| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.4921259842519685, |
| "eval_steps": 500, |
| "global_step": 3000, |
| "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 |
| }, |
| { |
| "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": 5.29375, |
| "completions/mean_terminated_length": 5.29375, |
| "completions/min_length": 5.05, |
| "completions/min_terminated_length": 5.05, |
| "entropy": 0.07264433649506827, |
| "epoch": 0.1673228346456693, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 4.4905000000000005e-06, |
| "loss": 0.016954894363880157, |
| "num_tokens": 303881.0, |
| "reward": 1.0861677438020707, |
| "reward_std": 0.009647212684649275, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0260498046875, |
| "rewards/reward_length/std": 0.003177844942547381, |
| "rewards/reward_semantic/mean": 0.2915674611926079, |
| "rewards/reward_semantic/std": 0.032339051365852356, |
| "step": 1020, |
| "step_time": 1.61673008425023 |
| }, |
| { |
| "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": 5.41875, |
| "completions/mean_terminated_length": 5.41875, |
| "completions/min_length": 4.75, |
| "completions/min_terminated_length": 4.75, |
| "entropy": 0.09382649142080482, |
| "epoch": 0.17060367454068243, |
| "frac_reward_zero_std": 0.65, |
| "grad_norm": 0.0, |
| "learning_rate": 4.4805000000000006e-06, |
| "loss": -0.023067578673362732, |
| "num_tokens": 309956.0, |
| "reward": 1.0835506588220596, |
| "reward_std": 0.012976862565119518, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0254150390625, |
| "rewards/reward_length/std": 0.006931392394471913, |
| "rewards/reward_semantic/mean": 0.28273809552192686, |
| "rewards/reward_semantic/std": 0.04399138689041138, |
| "step": 1040, |
| "step_time": 1.6394621164004093 |
| }, |
| { |
| "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": 5.825, |
| "completions/mean_terminated_length": 5.825, |
| "completions/min_length": 5.35, |
| "completions/min_terminated_length": 5.35, |
| "entropy": 0.061149569329199946, |
| "epoch": 0.17388451443569553, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 22.25, |
| "learning_rate": 4.470500000000001e-06, |
| "loss": 0.04459778964519501, |
| "num_tokens": 316344.0, |
| "reward": 1.0397380709648132, |
| "reward_std": 0.011474297757013118, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.023095703125, |
| "rewards/reward_length/std": 0.002799238113220781, |
| "rewards/reward_semantic/mean": 0.1363095223903656, |
| "rewards/reward_semantic/std": 0.038431893289089206, |
| "step": 1060, |
| "step_time": 1.6685051138990274 |
| }, |
| { |
| "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": 5.78125, |
| "completions/mean_terminated_length": 5.78125, |
| "completions/min_length": 5.55, |
| "completions/min_terminated_length": 5.55, |
| "entropy": 0.08575974361156113, |
| "epoch": 0.17716535433070865, |
| "frac_reward_zero_std": 0.85, |
| "grad_norm": 0.0, |
| "learning_rate": 4.460500000000001e-06, |
| "loss": 0.012008878588676452, |
| "num_tokens": 322629.0, |
| "reward": 1.0529980450868606, |
| "reward_std": 0.0035252992587629707, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0275390625, |
| "rewards/reward_length/std": 0.0012502899160608649, |
| "rewards/reward_semantic/mean": 0.18125, |
| "rewards/reward_semantic/std": 0.011572751402854919, |
| "step": 1080, |
| "step_time": 1.6754981747496458 |
| }, |
| { |
| "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": 5.175, |
| "completions/mean_terminated_length": 5.175, |
| "completions/min_length": 4.85, |
| "completions/min_terminated_length": 4.85, |
| "entropy": 0.1447469376193112, |
| "epoch": 0.18044619422572178, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.4505e-06, |
| "loss": 0.02233708053827286, |
| "num_tokens": 328841.0, |
| "reward": 1.052584233880043, |
| "reward_std": 0.019267462584684834, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0233154296875, |
| "rewards/reward_length/std": 0.004912395629798993, |
| "rewards/reward_semantic/mean": 0.17916666772216558, |
| "rewards/reward_semantic/std": 0.06471266187727451, |
| "step": 1100, |
| "step_time": 1.595077488649622 |
| }, |
| { |
| "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": 5.24375, |
| "completions/mean_terminated_length": 5.24375, |
| "completions/min_length": 4.95, |
| "completions/min_terminated_length": 4.95, |
| "entropy": 0.14285493081551975, |
| "epoch": 0.1837270341207349, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 4.4405e-06, |
| "loss": -0.039118018746376035, |
| "num_tokens": 334608.0, |
| "reward": 1.0435290575027465, |
| "reward_std": 0.00021757901486125776, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0294189453125, |
| "rewards/reward_length/std": 0.004351581819355488, |
| "rewards/reward_semantic/mean": 0.15, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 1120, |
| "step_time": 1.6498132331008493 |
| }, |
| { |
| "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": 5.34375, |
| "completions/mean_terminated_length": 5.34375, |
| "completions/min_length": 5.1, |
| "completions/min_terminated_length": 5.1, |
| "entropy": 0.1312446712197925, |
| "epoch": 0.18700787401574803, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 4.4305e-06, |
| "loss": 0.001287885382771492, |
| "num_tokens": 340591.0, |
| "reward": 1.0588012665510178, |
| "reward_std": 0.00015805717703187838, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.023974609375, |
| "rewards/reward_length/std": 0.003161165042547509, |
| "rewards/reward_semantic/mean": 0.2, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 1140, |
| "step_time": 1.5476819176994467 |
| }, |
| { |
| "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": 5.39375, |
| "completions/mean_terminated_length": 5.39375, |
| "completions/min_length": 4.75, |
| "completions/min_terminated_length": 4.75, |
| "entropy": 0.08150004155613715, |
| "epoch": 0.19028871391076116, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 15.8125, |
| "learning_rate": 4.4205e-06, |
| "loss": 0.04978489875793457, |
| "num_tokens": 346414.0, |
| "reward": 1.0645825117826462, |
| "reward_std": 0.023181272562942466, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.020849609375, |
| "rewards/reward_length/std": 0.0020763361942954363, |
| "rewards/reward_semantic/mean": 0.21875, |
| "rewards/reward_semantic/std": 0.07733173668384552, |
| "step": 1160, |
| "step_time": 1.5483456218503306 |
| }, |
| { |
| "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": 4.975, |
| "completions/mean_terminated_length": 4.975, |
| "completions/min_length": 4.55, |
| "completions/min_terminated_length": 4.55, |
| "entropy": 0.13536443736702494, |
| "epoch": 0.19356955380577429, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 11.5, |
| "learning_rate": 4.4105e-06, |
| "loss": 0.014298772811889649, |
| "num_tokens": 352282.0, |
| "reward": 1.0499975621700286, |
| "reward_std": 0.013283280395990004, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.025048828125, |
| "rewards/reward_length/std": 0.005499870236963034, |
| "rewards/reward_semantic/mean": 0.1708333333954215, |
| "rewards/reward_semantic/std": 0.044191432371735576, |
| "step": 1180, |
| "step_time": 1.5698805461998746 |
| }, |
| { |
| "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": 5.59375, |
| "completions/mean_terminated_length": 5.59375, |
| "completions/min_length": 5.2, |
| "completions/min_terminated_length": 5.2, |
| "entropy": 0.09904264745746331, |
| "epoch": 0.1968503937007874, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.4005000000000005e-06, |
| "loss": 0.018337526917457582, |
| "num_tokens": 358529.0, |
| "reward": 1.102414545416832, |
| "reward_std": 0.015537488225891138, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.026708984375, |
| "rewards/reward_length/std": 0.003325884399237111, |
| "rewards/reward_semantic/mean": 0.34583333432674407, |
| "rewards/reward_semantic/std": 0.051754919439554216, |
| "step": 1200, |
| "step_time": 1.6073716132496885 |
| }, |
| { |
| "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": 5.1375, |
| "completions/mean_terminated_length": 5.1375, |
| "completions/min_length": 4.7, |
| "completions/min_terminated_length": 4.7, |
| "entropy": 0.16791536924720277, |
| "epoch": 0.2001312335958005, |
| "frac_reward_zero_std": 0.35, |
| "grad_norm": 16.625, |
| "learning_rate": 4.3905000000000006e-06, |
| "loss": 0.023975622653961182, |
| "num_tokens": 364687.0, |
| "reward": 1.0299089789390563, |
| "reward_std": 0.01372892598665203, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.025927734375, |
| "rewards/reward_length/std": 0.005886835453566164, |
| "rewards/reward_semantic/mean": 0.10401785671710968, |
| "rewards/reward_semantic/std": 0.0461477629840374, |
| "step": 1220, |
| "step_time": 1.636508045949722 |
| }, |
| { |
| "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": 4.9375, |
| "completions/mean_terminated_length": 4.9375, |
| "completions/min_length": 4.2, |
| "completions/min_terminated_length": 4.2, |
| "entropy": 0.1175628273136681, |
| "epoch": 0.20341207349081364, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 0.0, |
| "learning_rate": 4.380500000000001e-06, |
| "loss": 0.0339366614818573, |
| "num_tokens": 370549.0, |
| "reward": 1.048176285624504, |
| "reward_std": 0.023710943618789317, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.020849609375, |
| "rewards/reward_length/std": 0.0026248104171827436, |
| "rewards/reward_semantic/mean": 0.1640625, |
| "rewards/reward_semantic/std": 0.07922998964786529, |
| "step": 1240, |
| "step_time": 1.6752965371496429 |
| }, |
| { |
| "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": 5.05, |
| "completions/max_terminated_length": 5.05, |
| "completions/mean_length": 4.80625, |
| "completions/mean_terminated_length": 4.80625, |
| "completions/min_length": 4.6, |
| "completions/min_terminated_length": 4.6, |
| "entropy": 0.09321834340953501, |
| "epoch": 0.20669291338582677, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 22.75, |
| "learning_rate": 4.3705e-06, |
| "loss": 0.0076515451073646545, |
| "num_tokens": 376662.0, |
| "reward": 1.031198051571846, |
| "reward_std": 0.003262797408387996, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.018896484375, |
| "rewards/reward_length/std": 0.0032897572789806874, |
| "rewards/reward_semantic/mean": 0.10714285746216774, |
| "rewards/reward_semantic/std": 0.01079898551106453, |
| "step": 1260, |
