{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 2000.0,
  "global_step": 3595,
  "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": 0.140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 397.359375,
      "completions/min_length": 244.0,
      "epoch": 0.0002781641168289291,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7349766492843628,
      "kl": 0.0,
      "learning_rate": 5.555555555555555e-09,
      "loss": 6.970367394387722e-09,
      "reward": 1.9015536308288574,
      "reward_std": 0.8292676210403442,
      "rewards/IngredientFormatReward/mean": 0.748828113079071,
      "rewards/IngredientFormatReward/std": 0.4224153757095337,
      "rewards/IngredientMatchReward/mean": 0.3902685046195984,
      "rewards/IngredientMatchReward/std": 0.28992533683776855,
      "rewards/IngredientQuantityMatchReward/mean": 0.5593319535255432,
      "rewards/IngredientQuantityMatchReward/std": 0.4461570084095001,
      "rewards/TotalKcalExactMatchReward/mean": 0.203125,
      "rewards/TotalKcalExactMatchReward/std": 0.40390563011169434,
      "step": 1
    },
    {
      "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": 0.09375,
      "completions/max_length": 513.0,
      "completions/mean_length": 403.083984375,
      "completions/min_length": 247.75,
      "epoch": 0.0013908205841446453,
      "frac_reward_zero_std": 0.015625,
      "grad_norm": 0.7494612336158752,
      "kl": 0.00029059583584967186,
      "learning_rate": 2.7777777777777774e-08,
      "loss": 1.1643016478046775e-05,
      "reward": 2.084854483604431,
      "reward_std": 0.7374771535396576,
      "rewards/IngredientFormatReward/mean": 0.8349934965372086,
      "rewards/IngredientFormatReward/std": 0.35871507972478867,
      "rewards/IngredientMatchReward/mean": 0.45674292743206024,
      "rewards/IngredientMatchReward/std": 0.3166300803422928,
      "rewards/IngredientQuantityMatchReward/mean": 0.5743680000305176,
      "rewards/IngredientQuantityMatchReward/std": 0.450081929564476,
      "rewards/TotalKcalExactMatchReward/mean": 0.21875,
      "rewards/TotalKcalExactMatchReward/std": 0.41126810014247894,
      "step": 5
    },
    {
      "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": 0.0625,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.8078125,
      "completions/min_length": 222.2,
      "epoch": 0.0027816411682892906,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7679515480995178,
      "kl": 0.00040537232052884065,
      "learning_rate": 5.555555555555555e-08,
      "loss": 1.6196584329009057e-05,
      "reward": 2.1525042057037354,
      "reward_std": 0.6553369760513306,
      "rewards/IngredientFormatReward/mean": 0.85098956823349,
      "rewards/IngredientFormatReward/std": 0.3357125997543335,
      "rewards/IngredientMatchReward/mean": 0.46298115253448485,
      "rewards/IngredientMatchReward/std": 0.2870068609714508,
      "rewards/IngredientQuantityMatchReward/mean": 0.5979084730148315,
      "rewards/IngredientQuantityMatchReward/std": 0.43828884363174436,
      "rewards/TotalKcalExactMatchReward/mean": 0.240625,
      "rewards/TotalKcalExactMatchReward/std": 0.42608442306518557,
      "step": 10
    },
    {
      "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": 0.08125,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.2984375,
      "completions/min_length": 217.4,
      "epoch": 0.004172461752433936,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7247325778007507,
      "kl": 0.00041362347001268064,
      "learning_rate": 8.333333333333333e-08,
      "loss": 1.6542989760637285e-05,
      "reward": 1.9850705146789551,
      "reward_std": 0.6654590249061585,
      "rewards/IngredientFormatReward/mean": 0.8356510400772095,
      "rewards/IngredientFormatReward/std": 0.3531181126832962,
      "rewards/IngredientMatchReward/mean": 0.460903400182724,
      "rewards/IngredientMatchReward/std": 0.30696226954460143,
      "rewards/IngredientQuantityMatchReward/mean": 0.4963285565376282,
      "rewards/IngredientQuantityMatchReward/std": 0.43468194007873534,
      "rewards/TotalKcalExactMatchReward/mean": 0.1921875,
      "rewards/TotalKcalExactMatchReward/std": 0.39231924414634706,
      "step": 15
    },
    {
      "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": 0.0671875,
      "completions/max_length": 513.0,
      "completions/mean_length": 402.853125,
      "completions/min_length": 269.8,
      "epoch": 0.005563282336578581,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.725919246673584,
      "kl": 0.00042192289620288646,
      "learning_rate": 1.111111111111111e-07,
      "loss": 1.6887777019292115e-05,
      "reward": 2.064033532142639,
      "reward_std": 0.7137743353843689,
      "rewards/IngredientFormatReward/mean": 0.8258333444595337,
      "rewards/IngredientFormatReward/std": 0.36316843032836915,
      "rewards/IngredientMatchReward/mean": 0.43542484641075135,
      "rewards/IngredientMatchReward/std": 0.28290987610816953,
      "rewards/IngredientQuantityMatchReward/mean": 0.5527753114700318,
      "rewards/IngredientQuantityMatchReward/std": 0.4324473083019257,
      "rewards/TotalKcalExactMatchReward/mean": 0.25,
      "rewards/TotalKcalExactMatchReward/std": 0.43261595964431765,
      "step": 20
    },
    {
      "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": 0.078125,
      "completions/max_length": 513.0,
      "completions/mean_length": 403.4921875,
      "completions/min_length": 247.4,
      "epoch": 0.006954102920723227,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7491269707679749,
      "kl": 0.00040630993462400513,
      "learning_rate": 1.3888888888888888e-07,
      "loss": 1.6278773546218873e-05,
      "reward": 2.092119598388672,
      "reward_std": 0.7215837955474853,
      "rewards/IngredientFormatReward/mean": 0.8207626461982727,
      "rewards/IngredientFormatReward/std": 0.3697893679141998,
      "rewards/IngredientMatchReward/mean": 0.4614254832267761,
      "rewards/IngredientMatchReward/std": 0.31788976192474366,
      "rewards/IngredientQuantityMatchReward/mean": 0.5443065166473389,
      "rewards/IngredientQuantityMatchReward/std": 0.4438231110572815,
      "rewards/TotalKcalExactMatchReward/mean": 0.265625,
      "rewards/TotalKcalExactMatchReward/std": 0.4396721661090851,
      "step": 25
    },
    {
      "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": 0.065625,
      "completions/max_length": 513.0,
      "completions/mean_length": 398.553125,
      "completions/min_length": 237.8,
      "epoch": 0.008344923504867872,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7440645694732666,
      "kl": 0.00041069387589232065,
      "learning_rate": 1.6666666666666665e-07,
      "loss": 1.6446597874164583e-05,
      "reward": 2.1207164525985718,
      "reward_std": 0.6690902352333069,
      "rewards/IngredientFormatReward/mean": 0.8505319833755494,
      "rewards/IngredientFormatReward/std": 0.3448954880237579,
      "rewards/IngredientMatchReward/mean": 0.49686943292617797,
      "rewards/IngredientMatchReward/std": 0.3237551271915436,
      "rewards/IngredientQuantityMatchReward/mean": 0.5108150601387024,
      "rewards/IngredientQuantityMatchReward/std": 0.45325875878334043,
      "rewards/TotalKcalExactMatchReward/mean": 0.2625,
      "rewards/TotalKcalExactMatchReward/std": 0.4397946834564209,
      "step": 30
    },
    {
      "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": 0.084375,
      "completions/max_length": 513.0,
      "completions/mean_length": 404.7203125,
      "completions/min_length": 257.4,
      "epoch": 0.009735744089012517,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7201761603355408,
      "kl": 0.00041231723917007913,
      "learning_rate": 1.9444444444444445e-07,
      "loss": 1.6507774125784637e-05,
      "reward": 2.106699252128601,
      "reward_std": 0.7597081899642945,
      "rewards/IngredientFormatReward/mean": 0.8413802146911621,
      "rewards/IngredientFormatReward/std": 0.34706807136535645,
      "rewards/IngredientMatchReward/mean": 0.43627233505249025,
      "rewards/IngredientMatchReward/std": 0.2960783183574677,
      "rewards/IngredientQuantityMatchReward/mean": 0.5540466845035553,
      "rewards/IngredientQuantityMatchReward/std": 0.4453916549682617,
      "rewards/TotalKcalExactMatchReward/mean": 0.275,
      "rewards/TotalKcalExactMatchReward/std": 0.4379255294799805,
      "step": 35
    },
    {
      "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": 0.084375,
      "completions/max_length": 513.0,
      "completions/mean_length": 402.0609375,
      "completions/min_length": 247.4,
      "epoch": 0.011126564673157162,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.74068683385849,
      "kl": 0.00042258204157405996,
      "learning_rate": 2.222222222222222e-07,
      "loss": 1.691014040261507e-05,
      "reward": 2.0093788385391234,
      "reward_std": 0.7664336800575257,
      "rewards/IngredientFormatReward/mean": 0.8254426956176758,
      "rewards/IngredientFormatReward/std": 0.3664437234401703,
      "rewards/IngredientMatchReward/mean": 0.4059288203716278,
      "rewards/IngredientMatchReward/std": 0.28713943660259245,
      "rewards/IngredientQuantityMatchReward/mean": 0.5623823046684265,
      "rewards/IngredientQuantityMatchReward/std": 0.45745145678520205,
      "rewards/TotalKcalExactMatchReward/mean": 0.215625,
      "rewards/TotalKcalExactMatchReward/std": 0.40993704795837405,
      "step": 40
    },
    {
      "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": 0.0734375,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.8671875,
      "completions/min_length": 220.4,
      "epoch": 0.012517385257301807,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7788537740707397,
      "kl": 0.0004178027720627142,
      "learning_rate": 2.5e-07,
      "loss": 1.670618075877428e-05,
      "reward": 1.9563416957855224,
      "reward_std": 0.6972125172615051,
      "rewards/IngredientFormatReward/mean": 0.822872006893158,
      "rewards/IngredientFormatReward/std": 0.3712240636348724,
      "rewards/IngredientMatchReward/mean": 0.44173184037208557,
      "rewards/IngredientMatchReward/std": 0.3116217702627182,
      "rewards/IngredientQuantityMatchReward/mean": 0.4729879081249237,
      "rewards/IngredientQuantityMatchReward/std": 0.45757158994674685,
      "rewards/TotalKcalExactMatchReward/mean": 0.21875,
      "rewards/TotalKcalExactMatchReward/std": 0.4113896369934082,
      "step": 45
    },
    {
      "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": 0.046875,
      "completions/max_length": 513.0,
      "completions/mean_length": 398.4078125,
      "completions/min_length": 218.6,
      "epoch": 0.013908205841446454,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8167436122894287,
      "kl": 0.0004406511514389422,
      "learning_rate": 2.7777777777777776e-07,
      "loss": 1.766979112289846e-05,
      "reward": 2.054394245147705,
      "reward_std": 0.6666451692581177,
      "rewards/IngredientFormatReward/mean": 0.840610122680664,
      "rewards/IngredientFormatReward/std": 0.3461251139640808,
      "rewards/IngredientMatchReward/mean": 0.46625000834465025,
      "rewards/IngredientMatchReward/std": 0.3063514709472656,
      "rewards/IngredientQuantityMatchReward/mean": 0.5053465902805329,
      "rewards/IngredientQuantityMatchReward/std": 0.45414825081825255,
      "rewards/TotalKcalExactMatchReward/mean": 0.2421875,
      "rewards/TotalKcalExactMatchReward/std": 0.4245078921318054,
      "step": 50
    },
    {
      "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": 0.0515625,
      "completions/max_length": 513.0,
      "completions/mean_length": 388.9828125,
      "completions/min_length": 222.0,
      "epoch": 0.015299026425591099,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8138473629951477,
      "kl": 0.00042133941533393224,
      "learning_rate": 3.055555555555556e-07,
      "loss": 1.6886042430996895e-05,
      "reward": 2.1590343952178954,
      "reward_std": 0.6529965162277221,
      "rewards/IngredientFormatReward/mean": 0.8830208301544189,
      "rewards/IngredientFormatReward/std": 0.3005488455295563,
      "rewards/IngredientMatchReward/mean": 0.47440544366836546,
      "rewards/IngredientMatchReward/std": 0.30408360362052916,
      "rewards/IngredientQuantityMatchReward/mean": 0.5719206094741821,
      "rewards/IngredientQuantityMatchReward/std": 0.4585339307785034,
      "rewards/TotalKcalExactMatchReward/mean": 0.2296875,
      "rewards/TotalKcalExactMatchReward/std": 0.41347005367279055,
      "step": 55
    },
    {
      "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": 0.0796875,
      "completions/max_length": 513.0,
      "completions/mean_length": 398.1203125,
      "completions/min_length": 222.4,
      "epoch": 0.016689847009735744,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7178951501846313,
      "kl": 0.0004636022935301298,
      "learning_rate": 3.333333333333333e-07,
      "loss": 1.8570698739495128e-05,
      "reward": 2.014821100234985,
      "reward_std": 0.7525708317756653,
      "rewards/IngredientFormatReward/mean": 0.8299218893051148,
      "rewards/IngredientFormatReward/std": 0.36488322615623475,
      "rewards/IngredientMatchReward/mean": 0.41852368116378785,
      "rewards/IngredientMatchReward/std": 0.2902585655450821,
      "rewards/IngredientQuantityMatchReward/mean": 0.5179380416870117,
      "rewards/IngredientQuantityMatchReward/std": 0.45164860486984254,
      "rewards/TotalKcalExactMatchReward/mean": 0.2484375,
      "rewards/TotalKcalExactMatchReward/std": 0.42951870560646055,
      "step": 60
    },
    {
      "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": 0.046875,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.471875,
      "completions/min_length": 257.8,
      "epoch": 0.01808066759388039,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7024106383323669,
      "kl": 0.00044692111259792,
      "learning_rate": 3.6111111111111107e-07,
      "loss": 1.784815249266103e-05,
      "reward": 2.1446351766586305,
      "reward_std": 0.6606986165046692,
      "rewards/IngredientFormatReward/mean": 0.8456249952316284,
      "rewards/IngredientFormatReward/std": 0.34710876941680907,
      "rewards/IngredientMatchReward/mean": 0.4586080014705658,
      "rewards/IngredientMatchReward/std": 0.2785313993692398,
      "rewards/IngredientQuantityMatchReward/mean": 0.5747771084308624,
      "rewards/IngredientQuantityMatchReward/std": 0.44896321892738345,
      "rewards/TotalKcalExactMatchReward/mean": 0.265625,
      "rewards/TotalKcalExactMatchReward/std": 0.44074881076812744,
      "step": 65
    },
    {
      "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": 0.040625,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.8890625,
      "completions/min_length": 239.4,
      "epoch": 0.019471488178025034,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7436627745628357,
      "kl": 0.0004441993540240219,
      "learning_rate": 3.888888888888889e-07,
      "loss": 1.7779297195374966e-05,
      "reward": 2.118137836456299,
      "reward_std": 0.7043254613876343,
      "rewards/IngredientFormatReward/mean": 0.8741369247436523,
      "rewards/IngredientFormatReward/std": 0.30888078510761263,
      "rewards/IngredientMatchReward/mean": 0.4107558369636536,
      "rewards/IngredientMatchReward/std": 0.28572845458984375,
      "rewards/IngredientQuantityMatchReward/mean": 0.5613700985908509,
      "rewards/IngredientQuantityMatchReward/std": 0.4577866315841675,
      "rewards/TotalKcalExactMatchReward/mean": 0.271875,
      "rewards/TotalKcalExactMatchReward/std": 0.44533634185791016,
      "step": 70
    },
    {
      "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": 0.065625,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.0078125,
      "completions/min_length": 223.4,
      "epoch": 0.02086230876216968,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7756255865097046,
      "kl": 0.0005053242901340127,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 2.017030492424965e-05,
      "reward": 2.087790060043335,
      "reward_std": 0.711814534664154,
      "rewards/IngredientFormatReward/mean": 0.8524739623069764,
      "rewards/IngredientFormatReward/std": 0.3421242475509644,
      "rewards/IngredientMatchReward/mean": 0.46218688488006593,
      "rewards/IngredientMatchReward/std": 0.30402821898460386,
      "rewards/IngredientQuantityMatchReward/mean": 0.5246916890144349,
      "rewards/IngredientQuantityMatchReward/std": 0.4510571718215942,
      "rewards/TotalKcalExactMatchReward/mean": 0.2484375,
      "rewards/TotalKcalExactMatchReward/std": 0.43120152950286866,
      "step": 75
    },
    {
      "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": 0.0765625,
      "completions/max_length": 513.0,
      "completions/mean_length": 400.8671875,
      "completions/min_length": 235.0,
      "epoch": 0.022253129346314324,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.770311713218689,
      "kl": 0.0005777321712230332,
      "learning_rate": 4.444444444444444e-07,
      "loss": 2.3143680300563575e-05,
      "reward": 2.0575687885284424,
      "reward_std": 0.77638738155365,
      "rewards/IngredientFormatReward/mean": 0.8024814009666443,
      "rewards/IngredientFormatReward/std": 0.38416465520858767,
      "rewards/IngredientMatchReward/mean": 0.4254823923110962,
      "rewards/IngredientMatchReward/std": 0.3045427978038788,
      "rewards/IngredientQuantityMatchReward/mean": 0.563979983329773,
      "rewards/IngredientQuantityMatchReward/std": 0.4503722727298737,
      "rewards/TotalKcalExactMatchReward/mean": 0.265625,
      "rewards/TotalKcalExactMatchReward/std": 0.4396756887435913,
      "step": 80
    },
    {
      "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": 0.0796875,
      "completions/max_length": 513.0,
      "completions/mean_length": 408.1390625,
      "completions/min_length": 238.0,
      "epoch": 0.02364394993045897,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7720329761505127,
      "kl": 0.0006533111230965005,
      "learning_rate": 4.722222222222222e-07,
      "loss": 2.6147160679101943e-05,
      "reward": 2.0888258457183837,
      "reward_std": 0.7742647767066956,
      "rewards/IngredientFormatReward/mean": 0.8335677027702332,
      "rewards/IngredientFormatReward/std": 0.36205244064331055,
      "rewards/IngredientMatchReward/mean": 0.47457787990570066,
      "rewards/IngredientMatchReward/std": 0.3164846241474152,
      "rewards/IngredientQuantityMatchReward/mean": 0.4963052570819855,
      "rewards/IngredientQuantityMatchReward/std": 0.4384254515171051,
      "rewards/TotalKcalExactMatchReward/mean": 0.284375,
      "rewards/TotalKcalExactMatchReward/std": 0.45113744139671325,
      "step": 85
    },
    {
      "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": 0.0578125,
      "completions/max_length": 513.0,
      "completions/mean_length": 389.85,
      "completions/min_length": 225.6,
      "epoch": 0.025034770514603615,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7358577251434326,
      "kl": 0.0007881541740061948,
      "learning_rate": 5e-07,
      "loss": 3.152899444103241e-05,
      "reward": 2.196377229690552,
      "reward_std": 0.6824372291564942,
      "rewards/IngredientFormatReward/mean": 0.8830841064453125,
      "rewards/IngredientFormatReward/std": 0.3072942852973938,
      "rewards/IngredientMatchReward/mean": 0.4858600676059723,
      "rewards/IngredientMatchReward/std": 0.3125761985778809,
      "rewards/IngredientQuantityMatchReward/mean": 0.5524331212043763,
      "rewards/IngredientQuantityMatchReward/std": 0.44864551424980165,
      "rewards/TotalKcalExactMatchReward/mean": 0.275,
      "rewards/TotalKcalExactMatchReward/std": 0.4422182202339172,
      "step": 90
    },
    {
      "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": 0.0703125,
      "completions/max_length": 513.0,
      "completions/mean_length": 393.0734375,
      "completions/min_length": 249.6,
      "epoch": 0.02642559109874826,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8563322424888611,
      "kl": 0.0014499979191896274,
      "learning_rate": 5.277777777777777e-07,
      "loss": 5.795812467113137e-05,
      "reward": 2.0926523447036742,
      "reward_std": 0.6976558685302734,
      "rewards/IngredientFormatReward/mean": 0.8504129528999329,
      "rewards/IngredientFormatReward/std": 0.3291980028152466,
      "rewards/IngredientMatchReward/mean": 0.47219123840332033,
      "rewards/IngredientMatchReward/std": 0.2914651095867157,
      "rewards/IngredientQuantityMatchReward/mean": 0.5294231832027435,
      "rewards/IngredientQuantityMatchReward/std": 0.4433592915534973,
      "rewards/TotalKcalExactMatchReward/mean": 0.240625,
      "rewards/TotalKcalExactMatchReward/std": 0.427162367105484,
      "step": 95
    },
    {
      "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": 0.096875,
      "completions/max_length": 513.0,
      "completions/mean_length": 407.1578125,
      "completions/min_length": 259.8,
      "epoch": 0.027816411682892908,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7285707592964172,
      "kl": 0.0023988128137716557,
      "learning_rate": 5.555555555555555e-07,
      "loss": 9.596748277544976e-05,
      "reward": 2.165234994888306,
      "reward_std": 0.7257884621620179,
      "rewards/IngredientFormatReward/mean": 0.8481119632720947,
      "rewards/IngredientFormatReward/std": 0.3413813829421997,
      "rewards/IngredientMatchReward/mean": 0.4407334983348846,
      "rewards/IngredientMatchReward/std": 0.29101337790489196,
      "rewards/IngredientQuantityMatchReward/mean": 0.5388895452022553,
      "rewards/IngredientQuantityMatchReward/std": 0.4558054804801941,
      "rewards/TotalKcalExactMatchReward/mean": 0.3375,
      "rewards/TotalKcalExactMatchReward/std": 0.4702342689037323,
      "step": 100
    },
    {
      "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": 0.08125,
      "completions/max_length": 513.0,
      "completions/mean_length": 406.5203125,
      "completions/min_length": 240.4,
      "epoch": 0.02920723226703755,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7217776775360107,
      "kl": 0.0016849263662152224,
      "learning_rate": 5.833333333333334e-07,
      "loss": 6.740582175552845e-05,
      "reward": 2.108404779434204,
      "reward_std": 0.7473301291465759,
      "rewards/IngredientFormatReward/mean": 0.8337239503860474,
      "rewards/IngredientFormatReward/std": 0.36305957436561587,
      "rewards/IngredientMatchReward/mean": 0.43829986453056335,
      "rewards/IngredientMatchReward/std": 0.2977118492126465,
      "rewards/IngredientQuantityMatchReward/mean": 0.520756047964096,
      "rewards/IngredientQuantityMatchReward/std": 0.44096989631652833,
      "rewards/TotalKcalExactMatchReward/mean": 0.315625,
      "rewards/TotalKcalExactMatchReward/std": 0.4639370679855347,
      "step": 105
    },
    {
      "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": 0.0671875,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.4265625,
      "completions/min_length": 245.2,
      "epoch": 0.030598052851182198,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.736385703086853,
      "kl": 0.003075429912132677,
      "learning_rate": 6.111111111111112e-07,
      "loss": 0.00012296217028051615,
      "reward": 2.197610855102539,
      "reward_std": 0.7173490405082703,
      "rewards/IngredientFormatReward/mean": 0.8491666674613952,
      "rewards/IngredientFormatReward/std": 0.3398062169551849,
      "rewards/IngredientMatchReward/mean": 0.41518911719322205,
      "rewards/IngredientMatchReward/std": 0.2829656690359116,
      "rewards/IngredientQuantityMatchReward/mean": 0.5645050764083862,
      "rewards/IngredientQuantityMatchReward/std": 0.45311381816864016,
      "rewards/TotalKcalExactMatchReward/mean": 0.36875,
      "rewards/TotalKcalExactMatchReward/std": 0.4792756617069244,
      "step": 110
    },
    {
      "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": 0.0640625,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.53125,
      "completions/min_length": 230.2,
      "epoch": 0.031988873435326845,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7664805054664612,
      "kl": 0.004706994362641126,
      "learning_rate": 6.388888888888888e-07,
      "loss": 0.00018819000106304884,
      "reward": 2.175532102584839,
      "reward_std": 0.7320732951164246,
      "rewards/IngredientFormatReward/mean": 0.8788281440734863,
      "rewards/IngredientFormatReward/std": 0.3098617732524872,
      "rewards/IngredientMatchReward/mean": 0.4196726202964783,
      "rewards/IngredientMatchReward/std": 0.2729521691799164,
      "rewards/IngredientQuantityMatchReward/mean": 0.5520314157009125,
      "rewards/IngredientQuantityMatchReward/std": 0.4562692940235138,
      "rewards/TotalKcalExactMatchReward/mean": 0.325,
      "rewards/TotalKcalExactMatchReward/std": 0.4657438635826111,
      "step": 115
    },
    {
      "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": 0.0546875,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.9921875,
      "completions/min_length": 246.6,
      "epoch": 0.03337969401947149,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8188135623931885,
      "kl": 0.004332279106165515,
      "learning_rate": 6.666666666666666e-07,
      "loss": 0.00017332370625808836,
      "reward": 2.2424280643463135,
      "reward_std": 0.6947030186653137,
      "rewards/IngredientFormatReward/mean": 0.8772433042526245,
      "rewards/IngredientFormatReward/std": 0.3062496542930603,
      "rewards/IngredientMatchReward/mean": 0.45630409717559817,
      "rewards/IngredientMatchReward/std": 0.3024283528327942,
      "rewards/IngredientQuantityMatchReward/mean": 0.5870055854320526,
      "rewards/IngredientQuantityMatchReward/std": 0.42272945046424865,
      "rewards/TotalKcalExactMatchReward/mean": 0.321875,
      "rewards/TotalKcalExactMatchReward/std": 0.4648774802684784,
      "step": 120
    },
    {
      "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": 0.040625,
      "completions/max_length": 513.0,
      "completions/mean_length": 390.7203125,
      "completions/min_length": 249.4,
      "epoch": 0.03477051460361613,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.816830575466156,
      "kl": 0.003736836545431288,
      "learning_rate": 6.944444444444444e-07,
      "loss": 0.0001495170406997204,
      "reward": 2.2428334236145018,
      "reward_std": 0.638411796092987,
      "rewards/IngredientFormatReward/mean": 0.9001041531562806,
      "rewards/IngredientFormatReward/std": 0.2765121221542358,
      "rewards/IngredientMatchReward/mean": 0.4665631353855133,
      "rewards/IngredientMatchReward/std": 0.30025785565376284,
      "rewards/IngredientQuantityMatchReward/mean": 0.5074161112308502,
      "rewards/IngredientQuantityMatchReward/std": 0.44353713989257815,
      "rewards/TotalKcalExactMatchReward/mean": 0.36875,
      "rewards/TotalKcalExactMatchReward/std": 0.475113707780838,
      "step": 125
    },
    {
      "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": 0.0546875,
      "completions/max_length": 513.0,
      "completions/mean_length": 397.80625,
      "completions/min_length": 251.8,
      "epoch": 0.03616133518776078,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8476716876029968,
      "kl": 0.004359716525505064,
      "learning_rate": 7.222222222222221e-07,
      "loss": 0.00017424174584448337,
      "reward": 2.316778326034546,
      "reward_std": 0.6683953762054443,
      "rewards/IngredientFormatReward/mean": 0.9030468583106994,
      "rewards/IngredientFormatReward/std": 0.2746707767248154,
      "rewards/IngredientMatchReward/mean": 0.44811282157897947,
      "rewards/IngredientMatchReward/std": 0.28702212870121,
      "rewards/IngredientQuantityMatchReward/mean": 0.5578061103820801,
      "rewards/IngredientQuantityMatchReward/std": 0.44357584714889525,
      "rewards/TotalKcalExactMatchReward/mean": 0.4078125,
      "rewards/TotalKcalExactMatchReward/std": 0.49025474190711976,
      "step": 130
    },
    {
      "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": 0.075,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.109375,
      "completions/min_length": 244.4,
      "epoch": 0.037552155771905425,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.774059534072876,
      "kl": 0.005080446207284694,
      "learning_rate": 7.5e-07,
      "loss": 0.00020329309627413749,
      "reward": 2.2883750915527346,
      "reward_std": 0.7694447875022888,
      "rewards/IngredientFormatReward/mean": 0.867638885974884,
      "rewards/IngredientFormatReward/std": 0.31823705434799193,
      "rewards/IngredientMatchReward/mean": 0.45627238750457766,
      "rewards/IngredientMatchReward/std": 0.30123019218444824,
      "rewards/IngredientQuantityMatchReward/mean": 0.5004013001918792,
      "rewards/IngredientQuantityMatchReward/std": 0.4554529309272766,
      "rewards/TotalKcalExactMatchReward/mean": 0.4640625,
      "rewards/TotalKcalExactMatchReward/std": 0.4989950478076935,
      "step": 135
    },
    {
      "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": 0.1015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 399.2828125,
      "completions/min_length": 229.2,
      "epoch": 0.03894297635605007,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7574522495269775,
      "kl": 0.006042543252988253,
      "learning_rate": 7.777777777777778e-07,
      "loss": 0.000241729780100286,
      "reward": 2.141042113304138,
      "reward_std": 0.749863576889038,
      "rewards/IngredientFormatReward/mean": 0.83494793176651,
      "rewards/IngredientFormatReward/std": 0.340142023563385,
      "rewards/IngredientMatchReward/mean": 0.4368985652923584,
      "rewards/IngredientMatchReward/std": 0.3157567739486694,
      "rewards/IngredientQuantityMatchReward/mean": 0.4723206400871277,
      "rewards/IngredientQuantityMatchReward/std": 0.4440523386001587,
      "rewards/TotalKcalExactMatchReward/mean": 0.396875,
      "rewards/TotalKcalExactMatchReward/std": 0.47984519600868225,
      "step": 140
    },
    {
      "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": 0.0609375,
      "completions/max_length": 513.0,
      "completions/mean_length": 403.6734375,
      "completions/min_length": 246.8,
      "epoch": 0.04033379694019471,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7979903817176819,
      "kl": 0.004112960568454583,
      "learning_rate": 8.055555555555556e-07,
      "loss": 0.00016456012381240725,
      "reward": 2.3432874202728273,
      "reward_std": 0.7342286825180053,
      "rewards/IngredientFormatReward/mean": 0.8814843773841858,
      "rewards/IngredientFormatReward/std": 0.31216220259666444,
      "rewards/IngredientMatchReward/mean": 0.4461422383785248,
      "rewards/IngredientMatchReward/std": 0.3000922918319702,
      "rewards/IngredientQuantityMatchReward/mean": 0.5172233939170837,
      "rewards/IngredientQuantityMatchReward/std": 0.4469065308570862,
      "rewards/TotalKcalExactMatchReward/mean": 0.4984375,
      "rewards/TotalKcalExactMatchReward/std": 0.497554749250412,
      "step": 145
    },
    {
      "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": 0.0671875,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.4578125,
      "completions/min_length": 206.8,
      "epoch": 0.04172461752433936,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7423848509788513,
      "kl": 0.004441153233346995,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.0001776395714841783,
      "reward": 2.220476579666138,
      "reward_std": 0.7300424218177796,
      "rewards/IngredientFormatReward/mean": 0.8623697757720947,
      "rewards/IngredientFormatReward/std": 0.31136985719203947,
      "rewards/IngredientMatchReward/mean": 0.4273505866527557,
      "rewards/IngredientMatchReward/std": 0.30395570397377014,
      "rewards/IngredientQuantityMatchReward/mean": 0.4557562291622162,
      "rewards/IngredientQuantityMatchReward/std": 0.4484573841094971,
      "rewards/TotalKcalExactMatchReward/mean": 0.475,
      "rewards/TotalKcalExactMatchReward/std": 0.49919587969779966,
      "step": 150
    },
    {
      "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": 0.05,
      "completions/max_length": 513.0,
      "completions/mean_length": 397.7890625,
      "completions/min_length": 233.8,
      "epoch": 0.043115438108484005,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7154651880264282,
      "kl": 0.004370697992271744,
      "learning_rate": 8.611111111111111e-07,
      "loss": 0.0001748448354192078,
      "reward": 2.4821091651916505,
      "reward_std": 0.7230484843254089,
      "rewards/IngredientFormatReward/mean": 0.908965790271759,
      "rewards/IngredientFormatReward/std": 0.27234546542167665,
      "rewards/IngredientMatchReward/mean": 0.480882340669632,
      "rewards/IngredientMatchReward/std": 0.29070753455162046,
      "rewards/IngredientQuantityMatchReward/mean": 0.5750735938549042,
      "rewards/IngredientQuantityMatchReward/std": 0.4402975857257843,
      "rewards/TotalKcalExactMatchReward/mean": 0.5171875,
      "rewards/TotalKcalExactMatchReward/std": 0.49578348994255067,
      "step": 155
    },
    {
      "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": 0.0796875,
      "completions/max_length": 513.0,
      "completions/mean_length": 403.9,
      "completions/min_length": 233.2,
      "epoch": 0.04450625869262865,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.8095347285270691,
      "kl": 0.006063173009897583,
      "learning_rate": 8.888888888888888e-07,
      "loss": 0.00024237227626144886,
      "reward": 2.404101753234863,
      "reward_std": 0.7797430753707886,
      "rewards/IngredientFormatReward/mean": 0.8703645825386047,
      "rewards/IngredientFormatReward/std": 0.31925941705703736,
      "rewards/IngredientMatchReward/mean": 0.4635703802108765,
      "rewards/IngredientMatchReward/std": 0.3105102479457855,
      "rewards/IngredientQuantityMatchReward/mean": 0.5482918679714203,
      "rewards/IngredientQuantityMatchReward/std": 0.44624740481376646,
      "rewards/TotalKcalExactMatchReward/mean": 0.521875,
      "rewards/TotalKcalExactMatchReward/std": 0.49667310118675234,
      "step": 160
    },
    {
      "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": 0.05625,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.25625,
      "completions/min_length": 241.0,
      "epoch": 0.0458970792767733,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7412997484207153,
      "kl": 0.006450092940940522,
      "learning_rate": 9.166666666666665e-07,
      "loss": 0.0002578888554126024,
      "reward": 2.505335474014282,
      "reward_std": 0.7264118909835815,
      "rewards/IngredientFormatReward/mean": 0.9001525521278382,
      "rewards/IngredientFormatReward/std": 0.2870828151702881,
      "rewards/IngredientMatchReward/mean": 0.4542801439762115,
      "rewards/IngredientMatchReward/std": 0.2940641105175018,
      "rewards/IngredientQuantityMatchReward/mean": 0.6071528434753418,
      "rewards/IngredientQuantityMatchReward/std": 0.44424931406974794,
      "rewards/TotalKcalExactMatchReward/mean": 0.54375,
      "rewards/TotalKcalExactMatchReward/std": 0.4922543168067932,
      "step": 165
    },
    {
      "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": 0.0484375,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.915625,
      "completions/min_length": 234.8,
      "epoch": 0.04728789986091794,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.8398517966270447,
      "kl": 0.006277643307112157,
      "learning_rate": 9.444444444444444e-07,
      "loss": 0.000251129362732172,
      "reward": 2.5534721851348876,
      "reward_std": 0.7245315790176392,
      "rewards/IngredientFormatReward/mean": 0.9078869223594666,
      "rewards/IngredientFormatReward/std": 0.2684895843267441,
      "rewards/IngredientMatchReward/mean": 0.48150365352630614,
      "rewards/IngredientMatchReward/std": 0.31550759077072144,
      "rewards/IngredientQuantityMatchReward/mean": 0.5640815615653991,
      "rewards/IngredientQuantityMatchReward/std": 0.42716373801231383,
      "rewards/TotalKcalExactMatchReward/mean": 0.6,
      "rewards/TotalKcalExactMatchReward/std": 0.48972519040107726,
      "step": 170
    },
    {
      "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": 0.05,
