| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.16404199475065617, |
| "eval_steps": 500, |
| "global_step": 1000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.05, |
| "completions/max_terminated_length": 6.05, |
| "completions/mean_length": 4.61875, |
| "completions/mean_terminated_length": 4.61875, |
| "completions/min_length": 3.5, |
| "completions/min_terminated_length": 3.5, |
| "entropy": 0.8203600860819279, |
| "epoch": 0.0032808398950131233, |
| "frac_reward_zero_std": 0.3, |
| "grad_norm": 5.8125, |
| "learning_rate": 4.990500000000001e-06, |
| "loss": 0.06519373655319213, |
| "num_tokens": 5635.0, |
| "reward": 1.1270193845033645, |
| "reward_std": 0.04764379555272171, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.050634765625, |
| "rewards/reward_length/std": 0.015013676788657903, |
| "rewards/reward_semantic/mean": 0.43183712139725683, |
| "rewards/reward_semantic/std": 0.15772892236709596, |
| "step": 20, |
| "step_time": 1.6850911805999202 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 7.15, |
| "completions/max_terminated_length": 7.15, |
| "completions/mean_length": 5.00625, |
| "completions/mean_terminated_length": 5.00625, |
| "completions/min_length": 3.65, |
| "completions/min_terminated_length": 3.65, |
| "entropy": 0.929626553109847, |
| "epoch": 0.006561679790026247, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 0.0, |
| "learning_rate": 4.9805e-06, |
| "loss": 0.07899624109268188, |
| "num_tokens": 11660.0, |
| "reward": 1.0556338340044022, |
| "reward_std": 0.017533331926097163, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0642578125, |
| "rewards/reward_length/std": 0.013992739858804271, |
| "rewards/reward_semantic/mean": 0.19615575280040504, |
| "rewards/reward_semantic/std": 0.05699479430913925, |
| "step": 40, |
| "step_time": 1.79874156090018 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.5, |
| "completions/max_terminated_length": 5.5, |
| "completions/mean_length": 4.325, |
| "completions/mean_terminated_length": 4.325, |
| "completions/min_length": 3.5, |
| "completions/min_terminated_length": 3.5, |
| "entropy": 0.7027173407775991, |
| "epoch": 0.00984251968503937, |
| "frac_reward_zero_std": 0.3, |
| "grad_norm": 0.0, |
| "learning_rate": 4.9705e-06, |
| "loss": 0.023067036271095277, |
| "num_tokens": 17440.0, |
| "reward": 1.1290264517068862, |
| "reward_std": 0.03574381650541909, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.052685546875, |
| "rewards/reward_length/std": 0.012657346948981286, |
| "rewards/reward_semantic/mean": 0.4388690486550331, |
| "rewards/reward_semantic/std": 0.11865786015987397, |
| "step": 60, |
| "step_time": 1.6641022313501708 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 7.05, |
| "completions/max_terminated_length": 7.05, |
| "completions/mean_length": 4.89375, |
| "completions/mean_terminated_length": 4.89375, |
| "completions/min_length": 3.45, |
| "completions/min_terminated_length": 3.45, |
| "entropy": 0.8099277728608285, |
| "epoch": 0.013123359580052493, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 2.578125, |
| "learning_rate": 4.9605e-06, |
| "loss": 0.12280056476593018, |
| "num_tokens": 23631.0, |
| "reward": 1.08654942214489, |
| "reward_std": 0.040316034654097166, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.06103515625, |
| "rewards/reward_length/std": 0.022364351549185812, |
| "rewards/reward_semantic/mean": 0.2986706353724003, |
| "rewards/reward_semantic/std": 0.1318330906331539, |
| "step": 80, |
| "step_time": 1.7635547111498453 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.75, |
| "completions/max_terminated_length": 6.75, |
| "completions/mean_length": 4.75, |
| "completions/mean_terminated_length": 4.75, |
| "completions/min_length": 3.4, |
| "completions/min_terminated_length": 3.4, |
| "entropy": 0.8052310052706162, |
| "epoch": 0.016404199475065617, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 3.171875, |
| "learning_rate": 4.9505e-06, |
| "loss": 0.007156229019165039, |
| "num_tokens": 29599.0, |
| "reward": 1.0940562933683395, |
| "reward_std": 0.030761140065442304, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.06103515625, |
| "rewards/reward_length/std": 0.017728388286195697, |
| "rewards/reward_semantic/mean": 0.3236935347318649, |
| "rewards/reward_semantic/std": 0.1025692880153656, |
| "step": 100, |
| "step_time": 1.6173622274000992 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.6, |
| "completions/max_terminated_length": 5.6, |
| "completions/mean_length": 4.125, |
| "completions/mean_terminated_length": 4.125, |
| "completions/min_length": 3.15, |
| "completions/min_terminated_length": 3.15, |
| "entropy": 0.7014230243818019, |
| "epoch": 0.01968503937007874, |
| "frac_reward_zero_std": 0.3, |
| "grad_norm": 0.0, |
| "learning_rate": 4.9405e-06, |
| "loss": 0.07218956351280212, |
| "num_tokens": 35107.0, |
| "reward": 1.1007445603609085, |
| "reward_std": 0.028740251205454113, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.04814453125, |
| "rewards/reward_length/std": 0.010129839461296797, |
| "rewards/reward_semantic/mean": 0.34383928664028646, |
| "rewards/reward_semantic/std": 0.09484268203377724, |
| "step": 120, |
| "step_time": 1.5441937706006683 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 7.25, |
| "completions/max_terminated_length": 7.25, |
| "completions/mean_length": 4.9625, |