| "step_time": 1.6717814109499158 |
| }, |
| { |
| "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": 5.2625, |
| "completions/mean_terminated_length": 5.2625, |
| "completions/min_length": 4.8, |
| "completions/min_terminated_length": 4.8, |
| "entropy": 0.10832190366709256, |
| "epoch": 0.2099737532808399, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.3605e-06, |
| "loss": 0.016128246486186982, |
| "num_tokens": 383024.0, |
| "reward": 1.013911136984825, |
| "reward_std": 0.00022039367468096316, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02177734375, |
| "rewards/reward_length/std": 0.004407888412242755, |
| "rewards/reward_semantic/mean": 0.05, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 1280, |
| "step_time": 1.6772221600500417 |
| }, |
| { |
| "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": 4.775, |
| "completions/mean_terminated_length": 4.775, |
| "completions/min_length": 4.15, |
| "completions/min_terminated_length": 4.15, |
| "entropy": 0.13252608955590403, |
| "epoch": 0.21325459317585302, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.3505e-06, |
| "loss": 0.06973940134048462, |
| "num_tokens": 389052.0, |
| "reward": 1.0591955691576005, |
| "reward_std": 0.02507804925335222, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0160888671875, |
| "rewards/reward_length/std": 0.001597413793206215, |
| "rewards/reward_semantic/mean": 0.2, |
| "rewards/reward_semantic/std": 0.0836443156003952, |
| "step": 1300, |
| "step_time": 1.815399284299383 |
| }, |
| { |
| "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": 5.25, |
| "completions/max_terminated_length": 5.25, |
| "completions/mean_length": 4.74375, |
| "completions/mean_terminated_length": 4.74375, |
| "completions/min_length": 4.35, |
| "completions/min_terminated_length": 4.35, |
| "entropy": 0.16815137023077115, |
| "epoch": 0.21653543307086615, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.3405e-06, |
| "loss": 0.039281854033470155, |
| "num_tokens": 395083.0, |
| "reward": 1.087931489944458, |
| "reward_std": 0.017514032226245035, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.019287109375, |
| "rewards/reward_length/std": 0.003718964639119804, |
| "rewards/reward_semantic/mean": 0.29631944447755815, |
| "rewards/reward_semantic/std": 0.05840769112110138, |
| "step": 1320, |
| "step_time": 1.5450484630991013 |
| }, |
| { |
| "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": 4.8, |
| "completions/max_terminated_length": 4.8, |
| "completions/mean_length": 4.5, |
| "completions/mean_terminated_length": 4.5, |
| "completions/min_length": 4.0, |
| "completions/min_terminated_length": 4.0, |
| "entropy": 0.11911952229129383, |
| "epoch": 0.21981627296587927, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.3305e-06, |
| "loss": 0.02354532480239868, |
| "num_tokens": 401019.0, |
| "reward": 1.0643640339374543, |
| "reward_std": 0.03804397364147007, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0252197265625, |
| "rewards/reward_length/std": 0.005615488672628999, |
| "rewards/reward_semantic/mean": 0.21875, |
| "rewards/reward_semantic/std": 0.12704137191176415, |
| "step": 1340, |
| "step_time": 1.511271317900355 |
| }, |
| { |
| "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": 4.95, |
| "completions/max_terminated_length": 4.95, |
| "completions/mean_length": 4.5125, |
| "completions/mean_terminated_length": 4.5125, |
| "completions/min_length": 4.15, |
| "completions/min_terminated_length": 4.15, |
| "entropy": 0.13438130084614386, |
| "epoch": 0.2230971128608924, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.3205e-06, |
| "loss": -0.014946562051773072, |
| "num_tokens": 406597.0, |
| "reward": 1.094342541694641, |
| "reward_std": 0.02126533806695079, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0218994140625, |
| "rewards/reward_length/std": 0.0038008941686712206, |
| "rewards/reward_semantic/mean": 0.3181250005960464, |
| "rewards/reward_semantic/std": 0.071246337890625, |
| "step": 1360, |
| "step_time": 1.5624590492494463 |
| }, |
| { |
| "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": 4.81875, |
| "completions/mean_terminated_length": 4.81875, |
| "completions/min_length": 4.25, |
| "completions/min_terminated_length": 4.25, |
| "entropy": 0.18390740080249088, |
| "epoch": 0.2263779527559055, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 5.71875, |
| "learning_rate": 4.3105000000000005e-06, |
| "loss": -0.006673748791217804, |
| "num_tokens": 412872.0, |
| "reward": 1.030611902475357, |
| "reward_std": 0.02634383789918502, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0259765625, |
| "rewards/reward_length/std": 0.00884399957722053, |
| "rewards/reward_semantic/mean": 0.10636904872953892, |
| "rewards/reward_semantic/std": 0.08875287175178528, |
| "step": 1380, |
| "step_time": 1.7178449202000592 |
| }, |
| { |
| "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": 4.53125, |
| "completions/mean_terminated_length": 4.53125, |
| "completions/min_length": 4.2, |
| "completions/min_terminated_length": 4.2, |
| "entropy": 0.15863893868372542, |
| "epoch": 0.22965879265091863, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.300500000000001e-06, |
| "loss": -0.001099221408367157, |
| "num_tokens": 418653.0, |
| "reward": 1.06249757707119, |
| "reward_std": 0.021931025733647403, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.025048828125, |
| "rewards/reward_length/std": 0.005685041821561754, |
| "rewards/reward_semantic/mean": 0.2125, |
| "rewards/reward_semantic/std": 0.07325703874230385, |
| "step": 1400, |
| "step_time": 1.5830605980497239 |
| }, |
| { |
| "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": 4.7, |
| "completions/max_terminated_length": 4.7, |
| "completions/mean_length": 4.1125, |
| "completions/mean_terminated_length": 4.1125, |
| "completions/min_length": 3.45, |
| "completions/min_terminated_length": 3.45, |
| "entropy": 0.25911560031236147, |
| "epoch": 0.23293963254593175, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 6.65625, |
| "learning_rate": 4.2905e-06, |
| "loss": 0.07240142226219178, |
| "num_tokens": 424375.0, |
| "reward": 1.091744390130043, |
| "reward_std": 0.02755639751521812, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0276123046875, |
| "rewards/reward_length/std": 0.0038219673559069635, |
| "rewards/reward_semantic/mean": 0.3104166679084301, |
| "rewards/reward_semantic/std": 0.09196329787373543, |
| "step": 1420, |
| "step_time": 1.6718641993996244 |
| }, |
| { |
| "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": 4.6, |
| "completions/max_terminated_length": 4.6, |
| "completions/mean_length": 4.225, |
| "completions/mean_terminated_length": 4.225, |
| "completions/min_length": 3.95, |
| "completions/min_terminated_length": 3.95, |
| "entropy": 0.15136082460521721, |
| "epoch": 0.23622047244094488, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.2805e-06, |
| "loss": 0.014926818013191224, |
| "num_tokens": 430275.0, |
| "reward": 1.119848296046257, |
| "reward_std": 0.00218765427962353, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.024462890625, |
| "rewards/reward_length/std": 0.005067214759765193, |
| "rewards/reward_semantic/mean": 0.40357142873108387, |
| "rewards/reward_semantic/std": 0.006613001227378845, |
| "step": 1440, |
| "step_time": 1.7020611779495993 |
| }, |
| { |
| "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": 4.8, |
| "completions/max_terminated_length": 4.8, |
| "completions/mean_length": 4.525, |
| "completions/mean_terminated_length": 4.525, |
| "completions/min_length": 3.95, |
| "completions/min_terminated_length": 3.95, |
| "entropy": 0.13695635193871566, |
| "epoch": 0.239501312335958, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.2705e-06, |
| "loss": -0.007017912715673447, |
| "num_tokens": 436319.0, |
| "reward": 1.1099511831998825, |
| "reward_std": 0.02091348784088041, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0259765625, |
| "rewards/reward_length/std": 0.0026839403493795545, |
| "rewards/reward_semantic/mean": 0.3708333333954215, |
| "rewards/reward_semantic/std": 0.06986123211681843, |
| "step": 1460, |
| "step_time": 1.5354945631996997 |
| }, |
| { |
| "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": 4.9, |
| "completions/max_terminated_length": 4.9, |
| "completions/mean_length": 4.4, |
| "completions/mean_terminated_length": 4.4, |
| "completions/min_length": 3.9, |
| "completions/min_terminated_length": 3.9, |
| "entropy": 0.21424212285055547, |
| "epoch": 0.24278215223097113, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.2605e-06, |
| "loss": 0.029949143528938293, |
| "num_tokens": 442239.0, |
| "reward": 1.1345235884189606, |
| "reward_std": 0.013351330516525196, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.03095703125, |
| "rewards/reward_length/std": 0.006451338075567037, |
| "rewards/reward_semantic/mean": 0.45357142984867094, |
| "rewards/reward_semantic/std": 0.043555130064487454, |
| "step": 1480, |
| "step_time": 1.5475482691500475 |
| }, |
| { |
| "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": 5.1, |
| "completions/max_terminated_length": 5.1, |
| "completions/mean_length": 4.6875, |
| "completions/mean_terminated_length": 4.6875, |
| "completions/min_length": 4.45, |
| "completions/min_terminated_length": 4.45, |
| "entropy": 0.12956037224794273, |
| "epoch": 0.24606299212598426, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.2505e-06, |
| "loss": 0.02256421893835068, |
| "num_tokens": 448397.0, |
| "reward": 1.07148334980011, |
| "reward_std": 0.014302821262390353, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.022119140625, |
| "rewards/reward_length/std": 0.006394583627115935, |
| "rewards/reward_semantic/mean": 0.24196428656578065, |
| "rewards/reward_semantic/std": 0.04743617326021195, |
| "step": 1500, |
| "step_time": 1.5719447982004566 |
| }, |
| { |
| "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": 4.925, |
| "completions/mean_terminated_length": 4.925, |
| "completions/min_length": 4.6, |
| "completions/min_terminated_length": 4.6, |
| "entropy": 0.10781946495189913, |
| "epoch": 0.24934383202099739, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 4.2405e-06, |
| "loss": -0.03193822503089905, |
| "num_tokens": 454585.0, |
| "reward": 1.0354276090860366, |
| "reward_std": 0.019653240226762136, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02001953125, |
| "rewards/reward_length/std": 0.004805501899681985, |