      "completions/max_length": 513.0,
      "completions/mean_length": 386.38125,
      "completions/min_length": 216.6,
      "epoch": 0.048678720445062586,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7972651720046997,
      "kl": 0.007527178205782548,
      "learning_rate": 9.722222222222222e-07,
      "loss": 0.00030102948658168317,
      "reward": 2.422546720504761,
      "reward_std": 0.6932518959045411,
      "rewards/IngredientFormatReward/mean": 0.9198958396911621,
      "rewards/IngredientFormatReward/std": 0.24611102044582367,
      "rewards/IngredientMatchReward/mean": 0.4539143145084381,
      "rewards/IngredientMatchReward/std": 0.30238354206085205,
      "rewards/IngredientQuantityMatchReward/mean": 0.473736572265625,
      "rewards/IngredientQuantityMatchReward/std": 0.44805272221565245,
      "rewards/TotalKcalExactMatchReward/mean": 0.575,
      "rewards/TotalKcalExactMatchReward/std": 0.48392468094825747,
      "step": 175
    },
    {
      "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": 0.0390625,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.946875,
      "completions/min_length": 210.4,
      "epoch": 0.05006954102920723,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7244554758071899,
      "kl": 0.006808792043011635,
      "learning_rate": 1e-06,
      "loss": 0.00027234144508838656,
      "reward": 2.585988426208496,
      "reward_std": 0.6814924240112304,
      "rewards/IngredientFormatReward/mean": 0.8969047546386719,
      "rewards/IngredientFormatReward/std": 0.28515351116657256,
      "rewards/IngredientMatchReward/mean": 0.4871840238571167,
      "rewards/IngredientMatchReward/std": 0.30355110168457033,
      "rewards/IngredientQuantityMatchReward/mean": 0.5706496775150299,
      "rewards/IngredientQuantityMatchReward/std": 0.4206697165966034,
      "rewards/TotalKcalExactMatchReward/mean": 0.63125,
      "rewards/TotalKcalExactMatchReward/std": 0.47843066453933714,
      "step": 180
    },
    {
      "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": 0.0203125,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.35,
      "completions/min_length": 258.6,
      "epoch": 0.05146036161335188,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7314364910125732,
      "kl": 0.007314420084003359,
      "learning_rate": 9.999947107075408e-07,
      "loss": 0.00029256632551550866,
      "reward": 2.662673282623291,
      "reward_std": 0.5554831802845002,
      "rewards/IngredientFormatReward/mean": 0.9673437595367431,
      "rewards/IngredientFormatReward/std": 0.15578847080469133,
      "rewards/IngredientMatchReward/mean": 0.5206299781799316,
      "rewards/IngredientMatchReward/std": 0.3142448902130127,
      "rewards/IngredientQuantityMatchReward/mean": 0.5731370329856873,
      "rewards/IngredientQuantityMatchReward/std": 0.43153069615364076,
      "rewards/TotalKcalExactMatchReward/mean": 0.6015625,
      "rewards/TotalKcalExactMatchReward/std": 0.4841706693172455,
      "step": 185
    },
    {
      "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": 0.0390625,
      "completions/max_length": 513.0,
      "completions/mean_length": 385.6,
      "completions/min_length": 228.6,
      "epoch": 0.05285118219749652,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7836043238639832,
      "kl": 0.008130288516986184,
      "learning_rate": 9.999788429420697e-07,
      "loss": 0.0003252896945923567,
      "reward": 2.6136358737945558,
      "reward_std": 0.6777990698814392,
      "rewards/IngredientFormatReward/mean": 0.9325000047683716,
      "rewards/IngredientFormatReward/std": 0.2264118403196335,
      "rewards/IngredientMatchReward/mean": 0.48915613889694215,
      "rewards/IngredientMatchReward/std": 0.30804571509361267,
      "rewards/IngredientQuantityMatchReward/mean": 0.5607298374176025,
      "rewards/IngredientQuantityMatchReward/std": 0.4353516399860382,
      "rewards/TotalKcalExactMatchReward/mean": 0.63125,
      "rewards/TotalKcalExactMatchReward/std": 0.47783060669898986,
      "step": 190
    },
    {
      "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": 0.0625,
      "completions/max_length": 513.0,
      "completions/mean_length": 400.6265625,
      "completions/min_length": 265.0,
      "epoch": 0.054242002781641166,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7703161239624023,
      "kl": 0.010244119248818606,
      "learning_rate": 9.999523970393036e-07,
      "loss": 0.0004097242839634418,
      "reward": 2.4793437004089354,
      "reward_std": 0.6277800083160401,
      "rewards/IngredientFormatReward/mean": 0.9137834787368775,
      "rewards/IngredientFormatReward/std": 0.25849959552288054,
      "rewards/IngredientMatchReward/mean": 0.45915488600730897,
      "rewards/IngredientMatchReward/std": 0.2793452024459839,
      "rewards/IngredientQuantityMatchReward/mean": 0.5282802939414978,
      "rewards/IngredientQuantityMatchReward/std": 0.45662105083465576,
      "rewards/TotalKcalExactMatchReward/mean": 0.578125,
      "rewards/TotalKcalExactMatchReward/std": 0.49128650426864623,
      "step": 195
    },
    {
      "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": 0.0453125,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.09375,
      "completions/min_length": 236.0,
      "epoch": 0.055632823365785816,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7537059783935547,
      "kl": 0.009191960797761568,
      "learning_rate": 9.999153735587632e-07,
      "loss": 0.00036758126225322485,
      "reward": 2.4740325927734377,
      "reward_std": 0.6981268048286438,
      "rewards/IngredientFormatReward/mean": 0.9252864718437195,
      "rewards/IngredientFormatReward/std": 0.24678438305854797,
      "rewards/IngredientMatchReward/mean": 0.45877419114112855,
      "rewards/IngredientMatchReward/std": 0.29859951734542844,
      "rewards/IngredientQuantityMatchReward/mean": 0.46965948939323426,
      "rewards/IngredientQuantityMatchReward/std": 0.44375336170196533,
      "rewards/TotalKcalExactMatchReward/mean": 0.6203125,
      "rewards/TotalKcalExactMatchReward/std": 0.4775801539421082,
      "step": 200
    },
    {
      "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": 0.0421875,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.365625,
      "completions/min_length": 241.8,
      "epoch": 0.05702364394993046,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7670131325721741,
      "kl": 0.008866281888913363,
      "learning_rate": 9.9986777328376e-07,
      "loss": 0.00035456912592053414,
      "reward": 2.679036331176758,
      "reward_std": 0.57293900847435,
      "rewards/IngredientFormatReward/mean": 0.9371874928474426,
      "rewards/IngredientFormatReward/std": 0.23141905665397644,
      "rewards/IngredientMatchReward/mean": 0.5184425234794616,
      "rewards/IngredientMatchReward/std": 0.30044159293174744,
      "rewards/IngredientQuantityMatchReward/mean": 0.5343438982963562,
      "rewards/IngredientQuantityMatchReward/std": 0.450323224067688,
      "rewards/TotalKcalExactMatchReward/mean": 0.6890625,
      "rewards/TotalKcalExactMatchReward/std": 0.4618146061897278,
      "step": 205
    },
    {
      "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": 0.0609375,
      "completions/max_length": 512.8,
      "completions/mean_length": 397.0625,
      "completions/min_length": 240.8,
      "epoch": 0.0584144645340751,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7266873121261597,
      "kl": 0.009309542400296777,
      "learning_rate": 9.998095972213817e-07,
      "loss": 0.0003722626715898514,
      "reward": 2.597199583053589,
      "reward_std": 0.6256489753723145,
      "rewards/IngredientFormatReward/mean": 0.91268230676651,
      "rewards/IngredientFormatReward/std": 0.25630738735198977,
      "rewards/IngredientMatchReward/mean": 0.47620717287063596,
      "rewards/IngredientMatchReward/std": 0.30730987787246705,
      "rewards/IngredientQuantityMatchReward/mean": 0.5395601093769073,
      "rewards/IngredientQuantityMatchReward/std": 0.44395468235015867,
      "rewards/TotalKcalExactMatchReward/mean": 0.66875,
      "rewards/TotalKcalExactMatchReward/std": 0.4664685130119324,
      "step": 210
    },
    {
      "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": 0.059375,
      "completions/max_length": 513.0,
      "completions/mean_length": 397.6953125,
      "completions/min_length": 250.6,
      "epoch": 0.059805285118219746,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7109785079956055,
      "kl": 0.009796890115831047,
      "learning_rate": 9.99740846602469e-07,
      "loss": 0.00039182775653898714,
      "reward": 2.7128491401672363,
      "reward_std": 0.5502410054206848,
      "rewards/IngredientFormatReward/mean": 0.9221875071525574,
      "rewards/IngredientFormatReward/std": 0.2499729573726654,
      "rewards/IngredientMatchReward/mean": 0.49059213399887086,
      "rewards/IngredientMatchReward/std": 0.3133565247058868,
      "rewards/IngredientQuantityMatchReward/mean": 0.625069534778595,
      "rewards/IngredientQuantityMatchReward/std": 0.4262022256851196,
      "rewards/TotalKcalExactMatchReward/mean": 0.675,
      "rewards/TotalKcalExactMatchReward/std": 0.4658839821815491,
      "step": 215
    },
    {
      "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": 0.021875,
      "completions/max_length": 513.0,
      "completions/mean_length": 377.0,
      "completions/min_length": 219.4,
      "epoch": 0.061196105702364396,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7904135584831238,
      "kl": 0.00917984291445464,
      "learning_rate": 9.996615228815905e-07,
      "loss": 0.0003672625869512558,
      "reward": 2.5808473587036134,
      "reward_std": 0.6128181099891663,
      "rewards/IngredientFormatReward/mean": 0.9564843654632569,
      "rewards/IngredientFormatReward/std": 0.18436979204416276,
      "rewards/IngredientMatchReward/mean": 0.44942651987075805,
      "rewards/IngredientMatchReward/std": 0.3052241802215576,
      "rewards/IngredientQuantityMatchReward/mean": 0.5499364614486695,
      "rewards/IngredientQuantityMatchReward/std": 0.45513352155685427,
      "rewards/TotalKcalExactMatchReward/mean": 0.625,
      "rewards/TotalKcalExactMatchReward/std": 0.47820115089416504,
      "step": 220
    },
    {
      "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": 0.0359375,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.703125,
      "completions/min_length": 261.0,
      "epoch": 0.06258692628650904,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.8294149041175842,
      "kl": 0.00947985349339433,
      "learning_rate": 9.995716277370112e-07,
      "loss": 0.00037920954637229445,
      "reward": 2.6771074295043946,
      "reward_std": 0.6180533349514008,
      "rewards/IngredientFormatReward/mean": 0.9501804351806641,
      "rewards/IngredientFormatReward/std": 0.19291509985923766,
      "rewards/IngredientMatchReward/mean": 0.5157575666904449,
      "rewards/IngredientMatchReward/std": 0.32128278613090516,
      "rewards/IngredientQuantityMatchReward/mean": 0.5752319931983948,
      "rewards/IngredientQuantityMatchReward/std": 0.4331511855125427,
      "rewards/TotalKcalExactMatchReward/mean": 0.6359375,
      "rewards/TotalKcalExactMatchReward/std": 0.48184582591056824,
      "step": 225
    },
    {
      "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": 0.0515625,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.38125,
      "completions/min_length": 256.2,
      "epoch": 0.06397774687065369,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8006162643432617,
      "kl": 0.013353917776839808,
      "learning_rate": 9.994711630706585e-07,
      "loss": 0.0005340860225260258,
      "reward": 2.593950796127319,
      "reward_std": 0.6663724660873414,
      "rewards/IngredientFormatReward/mean": 0.9172916650772095,
      "rewards/IngredientFormatReward/std": 0.2557332932949066,
      "rewards/IngredientMatchReward/mean": 0.5012678146362305,
      "rewards/IngredientMatchReward/std": 0.3088660776615143,
      "rewards/IngredientQuantityMatchReward/mean": 0.5988288640975952,
      "rewards/IngredientQuantityMatchReward/std": 0.43793771862983705,
      "rewards/TotalKcalExactMatchReward/mean": 0.5765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4930992126464844,
      "step": 230
    },
    {
      "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": 0.0171875,
      "completions/max_length": 509.8,
      "completions/mean_length": 388.740625,
      "completions/min_length": 247.4,
      "epoch": 0.06536856745479833,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.728783905506134,
      "kl": 0.010416155745042488,
      "learning_rate": 9.993601310080799e-07,
      "loss": 0.0004166138358414173,
      "reward": 2.6682600498199465,
      "reward_std": 0.5333548069000245,
      "rewards/IngredientFormatReward/mean": 0.9703906297683715,
      "rewards/IngredientFormatReward/std": 0.145054429769516,
      "rewards/IngredientMatchReward/mean": 0.521271800994873,
      "rewards/IngredientMatchReward/std": 0.2949683487415314,
      "rewards/IngredientQuantityMatchReward/mean": 0.5734726548194885,
      "rewards/IngredientQuantityMatchReward/std": 0.44596282243728635,
      "rewards/TotalKcalExactMatchReward/mean": 0.603125,
      "rewards/TotalKcalExactMatchReward/std": 0.4815952718257904,
      "step": 235
    },
    {
      "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": 0.0203125,
      "completions/max_length": 513.0,
      "completions/mean_length": 387.5265625,
      "completions/min_length": 242.2,
      "epoch": 0.06675938803894298,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7192296385765076,
      "kl": 0.010648410249268636,
      "learning_rate": 9.992385338983998e-07,
      "loss": 0.00042591681703925134,
      "reward": 2.7542993068695067,
      "reward_std": 0.5644841074943543,
      "rewards/IngredientFormatReward/mean": 0.9609635472297668,
      "rewards/IngredientFormatReward/std": 0.17166336327791215,
      "rewards/IngredientMatchReward/mean": 0.5329110741615295,
      "rewards/IngredientMatchReward/std": 0.2976407825946808,
      "rewards/IngredientQuantityMatchReward/mean": 0.593237179517746,
      "rewards/IngredientQuantityMatchReward/std": 0.4195017755031586,
      "rewards/TotalKcalExactMatchReward/mean": 0.6671875,
      "rewards/TotalKcalExactMatchReward/std": 0.4707712590694427,
      "step": 240
    },
    {
      "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": 0.0625,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.48125,
      "completions/min_length": 243.2,
      "epoch": 0.06815020862308763,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7437554597854614,
      "kl": 0.014695871755247936,
      "learning_rate": 9.99106374314269e-07,
      "loss": 0.0005878821946680545,
      "reward": 2.5849650382995604,
      "reward_std": 0.6724656701087952,
      "rewards/IngredientFormatReward/mean": 0.9083029747009277,
      "rewards/IngredientFormatReward/std": 0.27130608856678007,
      "rewards/IngredientMatchReward/mean": 0.5053019642829895,
      "rewards/IngredientMatchReward/std": 0.30541407465934756,
      "rewards/IngredientQuantityMatchReward/mean": 0.585422694683075,
      "rewards/IngredientQuantityMatchReward/std": 0.43134939670562744,
      "rewards/TotalKcalExactMatchReward/mean": 0.5859375,
      "rewards/TotalKcalExactMatchReward/std": 0.48427985310554506,
      "step": 245
    },
    {
      "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": 0.04375,
      "completions/max_length": 513.0,
      "completions/mean_length": 393.321875,
      "completions/min_length": 228.0,
      "epoch": 0.06954102920723226,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6927481889724731,
      "kl": 0.012796262564370409,
      "learning_rate": 9.989636550518103e-07,
      "loss": 0.0005117684602737426,
      "reward": 2.5995935440063476,
      "reward_std": 0.6334937334060669,
      "rewards/IngredientFormatReward/mean": 0.9379650235176087,
      "rewards/IngredientFormatReward/std": 0.22491171956062317,
      "rewards/IngredientMatchReward/mean": 0.5036893606185913,
      "rewards/IngredientMatchReward/std": 0.3012221992015839,
      "rewards/IngredientQuantityMatchReward/mean": 0.5235642194747925,
      "rewards/IngredientQuantityMatchReward/std": 0.43460506200790405,
      "rewards/TotalKcalExactMatchReward/mean": 0.634375,
      "rewards/TotalKcalExactMatchReward/std": 0.4736035823822021,
      "step": 250
    },
    {
      "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": 0.0171875,
      "completions/max_length": 510.8,
      "completions/mean_length": 385.353125,
      "completions/min_length": 243.6,
      "epoch": 0.07093184979137691,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7734192609786987,
      "kl": 0.012177031306782738,
      "learning_rate": 9.988103791305593e-07,
      "loss": 0.0004871044307947159,
      "reward": 2.7546799659729,
      "reward_std": 0.5723294258117676,
      "rewards/IngredientFormatReward/mean": 0.9548437476158143,
      "rewards/IngredientFormatReward/std": 0.1730358988046646,
      "rewards/IngredientMatchReward/mean": 0.5104142010211945,
      "rewards/IngredientMatchReward/std": 0.3049011051654816,
      "rewards/IngredientQuantityMatchReward/mean": 0.6144220471382141,
      "rewards/IngredientQuantityMatchReward/std": 0.4470917284488678,
      "rewards/TotalKcalExactMatchReward/mean": 0.675,
      "rewards/TotalKcalExactMatchReward/std": 0.4676419675350189,
      "step": 255
    },
    {
      "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": 0.053125,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.6890625,
      "completions/min_length": 265.2,
      "epoch": 0.07232267037552156,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6882739067077637,
      "kl": 0.013909409387269988,
      "learning_rate": 9.986465497934006e-07,
      "loss": 0.0005562674719840288,
      "reward": 2.643060731887817,
      "reward_std": 0.60300612449646,
      "rewards/IngredientFormatReward/mean": 0.9370833277702332,
      "rewards/IngredientFormatReward/std": 0.2204873889684677,
      "rewards/IngredientMatchReward/mean": 0.4994661629199982,
      "rewards/IngredientMatchReward/std": 0.28945026993751527,
      "rewards/IngredientQuantityMatchReward/mean": 0.5690113306045532,
      "rewards/IngredientQuantityMatchReward/std": 0.42663033604621886,
      "rewards/TotalKcalExactMatchReward/mean": 0.6375,
      "rewards/TotalKcalExactMatchReward/std": 0.4768159568309784,
      "step": 260
    },
    {
      "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": 0.0609375,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.25,
      "completions/min_length": 261.6,
      "epoch": 0.0737134909596662,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7281216979026794,
      "kl": 0.01698645217111334,
      "learning_rate": 9.984721705064993e-07,
      "loss": 0.0006792577914893627,
      "reward": 2.5794155597686768,
      "reward_std": 0.6507645726203919,
      "rewards/IngredientFormatReward/mean": 0.9251153349876404,
      "rewards/IngredientFormatReward/std": 0.24796229004859924,
      "rewards/IngredientMatchReward/mean": 0.4920498549938202,
      "rewards/IngredientMatchReward/std": 0.31408061981201174,
      "rewards/IngredientQuantityMatchReward/mean": 0.5185003936290741,
      "rewards/IngredientQuantityMatchReward/std": 0.44893131852149964,
      "rewards/TotalKcalExactMatchReward/mean": 0.64375,
      "rewards/TotalKcalExactMatchReward/std": 0.4780986487865448,
      "step": 265
    },
    {
      "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": 0.0375,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.65625,
      "completions/min_length": 219.6,
      "epoch": 0.07510431154381085,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.899786114692688,
      "kl": 0.01553950141533278,
      "learning_rate": 9.98287244959228e-07,
      "loss": 0.0006215041503310204,
      "reward": 2.607704925537109,
      "reward_std": 0.5998050808906555,
      "rewards/IngredientFormatReward/mean": 0.9451153397560119,
      "rewards/IngredientFormatReward/std": 0.20488475263118744,
      "rewards/IngredientMatchReward/mean": 0.5260571718215943,
      "rewards/IngredientMatchReward/std": 0.3056329846382141,
      "rewards/IngredientQuantityMatchReward/mean": 0.4787199139595032,
      "rewards/IngredientQuantityMatchReward/std": 0.4413680613040924,
      "rewards/TotalKcalExactMatchReward/mean": 0.6578125,
      "rewards/TotalKcalExactMatchReward/std": 0.46880462765693665,
      "step": 270
    },
    {
      "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": 0.0359375,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.5984375,
      "completions/min_length": 262.0,
      "epoch": 0.07649513212795549,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7272493243217468,
      "kl": 0.012873871141346171,
      "learning_rate": 9.980917770640873e-07,
      "loss": 0.0005149764940142632,
      "reward": 2.81902494430542,
      "reward_std": 0.5218873679637909,
      "rewards/IngredientFormatReward/mean": 0.9535379409790039,
      "rewards/IngredientFormatReward/std": 0.1974748343229294,
      "rewards/IngredientMatchReward/mean": 0.5688200712203979,
      "rewards/IngredientMatchReward/std": 0.31162298321723936,
      "rewards/IngredientQuantityMatchReward/mean": 0.5747918665409089,
      "rewards/IngredientQuantityMatchReward/std": 0.42267643809318545,
      "rewards/TotalKcalExactMatchReward/mean": 0.721875,
      "rewards/TotalKcalExactMatchReward/std": 0.4450909078121185,
      "step": 275
    },
    {
      "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": 0.034375,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.384375,
      "completions/min_length": 252.6,
      "epoch": 0.07788595271210014,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7545354962348938,
      "kl": 0.013803366833599284,
      "learning_rate": 9.978857709566249e-07,
      "loss": 0.0005521467886865139,
      "reward": 2.590224504470825,
      "reward_std": 0.5547021985054016,
      "rewards/IngredientFormatReward/mean": 0.9383593797683716,
      "rewards/IngredientFormatReward/std": 0.21592361330986024,
      "rewards/IngredientMatchReward/mean": 0.5045319378376008,
      "rewards/IngredientMatchReward/std": 0.2803065776824951,
      "rewards/IngredientQuantityMatchReward/mean": 0.5004582226276397,
      "rewards/IngredientQuantityMatchReward/std": 0.43875272274017335,
      "rewards/TotalKcalExactMatchReward/mean": 0.646875,
      "rewards/TotalKcalExactMatchReward/std": 0.4710656404495239,
      "step": 280
    },
    {
      "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": 0.0234375,
      "completions/max_length": 512.8,
      "completions/mean_length": 389.828125,
      "completions/min_length": 236.4,
      "epoch": 0.07927677329624479,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.1000856161117554,
      "kl": 0.015478781308047473,
      "learning_rate": 9.976692309953471e-07,
      "loss": 0.0006189643405377865,
      "reward": 2.813587999343872,
      "reward_std": 0.5521968245506287,
      "rewards/IngredientFormatReward/mean": 0.9665885329246521,
      "rewards/IngredientFormatReward/std": 0.15655259266495705,
      "rewards/IngredientMatchReward/mean": 0.5620684623718262,
      "rewards/IngredientMatchReward/std": 0.32483131885528566,
      "rewards/IngredientQuantityMatchReward/mean": 0.5911810338497162,
      "rewards/IngredientQuantityMatchReward/std": 0.44112812876701357,
      "rewards/TotalKcalExactMatchReward/mean": 0.69375,
      "rewards/TotalKcalExactMatchReward/std": 0.45805424451828003,
      "step": 285
    },
    {
      "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": 0.0484375,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.559375,
      "completions/min_length": 230.8,
      "epoch": 0.08066759388038942,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.8232524394989014,
      "kl": 0.015272686962271109,
      "learning_rate": 9.974421617616267e-07,
      "loss": 0.0006192252039909363,
      "reward": 2.5914158821105957,
      "reward_std": 0.6719521403312683,
      "rewards/IngredientFormatReward/mean": 0.9345498442649841,
      "rewards/IngredientFormatReward/std": 0.23133745789527893,
      "rewards/IngredientMatchReward/mean": 0.5013374328613281,
      "rewards/IngredientMatchReward/std": 0.31415660977363585,
      "rewards/IngredientQuantityMatchReward/mean": 0.4836536109447479,
      "rewards/IngredientQuantityMatchReward/std": 0.4528293967247009,
      "rewards/TotalKcalExactMatchReward/mean": 0.671875,
      "rewards/TotalKcalExactMatchReward/std": 0.4693707346916199,
      "step": 290
    },
    {
      "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": 0.01875,
      "completions/max_length": 508.6,
      "completions/mean_length": 384.0015625,
      "completions/min_length": 226.6,
      "epoch": 0.08205841446453407,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7752309441566467,
      "kl": 0.014664123253896832,
      "learning_rate": 9.97204568059606e-07,
      "loss": 0.0005867013707756997,
      "reward": 2.727179002761841,
      "reward_std": 0.5595415353775024,
      "rewards/IngredientFormatReward/mean": 0.9712499976158142,
      "rewards/IngredientFormatReward/std": 0.1503540590405464,
      "rewards/IngredientMatchReward/mean": 0.5309282064437866,
      "rewards/IngredientMatchReward/std": 0.29940311312675477,
      "rewards/IngredientQuantityMatchReward/mean": 0.5359383165836334,
      "rewards/IngredientQuantityMatchReward/std": 0.42868558764457704,
      "rewards/TotalKcalExactMatchReward/mean": 0.6890625,
      "rewards/TotalKcalExactMatchReward/std": 0.4502646565437317,
      "step": 295
    },
    {
      "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": 0.034375,
      "completions/max_length": 510.4,
      "completions/mean_length": 396.6265625,
      "completions/min_length": 260.4,
      "epoch": 0.08344923504867872,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7676609754562378,
      "kl": 0.014426854299381375,
      "learning_rate": 9.96956454916095e-07,
      "loss": 0.0005770421121269464,
      "reward": 2.58949761390686,
      "reward_std": 0.6066441535949707,
      "rewards/IngredientFormatReward/mean": 0.9593489646911622,
      "rewards/IngredientFormatReward/std": 0.17723449170589448,
      "rewards/IngredientMatchReward/mean": 0.5475229501724244,
      "rewards/IngredientMatchReward/std": 0.32086688876152036,
      "rewards/IngredientQuantityMatchReward/mean": 0.44668819308280944,
      "rewards/IngredientQuantityMatchReward/std": 0.4429606318473816,
      "rewards/TotalKcalExactMatchReward/mean": 0.6359375,
      "rewards/TotalKcalExactMatchReward/std": 0.4722549319267273,
      "step": 300
    },
    {
      "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": 0.0296875,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.1828125,
      "completions/min_length": 230.0,
      "epoch": 0.08484005563282336,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7131373882293701,
      "kl": 0.014628965401789174,
      "learning_rate": 9.96697827580466e-07,
      "loss": 0.0005851372610777616,
      "reward": 2.770707702636719,
      "reward_std": 0.6018207550048829,
      "rewards/IngredientFormatReward/mean": 0.9516406297683716,
      "rewards/IngredientFormatReward/std": 0.20158994495868682,
      "rewards/IngredientMatchReward/mean": 0.5231913506984711,
      "rewards/IngredientMatchReward/std": 0.31127086877822874,
      "rewards/IngredientQuantityMatchReward/mean": 0.6005632758140564,
      "rewards/IngredientQuantityMatchReward/std": 0.4462136089801788,
      "rewards/TotalKcalExactMatchReward/mean": 0.6953125,
      "rewards/TotalKcalExactMatchReward/std": 0.4513925492763519,
      "step": 305
    },
    {
      "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": 0.015625,
      "completions/max_length": 509.6,
      "completions/mean_length": 391.6234375,
      "completions/min_length": 258.0,
      "epoch": 0.08623087621696801,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7294876575469971,
      "kl": 0.015699848416261375,
      "learning_rate": 9.964286915245411e-07,
      "loss": 0.0006280275993049145,
      "reward": 2.776950168609619,
      "reward_std": 0.5536986947059631,
      "rewards/IngredientFormatReward/mean": 0.977734375,
      "rewards/IngredientFormatReward/std": 0.1404439002275467,
      "rewards/IngredientMatchReward/mean": 0.514191472530365,
      "rewards/IngredientMatchReward/std": 0.31364077925682066,
      "rewards/IngredientQuantityMatchReward/mean": 0.5928367972373962,
      "rewards/IngredientQuantityMatchReward/std": 0.44084068536758425,
      "rewards/TotalKcalExactMatchReward/mean": 0.6921875,
      "rewards/TotalKcalExactMatchReward/std": 0.45822349190711975,
      "step": 310
    },
    {
      "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": 0.025,
      "completions/max_length": 511.8,
      "completions/mean_length": 388.1203125,
      "completions/min_length": 233.6,
      "epoch": 0.08762169680111266,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.780403196811676,
      "kl": 0.016697630251292138,
      "learning_rate": 9.96149052442478e-07,
      "loss": 0.0006678056437522173,
      "reward": 2.6895166397094727,
      "reward_std": 0.5929653167724609,
      "rewards/IngredientFormatReward/mean": 0.9600744128227234,
      "rewards/IngredientFormatReward/std": 0.18856671154499055,
      "rewards/IngredientMatchReward/mean": 0.5147532820701599,
      "rewards/IngredientMatchReward/std": 0.28225860595703123,
      "rewards/IngredientQuantityMatchReward/mean": 0.5459389805793762,
      "rewards/IngredientQuantityMatchReward/std": 0.4366516828536987,
      "rewards/TotalKcalExactMatchReward/mean": 0.66875,
      "rewards/TotalKcalExactMatchReward/std": 0.4671382009983063,
      "step": 315
    },
    {
      "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": 0.009375,
      "completions/max_length": 504.8,
      "completions/mean_length": 386.215625,
      "completions/min_length": 266.8,
      "epoch": 0.0890125173852573,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7417155504226685,
      "kl": 0.015790742193348705,
      "learning_rate": 9.95858916250648e-07,
      "loss": 0.0006316334940493107,
      "reward": 2.8779314041137694,
      "reward_std": 0.5287329256534576,
      "rewards/IngredientFormatReward/mean": 0.9809635519981384,
      "rewards/IngredientFormatReward/std": 0.103397686034441,
      "rewards/IngredientMatchReward/mean": 0.5704042553901673,
      "rewards/IngredientMatchReward/std": 0.2879896104335785,
      "rewards/IngredientQuantityMatchReward/mean": 0.6140635967254638,
      "rewards/IngredientQuantityMatchReward/std": 0.41456308364868166,
      "rewards/TotalKcalExactMatchReward/mean": 0.7125,
      "rewards/TotalKcalExactMatchReward/std": 0.44940419793128966,
      "step": 320
    },
    {
      "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": 0.0328125,
      "completions/max_length": 512.8,
      "completions/mean_length": 398.8328125,
      "completions/min_length": 265.2,
      "epoch": 0.09040333796940195,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7298070192337036,
      "kl": 0.017347930185496806,
      "learning_rate": 9.955582890875117e-07,
      "loss": 0.0006938629318028689,
      "reward": 2.8600261211395264,
      "reward_std": 0.5498974919319153,
      "rewards/IngredientFormatReward/mean": 0.9581212997436523,
      "rewards/IngredientFormatReward/std": 0.17986520379781723,
      "rewards/IngredientMatchReward/mean": 0.58367520570755,
      "rewards/IngredientMatchReward/std": 0.3089649319648743,
      "rewards/IngredientQuantityMatchReward/mean": 0.6135420680046082,
      "rewards/IngredientQuantityMatchReward/std": 0.41773806810379027,
      "rewards/TotalKcalExactMatchReward/mean": 0.7046875,
      "rewards/TotalKcalExactMatchReward/std": 0.4514980256557465,
      "step": 325
    },
    {
      "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": 0.009375,
      "completions/max_length": 512.0,
      "completions/mean_length": 382.915625,
      "completions/min_length": 248.2,
      "epoch": 0.0917941585535466,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7022362351417542,
      "kl": 0.016667363268788905,
      "learning_rate": 9.952471773134892e-07,
      "loss": 0.0006667527835816145,
      "reward": 2.7937798500061035,
      "reward_std": 0.521451336145401,
      "rewards/IngredientFormatReward/mean": 0.9814843893051147,
      "rewards/IngredientFormatReward/std": 0.1265960529446602,
      "rewards/IngredientMatchReward/mean": 0.5398667573928833,
      "rewards/IngredientMatchReward/std": 0.3047673046588898,
      "rewards/IngredientQuantityMatchReward/mean": 0.5927411913871765,
      "rewards/IngredientQuantityMatchReward/std": 0.40651952624320986,
      "rewards/TotalKcalExactMatchReward/mean": 0.6796875,
      "rewards/TotalKcalExactMatchReward/std": 0.46350181102752686,
      "step": 330
    },
    {
      "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": 0.028125,
      "completions/max_length": 512.0,
      "completions/mean_length": 390.434375,
      "completions/min_length": 266.6,
      "epoch": 0.09318497913769123,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.8118611574172974,
      "kl": 0.01669728810666129,
      "learning_rate": 9.949255875108252e-07,
      "loss": 0.000667879544198513,
      "reward": 2.800106716156006,
      "reward_std": 0.5328549861907959,
      "rewards/IngredientFormatReward/mean": 0.9609114646911621,
      "rewards/IngredientFormatReward/std": 0.16961526293307544,
      "rewards/IngredientMatchReward/mean": 0.5768297553062439,
      "rewards/IngredientMatchReward/std": 0.30959353446960447,
      "rewards/IngredientQuantityMatchReward/mean": 0.5498654782772064,
      "rewards/IngredientQuantityMatchReward/std": 0.43173102736473085,
      "rewards/TotalKcalExactMatchReward/mean": 0.7125,
      "rewards/TotalKcalExactMatchReward/std": 0.45088860392570496,
      "step": 335
    },
    {
      "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": 0.03125,
      "completions/max_length": 506.0,
      "completions/mean_length": 386.9359375,
      "completions/min_length": 251.8,
      "epoch": 0.09457579972183588,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7292128801345825,
      "kl": 0.017025439580902458,
      "learning_rate": 9.945935264834493e-07,
      "loss": 0.0006810775958001614,
      "reward": 2.8443037033081056,
      "reward_std": 0.5523277521133423,
      "rewards/IngredientFormatReward/mean": 0.96328125,
      "rewards/IngredientFormatReward/std": 0.17349836379289627,
      "rewards/IngredientMatchReward/mean": 0.5613864183425903,
      "rewards/IngredientMatchReward/std": 0.30622108578681945,
      "rewards/IngredientQuantityMatchReward/mean": 0.6008860230445862,
      "rewards/IngredientQuantityMatchReward/std": 0.4337588012218475,
      "rewards/TotalKcalExactMatchReward/mean": 0.71875,
      "rewards/TotalKcalExactMatchReward/std": 0.4478823184967041,
      "step": 340
    },
    {
      "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": 0.0140625,
      "completions/max_length": 507.4,
      "completions/mean_length": 391.43125,
      "completions/min_length": 240.6,
      "epoch": 0.09596662030598054,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7324079871177673,
      "kl": 0.01828371067531407,
      "learning_rate": 9.942510012568337e-07,
      "loss": 0.0007314646616578102,
      "reward": 2.748303699493408,
      "reward_std": 0.538184404373169,
      "rewards/IngredientFormatReward/mean": 0.9805208444595337,
      "rewards/IngredientFormatReward/std": 0.1241001047194004,
      "rewards/IngredientMatchReward/mean": 0.5349696159362793,
      "rewards/IngredientMatchReward/std": 0.27805001139640806,
      "rewards/IngredientQuantityMatchReward/mean": 0.5281257867813111,
      "rewards/IngredientQuantityMatchReward/std": 0.44162182211875917,
      "rewards/TotalKcalExactMatchReward/mean": 0.7046875,
      "rewards/TotalKcalExactMatchReward/std": 0.45519277453422546,
      "step": 345
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 389.78125,
      "completions/min_length": 255.4,
      "epoch": 0.09735744089012517,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7865349650382996,
      "kl": 0.018346749048214407,
      "learning_rate": 9.938980190778424e-07,