| "completions/mean_terminated_length": 4.9625, |
| "completions/min_length": 3.5, |
| "completions/min_terminated_length": 3.5, |
| "entropy": 0.7183661407629188, |
| "epoch": 0.022965879265091863, |
| "frac_reward_zero_std": 0.15, |
| "grad_norm": 2.609375, |
| "learning_rate": 4.9305000000000005e-06, |
| "loss": 0.05585476756095886, |
| "num_tokens": 40989.0, |
| "reward": 1.1004581451416016, |
| "reward_std": 0.03431669266719837, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.059765625, |
| "rewards/reward_length/std": 0.017409664555452765, |
| "rewards/reward_semantic/mean": 0.34482143111526964, |
| "rewards/reward_semantic/std": 0.11257617846131325, |
| "step": 140, |
| "step_time": 1.7615427238992196 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 7.05, |
| "completions/max_terminated_length": 7.05, |
| "completions/mean_length": 4.89375, |
| "completions/mean_terminated_length": 4.89375, |
| "completions/min_length": 3.5, |
| "completions/min_terminated_length": 3.5, |
| "entropy": 0.6915914016732131, |
| "epoch": 0.026246719160104987, |
| "frac_reward_zero_std": 0.1, |
| "grad_norm": 2.859375, |
| "learning_rate": 4.920500000000001e-06, |
| "loss": 0.0472392201423645, |
| "num_tokens": 47220.0, |
| "reward": 1.0850708901882171, |
| "reward_std": 0.06178898175858194, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.065576171875, |
| "rewards/reward_length/std": 0.01732458183541894, |
| "rewards/reward_semantic/mean": 0.2944990089163184, |
| "rewards/reward_semantic/std": 0.2051414344459772, |
| "step": 160, |
| "step_time": 1.7252537605003453 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.25, |
| "completions/max_terminated_length": 6.25, |
| "completions/mean_length": 4.55625, |
| "completions/mean_terminated_length": 4.55625, |
| "completions/min_length": 3.35, |
| "completions/min_terminated_length": 3.35, |
| "entropy": 0.8429647823562846, |
| "epoch": 0.02952755905511811, |
| "frac_reward_zero_std": 0.05, |
| "grad_norm": 3.078125, |
| "learning_rate": 4.910500000000001e-06, |
| "loss": 0.08106310367584228, |
| "num_tokens": 53053.0, |
| "reward": 1.0679614007472993, |
| "reward_std": 0.03796071938340902, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0566650390625, |
| "rewards/reward_length/std": 0.014682336029363797, |
| "rewards/reward_semantic/mean": 0.23598214369267226, |
| "rewards/reward_semantic/std": 0.12518299631774427, |
| "step": 180, |
| "step_time": 1.7508945095991293 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.2, |
| "completions/max_terminated_length": 6.2, |
| "completions/mean_length": 4.5375, |
| "completions/mean_terminated_length": 4.5375, |
| "completions/min_length": 3.75, |
| "completions/min_terminated_length": 3.75, |
| "entropy": 0.5915892604841246, |
| "epoch": 0.03280839895013123, |
| "frac_reward_zero_std": 0.35, |
| "grad_norm": 3.71875, |
| "learning_rate": 4.9005e-06, |
| "loss": 0.0888818621635437, |
| "num_tokens": 58931.0, |
| "reward": 1.120530515909195, |
| "reward_std": 0.0405463597693597, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0493896484375, |
| "rewards/reward_length/std": 0.008753356069792061, |
| "rewards/reward_semantic/mean": 0.4100000007078052, |
| "rewards/reward_semantic/std": 0.13480054624378682, |
| "step": 200, |
| "step_time": 1.713693901949955 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.45, |
| "completions/max_terminated_length": 6.45, |
| "completions/mean_length": 5.15, |
| "completions/mean_terminated_length": 5.15, |
| "completions/min_length": 4.2, |
| "completions/min_terminated_length": 4.2, |
| "entropy": 0.6395838386631112, |
| "epoch": 0.03608923884514436, |
| "frac_reward_zero_std": 0.35, |
| "grad_norm": 2.625, |
| "learning_rate": 4.8905e-06, |
| "loss": 0.06417204141616821, |
| "num_tokens": 65139.0, |
| "reward": 1.1249804645776749, |
| "reward_std": 0.027039075889479135, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.056640625, |
| "rewards/reward_length/std": 0.011077208817005158, |
| "rewards/reward_semantic/mean": 0.4260416656732559, |
| "rewards/reward_semantic/std": 0.08868714347481728, |
| "step": 220, |
| "step_time": 1.7278446534002796 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.4, |
| "completions/max_terminated_length": 5.4, |
| "completions/mean_length": 4.2625, |
| "completions/mean_terminated_length": 4.2625, |
| "completions/min_length": 3.0, |
| "completions/min_terminated_length": 3.0, |
| "entropy": 0.5652526923571713, |
| "epoch": 0.03937007874015748, |
| "frac_reward_zero_std": 0.35, |
| "grad_norm": 8.125, |
| "learning_rate": 4.8805e-06, |
| "loss": 0.07150443196296692, |
| "num_tokens": 71221.0, |
| "reward": 1.1466667413711549, |
| "reward_std": 0.03061878081643954, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.063330078125, |
| "rewards/reward_length/std": 0.015704452409408987, |
| "rewards/reward_semantic/mean": 0.4994440995156765, |
| "rewards/reward_semantic/std": 0.10041465926915408, |
| "step": 240, |
| "step_time": 1.6952891380005894 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.3, |
| "completions/max_terminated_length": 6.3, |
| "completions/mean_length": 4.46875, |
| "completions/mean_terminated_length": 4.46875, |
| "completions/min_length": 3.35, |
| "completions/min_terminated_length": 3.35, |
| "entropy": 0.672205779588694, |
| "epoch": 0.04265091863517061, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 6.4375, |