| "rewards/reward_semantic/mean": 0.12142857313156127, |
| "rewards/reward_semantic/std": 0.06617598533630371, |
| "step": 1520, |
| "step_time": 1.5884246722999706 |
| }, |
| { |
| "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": 4.6875, |
| "completions/mean_terminated_length": 4.6875, |
| "completions/min_length": 4.4, |
| "completions/min_terminated_length": 4.4, |
| "entropy": 0.16822912358038594, |
| "epoch": 0.2526246719160105, |
| "frac_reward_zero_std": 0.65, |
| "grad_norm": 5.8125, |
| "learning_rate": 4.2305e-06, |
| "loss": 0.0058339819312095646, |
| "num_tokens": 460479.0, |
| "reward": 1.0865124613046646, |
| "reward_std": 0.023904979706276207, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0228759765625, |
| "rewards/reward_length/std": 0.004470406926702708, |
| "rewards/reward_semantic/mean": 0.2921875, |
| "rewards/reward_semantic/std": 0.07995888479053974, |
| "step": 1540, |
| "step_time": 1.5527578148497014 |
| }, |
| { |
| "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": 4.36875, |
| "completions/mean_terminated_length": 4.36875, |
| "completions/min_length": 3.95, |
| "completions/min_terminated_length": 3.95, |
| "entropy": 0.22550911762573378, |
| "epoch": 0.2559055118110236, |
| "frac_reward_zero_std": 0.55, |
| "grad_norm": 10.125, |
| "learning_rate": 4.2205000000000005e-06, |
| "loss": 0.005285773798823357, |
| "num_tokens": 466370.0, |
| "reward": 1.1038363248109817, |
| "reward_std": 0.022896537078850088, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0322021484375, |
| "rewards/reward_length/std": 0.006792640773346648, |
| "rewards/reward_semantic/mean": 0.3514880958944559, |
| "rewards/reward_semantic/std": 0.0760455921292305, |
| "step": 1560, |
| "step_time": 1.6276254383002198 |
| }, |
| { |
| "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": 5.35, |
| "completions/max_terminated_length": 5.35, |
| "completions/mean_length": 4.80625, |
| "completions/mean_terminated_length": 4.80625, |
| "completions/min_length": 4.4, |
| "completions/min_terminated_length": 4.4, |
| "entropy": 0.1503011735556356, |
| "epoch": 0.25918635170603677, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 4.2105e-06, |
| "loss": 0.010880699008703231, |
| "num_tokens": 472403.0, |
| "reward": 1.0855810642242432, |
| "reward_std": 0.01420300938298169, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02275390625, |
| "rewards/reward_length/std": 0.005383360758423805, |
| "rewards/reward_semantic/mean": 0.2890625, |
| "rewards/reward_semantic/std": 0.04729253724217415, |
| "step": 1580, |
| "step_time": 1.655612524249591 |
| }, |
| { |
| "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": 4.45, |
| "completions/max_terminated_length": 4.45, |
| "completions/mean_length": 4.11875, |
| "completions/mean_terminated_length": 4.11875, |
| "completions/min_length": 3.75, |
| "completions/min_terminated_length": 3.75, |
| "entropy": 0.16699727870654896, |
| "epoch": 0.26246719160104987, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.2005e-06, |
| "loss": 0.05660162568092346, |
| "num_tokens": 477918.0, |
| "reward": 1.1160965353250503, |
| "reward_std": 0.014429239320816124, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.019140625, |
| "rewards/reward_length/std": 0.002461307460907847, |
| "rewards/reward_semantic/mean": 0.39017857164144515, |
| "rewards/reward_semantic/std": 0.04821673259139061, |
| "step": 1600, |
| "step_time": 1.5680961081498026 |
| }, |
| { |
| "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": 4.95, |
| "completions/max_terminated_length": 4.95, |
| "completions/mean_length": 4.625, |
| "completions/mean_terminated_length": 4.625, |
| "completions/min_length": 4.35, |
| "completions/min_terminated_length": 4.35, |
| "entropy": 0.12586015183624114, |
| "epoch": 0.265748031496063, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.1905e-06, |
| "loss": 0.0030185095965862274, |
| "num_tokens": 483938.0, |
| "reward": 1.1076241761446, |
| "reward_std": 0.02165422995312838, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0251953125, |
| "rewards/reward_length/std": 0.0058576216455549, |
| "rewards/reward_semantic/mean": 0.3629464291036129, |
| "rewards/reward_semantic/std": 0.07241237610578537, |
| "step": 1620, |
| "step_time": 1.5842454133509818 |
| }, |
| { |
| "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": 4.5, |
| "completions/max_terminated_length": 4.5, |
| "completions/mean_length": 4.1625, |
| "completions/mean_terminated_length": 4.1625, |
| "completions/min_length": 3.9, |
| "completions/min_terminated_length": 3.9, |
| "entropy": 0.16225300922524183, |
| "epoch": 0.2690288713910761, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.1805e-06, |
| "loss": 0.02639976441860199, |
| "num_tokens": 489628.0, |
| "reward": 1.1402275443077088, |
| "reward_std": 0.020069138445251156, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02919921875, |
| "rewards/reward_length/std": 0.00430998255033046, |
| "rewards/reward_semantic/mean": 0.47229166738688944, |
| "rewards/reward_semantic/std": 0.06731179356575012, |
| "step": 1640, |
| "step_time": 1.5262142101497376 |
| }, |
| { |
| "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": 5.35625, |
| "completions/mean_terminated_length": 5.35625, |
| "completions/min_length": 5.0, |
| "completions/min_terminated_length": 5.0, |
| "entropy": 0.08895924828648276, |
| "epoch": 0.2723097112860892, |
| "frac_reward_zero_std": 0.65, |
| "grad_norm": 11.4375, |
| "learning_rate": 4.1705e-06, |
| "loss": -0.02690776586532593, |
| "num_tokens": 495533.0, |
| "reward": 1.077004411816597, |
| "reward_std": 0.0072896559962828174, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.025537109375, |
| "rewards/reward_length/std": 0.007682852516882122, |
| "rewards/reward_semantic/mean": 0.2609375, |
| "rewards/reward_semantic/std": 0.02414703443646431, |
| "step": 1660, |
| "step_time": 1.6573960603499471 |
| }, |
| { |
| "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": 5.1, |
| "completions/max_terminated_length": 5.1, |
| "completions/mean_length": 4.875, |
| "completions/mean_terminated_length": 4.875, |
| "completions/min_length": 4.4, |
| "completions/min_terminated_length": 4.4, |
| "entropy": 0.12892614689626497, |
| "epoch": 0.2755905511811024, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 4.1605e-06, |
| "loss": -0.00665338933467865, |
| "num_tokens": 501313.0, |
| "reward": 1.1107507348060608, |
| "reward_std": 0.01917153293034062, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0287353515625, |
| "rewards/reward_length/std": 0.004330105870030821, |
| "rewards/reward_semantic/mean": 0.373958333581686, |
| "rewards/reward_semantic/std": 0.06437318101525306, |
| "step": 1680, |
| "step_time": 1.8099103468489375 |
| }, |
| { |
| "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": 4.68125, |
| "completions/mean_terminated_length": 4.68125, |
| "completions/min_length": 4.45, |
| "completions/min_terminated_length": 4.45, |
| "entropy": 0.19288054770295276, |
| "epoch": 0.27887139107611547, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 4.1505e-06, |
| "loss": 0.07687278985977172, |
| "num_tokens": 507102.0, |
| "reward": 1.077316564321518, |
| "reward_std": 0.004241698494297452, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02509765625, |
| "rewards/reward_length/std": 0.00411116995383054, |
| "rewards/reward_semantic/mean": 0.2619047623127699, |
| "rewards/reward_semantic/std": 0.013468702137470246, |
| "step": 1700, |
| "step_time": 1.5861379748002946 |
| }, |
| { |
| "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": 5.1, |
| "completions/max_terminated_length": 5.1, |
| "completions/mean_length": 4.7125, |
| "completions/mean_terminated_length": 4.7125, |
| "completions/min_length": 4.4, |
| "completions/min_terminated_length": 4.4, |
| "entropy": 0.18708740206056973, |
| "epoch": 0.2821522309711286, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 4.1405e-06, |
| "loss": 0.026335787773132325, |
| "num_tokens": 513200.0, |
| "reward": 1.0792573064565658, |
| "reward_std": 0.012720743837417104, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0219970703125, |
| "rewards/reward_length/std": 0.005083874787669629, |
| "rewards/reward_semantic/mean": 0.26785714402794836, |
| "rewards/reward_semantic/std": 0.042097911983728406, |
| "step": 1720, |
| "step_time": 1.6246219198992549 |
| }, |
| { |
| "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": 4.85, |
| "completions/max_terminated_length": 4.85, |
| "completions/mean_length": 4.34375, |
| "completions/mean_terminated_length": 4.34375, |
| "completions/min_length": 4.1, |
| "completions/min_terminated_length": 4.1, |
| "entropy": 0.12696853005909362, |
| "epoch": 0.2854330708661417, |
| "frac_reward_zero_std": 0.85, |
| "grad_norm": 7.75, |
| "learning_rate": 4.1305e-06, |
| "loss": 0.0013355985283851624, |
| "num_tokens": 518775.0, |
| "reward": 1.1246948331594466, |
| "reward_std": 0.006485381873790175, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.018603515625, |
| "rewards/reward_length/std": 0.002097794343717396, |
| "rewards/reward_semantic/mean": 0.41875, |
| "rewards/reward_semantic/std": 0.021777565032243727, |
| "step": 1740, |
| "step_time": 1.5038381617996492 |
| }, |
| { |
| "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": 4.74375, |
| "completions/mean_terminated_length": 4.74375, |
| "completions/min_length": 3.55, |
| "completions/min_terminated_length": 3.55, |
| "entropy": 0.20232484516163823, |
| "epoch": 0.2887139107611549, |
| "frac_reward_zero_std": 0.45, |
| "grad_norm": 9.125, |
| "learning_rate": 4.120500000000001e-06, |
| "loss": 0.019324113428592683, |
| "num_tokens": 524710.0, |
| "reward": 1.0573926001787186, |
| "reward_std": 0.022785998944891618, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0240234375, |
| "rewards/reward_length/std": 0.007111385487951339, |
| "rewards/reward_semantic/mean": 0.1953125, |
| "rewards/reward_semantic/std": 0.07635064832866192, |
| "step": 1760, |
| "step_time": 1.5597154122505343 |
| }, |
| { |
| "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": 5.25, |
| "completions/max_terminated_length": 5.25, |
| "completions/mean_length": 4.1, |
| "completions/mean_terminated_length": 4.1, |
| "completions/min_length": 2.8, |
| "completions/min_terminated_length": 2.8, |
| "entropy": 0.34262509692343884, |
| "epoch": 0.291994750656168, |
| "frac_reward_zero_std": 0.4, |