      "loss": 0.0007338568568229676,
      "reward": 2.782780981063843,
      "reward_std": 0.5469257652759552,
      "rewards/IngredientFormatReward/mean": 0.9700781226158142,
      "rewards/IngredientFormatReward/std": 0.16300422549247742,
      "rewards/IngredientMatchReward/mean": 0.5200992047786712,
      "rewards/IngredientMatchReward/std": 0.30364001989364625,
      "rewards/IngredientQuantityMatchReward/mean": 0.545728611946106,
      "rewards/IngredientQuantityMatchReward/std": 0.4233491599559784,
      "rewards/TotalKcalExactMatchReward/mean": 0.746875,
      "rewards/TotalKcalExactMatchReward/std": 0.43145933747291565,
      "step": 350
    },
    {
      "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": 0.0140625,
      "completions/max_length": 508.2,
      "completions/mean_length": 394.4890625,
      "completions/min_length": 241.0,
      "epoch": 0.09874826147426982,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7335918545722961,
      "kl": 0.01754337821621448,
      "learning_rate": 9.935345874145795e-07,
      "loss": 0.0007017217110842466,
      "reward": 2.850579833984375,
      "reward_std": 0.4709952771663666,
      "rewards/IngredientFormatReward/mean": 0.978515625,
      "rewards/IngredientFormatReward/std": 0.12574003208428622,
      "rewards/IngredientMatchReward/mean": 0.5792435526847839,
      "rewards/IngredientMatchReward/std": 0.2941656231880188,
      "rewards/IngredientQuantityMatchReward/mean": 0.6225082039833069,
      "rewards/IngredientQuantityMatchReward/std": 0.4228747606277466,
      "rewards/TotalKcalExactMatchReward/mean": 0.6703125,
      "rewards/TotalKcalExactMatchReward/std": 0.4670146644115448,
      "step": 355
    },
    {
      "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": 0.0203125,
      "completions/max_length": 513.0,
      "completions/mean_length": 386.7578125,
      "completions/min_length": 233.4,
      "epoch": 0.10013908205841446,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7823643088340759,
      "kl": 0.019979644613340498,
      "learning_rate": 9.931607139562303e-07,
      "loss": 0.0007737993262708187,
      "reward": 2.771283721923828,
      "reward_std": 0.48842650055885317,
      "rewards/IngredientFormatReward/mean": 0.9694271087646484,
      "rewards/IngredientFormatReward/std": 0.16611113548278808,
      "rewards/IngredientMatchReward/mean": 0.549196434020996,
      "rewards/IngredientMatchReward/std": 0.3055261015892029,
      "rewards/IngredientQuantityMatchReward/mean": 0.5510977387428284,
      "rewards/IngredientQuantityMatchReward/std": 0.4223670423030853,
      "rewards/TotalKcalExactMatchReward/mean": 0.7015625,
      "rewards/TotalKcalExactMatchReward/std": 0.4430360376834869,
      "step": 360
    },
    {
      "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": 0.0234375,
      "completions/max_length": 510.4,
      "completions/mean_length": 386.5203125,
      "completions/min_length": 231.8,
      "epoch": 0.10152990264255911,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.725286066532135,
      "kl": 0.018332467472646385,
      "learning_rate": 9.92776406612899e-07,
      "loss": 0.0007331996224820614,
      "reward": 2.7579694747924806,
      "reward_std": 0.5351399898529052,
      "rewards/IngredientFormatReward/mean": 0.954687488079071,
      "rewards/IngredientFormatReward/std": 0.18398547321557998,
      "rewards/IngredientMatchReward/mean": 0.5577244639396668,
      "rewards/IngredientMatchReward/std": 0.28916783928871154,
      "rewards/IngredientQuantityMatchReward/mean": 0.5721200585365296,
      "rewards/IngredientQuantityMatchReward/std": 0.4316247820854187,
      "rewards/TotalKcalExactMatchReward/mean": 0.6734375,
      "rewards/TotalKcalExactMatchReward/std": 0.46652968525886535,
      "step": 365
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.6,
      "completions/mean_length": 385.4703125,
      "completions/min_length": 254.2,
      "epoch": 0.10292072322670376,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7083786129951477,
      "kl": 0.018818658008240164,
      "learning_rate": 9.923816735154415e-07,
      "loss": 0.0007529781199991703,
      "reward": 2.7624123096466064,
      "reward_std": 0.4925862789154053,
      "rewards/IngredientFormatReward/mean": 0.9794531226158142,
      "rewards/IngredientFormatReward/std": 0.1064148060977459,
      "rewards/IngredientMatchReward/mean": 0.5263802230358123,
      "rewards/IngredientMatchReward/std": 0.2988487720489502,
      "rewards/IngredientQuantityMatchReward/mean": 0.5237663745880127,
      "rewards/IngredientQuantityMatchReward/std": 0.4354355216026306,
      "rewards/TotalKcalExactMatchReward/mean": 0.7328125,
      "rewards/TotalKcalExactMatchReward/std": 0.4356215178966522,
      "step": 370
    },
    {
      "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": 0.03125,
      "completions/max_length": 508.8,
      "completions/mean_length": 393.2140625,
      "completions/min_length": 260.0,
      "epoch": 0.1043115438108484,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7111307978630066,
      "kl": 0.021116311359219254,
      "learning_rate": 9.91976523015293e-07,
      "loss": 0.0008446965366601944,
      "reward": 2.743468713760376,
      "reward_std": 0.5661642551422119,
      "rewards/IngredientFormatReward/mean": 0.9618340969085694,
      "rewards/IngredientFormatReward/std": 0.1569497775286436,
      "rewards/IngredientMatchReward/mean": 0.5765910625457764,
      "rewards/IngredientMatchReward/std": 0.30966550707817075,
      "rewards/IngredientQuantityMatchReward/mean": 0.5284810245037079,
      "rewards/IngredientQuantityMatchReward/std": 0.43139131665229796,
      "rewards/TotalKcalExactMatchReward/mean": 0.6765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4641339421272278,
      "step": 375
    },
    {
      "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": 0.0390625,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.6640625,
      "completions/min_length": 245.8,
      "epoch": 0.10570236439499305,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7693688869476318,
      "kl": 0.02106798751046881,
      "learning_rate": 9.915609636842911e-07,
      "loss": 0.0008427182212471962,
      "reward": 2.7324641227722166,
      "reward_std": 0.5566068172454834,
      "rewards/IngredientFormatReward/mean": 0.9533333301544189,
      "rewards/IngredientFormatReward/std": 0.2035890117287636,
      "rewards/IngredientMatchReward/mean": 0.5244115948677063,
      "rewards/IngredientMatchReward/std": 0.28770632147789,
      "rewards/IngredientQuantityMatchReward/mean": 0.5828442573547363,
      "rewards/IngredientQuantityMatchReward/std": 0.4409846901893616,
      "rewards/TotalKcalExactMatchReward/mean": 0.671875,
      "rewards/TotalKcalExactMatchReward/std": 0.46691768169403075,
      "step": 380
    },
    {
      "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": 0.0171875,
      "completions/max_length": 511.4,
      "completions/mean_length": 389.6796875,
      "completions/min_length": 251.2,
      "epoch": 0.1070931849791377,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6907310485839844,
      "kl": 0.020320191455539316,
      "learning_rate": 9.911350043144958e-07,
      "loss": 0.0008129674009978771,
      "reward": 2.8975246429443358,
      "reward_std": 0.4829355239868164,
      "rewards/IngredientFormatReward/mean": 0.964453125,
      "rewards/IngredientFormatReward/std": 0.14441812708973883,
      "rewards/IngredientMatchReward/mean": 0.5704482913017273,
      "rewards/IngredientMatchReward/std": 0.3032479345798492,
      "rewards/IngredientQuantityMatchReward/mean": 0.6235607147216797,
      "rewards/IngredientQuantityMatchReward/std": 0.40991583466529846,
      "rewards/TotalKcalExactMatchReward/mean": 0.7390625,
      "rewards/TotalKcalExactMatchReward/std": 0.4243670880794525,
      "step": 385
    },
    {
      "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": 0.0203125,
      "completions/max_length": 513.0,
      "completions/mean_length": 383.9171875,
      "completions/min_length": 242.2,
      "epoch": 0.10848400556328233,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7413637042045593,
      "kl": 0.019201053376309573,
      "learning_rate": 9.906986539180012e-07,
      "loss": 0.0007680667564272881,
      "reward": 2.87926664352417,
      "reward_std": 0.5338155388832092,
      "rewards/IngredientFormatReward/mean": 0.9664322853088378,
      "rewards/IngredientFormatReward/std": 0.16732385158538818,
      "rewards/IngredientMatchReward/mean": 0.540427815914154,
      "rewards/IngredientMatchReward/std": 0.30549808144569396,
      "rewards/IngredientQuantityMatchReward/mean": 0.6224065780639648,
      "rewards/IngredientQuantityMatchReward/std": 0.43633565306663513,
      "rewards/TotalKcalExactMatchReward/mean": 0.75,
      "rewards/TotalKcalExactMatchReward/std": 0.42876131534576417,
      "step": 390
    },
    {
      "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": 0.01875,
      "completions/max_length": 512.6,
      "completions/mean_length": 391.2921875,
      "completions/min_length": 259.4,
      "epoch": 0.10987482614742698,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6869736313819885,
      "kl": 0.019715036067645998,
      "learning_rate": 9.90251921726747e-07,
      "loss": 0.0007884832099080086,
      "reward": 2.91544189453125,
      "reward_std": 0.4922062695026398,
      "rewards/IngredientFormatReward/mean": 0.9751822829246521,
      "rewards/IngredientFormatReward/std": 0.13432122766971588,
      "rewards/IngredientMatchReward/mean": 0.5604154467582703,
      "rewards/IngredientMatchReward/std": 0.29352020025253295,
      "rewards/IngredientQuantityMatchReward/mean": 0.5985941112041473,
      "rewards/IngredientQuantityMatchReward/std": 0.41760724782943726,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.41098737716674805,
      "step": 395
    },
    {
      "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": 0.01875,
      "completions/max_length": 498.2,
      "completions/mean_length": 381.2296875,
      "completions/min_length": 242.6,
      "epoch": 0.11126564673157163,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7603372931480408,
      "kl": 0.02081260222475976,
      "learning_rate": 9.897948171923218e-07,
      "loss": 0.0008325819857418537,
      "reward": 2.7142991065979003,
      "reward_std": 0.5576951742172241,
      "rewards/IngredientFormatReward/mean": 0.9718750238418579,
      "rewards/IngredientFormatReward/std": 0.13910246267914772,
      "rewards/IngredientMatchReward/mean": 0.5314646005630493,
      "rewards/IngredientMatchReward/std": 0.30190695226192477,
      "rewards/IngredientQuantityMatchReward/mean": 0.4953344762325287,
      "rewards/IngredientQuantityMatchReward/std": 0.46032541394233706,
      "rewards/TotalKcalExactMatchReward/mean": 0.715625,
      "rewards/TotalKcalExactMatchReward/std": 0.4464111149311066,
      "step": 400
    },
    {
      "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": 0.0234375,
      "completions/max_length": 513.0,
      "completions/mean_length": 383.66875,
      "completions/min_length": 222.8,
      "epoch": 0.11265646731571627,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.740298867225647,
      "kl": 0.021776563371531665,
      "learning_rate": 9.893273499857642e-07,
      "loss": 0.000871221162378788,
      "reward": 2.886891746520996,
      "reward_std": 0.5212242007255554,
      "rewards/IngredientFormatReward/mean": 0.9692820072174072,
      "rewards/IngredientFormatReward/std": 0.1418784774839878,
      "rewards/IngredientMatchReward/mean": 0.5620102286338806,
      "rewards/IngredientMatchReward/std": 0.30565056800842283,
      "rewards/IngredientQuantityMatchReward/mean": 0.6118495345115662,
      "rewards/IngredientQuantityMatchReward/std": 0.42542343139648436,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.42996495962142944,
      "step": 405
    },
    {
      "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": 0.015625,
      "completions/max_length": 510.2,
      "completions/mean_length": 387.7109375,
      "completions/min_length": 241.8,
      "epoch": 0.11404728789986092,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7782815098762512,
      "kl": 0.01931304099271074,
      "learning_rate": 9.888495299973573e-07,
      "loss": 0.0007724935188889503,
      "reward": 2.9141572952270507,
      "reward_std": 0.49541053771972654,
      "rewards/IngredientFormatReward/mean": 0.9708593845367431,
      "rewards/IngredientFormatReward/std": 0.15200091004371644,
      "rewards/IngredientMatchReward/mean": 0.6145932674407959,
      "rewards/IngredientMatchReward/std": 0.27650038003921507,
      "rewards/IngredientQuantityMatchReward/mean": 0.6583922863006592,
      "rewards/IngredientQuantityMatchReward/std": 0.4239464819431305,
      "rewards/TotalKcalExactMatchReward/mean": 0.6703125,
      "rewards/TotalKcalExactMatchReward/std": 0.4658561825752258,
      "step": 410
    },
    {
      "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": 0.021875,
      "completions/max_length": 510.2,
      "completions/mean_length": 379.5578125,
      "completions/min_length": 243.4,
      "epoch": 0.11543810848400557,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7625982165336609,
      "kl": 0.021549660118762404,
      "learning_rate": 9.883613673364195e-07,
      "loss": 0.0008620643056929111,
      "reward": 2.9068836212158202,
      "reward_std": 0.5135960221290589,
      "rewards/IngredientFormatReward/mean": 0.9680729150772095,
      "rewards/IngredientFormatReward/std": 0.16229501515626907,
      "rewards/IngredientMatchReward/mean": 0.5588231742382049,
      "rewards/IngredientMatchReward/std": 0.2922717064619064,
      "rewards/IngredientQuantityMatchReward/mean": 0.6237375855445861,
      "rewards/IngredientQuantityMatchReward/std": 0.42023454904556273,
      "rewards/TotalKcalExactMatchReward/mean": 0.75625,
      "rewards/TotalKcalExactMatchReward/std": 0.428866970539093,
      "step": 415
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.4,
      "completions/mean_length": 378.0703125,
      "completions/min_length": 237.6,
      "epoch": 0.1168289290681502,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7636048793792725,
      "kl": 0.021920134348329158,
      "learning_rate": 9.878628723310913e-07,
      "loss": 0.0008768060244619847,
      "reward": 2.9490123748779298,
      "reward_std": 0.5071363091468811,
      "rewards/IngredientFormatReward/mean": 0.971484386920929,
      "rewards/IngredientFormatReward/std": 0.13410393372178078,
      "rewards/IngredientMatchReward/mean": 0.5426004528999329,
      "rewards/IngredientMatchReward/std": 0.29528992176055907,
      "rewards/IngredientQuantityMatchReward/mean": 0.6427400946617127,
      "rewards/IngredientQuantityMatchReward/std": 0.41394662857055664,
      "rewards/TotalKcalExactMatchReward/mean": 0.7921875,
      "rewards/TotalKcalExactMatchReward/std": 0.3955236434936523,
      "step": 420
    },
    {
      "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": 0.03125,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.3625,
      "completions/min_length": 252.2,
      "epoch": 0.11821974965229486,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7036556005477905,
      "kl": 0.021358940622303636,
      "learning_rate": 9.87354055528116e-07,
      "loss": 0.00085435900837183,
      "reward": 2.8730720043182374,
      "reward_std": 0.5078805685043335,
      "rewards/IngredientFormatReward/mean": 0.9601302146911621,
      "rewards/IngredientFormatReward/std": 0.18644485622644424,
      "rewards/IngredientMatchReward/mean": 0.5474907159805298,
      "rewards/IngredientMatchReward/std": 0.2955142557621002,
      "rewards/IngredientQuantityMatchReward/mean": 0.6092010855674743,
      "rewards/IngredientQuantityMatchReward/std": 0.4215419411659241,
      "rewards/TotalKcalExactMatchReward/mean": 0.75625,
      "rewards/TotalKcalExactMatchReward/std": 0.4189519166946411,
      "step": 425
    },
    {
      "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": 0.0171875,
      "completions/max_length": 512.4,
      "completions/mean_length": 394.6328125,
      "completions/min_length": 259.0,
      "epoch": 0.11961057023643949,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7593802809715271,
      "kl": 0.020391820115037262,
      "learning_rate": 9.868349276926173e-07,
      "loss": 0.000815667025744915,
      "reward": 3.0347278118133545,
      "reward_std": 0.4388155698776245,
      "rewards/IngredientFormatReward/mean": 0.9790736675262451,
      "rewards/IngredientFormatReward/std": 0.12180586419999599,
      "rewards/IngredientMatchReward/mean": 0.5882833242416382,
      "rewards/IngredientMatchReward/std": 0.28604965209960936,
      "rewards/IngredientQuantityMatchReward/mean": 0.687683379650116,
      "rewards/IngredientQuantityMatchReward/std": 0.39389697313308714,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.4065669775009155,
      "step": 430
    },
    {
      "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": 0.0296875,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.275,
      "completions/min_length": 262.0,
      "epoch": 0.12100139082058414,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7075633406639099,
      "kl": 0.0222692629089579,
      "learning_rate": 9.86305499807871e-07,
      "loss": 0.0008907708339393139,
      "reward": 2.8322206020355223,
      "reward_std": 0.45790958404541016,
      "rewards/IngredientFormatReward/mean": 0.9591406226158142,
      "rewards/IngredientFormatReward/std": 0.16454651802778245,
      "rewards/IngredientMatchReward/mean": 0.5731777012348175,
      "rewards/IngredientMatchReward/std": 0.2920497000217438,
      "rewards/IngredientQuantityMatchReward/mean": 0.5842772364616394,
      "rewards/IngredientQuantityMatchReward/std": 0.4166082203388214,
      "rewards/TotalKcalExactMatchReward/mean": 0.715625,
      "rewards/TotalKcalExactMatchReward/std": 0.43869020938873293,
      "step": 435
    },
    {
      "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": 0.0171875,
      "completions/max_length": 502.6,
      "completions/mean_length": 388.5390625,
      "completions/min_length": 224.8,
      "epoch": 0.12239221140472879,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7498156428337097,
      "kl": 0.021729087352287024,
      "learning_rate": 9.857657830750726e-07,
      "loss": 0.0008691315539181233,
      "reward": 2.9173073768615723,
      "reward_std": 0.469376403093338,
      "rewards/IngredientFormatReward/mean": 0.9806249976158142,
      "rewards/IngredientFormatReward/std": 0.10909082591533661,
      "rewards/IngredientMatchReward/mean": 0.5659883618354797,
      "rewards/IngredientMatchReward/std": 0.28756594061851504,
      "rewards/IngredientQuantityMatchReward/mean": 0.6035065531730652,
      "rewards/IngredientQuantityMatchReward/std": 0.4320640921592712,
      "rewards/TotalKcalExactMatchReward/mean": 0.7671875,
      "rewards/TotalKcalExactMatchReward/std": 0.4184816122055054,
      "step": 440
    },
    {
      "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": 0.03125,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.8625,
      "completions/min_length": 271.8,
      "epoch": 0.12378303198887343,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7427554130554199,
      "kl": 0.022438473138026892,
      "learning_rate": 9.852157889131007e-07,
      "loss": 0.0008975569158792496,
      "reward": 2.7711973667144774,
      "reward_std": 0.5837087631225586,
      "rewards/IngredientFormatReward/mean": 0.9584747195243836,
      "rewards/IngredientFormatReward/std": 0.19273171424865723,
      "rewards/IngredientMatchReward/mean": 0.5494079709053039,
      "rewards/IngredientMatchReward/std": 0.2919416844844818,
      "rewards/IngredientQuantityMatchReward/mean": 0.588314664363861,
      "rewards/IngredientQuantityMatchReward/std": 0.42805092930793764,
      "rewards/TotalKcalExactMatchReward/mean": 0.675,
      "rewards/TotalKcalExactMatchReward/std": 0.4670551478862762,
      "step": 445
    },
    {
      "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": 0.05,
      "completions/max_length": 513.0,
      "completions/mean_length": 400.6421875,
      "completions/min_length": 252.4,
      "epoch": 0.12517385257301808,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7527221441268921,
      "kl": 0.025158376456238328,
      "learning_rate": 9.846555289582756e-07,
      "loss": 0.0010063996538519858,
      "reward": 2.800438404083252,
      "reward_std": 0.5684214293956756,
      "rewards/IngredientFormatReward/mean": 0.9457663774490357,
      "rewards/IngredientFormatReward/std": 0.20938537269830704,
      "rewards/IngredientMatchReward/mean": 0.5764552235603333,
      "rewards/IngredientMatchReward/std": 0.31859682202339173,
      "rewards/IngredientQuantityMatchReward/mean": 0.6110292553901673,
      "rewards/IngredientQuantityMatchReward/std": 0.4231915295124054,
      "rewards/TotalKcalExactMatchReward/mean": 0.6671875,
      "rewards/TotalKcalExactMatchReward/std": 0.46875752210617067,
      "step": 450
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 389.359375,
      "completions/min_length": 236.0,
      "epoch": 0.12656467315716272,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7649770975112915,
      "kl": 0.023254029813688248,
      "learning_rate": 9.840850150641117e-07,
      "loss": 0.0009299272671341896,
      "reward": 2.8438170909881593,
      "reward_std": 0.48624941110610964,
      "rewards/IngredientFormatReward/mean": 0.9634375095367431,
      "rewards/IngredientFormatReward/std": 0.1753060534596443,
      "rewards/IngredientMatchReward/mean": 0.6139800667762756,
      "rewards/IngredientMatchReward/std": 0.30183743834495547,
      "rewards/IngredientQuantityMatchReward/mean": 0.525774621963501,
      "rewards/IngredientQuantityMatchReward/std": 0.42514413595199585,
      "rewards/TotalKcalExactMatchReward/mean": 0.740625,
      "rewards/TotalKcalExactMatchReward/std": 0.43582022190093994,
      "step": 455
    },
    {
      "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": 0.0515625,
      "completions/max_length": 513.0,
      "completions/mean_length": 403.421875,
      "completions/min_length": 276.4,
      "epoch": 0.12795549374130738,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 1.5184186697006226,
      "kl": 0.027110138500574976,
      "learning_rate": 9.835042593010686e-07,
      "loss": 0.001084165833890438,
      "reward": 2.839724397659302,
      "reward_std": 0.5796327233314514,
      "rewards/IngredientFormatReward/mean": 0.9483072996139527,
      "rewards/IngredientFormatReward/std": 0.20597576946020127,
      "rewards/IngredientMatchReward/mean": 0.5643192052841186,
      "rewards/IngredientMatchReward/std": 0.3047496736049652,
      "rewards/IngredientQuantityMatchReward/mean": 0.6192853331565857,
      "rewards/IngredientQuantityMatchReward/std": 0.421432638168335,
      "rewards/TotalKcalExactMatchReward/mean": 0.7078125,
      "rewards/TotalKcalExactMatchReward/std": 0.4554017841815948,
      "step": 460
    },
    {
      "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": 0.0171875,
      "completions/max_length": 508.2,
      "completions/mean_length": 388.903125,
      "completions/min_length": 250.0,
      "epoch": 0.12934631432545202,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6998506784439087,
      "kl": 0.023188903753180057,
      "learning_rate": 9.829132739562948e-07,
      "loss": 0.0009276852011680603,
      "reward": 2.9770085334777834,
      "reward_std": 0.5292151272296906,
      "rewards/IngredientFormatReward/mean": 0.9733333349227905,
      "rewards/IngredientFormatReward/std": 0.13938566744327546,
      "rewards/IngredientMatchReward/mean": 0.6149696111679077,
      "rewards/IngredientMatchReward/std": 0.3145597755908966,
      "rewards/IngredientQuantityMatchReward/mean": 0.6762055993080139,
      "rewards/IngredientQuantityMatchReward/std": 0.4129923641681671,
      "rewards/TotalKcalExactMatchReward/mean": 0.7125,
      "rewards/TotalKcalExactMatchReward/std": 0.453085857629776,
      "step": 465
    },
    {
      "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": 0.040625,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.6515625,
      "completions/min_length": 231.8,
      "epoch": 0.13073713490959665,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7366269826889038,
      "kl": 0.02417063304455951,
      "learning_rate": 9.823120715333675e-07,
      "loss": 0.0009667634963989258,
      "reward": 2.8089338302612306,
      "reward_std": 0.5877117991447449,
      "rewards/IngredientFormatReward/mean": 0.9498549342155457,
      "rewards/IngredientFormatReward/std": 0.20365288555622102,
      "rewards/IngredientMatchReward/mean": 0.5744320631027222,
      "rewards/IngredientMatchReward/std": 0.305409836769104,
      "rewards/IngredientQuantityMatchReward/mean": 0.5783968687057495,
      "rewards/IngredientQuantityMatchReward/std": 0.4373714983463287,
      "rewards/TotalKcalExactMatchReward/mean": 0.70625,
      "rewards/TotalKcalExactMatchReward/std": 0.4441069960594177,
      "step": 470
    },
    {
      "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": 0.015625,
      "completions/max_length": 512.4,
      "completions/mean_length": 392.5515625,
      "completions/min_length": 270.8,
      "epoch": 0.13212795549374132,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7693325281143188,
      "kl": 0.022445485449861736,
      "learning_rate": 9.817006647520284e-07,
      "loss": 0.0008979140780866146,
      "reward": 2.906987428665161,
      "reward_std": 0.4892770886421204,
      "rewards/IngredientFormatReward/mean": 0.9798437476158142,
      "rewards/IngredientFormatReward/std": 0.11782345212996007,
      "rewards/IngredientMatchReward/mean": 0.5608668208122254,
      "rewards/IngredientMatchReward/std": 0.29060490131378175,
      "rewards/IngredientQuantityMatchReward/mean": 0.635026752948761,
      "rewards/IngredientQuantityMatchReward/std": 0.4279853343963623,
      "rewards/TotalKcalExactMatchReward/mean": 0.73125,
      "rewards/TotalKcalExactMatchReward/std": 0.4427589774131775,
      "step": 475
    },
    {
      "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": 0.034375,
      "completions/max_length": 513.0,
      "completions/mean_length": 400.5890625,
      "completions/min_length": 251.4,
      "epoch": 0.13351877607788595,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7207197546958923,
      "kl": 0.023216928401961923,
      "learning_rate": 9.810790665479146e-07,
      "loss": 0.0009287634864449501,
      "reward": 2.842219924926758,
      "reward_std": 0.46143017411231996,
      "rewards/IngredientFormatReward/mean": 0.9690104246139526,
      "rewards/IngredientFormatReward/std": 0.16153716892004014,
      "rewards/IngredientMatchReward/mean": 0.5861247420310974,
      "rewards/IngredientMatchReward/std": 0.2880786955356598,
      "rewards/IngredientQuantityMatchReward/mean": 0.5964597225189209,
      "rewards/IngredientQuantityMatchReward/std": 0.42537760734558105,
      "rewards/TotalKcalExactMatchReward/mean": 0.690625,
      "rewards/TotalKcalExactMatchReward/std": 0.44918020367622374,
      "step": 480
    },
    {
      "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": 0.009375,
      "completions/max_length": 497.4,
      "completions/mean_length": 390.6265625,
      "completions/min_length": 247.2,
      "epoch": 0.1349095966620306,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7332528829574585,
      "kl": 0.024050121754407884,
      "learning_rate": 9.80447290072285e-07,
      "loss": 0.0009621541015803814,
      "reward": 2.9791000843048097,
      "reward_std": 0.5197779953479766,
      "rewards/IngredientFormatReward/mean": 0.9775520920753479,
      "rewards/IngredientFormatReward/std": 0.11693610101938248,
      "rewards/IngredientMatchReward/mean": 0.6081776976585388,
      "rewards/IngredientMatchReward/std": 0.31099793314933777,
      "rewards/IngredientQuantityMatchReward/mean": 0.6293077826499939,
      "rewards/IngredientQuantityMatchReward/std": 0.4286587595939636,
      "rewards/TotalKcalExactMatchReward/mean": 0.7640625,
      "rewards/TotalKcalExactMatchReward/std": 0.42141618132591246,
      "step": 485
    },
    {
      "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": 0.0328125,
      "completions/max_length": 512.8,
      "completions/mean_length": 400.8421875,
      "completions/min_length": 268.2,
      "epoch": 0.13630041724617525,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.706557035446167,
      "kl": 0.02498110675951466,
      "learning_rate": 9.798053486917416e-07,
      "loss": 0.000999157875776291,
      "reward": 2.904594373703003,
      "reward_std": 0.5024032175540925,
      "rewards/IngredientFormatReward/mean": 0.9638020753860473,
      "rewards/IngredientFormatReward/std": 0.17550685703754426,
      "rewards/IngredientMatchReward/mean": 0.5668285012245178,
      "rewards/IngredientMatchReward/std": 0.29408658146858213,
      "rewards/IngredientQuantityMatchReward/mean": 0.6130263328552246,
      "rewards/IngredientQuantityMatchReward/std": 0.42733028531074524,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.42357649803161623,
      "step": 490
    },
    {
      "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": 0.0140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.090625,
      "completions/min_length": 258.2,
      "epoch": 0.1376912378303199,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.701976478099823,
      "kl": 0.023832752835005522,
      "learning_rate": 9.791532559879475e-07,
      "loss": 0.000953151099383831,
      "reward": 2.8906358242034913,
      "reward_std": 0.5113727450370789,
      "rewards/IngredientFormatReward/mean": 0.9754389882087707,
      "rewards/IngredientFormatReward/std": 0.14507034569978713,
      "rewards/IngredientMatchReward/mean": 0.5291600346565246,
      "rewards/IngredientMatchReward/std": 0.2910259008407593,
      "rewards/IngredientQuantityMatchReward/mean": 0.5954118013381958,
      "rewards/IngredientQuantityMatchReward/std": 0.42795549631118773,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.38760373592376707,
      "step": 495
    },
    {
      "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": 0.025,
      "completions/max_length": 512.6,
      "completions/mean_length": 397.425,
      "completions/min_length": 252.0,
      "epoch": 0.13908205841446453,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7409101128578186,
      "kl": 0.02587097561918199,
      "learning_rate": 9.784910257573383e-07,
      "loss": 0.0010347630828619002,
      "reward": 2.857800817489624,
      "reward_std": 0.4655668616294861,
      "rewards/IngredientFormatReward/mean": 0.9700000047683716,
      "rewards/IngredientFormatReward/std": 0.1601935938000679,
      "rewards/IngredientMatchReward/mean": 0.601234495639801,
      "rewards/IngredientMatchReward/std": 0.2889039397239685,
      "rewards/IngredientQuantityMatchReward/mean": 0.5396912932395935,
      "rewards/IngredientQuantityMatchReward/std": 0.44021721482276915,
      "rewards/TotalKcalExactMatchReward/mean": 0.746875,
      "rewards/TotalKcalExactMatchReward/std": 0.43169044852256777,
      "step": 500
    },
    {
      "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": 0.0109375,
      "completions/max_length": 508.8,
      "completions/mean_length": 387.69375,
      "completions/min_length": 247.6,
      "epoch": 0.1404728789986092,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7838031053543091,
      "kl": 0.02346238095778972,
      "learning_rate": 9.778186720108318e-07,
      "loss": 0.0009385457262396813,
      "reward": 2.9772964477539063,
      "reward_std": 0.4686002552509308,
      "rewards/IngredientFormatReward/mean": 0.9853124976158142,
      "rewards/IngredientFormatReward/std": 0.09816802144050599,
      "rewards/IngredientMatchReward/mean": 0.5914366364479064,
      "rewards/IngredientMatchReward/std": 0.3054099023342133,
      "rewards/IngredientQuantityMatchReward/mean": 0.5958597302436829,
      "rewards/IngredientQuantityMatchReward/std": 0.4336180329322815,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.38523083329200747,
      "step": 505
    },
    {
      "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": 0.0140625,
      "completions/max_length": 506.6,
      "completions/mean_length": 394.96875,
      "completions/min_length": 266.4,
      "epoch": 0.14186369958275383,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6723412275314331,
      "kl": 0.02323896362213418,
      "learning_rate": 9.771362089735307e-07,
      "loss": 0.0009296289645135403,
      "reward": 2.9428701400756836,
      "reward_std": 0.42799751162528993,
      "rewards/IngredientFormatReward/mean": 0.9716517925262451,
      "rewards/IngredientFormatReward/std": 0.14892885833978653,
      "rewards/IngredientMatchReward/mean": 0.5667912840843201,
      "rewards/IngredientMatchReward/std": 0.2957393884658813,
      "rewards/IngredientQuantityMatchReward/mean": 0.6091145873069763,
      "rewards/IngredientQuantityMatchReward/std": 0.44083576202392577,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.392047244310379,
      "step": 510
    },
    {
      "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": 0.0234375,
      "completions/max_length": 510.6,
      "completions/mean_length": 389.0640625,
      "completions/min_length": 264.2,
      "epoch": 0.14325452016689846,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6448062658309937,
      "kl": 0.02515047767665237,
      "learning_rate": 9.76443651084421e-07,
      "loss": 0.0010061558336019517,
      "reward": 2.700420045852661,
      "reward_std": 0.4924412965774536,
      "rewards/IngredientFormatReward/mean": 0.9638020992279053,
      "rewards/IngredientFormatReward/std": 0.163498455286026,
      "rewards/IngredientMatchReward/mean": 0.5357006430625916,
      "rewards/IngredientMatchReward/std": 0.3113968729972839,
      "rewards/IngredientQuantityMatchReward/mean": 0.4884172558784485,
      "rewards/IngredientQuantityMatchReward/std": 0.44572351574897767,
      "rewards/TotalKcalExactMatchReward/mean": 0.7125,
      "rewards/TotalKcalExactMatchReward/std": 0.4334396183490753,
      "step": 515
    },
    {
      "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": 0.0171875,
      "completions/max_length": 504.4,
      "completions/mean_length": 389.028125,
      "completions/min_length": 240.6,
      "epoch": 0.14464534075104313,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6788337230682373,
      "kl": 0.024838618678040804,
      "learning_rate": 9.757410129960675e-07,
      "loss": 0.0009936664253473281,
      "reward": 2.8270545482635496,
      "reward_std": 0.48341248631477357,
      "rewards/IngredientFormatReward/mean": 0.9769010424613953,
      "rewards/IngredientFormatReward/std": 0.13005183711647988,
      "rewards/IngredientMatchReward/mean": 0.5910993456840515,
      "rewards/IngredientMatchReward/std": 0.28249636888504026,
      "rewards/IngredientQuantityMatchReward/mean": 0.5684291124343872,
      "rewards/IngredientQuantityMatchReward/std": 0.41839559674263,
      "rewards/TotalKcalExactMatchReward/mean": 0.690625,
      "rewards/TotalKcalExactMatchReward/std": 0.46186931133270265,
      "step": 520
    },
    {
      "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": 0.025,
      "completions/max_length": 507.8,
      "completions/mean_length": 395.584375,
      "completions/min_length": 251.8,
      "epoch": 0.14603616133518776,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6823793649673462,
      "kl": 0.023881654208526015,
      "learning_rate": 9.750283095743037e-07,
      "loss": 0.0009551938623189926,
      "reward": 2.8670840740203856,
      "reward_std": 0.45495287179946897,
      "rewards/IngredientFormatReward/mean": 0.9732961416244507,
      "rewards/IngredientFormatReward/std": 0.12547686100006103,
      "rewards/IngredientMatchReward/mean": 0.5820461392402649,
      "rewards/IngredientMatchReward/std": 0.293103963136673,
      "rewards/IngredientQuantityMatchReward/mean": 0.5586167931556701,
      "rewards/IngredientQuantityMatchReward/std": 0.43080027103424073,
      "rewards/TotalKcalExactMatchReward/mean": 0.753125,