| "learning_rate": 4.8705e-06, |
| "loss": 0.09235309958457946, |
| "num_tokens": 77416.0, |
| "reward": 1.0690006405115127, |
| "reward_std": 0.019543477255501783, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0533203125, |
| "rewards/reward_length/std": 0.017377534485422074, |
| "rewards/reward_semantic/mean": 0.23888888889923693, |
| "rewards/reward_semantic/std": 0.0630335971713066, |
| "step": 260, |
| "step_time": 1.7545264535998286 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.3, |
| "completions/max_terminated_length": 5.3, |
| "completions/mean_length": 4.525, |
| "completions/mean_terminated_length": 4.525, |
| "completions/min_length": 3.65, |
| "completions/min_terminated_length": 3.65, |
| "entropy": 0.5580058551760885, |
| "epoch": 0.045931758530183726, |
| "frac_reward_zero_std": 0.45, |
| "grad_norm": 4.0625, |
| "learning_rate": 4.8605e-06, |
| "loss": 0.02879692018032074, |
| "num_tokens": 83332.0, |
| "reward": 1.1027960389852525, |
| "reward_std": 0.011529068512027152, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0558349609375, |
| "rewards/reward_length/std": 0.008870826044585556, |
| "rewards/reward_semantic/mean": 0.3519593259319663, |
| "rewards/reward_semantic/std": 0.03782847560942173, |
| "step": 280, |
| "step_time": 1.5900321766996968 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.35, |
| "completions/max_terminated_length": 6.35, |
| "completions/mean_length": 4.56875, |
| "completions/mean_terminated_length": 4.56875, |
| "completions/min_length": 3.2, |
| "completions/min_terminated_length": 3.2, |
| "entropy": 0.8288409212742408, |
| "epoch": 0.04921259842519685, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 0.0, |
| "learning_rate": 4.8505e-06, |
| "loss": 0.044028478860855105, |
| "num_tokens": 89567.0, |
| "reward": 1.1026858359575271, |
| "reward_std": 0.03983475278655533, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.062744140625, |
| "rewards/reward_length/std": 0.018717235408257694, |
| "rewards/reward_semantic/mean": 0.35274351127445697, |
| "rewards/reward_semantic/std": 0.13158041909337043, |
| "step": 300, |
| "step_time": 1.7004065297001034 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.65, |
| "completions/max_terminated_length": 5.65, |
| "completions/mean_length": 4.5875, |
| "completions/mean_terminated_length": 4.5875, |
| "completions/min_length": 3.5, |
| "completions/min_terminated_length": 3.5, |
| "entropy": 0.7240128130373705, |
| "epoch": 0.05249343832020997, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 5.0625, |
| "learning_rate": 4.8405000000000005e-06, |
| "loss": 0.03284294605255127, |
| "num_tokens": 95157.0, |
| "reward": 1.0648824661970138, |
| "reward_std": 0.014824012622557347, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.055029296875, |
| "rewards/reward_length/std": 0.010819258564151824, |
| "rewards/reward_semantic/mean": 0.2254464277997613, |
| "rewards/reward_semantic/std": 0.04843873754143715, |
| "step": 320, |
| "step_time": 1.6396856073492927 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.25, |
| "completions/max_terminated_length": 6.25, |
| "completions/mean_length": 4.625, |
| "completions/mean_terminated_length": 4.625, |
| "completions/min_length": 3.35, |
| "completions/min_terminated_length": 3.35, |
| "entropy": 0.8246265778172528, |
| "epoch": 0.0557742782152231, |
| "frac_reward_zero_std": 0.35, |
| "grad_norm": 0.0, |
| "learning_rate": 4.830500000000001e-06, |
| "loss": 0.05039528012275696, |
| "num_tokens": 100857.0, |
| "reward": 1.0999272018671036, |
| "reward_std": 0.012020479088823777, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0510986328125, |
| "rewards/reward_length/std": 0.013131194305606187, |
| "rewards/reward_semantic/mean": 0.3416071431711316, |
| "rewards/reward_semantic/std": 0.03859963417053223, |
| "step": 340, |
| "step_time": 1.6233859046002181 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.75, |
| "completions/max_terminated_length": 5.75, |
| "completions/mean_length": 4.49375, |
| "completions/mean_terminated_length": 4.49375, |
| "completions/min_length": 3.7, |
| "completions/min_terminated_length": 3.7, |
| "entropy": 0.49938802123069764, |
| "epoch": 0.05905511811023622, |
| "frac_reward_zero_std": 0.3, |
| "grad_norm": 0.0, |
| "learning_rate": 4.8205e-06, |
| "loss": 0.01368042677640915, |
| "num_tokens": 106664.0, |
| "reward": 1.1010554283857346, |
| "reward_std": 0.045361624303768625, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0576416015625, |
| "rewards/reward_length/std": 0.010658931022044271, |
| "rewards/reward_semantic/mean": 0.3464583337306976, |
| "rewards/reward_semantic/std": 0.15053959861397742, |
| "step": 360, |
| "step_time": 1.6211198363504082 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.8, |
| "completions/max_terminated_length": 5.8, |
| "completions/mean_length": 4.74375, |
| "completions/mean_terminated_length": 4.74375, |
| "completions/min_length": 3.6, |
| "completions/min_terminated_length": 3.6, |
| "entropy": 0.6371999262647023, |
| "epoch": 0.062335958005249346, |
| "frac_reward_zero_std": 0.3, |
| "grad_norm": 9.1875, |
| "learning_rate": 4.8105e-06, |
| "loss": 0.03233610987663269, |
| "num_tokens": 112511.0, |
| "reward": 1.1021533191204071, |
| "reward_std": 0.02930783515548683, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.05693359375, |
| "rewards/reward_length/std": 0.012558965157950297, |