| "grad_norm": 16.5, |
| "learning_rate": 4.110500000000001e-06, |
| "loss": 0.11038769483566284, |
| "num_tokens": 530598.0, |
| "reward": 1.0436306297779083, |
| "reward_std": 0.015193032496608794, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0161376953125, |
| "rewards/reward_length/std": 0.00954111129976809, |
| "rewards/reward_semantic/mean": 0.14812500029802322, |
| "rewards/reward_semantic/std": 0.05001583844423294, |
| "step": 1780, |
| "step_time": 1.5629329150999183 |
| }, |
| { |
| "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": 4.0, |
| "completions/max_terminated_length": 4.0, |
| "completions/mean_length": 3.0375, |
| "completions/mean_terminated_length": 3.0375, |
| "completions/min_length": 2.4, |
| "completions/min_terminated_length": 2.4, |
| "entropy": 0.41122397423314394, |
| "epoch": 0.2952755905511811, |
| "frac_reward_zero_std": 0.25, |
| "grad_norm": 0.0, |
| "learning_rate": 4.100500000000001e-06, |
| "loss": 0.0228089839220047, |
| "num_tokens": 536380.0, |
| "reward": 1.096625980734825, |
| "reward_std": 0.039403259825485294, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.02373046875, |
| "rewards/reward_length/std": 0.00941192212048918, |
| "rewards/reward_semantic/mean": 0.326041666790843, |
| "rewards/reward_semantic/std": 0.13154361248016358, |
| "step": 1800, |
| "step_time": 1.4928159248503108 |
| }, |
| { |
| "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": 3.2, |
| "completions/max_terminated_length": 3.2, |
| "completions/mean_length": 2.73125, |
| "completions/mean_terminated_length": 2.73125, |
| "completions/min_length": 2.15, |
| "completions/min_terminated_length": 2.15, |
| "entropy": 0.41097258800436975, |
| "epoch": 0.29855643044619423, |
| "frac_reward_zero_std": 0.45, |
| "grad_norm": 37.75, |
| "learning_rate": 4.090500000000001e-06, |
| "loss": 0.03736543655395508, |
| "num_tokens": 541841.0, |
| "reward": 1.0882896214723587, |
| "reward_std": 0.029034473038336728, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.019921875, |
| "rewards/reward_length/std": 0.009078961075283587, |
| "rewards/reward_semantic/mean": 0.297619048692286, |
| "rewards/reward_semantic/std": 0.09641024619340896, |
| "step": 1820, |
| "step_time": 1.4359207120499378 |
| }, |
| { |
| "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": 4.6, |
| "completions/max_terminated_length": 4.6, |
| "completions/mean_length": 3.56875, |
| "completions/mean_terminated_length": 3.56875, |
| "completions/min_length": 2.8, |
| "completions/min_terminated_length": 2.8, |
| "entropy": 0.3877414964954369, |
| "epoch": 0.30183727034120733, |
| "frac_reward_zero_std": 0.45, |
| "grad_norm": 0.0, |
| "learning_rate": 4.0805e-06, |
| "loss": 0.10119327306747436, |
| "num_tokens": 547868.0, |
| "reward": 1.0956469804048539, |
| "reward_std": 0.012229273936827666, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.024560546875, |
| "rewards/reward_length/std": 0.008132253866642713, |
| "rewards/reward_semantic/mean": 0.3229166677221656, |
| "rewards/reward_semantic/std": 0.040395837649703024, |
| "step": 1840, |
| "step_time": 1.571252691800055 |
| }, |
| { |
| "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": 3.45, |
| "completions/max_terminated_length": 3.45, |
| "completions/mean_length": 2.40625, |
| "completions/mean_terminated_length": 2.40625, |
| "completions/min_length": 1.3, |
| "completions/min_terminated_length": 1.3, |
| "entropy": 0.3788038203958422, |
| "epoch": 0.3051181102362205, |
| "frac_reward_zero_std": 0.5, |
| "grad_norm": 10.0, |
| "learning_rate": 4.0705e-06, |
| "loss": 0.15065158605575563, |
| "num_tokens": 553645.0, |
| "reward": 1.02579345703125, |
| "reward_std": 0.007370092621567892, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.009130859375, |
| "rewards/reward_length/std": 0.009433019044809043, |
| "rewards/reward_semantic/mean": 0.0875, |
| "rewards/reward_semantic/std": 0.023145502805709837, |
| "step": 1860, |
| "step_time": 1.4640108326503651 |
| }, |
| { |
| "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": 1.85, |
| "completions/max_terminated_length": 1.85, |
| "completions/mean_length": 1.48125, |
| "completions/mean_terminated_length": 1.48125, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.1245657365056104, |
| "epoch": 0.3083989501312336, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 4.0605e-06, |
| "loss": 0.05667008757591248, |
| "num_tokens": 558986.0, |
| "reward": 1.0280249089002609, |
| "reward_std": 0.005344233532377984, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.002001953125, |
| "rewards/reward_length/std": 0.0013710828498005867, |
| "rewards/reward_semantic/mean": 0.09375, |
| "rewards/reward_semantic/std": 0.01767766922712326, |
| "step": 1880, |
| "step_time": 1.3200888238508015 |
| }, |
| { |
| "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": 1.7, |
| "completions/max_terminated_length": 1.7, |
| "completions/mean_length": 1.2875, |
| "completions/mean_terminated_length": 1.2875, |
| "completions/min_length": 1.05, |
| "completions/min_terminated_length": 1.05, |
| "entropy": 0.1832592948281672, |
| "epoch": 0.31167979002624674, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 4.0505000000000004e-06, |
| "loss": -0.03582154512405396, |
| "num_tokens": 564648.0, |
| "reward": 1.0297485411167144, |
| "reward_std": 0.022522959530761, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.005029296875, |
| "rewards/reward_length/std": 0.003776776511222124, |
| "rewards/reward_semantic/mean": 0.1, |
| "rewards/reward_semantic/std": 0.07554142624139785, |
| "step": 1900, |
| "step_time": 1.3407865802008927 |
| }, |
| { |
| "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": 1.75, |
| "completions/max_terminated_length": 1.75, |
| "completions/mean_length": 1.46875, |
| "completions/mean_terminated_length": 1.46875, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.19475372634042287, |
| "epoch": 0.31496062992125984, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 4.0405000000000005e-06, |
| "loss": -0.1091422200202942, |
| "num_tokens": 569867.0, |
| "reward": 1.0441931068897248, |
| "reward_std": 0.03400199599564076, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0036376953125, |
| "rewards/reward_length/std": 0.002957951743155718, |
| "rewards/reward_semantic/mean": 0.14791666716337204, |
| "rewards/reward_semantic/std": 0.11383298859000206, |
| "step": 1920, |
| "step_time": 1.369970721050413 |
| }, |
| { |
| "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": 1.7, |
| "completions/max_terminated_length": 1.7, |
| "completions/mean_length": 1.3875, |
| "completions/mean_terminated_length": 1.3875, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.17718661450198853, |
| "epoch": 0.318241469816273, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 4.030500000000001e-06, |
| "loss": -0.027825871109962465, |
| "num_tokens": 575025.0, |
| "reward": 1.0133480489253999, |
| "reward_std": 0.012130760264699347, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0044677734375, |
| "rewards/reward_length/std": 0.003881294187158346, |
| "rewards/reward_semantic/mean": 0.04523809552192688, |
| "rewards/reward_semantic/std": 0.0409621961414814, |
| "step": 1940, |
| "step_time": 1.3533337505006784 |
| }, |
| { |
| "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": 2.2, |
| "completions/max_terminated_length": 2.2, |
| "completions/mean_length": 1.575, |
| "completions/mean_terminated_length": 1.575, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.36525906809256414, |
| "epoch": 0.3215223097112861, |
| "frac_reward_zero_std": 0.55, |
| "grad_norm": 43.0, |
| "learning_rate": 4.020500000000001e-06, |
| "loss": 0.04816198050975799, |
| "num_tokens": 580237.0, |
| "reward": 1.0371252387762069, |
| "reward_std": 0.02014696542028105, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0074951171875, |
| "rewards/reward_length/std": 0.006795872491784394, |
| "rewards/reward_semantic/mean": 0.125, |
| "rewards/reward_semantic/std": 0.0667006328701973, |
| "step": 1960, |
| "step_time": 1.4318877942499058 |
| }, |
| { |
| "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": 2.65, |
| "completions/max_terminated_length": 2.65, |
| "completions/mean_length": 1.8625, |
| "completions/mean_terminated_length": 1.8625, |
| "completions/min_length": 1.45, |
| "completions/min_terminated_length": 1.45, |
| "entropy": 0.25888301711238454, |
| "epoch": 0.3248031496062992, |
| "frac_reward_zero_std": 0.55, |
| "grad_norm": 0.0, |
| "learning_rate": 4.010500000000001e-06, |
| "loss": 0.11174570322036743, |
| "num_tokens": 585599.0, |
| "reward": 1.0821655124425889, |
| "reward_std": 0.017148944691871292, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.006689453125, |
| "rewards/reward_length/std": 0.003263966436497867, |
| "rewards/reward_semantic/mean": 0.275, |
| "rewards/reward_semantic/std": 0.05713290348649025, |
| "step": 1980, |
| "step_time": 1.4646942404000582 |
| }, |
| { |
| "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": 2.35, |
| "completions/max_terminated_length": 2.35, |
| "completions/mean_length": 1.85, |
| "completions/mean_terminated_length": 1.85, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.25707166531865366, |
| "epoch": 0.32808398950131235, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 4.000500000000001e-06, |
| "loss": -0.0278127521276474, |
| "num_tokens": 591191.0, |
| "reward": 1.0355240941047668, |
| "reward_std": 0.007867054975940846, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0103515625, |
| "rewards/reward_length/std": 0.009302021842449904, |
| "rewards/reward_semantic/mean": 0.12013888955116273, |
| "rewards/reward_semantic/std": 0.02702435702085495, |
| "step": 2000, |
| "step_time": 1.426078681249146 |
| }, |
| { |
| "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": 2.4, |
| "completions/max_terminated_length": 2.4, |
| "completions/mean_length": 1.63125, |
| "completions/mean_terminated_length": 1.63125, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.28791044880636035, |
| "epoch": 0.33136482939632544, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 3.9905e-06, |
| "loss": 0.0719864845275879, |
| "num_tokens": 596868.0, |
| "reward": 1.0446875005960465, |
| "reward_std": 0.0211071026918944, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.00625, |
| "rewards/reward_length/std": 0.00591426498722285, |
| "rewards/reward_semantic/mean": 0.15, |
