step_paused_s": 91.0 }, { "step": 6, "length": 724.753906, "reward_all_samples": 0.578125, "groups_all_wrong": 6.0, "groups_all_right": 8.0, "rollout_time_s": 126.37808, "rollout_end": "2026-10-21 15:29:33", "reward": 0.578125, "cut_share": 0.332031, "logprob_abs_diff": 0.015061, "train_rollout_kl": 0.000676, "entropy": 0.0, "grad_norm": 0.494034, "clip_fraction": 0.0, "step_time_s": 372.8516, "log_probs_time_s": 70.87712, "train_time_s": 242.144518, "update_weights_time_s": 0.646327, "sleep_time_s": 0.726427, "wake_up_time_s": 0.745385, "step_end": "2026-10-21 15:33:36", "rollout_paused_s": 45.0, "step_paused_s": 182.0 }, { "step": 7, "length": 720.050781, "reward_all_samples": 0.628906, "groups_all_wrong": 6.0, "groups_all_right": 11.0, "rollout_time_s": 126.53523, "rollout_end": "2026-10-21 15:35:45", "reward": 0.628906, "cut_share": 0.339844, "logprob_abs_diff": 0.015419, "train_rollout_kl": 0.000698, "entropy": 0.0, "grad_norm": 0.411148, "clip_fraction": 0.0, "step_time_s": 372.650564, "log_probs_time_s": 70.626195, "train_time_s": 241.924006, "update_weights_time_s": 0.61746, "sleep_time_s": 0.71367, "wake_up_time_s": 0.750758, "step_end": "2026-10-21 15:39:49", "rollout_paused_s": 45.0, "step_paused_s": 182.0 }, { "step": 8, "length": 682.738281, "reward_all_samples": 0.65625, "groups_all_wrong": 6.0, "groups_all_right": 15.0, "rollout_time_s": 76.194952, "rollout_end": "2026-10-21 15:41:08", "reward": 0.65625, "cut_share": 0.320312, "logprob_abs_diff": 0.015317, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.416328, "clip_fraction": 0.0, "step_time_s": 320.387167, "log_probs_time_s": 70.054533, "train_time_s": 239.884954, "update_weights_time_s": 0.609158, "sleep_time_s": 0.711408, "wake_up_time_s": 0.764232, "step_end": "2026-10-21 15:45:09", "rollout_paused_s": 0.0, "step_paused_s": 136.0 }, { "step": 9, "length": 660.175781, "reward_all_samples": 0.71875, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 71.996209, "rollout_end": "2026-10-21 15:46:24", "reward": 0.71875, "cut_share": 0.234375, "logprob_abs_diff": 0.015291, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.427756, "clip_fraction": 0.0, "step_time_s": 313.904515, "log_probs_time_s": 69.298782, "train_time_s": 237.496803, "update_weights_time_s": 0.635694, "sleep_time_s": 0.741395, "wake_up_time_s": 0.841537, "step_end": "2026-10-21 15:50:23", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 10, "length": 678.5625, "reward_all_samples": 0.664062, "groups_all_wrong": 4.0, "groups_all_right": 8.0, "rollout_time_s": 75.023912, "rollout_end": "2026-10-21 15:51:41", "reward": 0.664062, "cut_share": 0.246094, "logprob_abs_diff": 0.015015, "train_rollout_kl": 0.000685, "entropy": 0.0, "grad_norm": 0.536493, "clip_fraction": 0.0, "step_time_s": 320.038815, "log_probs_time_s": 70.093228, "train_time_s": 240.725787, "update_weights_time_s": 0.608119, "sleep_time_s": 0.726594, "wake_up_time_s": 0.856655, "step_end": "2026-10-21 15:55:43", "rollout_paused_s": 0.0, "step_paused_s": 136.0 }, { "step": 11, "length": 641.589844, "reward_all_samples": 0.714844, "groups_all_wrong": 2.0, "groups_all_right": 13.0, "rollout_time_s": 115.54314, "rollout_end": "2026-10-21 15:57:41", "reward": 0.714844, "cut_share": 0.214844, "logprob_abs_diff": 0.014795, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.555996, "clip_fraction": 0.0, "step_time_s": 311.369043, "log_probs_time_s": 68.985171, "train_time_s": 191.458873, "update_weights_time_s": 0.612548, "sleep_time_s": 0.723311, "wake_up_time_s": 0.842346, "step_end": "2026-10-21 16:00:54", "rollout_paused_s": 45.0, "step_paused_s": 136.0 } ] }, "e6d-det-triton-repeat": { "steps": 12, "partial": false, "median": { "reward": 0.658203, "length": 704.06836, "cut_share": 0.289063, "reward_all_samples": 0.658203, "groups_all_wrong": 5.0, "groups_all_right": 10.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015147, "train_rollout_kl": 0.00069, "entropy": 0.0, "grad_norm": 0.464575, "clip_fraction": 0.0, "step_time_s": 315.977817, "rollout_time_s": 94.18865, "log_probs_time_s": 70.528448, "train_time_s": 199.432238, "update_weights_time_s": 0.637862, "sleep_time_s": 0.754619, "wake_up_time_s": 0.760007 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 81.860867, "steps": 7 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 797.996094, "reward_all_samples": 0.496094, "groups_all_wrong": 9.0, "groups_all_right": 5.0, "rollout_time_s": 97.491646, "rollout_end": "2026-10-21 16:06:46", "reward": 0.496094, "cut_share": 0.453125, "logprob_abs_diff": 0.015404, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.415758, "clip_fraction": 0.0, "step_time_s": 412.930723, "log_probs_time_s": 76.911588, "train_time_s": 256.945686, "update_weights_time_s": 0.609174, "sleep_time_s": 2.01289, "wake_up_time_s": 0.422605, "step_end": "2026-10-21 16:11:04", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 1, "length": 701.277344, "reward_all_samples": 0.710938, "groups_all_right": 10.0, "rollout_time_s": 80.355659, "rollout_end": "2026-10-21 16:12:31", "reward": 0.710938, "cut_share": 0.238281, "logprob_abs_diff": 0.015189, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.53062, "clip_fraction": 0.0, "step_time_s": 238.4666, "log_probs_time_s": 25.007112, "train_time_s": 149.787028, "update_weights_time_s": 0.64222, "sleep_time_s": 4.805743, "wake_up_time_s": 0.818562, "step_end": "2026-10-21 16:15:02", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 2, "length": 760.488281, "reward_all_samples": 0.632812, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 133.205171, "rollout_end": "2026-10-21 16:17:19", "reward": 0.632812, "cut_share": 0.304688, "logprob_abs_diff": 0.015489, "train_rollout_kl": 0.000707, "entropy": 0.0, "grad_norm": 0.48161, "clip_fraction": 0.0, "step_time_s": 382.951359, "log_probs_time_s": 71.81861, "train_time_s": 244.936985, "update_weights_time_s": 0.642874, "sleep_time_s": 1.226431, "wake_up_time_s": 0.743708, "step_end": "2026-10-21 16:21:25", "rollout_paused_s": 46.0, "step_paused_s": 182.0 }, { "step": 3, "length": 773.0625, "reward_all_samples": 0.628906, "groups_all_wrong": 4.0, "groups_all_right": 8.0, "rollout_time_s": 90.885655, "rollout_end": "2026-10-21 16:22:59", "reward": 0.628906, "cut_share": 0.375, "logprob_abs_diff": 0.014819, "train_rollout_kl": 0.000683, "entropy": 0.0, "grad_norm": 0.455845, "clip_fraction": 0.0, "step_time_s": 297.394008, "log_probs_time_s": 72.279896, "train_time_s": 201.773166, "update_weights_time_s": 0.643905, "sleep_time_s": 1.156767, "wake_up_time_s": 0.770599, "step_end": "2026-10-21 16:26:23", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 4, "length": 682.683594, "reward_all_samples": 0.726562, "groups_all_wrong": 1.0, "groups_all_right": 13.0, "rollout_time_s": 77.698673, "rollout_end": "2026-10-21 16:27:43", "reward": 0.726562, "cut_share": 0.265625, "logprob_abs_diff": 0.014915, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.449192, "clip_fraction": 0.0, "step_time_s": 275.918969, "log_probs_time_s": 69.786988, "train_time_s": 194.032917, "update_weights_time_s": 0.638756, "sleep_time_s": 0.725642, "wake_up_time_s": 0.75299, "step_end": "2026-10-21 16:30:59", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 5, "length": 705.929688, "reward_all_samples": 0.628906, "groups_all_wrong": 7.0, "groups_all_right": 12.0, "rollout_time_s": 81.860867, "rollout_end": "2026-10-21 16:32:23", "reward": 0.628906, "cut_share": 0.328125, "logprob_abs_diff": 0.015558, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.402908, "clip_fraction": 0.0, "step_time_s": 282.622415, "log_probs_time_s": 70.599901, "train_time_s": 196.606621, "update_weights_time_s": 0.609586, "sleep_time_s": 0.699489, "wake_up_time_s": 0.767024, "step_end": "2026-10-21 16:35:41", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 6, "length": 714.867188, "reward_all_samples": 0.589844, "groups_all_wrong": 6.0, "groups_all_right": 8.0, "rollout_time_s": 83.339681, "rollout_end": "2026-10-21 16:37:07", "reward": 0.589844, "cut_share": 0.347656, "logprob_abs_diff": 0.015106, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.465891, "clip_fraction": 0.0, "step_time_s": 284.758304, "log_probs_time_s": 70.795149, "train_time_s": 197.091311, "update_weights_time_s": 0.644024, "sleep_time_s": 0.754592, "wake_up_time_s": 0.839846, "step_end": "2026-10-21 16:40:26", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 7, "length": 727.09375, "reward_all_samples": 0.628906, "groups_all_wrong": 6.0, "groups_all_right": 10.0, "rollout_time_s": 129.961542, "rollout_end": "2026-10-21 16:42:39", "reward": 0.628906, "cut_share": 0.363281, "logprob_abs_diff": 0.015803, "train_rollout_kl": 0.000722, "entropy": 0.0, "grad_norm": 0.463879, "clip_fraction": 0.0, "step_time_s": 377.076852, "log_probs_time_s": 70.969614, "train_time_s": 242.815269, "update_weights_time_s": 0.628188, "sleep_time_s": 0.711721, "wake_up_time_s": 0.864049, "step_end": "2026-10-21 16:46:43", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 8, "length": 667.339844, "reward_all_samples": 0.6875, "groups_all_wrong": 5.0, "groups_all_right": 15.0, "rollout_time_s": 121.283223, "rollout_end": "2026-10-21 16:48:47", "reward": 0.6875, "cut_share": 0.273438, "logprob_abs_diff": 0.015106, "train_rollout_kl": 0.000669, "entropy": 0.0, "grad_norm": 0.381603, "clip_fraction": 0.0, "step_time_s": 319.400239, "log_probs_time_s": 69.719682, "train_time_s": 193.865019, "update_weights_time_s": 0.607266, "sleep_time_s": 0.695741, "wake_up_time_s": 0.738723, "step_end": "2026-10-21 16:52:02", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 9, "length": 672.878906, "reward_all_samples": 0.726562, "groups_all_wrong": 3.0, "groups_all_right": 11.0, "rollout_time_s": 121.928406, "rollout_end": "2026-10-21 16:54:07", "reward": 0.726562, "cut_share": 0.242188, "logprob_abs_diff": 0.01503, "train_rollout_kl": 0.00069, "entropy": 0.0, "grad_norm": 0.470384, "clip_fraction": 0.0, "step_time_s": 365.699343, "log_probs_time_s": 69.861362, "train_time_s": 239.469564, "update_weights_time_s": 0.608782, "sleep_time_s": 0.742961, "wake_up_time_s": 0.747656, "step_end": "2026-10-21 16:58:08", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 10, "length": 702.207031, "reward_all_samples": 0.703125, "groups_all_wrong": 5.0, "groups_all_right": 10.0, "rollout_time_s": 80.362041, "rollout_end": "2026-10-21 16:59:31", "reward": 0.703125, "cut_share": 0.25, "logprob_abs_diff": 0.015325, "train_rollout_kl": 0.000715, "entropy": 0.0, "grad_norm": 0.46527, "clip_fraction": 0.0, "step_time_s": 326.146424, "log_probs_time_s": 70.456995, "train_time_s": 241.44295, "update_weights_time_s": 0.636969, "sleep_time_s": 0.756215, "wake_up_time_s": 0.872165, "step_end": "2026-10-21 17:03:34", "rollout_paused_s": 0.0, "step_paused_s": 136.0 }, { "step": 11, "length": 637.023438, "reward_all_samples": 0.683594, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "rollout_time_s": 117.398197, "rollout_end": "2026-10-21 17:05:35", "reward": 0.683594, "cut_share": 0.242188, "logprob_abs_diff": 0.014814, "train_rollout_kl": 0.000683, "entropy": 0.0, "grad_norm": 0.513235, "clip_fraction": 0.0, "step_time_s": 312.555394, "log_probs_time_s": 68.905811, "train_time_s": 190.826867, "update_weights_time_s": 0.653086, "sleep_time_s": 0.754647, "wake_up_time_s": 0.744893, "step_end": "2026-10-21 17:08:47", "rollout_paused_s": 45.0, "step_paused_s": 135.0 } ] }, "e11-topp-topk": { "steps": 12, "partial": false, "median": { "reward": 0.679687, "length": 693.253907, "cut_share": 0.253906, "reward_all_samples": 0.679687, "groups_all_wrong": 4.0, "groups_all_right": 11.5, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.013503, "train_rollout_kl": 0.000594, "entropy": 0.0, "grad_norm": 0.389593, "clip_fraction": 0.0, "step_time_s": 266.616134, "rollout_time_s": 74.52995, "log_probs_time_s": 71.667465, "train_time_s": 176.053344, "update_weights_time_s": 0.62589, "sleep_time_s": 0.749736, "wake_up_time_s": 0.829136 }, "median_unpaused": { "step_time_s": { "median": 188.924586, "steps": 1 }, "rollout_time_s": { "median": 66.81487, "steps": 7 }, "log_probs_time_s": { "median": 27.730106, "steps": 1 }, "train_time_s": { "median": 113.556193, "steps": 1 }, "update_weights_time_s": { "median": 0.641821, "steps": 1 }, "sleep_time_s": { "median": 1.180037, "steps": 1 }, "wake_up_time_s": { "median": 0.843302, "steps": 1 } }, "per_step": [ { "step": 0, "length": 785.554688, "reward_all_samples": 0.578125, "groups_all_wrong": 5.0, "groups_all_right": 7.0, "rollout_time_s": 78.45305, "rollout_end": "2026-10-21 17:14:18", "reward": 0.578125, "cut_share": 0.398438, "logprob_abs_diff": 0.013574, "train_rollout_kl": 0.000601, "entropy": 0.0, "grad_norm": 0.405032, "clip_fraction": 0.0, "step_time_s": 354.240754, "log_probs_time_s": 78.853609, "train_time_s": 217.854948, "update_weights_time_s": 0.635685, "sleep_time_s": 