| "rewards/reward_semantic/mean": 0.3500000026077032, |
| "rewards/reward_semantic/std": 0.09658735543489456, |
| "step": 380, |
| "step_time": 1.6564469950499188 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.05, |
| "completions/max_terminated_length": 6.05, |
| "completions/mean_length": 4.33125, |
| "completions/mean_terminated_length": 4.33125, |
| "completions/min_length": 3.35, |
| "completions/min_terminated_length": 3.35, |
| "entropy": 0.7747498400121913, |
| "epoch": 0.06561679790026247, |
| "frac_reward_zero_std": 0.15, |
| "grad_norm": 16.0, |
| "learning_rate": 4.8005e-06, |
| "loss": 0.02885372340679169, |
| "num_tokens": 118188.0, |
| "reward": 1.1014230579137803, |
| "reward_std": 0.054096709901932626, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.054931640625, |
| "rewards/reward_length/std": 0.01498162968782708, |
| "rewards/reward_semantic/mean": 0.3472321443259716, |
| "rewards/reward_semantic/std": 0.17982315719127656, |
| "step": 400, |
| "step_time": 1.6521750739995695 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.55, |
| "completions/max_terminated_length": 5.55, |
| "completions/mean_length": 4.25, |
| "completions/mean_terminated_length": 4.25, |
| "completions/min_length": 3.25, |
| "completions/min_terminated_length": 3.25, |
| "entropy": 0.6093234239519916, |
| "epoch": 0.0688976377952756, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 8.1875, |
| "learning_rate": 4.7905e-06, |
| "loss": 0.07366046905517579, |
| "num_tokens": 123980.0, |
| "reward": 1.088520696759224, |
| "reward_std": 0.02127447940329148, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0520263671875, |
| "rewards/reward_length/std": 0.010646102635655552, |
| "rewards/reward_semantic/mean": 0.30374008174985645, |
| "rewards/reward_semantic/std": 0.06995791383087635, |
| "step": 420, |
| "step_time": 1.598766735350273 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.6, |
| "completions/max_terminated_length": 5.6, |
| "completions/mean_length": 4.4625, |
| "completions/mean_terminated_length": 4.4625, |
| "completions/min_length": 3.4, |
| "completions/min_terminated_length": 3.4, |
| "entropy": 0.6000309210250634, |
| "epoch": 0.07217847769028872, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 6.25, |
| "learning_rate": 4.7805e-06, |
| "loss": 0.02538236975669861, |
| "num_tokens": 130310.0, |
| "reward": 1.1284031838178634, |
| "reward_std": 0.030537298787385224, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0548828125, |
| "rewards/reward_length/std": 0.012618192343506963, |
| "rewards/reward_semantic/mean": 0.43715773820877074, |
| "rewards/reward_semantic/std": 0.10091984458267689, |
| "step": 440, |
| "step_time": 1.6581264014497719 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.75, |
| "completions/max_terminated_length": 6.75, |
| "completions/mean_length": 4.9, |
| "completions/mean_terminated_length": 4.9, |
| "completions/min_length": 3.45, |
| "completions/min_terminated_length": 3.45, |
| "entropy": 0.7827204793691636, |
| "epoch": 0.07545931758530183, |
| "frac_reward_zero_std": 0.15, |
| "grad_norm": 2.140625, |
| "learning_rate": 4.7705e-06, |
| "loss": 0.09432601332664489, |
| "num_tokens": 136206.0, |
| "reward": 1.061219373345375, |
| "reward_std": 0.04166477765902528, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.067041015625, |
| "rewards/reward_length/std": 0.016452802077401428, |
| "rewards/reward_semantic/mean": 0.21523809637874364, |
| "rewards/reward_semantic/std": 0.13769038021564484, |
| "step": 460, |
| "step_time": 1.763348830850282 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.95, |
| "completions/max_terminated_length": 5.95, |
| "completions/mean_length": 4.5125, |
| "completions/mean_terminated_length": 4.5125, |
| "completions/min_length": 3.3, |
| "completions/min_terminated_length": 3.3, |
| "entropy": 0.6544731822835275, |
| "epoch": 0.07874015748031496, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.0, |
| "learning_rate": 4.7605e-06, |
| "loss": 0.095479816198349, |
| "num_tokens": 142168.0, |
| "reward": 1.110293486714363, |
| "reward_std": 0.03730325770447962, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0526123046875, |
| "rewards/reward_length/std": 0.013677800679579378, |
| "rewards/reward_semantic/mean": 0.37641369104385375, |
| "rewards/reward_semantic/std": 0.12245006412267685, |
| "step": 480, |
| "step_time": 1.693691986000158 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.2, |
| "completions/max_terminated_length": 6.2, |
| "completions/mean_length": 4.5375, |
| "completions/mean_terminated_length": 4.5375, |
| "completions/min_length": 3.3, |
| "completions/min_terminated_length": 3.3, |
| "entropy": 0.7134705329648569, |
| "epoch": 0.08202099737532809, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 0.0, |
| "learning_rate": 4.7505000000000005e-06, |
| "loss": 0.04971726834774017, |
| "num_tokens": 148190.0, |
| "reward": 1.059982505440712, |
| "reward_std": 0.029997232874302428, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.053564453125, |
| "rewards/reward_length/std": 0.012155310017988086, |
| "rewards/reward_semantic/mean": 0.20886905267834663, |
| "rewards/reward_semantic/std": 0.0987570121884346, |
| "step": 500, |
| "step_time": 1.7915035136496953 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 4.75, |
| "completions/max_terminated_length": 4.75, |