| "rewards/reward_semantic/std": 0.07028125375509262, |
| "step": 2020, |
| "step_time": 1.474500678699951 |
| }, |
| { |
| "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": 1.95, |
| "completions/max_terminated_length": 1.95, |
| "completions/mean_length": 1.65625, |
| "completions/mean_terminated_length": 1.65625, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.2252592260585516, |
| "epoch": 0.3346456692913386, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.9805e-06, |
| "loss": 0.01632298231124878, |
| "num_tokens": 602421.0, |
| "reward": 1.0201035290956497, |
| "reward_std": 0.005319315989618189, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0107177734375, |
| "rewards/reward_length/std": 0.005171247851103544, |
| "rewards/reward_semantic/mean": 0.06879807692021131, |
| "rewards/reward_semantic/std": 0.01816960349678993, |
| "step": 2040, |
| "step_time": 1.3785421489007603 |
| }, |
| { |
| "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": 2.45, |
| "completions/max_terminated_length": 2.45, |
| "completions/mean_length": 1.69375, |
| "completions/mean_terminated_length": 1.69375, |
| "completions/min_length": 1.25, |
| "completions/min_terminated_length": 1.25, |
| "entropy": 0.22661980051198044, |
| "epoch": 0.3379265091863517, |
| "frac_reward_zero_std": 0.65, |
| "grad_norm": 16.75, |
| "learning_rate": 3.9705e-06, |
| "loss": -0.04810878336429596, |
| "num_tokens": 607916.0, |
| "reward": 1.0543103069067001, |
| "reward_std": 0.029334707593079656, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0075439453125, |
| "rewards/reward_length/std": 0.007363162655383349, |
| "rewards/reward_semantic/mean": 0.18229166641831399, |
| "rewards/reward_semantic/std": 0.09833509251475334, |
| "step": 2060, |
| "step_time": 1.4630046169510933 |
| }, |
| { |
| "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": 1.95, |
| "completions/max_terminated_length": 1.95, |
| "completions/mean_length": 1.68125, |
| "completions/mean_terminated_length": 1.68125, |
| "completions/min_length": 1.45, |
| "completions/min_terminated_length": 1.45, |
| "entropy": 0.18975868790876121, |
| "epoch": 0.34120734908136485, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.9605000000000005e-06, |
| "loss": 0.04834949374198914, |
| "num_tokens": 613481.0, |
| "reward": 1.0727307140827178, |
| "reward_std": 0.005326971549948212, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0078857421875, |
| "rewards/reward_length/std": 0.001163953309878707, |
| "rewards/reward_semantic/mean": 0.24375, |
| "rewards/reward_semantic/std": 0.01767766922712326, |
| "step": 2080, |
| "step_time": 1.3919913940497282 |
| }, |
| { |
| "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": 1.7, |
| "completions/max_terminated_length": 1.7, |
| "completions/mean_length": 1.45625, |
| "completions/mean_terminated_length": 1.45625, |
| "completions/min_length": 1.3, |
| "completions/min_terminated_length": 1.3, |
| "entropy": 0.26590851033106444, |
| "epoch": 0.34448818897637795, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.9505000000000005e-06, |
| "loss": 0.07636955976486207, |
| "num_tokens": 618530.0, |
| "reward": 1.0447363257408142, |
| "reward_std": 0.0001262445846805349, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0052734375, |
| "rewards/reward_length/std": 0.0025248861871659755, |
| "rewards/reward_semantic/mean": 0.15, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 2100, |
| "step_time": 1.37139456360037 |
| }, |
| { |
| "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": 1.35, |
| "completions/max_terminated_length": 1.35, |
| "completions/mean_length": 1.275, |
| "completions/mean_terminated_length": 1.275, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.10506298460531979, |
| "epoch": 0.34776902887139105, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 95.0, |
| "learning_rate": 3.940500000000001e-06, |
| "loss": -0.03741676807403564, |
| "num_tokens": 623814.0, |
| "reward": 1.026152342557907, |
| "reward_std": 0.011182133108377457, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.001953125, |
| "rewards/reward_length/std": 0.0010585635434836149, |
| "rewards/reward_semantic/mean": 0.0875, |
| "rewards/reward_semantic/std": 0.03745021224021912, |
| "step": 2120, |
| "step_time": 1.5095237725503465 |
| }, |
| { |
| "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": 1.9, |
| "completions/max_terminated_length": 1.9, |
| "completions/mean_length": 1.59375, |
| "completions/mean_terminated_length": 1.59375, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.14830810520070373, |
| "epoch": 0.3510498687664042, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.930500000000001e-06, |
| "loss": -0.01694631278514862, |
| "num_tokens": 629077.0, |
| "reward": 1.0316784739494325, |
| "reward_std": 0.014781178609700874, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0039306640625, |
| "rewards/reward_length/std": 0.0030176832806319, |
| "rewards/reward_semantic/mean": 0.10625, |
| "rewards/reward_semantic/std": 0.049022963643074034, |
| "step": 2140, |
| "step_time": 1.769969190950087 |
| }, |
| { |
| "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": 1.9, |
| "completions/max_terminated_length": 1.9, |
| "completions/mean_length": 1.4125, |
| "completions/mean_terminated_length": 1.4125, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.1759153566934401, |
| "epoch": 0.3543307086614173, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.920500000000001e-06, |
| "loss": -0.09908871650695801, |
| "num_tokens": 634439.0, |
| "reward": 1.0429795563220978, |
| "reward_std": 0.013900509846280329, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.004248046875, |
| "rewards/reward_length/std": 0.004071233188733458, |
| "rewards/reward_semantic/mean": 0.14397321436554195, |
| "rewards/reward_semantic/std": 0.046875466406345365, |
| "step": 2160, |
| "step_time": 1.5330110972505282 |
| }, |
| { |
| "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": 2.45, |
| "completions/max_terminated_length": 2.45, |
| "completions/mean_length": 1.85, |
| "completions/mean_terminated_length": 1.85, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.2783667841518763, |
| "epoch": 0.35761154855643046, |
| "frac_reward_zero_std": 0.6, |
| "grad_norm": 0.0, |
| "learning_rate": 3.9105e-06, |
| "loss": 0.051303112506866456, |
| "num_tokens": 640079.0, |
| "reward": 1.0466699212789536, |
| "reward_std": 0.01480488276865799, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0041015625, |
| "rewards/reward_length/std": 0.003491846891120076, |
| "rewards/reward_semantic/mean": 0.15625, |
| "rewards/reward_semantic/std": 0.049022963643074034, |
| "step": 2180, |
| "step_time": 1.5538693566009898 |
| }, |
| { |
| "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": 1.65, |
| "completions/max_terminated_length": 1.65, |
| "completions/mean_length": 1.45, |
| "completions/mean_terminated_length": 1.45, |
| "completions/min_length": 1.1, |
| "completions/min_terminated_length": 1.1, |
| "entropy": 0.11134458579908824, |
| "epoch": 0.36089238845144356, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.9005e-06, |
| "loss": -0.02030053585767746, |
| "num_tokens": 645399.0, |
| "reward": 1.054638671875, |
| "reward_std": 0.015008862101240084, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0072265625, |
| "rewards/reward_length/std": 0.004229071293957531, |
| "rewards/reward_semantic/mean": 0.1833333343267441, |
| "rewards/reward_semantic/std": 0.05029634982347488, |
| "step": 2200, |
| "step_time": 1.4331181380496674 |
| }, |
| { |
| "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": 1.25, |
| "completions/max_terminated_length": 1.25, |
| "completions/mean_length": 1.19375, |
| "completions/mean_terminated_length": 1.19375, |
| "completions/min_length": 1.0, |
| "completions/min_terminated_length": 1.0, |
| "entropy": 0.10332108132424764, |
| "epoch": 0.3641732283464567, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.8905e-06, |
| "loss": -0.028479492664337157, |
| "num_tokens": 650942.0, |
| "reward": 1.0243029773235321, |
| "reward_std": 0.01221160739660263, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0014404296875, |
| "rewards/reward_length/std": 0.0007069154875352978, |
| "rewards/reward_semantic/mean": 0.08125, |
| "rewards/reward_semantic/std": 0.0408231720328331, |
| "step": 2220, |
| "step_time": 1.4920091373000104 |
| }, |
| { |
| "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": 1.85, |
| "completions/max_terminated_length": 1.85, |
| "completions/mean_length": 1.7125, |
| "completions/mean_terminated_length": 1.7125, |
| "completions/min_length": 1.65, |
| "completions/min_terminated_length": 1.65, |
| "entropy": 0.11441818393213907, |
| "epoch": 0.3674540682414698, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.8805e-06, |
| "loss": -0.041957962512969973, |
| "num_tokens": 656416.0, |
| "reward": 1.074772948026657, |
| "reward_std": 0.01053754761815071, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.004541015625, |
| "rewards/reward_length/std": 0.0013810679549351335, |
| "rewards/reward_semantic/mean": 0.25, |
| "rewards/reward_semantic/std": 0.03535533845424652, |
| "step": 2240, |
| "step_time": 1.563767152150831 |
| }, |
| { |
| "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": 1.5, |
| "completions/max_terminated_length": 1.5, |
| "completions/mean_length": 1.28125, |
| "completions/mean_terminated_length": 1.28125, |
| "completions/min_length": 1.1, |
| "completions/min_terminated_length": 1.1, |
| "entropy": 0.2240779584128177, |
| "epoch": 0.37073490813648297, |
| "frac_reward_zero_std": 0.85, |
| "grad_norm": 0.0, |
| "learning_rate": 3.8705000000000005e-06, |
| "loss": -0.028519853949546814, |
| "num_tokens": 661837.0, |
| "reward": 1.0437194794416427, |
| "reward_std": 0.010648868084535934, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0068603515625, |
| "rewards/reward_length/std": 0.005185322277247906, |
| "rewards/reward_semantic/mean": 0.146875, |
| "rewards/reward_semantic/std": 0.036084233224391936, |
| "step": 2260, |
| "step_time": 1.5189277010996194 |
| }, |
| { |
| "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": 1.9, |
| "completions/max_terminated_length": 1.9, |
| "completions/mean_length": 1.45625, |
| "completions/mean_terminated_length": 1.45625, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.27324234248517315, |
| "epoch": 0.37401574803149606, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 56.5, |