      "rewards/TotalKcalExactMatchReward/std": 0.4183549165725708,
      "step": 525
    },
    {
      "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": 0.0109375,
      "completions/max_length": 505.2,
      "completions/mean_length": 390.6859375,
      "completions/min_length": 241.6,
      "epoch": 0.1474269819193324,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6847018599510193,
      "kl": 0.026205667469184846,
      "learning_rate": 9.74305555897917e-07,
      "loss": 0.0010480815544724464,
      "reward": 2.9751473903656005,
      "reward_std": 0.45612571835517884,
      "rewards/IngredientFormatReward/mean": 0.9853515863418579,
      "rewards/IngredientFormatReward/std": 0.08599194474518299,
      "rewards/IngredientMatchReward/mean": 0.6294636487960815,
      "rewards/IngredientMatchReward/std": 0.2975753009319305,
      "rewards/IngredientQuantityMatchReward/mean": 0.5993947505950927,
      "rewards/IngredientQuantityMatchReward/std": 0.4184840977191925,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.42358742356300355,
      "step": 530
    },
    {
      "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": 0.0203125,
      "completions/max_length": 513.0,
      "completions/mean_length": 386.5671875,
      "completions/min_length": 250.2,
      "epoch": 0.14881780250347706,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7143814563751221,
      "kl": 0.025549968774430453,
      "learning_rate": 9.735727672583297e-07,
      "loss": 0.0010221581906080246,
      "reward": 3.0114527225494383,
      "reward_std": 0.4811578273773193,
      "rewards/IngredientFormatReward/mean": 0.9715104222297668,
      "rewards/IngredientFormatReward/std": 0.15780526846647264,
      "rewards/IngredientMatchReward/mean": 0.5821434855461121,
      "rewards/IngredientMatchReward/std": 0.3114396154880524,
      "rewards/IngredientQuantityMatchReward/mean": 0.6687362551689148,
      "rewards/IngredientQuantityMatchReward/std": 0.4153988003730774,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.4070970892906189,
      "step": 535
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.7578125,
      "completions/min_length": 277.4,
      "epoch": 0.1502086230876217,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7409384250640869,
      "kl": 0.025168495741672813,
      "learning_rate": 9.728299591592752e-07,
      "loss": 0.0010109595954418183,
      "reward": 2.85003867149353,
      "reward_std": 0.43052482008934023,
      "rewards/IngredientFormatReward/mean": 0.983802080154419,
      "rewards/IngredientFormatReward/std": 0.11858281642198562,
      "rewards/IngredientMatchReward/mean": 0.5917337536811829,
      "rewards/IngredientMatchReward/std": 0.2788821399211884,
      "rewards/IngredientQuantityMatchReward/mean": 0.5541902899742126,
      "rewards/IngredientQuantityMatchReward/std": 0.4308002769947052,
      "rewards/TotalKcalExactMatchReward/mean": 0.7203125,
      "rewards/TotalKcalExactMatchReward/std": 0.4451679825782776,
      "step": 540
    },
    {
      "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": 0.01875,
      "completions/max_length": 508.4,
      "completions/mean_length": 397.5484375,
      "completions/min_length": 261.0,
      "epoch": 0.15159944367176634,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7681187987327576,
      "kl": 0.02836602572351694,
      "learning_rate": 9.72077147316471e-07,
      "loss": 0.0011343616992235185,
      "reward": 2.882566547393799,
      "reward_std": 0.49046033024787905,
      "rewards/IngredientFormatReward/mean": 0.9757552146911621,
      "rewards/IngredientFormatReward/std": 0.13507271111011504,
      "rewards/IngredientMatchReward/mean": 0.5827213525772095,
      "rewards/IngredientMatchReward/std": 0.2956298977136612,
      "rewards/IngredientQuantityMatchReward/mean": 0.6037775635719299,
      "rewards/IngredientQuantityMatchReward/std": 0.42004692554473877,
      "rewards/TotalKcalExactMatchReward/mean": 0.7203125,
      "rewards/TotalKcalExactMatchReward/std": 0.443309360742569,
      "step": 545
    },
    {
      "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": 0.0078125,
      "completions/max_length": 504.0,
      "completions/mean_length": 387.8671875,
      "completions/min_length": 250.6,
      "epoch": 0.15299026425591097,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7030057907104492,
      "kl": 0.026008617551997303,
      "learning_rate": 9.71314347657285e-07,
      "loss": 0.0010403728112578392,
      "reward": 2.8892337799072267,
      "reward_std": 0.48583821058273313,
      "rewards/IngredientFormatReward/mean": 0.9875,
      "rewards/IngredientFormatReward/std": 0.09785700291395187,
      "rewards/IngredientMatchReward/mean": 0.5934883534908295,
      "rewards/IngredientMatchReward/std": 0.28317511677742,
      "rewards/IngredientQuantityMatchReward/mean": 0.5379329025745392,
      "rewards/IngredientQuantityMatchReward/std": 0.4267972707748413,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.4082271695137024,
      "step": 550
    },
    {
      "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": 0.0390625,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.6796875,
      "completions/min_length": 256.6,
      "epoch": 0.15438108484005564,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7302446365356445,
      "kl": 0.02743447886314243,
      "learning_rate": 9.705415763203992e-07,
      "loss": 0.0010974638164043427,
      "reward": 2.793476915359497,
      "reward_std": 0.5167571306228638,
      "rewards/IngredientFormatReward/mean": 0.9539973974227905,
      "rewards/IngredientFormatReward/std": 0.1849471166729927,
      "rewards/IngredientMatchReward/mean": 0.5718799471855164,
      "rewards/IngredientMatchReward/std": 0.29791882634162903,
      "rewards/IngredientQuantityMatchReward/mean": 0.5785369992256164,
      "rewards/IngredientQuantityMatchReward/std": 0.4268043220043182,
      "rewards/TotalKcalExactMatchReward/mean": 0.6890625,
      "rewards/TotalKcalExactMatchReward/std": 0.4540803492069244,
      "step": 555
    },
    {
      "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": 0.03125,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.6734375,
      "completions/min_length": 274.4,
      "epoch": 0.15577190542420027,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6885735392570496,
      "kl": 0.025130884489044547,
      "learning_rate": 9.697588496554677e-07,
      "loss": 0.0010053345933556557,
      "reward": 2.993966054916382,
      "reward_std": 0.5204974055290222,
      "rewards/IngredientFormatReward/mean": 0.9615625023841858,
      "rewards/IngredientFormatReward/std": 0.17939773947000504,
      "rewards/IngredientMatchReward/mean": 0.6134548664093018,
      "rewards/IngredientMatchReward/std": 0.2914328217506409,
      "rewards/IngredientQuantityMatchReward/mean": 0.6142611861228943,
      "rewards/IngredientQuantityMatchReward/std": 0.4277330875396729,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.39028285145759584,
      "step": 560
    },
    {
      "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": 0.0125,
      "completions/max_length": 505.4,
      "completions/mean_length": 389.209375,
      "completions/min_length": 267.0,
      "epoch": 0.1571627260083449,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7373332381248474,
      "kl": 0.02833310468122363,
      "learning_rate": 9.689661842227719e-07,
      "loss": 0.0011331411078572273,
      "reward": 2.9521562576293947,
      "reward_std": 0.4717487573623657,
      "rewards/IngredientFormatReward/mean": 0.9821354150772095,
      "rewards/IngredientFormatReward/std": 0.1252045825123787,
      "rewards/IngredientMatchReward/mean": 0.5781901121139527,
      "rewards/IngredientMatchReward/std": 0.287116551399231,
      "rewards/IngredientQuantityMatchReward/mean": 0.6090183138847352,
      "rewards/IngredientQuantityMatchReward/std": 0.4261681318283081,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.40791478753089905,
      "step": 565
    },
    {
      "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": 0.01875,
      "completions/max_length": 509.4,
      "completions/mean_length": 389.15625,
      "completions/min_length": 257.8,
      "epoch": 0.15855354659248957,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7094981074333191,
      "kl": 0.026204002369195224,
      "learning_rate": 9.681635967928686e-07,
      "loss": 0.001048055663704872,
      "reward": 2.8415787696838377,
      "reward_std": 0.46802008152008057,
      "rewards/IngredientFormatReward/mean": 0.9778348207473755,
      "rewards/IngredientFormatReward/std": 0.12536680959165097,
      "rewards/IngredientMatchReward/mean": 0.5831461310386657,
      "rewards/IngredientMatchReward/std": 0.3007568359375,
      "rewards/IngredientQuantityMatchReward/mean": 0.5384102821350097,
      "rewards/IngredientQuantityMatchReward/std": 0.45931328535079957,
      "rewards/TotalKcalExactMatchReward/mean": 0.7421875,
      "rewards/TotalKcalExactMatchReward/std": 0.4242383182048798,
      "step": 570
    },
    {
      "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": 0.0234375,
      "completions/max_length": 509.2,
      "completions/mean_length": 389.178125,
      "completions/min_length": 263.2,
      "epoch": 0.1599443671766342,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.686071515083313,
      "kl": 0.02771181467687711,
      "learning_rate": 9.673511043462365e-07,
      "loss": 0.0011084911413490773,
      "reward": 2.981544256210327,
      "reward_std": 0.4777001917362213,
      "rewards/IngredientFormatReward/mean": 0.971484375,
      "rewards/IngredientFormatReward/std": 0.14188916683197023,
      "rewards/IngredientMatchReward/mean": 0.6547222256660461,
      "rewards/IngredientMatchReward/std": 0.29094510078430175,
      "rewards/IngredientQuantityMatchReward/mean": 0.5881502151489257,
      "rewards/IngredientQuantityMatchReward/std": 0.4195889294147491,
      "rewards/TotalKcalExactMatchReward/mean": 0.7671875,
      "rewards/TotalKcalExactMatchReward/std": 0.42057693004608154,
      "step": 575
    },
    {
      "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": 0.0234375,
      "completions/max_length": 513.0,
      "completions/mean_length": 388.5984375,
      "completions/min_length": 261.8,
      "epoch": 0.16133518776077885,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6606355309486389,
      "kl": 0.027671499620191754,
      "learning_rate": 9.665287240729166e-07,
      "loss": 0.0011070670560002327,
      "reward": 2.8866348266601562,
      "reward_std": 0.48432177901268003,
      "rewards/IngredientFormatReward/mean": 0.969348955154419,
      "rewards/IngredientFormatReward/std": 0.16037327349185942,
      "rewards/IngredientMatchReward/mean": 0.5834052681922912,
      "rewards/IngredientMatchReward/std": 0.2894372344017029,
      "rewards/IngredientQuantityMatchReward/mean": 0.6104430437088013,
      "rewards/IngredientQuantityMatchReward/std": 0.42995097041130065,
      "rewards/TotalKcalExactMatchReward/mean": 0.7234375,
      "rewards/TotalKcalExactMatchReward/std": 0.43584696054458616,
      "step": 580
    },
    {
      "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": 0.0359375,
      "completions/max_length": 509.2,
      "completions/mean_length": 392.1859375,
      "completions/min_length": 256.4,
      "epoch": 0.1627260083449235,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6764155626296997,
      "kl": 0.029804628575220704,
      "learning_rate": 9.656964733721475e-07,
      "loss": 0.0011925380676984787,
      "reward": 2.829722595214844,
      "reward_std": 0.5019806742668151,
      "rewards/IngredientFormatReward/mean": 0.9627343773841858,
      "rewards/IngredientFormatReward/std": 0.16002234518527986,
      "rewards/IngredientMatchReward/mean": 0.569929325580597,
      "rewards/IngredientMatchReward/std": 0.2886444807052612,
      "rewards/IngredientQuantityMatchReward/mean": 0.5314339637756348,
      "rewards/IngredientQuantityMatchReward/std": 0.42082526087760924,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4208135664463043,
      "step": 585
    },
    {
      "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": 0.0078125,
      "completions/max_length": 504.8,
      "completions/mean_length": 387.7296875,
      "completions/min_length": 246.6,
      "epoch": 0.16411682892906815,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7083275318145752,
      "kl": 0.02589824195019901,
      "learning_rate": 9.648543698519991e-07,
      "loss": 0.00103579331189394,
      "reward": 3.0169702053070067,
      "reward_std": 0.3853384256362915,
      "rewards/IngredientFormatReward/mean": 0.9901041507720947,
      "rewards/IngredientFormatReward/std": 0.06783321052789688,
      "rewards/IngredientMatchReward/mean": 0.613008451461792,
      "rewards/IngredientMatchReward/std": 0.3062650740146637,
      "rewards/IngredientQuantityMatchReward/mean": 0.648232614994049,
      "rewards/IngredientQuantityMatchReward/std": 0.4172336578369141,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4179598748683929,
      "step": 590
    },
    {
      "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": 0.0125,
      "completions/max_length": 500.0,
      "completions/mean_length": 388.96875,
      "completions/min_length": 222.4,
      "epoch": 0.16550764951321278,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6607844233512878,
      "kl": 0.02781644803471863,
      "learning_rate": 9.64002431328998e-07,
      "loss": 0.0011126678436994553,
      "reward": 2.947980499267578,
      "reward_std": 0.45839241743087766,
      "rewards/IngredientFormatReward/mean": 0.9763541698455811,
      "rewards/IngredientFormatReward/std": 0.1387864574790001,
      "rewards/IngredientMatchReward/mean": 0.5945783078670501,
      "rewards/IngredientMatchReward/std": 0.3033637821674347,
      "rewards/IngredientQuantityMatchReward/mean": 0.6051730632781982,
      "rewards/IngredientQuantityMatchReward/std": 0.4249738872051239,
      "rewards/TotalKcalExactMatchReward/mean": 0.771875,
      "rewards/TotalKcalExactMatchReward/std": 0.4144205808639526,
      "step": 595
    },
    {
      "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": 0.021875,
      "completions/max_length": 505.6,
      "completions/mean_length": 397.390625,
      "completions/min_length": 273.6,
      "epoch": 0.16689847009735745,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7403410077095032,
      "kl": 0.027860931726172565,
      "learning_rate": 9.63140675827753e-07,
      "loss": 0.0011144131422042847,
      "reward": 2.9059377670288087,
      "reward_std": 0.42294371128082275,
      "rewards/IngredientFormatReward/mean": 0.9762500047683715,
      "rewards/IngredientFormatReward/std": 0.1115086704492569,
      "rewards/IngredientMatchReward/mean": 0.5822805166244507,
      "rewards/IngredientMatchReward/std": 0.30057679414749144,
      "rewards/IngredientQuantityMatchReward/mean": 0.5958447217941284,
      "rewards/IngredientQuantityMatchReward/std": 0.42643269896507263,
      "rewards/TotalKcalExactMatchReward/mean": 0.7515625,
      "rewards/TotalKcalExactMatchReward/std": 0.42623581290245055,
      "step": 600
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.0,
      "completions/mean_length": 391.7703125,
      "completions/min_length": 271.6,
      "epoch": 0.16828929068150209,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6867882609367371,
      "kl": 1.017303869058378,
      "learning_rate": 9.62269121580571e-07,
      "loss": 0.040814584493637084,
      "reward": 2.9245573043823243,
      "reward_std": 0.40745121240615845,
      "rewards/IngredientFormatReward/mean": 0.9918750047683715,
      "rewards/IngredientFormatReward/std": 0.0676910012960434,
      "rewards/IngredientMatchReward/mean": 0.569877827167511,
      "rewards/IngredientMatchReward/std": 0.2939007043838501,
      "rewards/IngredientQuantityMatchReward/mean": 0.5737420082092285,
      "rewards/IngredientQuantityMatchReward/std": 0.4358260929584503,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.3946849763393402,
      "step": 605
    },
    {
      "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": 0.021875,
      "completions/max_length": 512.0,
      "completions/mean_length": 393.0453125,
      "completions/min_length": 263.8,
      "epoch": 0.16968011126564672,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.69991534948349,
      "kl": 0.027150334184989335,
      "learning_rate": 9.613877870270734e-07,
      "loss": 0.0010859736241400242,
      "reward": 2.950673246383667,
      "reward_std": 0.46758312582969663,
      "rewards/IngredientFormatReward/mean": 0.9743489623069763,
      "rewards/IngredientFormatReward/std": 0.1289362832903862,
      "rewards/IngredientMatchReward/mean": 0.6121106266975402,
      "rewards/IngredientMatchReward/std": 0.2894917517900467,
      "rewards/IngredientQuantityMatchReward/mean": 0.5939012289047241,
      "rewards/IngredientQuantityMatchReward/std": 0.44014000296592715,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.4208824336528778,
      "step": 610
    },
    {
      "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": 0.0171875,
      "completions/max_length": 510.2,
      "completions/mean_length": 390.50625,
      "completions/min_length": 252.6,
      "epoch": 0.17107093184979139,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7449471950531006,
      "kl": 0.02868313742801547,
      "learning_rate": 9.60496690813805e-07,
      "loss": 0.0011473964899778367,
      "reward": 2.9799468994140623,
      "reward_std": 0.4946486234664917,
      "rewards/IngredientFormatReward/mean": 0.9736718773841858,
      "rewards/IngredientFormatReward/std": 0.14484046772122383,
      "rewards/IngredientMatchReward/mean": 0.630078113079071,
      "rewards/IngredientMatchReward/std": 0.29432931542396545,
      "rewards/IngredientQuantityMatchReward/mean": 0.5996343314647674,
      "rewards/IngredientQuantityMatchReward/std": 0.4255487024784088,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.41085425615310667,
      "step": 615
    },
    {
      "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": 0.025,
      "completions/max_length": 509.0,
      "completions/mean_length": 389.6984375,
      "completions/min_length": 280.0,
      "epoch": 0.17246175243393602,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6722802519798279,
      "kl": 0.028958051186054944,
      "learning_rate": 9.595958517938399e-07,
      "loss": 0.0011583505198359489,
      "reward": 2.8262381076812746,
      "reward_std": 0.4465320944786072,
      "rewards/IngredientFormatReward/mean": 0.9739062547683716,
      "rewards/IngredientFormatReward/std": 0.12499363049864769,
      "rewards/IngredientMatchReward/mean": 0.565862488746643,
      "rewards/IngredientMatchReward/std": 0.29269099831581114,
      "rewards/IngredientQuantityMatchReward/mean": 0.5802194237709045,
      "rewards/IngredientQuantityMatchReward/std": 0.437412816286087,
      "rewards/TotalKcalExactMatchReward/mean": 0.70625,
      "rewards/TotalKcalExactMatchReward/std": 0.43868969678878783,
      "step": 620
    },
    {
      "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": 0.0234375,
      "completions/max_length": 512.2,
      "completions/mean_length": 391.6125,
      "completions/min_length": 269.6,
      "epoch": 0.17385257301808066,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7302356958389282,
      "kl": 0.028412992018274963,
      "learning_rate": 9.58685289026382e-07,
      "loss": 0.0011363377794623374,
      "reward": 2.9949509143829345,
      "reward_std": 0.456043142080307,
      "rewards/IngredientFormatReward/mean": 0.9746986746788024,
      "rewards/IngredientFormatReward/std": 0.1321501724421978,
      "rewards/IngredientMatchReward/mean": 0.6073921084403991,
      "rewards/IngredientMatchReward/std": 0.31130966544151306,
      "rewards/IngredientQuantityMatchReward/mean": 0.6112977027893066,
      "rewards/IngredientQuantityMatchReward/std": 0.4173098862171173,
      "rewards/TotalKcalExactMatchReward/mean": 0.8015625,
      "rewards/TotalKcalExactMatchReward/std": 0.39602966904640197,
      "step": 625
    },
    {
      "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": 0.0140625,
      "completions/max_length": 506.8,
      "completions/mean_length": 391.35,
      "completions/min_length": 248.6,
      "epoch": 0.17524339360222532,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.727810263633728,
      "kl": 0.0283606713404879,
      "learning_rate": 9.577650217763627e-07,
      "loss": 0.0011344054713845254,
      "reward": 2.885877180099487,
      "reward_std": 0.49911447763442995,
      "rewards/IngredientFormatReward/mean": 0.9827343702316285,
      "rewards/IngredientFormatReward/std": 0.11650933995842934,
      "rewards/IngredientMatchReward/mean": 0.5809616804122925,
      "rewards/IngredientMatchReward/std": 0.2889358103275299,
      "rewards/IngredientQuantityMatchReward/mean": 0.6143685460090638,
      "rewards/IngredientQuantityMatchReward/std": 0.4223813652992249,
      "rewards/TotalKcalExactMatchReward/mean": 0.7078125,
      "rewards/TotalKcalExactMatchReward/std": 0.4538150727748871,
      "step": 630
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.6,
      "completions/mean_length": 393.303125,
      "completions/min_length": 251.4,
      "epoch": 0.17663421418636996,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6584274172782898,
      "kl": 0.029251058853697033,
      "learning_rate": 9.56835069514032e-07,
      "loss": 0.0011701955460011958,
      "reward": 2.936937427520752,
      "reward_std": 0.40667486786842344,
      "rewards/IngredientFormatReward/mean": 0.9897916793823243,
      "rewards/IngredientFormatReward/std": 0.07407835721969605,
      "rewards/IngredientMatchReward/mean": 0.5751605868339539,
      "rewards/IngredientMatchReward/std": 0.2886354625225067,
      "rewards/IngredientQuantityMatchReward/mean": 0.6251101851463318,
      "rewards/IngredientQuantityMatchReward/std": 0.4174710988998413,
      "rewards/TotalKcalExactMatchReward/mean": 0.746875,
      "rewards/TotalKcalExactMatchReward/std": 0.4202993929386139,
      "step": 635
    },
    {
      "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": 0.0046875,
      "completions/max_length": 504.0,
      "completions/mean_length": 387.21875,
      "completions/min_length": 256.6,
      "epoch": 0.1780250347705146,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6709960103034973,
      "kl": 0.028330153250135482,
      "learning_rate": 9.558954519145486e-07,
      "loss": 0.0011332341469824313,
      "reward": 2.886409616470337,
      "reward_std": 0.4505835294723511,
      "rewards/IngredientFormatReward/mean": 0.9907552003860474,
      "rewards/IngredientFormatReward/std": 0.06122845821082592,
      "rewards/IngredientMatchReward/mean": 0.577808165550232,
      "rewards/IngredientMatchReward/std": 0.29142323732376096,
      "rewards/IngredientQuantityMatchReward/mean": 0.5990962386131287,
      "rewards/IngredientQuantityMatchReward/std": 0.4226303339004517,
      "rewards/TotalKcalExactMatchReward/mean": 0.71875,
      "rewards/TotalKcalExactMatchReward/std": 0.4469615936279297,
      "step": 640
    },
    {
      "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": 0.0140625,
      "completions/max_length": 507.2,
      "completions/mean_length": 384.828125,
      "completions/min_length": 226.8,
      "epoch": 0.17941585535465926,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.656096339225769,
      "kl": 0.028859713370911776,
      "learning_rate": 9.54946188857561e-07,
      "loss": 0.001154566276818514,
      "reward": 2.9817068099975588,
      "reward_std": 0.4848079979419708,
      "rewards/IngredientFormatReward/mean": 0.9825000047683716,
      "rewards/IngredientFormatReward/std": 0.10897158831357956,
      "rewards/IngredientMatchReward/mean": 0.6160367250442504,
      "rewards/IngredientMatchReward/std": 0.2831864595413208,
      "rewards/IngredientQuantityMatchReward/mean": 0.6144201159477234,
      "rewards/IngredientQuantityMatchReward/std": 0.42764689326286315,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.4078401982784271,
      "step": 645
    },
    {
      "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": 0.0140625,
      "completions/max_length": 506.8,
      "completions/mean_length": 389.315625,
      "completions/min_length": 266.2,
      "epoch": 0.1808066759388039,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.734682559967041,
      "kl": 0.02891454305499792,
      "learning_rate": 9.539873004267892e-07,
      "loss": 0.0011331122368574142,
      "reward": 2.9698495388031008,
      "reward_std": 0.45514548420906065,
      "rewards/IngredientFormatReward/mean": 0.9760416746139526,
      "rewards/IngredientFormatReward/std": 0.11592378914356231,
      "rewards/IngredientMatchReward/mean": 0.5972185134887695,
      "rewards/IngredientMatchReward/std": 0.29427924156188967,
      "rewards/IngredientQuantityMatchReward/mean": 0.6325268268585205,
      "rewards/IngredientQuantityMatchReward/std": 0.4187918961048126,
      "rewards/TotalKcalExactMatchReward/mean": 0.7640625,
      "rewards/TotalKcalExactMatchReward/std": 0.40054896771907805,
      "step": 650
    },
    {
      "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": 0.01875,
      "completions/max_length": 512.0,
      "completions/mean_length": 400.5484375,
      "completions/min_length": 270.0,
      "epoch": 0.18219749652294853,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7253914475440979,
      "kl": 0.030589212966151535,
      "learning_rate": 9.530188069095984e-07,
      "loss": 0.001223677024245262,
      "reward": 2.8579261302948,
      "reward_std": 0.5048339784145355,
      "rewards/IngredientFormatReward/mean": 0.9809375047683716,
      "rewards/IngredientFormatReward/std": 0.1079106204211712,
      "rewards/IngredientMatchReward/mean": 0.5796428680419922,
      "rewards/IngredientMatchReward/std": 0.2922792613506317,
      "rewards/IngredientQuantityMatchReward/mean": 0.5660957932472229,
      "rewards/IngredientQuantityMatchReward/std": 0.4285131454467773,
      "rewards/TotalKcalExactMatchReward/mean": 0.73125,
      "rewards/TotalKcalExactMatchReward/std": 0.4407511293888092,
      "step": 655
    },
    {
      "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": 0.0140625,
      "completions/max_length": 512.0,
      "completions/mean_length": 391.3125,
      "completions/min_length": 278.4,
      "epoch": 0.1835883171070932,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7367101311683655,
      "kl": 0.029405112494714558,
      "learning_rate": 9.520407287965706e-07,
      "loss": 0.0011763956397771835,
      "reward": 3.055669593811035,
      "reward_std": 0.4588906943798065,
      "rewards/IngredientFormatReward/mean": 0.984375,
      "rewards/IngredientFormatReward/std": 0.10880739241838455,
      "rewards/IngredientMatchReward/mean": 0.6255388259887695,
      "rewards/IngredientMatchReward/std": 0.28291895389556887,
      "rewards/IngredientQuantityMatchReward/mean": 0.6410683155059814,
      "rewards/IngredientQuantityMatchReward/std": 0.39713342785835265,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.3906814754009247,
      "step": 660
    },
    {
      "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": 0.009375,
      "completions/max_length": 510.4,
      "completions/mean_length": 391.725,
      "completions/min_length": 256.8,
      "epoch": 0.18497913769123783,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6968327164649963,
      "kl": 0.029885214404202998,
      "learning_rate": 9.510530867810705e-07,
      "loss": 0.0011955580674111843,
      "reward": 3.0517891883850097,
      "reward_std": 0.42123398184776306,
      "rewards/IngredientFormatReward/mean": 0.9899999976158143,
      "rewards/IngredientFormatReward/std": 0.08557360619306564,
      "rewards/IngredientMatchReward/mean": 0.6036179542541504,
      "rewards/IngredientMatchReward/std": 0.28850804567337035,
      "rewards/IngredientQuantityMatchReward/mean": 0.6831712603569031,
      "rewards/IngredientQuantityMatchReward/std": 0.4143078148365021,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.4119704604148865,
      "step": 665
    },
    {
      "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": 0.0109375,
      "completions/max_length": 508.2,
      "completions/mean_length": 396.1671875,
      "completions/min_length": 277.6,
      "epoch": 0.18636995827538247,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.704643189907074,
      "kl": 0.02852991851978004,
      "learning_rate": 9.50055901758808e-07,
      "loss": 0.00114138200879097,
      "reward": 2.9828824043273925,
      "reward_std": 0.40985679626464844,
      "rewards/IngredientFormatReward/mean": 0.9801543951034546,
      "rewards/IngredientFormatReward/std": 0.11804641857743263,
      "rewards/IngredientMatchReward/mean": 0.5995734095573425,
      "rewards/IngredientMatchReward/std": 0.2708533525466919,
      "rewards/IngredientQuantityMatchReward/mean": 0.6375296354293823,
      "rewards/IngredientQuantityMatchReward/std": 0.40709404945373534,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4051818013191223,
      "step": 670
    },
    {
      "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": 0.021875,
      "completions/max_length": 509.6,
      "completions/mean_length": 402.6734375,
      "completions/min_length": 274.2,
      "epoch": 0.18776077885952713,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.9530069828033447,
      "kl": 0.030310024251230062,
      "learning_rate": 9.49049194827396e-07,
      "loss": 0.0012126181274652482,
      "reward": 2.9908456325531008,
      "reward_std": 0.4299203336238861,
      "rewards/IngredientFormatReward/mean": 0.9764843821525574,
      "rewards/IngredientFormatReward/std": 0.12506350949406625,
      "rewards/IngredientMatchReward/mean": 0.619918167591095,
      "rewards/IngredientMatchReward/std": 0.29712865948677064,
      "rewards/IngredientQuantityMatchReward/mean": 0.6335056185722351,
      "rewards/IngredientQuantityMatchReward/std": 0.41634988188743594,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.42713966965675354,
      "step": 675
    },
    {
      "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": 0.0125,
      "completions/max_length": 510.0,
      "completions/mean_length": 397.6875,
      "completions/min_length": 253.6,
      "epoch": 0.18915159944367177,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6784078478813171,
      "kl": 0.029615208157338202,
      "learning_rate": 9.480329872859039e-07,
      "loss": 0.00118463933467865,
      "reward": 2.971098279953003,
      "reward_std": 0.45131182074546816,
      "rewards/IngredientFormatReward/mean": 0.9823176980018615,
      "rewards/IngredientFormatReward/std": 0.11630072444677353,
      "rewards/IngredientMatchReward/mean": 0.6179464221000671,
      "rewards/IngredientMatchReward/std": 0.30161031484603884,
      "rewards/IngredientQuantityMatchReward/mean": 0.5692716717720032,
      "rewards/IngredientQuantityMatchReward/std": 0.4135533392429352,
      "rewards/TotalKcalExactMatchReward/mean": 0.8015625,
      "rewards/TotalKcalExactMatchReward/std": 0.3857778489589691,
      "step": 680
    },
    {
      "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": 0.025,
      "completions/max_length": 513.0,
      "completions/mean_length": 399.109375,
      "completions/min_length": 272.2,
      "epoch": 0.1905424200278164,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7147411704063416,
      "kl": 0.0303641744190827,
      "learning_rate": 9.470073006344073e-07,
      "loss": 0.0012148549780249597,
      "reward": 2.9252397537231447,
      "reward_std": 0.4890766263008118,
      "rewards/IngredientFormatReward/mean": 0.9698437452316284,
      "rewards/IngredientFormatReward/std": 0.15541768670082093,
      "rewards/IngredientMatchReward/mean": 0.5790067076683044,
      "rewards/IngredientMatchReward/std": 0.310345721244812,
      "rewards/IngredientQuantityMatchReward/mean": 0.6013892769813538,
      "rewards/IngredientQuantityMatchReward/std": 0.4297428786754608,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.4112145364284515,
      "step": 685
    },
    {
      "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": 0.025,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.259375,
      "completions/min_length": 249.4,
      "epoch": 0.19193324061196107,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7676776647567749,
      "kl": 0.042987666395492855,
      "learning_rate": 9.459721565735328e-07,
      "loss": 0.0017181294038891791,
      "reward": 2.856154441833496,
      "reward_std": 0.518483692407608,
      "rewards/IngredientFormatReward/mean": 0.9700000047683716,
      "rewards/IngredientFormatReward/std": 0.15938313901424409,
      "rewards/IngredientMatchReward/mean": 0.5849032878875733,
      "rewards/IngredientMatchReward/std": 0.3013963460922241,
      "rewards/IngredientQuantityMatchReward/mean": 0.5918761491775513,
      "rewards/IngredientQuantityMatchReward/std": 0.43730496764183047,
      "rewards/TotalKcalExactMatchReward/mean": 0.709375,
      "rewards/TotalKcalExactMatchReward/std": 0.4479576826095581,
      "step": 690
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 393.9359375,
      "completions/min_length": 236.4,
      "epoch": 0.1933240611961057,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.714293897151947,
      "kl": 0.03143521044403315,
      "learning_rate": 9.449275770039993e-07,
      "loss": 0.001257462427020073,
      "reward": 2.9983524322509765,
      "reward_std": 0.49345011711120607,
      "rewards/IngredientFormatReward/mean": 0.9696093797683716,
      "rewards/IngredientFormatReward/std": 0.16928862631320954,
      "rewards/IngredientMatchReward/mean": 0.5991536498069763,
      "rewards/IngredientMatchReward/std": 0.32413737177848817,
      "rewards/IngredientQuantityMatchReward/mean": 0.6608394384384155,
      "rewards/IngredientQuantityMatchReward/std": 0.4169635772705078,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.4014790415763855,
      "step": 695
    },
    {
      "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": 0.0296875,
      "completions/max_length": 512.4,
      "completions/mean_length": 400.2984375,
      "completions/min_length": 245.2,
      "epoch": 0.19471488178025034,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7118370532989502,
      "kl": 0.02870661865454167,
      "learning_rate": 9.43873584026154e-07,
      "loss": 0.0011482758447527886,
      "reward": 2.994645929336548,
      "reward_std": 0.46502137184143066,
      "rewards/IngredientFormatReward/mean": 0.9630468726158142,
      "rewards/IngredientFormatReward/std": 0.15404580980539323,
      "rewards/IngredientMatchReward/mean": 0.6109213829040527,
      "rewards/IngredientMatchReward/std": 0.2893530786037445,
      "rewards/IngredientQuantityMatchReward/mean": 0.5847402095794678,
      "rewards/IngredientQuantityMatchReward/std": 0.425648158788681,
      "rewards/TotalKcalExactMatchReward/mean": 0.8359375,
      "rewards/TotalKcalExactMatchReward/std": 0.3676931142807007,
      "step": 700
    },
    {
      "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": 0.0109375,
      "completions/max_length": 510.0,
      "completions/mean_length": 396.5609375,
      "completions/min_length": 254.8,
      "epoch": 0.19610570236439498,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6505905985832214,
      "kl": 0.02851622838061303,
      "learning_rate": 9.428101999395055e-07,
      "loss": 0.0011407439596951008,
      "reward": 2.909848928451538,
      "reward_std": 0.44104580879211425,
      "rewards/IngredientFormatReward/mean": 0.9778682947158813,
      "rewards/IngredientFormatReward/std": 0.12648722752928734,
      "rewards/IngredientMatchReward/mean": 0.568949019908905,
      "rewards/IngredientMatchReward/std": 0.28893651366233825,
      "rewards/IngredientQuantityMatchReward/mean": 0.6161566257476807,
      "rewards/IngredientQuantityMatchReward/std": 0.4073750078678131,
      "rewards/TotalKcalExactMatchReward/mean": 0.746875,
      "rewards/TotalKcalExactMatchReward/std": 0.427871572971344,
      "step": 705
    },
    {
      "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": 0.0125,