| "completions/mean_length": 3.85625, |
| "completions/mean_terminated_length": 3.85625, |
| "completions/min_length": 3.15, |
| "completions/min_terminated_length": 3.15, |
| "entropy": 0.489788422039328, |
| "epoch": 0.08530183727034121, |
| "frac_reward_zero_std": 0.35, |
| "grad_norm": 0.0, |
| "learning_rate": 4.740500000000001e-06, |
| "loss": 0.015317876636981965, |
| "num_tokens": 153895.0, |
| "reward": 1.14158952832222, |
| "reward_std": 0.03168491964024724, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0512451171875, |
| "rewards/reward_length/std": 0.007844625366851687, |
| "rewards/reward_semantic/mean": 0.48050595819950104, |
| "rewards/reward_semantic/std": 0.10510801821947098, |
| "step": 520, |
| "step_time": 1.6148291991500627 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.15, |
| "completions/max_terminated_length": 6.15, |
| "completions/mean_length": 4.54375, |
| "completions/mean_terminated_length": 4.54375, |
| "completions/min_length": 3.6, |
| "completions/min_terminated_length": 3.6, |
| "entropy": 0.7991890870037424, |
| "epoch": 0.08858267716535433, |
| "frac_reward_zero_std": 0.3, |
| "grad_norm": 2.546875, |
| "learning_rate": 4.730500000000001e-06, |
| "loss": 0.05347582697868347, |
| "num_tokens": 159790.0, |
| "reward": 1.0938216686248778, |
| "reward_std": 0.01884553566815157, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0449951171875, |
| "rewards/reward_length/std": 0.007818706298712641, |
| "rewards/reward_semantic/mean": 0.3202380955219269, |
| "rewards/reward_semantic/std": 0.061923586577177045, |
| "step": 540, |
| "step_time": 1.8538054024499615 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.2, |
| "completions/max_terminated_length": 5.2, |
| "completions/mean_length": 4.2375, |
| "completions/mean_terminated_length": 4.2375, |
| "completions/min_length": 3.45, |
| "completions/min_terminated_length": 3.45, |
| "entropy": 0.5264639136769802, |
| "epoch": 0.09186351706036745, |
| "frac_reward_zero_std": 0.3, |
| "grad_norm": 4.40625, |
| "learning_rate": 4.720500000000001e-06, |
| "loss": 0.022994591295719145, |
| "num_tokens": 165812.0, |
| "reward": 1.130643653869629, |
| "reward_std": 0.035184593574376775, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0594482421875, |
| "rewards/reward_length/std": 0.012201186269521713, |
| "rewards/reward_semantic/mean": 0.44538690596818925, |
| "rewards/reward_semantic/std": 0.11669721417129039, |
| "step": 560, |
| "step_time": 1.5765380310496766 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.9, |
| "completions/max_terminated_length": 5.9, |
| "completions/mean_length": 4.53125, |
| "completions/mean_terminated_length": 4.53125, |
| "completions/min_length": 3.6, |
| "completions/min_terminated_length": 3.6, |
| "entropy": 0.6579255037364419, |
| "epoch": 0.09514435695538058, |
| "frac_reward_zero_std": 0.35, |
| "grad_norm": 4.625, |
| "learning_rate": 4.710500000000001e-06, |
| "loss": 0.03634168207645416, |
| "num_tokens": 171769.0, |
| "reward": 1.1139779269695282, |
| "reward_std": 0.028173982925363817, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0481201171875, |
| "rewards/reward_length/std": 0.011657711572479457, |
| "rewards/reward_semantic/mean": 0.38794642984867095, |
| "rewards/reward_semantic/std": 0.09298380687832833, |
| "step": 580, |
| "step_time": 1.6179808518501886 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.9, |
| "completions/max_terminated_length": 5.9, |
| "completions/mean_length": 4.25, |
| "completions/mean_terminated_length": 4.25, |
| "completions/min_length": 3.2, |
| "completions/min_terminated_length": 3.2, |
| "entropy": 0.7248269693967814, |
| "epoch": 0.0984251968503937, |
| "frac_reward_zero_std": 0.2, |
| "grad_norm": 10.3125, |
| "learning_rate": 4.700500000000001e-06, |
| "loss": 0.09872611761093139, |
| "num_tokens": 177649.0, |
| "reward": 1.0644860982894897, |
| "reward_std": 0.026886727681267077, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0540283203125, |
| "rewards/reward_length/std": 0.014043591287918388, |
| "rewards/reward_semantic/mean": 0.22395833432674409, |
| "rewards/reward_semantic/std": 0.0882976733148098, |
| "step": 600, |
| "step_time": 1.6426984294495923 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.4, |
| "completions/max_terminated_length": 5.4, |
| "completions/mean_length": 4.5625, |
| "completions/mean_terminated_length": 4.5625, |
| "completions/min_length": 3.55, |
| "completions/min_terminated_length": 3.55, |
| "entropy": 0.4465312559736049, |
| "epoch": 0.10170603674540682, |
| "frac_reward_zero_std": 0.4, |
| "grad_norm": 0.0, |
| "learning_rate": 4.6905e-06, |
| "loss": -0.010341504216194152, |
| "num_tokens": 183835.0, |
| "reward": 1.1265694797039032, |
| "reward_std": 0.028417593393533024, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0556640625, |
| "rewards/reward_length/std": 0.011586534441448748, |
| "rewards/reward_semantic/mean": 0.43117559552192686, |
| "rewards/reward_semantic/std": 0.0946267582476139, |
| "step": 620, |
| "step_time": 1.6642883875004373 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.8, |
| "completions/max_terminated_length": 6.8, |
| "completions/mean_length": 5.25, |
| "completions/mean_terminated_length": 5.25, |
| "completions/min_length": 3.8, |
| "completions/min_terminated_length": 3.8, |
| "entropy": 0.4096738053149238, |
| "epoch": 0.10498687664041995, |