2.134514, "wake_up_time_s": 0.458833, "step_end": "2026-10-21 17:17:57", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 1, "length": 688.984375, "reward_all_samples": 0.707031, "groups_all_right": 10.0, "rollout_time_s": 64.767511, "rollout_end": "2026-10-21 17:19:09", "reward": 0.707031, "cut_share": 0.25, "logprob_abs_diff": 0.013479, "train_rollout_kl": 0.000593, "entropy": 0.0, "grad_norm": 0.46582, "clip_fraction": 0.0, "step_time_s": 227.483525, "log_probs_time_s": 71.774455, "train_time_s": 154.858414, "update_weights_time_s": 0.604893, "sleep_time_s": 4.350461, "wake_up_time_s": 0.820666, "step_end": "2026-10-21 17:21:45", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 2, "length": 736.578125, "reward_all_samples": 0.660156, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 70.606849, "rollout_end": "2026-10-21 17:22:59", "reward": 0.660156, "cut_share": 0.246094, "logprob_abs_diff": 0.013545, "train_rollout_kl": 0.000595, "entropy": 0.0, "grad_norm": 0.374154, "clip_fraction": 0.0, "step_time_s": 188.924586, "log_probs_time_s": 27.730106, "train_time_s": 113.556193, "update_weights_time_s": 0.641821, "sleep_time_s": 1.180037, "wake_up_time_s": 0.843302, "step_end": "2026-10-21 17:24:54", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 3, "length": 762.859375, "reward_all_samples": 0.65625, "groups_all_wrong": 6.0, "groups_all_right": 11.0, "rollout_time_s": 164.761663, "rollout_end": "2026-10-21 17:27:42", "reward": 0.65625, "cut_share": 0.320312, "logprob_abs_diff": 0.012778, "train_rollout_kl": 0.000565, "entropy": 0.0, "grad_norm": 0.346096, "clip_fraction": 0.0, "step_time_s": 421.619535, "log_probs_time_s": 73.783756, "train_time_s": 252.08517, "update_weights_time_s": 0.607555, "sleep_time_s": 1.198721, "wake_up_time_s": 0.750541, "step_end": "2026-10-21 17:31:55", "rollout_paused_s": 90.0, "step_paused_s": 225.0 }, { "step": 4, "length": 663.84375, "reward_all_samples": 0.761719, "groups_all_right": 12.0, "rollout_time_s": 108.69013, "rollout_end": "2026-10-21 17:33:47", "reward": 0.761719, "cut_share": 0.222656, "logprob_abs_diff": 0.013059, "train_rollout_kl": 0.00059, "entropy": 0.0, "grad_norm": 0.460842, "clip_fraction": 0.0, "step_time_s": 265.066982, "log_probs_time_s": 25.58786, "train_time_s": 152.063018, "update_weights_time_s": 0.597083, "sleep_time_s": 0.709101, "wake_up_time_s": 0.839562, "step_end": "2026-10-21 17:36:21", "rollout_paused_s": 45.0, "step_paused_s": 90.0 }, { "step": 5, "length": 705.097656, "reward_all_samples": 0.644531, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 68.381478, "rollout_end": "2026-10-21 17:37:32", "reward": 0.644531, "cut_share": 0.320312, "logprob_abs_diff": 0.014058, "train_rollout_kl": 0.000626, "entropy": 0.0, "grad_norm": 0.373078, "clip_fraction": 0.0, "step_time_s": 275.306549, "log_probs_time_s": 72.168882, "train_time_s": 202.530076, "update_weights_time_s": 0.605333, "sleep_time_s": 0.702088, "wake_up_time_s": 0.839406, "step_end": "2026-10-21 17:40:56", "rollout_paused_s": 0.0, "step_paused_s": 92.0 }, { "step": 6, "length": 697.523438, "reward_all_samples": 0.613281, "groups_all_wrong": 5.0, "groups_all_right": 9.0, "rollout_time_s": 66.81487, "rollout_end": "2026-10-21 17:42:05", "reward": 0.613281, "cut_share": 0.308594, "logprob_abs_diff": 0.013528, "train_rollout_kl": 0.000599, "entropy": 0.0, "grad_norm": 0.464884, "clip_fraction": 0.0, "step_time_s": 272.061278, "log_probs_time_s": 71.791357, "train_time_s": 200.979753, "update_weights_time_s": 0.616096, "sleep_time_s": 0.702923, "wake_up_time_s": 0.852137, "step_end": "2026-10-21 17:45:28", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 7, "length": 718.175781, "reward_all_samples": 0.664062, "groups_all_wrong": 7.0, "groups_all_right": 16.0, "rollout_time_s": 115.046445, "rollout_end": "2026-10-21 17:47:26", "reward": 0.664062, "cut_share": 0.3125, "logprob_abs_diff": 0.013952, "train_rollout_kl": 0.000616, "entropy": 0.0, "grad_norm": 0.272119, "clip_fraction": 0.0, "step_time_s": 322.639473, "log_probs_time_s": 72.659927, "train_time_s": 202.83861, "update_weights_time_s": 0.644419, "sleep_time_s": 0.746123, "wake_up_time_s": 0.760171, "step_end": "2026-10-21 17:50:51", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 8, "length": 674.128906, "reward_all_samples": 0.714844, "groups_all_wrong": 4.0, "groups_all_right": 16.0, "rollout_time_s": 109.344226, "rollout_end": "2026-10-21 17:52:43", "reward": 0.714844, "cut_share": 0.257812, "logprob_abs_diff": 0.013672, "train_rollout_kl": 0.000619, "entropy": 0.0, "grad_norm": 0.362972, "clip_fraction": 0.0, "step_time_s": 268.165287, "log_probs_time_s": 71.560476, "train_time_s": 154.429214, "update_weights_time_s": 0.61165, "sleep_time_s": 0.700005, "wake_up_time_s": 0.839777, "step_end": "2026-10-21 17:55:19", "rollout_paused_s": 45.0, "step_paused_s": 90.0 }, { "step": 9, "length": 653.0, "reward_all_samples": 0.75, "groups_all_wrong": 2.0, "groups_all_right": 15.0, "rollout_time_s": 61.536393, "rollout_end": "2026-10-21 17:56:23", "reward": 0.75, "cut_share": 0.234375, "logprob_abs_diff": 0.013212, "train_rollout_kl": 0.000591, "entropy": 0.0, "grad_norm": 0.419531, "clip_fraction": 0.0, "step_time_s": 217.51069, "log_probs_time_s": 70.669331, "train_time_s": 151.611071, "update_weights_time_s": 0.656507, "sleep_time_s": 0.777017, "wake_up_time_s": 0.837606, "step_end": "2026-10-21 17:58:56", "rollout_paused_s": 0.0, "step_paused_s": 46.0 }, { "step": 10, "length": 641.332031, "reward_all_samples": 0.714844, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 59.367211, "rollout_end": "2026-10-21 17:59:58", "reward": 0.714844, "cut_share": 0.191406, "logprob_abs_diff": 0.013018, "train_rollout_kl": 0.000579, "entropy": 0.0, "grad_norm": 0.465217, "clip_fraction": 0.0, "step_time_s": 260.930298, "log_probs_time_s": 70.716456, "train_time_s": 197.248274, "update_weights_time_s": 0.648561, "sleep_time_s": 0.727616, "wake_up_time_s": 0.7317, "step_end": "2026-10-21 18:03:17", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 11, "length": 635.0, "reward_all_samples": 0.695312, "groups_all_wrong": 4.0, "groups_all_right": 16.0, "rollout_time_s": 104.630101, "rollout_end": "2026-10-21 18:05:05", "reward": 0.695312, "cut_share": 0.242188, "logprob_abs_diff": 0.013018, "train_rollout_kl": 0.000578, "entropy": 0.0, "grad_norm": 0.3658, "clip_fraction": 0.0, "step_time_s": 259.908695, "log_probs_time_s": 70.426342, "train_time_s": 151.030647, "update_weights_time_s": 0.641525, "sleep_time_s": 0.753349, "wake_up_time_s": 0.757731, "step_end": "2026-10-21 18:07:37", "rollout_paused_s": 45.0, "step_paused_s": 90.0 } ] }, "e3b-offload-kv": { "steps": 12, "partial": false, "median": { "reward": 0.664062, "length": 702.771485, "cut_share": 0.30664, "reward_all_samples": 0.664062, "groups_all_wrong": 5.0, "groups_all_right": 11.5, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015279, "train_rollout_kl": 0.000706, "entropy": 0.0, "grad_norm": 0.438669, "clip_fraction": 0.0, "step_time_s": 302.300845, "rollout_time_s": 97.455186, "log_probs_time_s": 70.186804, "train_time_s": 195.806856, "update_weights_time_s": 0.622222, "sleep_time_s": 0.729604, "wake_up_time_s": 0.845619 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 57.840473, "steps": 5 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 795.523438, "reward_all_samples": 0.519531, "groups_all_wrong": 7.0, "groups_all_right": 5.0, "rollout_time_s": 68.170627, "rollout_end": "2026-10-21 18:12:52", "reward": 0.519531, "cut_share": 0.414062, "logprob_abs_diff": 0.015201, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.440825, "clip_fraction": 0.0, "step_time_s": 332.775785, "log_probs_time_s": 77.031511, "train_time_s": 211.876444, "update_weights_time_s": 0.620385, "sleep_time_s": 1.975914, "wake_up_time_s": 0.406307, "step_end": "2026-10-21 18:16:25", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 1, "length": 709.507812, "reward_all_samples": 0.695312, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 57.840473, "rollout_end": "2026-10-21 18:17:30", "reward": 0.695312, "cut_share": 0.273438, "logprob_abs_diff": 0.015212, "train_rollout_kl": 0.000706, "entropy": 0.0, "grad_norm": 0.4188, "clip_fraction": 0.0, "step_time_s": 216.492661, "log_probs_time_s": 25.172482, "train_time_s": 150.457251, "update_weights_time_s": 0.605071, "sleep_time_s": 4.697936, "wake_up_time_s": 0.753834, "step_end": "2026-10-21 18:20:02", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 2, "length": 768.222656, "reward_all_samples": 0.605469, "groups_all_wrong": 5.0, "groups_all_right": 6.0, "rollout_time_s": 63.929799, "rollout_end": "2026-10-21 18:21:09", "reward": 0.605469, "cut_share": 0.324219, "logprob_abs_diff": 0.015508, "train_rollout_kl": 0.000712, "entropy": 0.0, "grad_norm": 0.492654, "clip_fraction": 0.0, "step_time_s": 269.037087, "log_probs_time_s": 72.026961, "train_time_s": 200.381723, "update_weights_time_s": 0.628758, "sleep_time_s": 1.234745, "wake_up_time_s": 0.846135, "step_end": "2026-10-21 18:24:31", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 3, "length": 780.960938, "reward_all_samples": 0.617188, "groups_all_wrong": 5.0, "groups_all_right": 8.0, "rollout_time_s": 110.972311, "rollout_end": "2026-10-21 18:26:25", "reward": 0.617188, "cut_share": 0.363281, "logprob_abs_diff": 0.014485, "train_rollout_kl": 0.000652, "entropy": 0.0, "grad_norm": 0.405817, "clip_fraction": 0.0, "step_time_s": 363.088594, "log_probs_time_s": 72.402939, "train_time_s": 247.453705, "update_weights_time_s": 0.657573, "sleep_time_s": 1.222744, "wake_up_time_s": 0.747571, "step_end": "2026-10-21 18:30:34", "rollout_paused_s": 46.0, "step_paused_s": 182.0 }, { "step": 4, "length": 688.21875, "reward_all_samples": 0.757812, "groups_all_right": 14.0, "rollout_time_s": 55.866306, "rollout_end": "2026-10-21 18:31:32", "reward": 0.757812, "cut_share": 0.234375, "logprob_abs_diff": 0.014679, "train_rollout_kl": 0.000666, "entropy": 0.0, "grad_norm": 0.452438, "clip_fraction": 0.0, "step_time_s": 254.788794, "log_probs_time_s": 69.980623, "train_time_s": 194.699617, "update_weights_time_s": 0.649703, "sleep_time_s": 0.729125, "wake_up_time_s": 0.852201, "step_end": "2026-10-21 18:34:48", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 5, "length": 697.777344, "reward_all_samples": 0.632812, "groups_all_wrong": 7.0, "groups_all_right": 13.0, "rollout_time_s": 102.989713, "rollout_end": "2026-10-21 18:36:34", "reward": 0.632812, "cut_share": 0.324219, "logprob_abs_diff": 0.015632, "train_rollout_kl": 0.00072, "entropy": 0.0, "grad_norm": 0.348742, "clip_fraction": 0.0, "step_time_s": 348.690135, "log_probs_time_s": 70.331427, "train_time_s": 241.417263, "update_weights_time_s": 0.611143, "sleep_time_s": 0.692809, "wake_up_time_s": 0.860335, "step_end": "2026-10-21 18:40:37", "rollout_paused_s": 46.0, "step_paused_s": 181.0 }, { "step": 6, "length": 711.023438, "reward_all_samples": 0.574219, "groups_all_wrong": 4.0, "groups_all_right": 8.0, "rollout_time_s": 104.422053, "rollout_end": "2026-10-21 18:42:24", "reward": 0.574219, "cut_share": 0.324219, "logprob_abs_diff": 0.015535, "train_rollout_kl": 0.000706, "entropy": 0.0, "grad_norm": 0.51305, "clip_fraction": 0.0, "step_time_s": 304.441644, "log_probs_time_s": 70.556912, "train_time_s": 195.695087, "update_weights_time_s": 0.646323, "sleep_time_s": 0.730084, "wake_up_time_s": 0.850596, "step_end": "2026-10-21 18:45:42", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 7, "length": 707.765625, "reward_all_samples": 0.65625, "groups_all_wrong": 5.0, "groups_all_right": 15.0, "rollout_time_s": 103.36516, "rollout_end": "2026-10-21 18:47:28", "reward": 0.65625, "cut_share": 0.316406, "logprob_abs_diff": 0.015977, "train_rollout_kl": 0.000749, "entropy": 0.0, "grad_norm": 0.364168, "clip_fraction": 0.0, "step_time_s": 303.595573, "log_probs_time_s": 70.501667, "train_time_s": 195.918626, "update_weights_time_s": 0.60972, "sleep_time_s": 0.720699, "wake_up_time_s": 0.876894, "step_end": "2026-10-21 18:50:45", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 8, "length": 684.789062, "reward_all_samples": 0.671875, "groups_all_wrong": 7.0, "groups_all_right": 12.0, "rollout_time_s": 101.28368, "rollout_end": "2026-10-21 18:52:29", "reward": 0.671875, "cut_share": 0.296875, "logprob_abs_diff": 0.015729, "train_rollout_kl": 0.000716, "entropy": 0.0, "grad_norm": 0.436513, "clip_fraction": 0.0, "step_time_s": 301.006117, "log_probs_time_s": 70.042182, "train_time_s": 195.482206, "update_weights_time_s": 0.617833, "sleep_time_s": 0.699416, "wake_up_time_s": 0.779236, "step_end": "2026-10-21 18:55:46", "rollout_paused_s": 46.0, "step_paused_s": 138.0 }, { "step": 9, "length": 658.804688, "reward_all_samples": 0.722656, "groups_all_wrong": 3.0, "groups_all_right": 11.0, "rollout_time_s": 97.996554, "rollout_end": "2026-10-21 18:57:27", "reward": 0.722656, "cut_share": 0.21875, "logprob_abs_diff": 