      "completions/max_length": 507.4,
      "completions/mean_length": 391.8578125,
      "completions/min_length": 227.0,
      "epoch": 0.19749652294853964,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6460047364234924,
      "kl": 0.032377722975797954,
      "learning_rate": 9.417374472422513e-07,
      "loss": 0.0012954175472259521,
      "reward": 2.9442656517028807,
      "reward_std": 0.5029145121574402,
      "rewards/IngredientFormatReward/mean": 0.9696875095367432,
      "rewards/IngredientFormatReward/std": 0.16133667081594466,
      "rewards/IngredientMatchReward/mean": 0.6018706560134888,
      "rewards/IngredientMatchReward/std": 0.2942827999591827,
      "rewards/IngredientQuantityMatchReward/mean": 0.6336450338363647,
      "rewards/IngredientQuantityMatchReward/std": 0.4295430064201355,
      "rewards/TotalKcalExactMatchReward/mean": 0.7390625,
      "rewards/TotalKcalExactMatchReward/std": 0.4340041160583496,
      "step": 710
    },
    {
      "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": 0.00625,
      "completions/max_length": 503.8,
      "completions/mean_length": 390.2859375,
      "completions/min_length": 284.8,
      "epoch": 0.19888734353268428,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6602519154548645,
      "kl": 0.028656962723471225,
      "learning_rate": 9.406553486308027e-07,
      "loss": 0.0011462993919849397,
      "reward": 2.86570086479187,
      "reward_std": 0.4690316140651703,
      "rewards/IngredientFormatReward/mean": 0.990416657924652,
      "rewards/IngredientFormatReward/std": 0.06260524652898311,
      "rewards/IngredientMatchReward/mean": 0.5725049614906311,
      "rewards/IngredientMatchReward/std": 0.3110696792602539,
      "rewards/IngredientQuantityMatchReward/mean": 0.588716697692871,
      "rewards/IngredientQuantityMatchReward/std": 0.4408229410648346,
      "rewards/TotalKcalExactMatchReward/mean": 0.7140625,
      "rewards/TotalKcalExactMatchReward/std": 0.44996110796928407,
      "step": 715
    },
    {
      "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": 0.0171875,
      "completions/max_length": 505.2,
      "completions/mean_length": 398.0546875,
      "completions/min_length": 244.6,
      "epoch": 0.20027816411682892,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6826504468917847,
      "kl": 0.03144657907541841,
      "learning_rate": 9.395639269993034e-07,
      "loss": 0.001257948949933052,
      "reward": 3.1293591499328612,
      "reward_std": 0.491245037317276,
      "rewards/IngredientFormatReward/mean": 0.9729166626930237,
      "rewards/IngredientFormatReward/std": 0.13929663002490997,
      "rewards/IngredientMatchReward/mean": 0.644785475730896,
      "rewards/IngredientMatchReward/std": 0.2948901176452637,
      "rewards/IngredientQuantityMatchReward/mean": 0.6647819876670837,
      "rewards/IngredientQuantityMatchReward/std": 0.39904108047485354,
      "rewards/TotalKcalExactMatchReward/mean": 0.846875,
      "rewards/TotalKcalExactMatchReward/std": 0.35378908812999726,
      "step": 720
    },
    {
      "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": 0.0390625,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.9,
      "completions/min_length": 255.8,
      "epoch": 0.20166898470097358,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6098330616950989,
      "kl": 0.03146871651988477,
      "learning_rate": 9.384632054391466e-07,
      "loss": 0.0012590238824486733,
      "reward": 2.974373722076416,
      "reward_std": 0.5352123498916626,
      "rewards/IngredientFormatReward/mean": 0.96015625,
      "rewards/IngredientFormatReward/std": 0.1843802809715271,
      "rewards/IngredientMatchReward/mean": 0.603830623626709,
      "rewards/IngredientMatchReward/std": 0.29342788457870483,
      "rewards/IngredientQuantityMatchReward/mean": 0.6353869080543518,
      "rewards/IngredientQuantityMatchReward/std": 0.39622330069541933,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.41315234899520875,
      "step": 725
    },
    {
      "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": 0.0234375,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.6984375,
      "completions/min_length": 251.6,
      "epoch": 0.20305980528511822,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7605530023574829,
      "kl": 0.03163579052779823,
      "learning_rate": 9.373532072384851e-07,
      "loss": 0.0012655130587518216,
      "reward": 2.9362371921539308,
      "reward_std": 0.5152683258056641,
      "rewards/IngredientFormatReward/mean": 0.9728515625,
      "rewards/IngredientFormatReward/std": 0.14956897497177124,
      "rewards/IngredientMatchReward/mean": 0.6218024492263794,
      "rewards/IngredientMatchReward/std": 0.27714348435401914,
      "rewards/IngredientQuantityMatchReward/mean": 0.6118955969810486,
      "rewards/IngredientQuantityMatchReward/std": 0.41608654856681826,
      "rewards/TotalKcalExactMatchReward/mean": 0.7296875,
      "rewards/TotalKcalExactMatchReward/std": 0.4384908080101013,
      "step": 730
    },
    {
      "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": 0.021875,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.328125,
      "completions/min_length": 267.2,
      "epoch": 0.20445062586926285,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6673536896705627,
      "kl": 0.0321547920582816,
      "learning_rate": 9.362339558817393e-07,
      "loss": 0.0012860743328928948,
      "reward": 2.9911890983581544,
      "reward_std": 0.5093835830688477,
      "rewards/IngredientFormatReward/mean": 0.9742559552192688,
      "rewards/IngredientFormatReward/std": 0.15330008417367935,
      "rewards/IngredientMatchReward/mean": 0.5867342472076416,
      "rewards/IngredientMatchReward/std": 0.3136568009853363,
      "rewards/IngredientQuantityMatchReward/mean": 0.6192614316940308,
      "rewards/IngredientQuantityMatchReward/std": 0.4280283749103546,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.39159988760948183,
      "step": 735
    },
    {
      "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": 0.025,
      "completions/max_length": 508.4,
      "completions/mean_length": 394.5765625,
      "completions/min_length": 253.8,
      "epoch": 0.20584144645340752,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7442731261253357,
      "kl": 0.0327463579364121,
      "learning_rate": 9.351054750491004e-07,
      "loss": 0.0013101151213049888,
      "reward": 2.9522375583648683,
      "reward_std": 0.4935662269592285,
      "rewards/IngredientFormatReward/mean": 0.9732142925262451,
      "rewards/IngredientFormatReward/std": 0.13986825197935104,
      "rewards/IngredientMatchReward/mean": 0.6265885353088378,
      "rewards/IngredientMatchReward/std": 0.30202017426490785,
      "rewards/IngredientQuantityMatchReward/mean": 0.563372266292572,
      "rewards/IngredientQuantityMatchReward/std": 0.4121869564056396,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.39316596984863283,
      "step": 740
    },
    {
      "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": 0.0125,
      "completions/max_length": 501.4,
      "completions/mean_length": 391.06875,
      "completions/min_length": 247.2,
      "epoch": 0.20723226703755215,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6689611673355103,
      "kl": 0.03534543386194855,
      "learning_rate": 9.339677886160291e-07,
      "loss": 0.0014133938588202,
      "reward": 2.8360384464263917,
      "reward_std": 0.46426122784614565,
      "rewards/IngredientFormatReward/mean": 0.9857031226158142,
      "rewards/IngredientFormatReward/std": 0.09839616119861602,
      "rewards/IngredientMatchReward/mean": 0.6112010240554809,
      "rewards/IngredientMatchReward/std": 0.2886400640010834,
      "rewards/IngredientQuantityMatchReward/mean": 0.5438217937946319,
      "rewards/IngredientQuantityMatchReward/std": 0.44390941858291627,
      "rewards/TotalKcalExactMatchReward/mean": 0.6953125,
      "rewards/TotalKcalExactMatchReward/std": 0.45529637932777406,
      "step": 745
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 399.2765625,
      "completions/min_length": 263.8,
      "epoch": 0.2086230876216968,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6407611966133118,
      "kl": 0.030646953172981738,
      "learning_rate": 9.328209206527502e-07,
      "loss": 0.0012261301279067993,
      "reward": 2.9185956954956054,
      "reward_std": 0.4603721797466278,
      "rewards/IngredientFormatReward/mean": 0.9682291746139526,
      "rewards/IngredientFormatReward/std": 0.16202533841133118,
      "rewards/IngredientMatchReward/mean": 0.5754563689231873,
      "rewards/IngredientMatchReward/std": 0.29731823205947877,
      "rewards/IngredientQuantityMatchReward/mean": 0.5983476281166077,
      "rewards/IngredientQuantityMatchReward/std": 0.412556517124176,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4096756637096405,
      "step": 750
    },
    {
      "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": 0.0125,
      "completions/max_length": 511.6,
      "completions/mean_length": 396.646875,
      "completions/min_length": 257.2,
      "epoch": 0.21001390820584145,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6754993200302124,
      "kl": 0.031907440279610455,
      "learning_rate": 9.31664895423744e-07,
      "loss": 0.0012764555402100086,
      "reward": 2.9990989208221435,
      "reward_std": 0.39861383438110354,
      "rewards/IngredientFormatReward/mean": 0.9821874976158143,
      "rewards/IngredientFormatReward/std": 0.10982606858015061,
      "rewards/IngredientMatchReward/mean": 0.6273164629936219,
      "rewards/IngredientMatchReward/std": 0.28818137645721437,
      "rewards/IngredientQuantityMatchReward/mean": 0.5927198767662049,
      "rewards/IngredientQuantityMatchReward/std": 0.4226322829723358,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.399892657995224,
      "step": 755
    },
    {
      "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": 0.01875,
      "completions/max_length": 509.2,
      "completions/mean_length": 397.95625,
      "completions/min_length": 269.4,
      "epoch": 0.2114047287899861,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6390153169631958,
      "kl": 0.03162261028774083,
      "learning_rate": 9.30499737387233e-07,
      "loss": 0.001264964509755373,
      "reward": 3.1403099060058595,
      "reward_std": 0.43912080526351926,
      "rewards/IngredientFormatReward/mean": 0.9791666746139527,
      "rewards/IngredientFormatReward/std": 0.12517903745174408,
      "rewards/IngredientMatchReward/mean": 0.6559573531150817,
      "rewards/IngredientMatchReward/std": 0.2926880419254303,
      "rewards/IngredientQuantityMatchReward/mean": 0.695810866355896,
      "rewards/IngredientQuantityMatchReward/std": 0.3989140212535858,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.3878820836544037,
      "step": 760
    },
    {
      "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": 0.015625,
      "completions/max_length": 512.0,
      "completions/mean_length": 398.6171875,
      "completions/min_length": 273.2,
      "epoch": 0.21279554937413073,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6783661246299744,
      "kl": 0.030737385945394637,
      "learning_rate": 9.293254711946633e-07,
      "loss": 0.0012293593958020211,
      "reward": 3.055101490020752,
      "reward_std": 0.4194644451141357,
      "rewards/IngredientFormatReward/mean": 0.9799218773841858,
      "rewards/IngredientFormatReward/std": 0.1248725950717926,
      "rewards/IngredientMatchReward/mean": 0.631891131401062,
      "rewards/IngredientMatchReward/std": 0.2970883846282959,
      "rewards/IngredientQuantityMatchReward/mean": 0.6479760408401489,
      "rewards/IngredientQuantityMatchReward/std": 0.4189946413040161,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.3833641231060028,
      "step": 765
    },
    {
      "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": 0.01875,
      "completions/max_length": 506.2,
      "completions/mean_length": 385.9453125,
      "completions/min_length": 250.6,
      "epoch": 0.2141863699582754,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7975168824195862,
      "kl": 0.034082498284988105,
      "learning_rate": 9.281421216901842e-07,
      "loss": 0.0013631395064294337,
      "reward": 2.953678321838379,
      "reward_std": 0.4751779854297638,
      "rewards/IngredientFormatReward/mean": 0.9660937547683716,
      "rewards/IngredientFormatReward/std": 0.14769160524010658,
      "rewards/IngredientMatchReward/mean": 0.6321670532226562,
      "rewards/IngredientMatchReward/std": 0.29352579116821287,
      "rewards/IngredientQuantityMatchReward/mean": 0.5944800078868866,
      "rewards/IngredientQuantityMatchReward/std": 0.4254342496395111,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.3947984278202057,
      "step": 770
    },
    {
      "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": 0.01875,
      "completions/max_length": 513.0,
      "completions/mean_length": 393.13125,
      "completions/min_length": 252.8,
      "epoch": 0.21557719054242003,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7856742143630981,
      "kl": 0.03544444795697928,
      "learning_rate": 9.269497139101223e-07,
      "loss": 0.0014177929610013963,
      "reward": 3.043894386291504,
      "reward_std": 0.4352409660816193,
      "rewards/IngredientFormatReward/mean": 0.9726301908493042,
      "rewards/IngredientFormatReward/std": 0.14467429965734482,
      "rewards/IngredientMatchReward/mean": 0.5969146966934205,
      "rewards/IngredientMatchReward/std": 0.28962491154670716,
      "rewards/IngredientQuantityMatchReward/mean": 0.6805995225906372,
      "rewards/IngredientQuantityMatchReward/std": 0.4172350108623505,
      "rewards/TotalKcalExactMatchReward/mean": 0.79375,
      "rewards/TotalKcalExactMatchReward/std": 0.40479028820991514,
      "step": 775
    },
    {
      "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": 0.0234375,
      "completions/max_length": 506.8,
      "completions/mean_length": 389.1390625,
      "completions/min_length": 251.8,
      "epoch": 0.21696801112656466,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6804711222648621,
      "kl": 0.036672121356241406,
      "learning_rate": 9.257482730824515e-07,
      "loss": 0.0014671500772237777,
      "reward": 2.9757599353790285,
      "reward_std": 0.45886964797973634,
      "rewards/IngredientFormatReward/mean": 0.9778125047683716,
      "rewards/IngredientFormatReward/std": 0.11689424216747284,
      "rewards/IngredientMatchReward/mean": 0.6214930772781372,
      "rewards/IngredientMatchReward/std": 0.3079483389854431,
      "rewards/IngredientQuantityMatchReward/mean": 0.5920794010162354,
      "rewards/IngredientQuantityMatchReward/std": 0.39307602643966677,
      "rewards/TotalKcalExactMatchReward/mean": 0.784375,
      "rewards/TotalKcalExactMatchReward/std": 0.4060091316699982,
      "step": 780
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.0,
      "completions/mean_length": 387.490625,
      "completions/min_length": 225.2,
      "epoch": 0.21835883171070933,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6856589913368225,
      "kl": 0.0358359067235142,
      "learning_rate": 9.245378246262592e-07,
      "loss": 0.0014331799000501632,
      "reward": 3.0270840644836428,
      "reward_std": 0.4414548695087433,
      "rewards/IngredientFormatReward/mean": 0.9877976179122925,
      "rewards/IngredientFormatReward/std": 0.08675730489194393,
      "rewards/IngredientMatchReward/mean": 0.6201382637023926,
      "rewards/IngredientMatchReward/std": 0.29792742133140565,
      "rewards/IngredientQuantityMatchReward/mean": 0.583210676908493,
      "rewards/IngredientQuantityMatchReward/std": 0.4317150294780731,
      "rewards/TotalKcalExactMatchReward/mean": 0.8359375,
      "rewards/TotalKcalExactMatchReward/std": 0.3566481202840805,
      "step": 785
    },
    {
      "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": 0.021875,
      "completions/max_length": 510.4,
      "completions/mean_length": 393.415625,
      "completions/min_length": 265.8,
      "epoch": 0.21974965229485396,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7207933664321899,
      "kl": 0.034403257770463824,
      "learning_rate": 9.233183941512093e-07,
      "loss": 0.001376278605312109,
      "reward": 2.964992141723633,
      "reward_std": 0.46569212675094607,
      "rewards/IngredientFormatReward/mean": 0.979296875,
      "rewards/IngredientFormatReward/std": 0.09206215813755989,
      "rewards/IngredientMatchReward/mean": 0.619907021522522,
      "rewards/IngredientMatchReward/std": 0.29936749339103697,
      "rewards/IngredientQuantityMatchReward/mean": 0.5767257273197174,
      "rewards/IngredientQuantityMatchReward/std": 0.41478478312492373,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.39120323956012726,
      "step": 790
    },
    {
      "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": 0.0171875,
      "completions/max_length": 497.6,
      "completions/mean_length": 385.078125,
      "completions/min_length": 243.4,
      "epoch": 0.2211404728789986,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7506348490715027,
      "kl": 0.03393740386236459,
      "learning_rate": 9.220900074569993e-07,
      "loss": 0.0013577956706285477,
      "reward": 2.997026968002319,
      "reward_std": 0.46164332032203675,
      "rewards/IngredientFormatReward/mean": 0.9756770968437195,
      "rewards/IngredientFormatReward/std": 0.11836831197142601,
      "rewards/IngredientMatchReward/mean": 0.5768706798553467,
      "rewards/IngredientMatchReward/std": 0.29588728547096255,
      "rewards/IngredientQuantityMatchReward/mean": 0.6960417270660401,
      "rewards/IngredientQuantityMatchReward/std": 0.42147263288497927,
      "rewards/TotalKcalExactMatchReward/mean": 0.7484375,
      "rewards/TotalKcalExactMatchReward/std": 0.42918060421943666,
      "step": 795
    },
    {
      "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": 0.0296875,
      "completions/max_length": 508.6,
      "completions/mean_length": 390.1328125,
      "completions/min_length": 251.8,
      "epoch": 0.22253129346314326,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7214834094047546,
      "kl": 0.035237916419282554,
      "learning_rate": 9.20852690532815e-07,
      "loss": 0.0014095636084675788,
      "reward": 2.8804359912872313,
      "reward_std": 0.4783861994743347,
      "rewards/IngredientFormatReward/mean": 0.969921875,
      "rewards/IngredientFormatReward/std": 0.15964724868535995,
      "rewards/IngredientMatchReward/mean": 0.5503304958343506,
      "rewards/IngredientMatchReward/std": 0.29477221965789796,
      "rewards/IngredientQuantityMatchReward/mean": 0.6039335608482361,
      "rewards/IngredientQuantityMatchReward/std": 0.4312254786491394,
      "rewards/TotalKcalExactMatchReward/mean": 0.75625,
      "rewards/TotalKcalExactMatchReward/std": 0.4224544405937195,
      "step": 800
    },
    {
      "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": 0.025,
      "completions/max_length": 510.6,
      "completions/mean_length": 391.7359375,
      "completions/min_length": 265.0,
      "epoch": 0.2239221140472879,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.662796676158905,
      "kl": 0.03562974405940622,
      "learning_rate": 9.196064695567809e-07,
      "loss": 0.0014253812842071056,
      "reward": 2.928542375564575,
      "reward_std": 0.4484716713428497,
      "rewards/IngredientFormatReward/mean": 0.9722395896911621,
      "rewards/IngredientFormatReward/std": 0.1353061430156231,
      "rewards/IngredientMatchReward/mean": 0.5792367458343506,
      "rewards/IngredientMatchReward/std": 0.2964724600315094,
      "rewards/IngredientQuantityMatchReward/mean": 0.6426909327507019,
      "rewards/IngredientQuantityMatchReward/std": 0.42488672733306887,
      "rewards/TotalKcalExactMatchReward/mean": 0.734375,
      "rewards/TotalKcalExactMatchReward/std": 0.4239367187023163,
      "step": 805
    },
    {
      "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": 0.021875,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.98125,
      "completions/min_length": 254.8,
      "epoch": 0.22531293463143254,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6553943753242493,
      "kl": 0.03560564590152353,
      "learning_rate": 9.183513708954057e-07,
      "loss": 0.0014241338707506658,
      "reward": 2.88757643699646,
      "reward_std": 0.46099640130996705,
      "rewards/IngredientFormatReward/mean": 0.9710416674613953,
      "rewards/IngredientFormatReward/std": 0.16456654220819472,
      "rewards/IngredientMatchReward/mean": 0.576538336277008,
      "rewards/IngredientMatchReward/std": 0.29418975710868833,
      "rewards/IngredientQuantityMatchReward/mean": 0.6118714094161988,
      "rewards/IngredientQuantityMatchReward/std": 0.42434260845184324,
      "rewards/TotalKcalExactMatchReward/mean": 0.728125,
      "rewards/TotalKcalExactMatchReward/std": 0.4403379440307617,
      "step": 810
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 389.453125,
      "completions/min_length": 255.2,
      "epoch": 0.2267037552155772,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7021270394325256,
      "kl": 0.040182554116472605,
      "learning_rate": 9.17087421103025e-07,
      "loss": 0.001607796922326088,
      "reward": 2.849439525604248,
      "reward_std": 0.45260084271430967,
      "rewards/IngredientFormatReward/mean": 0.9757291674613953,
      "rewards/IngredientFormatReward/std": 0.14797138273715973,
      "rewards/IngredientMatchReward/mean": 0.5811318755149841,
      "rewards/IngredientMatchReward/std": 0.2990729808807373,
      "rewards/IngredientQuantityMatchReward/mean": 0.6097660839557648,
      "rewards/IngredientQuantityMatchReward/std": 0.41675270199775694,
      "rewards/TotalKcalExactMatchReward/mean": 0.6828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4664593517780304,
      "step": 815
    },
    {
      "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": 0.0140625,
      "completions/max_length": 507.2,
      "completions/mean_length": 391.60625,
      "completions/min_length": 274.0,
      "epoch": 0.22809457579972184,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.713983952999115,
      "kl": 0.03592538379598409,
      "learning_rate": 9.158146469212393e-07,
      "loss": 0.0014371214434504508,
      "reward": 2.941563129425049,
      "reward_std": 0.4839325726032257,
      "rewards/IngredientFormatReward/mean": 0.9802864551544189,
      "rewards/IngredientFormatReward/std": 0.12603867128491403,
      "rewards/IngredientMatchReward/mean": 0.605546236038208,
      "rewards/IngredientMatchReward/std": 0.28818662762641906,
      "rewards/IngredientQuantityMatchReward/mean": 0.5791679739952087,
      "rewards/IngredientQuantityMatchReward/std": 0.4372753441333771,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4141713917255402,
      "step": 820
    },
    {
      "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": 0.021875,
      "completions/max_length": 513.0,
      "completions/mean_length": 402.3546875,
      "completions/min_length": 259.2,
      "epoch": 0.22948539638386647,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7204667329788208,
      "kl": 0.06811709401663393,
      "learning_rate": 9.145330752783481e-07,
      "loss": 0.0027270957827568052,
      "reward": 2.9213661193847655,
      "reward_std": 0.4193553149700165,
      "rewards/IngredientFormatReward/mean": 0.9707142949104309,
      "rewards/IngredientFormatReward/std": 0.15083834379911423,
      "rewards/IngredientMatchReward/mean": 0.569453752040863,
      "rewards/IngredientMatchReward/std": 0.2871735870838165,
      "rewards/IngredientQuantityMatchReward/mean": 0.6218230545520782,
      "rewards/IngredientQuantityMatchReward/std": 0.40918234586715696,
      "rewards/TotalKcalExactMatchReward/mean": 0.759375,
      "rewards/TotalKcalExactMatchReward/std": 0.422866028547287,
      "step": 825
    },
    {
      "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": 0.015625,
      "completions/max_length": 512.6,
      "completions/mean_length": 396.4234375,
      "completions/min_length": 233.4,
      "epoch": 0.23087621696801114,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6867784857749939,
      "kl": 0.038253002404235306,
      "learning_rate": 9.132427332887802e-07,
      "loss": 0.0015299856662750245,
      "reward": 2.9125558853149416,
      "reward_std": 0.44560866951942446,
      "rewards/IngredientFormatReward/mean": 0.9753385543823242,
      "rewards/IngredientFormatReward/std": 0.13101519346237184,
      "rewards/IngredientMatchReward/mean": 0.6180487394332885,
      "rewards/IngredientMatchReward/std": 0.2829564154148102,
      "rewards/IngredientQuantityMatchReward/mean": 0.6363561630249024,
      "rewards/IngredientQuantityMatchReward/std": 0.41775635480880735,
      "rewards/TotalKcalExactMatchReward/mean": 0.6828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4539078712463379,
      "step": 830
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.6,
      "completions/mean_length": 393.5390625,
      "completions/min_length": 267.8,
      "epoch": 0.23226703755215578,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6686710715293884,
      "kl": 0.03419446239713579,
      "learning_rate": 9.119436482525204e-07,
      "loss": 0.0013676073402166366,
      "reward": 2.9858155727386473,
      "reward_std": 0.4048700749874115,
      "rewards/IngredientFormatReward/mean": 0.9871354103088379,
      "rewards/IngredientFormatReward/std": 0.08223003149032593,
      "rewards/IngredientMatchReward/mean": 0.6355809688568115,
      "rewards/IngredientMatchReward/std": 0.27816588878631593,
      "rewards/IngredientQuantityMatchReward/mean": 0.6396617531776428,
      "rewards/IngredientQuantityMatchReward/std": 0.41819279789924624,
      "rewards/TotalKcalExactMatchReward/mean": 0.7234375,
      "rewards/TotalKcalExactMatchReward/std": 0.4450302064418793,
      "step": 835
    },
    {
      "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": 0.0125,
      "completions/max_length": 509.0,
      "completions/mean_length": 399.4953125,
      "completions/min_length": 257.4,
      "epoch": 0.2336578581363004,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6076139211654663,
      "kl": 0.03447474674321711,
      "learning_rate": 9.106358476545312e-07,
      "loss": 0.0013787638396024703,
      "reward": 3.042753314971924,
      "reward_std": 0.4141663730144501,
      "rewards/IngredientFormatReward/mean": 0.9867931604385376,
      "rewards/IngredientFormatReward/std": 0.08491213321685791,
      "rewards/IngredientMatchReward/mean": 0.6473660826683044,
      "rewards/IngredientMatchReward/std": 0.3052875936031342,
      "rewards/IngredientQuantityMatchReward/mean": 0.6289066672325134,
      "rewards/IngredientQuantityMatchReward/std": 0.4121295094490051,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.411685985326767,
      "step": 840
    },
    {
      "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": 0.01875,
      "completions/max_length": 499.6,
      "completions/mean_length": 400.1671875,
      "completions/min_length": 276.0,
      "epoch": 0.23504867872044508,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7678961753845215,
      "kl": 0.035930186160840094,
      "learning_rate": 9.093193591641721e-07,
      "loss": 0.0014372747391462326,
      "reward": 2.9903510570526124,
      "reward_std": 0.4582275986671448,
      "rewards/IngredientFormatReward/mean": 0.9776562452316284,
      "rewards/IngredientFormatReward/std": 0.10337670594453811,
      "rewards/IngredientMatchReward/mean": 0.6180661082267761,
      "rewards/IngredientMatchReward/std": 0.28336626291275024,
      "rewards/IngredientQuantityMatchReward/mean": 0.6196286797523498,
      "rewards/IngredientQuantityMatchReward/std": 0.4030408561229706,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.4152782797813416,
      "step": 845
    },
    {
      "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": 0.028125,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.3109375,
      "completions/min_length": 261.4,
      "epoch": 0.2364394993045897,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7015578150749207,
      "kl": 0.034239031421020624,
      "learning_rate": 9.079942106346137e-07,
      "loss": 0.001369218062609434,
      "reward": 2.9080883026123048,
      "reward_std": 0.5065367937088012,
      "rewards/IngredientFormatReward/mean": 0.9693750023841858,
      "rewards/IngredientFormatReward/std": 0.1656607583165169,
      "rewards/IngredientMatchReward/mean": 0.6202536940574646,
      "rewards/IngredientMatchReward/std": 0.31288120746612547,
      "rewards/IngredientQuantityMatchReward/mean": 0.5918970346450806,
      "rewards/IngredientQuantityMatchReward/std": 0.4122667610645294,
      "rewards/TotalKcalExactMatchReward/mean": 0.7265625,
      "rewards/TotalKcalExactMatchReward/std": 0.4379298210144043,
      "step": 850
    },
    {
      "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": 0.0140625,
      "completions/max_length": 511.8,
      "completions/mean_length": 395.1578125,
      "completions/min_length": 222.0,
      "epoch": 0.23783031988873435,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7002923488616943,
      "kl": 0.038634613458998504,
      "learning_rate": 9.066604301022485e-07,
      "loss": 0.0015293240547180175,
      "reward": 3.0154534339904786,
      "reward_std": 0.4475682139396667,
      "rewards/IngredientFormatReward/mean": 0.9821093797683715,
      "rewards/IngredientFormatReward/std": 0.0994268886744976,
      "rewards/IngredientMatchReward/mean": 0.6168365597724914,
      "rewards/IngredientMatchReward/std": 0.28127440214157107,
      "rewards/IngredientQuantityMatchReward/mean": 0.6883825063705444,
      "rewards/IngredientQuantityMatchReward/std": 0.394818377494812,
      "rewards/TotalKcalExactMatchReward/mean": 0.728125,
      "rewards/TotalKcalExactMatchReward/std": 0.4414012312889099,
      "step": 855
    },
    {
      "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": 0.015625,
      "completions/max_length": 509.2,
      "completions/mean_length": 390.896875,
      "completions/min_length": 248.2,
      "epoch": 0.23922114047287898,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6400624513626099,
      "kl": 0.03780011597555131,
      "learning_rate": 9.053180457860977e-07,
      "loss": 0.0015122408978641034,
      "reward": 2.980843448638916,
      "reward_std": 0.4466714680194855,
      "rewards/IngredientFormatReward/mean": 0.9730208158493042,
      "rewards/IngredientFormatReward/std": 0.1319844603538513,
      "rewards/IngredientMatchReward/mean": 0.6098846673965455,
      "rewards/IngredientMatchReward/std": 0.2906585156917572,
      "rewards/IngredientQuantityMatchReward/mean": 0.6370004773139953,
      "rewards/IngredientQuantityMatchReward/std": 0.422407865524292,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.4248836100101471,
      "step": 860
    },
    {
      "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": 0.01875,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.340625,
      "completions/min_length": 265.8,
      "epoch": 0.24061196105702365,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6853905320167542,
      "kl": 0.03545564888045192,
      "learning_rate": 9.039670860872144e-07,
      "loss": 0.0014183764345943929,
      "reward": 2.9594481945037843,
      "reward_std": 0.45008849501609804,
      "rewards/IngredientFormatReward/mean": 0.9819270730018616,
      "rewards/IngredientFormatReward/std": 0.12875129282474518,
      "rewards/IngredientMatchReward/mean": 0.5874163150787354,
      "rewards/IngredientMatchReward/std": 0.2873119831085205,
      "rewards/IngredientQuantityMatchReward/mean": 0.5994797468185424,
      "rewards/IngredientQuantityMatchReward/std": 0.44308061003684995,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.38857262432575224,
      "step": 865
    },
    {
      "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": 0.009375,
      "completions/max_length": 494.8,
      "completions/mean_length": 397.6859375,
      "completions/min_length": 257.8,
      "epoch": 0.24200278164116829,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.714350700378418,
      "kl": 0.03471857472322881,
      "learning_rate": 9.026075795880821e-07,
      "loss": 0.001388669479638338,
      "reward": 3.029931926727295,
      "reward_std": 0.39870219230651854,
      "rewards/IngredientFormatReward/mean": 0.9884895801544189,
      "rewards/IngredientFormatReward/std": 0.06684594321995974,
      "rewards/IngredientMatchReward/mean": 0.619605016708374,
      "rewards/IngredientMatchReward/std": 0.30052986145019533,
      "rewards/IngredientQuantityMatchReward/mean": 0.6343373656272888,
      "rewards/IngredientQuantityMatchReward/std": 0.43777188658714294,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.40009031295776365,
      "step": 870
    },
    {
      "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": 0.0234375,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.43125,
      "completions/min_length": 236.8,
      "epoch": 0.24339360222531292,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6846738457679749,
      "kl": 0.03683779882267117,
      "learning_rate": 9.012395550520109e-07,
      "loss": 0.0014737180434167385,
      "reward": 3.0627113342285157,
      "reward_std": 0.4080630362033844,
      "rewards/IngredientFormatReward/mean": 0.9711979389190674,
      "rewards/IngredientFormatReward/std": 0.15553708821535112,
      "rewards/IngredientMatchReward/mean": 0.6409418463706971,
      "rewards/IngredientMatchReward/std": 0.2837235748767853,
      "rewards/IngredientQuantityMatchReward/mean": 0.6411966204643249,
      "rewards/IngredientQuantityMatchReward/std": 0.42754223942756653,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.36027163565158843,
      "step": 875
    },
    {
      "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": 0.0234375,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.959375,
      "completions/min_length": 261.6,
      "epoch": 0.24478442280945759,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6933212280273438,
      "kl": 0.038240592228248715,
      "learning_rate": 8.998630414225283e-07,
      "loss": 0.001529487781226635,
      "reward": 2.945117950439453,
      "reward_std": 0.4785862982273102,
      "rewards/IngredientFormatReward/mean": 0.9691666722297668,
      "rewards/IngredientFormatReward/std": 0.16549014002084733,
      "rewards/IngredientMatchReward/mean": 0.5762072086334229,
      "rewards/IngredientMatchReward/std": 0.28135599493980407,
      "rewards/IngredientQuantityMatchReward/mean": 0.6075566649436951,
      "rewards/IngredientQuantityMatchReward/std": 0.41557438373565675,
      "rewards/TotalKcalExactMatchReward/mean": 0.7921875,
      "rewards/TotalKcalExactMatchReward/std": 0.39553323984146116,
      "step": 880
    },
    {
      "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": 0.021875,
      "completions/max_length": 509.0,
      "completions/mean_length": 398.0453125,
      "completions/min_length": 242.4,
      "epoch": 0.24617524339360222,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.687666118144989,
      "kl": 0.03755876780487597,
      "learning_rate": 8.984780678227668e-07,
      "loss": 0.001502517983317375,
      "reward": 2.8259692668914793,
      "reward_std": 0.48776999711990354,
      "rewards/IngredientFormatReward/mean": 0.9772265672683715,
      "rewards/IngredientFormatReward/std": 0.12334810197353363,
      "rewards/IngredientMatchReward/mean": 0.5587667465209961,
      "rewards/IngredientMatchReward/std": 0.31351504027843474,
      "rewards/IngredientQuantityMatchReward/mean": 0.5915385007858276,
      "rewards/IngredientQuantityMatchReward/std": 0.4429964065551758,
      "rewards/TotalKcalExactMatchReward/mean": 0.6984375,
      "rewards/TotalKcalExactMatchReward/std": 0.45358685255050657,
      "step": 885
    },
    {
      "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": 0.0234375,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.665625,
      "completions/min_length": 274.0,
      "epoch": 0.24756606397774686,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6766310334205627,