| "frac_reward_zero_std": 0.4, |
| "grad_norm": 0.0, |
| "learning_rate": 4.6805e-06, |
| "loss": 0.05256671905517578, |
| "num_tokens": 190283.0, |
| "reward": 1.1175170987844467, |
| "reward_std": 0.015989218310278375, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.049658203125, |
| "rewards/reward_length/std": 0.008098985790275038, |
| "rewards/reward_semantic/mean": 0.4000000020489097, |
| "rewards/reward_semantic/std": 0.05260697938501835, |
| "step": 640, |
| "step_time": 1.7202262992497708 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.0, |
| "completions/max_terminated_length": 5.0, |
| "completions/mean_length": 4.40625, |
| "completions/mean_terminated_length": 4.40625, |
| "completions/min_length": 3.75, |
| "completions/min_terminated_length": 3.75, |
| "entropy": 0.44533549200182276, |
| "epoch": 0.10826771653543307, |
| "frac_reward_zero_std": 0.45, |
| "grad_norm": 28.0, |
| "learning_rate": 4.6705000000000004e-06, |
| "loss": 0.000417201966047287, |
| "num_tokens": 196284.0, |
| "reward": 1.1231877475976944, |
| "reward_std": 0.02121658924515941, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0481201171875, |
| "rewards/reward_length/std": 0.007935813744552433, |
| "rewards/reward_semantic/mean": 0.4186458334326744, |
| "rewards/reward_semantic/std": 0.06994377076625824, |
| "step": 660, |
| "step_time": 1.5930430018996957 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.95, |
| "completions/max_terminated_length": 5.95, |
| "completions/mean_length": 5.04375, |
| "completions/mean_terminated_length": 5.04375, |
| "completions/min_length": 3.95, |
| "completions/min_terminated_length": 3.95, |
| "entropy": 0.4285797670326247, |
| "epoch": 0.1115485564304462, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 8.1875, |
| "learning_rate": 4.6605000000000005e-06, |
| "loss": -0.04365907907485962, |
| "num_tokens": 202619.0, |
| "reward": 1.0846716284751892, |
| "reward_std": 0.020432764168072028, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0378173828125, |
| "rewards/reward_length/std": 0.01130076235276647, |
| "rewards/reward_semantic/mean": 0.28854166716337204, |
| "rewards/reward_semantic/std": 0.06747942492365837, |
| "step": 680, |
| "step_time": 1.6218179324499942 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.15, |
| "completions/max_terminated_length": 5.15, |
| "completions/mean_length": 4.6375, |
| "completions/mean_terminated_length": 4.6375, |
| "completions/min_length": 4.3, |
| "completions/min_terminated_length": 4.3, |
| "entropy": 0.15305614546541618, |
| "epoch": 0.11482939632545931, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 6.15625, |
| "learning_rate": 4.650500000000001e-06, |
| "loss": -0.022796785831451415, |
| "num_tokens": 208433.0, |
| "reward": 1.115874022245407, |
| "reward_std": 0.013541911324136891, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.03251953125, |
| "rewards/reward_length/std": 0.0023872115067206322, |
| "rewards/reward_semantic/mean": 0.391666666790843, |
| "rewards/reward_semantic/std": 0.045165380090475084, |
| "step": 700, |
| "step_time": 1.5669768184995518 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.6, |
| "completions/max_terminated_length": 5.6, |
| "completions/mean_length": 4.7125, |
| "completions/mean_terminated_length": 4.7125, |
| "completions/min_length": 3.95, |
| "completions/min_terminated_length": 3.95, |
| "entropy": 0.3805655787051364, |
| "epoch": 0.11811023622047244, |
| "frac_reward_zero_std": 0.4, |
| "grad_norm": 0.0, |
| "learning_rate": 4.640500000000001e-06, |
| "loss": 0.004567959904670715, |
| "num_tokens": 214115.0, |
| "reward": 1.0923460215330123, |
| "reward_std": 0.030167541204718873, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0387939453125, |
| "rewards/reward_length/std": 0.00844412127044052, |
| "rewards/reward_semantic/mean": 0.3142857149243355, |
| "rewards/reward_semantic/std": 0.10010053664445877, |
| "step": 720, |
| "step_time": 1.6804628341502394 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.4, |
| "completions/max_terminated_length": 5.4, |
| "completions/mean_length": 4.93125, |
| "completions/mean_terminated_length": 4.93125, |
| "completions/min_length": 4.5, |
| "completions/min_terminated_length": 4.5, |
| "entropy": 0.25949509698621115, |
| "epoch": 0.12139107611548557, |
| "frac_reward_zero_std": 0.55, |
| "grad_norm": 21.125, |
| "learning_rate": 4.630500000000001e-06, |
| "loss": 0.022067296504974365, |
| "num_tokens": 220184.0, |
| "reward": 1.1069529592990874, |
| "reward_std": 0.022339693034882657, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0404052734375, |
| "rewards/reward_length/std": 0.006306704942835495, |
| "rewards/reward_semantic/mean": 0.3632440477609634, |
| "rewards/reward_semantic/std": 0.07467387840151787, |
| "step": 740, |
| "step_time": 1.6020566637997036 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.75, |
| "completions/max_terminated_length": 5.75, |
| "completions/mean_length": 5.2375, |
| "completions/mean_terminated_length": 5.2375, |
| "completions/min_length": 4.85, |
| "completions/min_terminated_length": 4.85, |
| "entropy": 0.13271254739520372, |
| "epoch": 0.12467191601049869, |
| "frac_reward_zero_std": 0.65, |
| "grad_norm": 0.0, |
| "learning_rate": 4.620500000000001e-06, |
| "loss": -0.011591804772615432, |
| "num_tokens": 225926.0, |
| "reward": 1.1052490204572678, |