| "learning_rate": 3.8605000000000006e-06, |
| "loss": 0.0488494873046875, |
| "num_tokens": 667166.0, |
| "reward": 1.044693610072136, |
| "reward_std": 0.014641210397530812, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0061279296875, |
| "rewards/reward_length/std": 0.004734251950867474, |
| "rewards/reward_semantic/mean": 0.15, |
| "rewards/reward_semantic/std": 0.0487184002995491, |
| "step": 2280, |
| "step_time": 1.5014171188499859 |
| }, |
| { |
| "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": 1.75, |
| "completions/max_terminated_length": 1.75, |
| "completions/mean_length": 1.375, |
| "completions/mean_terminated_length": 1.375, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.1271202877673204, |
| "epoch": 0.37729658792650916, |
| "frac_reward_zero_std": 0.85, |
| "grad_norm": 0.0, |
| "learning_rate": 3.850500000000001e-06, |
| "loss": 0.06897670030593872, |
| "num_tokens": 672538.0, |
| "reward": 1.0148681670427322, |
| "reward_std": 0.00010973085154546425, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.00263671875, |
| "rewards/reward_length/std": 0.002194640040397644, |
| "rewards/reward_semantic/mean": 0.05, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 2300, |
| "step_time": 1.6312724257998525 |
| }, |
| { |
| "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": 2.0, |
| "completions/max_terminated_length": 2.0, |
| "completions/mean_length": 1.85625, |
| "completions/mean_terminated_length": 1.85625, |
| "completions/min_length": 1.7, |
| "completions/min_terminated_length": 1.7, |
| "entropy": 0.13609613068401813, |
| "epoch": 0.3805774278215223, |
| "frac_reward_zero_std": 0.85, |
| "grad_norm": 64.0, |
| "learning_rate": 3.840500000000001e-06, |
| "loss": -0.0012609858065843582, |
| "num_tokens": 678035.0, |
| "reward": 1.0820971757173539, |
| "reward_std": 0.010550401956425048, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.008056640625, |
| "rewards/reward_length/std": 0.002859796490520239, |
| "rewards/reward_semantic/mean": 0.275, |
| "rewards/reward_semantic/std": 0.03535533845424652, |
| "step": 2320, |
| "step_time": 1.3917430368497663 |
| }, |
| { |
| "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": 1.55, |
| "completions/max_terminated_length": 1.55, |
| "completions/mean_length": 1.5375, |
| "completions/mean_terminated_length": 1.5375, |
| "completions/min_length": 1.45, |
| "completions/min_terminated_length": 1.45, |
| "entropy": 0.08209456286567729, |
| "epoch": 0.3838582677165354, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.8305e-06, |
| "loss": -0.016475823521614075, |
| "num_tokens": 683401.0, |
| "reward": 1.0644750893115997, |
| "reward_std": 0.007920423522591591, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.004248046875, |
| "rewards/reward_length/std": 0.0006905340123921633, |
| "rewards/reward_semantic/mean": 0.215625, |
| "rewards/reward_semantic/std": 0.02651650384068489, |
| "step": 2340, |
| "step_time": 1.4020877050495073 |
| }, |
| { |
| "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": 1.15, |
| "completions/max_terminated_length": 1.15, |
| "completions/mean_length": 1.05625, |
| "completions/mean_terminated_length": 1.05625, |
| "completions/min_length": 1.0, |
| "completions/min_terminated_length": 1.0, |
| "entropy": 0.13131917668215465, |
| "epoch": 0.38713910761154857, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.8205e-06, |
| "loss": 0.03290719091892243, |
| "num_tokens": 688602.0, |
| "reward": 0.9999377459287644, |
| "reward_std": 8.158853597706184e-05, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0012451171875, |
| "rewards/reward_length/std": 0.0016317619476467371, |
| "rewards/reward_semantic/mean": 0.0, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 2360, |
| "step_time": 1.4038675237999996 |
| }, |
| { |
| "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": 1.7, |
| "completions/max_terminated_length": 1.7, |
| "completions/mean_length": 1.49375, |
| "completions/mean_terminated_length": 1.49375, |
| "completions/min_length": 1.25, |
| "completions/min_terminated_length": 1.25, |
| "entropy": 0.1492565186897991, |
| "epoch": 0.39041994750656167, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 40.25, |
| "learning_rate": 3.8105e-06, |
| "loss": 0.020195800065994262, |
| "num_tokens": 693929.0, |
| "reward": 1.0560571283102036, |
| "reward_std": 0.010654939358937554, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.003857421875, |
| "rewards/reward_length/std": 0.0022097086533904076, |
| "rewards/reward_semantic/mean": 0.1875, |
| "rewards/reward_semantic/std": 0.03535533845424652, |
| "step": 2380, |
| "step_time": 1.463225544049419 |
| }, |
| { |
| "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": 1.95, |
| "completions/max_terminated_length": 1.95, |
| "completions/mean_length": 1.69375, |
| "completions/mean_terminated_length": 1.69375, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.17988212132040643, |
| "epoch": 0.3937007874015748, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.8005000000000003e-06, |
| "loss": -0.07997376918792724, |
| "num_tokens": 699448.0, |
| "reward": 1.0762988328933716, |
| "reward_std": 0.029977351016714237, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0115234375, |
| "rewards/reward_length/std": 0.005801353743299842, |
| "rewards/reward_semantic/mean": 0.25625, |
| "rewards/reward_semantic/std": 0.10064117759466171, |
| "step": 2400, |
| "step_time": 1.4135737273500126 |
| }, |
| { |
| "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": 1.75, |
| "completions/max_terminated_length": 1.75, |
| "completions/mean_length": 1.46875, |
| "completions/mean_terminated_length": 1.46875, |
| "completions/min_length": 1.25, |
| "completions/min_terminated_length": 1.25, |
| "entropy": 0.31917830523161683, |
| "epoch": 0.3969816272965879, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7905000000000004e-06, |
| "loss": 0.03466253280639649, |
| "num_tokens": 704851.0, |
| "reward": 1.039982905983925, |
| "reward_std": 0.010567388724302873, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.006591796875, |
| "rewards/reward_length/std": 0.004751381115056574, |
| "rewards/reward_semantic/mean": 0.134375, |
| "rewards/reward_semantic/std": 0.034716295450925826, |
| "step": 2420, |
| "step_time": 1.4708769475500958 |
| }, |
| { |
| "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": 2.2, |
| "completions/max_terminated_length": 2.2, |
| "completions/mean_length": 1.725, |
| "completions/mean_terminated_length": 1.725, |
| "completions/min_length": 1.6, |
| "completions/min_terminated_length": 1.6, |
| "entropy": 0.21973865428371936, |
| "epoch": 0.400262467191601, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7805000000000005e-06, |
| "loss": -0.04793748557567597, |
| "num_tokens": 710143.0, |
| "reward": 1.039420160651207, |
| "reward_std": 0.011744571132294369, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0115966796875, |
| "rewards/reward_length/std": 0.0046345341950654985, |
| "rewards/reward_semantic/mean": 0.13333333432674407, |
| "rewards/reward_semantic/std": 0.03922454360872507, |
| "step": 2440, |
| "step_time": 1.4297261677495954 |
| }, |
| { |
| "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": 2.3, |
| "completions/max_terminated_length": 2.3, |
| "completions/mean_length": 2.10625, |
| "completions/mean_terminated_length": 2.10625, |
| "completions/min_length": 2.05, |
| "completions/min_terminated_length": 2.05, |
| "entropy": 0.15823856241768225, |
| "epoch": 0.4035433070866142, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7705e-06, |
| "loss": -0.02957601547241211, |
| "num_tokens": 715576.0, |
| "reward": 1.0840722680091859, |
| "reward_std": 0.005249162093969062, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0060546875, |
| "rewards/reward_length/std": 0.0020936472807079553, |
| "rewards/reward_semantic/mean": 0.28125, |
| "rewards/reward_semantic/std": 0.01767766922712326, |
| "step": 2460, |
| "step_time": 1.5116692341002818 |
| }, |
| { |
| "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": 2.05, |
| "completions/max_terminated_length": 2.05, |
| "completions/mean_length": 1.55, |
| "completions/mean_terminated_length": 1.55, |
| "completions/min_length": 1.05, |
| "completions/min_terminated_length": 1.05, |
| "entropy": 0.289413814983709, |
| "epoch": 0.4068241469816273, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 35.5, |
| "learning_rate": 3.7605000000000002e-06, |
| "loss": 0.06704224348068237, |
| "num_tokens": 720880.0, |
| "reward": 1.0202246069908143, |
| "reward_std": 0.006841066660126671, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0080078125, |
| "rewards/reward_length/std": 0.0059657159727066755, |
| "rewards/reward_semantic/mean": 0.06875, |
| "rewards/reward_semantic/std": 0.022160132229328156, |
| "step": 2480, |
| "step_time": 1.5389296520498648 |
| }, |
| { |
| "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": 2.85, |
| "completions/max_terminated_length": 2.85, |
| "completions/mean_length": 2.1875, |
| "completions/mean_terminated_length": 2.1875, |
| "completions/min_length": 1.8, |
| "completions/min_terminated_length": 1.8, |
| "entropy": 0.18050932437181472, |
| "epoch": 0.41010498687664043, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7505000000000003e-06, |
| "loss": -0.027880549430847168, |
| "num_tokens": 726318.0, |
| "reward": 1.109132081270218, |
| "reward_std": 0.018227603723062202, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0142333984375, |
| "rewards/reward_length/std": 0.007133095117751509, |
| "rewards/reward_semantic/mean": 0.3661458343267441, |
| "rewards/reward_semantic/std": 0.0611537054181099, |
| "step": 2500, |
| "step_time": 1.7098993262005024 |
| }, |
| { |
| "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": 3.3, |
| "completions/max_terminated_length": 3.3, |
| "completions/mean_length": 2.6, |
| "completions/mean_terminated_length": 2.6, |
| "completions/min_length": 2.35, |
| "completions/min_terminated_length": 2.35, |
| "entropy": 0.2633174742150004, |
| "epoch": 0.41338582677165353, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7405000000000004e-06, |
| "loss": -0.05326451659202576, |
| "num_tokens": 731942.0, |
| "reward": 1.1188932448625564, |
| "reward_std": 0.014252605962974485, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.01796875, |
| "rewards/reward_length/std": 0.006979469233192504, |