0.015345, "train_rollout_kl": 0.000709, "entropy": 0.0, "grad_norm": 0.470296, "clip_fraction": 0.0, "step_time_s": 339.50324, "log_probs_time_s": 69.205187, "train_time_s": 237.199819, "update_weights_time_s": 0.619705, "sleep_time_s": 0.701618, "wake_up_time_s": 0.859169, "step_end": "2026-10-21 19:01:26", "rollout_paused_s": 45.0, "step_paused_s": 181.0 }, { "step": 10, "length": 651.125, "reward_all_samples": 0.707031, "groups_all_wrong": 3.0, "groups_all_right": 7.0, "rollout_time_s": 96.913818, "rollout_end": "2026-10-21 19:03:05", "reward": 0.707031, "cut_share": 0.207031, "logprob_abs_diff": 0.014822, "train_rollout_kl": 0.000679, "entropy": 0.0, "grad_norm": 0.570029, "clip_fraction": 0.0, "step_time_s": 294.176421, "log_probs_time_s": 23.978618, "train_time_s": 193.172009, "update_weights_time_s": 0.62406, "sleep_time_s": 0.706118, "wake_up_time_s": 0.763773, "step_end": "2026-10-21 19:06:20", "rollout_paused_s": 46.0, "step_paused_s": 138.0 }, { "step": 11, "length": 641.585938, "reward_all_samples": 0.707031, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 51.703855, "rollout_end": "2026-10-21 19:07:14", "reward": 0.707031, "cut_share": 0.234375, "logprob_abs_diff": 0.015088, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.407825, "clip_fraction": 0.0, "step_time_s": 201.815629, "log_probs_time_s": 69.002554, "train_time_s": 145.756861, "update_weights_time_s": 0.6627, "sleep_time_s": 0.742918, "wake_up_time_s": 0.845104, "step_end": "2026-10-21 19:09:42", "rollout_paused_s": 0.0, "step_paused_s": 45.0 } ] }, "e3c-no-offload-rollout": { "steps": 12, "partial": false, "median": { "reward": 0.664062, "length": 678.25586, "cut_share": 0.283203, "reward_all_samples": 0.664062, "groups_all_wrong": 4.0, "groups_all_right": 11.5, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.01539, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.441683, "clip_fraction": 0.0, "step_time_s": 290.546241, "rollout_time_s": 63.161759, "log_probs_time_s": 69.649521, "train_time_s": 196.552136, "update_weights_time_s": 0.642703, "sleep_time_s": 0.732117, "wake_up_time_s": 0.824997 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 51.955779, "steps": 7 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 796.738281, "reward_all_samples": 0.527344, "groups_all_wrong": 7.0, "groups_all_right": 8.0, "rollout_time_s": 68.125865, "rollout_end": "2026-10-21 19:15:03", "reward": 0.527344, "cut_share": 0.425781, "logprob_abs_diff": 0.015117, "train_rollout_kl": 0.000685, "entropy": 0.0, "grad_norm": 0.434932, "clip_fraction": 0.0, "step_time_s": 326.957615, "log_probs_time_s": 76.481627, "train_time_s": 255.390719, "update_weights_time_s": 0.636944, "sleep_time_s": 2.303358, "wake_up_time_s": 0.453646, "step_end": "2026-10-21 19:19:19", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 1, "length": 688.496094, "reward_all_samples": 0.6875, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "rollout_time_s": 100.99898, "rollout_end": "2026-10-21 19:21:05", "reward": 0.6875, "cut_share": 0.273438, "logprob_abs_diff": 0.015041, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.424787, "clip_fraction": 0.0, "step_time_s": 255.355196, "log_probs_time_s": 24.523497, "train_time_s": 148.352944, "update_weights_time_s": 0.640517, "sleep_time_s": 4.340848, "wake_up_time_s": 0.837195, "step_end": "2026-10-21 19:23:35", "rollout_paused_s": 45.0, "step_paused_s": 90.0 }, { "step": 2, "length": 759.398438, "reward_all_samples": 0.632812, "groups_all_wrong": 5.0, "groups_all_right": 9.0, "rollout_time_s": 109.661573, "rollout_end": "2026-10-21 19:25:26", "reward": 0.632812, "cut_share": 0.292969, "logprob_abs_diff": 0.015344, "train_rollout_kl": 0.000689, "entropy": 0.0, "grad_norm": 0.424813, "clip_fraction": 0.0, "step_time_s": 311.238224, "log_probs_time_s": 71.492417, "train_time_s": 198.884308, "update_weights_time_s": 0.626323, "sleep_time_s": 1.117086, "wake_up_time_s": 0.750629, "step_end": "2026-10-21 19:28:46", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 3, "length": 770.09375, "reward_all_samples": 0.621094, "groups_all_wrong": 4.0, "groups_all_right": 8.0, "rollout_time_s": 110.02861, "rollout_end": "2026-10-21 19:30:38", "reward": 0.621094, "cut_share": 0.355469, "logprob_abs_diff": 0.014813, "train_rollout_kl": 0.000672, "entropy": 0.0, "grad_norm": 0.448434, "clip_fraction": 0.0, "step_time_s": 313.361367, "log_probs_time_s": 71.921494, "train_time_s": 200.682484, "update_weights_time_s": 0.61656, "sleep_time_s": 1.085217, "wake_up_time_s": 0.735982, "step_end": "2026-10-21 19:33:59", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 4, "length": 676.328125, "reward_all_samples": 0.746094, "groups_all_right": 13.0, "rollout_time_s": 100.091031, "rollout_end": "2026-10-21 19:35:41", "reward": 0.746094, "cut_share": 0.242188, "logprob_abs_diff": 0.01493, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.489605, "clip_fraction": 0.0, "step_time_s": 340.694351, "log_probs_time_s": 69.468552, "train_time_s": 238.190114, "update_weights_time_s": 0.650679, "sleep_time_s": 0.718687, "wake_up_time_s": 0.847695, "step_end": "2026-10-21 19:39:40", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 5, "length": 680.183594, "reward_all_samples": 0.640625, "groups_all_wrong": 6.0, "groups_all_right": 10.0, "rollout_time_s": 101.261841, "rollout_end": "2026-10-21 19:41:22", "reward": 0.640625, "cut_share": 0.320312, "logprob_abs_diff": 0.01584, "train_rollout_kl": 0.00073, "entropy": 0.0, "grad_norm": 0.453823, "clip_fraction": 0.0, "step_time_s": 297.854513, "log_probs_time_s": 69.830491, "train_time_s": 194.219963, "update_weights_time_s": 0.648175, "sleep_time_s": 0.710311, "wake_up_time_s": 0.824296, "step_end": "2026-10-21 19:44:38", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 6, "length": 674.183594, "reward_all_samples": 0.605469, "groups_all_wrong": 6.0, "groups_all_right": 10.0, "rollout_time_s": 55.636852, "rollout_end": "2026-10-21 19:45:35", "reward": 0.605469, "cut_share": 0.285156, "logprob_abs_diff": 0.015435, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.508712, "clip_fraction": 0.0, "step_time_s": 250.723355, "log_probs_time_s": 69.450728, "train_time_s": 192.746768, "update_weights_time_s": 0.616821, "sleep_time_s": 0.703891, "wake_up_time_s": 0.825697, "step_end": "2026-10-21 19:48:48", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 7, "length": 702.125, "reward_all_samples": 0.621094, "groups_all_wrong": 6.0, "groups_all_right": 12.0, "rollout_time_s": 58.197653, "rollout_end": "2026-10-21 19:49:48", "reward": 0.621094, "cut_share": 0.335938, "logprob_abs_diff": 0.016048, "train_rollout_kl": 0.000722, "entropy": 0.0, "grad_norm": 0.431432, "clip_fraction": 0.0, "step_time_s": 300.355492, "log_probs_time_s": 115.396689, "train_time_s": 239.74477, "update_weights_time_s": 0.662007, "sleep_time_s": 0.718447, "wake_up_time_s": 0.832154, "step_end": "2026-10-21 19:53:49", "rollout_paused_s": 0.0, "step_paused_s": 136.0 }, { "step": 8, "length": 644.023438, "reward_all_samples": 0.703125, "groups_all_wrong": 4.0, "groups_all_right": 15.0, "rollout_time_s": 51.955779, "rollout_end": "2026-10-21 19:54:42", "reward": 0.703125, "cut_share": 0.28125, "logprob_abs_diff": 0.015648, "train_rollout_kl": 0.000719, "entropy": 0.0, "grad_norm": 0.40155, "clip_fraction": 0.0, "step_time_s": 246.282471, "log_probs_time_s": 68.923921, "train_time_s": 192.048988, "update_weights_time_s": 0.648558, "sleep_time_s": 0.717283, "wake_up_time_s": 0.721666, "step_end": "2026-10-21 19:57:55", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 9, "length": 598.429688, "reward_all_samples": 0.734375, "groups_all_wrong": 2.0, "groups_all_right": 12.0, "rollout_time_s": 48.206567, "rollout_end": "2026-10-21 19:58:45", "reward": 0.734375, "cut_share": 0.207031, "logprob_abs_diff": 0.015192, "train_rollout_kl": 0.000684, "entropy": 0.0, "grad_norm": 0.499291, "clip_fraction": 0.0, "step_time_s": 238.455482, "log_probs_time_s": 22.459674, "train_time_s": 187.860137, "update_weights_time_s": 0.633008, "sleep_time_s": 0.73172, "wake_up_time_s": 0.839848, "step_end": "2026-10-21 20:01:53", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 10, "length": 599.191406, "reward_all_samples": 0.730469, "groups_all_wrong": 3.0, "groups_all_right": 11.0, "rollout_time_s": 46.734233, "rollout_end": "2026-10-21 20:02:42", "reward": 0.730469, "cut_share": 0.15625, "logprob_abs_diff": 0.015678, "train_rollout_kl": 0.000763, "entropy": 0.0, "grad_norm": 0.531792, "clip_fraction": 0.0, "step_time_s": 283.237968, "log_probs_time_s": 113.275561, "train_time_s": 234.184966, "update_weights_time_s": 0.644889, "sleep_time_s": 0.732513, "wake_up_time_s": 0.746731, "step_end": "2026-10-21 20:06:37", "rollout_paused_s": 0.0, "step_paused_s": 137.0 }, { "step": 11, "length": 579.535156, "reward_all_samples": 0.703125, "groups_all_wrong": 4.0, "groups_all_right": 16.0, "rollout_time_s": 46.049489, "rollout_end": "2026-10-21 20:07:24", "reward": 0.703125, "cut_share": 0.214844, "logprob_abs_diff": 0.015456, "train_rollout_kl": 0.000715, "entropy": 0.0, "grad_norm": 0.39665, "clip_fraction": 0.0, "step_time_s": 234.675804, "log_probs_time_s": 67.327168, "train_time_s": 186.171356, "update_weights_time_s": 0.667765, "sleep_time_s": 0.746711, "wake_up_time_s": 0.858902, "step_end": "2026-10-21 20:10:31", "rollout_paused_s": 0.0, "step_paused_s": 90.0 } ] }, "e3d-no-offload-train": { "steps": 12, "partial": false, "median": { "reward": 0.640625, "length": 688.396485, "cut_share": 0.285157, "reward_all_samples": 0.640625, "groups_all_wrong": 5.0, "groups_all_right": 12.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015193, "train_rollout_kl": 0.000691, "entropy": 0.0, "grad_norm": 0.459853, "clip_fraction": 0.0, "step_time_s": 301.363104, "rollout_time_s": 81.176181, "log_probs_time_s": 70.506109, "train_time_s": 206.377922, "update_weights_time_s": 0.518008, "sleep_time_s": null, "wake_up_time_s": null }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 59.770268, "steps": 6 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 781.105469, "reward_all_samples": 0.535156, "groups_all_wrong": 9.0, "groups_all_right": 6.0, "rollout_time_s": 66.541497, "rollout_end": "2026-10-21 20:15:44", "reward": 0.535156, "cut_share": 0.40625, "logprob_abs_diff": 0.015159, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.449215, "clip_fraction": 0.0, "step_time_s": 330.619319, "log_probs_time_s": 77.499846, "train_time_s": 212.119071, "update_weights_time_s": 0.518514, "step_end": "2026-10-21 20:19:17", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 1, "length": 690.773438, "reward_all_samples": 0.679688, "groups_all_wrong": 2.0, "groups_all_right": 13.0, "rollout_time_s": 101.37976, "rollout_end": "2026-10-21 20:21:01", "reward": 0.679688, "cut_share": 0.269531, "logprob_abs_diff": 0.015193, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.478249, "clip_fraction": 0.0, "step_time_s": 344.213096, "log_probs_time_s": 70.586693, "train_time_s": 239.850244, "update_weights_time_s": 0.520565, "step_end": "2026-10-21 20:25:01", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 2, "length": 770.148438, "reward_all_samples": 0.609375, "groups_all_wrong": 3.0, "groups_all_right": 7.0, "rollout_time_s": 109.50371, "rollout_end": "2026-10-21 20:26:53", "reward": 0.609375, "cut_share": 0.339844, "logprob_abs_diff": 0.015348, "train_rollout_kl": 0.000694, "entropy": 0.0, "grad_norm": 0.470491, "clip_fraction": 0.0, "step_time_s": 313.25755, "log_probs_time_s": 117.793774, "train_time_s": 200.636773, "update_weights_time_s": 0.505672, "step_end": "2026-10-21 20:30:15", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 3, "length": 774.679688, "reward_all_samples": 0.617188, "groups_all_wrong": 5.0, "groups_all_right": 8.0, "rollout_time_s": 65.216353, "rollout_end": "2026-10-21 20:31:22", "reward": 0.617188, "cut_share": 0.347656, "logprob_abs_diff": 0.014743, "train_rollout_kl": 0.000657, "entropy": 0.0, "grad_norm": 0.424487, "clip_fraction": 0.0, "step_time_s": 315.685631, "log_probs_time_s": 72.84545, "train_time_s": 247.418934, "update_weights_time_s": 0.529174, "step_end": "2026-10-21 20:35:30", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 4, "length": 686.019531, "reward_all_samples": 0.757812, "groups_all_right": 12.0, "rollout_time_s": 55.688312, "rollout_end": "2026-10-21 20:36:28", "reward": 0.757812, "cut_share": 0.25, "logprob_abs_diff": 0.01486, "train_rollout_kl": 0.000677, "entropy": 0.0, "grad_norm": 0.520845, "clip_fraction": 0.0, "step_time_s": 252.588469, "log_probs_time_s": 70.236422, "train_time_s": 193.942993, "update_weights_time_s": 0.495623, "step_end": "2026-10-21 20:39:43", "rollout_paused_s": 0.0, "step_paused_s": 92.0 }, { "step": 5, "length": 718.058594, "reward_all_samples": 0.621094, "groups_all_wrong": 7.0, "groups_all_right": 13.0, "rollout_time_s": 