| "reward_std": 0.012424617203214439, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.03251953125, |
| "rewards/reward_length/std": 0.0069050868391059336, |
| "rewards/reward_semantic/mean": 0.35625, |
| "rewards/reward_semantic/std": 0.04110970795154571, |
| "step": 760, |
| "step_time": 1.5845169639495906 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.85, |
| "completions/max_terminated_length": 5.85, |
| "completions/mean_length": 5.29375, |
| "completions/mean_terminated_length": 5.29375, |
| "completions/min_length": 4.7, |
| "completions/min_terminated_length": 4.7, |
| "entropy": 0.2635622876246998, |
| "epoch": 0.1279527559055118, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.6105e-06, |
| "loss": -0.012393039464950562, |
| "num_tokens": 231541.0, |
| "reward": 1.0538557916879654, |
| "reward_std": 0.024872867923113516, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.03642578125, |
| "rewards/reward_length/std": 0.006841688393615186, |
| "rewards/reward_semantic/mean": 0.1855902776122093, |
| "rewards/reward_semantic/std": 0.08292527124285698, |
| "step": 780, |
| "step_time": 1.5691503484998066 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.65, |
| "completions/max_terminated_length": 5.65, |
| "completions/mean_length": 5.26875, |
| "completions/mean_terminated_length": 5.26875, |
| "completions/min_length": 4.95, |
| "completions/min_terminated_length": 4.95, |
| "entropy": 0.1539801352791983, |
| "epoch": 0.13123359580052493, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 4.6005e-06, |
| "loss": 0.02297772765159607, |
| "num_tokens": 237712.0, |
| "reward": 1.0723930418491363, |
| "reward_std": 0.006485555377730634, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.033984375, |
| "rewards/reward_length/std": 0.004696746869012713, |
| "rewards/reward_semantic/mean": 0.2469742067158222, |
| "rewards/reward_semantic/std": 0.021238823607563972, |
| "step": 800, |
| "step_time": 1.5607441842005756 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.85, |
| "completions/max_terminated_length": 5.85, |
| "completions/mean_length": 5.025, |
| "completions/mean_terminated_length": 5.025, |
| "completions/min_length": 4.5, |
| "completions/min_terminated_length": 4.5, |
| "entropy": 0.22530270127426774, |
| "epoch": 0.13451443569553806, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.0, |
| "learning_rate": 4.5905e-06, |
| "loss": -0.00012428686022758484, |
| "num_tokens": 243900.0, |
| "reward": 1.0480462223291398, |
| "reward_std": 0.014764534048299537, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0357421875, |
| "rewards/reward_length/std": 0.010764729115180671, |
| "rewards/reward_semantic/mean": 0.16611111108213664, |
| "rewards/reward_semantic/std": 0.049135804176330566, |
| "step": 820, |
| "step_time": 1.6110364599507192 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.2, |
| "completions/max_terminated_length": 6.2, |
| "completions/mean_length": 5.25625, |
| "completions/mean_terminated_length": 5.25625, |
| "completions/min_length": 4.8, |
| "completions/min_terminated_length": 4.8, |
| "entropy": 0.31118827113505176, |
| "epoch": 0.1377952755905512, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.0, |
| "learning_rate": 4.5805000000000004e-06, |
| "loss": 0.01610923409461975, |
| "num_tokens": 250069.0, |
| "reward": 1.0784188359975815, |
| "reward_std": 0.02915856411345885, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0343017578125, |
| "rewards/reward_length/std": 0.008795349719002843, |
| "rewards/reward_semantic/mean": 0.26711309850215914, |
| "rewards/reward_semantic/std": 0.09740542992949486, |
| "step": 840, |
| "step_time": 1.6490229548499884 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.25, |
| "completions/max_terminated_length": 6.25, |
| "completions/mean_length": 5.48125, |
| "completions/mean_terminated_length": 5.48125, |
| "completions/min_length": 4.8, |
| "completions/min_terminated_length": 4.8, |
| "entropy": 0.19009512941393042, |
| "epoch": 0.1410761154855643, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.5705000000000005e-06, |
| "loss": -0.048175308108329776, |
| "num_tokens": 256186.0, |
| "reward": 1.1095454066991806, |
| "reward_std": 0.03285904193180613, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0268798828125, |
| "rewards/reward_length/std": 0.009483330580405892, |
| "rewards/reward_semantic/mean": 0.36963141039013864, |
| "rewards/reward_semantic/std": 0.11041291058063507, |
| "step": 860, |
| "step_time": 1.84969353199931 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.7, |
| "completions/max_terminated_length": 5.7, |
| "completions/mean_length": 4.79375, |
| "completions/mean_terminated_length": 4.79375, |
| "completions/min_length": 4.4, |
| "completions/min_terminated_length": 4.4, |
| "entropy": 0.21118192300355076, |
| "epoch": 0.14435695538057744, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.560500000000001e-06, |
| "loss": 0.050315862894058226, |
| "num_tokens": 262425.0, |
| "reward": 1.0997518181800843, |
| "reward_std": 0.022761242519482038, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.031689453125, |
| "rewards/reward_length/std": 0.00647321484866552, |
| "rewards/reward_semantic/mean": 0.3377877000719309, |
| "rewards/reward_semantic/std": 0.07582005113363266, |
| "step": 880, |
| "step_time": 1.6589296972999363 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.8, |