| "rewards/reward_semantic/mean": 0.399305556062609, |
| "rewards/reward_semantic/std": 0.04833216369152069, |
| "step": 2520, |
| "step_time": 1.5303475668495594 |
| }, |
| { |
| "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": 1.85, |
| "completions/max_terminated_length": 1.85, |
| "completions/mean_length": 1.59375, |
| "completions/mean_terminated_length": 1.59375, |
| "completions/min_length": 1.45, |
| "completions/min_terminated_length": 1.45, |
| "entropy": 0.18814916715491564, |
| "epoch": 0.4166666666666667, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7305e-06, |
| "loss": 0.0461560845375061, |
| "num_tokens": 737493.0, |
| "reward": 1.0595434546470641, |
| "reward_std": 0.008125243066751863, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.009130859375, |
| "rewards/reward_length/std": 0.0031921896385028957, |
| "rewards/reward_semantic/mean": 0.2, |
| "rewards/reward_semantic/std": 0.026726123690605164, |
| "step": 2540, |
| "step_time": 1.4838354332499875 |
| }, |
| { |
| "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": 2.0, |
| "completions/max_terminated_length": 2.0, |
| "completions/mean_length": 1.25, |
| "completions/mean_terminated_length": 1.25, |
| "completions/min_length": 1.1, |
| "completions/min_terminated_length": 1.1, |
| "entropy": 0.3459064724011114, |
| "epoch": 0.4199475065616798, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7205e-06, |
| "loss": 0.09954692721366883, |
| "num_tokens": 742725.0, |
| "reward": 1.030424803495407, |
| "reward_std": 0.0018953357473947107, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.00400390625, |
| "rewards/reward_length/std": 0.004761095647700131, |
| "rewards/reward_semantic/mean": 0.1020833333954215, |
| "rewards/reward_semantic/std": 0.00589255727827549, |
| "step": 2560, |
| "step_time": 1.462830059549742 |
| }, |
| { |
| "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": 2.4, |
| "completions/max_terminated_length": 2.4, |
| "completions/mean_length": 1.90625, |
| "completions/mean_terminated_length": 1.90625, |
| "completions/min_length": 1.6, |
| "completions/min_terminated_length": 1.6, |
| "entropy": 0.21407992226741043, |
| "epoch": 0.42322834645669294, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 68.0, |
| "learning_rate": 3.7105000000000003e-06, |
| "loss": 0.07427935600280762, |
| "num_tokens": 748158.0, |
| "reward": 1.0569039553403854, |
| "reward_std": 0.0054210009300732056, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.008349609375, |
| "rewards/reward_length/std": 0.0038837586529552937, |
| "rewards/reward_semantic/mean": 0.19107142686843873, |
| "rewards/reward_semantic/std": 0.01821078509092331, |
| "step": 2580, |
| "step_time": 1.4073684682496606 |
| }, |
| { |
| "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": 1.5, |
| "completions/max_terminated_length": 1.5, |
| "completions/mean_length": 1.475, |
| "completions/mean_terminated_length": 1.475, |
| "completions/min_length": 1.3, |
| "completions/min_terminated_length": 1.3, |
| "entropy": 0.164965115715313, |
| "epoch": 0.42650918635170604, |
| "frac_reward_zero_std": 0.95, |
| "grad_norm": 0.0, |
| "learning_rate": 3.7005000000000004e-06, |
| "loss": -0.01373627632856369, |
| "num_tokens": 753554.0, |
| "reward": 1.057954102754593, |
| "reward_std": 0.005289490893483162, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.00341796875, |
| "rewards/reward_length/std": 0.00027621358167380096, |
| "rewards/reward_semantic/mean": 0.19375, |
| "rewards/reward_semantic/std": 0.01767766922712326, |
| "step": 2600, |
| "step_time": 1.5095879755494024 |
| }, |
| { |
| "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": 1.75, |
| "completions/max_terminated_length": 1.75, |
| "completions/mean_length": 1.4, |
| "completions/mean_terminated_length": 1.4, |
| "completions/min_length": 1.2, |
| "completions/min_terminated_length": 1.2, |
| "entropy": 0.1770161235574051, |
| "epoch": 0.42979002624671914, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 53.0, |
| "learning_rate": 3.6905e-06, |
| "loss": 0.07016395330429077, |
| "num_tokens": 758970.0, |
| "reward": 1.0241906732320785, |
| "reward_std": 0.00791277912503574, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0036865234375, |
| "rewards/reward_length/std": 0.0040017103776335715, |
| "rewards/reward_semantic/mean": 0.08125, |
| "rewards/reward_semantic/std": 0.025877460837364197, |
| "step": 2620, |
| "step_time": 1.456911081150156 |
| }, |
| { |
| "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": 1.95, |
| "completions/max_terminated_length": 1.95, |
| "completions/mean_length": 1.7375, |
| "completions/mean_terminated_length": 1.7375, |
| "completions/min_length": 1.5, |
| "completions/min_terminated_length": 1.5, |
| "entropy": 0.09709408191993134, |
| "epoch": 0.4330708661417323, |
| "frac_reward_zero_std": 0.85, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6805e-06, |
| "loss": -0.04094939231872559, |
| "num_tokens": 764760.0, |
| "reward": 1.086412352323532, |
| "reward_std": 0.014678525365889072, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0092529296875, |
| "rewards/reward_length/std": 0.003191290353424847, |
| "rewards/reward_semantic/mean": 0.2895833343267441, |
| "rewards/reward_semantic/std": 0.04944768697023392, |
| "step": 2640, |
| "step_time": 1.5366251467001348 |
| }, |
| { |
| "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": 2.25, |
| "completions/max_terminated_length": 2.25, |
| "completions/mean_length": 1.7875, |
| "completions/mean_terminated_length": 1.7875, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.23105483847903088, |
| "epoch": 0.4363517060367454, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6705000000000003e-06, |
| "loss": -0.030925416946411134, |
| "num_tokens": 770222.0, |
| "reward": 1.0502401351928712, |
| "reward_std": 0.013737828232115135, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.009326171875, |
| "rewards/reward_length/std": 0.006337474985048175, |
| "rewards/reward_semantic/mean": 0.1690214669331908, |
| "rewards/reward_semantic/std": 0.046560603007674214, |
| "step": 2660, |
| "step_time": 1.4115802160996282 |
| }, |
| { |
| "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": 2.05, |
| "completions/max_terminated_length": 2.05, |
| "completions/mean_length": 1.88125, |
| "completions/mean_terminated_length": 1.88125, |
| "completions/min_length": 1.7, |
| "completions/min_terminated_length": 1.7, |
| "entropy": 0.0847354463650845, |
| "epoch": 0.43963254593175854, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6605000000000004e-06, |
| "loss": -0.04883143007755279, |
| "num_tokens": 775739.0, |
| "reward": 1.08895263671875, |
| "reward_std": 0.010505473986268044, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.008447265625, |
| "rewards/reward_length/std": 0.0028636682778596877, |
| "rewards/reward_semantic/mean": 0.29791666865348815, |
| "rewards/reward_semantic/std": 0.03549508601427078, |
| "step": 2680, |
| "step_time": 1.5979070248503375 |
| }, |
| { |
| "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": 2.45, |
| "completions/max_terminated_length": 2.45, |
| "completions/mean_length": 2.09375, |
| "completions/mean_terminated_length": 2.09375, |
| "completions/min_length": 1.75, |
| "completions/min_terminated_length": 1.75, |
| "entropy": 0.1793067864375189, |
| "epoch": 0.44291338582677164, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 51.0, |
| "learning_rate": 3.6505e-06, |
| "loss": -0.05965709686279297, |
| "num_tokens": 781210.0, |
| "reward": 1.072282800078392, |
| "reward_std": 0.017886035761330278, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0110107421875, |
| "rewards/reward_length/std": 0.005271141580305993, |
| "rewards/reward_semantic/mean": 0.24277777783572674, |
| "rewards/reward_semantic/std": 0.06017936840653419, |
| "step": 2700, |
| "step_time": 1.5664898869004902 |
| }, |
| { |
| "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": 1.9, |
| "completions/max_terminated_length": 1.9, |
| "completions/mean_length": 1.59375, |
| "completions/mean_terminated_length": 1.59375, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.31023325126734563, |
| "epoch": 0.4461942257217848, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6405e-06, |
| "loss": 0.029807302355766296, |
| "num_tokens": 786593.0, |
| "reward": 1.0378039538860322, |
| "reward_std": 0.009880722660454921, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0064208984375, |
| "rewards/reward_length/std": 0.005932193575426936, |
| "rewards/reward_semantic/mean": 0.1270833333954215, |
| "rewards/reward_semantic/std": 0.03261868096888065, |
| "step": 2720, |
| "step_time": 1.547453101550309 |
| }, |
| { |
| "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": 1.75, |
| "completions/max_terminated_length": 1.75, |
| "completions/mean_length": 1.675, |
| "completions/mean_terminated_length": 1.675, |
| "completions/min_length": 1.6, |
| "completions/min_terminated_length": 1.6, |
| "entropy": 0.11798639885819284, |
| "epoch": 0.4494750656167979, |
| "frac_reward_zero_std": 0.95, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6305000000000002e-06, |
| "loss": -0.027998325228691102, |
| "num_tokens": 792021.0, |
| "reward": 1.0469231307506561, |
| "reward_std": 0.002181190438568592, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.00439453125, |
| "rewards/reward_length/std": 0.002192377485334873, |
| "rewards/reward_semantic/mean": 0.15714285746216775, |
| "rewards/reward_semantic/std": 0.007636035978794098, |
| "step": 2740, |
| "step_time": 1.5446192357003383 |
| }, |
| { |
| "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": 1.95, |
| "completions/max_terminated_length": 1.95, |
| "completions/mean_length": 1.65, |
| "completions/mean_terminated_length": 1.65, |
| "completions/min_length": 1.45, |
| "completions/min_terminated_length": 1.45, |
| "entropy": 0.21348486854840304, |
| "epoch": 0.452755905511811, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6205000000000003e-06, |
| "loss": 0.004397254809737206, |
| "num_tokens": 797573.0, |
| "reward": 1.0833984375, |
| "reward_std": 0.025374090240802616, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.00703125, |
| "rewards/reward_length/std": 0.004363904730416835, |
| "rewards/reward_semantic/mean": 0.27916666865348816, |
| "rewards/reward_semantic/std": 0.08451805114746094, |
| "step": 2760, |