59.742037, "rollout_end": "2026-10-21 20:40:45", "reward": 0.621094, "cut_share": 0.34375, "logprob_abs_diff": 0.016098, "train_rollout_kl": 0.000732, "entropy": 0.0, "grad_norm": 0.348492, "clip_fraction": 0.0, "step_time_s": 259.418856, "log_probs_time_s": 71.093849, "train_time_s": 196.566684, "update_weights_time_s": 0.521918, "step_end": "2026-10-21 20:44:02", "rollout_paused_s": 0.0, "step_paused_s": 91.0 }, { "step": 6, "length": 679.632812, "reward_all_samples": 0.578125, "groups_all_wrong": 7.0, "groups_all_right": 8.0, "rollout_time_s": 101.11114, "rollout_end": "2026-10-21 20:45:46", "reward": 0.578125, "cut_share": 0.273438, "logprob_abs_diff": 0.015105, "train_rollout_kl": 0.000685, "entropy": 0.0, "grad_norm": 0.470677, "clip_fraction": 0.0, "step_time_s": 343.347974, "log_probs_time_s": 70.323138, "train_time_s": 239.094637, "update_weights_time_s": 0.534251, "step_end": "2026-10-21 20:49:46", "rollout_paused_s": 45.0, "step_paused_s": 182.0 }, { "step": 7, "length": 714.902344, "reward_all_samples": 0.613281, "groups_all_wrong": 6.0, "groups_all_right": 11.0, "rollout_time_s": 59.798498, "rollout_end": "2026-10-21 20:50:48", "reward": 0.613281, "cut_share": 0.367188, "logprob_abs_diff": 0.01572, "train_rollout_kl": 0.000709, "entropy": 0.0, "grad_norm": 0.385963, "clip_fraction": 0.0, "step_time_s": 304.139709, "log_probs_time_s": 71.110737, "train_time_s": 241.376109, "update_weights_time_s": 0.501519, "step_end": "2026-10-21 20:54:50", "rollout_paused_s": 0.0, "step_paused_s": 136.0 }, { "step": 8, "length": 678.835938, "reward_all_samples": 0.660156, "groups_all_wrong": 5.0, "groups_all_right": 12.0, "rollout_time_s": 100.804708, "rollout_end": "2026-10-21 20:56:33", "reward": 0.660156, "cut_share": 0.296875, "logprob_abs_diff": 0.015369, "train_rollout_kl": 0.000682, "entropy": 0.0, "grad_norm": 0.413819, "clip_fraction": 0.0, "step_time_s": 298.586499, "log_probs_time_s": 70.425525, "train_time_s": 194.680773, "update_weights_time_s": 0.517502, "step_end": "2026-10-21 20:59:49", "rollout_paused_s": 45.0, "step_paused_s": 137.0 }, { "step": 9, "length": 635.941406, "reward_all_samples": 0.714844, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 95.940609, "rollout_end": "2026-10-21 21:01:27", "reward": 0.714844, "cut_share": 0.203125, "logprob_abs_diff": 0.015342, "train_rollout_kl": 0.000725, "entropy": 0.0, "grad_norm": 0.444483, "clip_fraction": 0.0, "step_time_s": 289.93292, "log_probs_time_s": 69.067626, "train_time_s": 190.91572, "update_weights_time_s": 0.489885, "step_end": "2026-10-21 21:04:38", "rollout_paused_s": 45.9, "step_paused_s": 137.0 }, { "step": 10, "length": 643.09375, "reward_all_samples": 0.742188, "groups_all_wrong": 3.0, "groups_all_right": 13.0, "rollout_time_s": 95.810864, "rollout_end": "2026-10-21 21:06:17", "reward": 0.742188, "cut_share": 0.164062, "logprob_abs_diff": 0.015193, "train_rollout_kl": 0.000714, "entropy": 0.0, "grad_norm": 0.501545, "clip_fraction": 0.0, "step_time_s": 290.957192, "log_probs_time_s": 69.532095, "train_time_s": 192.018959, "update_weights_time_s": 0.528144, "step_end": "2026-10-21 21:09:29", "rollout_paused_s": 45.0, "step_paused_s": 136.0 }, { "step": 11, "length": 622.9375, "reward_all_samples": 0.6875, "groups_all_wrong": 2.0, "groups_all_right": 12.0, "rollout_time_s": 49.839419, "rollout_end": "2026-10-21 21:10:22", "reward": 0.6875, "cut_share": 0.242188, "logprob_abs_diff": 0.015083, "train_rollout_kl": 0.000688, "entropy": 0.0, "grad_norm": 0.493643, "clip_fraction": 0.0, "step_time_s": 287.887072, "log_probs_time_s": 68.793268, "train_time_s": 235.069756, "update_weights_time_s": 0.510606, "step_end": "2026-10-21 21:14:17", "rollout_paused_s": 0.0, "step_paused_s": 138.0 } ] }, "e4-0.6b": { "steps": 4, "partial": false, "median": { "reward": 0.003906, "length": 255.960938, "cut_share": 0.996094, "reward_all_samples": 0.003906, "groups_all_wrong": 31.5, "groups_all_right": null, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015273, "train_rollout_kl": 0.000731, "entropy": 0.0, "grad_norm": 0.08038, "clip_fraction": 0.0, "step_time_s": 112.858062, "rollout_time_s": 12.395647, "log_probs_time_s": 14.299919, "train_time_s": 71.919622, "update_weights_time_s": 0.641702, "sleep_time_s": 1.589914, "wake_up_time_s": 0.805392 }, "median_unpaused": { "step_time_s": { "median": 91.784298, "steps": 3 }, "rollout_time_s": { "median": 12.292029, "steps": 3 }, "log_probs_time_s": { "median": 14.270588, "steps": 3 }, "train_time_s": { "median": 71.786645, "steps": 3 }, "update_weights_time_s": { "median": 0.648656, "steps": 3 }, "sleep_time_s": { "median": 1.979656, "steps": 3 }, "wake_up_time_s": { "median": 0.844002, "steps": 3 } }, "per_step": [ { "step": 0, "length": 255.863281, "reward_all_samples": 0.011719, "groups_all_wrong": 31.0, "rollout_time_s": 12.499266, "rollout_end": "2026-10-22 05:28:51", "reward": 0.011719, "cut_share": 0.992188, "logprob_abs_diff": 0.015136, "train_rollout_kl": 0.000723, "entropy": 0.0, "grad_norm": 0.18585, "clip_fraction": 0.0, "step_time_s": 145.696616, "log_probs_time_s": 18.763459, "train_time_s": 80.958204, "update_weights_time_s": 0.648656, "sleep_time_s": 1.979656, "wake_up_time_s": 0.389952, "step_end": "2026-10-22 05:30:13", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 1, "length": 255.921875, "reward_all_samples": 0.007812, "groups_all_wrong": 31.0, "rollout_time_s": 12.292029, "rollout_end": "2026-10-22 05:30:31", "reward": 0.007812, "cut_share": 0.992188, "logprob_abs_diff": 0.015411, "train_rollout_kl": 0.000754, "entropy": 0.0, "grad_norm": 0.16076, "clip_fraction": 0.0, "step_time_s": 91.784298, "log_probs_time_s": 14.270588, "train_time_s": 71.786645, "update_weights_time_s": 0.633483, "sleep_time_s": 4.113117, "wake_up_time_s": 0.844002, "step_end": "2026-10-22 05:31:45", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 2, "length": 256.0, "reward_all_samples": 0.0, "groups_all_wrong": 32.0, "rollout_time_s": 12.022492, "rollout_end": "2026-10-22 05:32:00", "reward": 0.0, "cut_share": 1.0, "logprob_abs_diff": 0.01543, "train_rollout_kl": 0.000738, "entropy": 0.0, "grad_norm": 0.0, "clip_fraction": 0.0, "step_time_s": 88.224191, "log_probs_time_s": 14.144441, "train_time_s": 71.405636, "update_weights_time_s": 0.654508, "sleep_time_s": 1.200172, "wake_up_time_s": 0.859584, "step_end": "2026-10-22 05:33:13", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 3, "length": 256.0, "reward_all_samples": 0.0, "groups_all_wrong": 32.0, "rollout_time_s": 57.422751, "rollout_end": "2026-10-22 05:34:13", "reward": 0.0, "cut_share": 1.0, "logprob_abs_diff": 0.014775, "train_rollout_kl": 0.000708, "entropy": 0.0, "grad_norm": 0.0, "clip_fraction": 0.0, "step_time_s": 133.931826, "log_probs_time_s": 14.329251, "train_time_s": 72.052598, "update_weights_time_s": 0.634748, "sleep_time_s": 1.081404, "wake_up_time_s": 0.766783, "step_end": "2026-10-22 05:35:27", "rollout_paused_s": 45.0, "step_paused_s": 45.0 } ] }, "e4-1.7b": { "steps": 4, "partial": false, "median": { "reward": 0.0, "length": 256.0, "cut_share": 1.0, "reward_all_samples": 0.0, "groups_all_wrong": 32.0, "groups_all_right": null, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.01115, "train_rollout_kl": 0.001167, "entropy": 0.0, "grad_norm": 0.0, "clip_fraction": 0.0, "step_time_s": 240.703769, "rollout_time_s": 29.995109, "log_probs_time_s": 29.691584, "train_time_s": 185.833419, "update_weights_time_s": 0.907601, "sleep_time_s": 3.831921, "wake_up_time_s": 1.436627 }, "median_unpaused": { "step_time_s": { "median": 240.703769, "steps": 4 }, "rollout_time_s": { "median": 29.995109, "steps": 4 }, "log_probs_time_s": { "median": 29.691584, "steps": 4 }, "train_time_s": { "median": 185.833419, "steps": 4 }, "update_weights_time_s": { "median": 0.907601, "steps": 4 }, "sleep_time_s": { "median": 3.831921, "steps": 4 }, "wake_up_time_s": { "median": 1.436627, "steps": 4 } }, "per_step": [ { "step": 0, "length": 255.921875, "reward_all_samples": 0.003906, "groups_all_wrong": 31.0, "rollout_time_s": 16.455604, "rollout_end": "2026-10-22 05:40:58", "reward": 0.003906, "cut_share": 0.996094, "logprob_abs_diff": 0.010914, "train_rollout_kl": 0.001148, "entropy": 0.0, "grad_norm": 0.265367, "clip_fraction": 0.0, "step_time_s": 242.543478, "log_probs_time_s": 33.103015, "train_time_s": 168.491639, "update_weights_time_s": 0.953785, "sleep_time_s": 4.461439, "wake_up_time_s": 0.634639, "step_end": "2026-10-22 05:43:47", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 1, "length": 256.0, "reward_all_samples": 0.0, "groups_all_wrong": 32.0, "rollout_time_s": 21.362916, "rollout_end": "2026-10-22 05:44:23", "reward": 0.0, "cut_share": 1.0, "logprob_abs_diff": 0.011152, "train_rollout_kl": 0.001187, "entropy": 0.0, "grad_norm": 0.0, "clip_fraction": 0.0, "step_time_s": 216.305738, "log_probs_time_s": 29.582352, "train_time_s": 178.364255, "update_weights_time_s": 0.920276, "sleep_time_s": 11.821312, "wake_up_time_s": 1.568311, "step_end": "2026-10-22 05:47:24", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 2, "length": 256.0, "reward_all_samples": 0.0, "groups_all_wrong": 32.0, "rollout_time_s": 38.627301, "rollout_end": "2026-10-22 05:48:07", "reward": 0.0, "cut_share": 1.0, "logprob_abs_diff": 0.011816, "train_rollout_kl": 0.001278, "entropy": 0.0, "grad_norm": 0.0, "clip_fraction": 0.0, "step_time_s": 238.864061, "log_probs_time_s": 29.413388, "train_time_s": 193.302584, "update_weights_time_s": 0.885162, "sleep_time_s": 2.62803, "wake_up_time_s": 1.448952, "step_end": "2026-10-22 05:51:23", "rollout_paused_s": 0.0, "step_paused_s": 0.0 }, { "step": 3, "length": 256.0, "reward_all_samples": 0.0, "groups_all_wrong": 32.0, "rollout_time_s": 54.108995, "rollout_end": "2026-10-22 05:52:22", "reward": 0.0, "cut_share": 1.0, "logprob_abs_diff": 0.011148, "train_rollout_kl": 0.00109, "entropy": 0.0, "grad_norm": 0.0, "clip_fraction": 0.0, "step_time_s": 257.452137, "log_probs_time_s": 29.800816, "train_time_s": 195.939775, "update_weights_time_s": 0.894927, "sleep_time_s": 3.202403, "wake_up_time_s": 1.424301, "step_end": "2026-10-22 05:55:40", "rollout_paused_s": 0.0, "step_paused_s": 0.0 } ] }, "e2-n4": { "steps": 24, "partial": false, "median": { "reward": 0.679688, "length": 691.628906, "cut_share": 0.292968, "reward_all_samples": 0.679688, "groups_all_wrong": 4.5, "groups_all_right": 15.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015359, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.499038, "clip_fraction": 0.0, "step_time_s": 135.434307, "rollout_time_s": 32.493577, "log_probs_time_s": 15.681386, "train_time_s": 97.533317, "update_weights_time_s": 0.644969, "sleep_time_s": 0.733437, "wake_up_time_s": 0.771092 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 32.493577, "steps": 14 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 795.617188, "reward_all_samples": 0.53125, "groups_all_wrong": 7.0, "groups_all_right": 9.0, "rollout_time_s": 38.017989, "rollout_end": "2026-10-22 06:18:16", "reward": 0.53125, "cut_share": 0.414062, "logprob_abs_diff": 0.015576, "train_rollout_kl": 0.000711, "entropy": 0.0, "grad_norm": 0.539012, "clip_fraction": 0.0, "step_time_s": 201.142803, "log_probs_time_s": 18.264516, "train_time_s": 111.03121, "update_weights_time_s": 0.655416, "sleep_time_s": 2.001901, "wake_up_time_s": 0.446094, "step_end": "2026-10-22 06:20:08", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 1, "length": 707.851562, "reward_all_samples": 0.6875, "groups_all_wrong": 3.0, "groups_all_right": 15.0, "rollout_time_s": 32.436226, "rollout_end": "2026-10-22 06:20:47", "reward": 0.6875, "cut_share": 0.304688, "logprob_abs_diff": 0.015491, "train_rollout_kl": 0.000713, "entropy": 0.0, "grad_norm": 0.538014, "clip_fraction": 0.0, "step_time_s": 138.047897, "log_probs_time_s": 58.001814, "train_time_s": 97.915543, "update_weights_time_s": 0.637731, "sleep_time_s": 4.224224, "wake_up_time_s": 0.741243, "step_end": "2026-10-22 06:22:26", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 2, "length": 778.09375, "reward_all_samples": 0.5625, "groups_all_wrong": 6.0, "groups_all_right": 7.0, "rollout_time_s": 81.747663, "rollout_end": "2026-10-22 06:23:51", "reward": 0.5625, "cut_share": 0.40625, "logprob_abs_diff": 0.015591, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.571227, "clip_fraction": 0.0, "step_time_s": 232.776508, "log_probs_time_s": 58.810598, "train_time_s": 146.047084, "update_weights_time_s": 0.646976, "sleep_time_s": 1.166306, "wake_up_time_s": 0.741958, "step_end": "2026-10-22 06:26:19", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 3, "length": 769.65625, "reward_all_samples": 0.617188, "groups_all_wrong": 5.0, "groups_all_right": 13.0, "rollout_time_s": 35.902363, "rollout_end": "2026-10-22 06:26:58", "reward": 0.617188, "cut_share": 0.335938, "logprob_abs_diff": 0.014692, "train_rollout_kl": 0.000677, "entropy": 0.0, "grad_norm": 