      "kl": 0.037341193715110424,
      "learning_rate": 8.970846635548482e-07,
      "loss": 0.0014935942366719247,
      "reward": 2.9456982135772707,
      "reward_std": 0.4546304702758789,
      "rewards/IngredientFormatReward/mean": 0.9774999976158142,
      "rewards/IngredientFormatReward/std": 0.14295051991939545,
      "rewards/IngredientMatchReward/mean": 0.5852505087852478,
      "rewards/IngredientMatchReward/std": 0.27505454421043396,
      "rewards/IngredientQuantityMatchReward/mean": 0.593885064125061,
      "rewards/IngredientQuantityMatchReward/std": 0.4204001009464264,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.39809203147888184,
      "step": 890
    },
    {
      "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": 0.01875,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.15625,
      "completions/min_length": 263.8,
      "epoch": 0.24895688456189152,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.637706458568573,
      "kl": 0.034991304646246134,
      "learning_rate": 8.956828580992633e-07,
      "loss": 0.0013996277004480362,
      "reward": 2.9801607608795164,
      "reward_std": 0.44261011481285095,
      "rewards/IngredientFormatReward/mean": 0.9754166722297668,
      "rewards/IngredientFormatReward/std": 0.14382019639015198,
      "rewards/IngredientMatchReward/mean": 0.5979985237121582,
      "rewards/IngredientMatchReward/std": 0.29598821997642516,
      "rewards/IngredientQuantityMatchReward/mean": 0.6020581245422363,
      "rewards/IngredientQuantityMatchReward/std": 0.44247824549674986,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.39540945887565615,
      "step": 895
    },
    {
      "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": 0.021875,
      "completions/max_length": 510.4,
      "completions/mean_length": 399.5375,
      "completions/min_length": 264.8,
      "epoch": 0.25034770514603616,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7266840934753418,
      "kl": 0.03753926283679902,
      "learning_rate": 8.942726811142478e-07,
      "loss": 0.0015015792101621627,
      "reward": 2.8965949535369875,
      "reward_std": 0.422959041595459,
      "rewards/IngredientFormatReward/mean": 0.975,
      "rewards/IngredientFormatReward/std": 0.1332007184624672,
      "rewards/IngredientMatchReward/mean": 0.6016995429992675,
      "rewards/IngredientMatchReward/std": 0.2900406777858734,
      "rewards/IngredientQuantityMatchReward/mean": 0.5808328747749328,
      "rewards/IngredientQuantityMatchReward/std": 0.45071707367897035,
      "rewards/TotalKcalExactMatchReward/mean": 0.7390625,
      "rewards/TotalKcalExactMatchReward/std": 0.4362715780735016,
      "step": 900
    },
    {
      "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": 0.025,
      "completions/max_length": 508.2,
      "completions/mean_length": 394.2484375,
      "completions/min_length": 261.0,
      "epoch": 0.2517385257301808,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7164000272750854,
      "kl": 0.04179991343989968,
      "learning_rate": 8.928541624351562e-07,
      "loss": 0.001672225072979927,
      "reward": 3.075339603424072,
      "reward_std": 0.454181969165802,
      "rewards/IngredientFormatReward/mean": 0.9723437547683715,
      "rewards/IngredientFormatReward/std": 0.13345520794391633,
      "rewards/IngredientMatchReward/mean": 0.6349938035011291,
      "rewards/IngredientMatchReward/std": 0.3050593674182892,
      "rewards/IngredientQuantityMatchReward/mean": 0.6820645570755005,
      "rewards/IngredientQuantityMatchReward/std": 0.40236865282058715,
      "rewards/TotalKcalExactMatchReward/mean": 0.7859375,
      "rewards/TotalKcalExactMatchReward/std": 0.3957382678985596,
      "step": 905
    },
    {
      "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": 0.0109375,
      "completions/max_length": 501.0,
      "completions/mean_length": 384.8,
      "completions/min_length": 225.2,
      "epoch": 0.25312934631432543,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7135931849479675,
      "kl": 0.03884520079009235,
      "learning_rate": 8.914273320738288e-07,
      "loss": 0.0015718767419457436,
      "reward": 2.922934722900391,
      "reward_std": 0.4588244318962097,
      "rewards/IngredientFormatReward/mean": 0.9789843797683716,
      "rewards/IngredientFormatReward/std": 0.11647437512874603,
      "rewards/IngredientMatchReward/mean": 0.6133823275566102,
      "rewards/IngredientMatchReward/std": 0.30278020203113554,
      "rewards/IngredientQuantityMatchReward/mean": 0.6290055751800537,
      "rewards/IngredientQuantityMatchReward/std": 0.42403936982154844,
      "rewards/TotalKcalExactMatchReward/mean": 0.7015625,
      "rewards/TotalKcalExactMatchReward/std": 0.452223014831543,
      "step": 910
    },
    {
      "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": 0.025,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.36875,
      "completions/min_length": 258.8,
      "epoch": 0.2545201668984701,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6942304372787476,
      "kl": 0.03717799754813313,
      "learning_rate": 8.89992220217958e-07,
      "loss": 0.0014871499501168729,
      "reward": 3.0345030784606934,
      "reward_std": 0.5153680682182312,
      "rewards/IngredientFormatReward/mean": 0.9729166507720948,
      "rewards/IngredientFormatReward/std": 0.157105515897274,
      "rewards/IngredientMatchReward/mean": 0.6512928009033203,
      "rewards/IngredientMatchReward/std": 0.287622344493866,
      "rewards/IngredientQuantityMatchReward/mean": 0.6587311625480652,
      "rewards/IngredientQuantityMatchReward/std": 0.3968525528907776,
      "rewards/TotalKcalExactMatchReward/mean": 0.7515625,
      "rewards/TotalKcalExactMatchReward/std": 0.42349056601524354,
      "step": 915
    },
    {
      "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": 0.01875,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.059375,
      "completions/min_length": 245.6,
      "epoch": 0.25591098748261476,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7068426012992859,
      "kl": 0.036225578258745374,
      "learning_rate": 8.885488572304488e-07,
      "loss": 0.0014492036774754525,
      "reward": 3.032380723953247,
      "reward_std": 0.4576831817626953,
      "rewards/IngredientFormatReward/mean": 0.9752864480018616,
      "rewards/IngredientFormatReward/std": 0.1494772642850876,
      "rewards/IngredientMatchReward/mean": 0.5961935997009278,
      "rewards/IngredientMatchReward/std": 0.2839872181415558,
      "rewards/IngredientQuantityMatchReward/mean": 0.6327756881713867,
      "rewards/IngredientQuantityMatchReward/std": 0.402322381734848,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3686569333076477,
      "step": 920
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.165625,
      "completions/min_length": 255.8,
      "epoch": 0.2573018080667594,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6422480344772339,
      "kl": 0.03873362694866955,
      "learning_rate": 8.870972736487774e-07,
      "loss": 0.0015496846288442611,
      "reward": 3.0351743698120117,
      "reward_std": 0.45523384809494016,
      "rewards/IngredientFormatReward/mean": 0.9772916674613953,
      "rewards/IngredientFormatReward/std": 0.13808016031980513,
      "rewards/IngredientMatchReward/mean": 0.6586216449737549,
      "rewards/IngredientMatchReward/std": 0.2872202157974243,
      "rewards/IngredientQuantityMatchReward/mean": 0.602386087179184,
      "rewards/IngredientQuantityMatchReward/std": 0.4141529083251953,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.3945878803730011,
      "step": 925
    },
    {
      "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": 0.0125,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.778125,
      "completions/min_length": 267.2,
      "epoch": 0.25869262865090403,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7084170579910278,
      "kl": 0.03909153123386204,
      "learning_rate": 8.856375001843441e-07,
      "loss": 0.0015634719282388687,
      "reward": 2.907779502868652,
      "reward_std": 0.4718652844429016,
      "rewards/IngredientFormatReward/mean": 0.9831250071525574,
      "rewards/IngredientFormatReward/std": 0.1244356945157051,
      "rewards/IngredientMatchReward/mean": 0.5946900010108948,
      "rewards/IngredientMatchReward/std": 0.2881698727607727,
      "rewards/IngredientQuantityMatchReward/mean": 0.5846520006656647,
      "rewards/IngredientQuantityMatchReward/std": 0.42135303020477294,
      "rewards/TotalKcalExactMatchReward/mean": 0.7453125,
      "rewards/TotalKcalExactMatchReward/std": 0.4285692572593689,
      "step": 930
    },
    {
      "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": 0.0078125,
      "completions/max_length": 510.8,
      "completions/mean_length": 398.696875,
      "completions/min_length": 265.8,
      "epoch": 0.26008344923504867,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7023848295211792,
      "kl": 0.0367943222168833,
      "learning_rate": 8.84169567721824e-07,
      "loss": 0.0014716638252139091,
      "reward": 2.995437002182007,
      "reward_std": 0.4420520782470703,
      "rewards/IngredientFormatReward/mean": 0.9889955401420594,
      "rewards/IngredientFormatReward/std": 0.08887074701488018,
      "rewards/IngredientMatchReward/mean": 0.621924614906311,
      "rewards/IngredientMatchReward/std": 0.2982144713401794,
      "rewards/IngredientQuantityMatchReward/mean": 0.6204544305801392,
      "rewards/IngredientQuantityMatchReward/std": 0.41932734847068787,
      "rewards/TotalKcalExactMatchReward/mean": 0.7640625,
      "rewards/TotalKcalExactMatchReward/std": 0.42419267892837526,
      "step": 935
    },
    {
      "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": 0.01875,
      "completions/max_length": 511.6,
      "completions/mean_length": 392.1734375,
      "completions/min_length": 251.2,
      "epoch": 0.2614742698191933,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6219667196273804,
      "kl": 0.038923771027475594,
      "learning_rate": 8.826935073185133e-07,
      "loss": 0.0015569287352263927,
      "reward": 2.885612678527832,
      "reward_std": 0.47264568209648133,
      "rewards/IngredientFormatReward/mean": 0.9668750047683716,
      "rewards/IngredientFormatReward/std": 0.15271946042776108,
      "rewards/IngredientMatchReward/mean": 0.5864410042762757,
      "rewards/IngredientMatchReward/std": 0.31077885031700136,
      "rewards/IngredientQuantityMatchReward/mean": 0.563546645641327,
      "rewards/IngredientQuantityMatchReward/std": 0.4403660535812378,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.40360012352466584,
      "step": 940
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.6859375,
      "completions/min_length": 261.0,
      "epoch": 0.26286509040333794,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6733676195144653,
      "kl": 0.04225626492407173,
      "learning_rate": 8.812093502036731e-07,
      "loss": 0.0016901016235351563,
      "reward": 3.065718173980713,
      "reward_std": 0.5067483425140381,
      "rewards/IngredientFormatReward/mean": 0.9720312595367432,
      "rewards/IngredientFormatReward/std": 0.1448599338531494,
      "rewards/IngredientMatchReward/mean": 0.6443012237548829,
      "rewards/IngredientMatchReward/std": 0.30254427194595335,
      "rewards/IngredientQuantityMatchReward/mean": 0.6712606906890869,
      "rewards/IngredientQuantityMatchReward/std": 0.3889262616634369,
      "rewards/TotalKcalExactMatchReward/mean": 0.778125,
      "rewards/TotalKcalExactMatchReward/std": 0.4119017422199249,
      "step": 945
    },
    {
      "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": 0.040625,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.7359375,
      "completions/min_length": 269.4,
      "epoch": 0.26425591098748263,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6687275767326355,
      "kl": 0.04134470855351537,
      "learning_rate": 8.79717127777867e-07,
      "loss": 0.0016542974859476089,
      "reward": 2.9952784061431883,
      "reward_std": 0.4762373924255371,
      "rewards/IngredientFormatReward/mean": 0.9588392972946167,
      "rewards/IngredientFormatReward/std": 0.18952201455831527,
      "rewards/IngredientMatchReward/mean": 0.5773170948028564,
      "rewards/IngredientMatchReward/std": 0.30151690244674684,
      "rewards/IngredientQuantityMatchReward/mean": 0.6888094902038574,
      "rewards/IngredientQuantityMatchReward/std": 0.4004581332206726,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.40736610889434816,
      "step": 950
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 397.571875,
      "completions/min_length": 241.2,
      "epoch": 0.26564673157162727,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6472366452217102,
      "kl": 0.06367715783417224,
      "learning_rate": 8.782168716122987e-07,
      "loss": 0.0025473890826106073,
      "reward": 3.036988115310669,
      "reward_std": 0.45366032123565675,
      "rewards/IngredientFormatReward/mean": 0.9667447924613952,
      "rewards/IngredientFormatReward/std": 0.16739132702350618,
      "rewards/IngredientMatchReward/mean": 0.6112258076667786,
      "rewards/IngredientMatchReward/std": 0.29761743545532227,
      "rewards/IngredientQuantityMatchReward/mean": 0.6605799674987793,
      "rewards/IngredientQuantityMatchReward/std": 0.41313338875770567,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.3953717231750488,
      "step": 955
    },
    {
      "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": 0.0125,
      "completions/max_length": 509.8,
      "completions/mean_length": 398.71875,
      "completions/min_length": 282.6,
      "epoch": 0.2670375521557719,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.5986440777778625,
      "kl": 0.036805032892152666,
      "learning_rate": 8.767086134481425e-07,
      "loss": 0.0014722667634487151,
      "reward": 2.9474908351898192,
      "reward_std": 0.43001130819320676,
      "rewards/IngredientFormatReward/mean": 0.98203125,
      "rewards/IngredientFormatReward/std": 0.12097890824079513,
      "rewards/IngredientMatchReward/mean": 0.6146249771118164,
      "rewards/IngredientMatchReward/std": 0.28172118961811066,
      "rewards/IngredientQuantityMatchReward/mean": 0.6195845484733582,
      "rewards/IngredientQuantityMatchReward/std": 0.42852511405944826,
      "rewards/TotalKcalExactMatchReward/mean": 0.73125,
      "rewards/TotalKcalExactMatchReward/std": 0.4326680898666382,
      "step": 960
    },
    {
      "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": 0.0234375,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.1828125,
      "completions/min_length": 271.2,
      "epoch": 0.26842837273991654,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6705757975578308,
      "kl": 0.03636681609787047,
      "learning_rate": 8.751923851958727e-07,
      "loss": 0.0014547625556588172,
      "reward": 2.9844186305999756,
      "reward_std": 0.43761271238327026,
      "rewards/IngredientFormatReward/mean": 0.9759374976158142,
      "rewards/IngredientFormatReward/std": 0.14842937886714935,
      "rewards/IngredientMatchReward/mean": 0.6150954723358154,
      "rewards/IngredientMatchReward/std": 0.30773370862007143,
      "rewards/IngredientQuantityMatchReward/mean": 0.629323172569275,
      "rewards/IngredientQuantityMatchReward/std": 0.41692014336586,
      "rewards/TotalKcalExactMatchReward/mean": 0.7640625,
      "rewards/TotalKcalExactMatchReward/std": 0.42084046006202697,
      "step": 965
    },
    {
      "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": 0.009375,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.0953125,
      "completions/min_length": 245.0,
      "epoch": 0.2698191933240612,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6692220568656921,
      "kl": 0.03661972593981773,
      "learning_rate": 8.736682189345879e-07,
      "loss": 0.0014646416530013084,
      "reward": 3.068998193740845,
      "reward_std": 0.3991569995880127,
      "rewards/IngredientFormatReward/mean": 0.9866927027702331,
      "rewards/IngredientFormatReward/std": 0.10508427172899246,
      "rewards/IngredientMatchReward/mean": 0.6692838549613953,
      "rewards/IngredientMatchReward/std": 0.29518154859542844,
      "rewards/IngredientQuantityMatchReward/mean": 0.6364591598510743,
      "rewards/IngredientQuantityMatchReward/std": 0.41511212587356566,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.39989053606987,
      "step": 970
    },
    {
      "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": 0.0265625,
      "completions/max_length": 511.8,
      "completions/mean_length": 403.2828125,
      "completions/min_length": 268.6,
      "epoch": 0.2712100139082058,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6330889463424683,
      "kl": 0.034678391995839775,
      "learning_rate": 8.721361469113322e-07,
      "loss": 0.0013872714713215827,
      "reward": 3.068177604675293,
      "reward_std": 0.45619659423828124,
      "rewards/IngredientFormatReward/mean": 0.9731250047683716,
      "rewards/IngredientFormatReward/std": 0.13897247463464737,
      "rewards/IngredientMatchReward/mean": 0.6402989029884338,
      "rewards/IngredientMatchReward/std": 0.27887474894523623,
      "rewards/IngredientQuantityMatchReward/mean": 0.6235036015510559,
      "rewards/IngredientQuantityMatchReward/std": 0.40562615394592283,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.3756143867969513,
      "step": 975
    },
    {
      "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": 0.015625,
      "completions/max_length": 510.4,
      "completions/mean_length": 391.396875,
      "completions/min_length": 250.6,
      "epoch": 0.2726008344923505,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7315894365310669,
      "kl": 0.03695494781713933,
      "learning_rate": 8.705962015404142e-07,
      "loss": 0.0014784792438149452,
      "reward": 3.024481010437012,
      "reward_std": 0.43837874531745913,
      "rewards/IngredientFormatReward/mean": 0.9778645753860473,
      "rewards/IngredientFormatReward/std": 0.1255320355296135,
      "rewards/IngredientMatchReward/mean": 0.6787772893905639,
      "rewards/IngredientMatchReward/std": 0.2739121735095978,
      "rewards/IngredientQuantityMatchReward/mean": 0.6569016218185425,
      "rewards/IngredientQuantityMatchReward/std": 0.4188307225704193,
      "rewards/TotalKcalExactMatchReward/mean": 0.7109375,
      "rewards/TotalKcalExactMatchReward/std": 0.43596926927566526,
      "step": 980
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.8,
      "completions/mean_length": 394.14375,
      "completions/min_length": 246.2,
      "epoch": 0.27399165507649514,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.679327666759491,
      "kl": 0.03838320455979556,
      "learning_rate": 8.690484154027191e-07,
      "loss": 0.0015355561859905719,
      "reward": 3.014692544937134,
      "reward_std": 0.4545248806476593,
      "rewards/IngredientFormatReward/mean": 0.9801562428474426,
      "rewards/IngredientFormatReward/std": 0.11655383929610252,
      "rewards/IngredientMatchReward/mean": 0.5849051475524902,
      "rewards/IngredientMatchReward/std": 0.27766265869140627,
      "rewards/IngredientQuantityMatchReward/mean": 0.6105686783790588,
      "rewards/IngredientQuantityMatchReward/std": 0.439173287153244,
      "rewards/TotalKcalExactMatchReward/mean": 0.8390625,
      "rewards/TotalKcalExactMatchReward/std": 0.3558571696281433,
      "step": 985
    },
    {
      "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": 0.0140625,
      "completions/max_length": 503.2,
      "completions/mean_length": 394.9859375,
      "completions/min_length": 292.0,
      "epoch": 0.2753824756606398,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6654589176177979,
      "kl": 0.03780271227005869,
      "learning_rate": 8.674928212450214e-07,
      "loss": 0.0015188731253147126,
      "reward": 2.919002819061279,
      "reward_std": 0.4351273596286774,
      "rewards/IngredientFormatReward/mean": 0.9859375,
      "rewards/IngredientFormatReward/std": 0.08695523887872696,
      "rewards/IngredientMatchReward/mean": 0.6129117131233215,
      "rewards/IngredientMatchReward/std": 0.30980420112609863,
      "rewards/IngredientQuantityMatchReward/mean": 0.5967161536216736,
      "rewards/IngredientQuantityMatchReward/std": 0.4323438584804535,
      "rewards/TotalKcalExactMatchReward/mean": 0.7234375,
      "rewards/TotalKcalExactMatchReward/std": 0.44485429525375364,
      "step": 990
    },
    {
      "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": 0.0109375,
      "completions/max_length": 505.4,
      "completions/mean_length": 398.0296875,
      "completions/min_length": 250.4,
      "epoch": 0.2767732962447844,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7313476800918579,
      "kl": 0.039417910808697346,
      "learning_rate": 8.659294519792908e-07,
      "loss": 0.001576758362352848,
      "reward": 3.0119237899780273,
      "reward_std": 0.45336784720420836,
      "rewards/IngredientFormatReward/mean": 0.9878125071525574,
      "rewards/IngredientFormatReward/std": 0.07935706898570061,
      "rewards/IngredientMatchReward/mean": 0.6188765048980713,
      "rewards/IngredientMatchReward/std": 0.29016692042350767,
      "rewards/IngredientQuantityMatchReward/mean": 0.6177348375320435,
      "rewards/IngredientQuantityMatchReward/std": 0.43465182185173035,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.3942445933818817,
      "step": 995
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.0,
      "completions/mean_length": 394.6484375,
      "completions/min_length": 251.0,
      "epoch": 0.27816411682892905,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6560540199279785,
      "kl": 0.037446612119674684,
      "learning_rate": 8.643583406819964e-07,
      "loss": 0.0014979524537920951,
      "reward": 3.034720516204834,
      "reward_std": 0.44605728387832644,
      "rewards/IngredientFormatReward/mean": 0.9841778397560119,
      "rewards/IngredientFormatReward/std": 0.09864606074988842,
      "rewards/IngredientMatchReward/mean": 0.6110633611679077,
      "rewards/IngredientMatchReward/std": 0.2937765657901764,
      "rewards/IngredientQuantityMatchReward/mean": 0.6644793748855591,
      "rewards/IngredientQuantityMatchReward/std": 0.4109896540641785,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.41746620535850526,
      "step": 1000
    },
    {
      "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": 0.021875,
      "completions/max_length": 512.4,
      "completions/mean_length": 392.9859375,
      "completions/min_length": 262.4,
      "epoch": 0.2795549374130737,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7340788841247559,
      "kl": 0.039979275036603215,
      "learning_rate": 8.627795205934068e-07,
      "loss": 0.0015989819541573525,
      "reward": 2.9803890228271483,
      "reward_std": 0.4077727794647217,
      "rewards/IngredientFormatReward/mean": 0.9702473878860474,
      "rewards/IngredientFormatReward/std": 0.13495282642543316,
      "rewards/IngredientMatchReward/mean": 0.6199850082397461,
      "rewards/IngredientMatchReward/std": 0.28927130699157716,
      "rewards/IngredientQuantityMatchReward/mean": 0.6557816982269287,
      "rewards/IngredientQuantityMatchReward/std": 0.426509028673172,
      "rewards/TotalKcalExactMatchReward/mean": 0.734375,
      "rewards/TotalKcalExactMatchReward/std": 0.43847213983535765,
      "step": 1005
    },
    {
      "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": 0.0234375,
      "completions/max_length": 508.6,
      "completions/mean_length": 393.7125,
      "completions/min_length": 267.4,
      "epoch": 0.2809457579972184,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6542072296142578,
      "kl": 0.03828480986412615,
      "learning_rate": 8.611930251168866e-07,
      "loss": 0.0015314262360334395,
      "reward": 2.9594085693359373,
      "reward_std": 0.3793847501277924,
      "rewards/IngredientFormatReward/mean": 0.9762500047683715,
      "rewards/IngredientFormatReward/std": 0.1294731840491295,
      "rewards/IngredientMatchReward/mean": 0.5929135680198669,
      "rewards/IngredientMatchReward/std": 0.2810113072395325,
      "rewards/IngredientQuantityMatchReward/mean": 0.6214950919151306,
      "rewards/IngredientQuantityMatchReward/std": 0.42402665615081786,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.4016814470291138,
      "step": 1010
    },
    {
      "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": 0.0171875,
      "completions/max_length": 505.4,
      "completions/mean_length": 387.6640625,
      "completions/min_length": 250.4,
      "epoch": 0.282336578581363,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6601881980895996,
      "kl": 0.0400760383810848,
      "learning_rate": 8.595988878181902e-07,
      "loss": 0.0016030745580792427,
      "reward": 3.040871238708496,
      "reward_std": 0.4805952966213226,
      "rewards/IngredientFormatReward/mean": 0.9822395801544189,
      "rewards/IngredientFormatReward/std": 0.10235036537051201,
      "rewards/IngredientMatchReward/mean": 0.6395058393478393,
      "rewards/IngredientMatchReward/std": 0.2921601474285126,
      "rewards/IngredientQuantityMatchReward/mean": 0.6347507834434509,
      "rewards/IngredientQuantityMatchReward/std": 0.4206187427043915,
      "rewards/TotalKcalExactMatchReward/mean": 0.784375,
      "rewards/TotalKcalExactMatchReward/std": 0.41050392389297485,
      "step": 1015
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.4,
      "completions/mean_length": 398.084375,
      "completions/min_length": 273.4,
      "epoch": 0.28372739916550765,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6317110061645508,
      "kl": 0.036205627908930185,
      "learning_rate": 8.579971424247512e-07,
      "loss": 0.0014481697231531142,
      "reward": 2.9704572200775146,
      "reward_std": 0.43727081418037417,
      "rewards/IngredientFormatReward/mean": 0.9856250047683716,
      "rewards/IngredientFormatReward/std": 0.10253802984952927,
      "rewards/IngredientMatchReward/mean": 0.6059505343437195,
      "rewards/IngredientMatchReward/std": 0.2872993677854538,
      "rewards/IngredientQuantityMatchReward/mean": 0.5570067405700684,
      "rewards/IngredientQuantityMatchReward/std": 0.44095311164855955,
      "rewards/TotalKcalExactMatchReward/mean": 0.821875,
      "rewards/TotalKcalExactMatchReward/std": 0.3698591083288193,
      "step": 1020
    },
    {
      "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": 0.028125,
      "completions/max_length": 507.8,
      "completions/mean_length": 396.3828125,
      "completions/min_length": 228.4,
      "epoch": 0.2851182197496523,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6201499700546265,
      "kl": 0.03740023162681609,
      "learning_rate": 8.563878228249685e-07,
      "loss": 0.001496100425720215,
      "reward": 2.8762183666229246,
      "reward_std": 0.49682642221450807,
      "rewards/IngredientFormatReward/mean": 0.9622656106948853,
      "rewards/IngredientFormatReward/std": 0.14582059532403946,
      "rewards/IngredientMatchReward/mean": 0.5401600360870361,
      "rewards/IngredientMatchReward/std": 0.3013901233673096,
      "rewards/IngredientQuantityMatchReward/mean": 0.6019176304340362,
      "rewards/IngredientQuantityMatchReward/std": 0.43271024227142335,
      "rewards/TotalKcalExactMatchReward/mean": 0.771875,
      "rewards/TotalKcalExactMatchReward/std": 0.4094280481338501,
      "step": 1025
    },
    {
      "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": 0.0109375,
      "completions/max_length": 513.0,
      "completions/mean_length": 399.1984375,
      "completions/min_length": 283.4,
      "epoch": 0.2865090403337969,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6777713894844055,
      "kl": 0.037024815729819235,
      "learning_rate": 8.547709630674908e-07,
      "loss": 0.0014808710664510726,
      "reward": 2.9829100131988526,
      "reward_std": 0.47250267267227175,
      "rewards/IngredientFormatReward/mean": 0.9870833396911621,
      "rewards/IngredientFormatReward/std": 0.10598021894693374,
      "rewards/IngredientMatchReward/mean": 0.619783627986908,
      "rewards/IngredientMatchReward/std": 0.283379340171814,
      "rewards/IngredientQuantityMatchReward/mean": 0.5760430872440339,
      "rewards/IngredientQuantityMatchReward/std": 0.4405200660228729,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3969113349914551,
      "step": 1030
    },
    {
      "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": 0.0171875,
      "completions/max_length": 511.0,
      "completions/mean_length": 397.3640625,
      "completions/min_length": 270.2,
      "epoch": 0.28789986091794156,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6470330357551575,
      "kl": 0.039516926556825635,
      "learning_rate": 8.531465973604944e-07,
      "loss": 0.0015805866569280624,
      "reward": 2.9705513954162597,
      "reward_std": 0.47205381989479067,
      "rewards/IngredientFormatReward/mean": 0.9770312547683716,
      "rewards/IngredientFormatReward/std": 0.12652091942727567,
      "rewards/IngredientMatchReward/mean": 0.5878050565719605,
      "rewards/IngredientMatchReward/std": 0.2737370431423187,
      "rewards/IngredientQuantityMatchReward/mean": 0.6104025483131409,
      "rewards/IngredientQuantityMatchReward/std": 0.4259393811225891,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.3765485107898712,
      "step": 1035
    },
    {
      "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": 0.0140625,
      "completions/max_length": 509.6,
      "completions/mean_length": 396.15,
      "completions/min_length": 267.6,
      "epoch": 0.28929068150208626,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7152348160743713,
      "kl": 0.03850674827117473,
      "learning_rate": 8.515147600709603e-07,
      "loss": 0.0015403028577566148,
      "reward": 2.8930485248565674,
      "reward_std": 0.48774529099464414,
      "rewards/IngredientFormatReward/mean": 0.9825520753860474,
      "rewards/IngredientFormatReward/std": 0.11014169603586196,
      "rewards/IngredientMatchReward/mean": 0.5573319673538208,
      "rewards/IngredientMatchReward/std": 0.3069951593875885,
      "rewards/IngredientQuantityMatchReward/mean": 0.5453519821166992,
      "rewards/IngredientQuantityMatchReward/std": 0.4497275948524475,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.38788450360298155,
      "step": 1040
    },
    {
      "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": 0.03125,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.728125,
      "completions/min_length": 243.6,
      "epoch": 0.2906815020862309,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.714313805103302,
      "kl": 0.07018633377738297,
      "learning_rate": 8.498754857239471e-07,
      "loss": 0.0028061190620064735,
      "reward": 2.914971446990967,
      "reward_std": 0.48313209414482117,
      "rewards/IngredientFormatReward/mean": 0.9662500023841858,
      "rewards/IngredientFormatReward/std": 0.17367809414863586,
      "rewards/IngredientMatchReward/mean": 0.6251264691352845,
      "rewards/IngredientMatchReward/std": 0.3060845613479614,
      "rewards/IngredientQuantityMatchReward/mean": 0.6126574218273163,
      "rewards/IngredientQuantityMatchReward/std": 0.4163425207138062,
      "rewards/TotalKcalExactMatchReward/mean": 0.7109375,
      "rewards/TotalKcalExactMatchReward/std": 0.4510436117649078,
      "step": 1045
    },
    {
      "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": 0.00625,
      "completions/max_length": 509.8,
      "completions/mean_length": 387.565625,
      "completions/min_length": 240.4,
      "epoch": 0.29207232267037553,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6646053791046143,
      "kl": 0.040469748759642245,
      "learning_rate": 8.482288090018608e-07,
      "loss": 0.0016189534217119217,
      "reward": 3.0161834716796876,
      "reward_std": 0.4042413294315338,
      "rewards/IngredientFormatReward/mean": 0.98828125,
      "rewards/IngredientFormatReward/std": 0.07882513403892517,
      "rewards/IngredientMatchReward/mean": 0.6237146496772766,
      "rewards/IngredientMatchReward/std": 0.2907561004161835,
      "rewards/IngredientQuantityMatchReward/mean": 0.5995000958442688,
      "rewards/IngredientQuantityMatchReward/std": 0.41379945278167723,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.39606988430023193,
      "step": 1050
    },
    {
      "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": 0.015625,
      "completions/max_length": 509.6,
      "completions/mean_length": 398.89375,
      "completions/min_length": 272.2,
      "epoch": 0.29346314325452016,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7717159986495972,
      "kl": 0.03940560757182539,
      "learning_rate": 8.465747647437204e-07,
      "loss": 0.0015761561691761017,
      "reward": 3.0119256019592284,
      "reward_std": 0.4466732025146484,
      "rewards/IngredientFormatReward/mean": 0.982239592075348,
      "rewards/IngredientFormatReward/std": 0.0915367141366005,
      "rewards/IngredientMatchReward/mean": 0.6272364735603333,
      "rewards/IngredientMatchReward/std": 0.3005757510662079,
      "rewards/IngredientQuantityMatchReward/mean": 0.5852620124816894,
      "rewards/IngredientQuantityMatchReward/std": 0.4200045645236969,
      "rewards/TotalKcalExactMatchReward/mean": 0.8171875,
      "rewards/TotalKcalExactMatchReward/std": 0.37832899689674376,
      "step": 1055
    },
    {
      "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": 0.0171875,
      "completions/max_length": 509.8,
      "completions/mean_length": 396.9375,
      "completions/min_length": 273.2,
      "epoch": 0.2948539638386648,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.73525470495224,
      "kl": 0.04403948448598385,
      "learning_rate": 8.44913387944421e-07,
      "loss": 0.0017619559541344643,
      "reward": 2.986718940734863,
      "reward_std": 0.44211495518684385,
      "rewards/IngredientFormatReward/mean": 0.9837351202964782,
      "rewards/IngredientFormatReward/std": 0.10179438814520836,
      "rewards/IngredientMatchReward/mean": 0.6092330098152161,
      "rewards/IngredientMatchReward/std": 0.3025040805339813,
      "rewards/IngredientQuantityMatchReward/mean": 0.6500007748603821,
      "rewards/IngredientQuantityMatchReward/std": 0.4024610221385956,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.4365465521812439,
      "step": 1060
    },
    {
      "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": 0.0390625,
      "completions/max_length": 511.2,
      "completions/mean_length": 400.3953125,
      "completions/min_length": 285.6,
      "epoch": 0.29624478442280944,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7194033265113831,
      "kl": 0.04031976000405848,
      "learning_rate": 8.43244713753994e-07,
      "loss": 0.0016128033399581908,
      "reward": 2.9168705463409426,
      "reward_std": 0.48592537045478823,
      "rewards/IngredientFormatReward/mean": 0.9639881014823913,
      "rewards/IngredientFormatReward/std": 0.151922045648098,
      "rewards/IngredientMatchReward/mean": 0.6120442748069763,
      "rewards/IngredientMatchReward/std": 0.30550934076309205,
      "rewards/IngredientQuantityMatchReward/mean": 0.5580257177352905,
      "rewards/IngredientQuantityMatchReward/std": 0.4364677667617798,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3887697756290436,
      "step": 1065
    },
    {
      "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": 0.00625,
      "completions/max_length": 506.8,
      "completions/mean_length": 390.7953125,
      "completions/min_length": 247.0,
      "epoch": 0.29763560500695413,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7260541319847107,
      "kl": 0.03924874525982887,
      "learning_rate": 8.415687774768626e-07,
      "loss": 0.001570044457912445,
      "reward": 3.0964519023895263,
      "reward_std": 0.43312625885009765,
      "rewards/IngredientFormatReward/mean": 0.9788541674613953,
      "rewards/IngredientFormatReward/std": 0.12116530686616897,
      "rewards/IngredientMatchReward/mean": 0.6652752995491028,
      "rewards/IngredientMatchReward/std": 0.2774786114692688,
      "rewards/IngredientQuantityMatchReward/mean": 0.6491973519325256,