| "step_time": 1.5169254192499466 |
| }, |
| { |
| "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": 2.35, |
| "completions/max_terminated_length": 2.35, |
| "completions/mean_length": 2.11875, |
| "completions/mean_terminated_length": 2.11875, |
| "completions/min_length": 1.9, |
| "completions/min_terminated_length": 1.9, |
| "entropy": 0.11082232519038371, |
| "epoch": 0.45603674540682415, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6105e-06, |
| "loss": 0.03271139860153198, |
| "num_tokens": 803048.0, |
| "reward": 1.1008435100317002, |
| "reward_std": 0.006947353344003204, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0081298828125, |
| "rewards/reward_length/std": 0.0022439215099439025, |
| "rewards/reward_semantic/mean": 0.3375, |
| "rewards/reward_semantic/std": 0.023145502805709837, |
| "step": 2780, |
| "step_time": 1.6986804411993943 |
| }, |
| { |
| "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": 2.2, |
| "completions/max_terminated_length": 2.2, |
| "completions/mean_length": 1.98125, |
| "completions/mean_terminated_length": 1.98125, |
| "completions/min_length": 1.95, |
| "completions/min_terminated_length": 1.95, |
| "entropy": 0.1825924801894871, |
| "epoch": 0.45931758530183725, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.6005e-06, |
| "loss": -0.03052682280540466, |
| "num_tokens": 808357.0, |
| "reward": 1.120726317167282, |
| "reward_std": 0.0034699331779847855, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0104736328125, |
| "rewards/reward_length/std": 0.0020025484496727585, |
| "rewards/reward_semantic/mean": 0.404166666790843, |
| "rewards/reward_semantic/std": 0.01178511455655098, |
| "step": 2800, |
| "step_time": 1.7920531756996751 |
| }, |
| { |
| "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": 2.25, |
| "completions/max_terminated_length": 2.25, |
| "completions/mean_length": 1.8375, |
| "completions/mean_terminated_length": 1.8375, |
| "completions/min_length": 1.5, |
| "completions/min_terminated_length": 1.5, |
| "entropy": 0.2862564872033545, |
| "epoch": 0.4625984251968504, |
| "frac_reward_zero_std": 0.7, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5905e-06, |
| "loss": -0.034829136729240415, |
| "num_tokens": 814131.0, |
| "reward": 1.064993718266487, |
| "reward_std": 0.015606581326574087, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.01083984375, |
| "rewards/reward_length/std": 0.007136135920882225, |
| "rewards/reward_semantic/mean": 0.21845238134264947, |
| "rewards/reward_semantic/std": 0.05225111655890942, |
| "step": 2820, |
| "step_time": 1.4031961227998182 |
| }, |
| { |
| "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": 1.45, |
| "completions/max_terminated_length": 1.45, |
| "completions/mean_length": 1.2625, |
| "completions/mean_terminated_length": 1.2625, |
| "completions/min_length": 1.1, |
| "completions/min_terminated_length": 1.1, |
| "entropy": 0.17255667096178512, |
| "epoch": 0.4658792650918635, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5805000000000003e-06, |
| "loss": -0.047430571913719174, |
| "num_tokens": 819389.0, |
| "reward": 1.0400073289871217, |
| "reward_std": 0.0159482779155951, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.006103515625, |
| "rewards/reward_length/std": 0.0030561807565391065, |
| "rewards/reward_semantic/mean": 0.134375, |
| "rewards/reward_semantic/std": 0.053519590944051745, |
| "step": 2840, |
| "step_time": 1.4340180885501468 |
| }, |
| { |
| "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": 1.65, |
| "completions/max_terminated_length": 1.65, |
| "completions/mean_length": 1.4375, |
| "completions/mean_terminated_length": 1.4375, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.15011761501446017, |
| "epoch": 0.46916010498687666, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5705e-06, |
| "loss": 0.014107689261436462, |
| "num_tokens": 824803.0, |
| "reward": 1.0317797899246215, |
| "reward_std": 0.0034630337380804122, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.001904296875, |
| "rewards/reward_length/std": 0.0007282741600647569, |
| "rewards/reward_semantic/mean": 0.10625, |
| "rewards/reward_semantic/std": 0.011572751402854919, |
| "step": 2860, |
| "step_time": 1.641624565299935 |
| }, |
| { |
| "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": 1.7, |
| "completions/max_terminated_length": 1.7, |
| "completions/mean_length": 1.325, |
| "completions/mean_terminated_length": 1.325, |
| "completions/min_length": 1.25, |
| "completions/min_terminated_length": 1.25, |
| "entropy": 0.2041399403475225, |
| "epoch": 0.47244094488188976, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5605e-06, |
| "loss": 0.076927649974823, |
| "num_tokens": 830351.0, |
| "reward": 1.0223352044820786, |
| "reward_std": 0.00015387315361294895, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0032958984375, |
| "rewards/reward_length/std": 0.0030774822225794196, |
| "rewards/reward_semantic/mean": 0.075, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 2880, |
| "step_time": 1.3945468248992257 |
| }, |
| { |
| "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": 1.85, |
| "completions/max_terminated_length": 1.85, |
| "completions/mean_length": 1.6, |
| "completions/mean_terminated_length": 1.6, |
| "completions/min_length": 1.4, |
| "completions/min_terminated_length": 1.4, |
| "entropy": 0.20412995948863683, |
| "epoch": 0.4757217847769029, |
| "frac_reward_zero_std": 0.75, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5505e-06, |
| "loss": 0.029290300607681275, |
| "num_tokens": 835583.0, |
| "reward": 1.0744787573814392, |
| "reward_std": 0.008839267375878989, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0104248046875, |
| "rewards/reward_length/std": 0.0028774956008419394, |
| "rewards/reward_semantic/mean": 0.25, |
| "rewards/reward_semantic/std": 0.02925042062997818, |
| "step": 2900, |
| "step_time": 1.4834619632991235 |
| }, |
| { |
| "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": 1.5, |
| "completions/max_terminated_length": 1.5, |
| "completions/mean_length": 1.3875, |
| "completions/mean_terminated_length": 1.3875, |
| "completions/min_length": 1.3, |
| "completions/min_terminated_length": 1.3, |
| "entropy": 0.214178694797738, |
| "epoch": 0.479002624671916, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 37.5, |
| "learning_rate": 3.5405000000000002e-06, |
| "loss": 0.016094771027565003, |
| "num_tokens": 841037.0, |
| "reward": 1.0241186529397965, |
| "reward_std": 0.005358124604390469, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.005126953125, |
| "rewards/reward_length/std": 0.0019251078832894565, |
| "rewards/reward_semantic/mean": 0.08125, |
| "rewards/reward_semantic/std": 0.01767766922712326, |
| "step": 2920, |
| "step_time": 1.4553790288999153 |
| }, |
| { |
| "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": 2.3, |
| "completions/max_terminated_length": 2.3, |
| "completions/mean_length": 2.25625, |
| "completions/mean_terminated_length": 2.25625, |
| "completions/min_length": 2.25, |
| "completions/min_terminated_length": 2.25, |
| "entropy": 0.10389214321330656, |
| "epoch": 0.4822834645669291, |
| "frac_reward_zero_std": 0.95, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5305e-06, |
| "loss": 0.010661421716213227, |
| "num_tokens": 846790.0, |
| "reward": 1.1047106981277466, |
| "reward_std": 1.726334885461256e-05, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0057861328125, |
| "rewards/reward_length/std": 0.00034526700619608164, |
| "rewards/reward_semantic/mean": 0.35, |
| "rewards/reward_semantic/std": 0.0, |
| "step": 2940, |
| "step_time": 1.5068046017999221 |
| }, |
| { |
| "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": 1.7, |
| "completions/max_terminated_length": 1.7, |
| "completions/mean_length": 1.39375, |
| "completions/mean_terminated_length": 1.39375, |
| "completions/min_length": 1.25, |
| "completions/min_terminated_length": 1.25, |
| "entropy": 0.17968989024229814, |
| "epoch": 0.48556430446194226, |
| "frac_reward_zero_std": 0.8, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5205e-06, |
| "loss": -0.024578049778938293, |
| "num_tokens": 852165.0, |
| "reward": 1.0689611703157424, |
| "reward_std": 0.012143709098745603, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0082763671875, |
| "rewards/reward_length/std": 0.004592829616740346, |
| "rewards/reward_semantic/mean": 0.23125, |
| "rewards/reward_semantic/std": 0.0408231720328331, |
| "step": 2960, |
| "step_time": 1.367187066900442 |
| }, |
| { |
| "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": 2.05, |
| "completions/max_terminated_length": 2.05, |
| "completions/mean_length": 1.9875, |
| "completions/mean_terminated_length": 1.9875, |
| "completions/min_length": 1.8, |
| "completions/min_terminated_length": 1.8, |
| "entropy": 0.11668806873640278, |
| "epoch": 0.48884514435695536, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5105e-06, |
| "loss": -0.004691676050424576, |
| "num_tokens": 857739.0, |
| "reward": 1.085783690214157, |
| "reward_std": 0.006893926800694316, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.009326171875, |
| "rewards/reward_length/std": 0.0022603030316531657, |
| "rewards/reward_semantic/mean": 0.2875, |
| "rewards/reward_semantic/std": 0.023145502805709837, |
| "step": 2980, |
| "step_time": 1.5960781566998776 |
| }, |
| { |
| "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": 1.8, |
| "completions/max_terminated_length": 1.8, |
| "completions/mean_length": 1.66875, |
| "completions/mean_terminated_length": 1.66875, |
| "completions/min_length": 1.55, |
| "completions/min_terminated_length": 1.55, |
| "entropy": 0.13652693098702, |
| "epoch": 0.4921259842519685, |
| "frac_reward_zero_std": 0.9, |
| "grad_norm": 0.0, |
| "learning_rate": 3.5005e-06, |
| "loss": -0.047617864608764646, |
| "num_tokens": 863238.0, |
| "reward": 1.0803625404834747, |
| "reward_std": 0.015734923630952836, |
| "rewards/reward_grounded/mean": 1.0, |
| "rewards/reward_grounded/std": 0.0, |
| "rewards/reward_length/mean": -0.0052490234375, |
| "rewards/reward_length/std": 0.0009230148047208786, |
| "rewards/reward_semantic/mean": 0.26875, |
| "rewards/reward_semantic/std": 0.05260358452796936, |
| "step": 3000, |
| "step_time": 1.6425139239494455 |
| } |
| ], |
| "logging_steps": 20, |
| "max_steps": 10000, |
| "num_input_tokens_seen": 863238, |
| "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 |
| } |
|
|