0.472146, "clip_fraction": 0.0, "step_time_s": 187.255561, "log_probs_time_s": 58.895747, "train_time_s": 146.661239, "update_weights_time_s": 0.645329, "sleep_time_s": 1.220616, "wake_up_time_s": 0.856935, "step_end": "2026-10-22 06:29:26", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 4, "length": 691.4375, "reward_all_samples": 0.78125, "groups_all_wrong": 1.0, "groups_all_right": 17.0, "rollout_time_s": 31.824405, "rollout_end": "2026-10-22 06:30:01", "reward": 0.78125, "cut_share": 0.265625, "logprob_abs_diff": 0.014729, "train_rollout_kl": 0.000674, "entropy": 0.0, "grad_norm": 0.506562, "clip_fraction": 0.0, "step_time_s": 133.80091, "log_probs_time_s": 57.695053, "train_time_s": 97.728478, "update_weights_time_s": 0.603197, "sleep_time_s": 0.703009, "wake_up_time_s": 0.845654, "step_end": "2026-10-22 06:31:40", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 5, "length": 691.820312, "reward_all_samples": 0.609375, "groups_all_wrong": 8.0, "groups_all_right": 15.0, "rollout_time_s": 31.985356, "rollout_end": "2026-10-22 06:32:15", "reward": 0.609375, "cut_share": 0.328125, "logprob_abs_diff": 0.015386, "train_rollout_kl": 0.000683, "entropy": 0.0, "grad_norm": 0.403082, "clip_fraction": 0.0, "step_time_s": 134.418552, "log_probs_time_s": 57.971927, "train_time_s": 98.233773, "update_weights_time_s": 0.646132, "sleep_time_s": 0.725208, "wake_up_time_s": 0.854484, "step_end": "2026-10-22 06:33:54", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 6, "length": 726.046875, "reward_all_samples": 0.625, "groups_all_wrong": 6.0, "groups_all_right": 13.0, "rollout_time_s": 78.837076, "rollout_end": "2026-10-22 06:35:16", "reward": 0.625, "cut_share": 0.3125, "logprob_abs_diff": 0.015054, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.491514, "clip_fraction": 0.0, "step_time_s": 182.03335, "log_probs_time_s": 12.82297, "train_time_s": 98.884114, "update_weights_time_s": 0.609954, "sleep_time_s": 0.722741, "wake_up_time_s": 0.855891, "step_end": "2026-10-22 06:36:56", "rollout_paused_s": 46.0, "step_paused_s": 91.0 }, { "step": 7, "length": 717.351562, "reward_all_samples": 0.671875, "groups_all_wrong": 5.0, "groups_all_right": 14.0, "rollout_time_s": 33.343777, "rollout_end": "2026-10-22 06:37:32", "reward": 0.671875, "cut_share": 0.335938, "logprob_abs_diff": 0.01551, "train_rollout_kl": 0.000713, "entropy": 0.0, "grad_norm": 0.523862, "clip_fraction": 0.0, "step_time_s": 136.450062, "log_probs_time_s": 58.160734, "train_time_s": 98.846323, "update_weights_time_s": 0.650563, "sleep_time_s": 0.735037, "wake_up_time_s": 0.783199, "step_end": "2026-10-22 06:39:13", "rollout_paused_s": 0.0, "step_paused_s": 46.0 }, { "step": 8, "length": 672.359375, "reward_all_samples": 0.6875, "groups_all_wrong": 7.0, "groups_all_right": 17.0, "rollout_time_s": 30.970884, "rollout_end": "2026-10-22 06:39:46", "reward": 0.6875, "cut_share": 0.304688, "logprob_abs_diff": 0.015528, "train_rollout_kl": 0.000708, "entropy": 0.0, "grad_norm": 0.414747, "clip_fraction": 0.0, "step_time_s": 132.504849, "log_probs_time_s": 12.354147, "train_time_s": 97.336181, "update_weights_time_s": 0.651588, "sleep_time_s": 0.71872, "wake_up_time_s": 0.839341, "step_end": "2026-10-22 06:41:25", "rollout_paused_s": 0.0, "step_paused_s": 46.0 }, { "step": 9, "length": 660.007812, "reward_all_samples": 0.664062, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 75.341205, "rollout_end": "2026-10-22 06:42:43", "reward": 0.664062, "cut_share": 0.289062, "logprob_abs_diff": 0.015331, "train_rollout_kl": 0.000706, "entropy": 0.0, "grad_norm": 0.629005, "clip_fraction": 0.0, "step_time_s": 175.88246, "log_probs_time_s": 11.952682, "train_time_s": 96.370345, "update_weights_time_s": 0.636545, "sleep_time_s": 0.739015, "wake_up_time_s": 0.747518, "step_end": "2026-10-22 06:44:21", "rollout_paused_s": 46.0, "step_paused_s": 92.0 }, { "step": 10, "length": 656.40625, "reward_all_samples": 0.710938, "groups_all_wrong": 5.0, "groups_all_right": 16.0, "rollout_time_s": 29.504539, "rollout_end": "2026-10-22 06:44:53", "reward": 0.710938, "cut_share": 0.171875, "logprob_abs_diff": 0.014885, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.475224, "clip_fraction": 0.0, "step_time_s": 130.799886, "log_probs_time_s": 57.478707, "train_time_s": 97.02696, "update_weights_time_s": 0.644608, "sleep_time_s": 0.750057, "wake_up_time_s": 0.765397, "step_end": "2026-10-22 06:46:32", "rollout_paused_s": 0.0, "step_paused_s": 46.0 }, { "step": 11, "length": 662.429688, "reward_all_samples": 0.664062, "groups_all_wrong": 6.0, "groups_all_right": 15.0, "rollout_time_s": 75.661155, "rollout_end": "2026-10-22 06:47:50", "reward": 0.664062, "cut_share": 0.257812, "logprob_abs_diff": 0.01513, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.477998, "clip_fraction": 0.0, "step_time_s": 176.301298, "log_probs_time_s": 57.394963, "train_time_s": 96.416104, "update_weights_time_s": 0.62441, "sleep_time_s": 0.725492, "wake_up_time_s": 0.760872, "step_end": "2026-10-22 06:49:28", "rollout_paused_s": 46.0, "step_paused_s": 92.0 }, { "step": 12, "length": 705.59375, "reward_all_samples": 0.625, "groups_all_wrong": 6.0, "groups_all_right": 13.0, "rollout_time_s": 32.550928, "rollout_end": "2026-10-22 06:50:04", "reward": 0.625, "cut_share": 0.289062, "logprob_abs_diff": 0.015102, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.560825, "clip_fraction": 0.0, "step_time_s": 180.768464, "log_probs_time_s": 57.966406, "train_time_s": 144.002183, "update_weights_time_s": 0.647575, "sleep_time_s": 0.745795, "wake_up_time_s": 0.734715, "step_end": "2026-10-22 06:52:29", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 13, "length": 671.085938, "reward_all_samples": 0.726562, "groups_all_wrong": 4.0, "groups_all_right": 17.0, "rollout_time_s": 31.086641, "rollout_end": "2026-10-22 06:53:03", "reward": 0.726562, "cut_share": 0.273438, "logprob_abs_diff": 0.014832, "train_rollout_kl": 0.000681, "entropy": 0.0, "grad_norm": 0.450333, "clip_fraction": 0.0, "step_time_s": 132.725842, "log_probs_time_s": 12.252874, "train_time_s": 97.338156, "update_weights_time_s": 0.643341, "sleep_time_s": 0.722668, "wake_up_time_s": 0.846998, "step_end": "2026-10-22 06:54:42", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 14, "length": 729.640625, "reward_all_samples": 0.609375, "groups_all_wrong": 7.0, "groups_all_right": 15.0, "rollout_time_s": 34.154217, "rollout_end": "2026-10-22 06:55:19", "reward": 0.609375, "cut_share": 0.3125, "logprob_abs_diff": 0.016151, "train_rollout_kl": 0.000726, "entropy": 0.0, "grad_norm": 0.409368, "clip_fraction": 0.0, "step_time_s": 137.453252, "log_probs_time_s": 12.818958, "train_time_s": 99.192697, "update_weights_time_s": 0.648548, "sleep_time_s": 0.737645, "wake_up_time_s": 0.749182, "step_end": "2026-10-22 06:56:59", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 15, "length": 662.921875, "reward_all_samples": 0.734375, "groups_all_wrong": 4.0, "groups_all_right": 17.0, "rollout_time_s": 29.654162, "rollout_end": "2026-10-22 06:57:32", "reward": 0.734375, "cut_share": 0.15625, "logprob_abs_diff": 0.014753, "train_rollout_kl": 0.000687, "entropy": 0.0, "grad_norm": 0.444428, "clip_fraction": 0.0, "step_time_s": 175.909669, "log_probs_time_s": 57.43987, "train_time_s": 142.034851, "update_weights_time_s": 0.633362, "sleep_time_s": 0.720173, "wake_up_time_s": 0.758865, "step_end": "2026-10-22 06:59:55", "rollout_paused_s": 0.0, "step_paused_s": 90.0 }, { "step": 16, "length": 744.835938, "reward_all_samples": 0.59375, "groups_all_wrong": 4.0, "groups_all_right": 10.0, "rollout_time_s": 34.435756, "rollout_end": "2026-10-22 07:00:32", "reward": 0.59375, "cut_share": 0.335938, "logprob_abs_diff": 0.015398, "train_rollout_kl": 0.000683, "entropy": 0.0, "grad_norm": 0.578723, "clip_fraction": 0.0, "step_time_s": 138.205484, "log_probs_time_s": 58.394097, "train_time_s": 99.73013, "update_weights_time_s": 0.614058, "sleep_time_s": 0.695588, "wake_up_time_s": 0.753783, "step_end": "2026-10-22 07:02:13", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 17, "length": 744.234375, "reward_all_samples": 0.59375, "groups_all_wrong": 6.0, "groups_all_right": 14.0, "rollout_time_s": 34.723223, "rollout_end": "2026-10-22 07:02:51", "reward": 0.59375, "cut_share": 0.359375, "logprob_abs_diff": 0.015576, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.456027, "clip_fraction": 0.0, "step_time_s": 93.302586, "log_probs_time_s": 13.098256, "train_time_s": 54.448705, "update_weights_time_s": 0.656775, "sleep_time_s": 0.729179, "wake_up_time_s": 0.760563, "step_end": "2026-10-22 07:03:47", "rollout_paused_s": 0.0, "step_paused_s": 46.0 }, { "step": 18, "length": 665.453125, "reward_all_samples": 0.710938, "groups_all_wrong": 4.0, "groups_all_right": 16.0, "rollout_time_s": 30.927283, "rollout_end": "2026-10-22 07:04:20", "reward": 0.710938, "cut_share": 0.25, "logprob_abs_diff": 0.015534, "train_rollout_kl": 0.000724, "entropy": 0.0, "grad_norm": 0.48593, "clip_fraction": 0.0, "step_time_s": 87.066544, "log_probs_time_s": 12.236744, "train_time_s": 51.911438, "update_weights_time_s": 0.652277, "sleep_time_s": 0.735787, "wake_up_time_s": 0.844277, "step_end": "2026-10-22 07:05:14", "rollout_paused_s": 30.9, "step_paused_s": 87.1 }, { "step": 19, "length": 716.882812, "reward_all_samples": 0.703125, "groups_all_wrong": 4.0, "groups_all_right": 16.0, "rollout_time_s": 33.019113, "rollout_end": "2026-10-22 07:05:49", "reward": 0.703125, "cut_share": 0.296875, "logprob_abs_diff": 0.015099, "train_rollout_kl": 0.000682, "entropy": 0.0, "grad_norm": 0.456453, "clip_fraction": 0.0, "step_time_s": 90.580336, "log_probs_time_s": 12.782735, "train_time_s": 53.369736, "update_weights_time_s": 0.662006, "sleep_time_s": 0.769306, "wake_up_time_s": 0.756362, "step_end": "2026-10-22 07:06:44", "rollout_paused_s": 33.0, "step_paused_s": 90.6 }, { "step": 20, "length": 623.570312, "reward_all_samples": 0.804688, "groups_all_wrong": 2.0, "groups_all_right": 18.0, "rollout_time_s": 27.89495, "rollout_end": "2026-10-22 07:07:15", "reward": 0.804688, "cut_share": 0.125, "logprob_abs_diff": 0.013883, "train_rollout_kl": 0.000642, "entropy": 0.0, "grad_norm": 0.542843, "clip_fraction": 0.0, "step_time_s": 83.057476, "log_probs_time_s": 11.777903, "train_time_s": 50.814181, "update_weights_time_s": 0.622573, "sleep_time_s": 0.731838, "wake_up_time_s": 0.848305, "step_end": "2026-10-22 07:08:07", "rollout_paused_s": 27.9, "step_paused_s": 83.1 }, { "step": 21, "length": 661.382812, "reward_all_samples": 0.6875, "groups_all_wrong": 4.0, "groups_all_right": 12.0, "rollout_time_s": 30.617487, "rollout_end": "2026-10-22 07:08:41", "reward": 0.6875, "cut_share": 0.210938, "logprob_abs_diff": 0.014444, "train_rollout_kl": 0.000663, "entropy": 0.0, "grad_norm": 0.54209, "clip_fraction": 0.0, "step_time_s": 86.901661, "log_probs_time_s": 12.178797, "train_time_s": 52.027694, "update_weights_time_s": 0.645506, "sleep_time_s": 0.728919, "wake_up_time_s": 0.856605, "step_end": "2026-10-22 07:09:34", "rollout_paused_s": 30.6, "step_paused_s": 86.9 }, { "step": 22, "length": 607.054688, "reward_all_samples": 0.710938, "groups_all_wrong": 2.0, "groups_all_right": 12.0, "rollout_time_s": 27.396003, "rollout_end": "2026-10-22 07:10:04", "reward": 0.710938, "cut_share": 0.15625, "logprob_abs_diff": 0.015676, "train_rollout_kl": 0.00076, "entropy": 0.0, "grad_norm": 0.638271, "clip_fraction": 0.0, "step_time_s": 81.378025, "log_probs_time_s": 11.583917, "train_time_s": 49.825203, "update_weights_time_s": 0.62886, "sleep_time_s": 0.72095, "wake_up_time_s": 0.776788, "step_end": "2026-10-22 07:10:56", "rollout_paused_s": 27.4, "step_paused_s": 81.4 }, { "step": 23, "length": 619.429688, "reward_all_samples": 0.703125, "groups_all_wrong": 4.0, "groups_all_right": 18.0, "rollout_time_s": 28.625033, "rollout_end": "2026-10-22 07:11:27", "reward": 0.703125, "cut_share": 0.234375, "logprob_abs_diff": 0.015732, "train_rollout_kl": 0.000752, "entropy": 0.0, "grad_norm": 0.546791, "clip_fraction": 0.0, "step_time_s": 82.871769, "log_probs_time_s": 11.630265, "train_time_s": 49.994847, "update_weights_time_s": 0.6363, "sleep_time_s": 0.754588, "wake_up_time_s": 0.867991, "step_end": "2026-10-22 07:12:18", "rollout_paused_s": 28.6, "step_paused_s": 82.9 } ] }, "e2-n16": { "steps": 24, "partial": true, "median": { "reward": 0.691407, "length": 646.591797, "cut_share": 0.243164, "reward_all_samples": 0.691407, "groups_all_wrong": 3.0, "groups_all_right": 8.