      "rewards/IngredientQuantityMatchReward/std": 0.42371960282325744,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.3943279206752777,
      "step": 1070
    },
    {
      "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": 0.0234375,
      "completions/max_length": 506.2,
      "completions/mean_length": 397.975,
      "completions/min_length": 260.6,
      "epoch": 0.29902642559109877,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6908602118492126,
      "kl": 0.038830549083650114,
      "learning_rate": 8.39885614571095e-07,
      "loss": 0.0015533534809947015,
      "reward": 2.9613656997680664,
      "reward_std": 0.4897449791431427,
      "rewards/IngredientFormatReward/mean": 0.966796875,
      "rewards/IngredientFormatReward/std": 0.14599068462848663,
      "rewards/IngredientMatchReward/mean": 0.6265476107597351,
      "rewards/IngredientMatchReward/std": 0.28194483518600466,
      "rewards/IngredientQuantityMatchReward/mean": 0.6023962616920471,
      "rewards/IngredientQuantityMatchReward/std": 0.43221556544303896,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.409938383102417,
      "step": 1075
    },
    {
      "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": 0.009375,
      "completions/max_length": 505.8,
      "completions/mean_length": 395.203125,
      "completions/min_length": 283.6,
      "epoch": 0.3004172461752434,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7245336771011353,
      "kl": 0.038045503990724684,
      "learning_rate": 8.38195260647655e-07,
      "loss": 0.0015217063948512078,
      "reward": 3.103727436065674,
      "reward_std": 0.4260666728019714,
      "rewards/IngredientFormatReward/mean": 0.9903125047683716,
      "rewards/IngredientFormatReward/std": 0.06337217316031456,
      "rewards/IngredientMatchReward/mean": 0.6669407367706299,
      "rewards/IngredientMatchReward/std": 0.2760904639959335,
      "rewards/IngredientQuantityMatchReward/mean": 0.6402241945266723,
      "rewards/IngredientQuantityMatchReward/std": 0.41369547247886657,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.38739696741104124,
      "step": 1080
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.4046875,
      "completions/min_length": 261.0,
      "epoch": 0.30180806675938804,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6876221299171448,
      "kl": 0.041109291440807286,
      "learning_rate": 8.364977514696474e-07,
      "loss": 0.0016444934532046317,
      "reward": 2.9648212432861327,
      "reward_std": 0.4610382914543152,
      "rewards/IngredientFormatReward/mean": 0.9824032783508301,
      "rewards/IngredientFormatReward/std": 0.11955612003803254,
      "rewards/IngredientMatchReward/mean": 0.5959220051765441,
      "rewards/IngredientMatchReward/std": 0.2947571128606796,
      "rewards/IngredientQuantityMatchReward/mean": 0.5974334716796875,
      "rewards/IngredientQuantityMatchReward/std": 0.4393026113510132,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.39993913769721984,
      "step": 1085
    },
    {
      "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": 0.0078125,
      "completions/max_length": 503.6,
      "completions/mean_length": 394.2390625,
      "completions/min_length": 263.4,
      "epoch": 0.3031988873435327,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6638349294662476,
      "kl": 0.04468865282833576,
      "learning_rate": 8.347931229515624e-07,
      "loss": 0.0017881166189908982,
      "reward": 2.975427198410034,
      "reward_std": 0.3665485680103302,
      "rewards/IngredientFormatReward/mean": 0.9866145730018616,
      "rewards/IngredientFormatReward/std": 0.10305822119116784,
      "rewards/IngredientMatchReward/mean": 0.6167863607406616,
      "rewards/IngredientMatchReward/std": 0.2862016916275024,
      "rewards/IngredientQuantityMatchReward/mean": 0.6376512527465821,
      "rewards/IngredientQuantityMatchReward/std": 0.4232665181159973,
      "rewards/TotalKcalExactMatchReward/mean": 0.734375,
      "rewards/TotalKcalExactMatchReward/std": 0.43666509389877317,
      "step": 1090
    },
    {
      "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": 0.009375,
      "completions/max_length": 505.4,
      "completions/mean_length": 389.9375,
      "completions/min_length": 274.2,
      "epoch": 0.3045897079276773,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6407614946365356,
      "kl": 0.044913537241518496,
      "learning_rate": 8.330814111585147e-07,
      "loss": 0.0017966246232390404,
      "reward": 2.969407796859741,
      "reward_std": 0.4416906535625458,
      "rewards/IngredientFormatReward/mean": 0.9874553680419922,
      "rewards/IngredientFormatReward/std": 0.09603629484772683,
      "rewards/IngredientMatchReward/mean": 0.6119326591491699,
      "rewards/IngredientMatchReward/std": 0.3116526186466217,
      "rewards/IngredientQuantityMatchReward/mean": 0.6106447577476501,
      "rewards/IngredientQuantityMatchReward/std": 0.42712777853012085,
      "rewards/TotalKcalExactMatchReward/mean": 0.759375,
      "rewards/TotalKcalExactMatchReward/std": 0.42226200103759765,
      "step": 1095
    },
    {
      "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": 0.0078125,
      "completions/max_length": 502.2,
      "completions/mean_length": 389.5390625,
      "completions/min_length": 252.4,
      "epoch": 0.30598052851182195,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.65201735496521,
      "kl": 0.039539157832041386,
      "learning_rate": 8.313626523054817e-07,
      "loss": 0.001581670343875885,
      "reward": 3.082692050933838,
      "reward_std": 0.4257974922657013,
      "rewards/IngredientFormatReward/mean": 0.9868750095367431,
      "rewards/IngredientFormatReward/std": 0.07427057325839996,
      "rewards/IngredientMatchReward/mean": 0.6660503268241882,
      "rewards/IngredientMatchReward/std": 0.2906301259994507,
      "rewards/IngredientQuantityMatchReward/mean": 0.6297666907310486,
      "rewards/IngredientQuantityMatchReward/std": 0.4138112425804138,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.4003869414329529,
      "step": 1100
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.6,
      "completions/mean_length": 387.5453125,
      "completions/min_length": 263.6,
      "epoch": 0.30737134909596664,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.5957731604576111,
      "kl": 0.04335511876270175,
      "learning_rate": 8.296368827565364e-07,
      "loss": 0.0017340987920761108,
      "reward": 2.981190013885498,
      "reward_std": 0.45507683157920836,
      "rewards/IngredientFormatReward/mean": 0.9846875190734863,
      "rewards/IngredientFormatReward/std": 0.10660690441727638,
      "rewards/IngredientMatchReward/mean": 0.5576079070568085,
      "rewards/IngredientMatchReward/std": 0.27225913405418395,
      "rewards/IngredientQuantityMatchReward/mean": 0.6248321235179901,
      "rewards/IngredientQuantityMatchReward/std": 0.41838674545288085,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3690988004207611,
      "step": 1105
    },
    {
      "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": 0.00625,
      "completions/max_length": 507.0,
      "completions/mean_length": 393.3875,
      "completions/min_length": 273.8,
      "epoch": 0.3087621696801113,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6201445460319519,
      "kl": 0.042535467795096336,
      "learning_rate": 8.279041390240781e-07,
      "loss": 0.0017013853415846826,
      "reward": 3.088879632949829,
      "reward_std": 0.35922372341156006,
      "rewards/IngredientFormatReward/mean": 0.9940624952316284,
      "rewards/IngredientFormatReward/std": 0.04400581270456314,
      "rewards/IngredientMatchReward/mean": 0.6234399914741516,
      "rewards/IngredientMatchReward/std": 0.27694119811058043,
      "rewards/IngredientQuantityMatchReward/mean": 0.6651272177696228,
      "rewards/IngredientQuantityMatchReward/std": 0.42454165816307066,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3904335916042328,
      "step": 1110
    },
    {
      "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": 0.0109375,
      "completions/max_length": 508.8,
      "completions/mean_length": 387.084375,
      "completions/min_length": 260.0,
      "epoch": 0.3101529902642559,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7132757306098938,
      "kl": 0.04249203451909125,
      "learning_rate": 8.261644577680602e-07,
      "loss": 0.0017001090571284294,
      "reward": 3.0332549095153807,
      "reward_std": 0.4516888916492462,
      "rewards/IngredientFormatReward/mean": 0.9883593797683716,
      "rewards/IngredientFormatReward/std": 0.08581716865301132,
      "rewards/IngredientMatchReward/mean": 0.619999372959137,
      "rewards/IngredientMatchReward/std": 0.2996371924877167,
      "rewards/IngredientQuantityMatchReward/mean": 0.6498961329460144,
      "rewards/IngredientQuantityMatchReward/std": 0.4136515438556671,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.4019003748893738,
      "step": 1115
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.2,
      "completions/mean_length": 393.415625,
      "completions/min_length": 265.8,
      "epoch": 0.31154381084840055,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6780772805213928,
      "kl": 0.04389502888079733,
      "learning_rate": 8.244178757952147e-07,
      "loss": 0.0017559628933668137,
      "reward": 3.0198182582855226,
      "reward_std": 0.41773988008499147,
      "rewards/IngredientFormatReward/mean": 0.9877343773841858,
      "rewards/IngredientFormatReward/std": 0.09690560251474381,
      "rewards/IngredientMatchReward/mean": 0.6150545716285706,
      "rewards/IngredientMatchReward/std": 0.27234477996826173,
      "rewards/IngredientQuantityMatchReward/mean": 0.663904333114624,
      "rewards/IngredientQuantityMatchReward/std": 0.39827964901924134,
      "rewards/TotalKcalExactMatchReward/mean": 0.753125,
      "rewards/TotalKcalExactMatchReward/std": 0.4100442737340927,
      "step": 1120
    },
    {
      "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": 0.00625,
      "completions/max_length": 511.2,
      "completions/mean_length": 386.2671875,
      "completions/min_length": 242.4,
      "epoch": 0.3129346314325452,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.689258337020874,
      "kl": 0.04299152567982674,
      "learning_rate": 8.226644300582729e-07,
      "loss": 0.0017199600115418434,
      "reward": 3.042409086227417,
      "reward_std": 0.4061857044696808,
      "rewards/IngredientFormatReward/mean": 0.9890625,
      "rewards/IngredientFormatReward/std": 0.09063329249620437,
      "rewards/IngredientMatchReward/mean": 0.6030859589576721,
      "rewards/IngredientMatchReward/std": 0.3205138087272644,
      "rewards/IngredientQuantityMatchReward/mean": 0.6033856332302093,
      "rewards/IngredientQuantityMatchReward/std": 0.4283260703086853,
      "rewards/TotalKcalExactMatchReward/mean": 0.846875,
      "rewards/TotalKcalExactMatchReward/std": 0.3567110180854797,
      "step": 1125
    },
    {
      "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": 0.009375,
      "completions/max_length": 510.2,
      "completions/mean_length": 389.278125,
      "completions/min_length": 264.0,
      "epoch": 0.3143254520166898,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7026454210281372,
      "kl": 0.043634300702251494,
      "learning_rate": 8.209041576551841e-07,
      "loss": 0.0017453931272029878,
      "reward": 3.039749765396118,
      "reward_std": 0.39314189553260803,
      "rewards/IngredientFormatReward/mean": 0.9890625,
      "rewards/IngredientFormatReward/std": 0.09063329249620437,
      "rewards/IngredientMatchReward/mean": 0.6170356154441834,
      "rewards/IngredientMatchReward/std": 0.28204314708709716,
      "rewards/IngredientQuantityMatchReward/mean": 0.6649016618728638,
      "rewards/IngredientQuantityMatchReward/std": 0.41771509647369387,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.412118011713028,
      "step": 1130
    },
    {
      "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": 0.0046875,
      "completions/max_length": 507.6,
      "completions/mean_length": 390.76875,
      "completions/min_length": 257.2,
      "epoch": 0.3157162726008345,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6993815302848816,
      "kl": 0.04059697471093386,
      "learning_rate": 8.191370958283304e-07,
      "loss": 0.0016238151118159294,
      "reward": 3.169062376022339,
      "reward_std": 0.3827301383018494,
      "rewards/IngredientFormatReward/mean": 0.99453125,
      "rewards/IngredientFormatReward/std": 0.059600303322076796,
      "rewards/IngredientMatchReward/mean": 0.6575991868972778,
      "rewards/IngredientMatchReward/std": 0.27744790315628054,
      "rewards/IngredientQuantityMatchReward/mean": 0.6934943556785583,
      "rewards/IngredientQuantityMatchReward/std": 0.39465083479881286,
      "rewards/TotalKcalExactMatchReward/mean": 0.8234375,
      "rewards/TotalKcalExactMatchReward/std": 0.37338051199913025,
      "step": 1135
    },
    {
      "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": 0.003125,
      "completions/max_length": 503.4,
      "completions/mean_length": 395.5109375,
      "completions/min_length": 277.8,
      "epoch": 0.31710709318497915,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6774301528930664,
      "kl": 0.03972219394054264,
      "learning_rate": 8.173632819637388e-07,
      "loss": 0.001589289866387844,
      "reward": 3.1047873497009277,
      "reward_std": 0.393167108297348,
      "rewards/IngredientFormatReward/mean": 0.9953125,
      "rewards/IngredientFormatReward/std": 0.04257904887199402,
      "rewards/IngredientMatchReward/mean": 0.6563039422035217,
      "rewards/IngredientMatchReward/std": 0.29307783842086793,
      "rewards/IngredientQuantityMatchReward/mean": 0.618795907497406,
      "rewards/IngredientQuantityMatchReward/std": 0.43244861960411074,
      "rewards/TotalKcalExactMatchReward/mean": 0.834375,
      "rewards/TotalKcalExactMatchReward/std": 0.3656123340129852,
      "step": 1140
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.8,
      "completions/mean_length": 387.9,
      "completions/min_length": 241.2,
      "epoch": 0.3184979137691238,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6518785357475281,
      "kl": 0.044394610333256423,
      "learning_rate": 8.155827535902911e-07,
      "loss": 0.0017760045826435088,
      "reward": 3.0225061416625976,
      "reward_std": 0.3792412966489792,
      "rewards/IngredientFormatReward/mean": 0.9885416746139526,
      "rewards/IngredientFormatReward/std": 0.09652584828436375,
      "rewards/IngredientMatchReward/mean": 0.6106491684913635,
      "rewards/IngredientMatchReward/std": 0.28019180297851565,
      "rewards/IngredientQuantityMatchReward/mean": 0.6498778104782105,
      "rewards/IngredientQuantityMatchReward/std": 0.4284306287765503,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.3934566378593445,
      "step": 1145
    },
    {
      "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": 0.0171875,
      "completions/max_length": 512.2,
      "completions/mean_length": 395.36875,
      "completions/min_length": 268.2,
      "epoch": 0.3198887343532684,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.75348961353302,
      "kl": 0.04309874954633415,
      "learning_rate": 8.137955483789278e-07,
      "loss": 0.0017238954082131387,
      "reward": 3.0311806201934814,
      "reward_std": 0.4421843349933624,
      "rewards/IngredientFormatReward/mean": 0.9781510472297669,
      "rewards/IngredientFormatReward/std": 0.11678568720817566,
      "rewards/IngredientMatchReward/mean": 0.6269333004951477,
      "rewards/IngredientMatchReward/std": 0.28258291482925413,
      "rewards/IngredientQuantityMatchReward/mean": 0.6464088559150696,
      "rewards/IngredientQuantityMatchReward/std": 0.40041940212249755,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.4114561975002289,
      "step": 1150
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.8,
      "completions/mean_length": 394.959375,
      "completions/min_length": 259.4,
      "epoch": 0.32127955493741306,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6728385090827942,
      "kl": 0.041832944145426154,
      "learning_rate": 8.120017041418537e-07,
      "loss": 0.0016738016158342362,
      "reward": 3.038546895980835,
      "reward_std": 0.36745981574058534,
      "rewards/IngredientFormatReward/mean": 0.990625,
      "rewards/IngredientFormatReward/std": 0.08340958207845688,
      "rewards/IngredientMatchReward/mean": 0.6225868225097656,
      "rewards/IngredientMatchReward/std": 0.2717043608427048,
      "rewards/IngredientQuantityMatchReward/mean": 0.617522656917572,
      "rewards/IngredientQuantityMatchReward/std": 0.4125868022441864,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.3869707822799683,
      "step": 1155
    },
    {
      "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": 0.025,
      "completions/max_length": 506.2,
      "completions/mean_length": 388.2078125,
      "completions/min_length": 245.4,
      "epoch": 0.3226703755215577,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6846386194229126,
      "kl": 0.053002631710842255,
      "learning_rate": 8.102012588317355e-07,
      "loss": 0.002119516022503376,
      "reward": 3.0419249057769777,
      "reward_std": 0.4614200294017792,
      "rewards/IngredientFormatReward/mean": 0.9736830472946167,
      "rewards/IngredientFormatReward/std": 0.13447005301713943,
      "rewards/IngredientMatchReward/mean": 0.6277988672256469,
      "rewards/IngredientMatchReward/std": 0.2982490539550781,
      "rewards/IngredientQuantityMatchReward/mean": 0.6357555866241456,
      "rewards/IngredientQuantityMatchReward/std": 0.4329476416110992,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.38714192509651185,
      "step": 1160
    },
    {
      "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": 0.0109375,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.6859375,
      "completions/min_length": 257.2,
      "epoch": 0.3240611961057024,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6272140145301819,
      "kl": 0.047240815870463845,
      "learning_rate": 8.083942505409008e-07,
      "loss": 0.0018897855654358864,
      "reward": 2.928142023086548,
      "reward_std": 0.4312022864818573,
      "rewards/IngredientFormatReward/mean": 0.985740327835083,
      "rewards/IngredientFormatReward/std": 0.10639770179986954,
      "rewards/IngredientMatchReward/mean": 0.5962034702301026,
      "rewards/IngredientMatchReward/std": 0.29347813725471494,
      "rewards/IngredientQuantityMatchReward/mean": 0.557135808467865,
      "rewards/IngredientQuantityMatchReward/std": 0.4391862154006958,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.4047076404094696,
      "step": 1165
    },
    {
      "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": 0.01875,
      "completions/max_length": 512.6,
      "completions/mean_length": 405.865625,
      "completions/min_length": 270.8,
      "epoch": 0.325452016689847,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7186459302902222,
      "kl": 0.04450485638808459,
      "learning_rate": 8.065807175005306e-07,
      "loss": 0.0017805084586143493,
      "reward": 3.0247393608093263,
      "reward_std": 0.43825086355209353,
      "rewards/IngredientFormatReward/mean": 0.9800000071525574,
      "rewards/IngredientFormatReward/std": 0.11633365899324417,
      "rewards/IngredientMatchReward/mean": 0.6262376070022583,
      "rewards/IngredientMatchReward/std": 0.2948880285024643,
      "rewards/IngredientQuantityMatchReward/mean": 0.6560018181800842,
      "rewards/IngredientQuantityMatchReward/std": 0.3940140724182129,
      "rewards/TotalKcalExactMatchReward/mean": 0.7625,
      "rewards/TotalKcalExactMatchReward/std": 0.4198298633098602,
      "step": 1170
    },
    {
      "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": 0.028125,
      "completions/max_length": 513.0,
      "completions/mean_length": 403.5625,
      "completions/min_length": 279.8,
      "epoch": 0.32684283727399166,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.652772843837738,
      "kl": 0.045967866806313394,
      "learning_rate": 8.047606980798515e-07,
      "loss": 0.001838766410946846,
      "reward": 2.9285586357116697,
      "reward_std": 0.4667127549648285,
      "rewards/IngredientFormatReward/mean": 0.9732142925262451,
      "rewards/IngredientFormatReward/std": 0.1576545402407646,
      "rewards/IngredientMatchReward/mean": 0.6107007622718811,
      "rewards/IngredientMatchReward/std": 0.2908420205116272,
      "rewards/IngredientQuantityMatchReward/mean": 0.6087061285972595,
      "rewards/IngredientQuantityMatchReward/std": 0.40912932753562925,
      "rewards/TotalKcalExactMatchReward/mean": 0.7359375,
      "rewards/TotalKcalExactMatchReward/std": 0.4145884871482849,
      "step": 1175
    },
    {
      "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": 0.0265625,
      "completions/max_length": 511.4,
      "completions/mean_length": 394.40625,
      "completions/min_length": 263.8,
      "epoch": 0.3282336578581363,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.625441312789917,
      "kl": 0.06037823753431439,
      "learning_rate": 8.029342307853238e-07,
      "loss": 0.0024156935513019564,
      "reward": 3.06034779548645,
      "reward_std": 0.47609464526176454,
      "rewards/IngredientFormatReward/mean": 0.9684374928474426,
      "rewards/IngredientFormatReward/std": 0.14431809410452842,
      "rewards/IngredientMatchReward/mean": 0.6360243082046508,
      "rewards/IngredientMatchReward/std": 0.2919568657875061,
      "rewards/IngredientQuantityMatchReward/mean": 0.6730734944343567,
      "rewards/IngredientQuantityMatchReward/std": 0.4052798330783844,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4118169665336609,
      "step": 1180
    },
    {
      "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": 0.0046875,
      "completions/max_length": 499.0,
      "completions/mean_length": 388.5265625,
      "completions/min_length": 250.0,
      "epoch": 0.32962447844228093,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6767213940620422,
      "kl": 0.04451586329378188,
      "learning_rate": 8.011013542598258e-07,
      "loss": 0.0017808981239795685,
      "reward": 3.0574079513549806,
      "reward_std": 0.4540375053882599,
      "rewards/IngredientFormatReward/mean": 0.9915625095367432,
      "rewards/IngredientFormatReward/std": 0.06801376938819885,
      "rewards/IngredientMatchReward/mean": 0.6417565941810608,
      "rewards/IngredientMatchReward/std": 0.28428298234939575,
      "rewards/IngredientQuantityMatchReward/mean": 0.6647138357162475,
      "rewards/IngredientQuantityMatchReward/std": 0.4178314387798309,
      "rewards/TotalKcalExactMatchReward/mean": 0.759375,
      "rewards/TotalKcalExactMatchReward/std": 0.4165681004524231,
      "step": 1185
    },
    {
      "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": 0.034375,
      "completions/max_length": 510.0,
      "completions/mean_length": 400.0953125,
      "completions/min_length": 257.8,
      "epoch": 0.33101529902642557,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7180148363113403,
      "kl": 0.04539623921737075,
      "learning_rate": 7.992621072818375e-07,
      "loss": 0.0018160387873649598,
      "reward": 3.049036645889282,
      "reward_std": 0.5000029981136322,
      "rewards/IngredientFormatReward/mean": 0.9583556532859803,
      "rewards/IngredientFormatReward/std": 0.1660301133990288,
      "rewards/IngredientMatchReward/mean": 0.6038318634033203,
      "rewards/IngredientMatchReward/std": 0.29093838334083555,
      "rewards/IngredientQuantityMatchReward/mean": 0.6680991888046265,
      "rewards/IngredientQuantityMatchReward/std": 0.4054907917976379,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.37279661893844607,
      "step": 1190
    },
    {
      "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": 0.015625,
      "completions/max_length": 510.0,
      "completions/mean_length": 392.565625,
      "completions/min_length": 250.8,
      "epoch": 0.33240611961057026,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7087981700897217,
      "kl": 0.04504902786575258,
      "learning_rate": 7.974165287646199e-07,
      "loss": 0.0018019940704107284,
      "reward": 3.0593740463256838,
      "reward_std": 0.4065113961696625,
      "rewards/IngredientFormatReward/mean": 0.9820684552192688,
      "rewards/IngredientFormatReward/std": 0.1045232132077217,
      "rewards/IngredientMatchReward/mean": 0.6230959236621857,
      "rewards/IngredientMatchReward/std": 0.30069578289985655,
      "rewards/IngredientQuantityMatchReward/mean": 0.6588971614837646,
      "rewards/IngredientQuantityMatchReward/std": 0.40588967204093934,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.38163130879402163,
      "step": 1195
    },
    {
      "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": 0.015625,
      "completions/max_length": 502.2,
      "completions/mean_length": 392.3703125,
      "completions/min_length": 242.8,
      "epoch": 0.3337969401947149,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.7028740048408508,
      "kl": 0.044297435646876694,
      "learning_rate": 7.955646577553909e-07,
      "loss": 0.001771971769630909,
      "reward": 2.929919147491455,
      "reward_std": 0.3947701930999756,
      "rewards/IngredientFormatReward/mean": 0.9732142925262451,
      "rewards/IngredientFormatReward/std": 0.1214448481798172,
      "rewards/IngredientMatchReward/mean": 0.6054377436637879,
      "rewards/IngredientMatchReward/std": 0.2721258968114853,
      "rewards/IngredientQuantityMatchReward/mean": 0.612204658985138,
      "rewards/IngredientQuantityMatchReward/std": 0.41552995443344115,
      "rewards/TotalKcalExactMatchReward/mean": 0.7390625,
      "rewards/TotalKcalExactMatchReward/std": 0.43984751105308534,
      "step": 1200
    },
    {
      "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": 0.0140625,
      "completions/max_length": 511.4,
      "completions/mean_length": 393.684375,
      "completions/min_length": 260.8,
      "epoch": 0.33518776077885953,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.717398464679718,
      "kl": 0.04384518137667328,
      "learning_rate": 7.937065334345002e-07,
      "loss": 0.001753794401884079,
      "reward": 3.049497127532959,
      "reward_std": 0.421299946308136,
      "rewards/IngredientFormatReward/mean": 0.985546886920929,
      "rewards/IngredientFormatReward/std": 0.10045162588357925,
      "rewards/IngredientMatchReward/mean": 0.6495393276214599,
      "rewards/IngredientMatchReward/std": 0.29075068831443784,
      "rewards/IngredientQuantityMatchReward/mean": 0.670660924911499,
      "rewards/IngredientQuantityMatchReward/std": 0.41321573257446287,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.43112467527389525,
      "step": 1205
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.015625,
      "completions/min_length": 255.6,
      "epoch": 0.33657858136300417,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7146852016448975,
      "kl": 0.11223286460153759,
      "learning_rate": 7.918421951145992e-07,
      "loss": 0.004489917308092117,
      "reward": 3.091847562789917,
      "reward_std": 0.3895574867725372,
      "rewards/IngredientFormatReward/mean": 0.9815104007720947,
      "rewards/IngredientFormatReward/std": 0.11061891429126262,
      "rewards/IngredientMatchReward/mean": 0.6332726240158081,
      "rewards/IngredientMatchReward/std": 0.284361732006073,
      "rewards/IngredientQuantityMatchReward/mean": 0.6911271452903748,
      "rewards/IngredientQuantityMatchReward/std": 0.3966558754444122,
      "rewards/TotalKcalExactMatchReward/mean": 0.7859375,
      "rewards/TotalKcalExactMatchReward/std": 0.3858437240123749,
      "step": 1210
    },
    {
      "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": 0.00625,
      "completions/max_length": 506.0,
      "completions/mean_length": 391.6609375,
      "completions/min_length": 269.2,
      "epoch": 0.3379694019471488,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6622430086135864,
      "kl": 0.04423058428801596,
      "learning_rate": 7.899716822398107e-07,
      "loss": 0.001769232377409935,
      "reward": 3.026241827011108,
      "reward_std": 0.39915978312492373,
      "rewards/IngredientFormatReward/mean": 0.9831510305404663,
      "rewards/IngredientFormatReward/std": 0.09449614509940148,
      "rewards/IngredientMatchReward/mean": 0.6212599039077759,
      "rewards/IngredientMatchReward/std": 0.2815777659416199,
      "rewards/IngredientQuantityMatchReward/mean": 0.6343308687210083,
      "rewards/IngredientQuantityMatchReward/std": 0.4104671120643616,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.4089196860790253,
      "step": 1215
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 399.0421875,
      "completions/min_length": 268.8,
      "epoch": 0.33936022253129344,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6995375752449036,
      "kl": 0.046378902695141735,
      "learning_rate": 7.880950343848933e-07,
      "loss": 0.001855262741446495,
      "reward": 2.894041156768799,
      "reward_std": 0.4899939954280853,
      "rewards/IngredientFormatReward/mean": 0.9703645706176758,
      "rewards/IngredientFormatReward/std": 0.15981422364711761,
      "rewards/IngredientMatchReward/mean": 0.6005760192871094,
      "rewards/IngredientMatchReward/std": 0.29874255061149596,
      "rewards/IngredientQuantityMatchReward/mean": 0.5606004893779755,
      "rewards/IngredientQuantityMatchReward/std": 0.4271669626235962,
      "rewards/TotalKcalExactMatchReward/mean": 0.7625,
      "rewards/TotalKcalExactMatchReward/std": 0.4213130235671997,
      "step": 1220
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.6,
      "completions/mean_length": 402.053125,
      "completions/min_length": 265.8,
      "epoch": 0.34075104311543813,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6287099719047546,
      "kl": 0.3435618258547038,
      "learning_rate": 7.862122912544041e-07,
      "loss": 0.013743305206298828,
      "reward": 3.0836498737335205,
      "reward_std": 0.4236464619636536,
      "rewards/IngredientFormatReward/mean": 0.9829278349876404,
      "rewards/IngredientFormatReward/std": 0.11173093281686305,
      "rewards/IngredientMatchReward/mean": 0.6336050748825073,
      "rewards/IngredientMatchReward/std": 0.2927217662334442,
      "rewards/IngredientQuantityMatchReward/mean": 0.6374294281005859,
      "rewards/IngredientQuantityMatchReward/std": 0.40337653160095216,
      "rewards/TotalKcalExactMatchReward/mean": 0.8296875,
      "rewards/TotalKcalExactMatchReward/std": 0.3659061789512634,
      "step": 1225
    },
    {
      "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": 0.0125,
      "completions/max_length": 504.2,
      "completions/mean_length": 395.1078125,
      "completions/min_length": 247.0,
      "epoch": 0.34214186369958277,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6511803269386292,
      "kl": 0.043997203628532586,
      "learning_rate": 7.843234926818594e-07,
      "loss": 0.001759953796863556,
      "reward": 2.9981056690216064,
      "reward_std": 0.4053235173225403,
      "rewards/IngredientFormatReward/mean": 0.981276023387909,
      "rewards/IngredientFormatReward/std": 0.1175081841647625,
      "rewards/IngredientMatchReward/mean": 0.5940014004707337,
      "rewards/IngredientMatchReward/std": 0.2828234672546387,
      "rewards/IngredientQuantityMatchReward/mean": 0.6415782332420349,
      "rewards/IngredientQuantityMatchReward/std": 0.4185176372528076,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.4035619616508484,
      "step": 1230
    },
    {
      "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": 0.0171875,
      "completions/max_length": 509.0,
      "completions/mean_length": 402.0828125,
      "completions/min_length": 274.8,
      "epoch": 0.3435326842837274,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.8296776413917542,
      "kl": 0.04468656564131379,
      "learning_rate": 7.824286786288919e-07,
      "loss": 0.0017875529825687408,
      "reward": 3.0703739643096926,
      "reward_std": 0.4063426792621613,
      "rewards/IngredientFormatReward/mean": 0.9804166793823242,
      "rewards/IngredientFormatReward/std": 0.11292581036686897,
      "rewards/IngredientMatchReward/mean": 0.6367730379104615,
      "rewards/IngredientMatchReward/std": 0.28677576780319214,
      "rewards/IngredientQuantityMatchReward/mean": 0.6484968185424804,
      "rewards/IngredientQuantityMatchReward/std": 0.4008986532688141,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.39468390345573423,
      "step": 1235
    },
    {
      "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": 0.009375,
      "completions/max_length": 508.8,
      "completions/mean_length": 389.0421875,
      "completions/min_length": 265.6,
      "epoch": 0.34492350486787204,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7355208396911621,
      "kl": 0.07059146789833903,
      "learning_rate": 7.805278891844033e-07,
      "loss": 0.0028071509674191474,
      "reward": 3.00608606338501,
      "reward_std": 0.4155315697193146,
      "rewards/IngredientFormatReward/mean": 0.9853124976158142,
      "rewards/IngredientFormatReward/std": 0.11259681433439254,
      "rewards/IngredientMatchReward/mean": 0.6296162247657776,
      "rewards/IngredientMatchReward/std": 0.3042337715625763,
      "rewards/IngredientQuantityMatchReward/mean": 0.6302198767662048,
      "rewards/IngredientQuantityMatchReward/std": 0.43931986689567565,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.42408340573310854,
      "step": 1240
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.45625,
      "completions/min_length": 270.4,
      "epoch": 0.3463143254520167,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7564507126808167,
      "kl": 0.04235607595182955,
      "learning_rate": 7.786211645637195e-07,
      "loss": 0.0016942832618951798,
      "reward": 3.1188386917114257,
      "reward_std": 0.4350468873977661,
      "rewards/IngredientFormatReward/mean": 0.9809375047683716,
      "rewards/IngredientFormatReward/std": 0.12780899554491043,
      "rewards/IngredientMatchReward/mean": 0.646865701675415,
      "rewards/IngredientMatchReward/std": 0.2780866503715515,
      "rewards/IngredientQuantityMatchReward/mean": 0.6629104495048523,
      "rewards/IngredientQuantityMatchReward/std": 0.4089032709598541,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.36764142513275144,
      "step": 1245
    },
    {
      "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": 0.025,
      "completions/max_length": 513.0,
      "completions/mean_length": 404.4109375,
      "completions/min_length": 265.6,
      "epoch": 0.3477051460361613,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6491013765335083,
      "kl": 0.04551801586057991,
      "learning_rate": 7.76708545107737e-07,
      "loss": 0.0018206179141998292,
      "reward": 3.0424689769744875,
      "reward_std": 0.44069910049438477,
      "rewards/IngredientFormatReward/mean": 0.9783556699752808,
      "rewards/IngredientFormatReward/std": 0.13486974984407424,
      "rewards/IngredientMatchReward/mean": 0.6330344676971436,
      "rewards/IngredientMatchReward/std": 0.29515123963356016,
      "rewards/IngredientQuantityMatchReward/mean": 0.5982663631439209,
      "rewards/IngredientQuantityMatchReward/std": 0.4426428616046906,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.37421483397483823,
      "step": 1250
    },
    {
      "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": 0.0296875,
      "completions/max_length": 507.4,
      "completions/mean_length": 401.31875,
      "completions/min_length": 281.8,
      "epoch": 0.34909596662030595,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6339758038520813,
      "kl": 0.045095857931301,
      "learning_rate": 7.747900712820703e-07,
      "loss": 0.001803934946656227,
      "reward": 3.072742462158203,
      "reward_std": 0.4893173336982727,
      "rewards/IngredientFormatReward/mean": 0.965625,
      "rewards/IngredientFormatReward/std": 0.15386179387569426,
      "rewards/IngredientMatchReward/mean": 0.6180078268051148,