| "completions/max_terminated_length": 5.8, |
| "completions/mean_length": 4.91875, |
| "completions/mean_terminated_length": 4.91875, |
| "completions/min_length": 4.15, |
| "completions/min_terminated_length": 4.15, |
| "entropy": 0.3470641552201414, |
| "epoch": 0.14763779527559054, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.0, |
| "learning_rate": 4.550500000000001e-06, |
| "loss": 0.012325967103242874, |
| "num_tokens": 268468.0, |
| "reward": 1.0776826739311218, |
| "reward_std": 0.018389373533136678, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0361083984375, |
| "rewards/reward_length/std": 0.010244967485778033, |
| "rewards/reward_semantic/mean": 0.26496031805872916, |
| "rewards/reward_semantic/std": 0.060299596190452574, |
| "step": 900, |
| "step_time": 1.6155126945997835 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.0, |
| "completions/max_terminated_length": 6.0, |
| "completions/mean_length": 5.2, |
| "completions/mean_terminated_length": 5.2, |
| "completions/min_length": 4.45, |
| "completions/min_terminated_length": 4.45, |
| "entropy": 0.2223609389172907, |
| "epoch": 0.15091863517060367, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 29.5, |
| "learning_rate": 4.540500000000001e-06, |
| "loss": 0.008905045688152313, |
| "num_tokens": 274548.0, |
| "reward": 1.0319956064224243, |
| "reward_std": 0.01447186817167676, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.027587890625, |
| "rewards/reward_length/std": 0.004147578129777685, |
| "rewards/reward_semantic/mean": 0.1112500011920929, |
| "rewards/reward_semantic/std": 0.04789424017071724, |
| "step": 920, |
| "step_time": 1.6041553608993127 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.25, |
| "completions/max_terminated_length": 6.25, |
| "completions/mean_length": 5.5, |
| "completions/mean_terminated_length": 5.5, |
| "completions/min_length": 4.8, |
| "completions/min_terminated_length": 4.8, |
| "entropy": 0.15990385009681632, |
| "epoch": 0.1541994750656168, |
| "frac_reward_zero_std": 0.65, |
| "grad_norm": 25.75, |
| "learning_rate": 4.5305e-06, |
| "loss": 0.0012000683695077897, |
| "num_tokens": 280548.0, |
| "reward": 1.0634800434112548, |
| "reward_std": 0.01882710763893556, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02861328125, |
| "rewards/reward_length/std": 0.0077044784789904956, |
| "rewards/reward_semantic/mean": 0.21636904776096344, |
| "rewards/reward_semantic/std": 0.06269824169576169, |
| "step": 940, |
| "step_time": 1.591113177501029 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 6.0, |
| "completions/max_terminated_length": 6.0, |
| "completions/mean_length": 5.48125, |
| "completions/mean_terminated_length": 5.48125, |
| "completions/min_length": 5.0, |
| "completions/min_terminated_length": 5.0, |
| "entropy": 0.1008574430443332, |
| "epoch": 0.15748031496062992, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.5205e-06, |
| "loss": 0.0621263861656189, |
| "num_tokens": 286377.0, |
| "reward": 1.1116095036268234, |
| "reward_std": 0.022756787104299292, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02861328125, |
| "rewards/reward_length/std": 0.004306197509868071, |
| "rewards/reward_semantic/mean": 0.3768005946651101, |
| "rewards/reward_semantic/std": 0.07597103714942932, |
| "step": 960, |
| "step_time": 1.6255918225498136 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.7, |
| "completions/max_terminated_length": 5.7, |
| "completions/mean_length": 5.4625, |
| "completions/mean_terminated_length": 5.4625, |
| "completions/min_length": 4.9, |
| "completions/min_terminated_length": 4.9, |
| "entropy": 0.12628535750282027, |
| "epoch": 0.16076115485564305, |
| "frac_reward_zero_std": 0.65, |
| "grad_norm": 0.0, |
| "learning_rate": 4.5105e-06, |
| "loss": 0.0023237094283103943, |
| "num_tokens": 292251.0, |
| "reward": 1.0658237963914872, |
| "reward_std": 0.012343478866387158, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.03173828125, |
| "rewards/reward_length/std": 0.007525579165667295, |
| "rewards/reward_semantic/mean": 0.2247023830190301, |
| "rewards/reward_semantic/std": 0.040503402799367906, |
| "step": 980, |
| "step_time": 1.583995696999318 |
| }, |
| { |
| "clip_ratio/high_max": 0.0, |
| "clip_ratio/high_mean": 0.0, |
| "clip_ratio/low_mean": 0.0, |
| "clip_ratio/low_min": 0.0, |
| "clip_ratio/region_mean": 0.0, |
| "completions/clipped_ratio": 0.0, |
| "completions/max_length": 5.55, |
| "completions/max_terminated_length": 5.55, |
| "completions/mean_length": 5.24375, |
| "completions/mean_terminated_length": 5.24375, |
| "completions/min_length": 4.85, |
| "completions/min_terminated_length": 4.85, |
| "entropy": 0.09762425903263647, |
| "epoch": 0.16404199475065617, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 17.375, |
| "learning_rate": 4.5005e-06, |
| "loss": -0.0012902652844786644, |
| "num_tokens": 297890.0, |
| "reward": 1.0763329923152924, |
| "reward_std": 0.013358409999636934, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02958984375, |
| "rewards/reward_length/std": 0.006560613529291004, |
| "rewards/reward_semantic/mean": 0.259375, |
| "rewards/reward_semantic/std": 0.04419417306780815, |
| "step": 1000, |
| "step_time": 1.56278663364983 |
| } |
| ], |
| "logging_steps": 20, |
| "max_steps": 10000, |
| "num_input_tokens_seen": 297890, |
| "num_train_epochs": 2, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