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015206, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.396374, "clip_fraction": 0.0, "step_time_s": 627.147432, "rollout_time_s": 188.187084, "log_probs_time_s": 136.370014, "train_time_s": 427.107799, "update_weights_time_s": 0.645126, "sleep_time_s": 0.736965, "wake_up_time_s": 0.849002 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": null, "steps": 0 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 809.396484, "reward_all_samples": 0.5, "groups_all_wrong": 6.0, "groups_all_right": 4.0, "rollout_time_s": 182.863885, "rollout_end": "2026-10-22 07:19:30", "reward": 0.5, "cut_share": 0.451172, "logprob_abs_diff": 0.015469, "train_rollout_kl": 0.000701, "entropy": 0.0, "grad_norm": 0.383024, "clip_fraction": 0.0, "step_time_s": 741.821803, "log_probs_time_s": 150.554326, "train_time_s": 506.277206, "update_weights_time_s": 0.64648, "sleep_time_s": 2.002712, "wake_up_time_s": 0.405212, "step_end": "2026-10-22 07:27:57", "rollout_paused_s": 140.9, "step_paused_s": 464.8 }, { "step": 1, "length": 697.986328, "reward_all_samples": 0.683594, "groups_all_wrong": 1.0, "groups_all_right": 7.0, "rollout_time_s": 201.675581, "rollout_end": "2026-10-22 07:31:26", "reward": 0.683594, "cut_share": 0.273438, "logprob_abs_diff": 0.015225, "train_rollout_kl": 0.000694, "entropy": 0.0, "grad_norm": 0.409023, "clip_fraction": 0.0, "step_time_s": 644.974401, "log_probs_time_s": 140.364237, "train_time_s": 434.986435, "update_weights_time_s": 0.642327, "sleep_time_s": 4.510744, "wake_up_time_s": 0.753021, "step_end": "2026-10-22 07:38:42", "rollout_paused_s": 92.0, "step_paused_s": 319.0 }, { "step": 2, "length": 773.71875, "reward_all_samples": 0.589844, "groups_all_wrong": 4.0, "groups_all_right": 3.0, "rollout_time_s": 261.361543, "rollout_end": "2026-10-22 07:43:07", "reward": 0.589844, "cut_share": 0.326172, "logprob_abs_diff": 0.015541, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.381372, "clip_fraction": 0.0, "step_time_s": 759.859751, "log_probs_time_s": 144.334137, "train_time_s": 493.512438, "update_weights_time_s": 0.620077, "sleep_time_s": 1.12945, "wake_up_time_s": 0.84961, "step_end": "2026-10-22 07:51:22", "rollout_paused_s": 135.0, "step_paused_s": 405.0 }, { "step": 3, "length": 784.189453, "reward_all_samples": 0.585938, "groups_all_wrong": 5.0, "groups_all_right": 6.0, "rollout_time_s": 266.36966, "rollout_end": "2026-10-22 07:55:52", "reward": 0.585938, "cut_share": 0.388672, "logprob_abs_diff": 0.014962, "train_rollout_kl": 0.000682, "entropy": 0.0, "grad_norm": 0.358876, "clip_fraction": 0.0, "step_time_s": 766.367221, "log_probs_time_s": 144.609893, "train_time_s": 495.105885, "update_weights_time_s": 0.658433, "sleep_time_s": 1.18958, "wake_up_time_s": 0.752454, "step_end": "2026-10-22 08:04:09", "rollout_paused_s": 136.0, "step_paused_s": 407.0 }, { "step": 4, "length": 669.710938, "reward_all_samples": 0.736328, "groups_all_right": 7.0, "rollout_time_s": 194.797572, "rollout_end": "2026-10-22 08:07:26", "reward": 0.736328, "cut_share": 0.240234, "logprob_abs_diff": 0.01491, "train_rollout_kl": 0.000687, "entropy": 0.0, "grad_norm": 0.405688, "clip_fraction": 0.0, "step_time_s": 630.821554, "log_probs_time_s": 93.751835, "train_time_s": 431.554062, "update_weights_time_s": 0.649052, "sleep_time_s": 0.737098, "wake_up_time_s": 0.842535, "step_end": "2026-10-22 08:14:39", "rollout_paused_s": 90.0, "step_paused_s": 316.0 }, { "step": 5, "length": 684.003906, "reward_all_samples": 0.630859, "groups_all_wrong": 6.0, "groups_all_right": 9.0, "rollout_time_s": 243.535727, "rollout_end": "2026-10-22 08:18:46", "reward": 0.630859, "cut_share": 0.292969, "logprob_abs_diff": 0.015687, "train_rollout_kl": 0.000715, "entropy": 0.0, "grad_norm": 0.34021, "clip_fraction": 0.0, "step_time_s": 638.214463, "log_probs_time_s": 94.816711, "train_time_s": 389.96151, "update_weights_time_s": 0.642775, "sleep_time_s": 0.736764, "wake_up_time_s": 0.749498, "step_end": "2026-10-22 08:25:18", "rollout_paused_s": 135.0, "step_paused_s": 315.0 }, { "step": 6, "length": 689.613281, "reward_all_samples": 0.621094, "groups_all_wrong": 5.0, "groups_all_right": 9.0, "rollout_time_s": 199.057463, "rollout_end": "2026-10-22 08:28:39", "reward": 0.621094, "cut_share": 0.291016, "logprob_abs_diff": 0.015186, "train_rollout_kl": 0.000688, "entropy": 0.0, "grad_norm": 0.37376, "clip_fraction": 0.0, "step_time_s": 638.629826, "log_probs_time_s": 140.047825, "train_time_s": 435.087277, "update_weights_time_s": 0.628541, "sleep_time_s": 0.715681, "wake_up_time_s": 0.867759, "step_end": "2026-10-22 08:35:56", "rollout_paused_s": 92.0, "step_paused_s": 318.0 }, { "step": 7, "length": 714.033203, "reward_all_samples": 0.642578, "groups_all_wrong": 5.0, "groups_all_right": 7.0, "rollout_time_s": 251.931509, "rollout_end": "2026-10-22 08:40:11", "reward": 0.642578, "cut_share": 0.351562, "logprob_abs_diff": 0.015813, "train_rollout_kl": 0.000721, "entropy": 0.0, "grad_norm": 0.365396, "clip_fraction": 0.0, "step_time_s": 649.546051, "log_probs_time_s": 141.255473, "train_time_s": 393.098952, "update_weights_time_s": 0.66322, "sleep_time_s": 0.731254, "wake_up_time_s": 0.850106, "step_end": "2026-10-22 08:46:46", "rollout_paused_s": 135.0, "step_paused_s": 315.0 }, { "step": 8, "length": 669.289062, "reward_all_samples": 0.669922, "groups_all_wrong": 3.0, "groups_all_right": 7.0, "rollout_time_s": 239.152666, "rollout_end": "2026-10-22 08:50:48", "reward": 0.669922, "cut_share": 0.292969, "logprob_abs_diff": 0.015985, "train_rollout_kl": 0.000772, "entropy": 0.0, "grad_norm": 0.416747, "clip_fraction": 0.0, "step_time_s": 631.95881, "log_probs_time_s": 139.447289, "train_time_s": 388.317187, "update_weights_time_s": 0.646066, "sleep_time_s": 0.733896, "wake_up_time_s": 0.85198, "step_end": "2026-10-22 08:57:18", "rollout_paused_s": 135.0, "step_paused_s": 316.0 }, { "step": 9, "length": 631.652344, "reward_all_samples": 0.720703, "groups_all_wrong": 1.0, "groups_all_right": 8.0, "rollout_time_s": 144.922626, "rollout_end": "2026-10-22 08:59:46", "reward": 0.720703, "cut_share": 0.214844, "logprob_abs_diff": 0.015088, "train_rollout_kl": 0.000701, "entropy": 0.0, "grad_norm": 0.412413, "clip_fraction": 0.0, "step_time_s": 619.99336, "log_probs_time_s": 136.997151, "train_time_s": 470.431791, "update_weights_time_s": 0.647865, "sleep_time_s": 0.764237, "wake_up_time_s": 0.857574, "step_end": "2026-10-22 09:07:38", "rollout_paused_s": 45.0, "step_paused_s": 317.0 }, { "step": 10, "length": 639.203125, "reward_all_samples": 0.730469, "groups_all_wrong": 2.0, "groups_all_right": 5.0, "rollout_time_s": 186.773551, "rollout_end": "2026-10-22 09:10:47", "reward": 0.730469, "cut_share": 0.173828, "logprob_abs_diff": 0.014874, "train_rollout_kl": 0.000685, "entropy": 0.0, "grad_norm": 0.434551, "clip_fraction": 0.0, "step_time_s": 621.520313, "log_probs_time_s": 138.184434, "train_time_s": 430.157576, "update_weights_time_s": 0.614504, "sleep_time_s": 0.708184, "wake_up_time_s": 0.856858, "step_end": "2026-10-22 09:17:59", "rollout_paused_s": 90.0, "step_paused_s": 316.0 }, { "step": 11, "length": 610.480469, "reward_all_samples": 0.699219, "groups_all_wrong": 3.0, "groups_all_right": 12.0, "rollout_time_s": 181.394194, "rollout_end": "2026-10-22 09:21:03", "reward": 0.699219, "cut_share": 0.199219, "logprob_abs_diff": 0.014893, "train_rollout_kl": 0.000681, "entropy": 0.0, "grad_norm": 0.43528, "clip_fraction": 0.0, "step_time_s": 653.680586, "log_probs_time_s": 136.131527, "train_time_s": 467.792651, "update_weights_time_s": 0.622507, "sleep_time_s": 0.757552, "wake_up_time_s": 0.848394, "step_end": "2026-10-22 09:28:53", "rollout_paused_s": 90.0, "step_paused_s": 363.0 }, { "step": 12, "length": 643.179688, "reward_all_samples": 0.646484, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 189.600618, "rollout_end": "2026-10-22 09:32:05", "reward": 0.646484, "cut_share": 0.261719, "logprob_abs_diff": 0.015257, "train_rollout_kl": 0.000706, "entropy": 0.0, "grad_norm": 0.398366, "clip_fraction": 0.0, "step_time_s": 623.361319, "log_probs_time_s": 92.751356, "train_time_s": 429.291548, "update_weights_time_s": 0.61982, "sleep_time_s": 0.736787, "wake_up_time_s": 0.873528, "step_end": "2026-10-22 09:39:16", "rollout_paused_s": 92.0, "step_paused_s": 319.0 }, { "step": 13, "length": 612.558594, "reward_all_samples": 0.722656, "groups_all_right": 12.0, "rollout_time_s": 183.160589, "rollout_end": "2026-10-22 09:42:22", "reward": 0.722656, "cut_share": 0.21875, "logprob_abs_diff": 0.01485, "train_rollout_kl": 0.000679, "entropy": 0.0, "grad_norm": 0.394382, "clip_fraction": 0.0, "step_time_s": 612.635966, "log_probs_time_s": 136.566612, "train_time_s": 424.924051, "update_weights_time_s": 0.620034, "sleep_time_s": 0.713538, "wake_up_time_s": 0.755856, "step_end": "2026-10-22 09:49:29", "rollout_paused_s": 91.0, "step_paused_s": 320.0 }, { "step": 14, "length": 650.003906, "reward_all_samples": 0.636719, "groups_all_wrong": 2.0, "groups_all_right": 7.0, "rollout_time_s": 190.383927, "rollout_end": "2026-10-22 09:52:42", "reward": 0.636719, "cut_share": 0.251953, "logprob_abs_diff": 0.015909, "train_rollout_kl": 0.000726, "entropy": 0.0, "grad_norm": 0.447557, "clip_fraction": 0.0, "step_time_s": 624.729262, "log_probs_time_s": 138.434769, "train_time_s": 429.754025, "update_weights_time_s": 0.647757, "sleep_time_s": 0.747118, "wake_up_time_s": 0.768512, "step_end": "2026-10-22 09:59:54", "rollout_paused_s": 92.0, "step_paused_s": 318.0 }, { "step": 15, "length": 591.84375, "reward_all_samples": 0.742188, "groups_all_wrong": 3.0, "groups_all_right": 11.0, "rollout_time_s": 177.180059, "rollout_end": "2026-10-22 10:02:54", "reward": 0.742188, "cut_share": 0.158203, "logprob_abs_diff": 0.014977, "train_rollout_kl": 0.000693, "entropy": 0.0, "grad_norm": 0.373562, "clip_fraction": 0.0, "step_time_s": 512.894813, "log_probs_time_s": 135.548489, "train_time_s": 331.35956, "update_weights_time_s": 0.644906, "sleep_time_s": 0.736832, "wake_up_time_s": 0.756344, "step_end": "2026-10-22 10:08:27", "rollout_paused_s": 91.0, "step_paused_s": 226.0 }, { "step": 16, "length": 656.705078, "reward_all_samples": 0.638672, "groups_all_wrong": 3.0, "groups_all_right": 8.0, "rollout_time_s": 235.221004, "rollout_end": "2026-10-22 10:12:25", "reward": 0.638672, "cut_share": 0.246094, "logprob_abs_diff": 0.015026, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.391209, "clip_fraction": 0.0, "step_time_s": 625.664505, "log_probs_time_s": 93.412132, "train_time_s": 385.741158, "update_weights_time_s": 0.645347, "sleep_time_s": 0.752707, "wake_up_time_s": 0.847952, "step_end": "2026-10-22 10:18:52", "rollout_paused_s": 137.0, "step_paused_s": 318.0 }, { "step": 17, "length": 661.013672, "reward_all_samples": 0.650391, "groups_all_wrong": 3.0, "groups_all_right": 9.0, "rollout_time_s": 238.07393, "rollout_end": "2026-10-22 10:22:53", "reward": 0.650391, "cut_share": 0.277344, "logprob_abs_diff": 0.01573, "train_rollout_kl": 0.000731, "entropy": 0.0, "grad_norm": 0.38363, "clip_fraction": 0.0, "step_time_s": 628.630358, "log_probs_time_s": 93.664483, "train_time_s": 386.041944, "update_weights_time_s": 0.65136, "sleep_time_s": 0.704322, "wake_up_time_s": 0.75541, "step_end": "2026-10-22 10:29:21", "rollout_paused_s": 138.0, "step_paused_s": 320.0 }, { "step": 18, "length": 607.09375, "reward_all_samples": 0.732422, "groups_all_wrong": 1.0, "groups_all_right": 6.0, "rollout_time_s": 180.63482, "rollout_end": "2026-10-22 10:32:24", "reward": 0.732422, "cut_share": 0.203125, "logprob_abs_diff": 0.015379, "train_rollout_kl": 0.000716, "entropy": 0.0, "grad_norm": 0.474663, "clip_fraction": 0.0, "step_time_s": 609.221817, "log_probs_time_s": 136.402226, "train_time_s": 424.12447, "update_weights_time_s": 0.611687, "sleep_time_s": 0.721531, "wake_up_time_s": 0.773529, "step_end": "2026-10-22 10:39:30", "rollout_paused_s": 92.0, "step_paused_s": 318.0 }, { "step": 19, "length": 614.111328, "reward_all_samples": 0.736328, "groups_all_wrong": 2.0, "groups_all_right": 8.0, "rollout_time_s": 181.143502, "rollout_end": "2026-10-22 10:42:34", "reward": 0.736328, "cut_share": 0.152344, "logprob_abs_diff": 0.014989, "train_rollout_kl": 0.000695, "entropy": 0.0, "grad_norm": 0.423591, "clip_fraction": 0.0, "step_time_s": 610.249835, "log_probs_time_s": 136.337802, "train_time_s": 424.385025, "update_weights_time_s": 0.655314, "sleep_time_s": 0.749448, "wake_up_time_s": 0.871176, "step_end": "2026-10-22 10:49:40", "rollout_paused_s": 90.0, "step_paused_s": 317.0 }, { "step": 20, "length": 548.660156, "reward_all_samples": 0.808594, "groups_all_wrong": 2.0, "groups_all_right": 12.0, "rollout_time_s": 171.438688, "rollout_end": "2026-10-22 10:52:35", "reward": 0.808594, "cut_share": 0.119141, "logprob_abs_diff": 0.013973, "train_rollout_kl": 0.000658, "entropy": 0.0, "grad_norm": 0.376293, "clip_fraction": 0.0, "step_time_s": 590.736964, "log_probs_time_s": 133.477458, "train_time_s": 414.62774, "update_weights_time_s": 0.657922, "sleep_time_s": 0.759885, "wake_up_time_s": 0.864791, "step_end": "2026-10-22 10:59:31", "rollout_paused_s": 92.0, "step_paused_s": 319.0 }, { "step": 21, "length": 582.529297, "reward_all_samples": 0.714844, "groups_all_wrong": 1.0, "groups_all_right": 10.0, "rollout_time_s": 175.83057, "rollout_end": "2026-10-22 11:02:30", "reward": 0.714844, "cut_share": 0.160156, "logprob_abs_diff": 0.014631, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.386511, "clip_fraction": 0.0, "step_time_s": 646.980726, "log_probs_time_s": 89.964189, "train_time_s": 466.484547, "update_weights_time_s": 0.657086, "sleep_time_s": 0.785005, "wake_up_time_s": 0.853745, "step_end": "2026-10-22 11:10:18", "rollout_paused_s": 92.0, "step_paused_s": 364.0 }, { "step": 22, "length": 538.726562, "reward_all_samples": 0.761719, "groups_all_wrong": 1.0, "groups_all_right": 10.0, "rollout_time_s": 125.617671, "rollout_end": "2026-10-22 11:12:26", "reward": 0.761719, "cut_share": 0.136719, "logprob_abs_diff": 0.015403, "train_rollout_kl": 0.000717, "entropy": 0.0, "grad_norm": 0.464681, "clip_fraction": 0.0, "step_time_s": 380.167714, "log_probs_time_s": 88.017085, "train_time_s": 250.008935, "update_weights_time_s": 0.612648, "sleep_time_s": 0.72807, "wake_up_time_s": 0.853951, "step_end": "2026-10-22 11:16:38", "rollout_paused_s": 45.0, "step_paused_s": 233.0 }, { "step": 23, "length": 521.550781, "reward_all_samples": 0.726562, "groups_all_wrong": 2.0, "groups_all_right": 11.0, "rollout_time_s": 79.822141, "rollout_end": "2026-10-22 11:18:01", "reward": 0.726562, "cut_share": 0.166016, "logprob_abs_diff": 0.015432, "train_rollout_kl": 0.00072, "entropy": 0.0, "grad_norm": 0.458453, "clip_fraction": 0.0, "step_time_s": 271.384022, "log_probs_time_s": 41.845525, "train_time_s": 186.978473, "update_weights_time_s": 0.6264, "sleep_time_s": 0.722464, "wake_up_time_s": 0.89574, "step_end": "2026-10-22 11:21:10", "rollout_paused_s": 79.8, "step_paused_s": 271.4 } ] }, "e3e-base-frac025": { "steps": 12, "partial": false, "median": { "reward": 0.642578, "length": 704.75, "cut_share": 0.308594, "reward_all_samples": 0.642578, "groups_all_wrong": 4.5, "groups_all_right": 11.0, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.015318, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.43392, "clip_fraction": 0.0, "step_time_s": 327.841134, "rollout_time_s": 102.029164, "log_probs_time_s": 70.447114, "train_time_s": 240.348381, "update_weights_time_s": 0.62265, "sleep_time_s": 0.730943, "wake_up_time_s": 0.762332 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 60.677347, "steps": 4 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 804.65625, "reward_all_samples": 0.515625, "groups_all_wrong": 6.0, "groups_all_right": 7.0, "rollout_time_s": 69.403877, "rollout_end": "2026-10-22 11:43:49", "reward": 0.515625, "cut_share": 0.464844, "logprob_abs_diff": 0.015317, "train_rollout_kl": 0.000694, "entropy": 0.0, "grad_norm": 0.486586, "clip_fraction": 0.0, "step_time_s": 289.097431, "log_probs_time_s": 77.900015, "train_time_s": 168.083997, "update_weights_time_s": 0.610259, "sleep_time_s": 2.214079, "wake_up_time_s": 0.470545, "step_end": "2026-10-22 11:46:38", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 1, "length": 700.496094, "reward_all_samples": 0.707031, "groups_all_wrong": 3.0, "groups_all_right": 14.0, "rollout_time_s": 58.501218, "rollout_end": "2026-10-22 11:47:42", "reward": 0.707031, "cut_share": 0.265625, "logprob_abs_diff": 0.015326, "train_rollout_kl": 0.000723, "entropy": 0.0, "grad_norm": 0.449505, "clip_fraction": 0.0, "step_time_s": 306.293145, "log_probs_time_s": 70.40637, "train_time_s": 240.343924, "update_weights_time_s": 0.655655, "sleep_time_s": 4.485561, "wake_up_time_s": 0.76703, "step_end": "2026-10-22 11:51:44", "rollout_paused_s": 0.0, "step_paused_s": 137.0 }, { "step": 2, "length": 774.730469, "reward_all_samples": 0.582031, "groups_all_wrong": 4.0, "groups_all_right": 9.0, "rollout_time_s": 111.882275, "rollout_end": "2026-10-22 11:53:39", "reward": 0.582031, "cut_share": 0.355469, "logprob_abs_diff": 0.015502, "train_rollout_kl": 0.0007, "entropy": 0.0, "grad_norm": 0.421086, "clip_fraction": 0.0, "step_time_s": 362.93212, "log_probs_time_s": 117.635871, "train_time_s": 246.950758, "update_weights_time_s": 0.611474, "sleep_time_s": 1.209601, "wake_up_time_s": 0.757634, "step_end": "2026-10-22 11:57:47", "rollout_paused_s": 46.0, "step_paused_s": 181.0 }, { "step": 3, "length": 807.472656, "reward_all_samples": 0.59375, "groups_all_wrong": 5.0, "groups_all_right": 8.0, "rollout_time_s": 115.857764, "rollout_end": "2026-10-22 11:59:46", "reward": 0.59375, "cut_share": 0.421875, "logprob_abs_diff": 0.014958, "train_rollout_kl": 0.000675, "entropy": 0.0, "grad_norm": 0.435449, "clip_fraction": 0.0, "step_time_s": 415.154186, "log_probs_time_s": 73.203066, "train_time_s": 295.213698, "update_weights_time_s": 0.616625, "sleep_time_s": 1.084744, "wake_up_time_s": 0.84564, "step_end": "2026-10-22 12:04:42", "rollout_paused_s": 45.0, "step_paused_s": 226.0 }, { "step": 4, "length": 687.996094, "reward_all_samples": 0.738281, "groups_all_wrong": 1.0, "groups_all_right": 13.0, "rollout_time_s": 101.020099, "rollout_end": "2026-10-22 12:06:26", "reward": 0.738281, "cut_share": 0.246094, "logprob_abs_diff": 0.014694, "train_rollout_kl": 0.000677, "entropy": 0.0, "grad_norm": 0.431214, "clip_fraction": 0.0, "step_time_s": 345.058501, "log_probs_time_s": 70.151538, "train_time_s": 240.352838, "update_weights_time_s": 0.611566, "sleep_time_s": 0.709863, "wake_up_time_s": 0.860476, "step_end": "2026-10-22 12:10:27", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 5, "length": 710.867188, "reward_all_samples": 0.628906, "groups_all_wrong": 6.0, "groups_all_right": 10.0, "rollout_time_s": 105.420364, "rollout_end": "2026-10-22 12:12:15", "reward": 0.628906, "cut_share": 0.324219, "logprob_abs_diff": 0.015626, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.432391, "clip_fraction": 0.0, "step_time_s": 305.96873, "log_probs_time_s": 70.615822, "train_time_s": 196.9952, "update_weights_time_s": 0.643277, "sleep_time_s": 0.722378, "wake_up_time_s": 0.753141, "step_end": "2026-10-22 12:15:33", "rollout_paused_s": 45.0, "step_paused_s": 135.0 }, { "step": 6, "length": 709.003906, "reward_all_samples": 0.601562, "groups_all_wrong": 6.0, "groups_all_right": 12.0, "rollout_time_s": 150.498763, "rollout_end": "2026-10-22 12:18:06", "reward": 0.601562, "cut_share": 0.324219, "logprob_abs_diff": 0.015318, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.384966, "clip_fraction": 0.0, "step_time_s": 350.010906, "log_probs_time_s": 70.487857, "train_time_s": 195.800665, "update_weights_time_s": 0.631169, "sleep_time_s": 0.703609, "wake_up_time_s": 0.848285, "step_end": "2026-10-22 12:21:23", "rollout_paused_s": 90.0, "step_paused_s": 181.0 }, { "step": 7, "length": 738.191406, "reward_all_samples": 0.621094, "groups_all_wrong": 6.0, "groups_all_right": 11.0, "rollout_time_s": 62.853477, "rollout_end": "2026-10-22 12:22:28", "reward": 0.621094, "cut_share": 0.371094, "logprob_abs_diff": 0.015334, "train_rollout_kl": 0.000688, "entropy": 0.0, "grad_norm": 0.407978, "clip_fraction": 0.0, "step_time_s": 310.623768, "log_probs_time_s": 71.405864, "train_time_s": 244.066852, "update_weights_time_s": 0.627704, "sleep_time_s": 0.713591, "wake_up_time_s": 0.889761, "step_end": "2026-10-22 12:26:34", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 8, "length": 688.816406, "reward_all_samples": 0.664062, "groups_all_wrong": 7.0, "groups_all_right": 13.0, "rollout_time_s": 55.838781, "rollout_end": "2026-10-22 12:27:32", "reward": 0.664062, "cut_share": 0.292969, "logprob_abs_diff": 0.015375, "train_rollout_kl": 0.000741, "entropy": 0.0, "grad_norm": 0.356802, "clip_fraction": 0.0, "step_time_s": 255.102975, "log_probs_time_s": 70.213237, "train_time_s": 195.65096, "update_weights_time_s": 0.617597, "sleep_time_s": 0.725241, "wake_up_time_s": 0.753467, "step_end": "2026-10-22 12:30:49", "rollout_paused_s": 0.0, "step_paused_s": 92.0 }, { "step": 9, "length": 686.90625, "reward_all_samples": 0.683594, "groups_all_wrong": 3.0, "groups_all_right": 8.0, "rollout_time_s": 102.186108, "rollout_end": "2026-10-22 12:32:33", "reward": 0.683594, "cut_share": 0.269531, "logprob_abs_diff": 0.015271, "train_rollout_kl": 0.000705, "entropy": 0.0, "grad_norm": 0.506407, "clip_fraction": 0.0, "step_time_s": 346.106339, "log_probs_time_s": 70.065278, "train_time_s": 240.370952, "update_weights_time_s": 0.631154, "sleep_time_s": 0.719253, "wake_up_time_s": 0.749962, "step_end": "2026-10-22 12:36:35", "rollout_paused_s": 46.0, "step_paused_s": 182.0 }, { "step": 10, "length": 686.875, "reward_all_samples": 0.707031, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 101.872219, "rollout_end": "2026-10-22 12:38:19", "reward": 0.707031, "cut_share": 0.238281, "logprob_abs_diff": 0.015155, "train_rollout_kl": 0.000716, "entropy": 0.0, "grad_norm": 0.489586, "clip_fraction": 0.0, "step_time_s": 346.394768, "log_probs_time_s": 70.260758, "train_time_s": 240.865486, "update_weights_time_s": 0.664465, "sleep_time_s": 0.754834, "wake_up_time_s": 0.772762, "step_end": "2026-10-22 12:42:21", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 11, "length": 675.582031, "reward_all_samples": 0.65625, "groups_all_wrong": 2.0, "groups_all_right": 11.0, "rollout_time_s": 145.570436, "rollout_end": "2026-10-22 12:44:49", "reward": 0.65625, "cut_share": 0.269531, "logprob_abs_diff": 0.014686, "train_rollout_kl": 0.00067, "entropy": 0.0, "grad_norm": 0.563065, "clip_fraction": 0.0, "step_time_s": 297.808983, "log_probs_time_s": 24.403078, "train_time_s": 148.627702, "update_weights_time_s": 0.612862, "sleep_time_s": 0.736645, "wake_up_time_s": 0.733582, "step_end": "2026-10-22 12:47:19", "rollout_paused_s": 92.0, "step_paused_s": 138.0 } ] }, "e3f-no-offload-rollout-frac025": { "steps": 12, "partial": false, "median": { "reward": 0.65039, "length": 703.804688, "cut_share": 0.298828, "reward_all_samples": 0.65039, "groups_all_wrong": 5.0, "groups_all_right": 11.5, "dropped_all_wrong": null, "dropped_all_right": null, "logprob_abs_diff": 0.0152, "train_rollout_kl": 0.000692, "entropy": 0.0, "grad_norm": 0.451789, "clip_fraction": 0.0, "step_time_s": 337.202225, "rollout_time_s": 102.197377, "log_probs_time_s": 69.664549, "train_time_s": 239.053403, "update_weights_time_s": 0.643786, "sleep_time_s": 0.751591, "wake_up_time_s": 0.824172 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": 56.699827, "steps": 3 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 796.667969, "reward_all_samples": 0.535156, "groups_all_wrong": 8.0, "groups_all_right": 7.0, "rollout_time_s": 68.043557, "rollout_end": "2026-10-22 12:51:45", "reward": 0.535156, "cut_share": 0.402344, "logprob_abs_diff": 0.015182, "train_rollout_kl": 0.00068, "entropy": 0.0, "grad_norm": 0.440103, "clip_fraction": 0.0, "step_time_s": 327.603357, "log_probs_time_s": 77.002031, "train_time_s": 256.103493, "update_weights_time_s": 0.641524, "sleep_time_s": 2.059983, "wake_up_time_s": 0.4147, "step_end": "2026-10-22 12:56:02", "rollout_paused_s": 0.0, "step_paused_s": 135.0 }, { "step": 1, "length": 702.066406, "reward_all_samples": 0.699219, "groups_all_wrong": 1.0, "groups_all_right": 13.0, "rollout_time_s": 56.699827, "rollout_end": "2026-10-22 12:57:03", "reward": 0.699219, "cut_share": 0.265625, "logprob_abs_diff": 0.015022, "train_rollout_kl": 0.000682, "entropy": 0.0, "grad_norm": 0.504599, "clip_fraction": 0.0, "step_time_s": 211.848376, "log_probs_time_s": 24.619503, "train_time_s": 149.086437, "update_weights_time_s": 0.646048, "sleep_time_s": 4.405957, "wake_up_time_s": 0.824232, "step_end": "2026-10-22 12:59:33", "rollout_paused_s": 0.0, "step_paused_s": 45.0 }, { "step": 2, "length": 760.789062, "reward_all_samples": 0.605469, "groups_all_wrong": 4.0, "groups_all_right": 11.0, "rollout_time_s": 153.96347, "rollout_end": "2026-10-22 13:02:09", "reward": 0.605469, "cut_share": 0.359375, "logprob_abs_diff": 0.015465, "train_rollout_kl": 0.000713, "entropy": 0.0, "grad_norm": 0.461714, "clip_fraction": 0.0, "step_time_s": 400.884683, "log_probs_time_s": 71.46366, "train_time_s": 244.095802, "update_weights_time_s": 0.646078, "sleep_time_s": 1.155043, "wake_up_time_s": 0.832505, "step_end": "2026-10-22 13:06:14", "rollout_paused_s": 90.0, "step_paused_s": 227.0 }, { "step": 3, "length": 787.191406, "reward_all_samples": 0.601562, "groups_all_wrong": 5.0, "groups_all_right": 8.0, "rollout_time_s": 111.719587, "rollout_end": "2026-10-22 13:08:08", "reward": 0.601562, "cut_share": 0.371094, "logprob_abs_diff": 0.014784, "train_rollout_kl": 0.000682, "entropy": 0.0, "grad_norm": 0.46288, "clip_fraction": 0.0, "step_time_s": 361.452941, "log_probs_time_s": 72.22578, "train_time_s": 247.069297, "update_weights_time_s": 0.625821, "sleep_time_s": 1.098982, "wake_up_time_s": 0.733054, "step_end": "2026-10-22 13:12:16", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 4, "length": 688.648438, "reward_all_samples": 0.746094, "groups_all_right": 13.0, "rollout_time_s": 146.159821, "rollout_end": "2026-10-22 13:14:43", "reward": 0.746094, "cut_share": 0.230469, "logprob_abs_diff": 0.014838, "train_rollout_kl": 0.000675, "entropy": 0.0, "grad_norm": 0.449966, "clip_fraction": 0.0, "step_time_s": 296.201222, "log_probs_time_s": 24.254319, "train_time_s": 147.718359, "update_weights_time_s": 0.623464, "sleep_time_s": 0.69774, "wake_up_time_s": 0.814301, "step_end": "2026-10-22 13:17:12", "rollout_paused_s": 91.0, "step_paused_s": 137.0 }, { "step": 5, "length": 705.542969, "reward_all_samples": 0.621094, "groups_all_wrong": 6.0, "groups_all_right": 12.0, "rollout_time_s": 148.802578, "rollout_end": "2026-10-22 13:19:42", "reward": 0.621094, "cut_share": 0.332031, "logprob_abs_diff": 0.015767, "train_rollout_kl": 0.000714, "entropy": 0.0, "grad_norm": 0.376376, "clip_fraction": 0.0, "step_time_s": 347.295317, "log_probs_time_s": 25.059161, "train_time_s": 196.068905, "update_weights_time_s": 0.670371, "sleep_time_s": 0.735292, "wake_up_time_s": 0.824113, "step_end": "2026-10-22 13:22:59", "rollout_paused_s": 90.0, "step_paused_s": 182.0 }, { "step": 6, "length": 707.542969, "reward_all_samples": 0.605469, "groups_all_wrong": 5.0, "groups_all_right": 10.0, "rollout_time_s": 148.921975, "rollout_end": "2026-10-22 13:25:30", "reward": 0.605469, "cut_share": 0.304688, "logprob_abs_diff": 0.015219, "train_rollout_kl": 0.000697, "entropy": 0.0, "grad_norm": 0.482804, "clip_fraction": 0.0, "step_time_s": 391.65117, "log_probs_time_s": 70.155569, "train_time_s": 240.361371, "update_weights_time_s": 0.622218, "sleep_time_s": 0.712872, "wake_up_time_s": 0.841116, "step_end": "2026-10-22 13:29:31", "rollout_paused_s": 90.0, "step_paused_s": 227.0 }, { "step": 7, "length": 733.824219, "reward_all_samples": 0.628906, "groups_all_wrong": 6.0, "groups_all_right": 11.0, "rollout_time_s": 106.473987, "rollout_end": "2026-10-22 13:31:19", "reward": 0.628906, "cut_share": 0.359375, "logprob_abs_diff": 0.015882, "train_rollout_kl": 0.000719, "entropy": 0.0, "grad_norm": 0.39563, "clip_fraction": 0.0, "step_time_s": 350.991424, "log_probs_time_s": 116.171187, "train_time_s": 242.257106, "update_weights_time_s": 0.611152, "sleep_time_s": 0.693251, "wake_up_time_s": 0.749166, "step_end": "2026-10-22 13:35:22", "rollout_paused_s": 45.0, "step_paused_s": 180.0 }, { "step": 8, "length": 680.664062, "reward_all_samples": 0.671875, "groups_all_wrong": 5.0, "groups_all_right": 15.0, "rollout_time_s": 55.155715, "rollout_end": "2026-10-22 13:36:18", "reward": 0.671875, "cut_share": 0.292969, "logprob_abs_diff": 0.015403, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.403662, "clip_fraction": 0.0, "step_time_s": 296.955293, "log_probs_time_s": 69.838416, "train_time_s": 239.430747, "update_weights_time_s": 0.633425, "sleep_time_s": 0.691311, "wake_up_time_s": 0.848611, "step_end": "2026-10-22 13:40:19", "rollout_paused_s": 0.0, "step_paused_s": 136.0 }, { "step": 9, "length": 646.03125, "reward_all_samples": 0.730469, "groups_all_wrong": 2.0, "groups_all_right": 13.0, "rollout_time_s": 96.7452, "rollout_end": "2026-10-22 13:41:57", "reward": 0.730469, "cut_share": 0.230469, "logprob_abs_diff": 0.01525, "train_rollout_kl": 0.000701, "entropy": 0.0, "grad_norm": 0.453612, "clip_fraction": 0.0, "step_time_s": 335.362951, "log_probs_time_s": 68.773657, "train_time_s": 236.250461, "update_weights_time_s": 0.665129, "sleep_time_s": 0.769895, "wake_up_time_s": 0.742214, "step_end": "2026-10-22 13:45:54", "rollout_paused_s": 45.0, "step_paused_s": 182.0 }, { "step": 10, "length": 661.332031, "reward_all_samples": 0.710938, "groups_all_wrong": 3.0, "groups_all_right": 10.0, "rollout_time_s": 97.920768, "rollout_end": "2026-10-22 13:47:34", "reward": 0.710938, "cut_share": 0.210938, "logprob_abs_diff": 0.014977, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.498077, "clip_fraction": 0.0, "step_time_s": 339.0415, "log_probs_time_s": 69.490682, "train_time_s": 238.676058, "update_weights_time_s": 0.661206, "sleep_time_s": 0.7453, "wake_up_time_s": 0.846136, "step_end": "2026-10-22 13:51:33", "rollout_paused_s": 46.0, "step_paused_s": 182.0 }, { "step": 11, "length": 621.5, "reward_all_samples": 0.714844, "groups_all_wrong": 4.0, "groups_all_right": 15.0, "rollout_time_s": 94.551315, "rollout_end": "2026-10-22 13:53:09", "reward": 0.714844, "cut_share": 0.234375, "logprob_abs_diff": 0.014734, "train_rollout_kl": 0.000668, "entropy": 0.0, "grad_norm": 0.377234, "clip_fraction": 0.0, "step_time_s": 331.402758, "log_probs_time_s": 68.312382, "train_time_s": 234.392806, "update_weights_time_s": 0.656993, "sleep_time_s": 0.757882, "wake_up_time_s": 0.848506, "step_end": "2026-10-22 13:57:05", "rollout_paused_s": 46.0, "step_paused_s": 184.0 } ] }, "e2b-dapo-filter": { "steps": 12, "partial": true, "median": { "reward": 0.617188, "length": 754.259766, "cut_share": 0.322265, "reward_all_samples": 0.67847, "groups_all_wrong": null, "groups_all_right": null, "dropped_all_wrong": 7.0, "dropped_all_right": 29.5, "logprob_abs_diff": 0.015305, "train_rollout_kl": 0.000699, "entropy": 0.0, "grad_norm": 0.637341, "clip_fraction": 0.0, "step_time_s": 593.055625, "rollout_time_s": 340.682244, "log_probs_time_s": 71.860157, "train_time_s": 248.551075, "update_weights_time_s": 0.635537, "sleep_time_s": 0.74226, "wake_up_time_s": 0.847054 }, "median_unpaused": { "step_time_s": { "median": null, "steps": 0 }, "rollout_time_s": { "median": null, "steps": 0 }, "log_probs_time_s": { "median": null, "steps": 0 }, "train_time_s": { "median": null, "steps": 0 }, "update_weights_time_s": { "median": null, "steps": 0 }, "sleep_time_s": { "median": null, "steps": 0 }, "wake_up_time_s": { "median": null, "steps": 0 } }, "per_step": [ { "step": 0, "length": 786.558594, "reward_all_samples": 0.659722, "dropped_all_wrong": 10.0, "dropped_all_right": 26.0, "rollout_time_s": 322.463871, "rollout_end": "2026-10-22 14:05:35", "reward": 0.605469, "cut_share": 0.355469, "logprob_abs_diff": 0.014765, "train_rollout_kl": 0.000686, "entropy": 0.0, "grad_norm": 0.611441, "clip_fraction": 0.0, "step_time_s": 632.43666, "log_probs_time_s": 77.037314, "train_time_s": 257.1603, "update_weights_time_s": 0.61971, "sleep_time_s": 2.015746, "wake_up_time_s": 0.422934, "step_end": "2026-10-22 14:09:54", "rollout_paused_s": 135.0, "step_paused_s": 272.0 }, { "step": 1, "length": 798.324219, "reward_all_samples": 0.619792, "dropped_all_wrong": 7.0, "dropped_all_right": 9.0, "rollout_time_s": 185.490095, "rollout_end": "2026-10-22 14:13:06", "reward": 0.648438, "cut_share": 0.382812, "logprob_abs_diff": 0.014919, "train_rollout_kl": 0.000676, "entropy": 0.0, "grad_norm": 0.577685, "clip_fraction": 0.0, "step_time_s": 442.227279, "log_probs_time_s": 72.824242, "train_time_s": 248.627842, "update_weights_time_s": 0.651885, "sleep_time_s": 4.56825, "wake_up_time_s": 0.761879, "step_end": "2026-10-22 14:17:16", "rollout_paused_s": 92.0, "step_paused_s": 227.0 }, { "step": 2, "length": 724.726562, "reward_all_samples": 0.724265, "dropped_all_wrong": 6.0, "dropped_all_right": 30.0, "rollout_time_s": 346.407786, "rollout_end": "2026-10-22 14:23:06", "reward": 0.601562, "cut_share": 0.300781, "logprob_abs_diff": 0.015434, "train_rollout_kl": 0.000708, "entropy": 0.0, "grad_norm": 0.696178, "clip_fraction": 0.0, "step_time_s": 594.824077, "log_probs_time_s": 71.03706, "train_time_s": 243.608562, "update_weights_time_s": 0.616702, "sleep_time_s": 1.104151, "wake_up_time_s": 0.845058, "step_end": "2026-10-22 14:27:11", "rollout_paused_s": 182.0, "step_paused_s": 318.0 }, { "step": 3, "length": 692.722656, "reward_all_samples": 0.685065, "dropped_all_wrong": 13.0, "dropped_all_right": 31.0, "rollout_time_s": 343.830449, "rollout_end": "2026-10-22 14:32:58", "reward": 0.652344, "cut_share": 0.261719, "logprob_abs_diff": 0.015308, "train_rollout_kl": 0.000704, "entropy": 0.0, "grad_norm": 0.677193, "clip_fraction": 0.0, "step_time_s": 589.975367, "log_probs_time_s": 70.377278, "train_time_s": 241.291393, "update_weights_time_s": 0.629017, "sleep_time_s": 1.220888, "wake_up_time_s": 0.867039, "step_end": "2026-10-22 14:37:01", "rollout_paused_s": 181.0, "step_paused_s": 317.0 }, { "step": 4, "length": 707.355469, "reward_all_samples": 0.725746, "dropped_all_wrong": 6.0, "dropped_all_right": 29.0, "rollout_time_s": 336.603237, "rollout_end": "2026-10-22 14:42:40", "reward": 0.613281, "cut_share": 0.285156, "logprob_abs_diff": 0.015017, "train_rollout_kl": 0.000693, "entropy": 0.0, "grad_norm": 0.732289, "clip_fraction": 0.0, "step_time_s": 627.950909, "log_probs_time_s": 115.759101, "train_time_s": 286.858672, "update_weights_time_s": 0.640836, "sleep_time_s": 0.741474, "wake_up_time_s": 0.855584, "step_end": "2026-10-22 14:47:29", "rollout_paused_s": 180.0, "step_paused_s": 361.0 }, { "step": 5, "length": 799.535156, "reward_all_samples": 0.66129, "dropped_all_wrong": 7.0, "dropped_all_right": 23.0, "rollout_time_s": 364.286047, "rollout_end": "2026-10-22 14:53:36", "reward": 0.5625, "cut_share": 0.386719, "logprob_abs_diff": 0.015845, "train_rollout_kl": 0.000711, "entropy": 0.0, "grad_norm": 0.619081, "clip_fraction": 0.0, "step_time_s": 617.033586, "log_probs_time_s": 118.137637, "train_time_s": 248.474308, "update_weights_time_s": 0.639546, "sleep_time_s": 0.748076, "wake_up_time_s": 0.747777, "step_end": "2026-10-22 14:57:46", "rollout_paused_s": 182.0, "step_paused_s": 318.0 }, { "step": 6, "length": 743.585938, "reward_all_samples": 0.701014, "dropped_all_wrong": 11.0, "dropped_all_right": 31.0, "rollout_time_s": 403.003726, "rollout_end": "2026-10-22 15:04:31", "reward": 0.652344, "cut_share": 0.3125, "logprob_abs_diff": 0.015301, "train_rollout_kl": 0.000694, "entropy": 0.0, "grad_norm": 0.632371, "clip_fraction": 0.0, "step_time_s": 606.82753, "log_probs_time_s": 71.453765, "train_time_s": 199.509335, "update_weights_time_s": 0.627784, "sleep_time_s": 0.708513, "wake_up_time_s": 0.851035, "step_end": "2026-10-22 15:07:52", "rollout_paused_s": 225.0, "step_paused_s": 317.0 }, { "step": 7, "length": 767.808594, "reward_all_samples": 0.671875, "dropped_all_wrong": 4.0, "dropped_all_right": 12.0, "rollout_time_s": 177.918424, "rollout_end": "2026-10-22 15:10:53", "reward": 0.632812, "cut_share": 0.332031, "logprob_abs_diff": 0.015145, "train_rollout_kl": 0.000696, "entropy": 0.0, "grad_norm": 0.688081, "clip_fraction": 0.0, "step_time_s": 472.97018, "log_probs_time_s": 71.791037, "train_time_s": 290.790328, "update_weights_time_s": 0.629322, "sleep_time_s": 0.711543, "wake_up_time_s": 0.760769, "step_end": "2026-10-22 15:15:45", "rollout_paused_s": 92.0, "step_paused_s": 273.0 }, { "step": 8, "length": 813.582031, "reward_all_samples": 0.638021, "dropped_all_wrong": 3.0, "dropped_all_right": 13.0, "rollout_time_s": 179.743647, "rollout_end": "2026-10-22 15:18:48", "reward": 0.550781, "cut_share": 0.410156, "logprob_abs_diff": 0.015797, "train_rollout_kl": 0.000702, "entropy": 0.0, "grad_norm": 0.580255, "clip_fraction": 0.0, "step_time_s": 433.996773, "log_probs_time_s": 27.808323, "train_time_s": 249.939218, "update_weights_time_s": 0.644129, "sleep_time_s": 0.721087, "wake_up_time_s": 0.825344, "step_end": "2026-10-22 15:22:59", "rollout_paused_s": 91.0, "step_paused_s": 227.0 }, { "step": 9, "length": 702.390625, "reward_all_samples": 0.753571, "dropped_all_wrong": 6.0, "dropped_all_right": 31.0, "rollout_time_s": 344.710526, "rollout_end": "2026-10-22 15:28:47", "reward": 0.675781, "cut_share": 0.253906, "logprob_abs_diff": 0.015977, "train_rollout_kl": 0.000717, "entropy": 0.0, "grad_norm": 0.723046, "clip_fraction": 0.0, "step_time_s": 591.287173, "log_probs_time_s": 70.665529, "train_time_s": 242.19443, "update_weights_time_s": 0.644878, "sleep_time_s": 0.74086, "wake_up_time_s": 0.860636, "step_end": "2026-10-22 15:32:51", "rollout_paused_s": 180.0, "step_paused_s": 317.0 }, { "step": 10, "length": 756.121094, "reward_all_samples": 0.723684, "dropped_all_wrong": 8.0, "dropped_all_right": 35.0, "rollout_time_s": 338.30519, "rollout_end": "2026-10-22 15:38:32", "reward": 0.621094, "cut_share": 0.34375, "logprob_abs_diff": 0.015282, "train_rollout_kl": 0.000689, "entropy": 0.0, "grad_norm": 0.630947, "clip_fraction": 0.0, "step_time_s": 542.693492, "log_probs_time_s": 71.929276, "train_time_s": 199.827899, "update_weights_time_s": 0.662013, "sleep_time_s": 0.739895, "wake_up_time_s": 0.854881, "step_end": "2026-10-22 15:41:53", "rollout_paused_s": 180.0, "step_paused_s": 271.0 }, { "step": 11, "length": 752.398438, "reward_all_samples": 0.66276, "dropped_all_wrong": 12.0, "dropped_all_right": 36.0, "rollout_time_s": 343.059298, "rollout_end": "2026-10-22 15:47:39", "reward": 0.613281, "cut_share": 0.304688, "logprob_abs_diff": 0.015519, "train_rollout_kl": 0.000707, "entropy": 0.0, "grad_norm": 0.64231, "clip_fraction": 0.0, "step_time_s": 637.882823, "log_probs_time_s": 117.079754, "train_time_s": 290.44426, "update_weights_time_s": 0.631529, "sleep_time_s": 0.743047, "wake_up_time_s": 0.849049, "step_end": "2026-10-22 15:52:31", "rollout_paused_s": 184.0, "step_paused_s": 366.0 } ] } } }