      "rewards/IngredientMatchReward/std": 0.2904675453901291,
      "rewards/IngredientQuantityMatchReward/mean": 0.6781721353530884,
      "rewards/IngredientQuantityMatchReward/std": 0.41462584733963015,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.3824631690979004,
      "step": 1255
    },
    {
      "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": 0.0296875,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.753125,
      "completions/min_length": 249.4,
      "epoch": 0.35048678720445064,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6409927606582642,
      "kl": 0.04350223378278315,
      "learning_rate": 7.728657836761959e-07,
      "loss": 0.0017400886863470078,
      "reward": 3.152744436264038,
      "reward_std": 0.45829938650131224,
      "rewards/IngredientFormatReward/mean": 0.9662500023841858,
      "rewards/IngredientFormatReward/std": 0.17663300186395645,
      "rewards/IngredientMatchReward/mean": 0.701597261428833,
      "rewards/IngredientMatchReward/std": 0.28799739480018616,
      "rewards/IngredientQuantityMatchReward/mean": 0.6583346724510193,
      "rewards/IngredientQuantityMatchReward/std": 0.40607632994651793,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.3737199783325195,
      "step": 1260
    },
    {
      "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": 0.04375,
      "completions/max_length": 513.0,
      "completions/mean_length": 403.8640625,
      "completions/min_length": 273.0,
      "epoch": 0.3518776077885953,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6250739097595215,
      "kl": 0.045182203454896805,
      "learning_rate": 7.709357230025937e-07,
      "loss": 0.0018076343461871148,
      "reward": 2.949186897277832,
      "reward_std": 0.45738943815231325,
      "rewards/IngredientFormatReward/mean": 0.9545665860176087,
      "rewards/IngredientFormatReward/std": 0.19560373723506927,
      "rewards/IngredientMatchReward/mean": 0.6393668532371521,
      "rewards/IngredientMatchReward/std": 0.3034960389137268,
      "rewards/IngredientQuantityMatchReward/mean": 0.5896284222602844,
      "rewards/IngredientQuantityMatchReward/std": 0.4419006407260895,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4109238743782043,
      "step": 1265
    },
    {
      "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": 0.009375,
      "completions/max_length": 510.2,
      "completions/mean_length": 393.5328125,
      "completions/min_length": 264.8,
      "epoch": 0.3532684283727399,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6803897023200989,
      "kl": 0.04348803351167589,
      "learning_rate": 7.689999300958851e-07,
      "loss": 0.001739520952105522,
      "reward": 3.1484294891357423,
      "reward_std": 0.3893423616886139,
      "rewards/IngredientFormatReward/mean": 0.9821093797683715,
      "rewards/IngredientFormatReward/std": 0.09933126196265221,
      "rewards/IngredientMatchReward/mean": 0.6269580841064453,
      "rewards/IngredientMatchReward/std": 0.3178780019283295,
      "rewards/IngredientQuantityMatchReward/mean": 0.6643620371818543,
      "rewards/IngredientQuantityMatchReward/std": 0.40709378123283385,
      "rewards/TotalKcalExactMatchReward/mean": 0.875,
      "rewards/TotalKcalExactMatchReward/std": 0.320685601234436,
      "step": 1270
    },
    {
      "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": 0.0140625,
      "completions/max_length": 512.6,
      "completions/mean_length": 396.5296875,
      "completions/min_length": 276.2,
      "epoch": 0.35465924895688455,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6363331079483032,
      "kl": 0.04308997318148613,
      "learning_rate": 7.670584459119694e-07,
      "loss": 0.0017237666994333266,
      "reward": 2.969732904434204,
      "reward_std": 0.4224898397922516,
      "rewards/IngredientFormatReward/mean": 0.9850520849227905,
      "rewards/IngredientFormatReward/std": 0.10458039119839668,
      "rewards/IngredientMatchReward/mean": 0.568692970275879,
      "rewards/IngredientMatchReward/std": 0.29052608013153075,
      "rewards/IngredientQuantityMatchReward/mean": 0.6659879565238953,
      "rewards/IngredientQuantityMatchReward/std": 0.43609716296195983,
      "rewards/TotalKcalExactMatchReward/mean": 0.75,
      "rewards/TotalKcalExactMatchReward/std": 0.4251108765602112,
      "step": 1275
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.4,
      "completions/mean_length": 394.26875,
      "completions/min_length": 244.4,
      "epoch": 0.3560500695410292,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.7387193441390991,
      "kl": 0.04524663793854415,
      "learning_rate": 7.651113115271572e-07,
      "loss": 0.0018100041896104812,
      "reward": 2.9772053241729735,
      "reward_std": 0.3859552562236786,
      "rewards/IngredientFormatReward/mean": 0.9921875,
      "rewards/IngredientFormatReward/std": 0.05527795404195786,
      "rewards/IngredientMatchReward/mean": 0.6179451823234559,
      "rewards/IngredientMatchReward/std": 0.2892803192138672,
      "rewards/IngredientQuantityMatchReward/mean": 0.5951976597309112,
      "rewards/IngredientQuantityMatchReward/std": 0.4320973575115204,
      "rewards/TotalKcalExactMatchReward/mean": 0.771875,
      "rewards/TotalKcalExactMatchReward/std": 0.41942211985588074,
      "step": 1280
    },
    {
      "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": 0.0109375,
      "completions/max_length": 512.4,
      "completions/mean_length": 396.628125,
      "completions/min_length": 286.8,
      "epoch": 0.3574408901251738,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6847825050354004,
      "kl": 0.04578770115040243,
      "learning_rate": 7.631585681373013e-07,
      "loss": 0.00183718241751194,
      "reward": 3.059334373474121,
      "reward_std": 0.3995037257671356,
      "rewards/IngredientFormatReward/mean": 0.9859635353088378,
      "rewards/IngredientFormatReward/std": 0.10678645819425583,
      "rewards/IngredientMatchReward/mean": 0.655921995639801,
      "rewards/IngredientMatchReward/std": 0.26984080076217654,
      "rewards/IngredientQuantityMatchReward/mean": 0.5893237948417663,
      "rewards/IngredientQuantityMatchReward/std": 0.4238016366958618,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.37518492341041565,
      "step": 1285
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 404.5703125,
      "completions/min_length": 256.6,
      "epoch": 0.3588317107093185,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6697726845741272,
      "kl": 0.05169771097134799,
      "learning_rate": 7.612002570569254e-07,
      "loss": 0.0020679634064435957,
      "reward": 3.087264966964722,
      "reward_std": 0.41781227588653563,
      "rewards/IngredientFormatReward/mean": 0.9698958277702332,
      "rewards/IngredientFormatReward/std": 0.15693533271551133,
      "rewards/IngredientMatchReward/mean": 0.6202176332473754,
      "rewards/IngredientMatchReward/std": 0.29056022465229037,
      "rewards/IngredientQuantityMatchReward/mean": 0.6815265893936158,
      "rewards/IngredientQuantityMatchReward/std": 0.40200709700584414,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.38562028408050536,
      "step": 1290
    },
    {
      "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": 0.009375,
      "completions/max_length": 497.2,
      "completions/mean_length": 388.734375,
      "completions/min_length": 261.8,
      "epoch": 0.36022253129346316,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6716930866241455,
      "kl": 0.04933963837102055,
      "learning_rate": 7.592364197183494e-07,
      "loss": 0.001973751001060009,
      "reward": 3.0686947822570803,
      "reward_std": 0.42163644433021547,
      "rewards/IngredientFormatReward/mean": 0.9865829586982727,
      "rewards/IngredientFormatReward/std": 0.07807523384690285,
      "rewards/IngredientMatchReward/mean": 0.6540155410766602,
      "rewards/IngredientMatchReward/std": 0.28045170605182645,
      "rewards/IngredientQuantityMatchReward/mean": 0.6734087467193604,
      "rewards/IngredientQuantityMatchReward/std": 0.4206121265888214,
      "rewards/TotalKcalExactMatchReward/mean": 0.7546875,
      "rewards/TotalKcalExactMatchReward/std": 0.42576863765716555,
      "step": 1295
    },
    {
      "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": 0.00625,
      "completions/max_length": 507.0,
      "completions/mean_length": 388.784375,
      "completions/min_length": 261.2,
      "epoch": 0.3616133518776078,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6620562076568604,
      "kl": 0.045209857448935506,
      "learning_rate": 7.572670976708136e-07,
      "loss": 0.0018085699528455734,
      "reward": 3.1241397857666016,
      "reward_std": 0.36960124373435976,
      "rewards/IngredientFormatReward/mean": 0.9946874976158142,
      "rewards/IngredientFormatReward/std": 0.04299455732107162,
      "rewards/IngredientMatchReward/mean": 0.6504470586776734,
      "rewards/IngredientMatchReward/std": 0.2744769275188446,
      "rewards/IngredientQuantityMatchReward/mean": 0.6727553248405457,
      "rewards/IngredientQuantityMatchReward/std": 0.38103885650634767,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3866784691810608,
      "step": 1300
    },
    {
      "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": 0.00625,
      "completions/max_length": 504.6,
      "completions/mean_length": 393.6125,
      "completions/min_length": 282.0,
      "epoch": 0.36300417246175243,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.706493079662323,
      "kl": 0.04794170141685754,
      "learning_rate": 7.55292332579599e-07,
      "loss": 0.0019177064299583436,
      "reward": 3.060842227935791,
      "reward_std": 0.37944849729537966,
      "rewards/IngredientFormatReward/mean": 0.9922395706176758,
      "rewards/IngredientFormatReward/std": 0.05438530929386616,
      "rewards/IngredientMatchReward/mean": 0.6182112097740173,
      "rewards/IngredientMatchReward/std": 0.27092549204826355,
      "rewards/IngredientQuantityMatchReward/mean": 0.6347665190696716,
      "rewards/IngredientQuantityMatchReward/std": 0.41198580265045165,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.37279285192489625,
      "step": 1305
    },
    {
      "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": 0.0109375,
      "completions/max_length": 507.8,
      "completions/mean_length": 392.8203125,
      "completions/min_length": 267.8,
      "epoch": 0.36439499304589706,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6756166815757751,
      "kl": 0.0481220202986151,
      "learning_rate": 7.533121662251458e-07,
      "loss": 0.0019247991964221,
      "reward": 3.113231086730957,
      "reward_std": 0.36561869978904726,
      "rewards/IngredientFormatReward/mean": 0.9901692628860473,
      "rewards/IngredientFormatReward/std": 0.05964524168521166,
      "rewards/IngredientMatchReward/mean": 0.6073189616203308,
      "rewards/IngredientMatchReward/std": 0.30605280995368955,
      "rewards/IngredientQuantityMatchReward/mean": 0.7329302668571472,
      "rewards/IngredientQuantityMatchReward/std": 0.3724480032920837,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4047005772590637,
      "step": 1310
    },
    {
      "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": 0.0109375,
      "completions/max_length": 503.2,
      "completions/mean_length": 390.128125,
      "completions/min_length": 258.2,
      "epoch": 0.3657858136300417,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6606735587120056,
      "kl": 0.05145847264211625,
      "learning_rate": 7.513266405021703e-07,
      "loss": 0.0020576369017362593,
      "reward": 2.954634428024292,
      "reward_std": 0.4394294261932373,
      "rewards/IngredientFormatReward/mean": 0.9875,
      "rewards/IngredientFormatReward/std": 0.09063890352845191,
      "rewards/IngredientMatchReward/mean": 0.6141071438789367,
      "rewards/IngredientMatchReward/std": 0.2969283401966095,
      "rewards/IngredientQuantityMatchReward/mean": 0.6108398795127868,
      "rewards/IngredientQuantityMatchReward/std": 0.4317851722240448,
      "rewards/TotalKcalExactMatchReward/mean": 0.7421875,
      "rewards/TotalKcalExactMatchReward/std": 0.41644211411476134,
      "step": 1315
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.0,
      "completions/mean_length": 393.3578125,
      "completions/min_length": 256.4,
      "epoch": 0.3671766342141864,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6441048383712769,
      "kl": 0.04711353066377342,
      "learning_rate": 7.49335797418777e-07,
      "loss": 0.0018845852464437485,
      "reward": 3.0771498680114746,
      "reward_std": 0.4205670475959778,
      "rewards/IngredientFormatReward/mean": 0.9825632333755493,
      "rewards/IngredientFormatReward/std": 0.09538376778364181,
      "rewards/IngredientMatchReward/mean": 0.6370135188102722,
      "rewards/IngredientMatchReward/std": 0.2930262327194214,
      "rewards/IngredientQuantityMatchReward/mean": 0.68101065158844,
      "rewards/IngredientQuantityMatchReward/std": 0.39768595099449155,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4095042526721954,
      "step": 1320
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 399.4453125,
      "completions/min_length": 271.6,
      "epoch": 0.36856745479833103,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6368516683578491,
      "kl": 0.05054594988469034,
      "learning_rate": 7.473396790955714e-07,
      "loss": 0.0020216893404722213,
      "reward": 3.068678379058838,
      "reward_std": 0.4033094227313995,
      "rewards/IngredientFormatReward/mean": 0.9831770777702331,
      "rewards/IngredientFormatReward/std": 0.12154540866613388,
      "rewards/IngredientMatchReward/mean": 0.6302232027053833,
      "rewards/IngredientMatchReward/std": 0.2985549569129944,
      "rewards/IngredientQuantityMatchReward/mean": 0.6443406462669372,
      "rewards/IngredientQuantityMatchReward/std": 0.4117864727973938,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.3913091003894806,
      "step": 1325
    },
    {
      "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": 0.0234375,
      "completions/max_length": 509.6,
      "completions/mean_length": 392.634375,
      "completions/min_length": 273.8,
      "epoch": 0.36995827538247567,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7256594300270081,
      "kl": 0.04610037456732243,
      "learning_rate": 7.453383277647678e-07,
      "loss": 0.001843712106347084,
      "reward": 3.0510849952697754,
      "reward_std": 0.42244759798049925,
      "rewards/IngredientFormatReward/mean": 0.9765885472297668,
      "rewards/IngredientFormatReward/std": 0.1349714256823063,
      "rewards/IngredientMatchReward/mean": 0.6786749958992004,
      "rewards/IngredientMatchReward/std": 0.29368536472320556,
      "rewards/IngredientQuantityMatchReward/mean": 0.6880090117454529,
      "rewards/IngredientQuantityMatchReward/std": 0.3833075225353241,
      "rewards/TotalKcalExactMatchReward/mean": 0.7078125,
      "rewards/TotalKcalExactMatchReward/std": 0.45246235728263856,
      "step": 1330
    },
    {
      "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": 0.0109375,
      "completions/max_length": 513.0,
      "completions/mean_length": 390.9390625,
      "completions/min_length": 239.6,
      "epoch": 0.3713490959666203,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7128857970237732,
      "kl": 0.047564028692431745,
      "learning_rate": 7.433317857692962e-07,
      "loss": 0.0019028615206480026,
      "reward": 3.018064594268799,
      "reward_std": 0.41078593134880065,
      "rewards/IngredientFormatReward/mean": 0.9841666698455811,
      "rewards/IngredientFormatReward/std": 0.1172505185008049,
      "rewards/IngredientMatchReward/mean": 0.595754599571228,
      "rewards/IngredientMatchReward/std": 0.28117249011993406,
      "rewards/IngredientQuantityMatchReward/mean": 0.6256433725357056,
      "rewards/IngredientQuantityMatchReward/std": 0.41258437037467954,
      "rewards/TotalKcalExactMatchReward/mean": 0.8125,
      "rewards/TotalKcalExactMatchReward/std": 0.3811466634273529,
      "step": 1335
    },
    {
      "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": 0.0046875,
      "completions/max_length": 504.0,
      "completions/mean_length": 393.603125,
      "completions/min_length": 281.0,
      "epoch": 0.37273991655076494,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.633821427822113,
      "kl": 0.04494093197863549,
      "learning_rate": 7.413200955619065e-07,
      "loss": 0.0017976880073547364,
      "reward": 3.161479616165161,
      "reward_std": 0.3582748889923096,
      "rewards/IngredientFormatReward/mean": 0.9943489551544189,
      "rewards/IngredientFormatReward/std": 0.05378681570291519,
      "rewards/IngredientMatchReward/mean": 0.6633042454719543,
      "rewards/IngredientMatchReward/std": 0.2839880406856537,
      "rewards/IngredientQuantityMatchReward/mean": 0.6960138440132141,
      "rewards/IngredientQuantityMatchReward/std": 0.401453024148941,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.3868353426456451,
      "step": 1340
    },
    {
      "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": 0.025,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.909375,
      "completions/min_length": 269.6,
      "epoch": 0.3741307371349096,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6353825330734253,
      "kl": 0.047297230316326024,
      "learning_rate": 7.393032997042697e-07,
      "loss": 0.0018922967836260795,
      "reward": 2.9042635917663575,
      "reward_std": 0.45287638902664185,
      "rewards/IngredientFormatReward/mean": 0.9635937571525574,
      "rewards/IngredientFormatReward/std": 0.1764824703335762,
      "rewards/IngredientMatchReward/mean": 0.6264143109321594,
      "rewards/IngredientMatchReward/std": 0.2897026062011719,
      "rewards/IngredientQuantityMatchReward/mean": 0.5861305117607116,
      "rewards/IngredientQuantityMatchReward/std": 0.4275715291500092,
      "rewards/TotalKcalExactMatchReward/mean": 0.728125,
      "rewards/TotalKcalExactMatchReward/std": 0.44556268453598025,
      "step": 1345
    },
    {
      "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": 0.0125,
      "completions/max_length": 503.4,
      "completions/mean_length": 391.7609375,
      "completions/min_length": 268.2,
      "epoch": 0.37552155771905427,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6813985109329224,
      "kl": 0.04787277162540704,
      "learning_rate": 7.372814408660788e-07,
      "loss": 0.0019150124862790108,
      "reward": 3.0159966945648193,
      "reward_std": 0.4248832643032074,
      "rewards/IngredientFormatReward/mean": 0.9875520706176758,
      "rewards/IngredientFormatReward/std": 0.0813837306573987,
      "rewards/IngredientMatchReward/mean": 0.629173481464386,
      "rewards/IngredientMatchReward/std": 0.3017181038856506,
      "rewards/IngredientQuantityMatchReward/mean": 0.6195836544036866,
      "rewards/IngredientQuantityMatchReward/std": 0.403140127658844,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.41130141019821165,
      "step": 1350
    },
    {
      "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": 0.021875,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.6796875,
      "completions/min_length": 268.6,
      "epoch": 0.3769123783031989,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7041727304458618,
      "kl": 0.04741974025964737,
      "learning_rate": 7.352545618241442e-07,
      "loss": 0.0018971532583236695,
      "reward": 2.937150573730469,
      "reward_std": 0.47615134716033936,
      "rewards/IngredientFormatReward/mean": 0.9778645873069763,
      "rewards/IngredientFormatReward/std": 0.1367991030216217,
      "rewards/IngredientMatchReward/mean": 0.6257552027702331,
      "rewards/IngredientMatchReward/std": 0.2937148153781891,
      "rewards/IngredientQuantityMatchReward/mean": 0.5600932717323304,
      "rewards/IngredientQuantityMatchReward/std": 0.4284715175628662,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.3832641303539276,
      "step": 1355
    },
    {
      "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": 0.0203125,
      "completions/max_length": 511.2,
      "completions/mean_length": 399.834375,
      "completions/min_length": 275.0,
      "epoch": 0.37830319888734354,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6130369901657104,
      "kl": 0.04342401102185249,
      "learning_rate": 7.332227054614904e-07,
      "loss": 0.0017366718500852584,
      "reward": 3.0526763916015627,
      "reward_std": 0.42766509056091306,
      "rewards/IngredientFormatReward/mean": 0.9768749952316285,
      "rewards/IngredientFormatReward/std": 0.12218592166900635,
      "rewards/IngredientMatchReward/mean": 0.6191027879714965,
      "rewards/IngredientMatchReward/std": 0.3084140956401825,
      "rewards/IngredientQuantityMatchReward/mean": 0.6410735964775085,
      "rewards/IngredientQuantityMatchReward/std": 0.41126957535743713,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.3846336603164673,
      "step": 1360
    },
    {
      "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": 0.0078125,
      "completions/max_length": 506.0,
      "completions/mean_length": 391.740625,
      "completions/min_length": 240.6,
      "epoch": 0.3796940194714882,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7070378065109253,
      "kl": 0.04717907048761845,
      "learning_rate": 7.311859147664472e-07,
      "loss": 0.00188693106174469,
      "reward": 3.010082483291626,
      "reward_std": 0.44024962186813354,
      "rewards/IngredientFormatReward/mean": 0.9850000023841858,
      "rewards/IngredientFormatReward/std": 0.09434654638171196,
      "rewards/IngredientMatchReward/mean": 0.6260190844535828,
      "rewards/IngredientMatchReward/std": 0.2963875949382782,
      "rewards/IngredientQuantityMatchReward/mean": 0.6240634083747864,
      "rewards/IngredientQuantityMatchReward/std": 0.41469027996063235,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.41165693998336794,
      "step": 1365
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.61875,
      "completions/min_length": 267.8,
      "epoch": 0.3810848400556328,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6732054948806763,
      "kl": 0.5468679509591311,
      "learning_rate": 7.291442328317413e-07,
      "loss": 0.021924957633018494,
      "reward": 2.8937462329864503,
      "reward_std": 0.4447342872619629,
      "rewards/IngredientFormatReward/mean": 0.9809635400772094,
      "rewards/IngredientFormatReward/std": 0.12264752835035324,
      "rewards/IngredientMatchReward/mean": 0.6312493920326233,
      "rewards/IngredientMatchReward/std": 0.2771876215934753,
      "rewards/IngredientQuantityMatchReward/mean": 0.5799708127975464,
      "rewards/IngredientQuantityMatchReward/std": 0.43602352142333983,
      "rewards/TotalKcalExactMatchReward/mean": 0.7015625,
      "rewards/TotalKcalExactMatchReward/std": 0.4481926143169403,
      "step": 1370
    },
    {
      "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": 0.01875,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.4375,
      "completions/min_length": 286.0,
      "epoch": 0.38247566063977745,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6947128772735596,
      "kl": 0.051231970288790764,
      "learning_rate": 7.270977028535846e-07,
      "loss": 0.0020487431436777117,
      "reward": 3.081242084503174,
      "reward_std": 0.42529494166374204,
      "rewards/IngredientFormatReward/mean": 0.9795535802841187,
      "rewards/IngredientFormatReward/std": 0.13361096233129502,
      "rewards/IngredientMatchReward/mean": 0.6574689984321594,
      "rewards/IngredientMatchReward/std": 0.27973159551620486,
      "rewards/IngredientQuantityMatchReward/mean": 0.6489070296287537,
      "rewards/IngredientQuantityMatchReward/std": 0.41112475991249087,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.3904457151889801,
      "step": 1375
    },
    {
      "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": 0.01875,
      "completions/max_length": 512.2,
      "completions/mean_length": 388.3703125,
      "completions/min_length": 249.2,
      "epoch": 0.38386648122392214,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6509252190589905,
      "kl": 0.046685748361051084,
      "learning_rate": 7.250463681307588e-07,
      "loss": 0.001867828331887722,
      "reward": 3.0991830825805664,
      "reward_std": 0.4528281927108765,
      "rewards/IngredientFormatReward/mean": 0.980078125,
      "rewards/IngredientFormatReward/std": 0.1193616308271885,
      "rewards/IngredientMatchReward/mean": 0.6413913607597351,
      "rewards/IngredientMatchReward/std": 0.29768499433994294,
      "rewards/IngredientQuantityMatchReward/mean": 0.6730260252952576,
      "rewards/IngredientQuantityMatchReward/std": 0.4107559084892273,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.3941961944103241,
      "step": 1380
    },
    {
      "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": 0.009375,
      "completions/max_length": 501.4,
      "completions/mean_length": 390.2609375,
      "completions/min_length": 258.2,
      "epoch": 0.3852573018080668,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6028541922569275,
      "kl": 0.0516342549584806,
      "learning_rate": 7.229902720637013e-07,
      "loss": 0.0020651400089263918,
      "reward": 3.0714120864868164,
      "reward_std": 0.4244008719921112,
      "rewards/IngredientFormatReward/mean": 0.9860342264175415,
      "rewards/IngredientFormatReward/std": 0.08796465378254652,
      "rewards/IngredientMatchReward/mean": 0.6268959999084472,
      "rewards/IngredientMatchReward/std": 0.30808745622634887,
      "rewards/IngredientQuantityMatchReward/mean": 0.6631693601608276,
      "rewards/IngredientQuantityMatchReward/std": 0.4120981454849243,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.3891636371612549,
      "step": 1385
    },
    {
      "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": 0.028125,
      "completions/max_length": 512.8,
      "completions/mean_length": 394.9046875,
      "completions/min_length": 259.6,
      "epoch": 0.3866481223922114,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.693778395652771,
      "kl": 0.04984358986839652,
      "learning_rate": 7.20929458153586e-07,
      "loss": 0.0019936652854084967,
      "reward": 2.9926676750183105,
      "reward_std": 0.4538565337657928,
      "rewards/IngredientFormatReward/mean": 0.9710416793823242,
      "rewards/IngredientFormatReward/std": 0.1352289214730263,
      "rewards/IngredientMatchReward/mean": 0.5942381978034973,
      "rewards/IngredientMatchReward/std": 0.296366685628891,
      "rewards/IngredientQuantityMatchReward/mean": 0.5883253335952758,
      "rewards/IngredientQuantityMatchReward/std": 0.4092931866645813,
      "rewards/TotalKcalExactMatchReward/mean": 0.8390625,
      "rewards/TotalKcalExactMatchReward/std": 0.36484946608543395,
      "step": 1390
    },
    {
      "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": 0.0171875,
      "completions/max_length": 501.6,
      "completions/mean_length": 397.6265625,
      "completions/min_length": 278.0,
      "epoch": 0.38803894297635605,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6608783006668091,
      "kl": 0.04760516933165491,
      "learning_rate": 7.188639700014025e-07,
      "loss": 0.0019130157306790352,
      "reward": 3.1505064487457277,
      "reward_std": 0.3983344674110413,
      "rewards/IngredientFormatReward/mean": 0.9819270849227906,
      "rewards/IngredientFormatReward/std": 0.1014548234641552,
      "rewards/IngredientMatchReward/mean": 0.6463622331619263,
      "rewards/IngredientMatchReward/std": 0.29087430238723755,
      "rewards/IngredientQuantityMatchReward/mean": 0.6690921902656555,
      "rewards/IngredientQuantityMatchReward/std": 0.4150608241558075,
      "rewards/TotalKcalExactMatchReward/mean": 0.853125,
      "rewards/TotalKcalExactMatchReward/std": 0.3509778559207916,
      "step": 1395
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.4,
      "completions/mean_length": 395.771875,
      "completions/min_length": 267.4,
      "epoch": 0.3894297635605007,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7344890236854553,
      "kl": 0.048899111011996864,
      "learning_rate": 7.167938513070344e-07,
      "loss": 0.001956099271774292,
      "reward": 2.9775744438171388,
      "reward_std": 0.34394469261169436,
      "rewards/IngredientFormatReward/mean": 0.989661455154419,
      "rewards/IngredientFormatReward/std": 0.06796687468886375,
      "rewards/IngredientMatchReward/mean": 0.6366368889808655,
      "rewards/IngredientMatchReward/std": 0.28161700963974,
      "rewards/IngredientQuantityMatchReward/mean": 0.571588671207428,
      "rewards/IngredientQuantityMatchReward/std": 0.4268614649772644,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.4140839040279388,
      "step": 1400
    },
    {
      "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": 0.0046875,
      "completions/max_length": 503.6,
      "completions/mean_length": 391.0953125,
      "completions/min_length": 270.0,
      "epoch": 0.3908205841446453,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6491956114768982,
      "kl": 0.047348243673332034,
      "learning_rate": 7.147191458683348e-07,
      "loss": 0.0018939247354865074,
      "reward": 3.1373552322387694,
      "reward_std": 0.39182621240615845,
      "rewards/IngredientFormatReward/mean": 0.9948660731315613,
      "rewards/IngredientFormatReward/std": 0.046136388555169106,
      "rewards/IngredientMatchReward/mean": 0.6508209228515625,
      "rewards/IngredientMatchReward/std": 0.2824489861726761,
      "rewards/IngredientQuantityMatchReward/mean": 0.672918152809143,
      "rewards/IngredientQuantityMatchReward/std": 0.42214601039886473,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.38358521461486816,
      "step": 1405
    },
    {
      "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": 0.0078125,
      "completions/max_length": 511.8,
      "completions/mean_length": 393.0140625,
      "completions/min_length": 285.4,
      "epoch": 0.39221140472878996,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6753212809562683,
      "kl": 0.047298068576492366,
      "learning_rate": 7.126398975801988e-07,
      "loss": 0.001892334222793579,
      "reward": 2.993410110473633,
      "reward_std": 0.42139936685562135,
      "rewards/IngredientFormatReward/mean": 0.991614580154419,
      "rewards/IngredientFormatReward/std": 0.07206480577588081,
      "rewards/IngredientMatchReward/mean": 0.6076562523841857,
      "rewards/IngredientMatchReward/std": 0.28833829760551455,
      "rewards/IngredientQuantityMatchReward/mean": 0.6097642540931701,
      "rewards/IngredientQuantityMatchReward/std": 0.4363261044025421,
      "rewards/TotalKcalExactMatchReward/mean": 0.784375,
      "rewards/TotalKcalExactMatchReward/std": 0.3828993201255798,
      "step": 1410
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.2,
      "completions/mean_length": 391.2296875,
      "completions/min_length": 264.0,
      "epoch": 0.39360222531293465,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6234867572784424,
      "kl": 0.04924510028213262,
      "learning_rate": 7.105561504336355e-07,
      "loss": 0.001970212534070015,
      "reward": 3.0320447444915772,
      "reward_std": 0.4181003987789154,
      "rewards/IngredientFormatReward/mean": 0.9856250047683716,
      "rewards/IngredientFormatReward/std": 0.08855490759015083,
      "rewards/IngredientMatchReward/mean": 0.6343092918395996,
      "rewards/IngredientMatchReward/std": 0.30618330240249636,
      "rewards/IngredientQuantityMatchReward/mean": 0.641797935962677,
      "rewards/IngredientQuantityMatchReward/std": 0.4155651867389679,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.4193932831287384,
      "step": 1415
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.4,
      "completions/mean_length": 395.6515625,
      "completions/min_length": 264.2,
      "epoch": 0.3949930458970793,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6549289226531982,
      "kl": 0.048028711089864375,
      "learning_rate": 7.084679485148375e-07,
      "loss": 0.0019211186096072196,
      "reward": 3.0376140594482424,
      "reward_std": 0.3858941853046417,
      "rewards/IngredientFormatReward/mean": 0.9880208253860474,
      "rewards/IngredientFormatReward/std": 0.08144994545727968,
      "rewards/IngredientMatchReward/mean": 0.5925304055213928,
      "rewards/IngredientMatchReward/std": 0.3050950050354004,
      "rewards/IngredientQuantityMatchReward/mean": 0.6211253523826599,
      "rewards/IngredientQuantityMatchReward/std": 0.4290615260601044,
      "rewards/TotalKcalExactMatchReward/mean": 0.8359375,
      "rewards/TotalKcalExactMatchReward/std": 0.35664623975753784,
      "step": 1420
    },
    {
      "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": 0.0203125,
      "completions/max_length": 511.4,
      "completions/mean_length": 392.915625,
      "completions/min_length": 250.6,
      "epoch": 0.3963838664812239,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.5656110048294067,
      "kl": 0.061237924033775926,
      "learning_rate": 7.06375336004247e-07,
      "loss": 0.0024501249194145204,
      "reward": 3.006493091583252,
      "reward_std": 0.47243286967277526,
      "rewards/IngredientFormatReward/mean": 0.9774218797683716,
      "rewards/IngredientFormatReward/std": 0.12839448899030687,
      "rewards/IngredientMatchReward/mean": 0.6348812222480774,
      "rewards/IngredientMatchReward/std": 0.301179563999176,
      "rewards/IngredientQuantityMatchReward/mean": 0.594189977645874,
      "rewards/IngredientQuantityMatchReward/std": 0.42939573526382446,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3994594097137451,
      "step": 1425
    },
    {
      "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": 0.0234375,
      "completions/max_length": 507.8,
      "completions/mean_length": 391.134375,
      "completions/min_length": 266.0,
      "epoch": 0.39777468706536856,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6568824648857117,
      "kl": 0.05042944210581481,
      "learning_rate": 7.042783571756228e-07,
      "loss": 0.0020168770104646684,
      "reward": 3.0168625831604006,
      "reward_std": 0.46565952301025393,
      "rewards/IngredientFormatReward/mean": 0.9702976226806641,
      "rewards/IngredientFormatReward/std": 0.1451386496424675,
      "rewards/IngredientMatchReward/mean": 0.6251142144203186,
      "rewards/IngredientMatchReward/std": 0.2843863904476166,
      "rewards/IngredientQuantityMatchReward/mean": 0.6386382818222046,
      "rewards/IngredientQuantityMatchReward/std": 0.4032295286655426,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.40105786323547366,
      "step": 1430
    },
    {
      "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": 0.0078125,
      "completions/max_length": 510.2,
      "completions/mean_length": 391.8359375,
      "completions/min_length": 276.4,
      "epoch": 0.3991655076495132,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6645727753639221,
      "kl": 0.05089432932436466,
      "learning_rate": 7.021770563951017e-07,
      "loss": 0.002035984769463539,
      "reward": 3.065924549102783,
      "reward_std": 0.42762036323547364,
      "rewards/IngredientFormatReward/mean": 0.990416657924652,
      "rewards/IngredientFormatReward/std": 0.07949181646108627,
      "rewards/IngredientMatchReward/mean": 0.6396770238876343,
      "rewards/IngredientMatchReward/std": 0.2999457597732544,
      "rewards/IngredientQuantityMatchReward/mean": 0.6295809030532837,
      "rewards/IngredientQuantityMatchReward/std": 0.4113676190376282,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3949193716049194,
      "step": 1435
    },
    {
      "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": 0.0171875,
      "completions/max_length": 510.4,
      "completions/mean_length": 393.7078125,
      "completions/min_length": 288.2,
      "epoch": 0.40055632823365783,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6648370623588562,
      "kl": 0.051764291478320956,
      "learning_rate": 7.000714781202613e-07,
      "loss": 0.0020709920674562452,
      "reward": 3.0830320358276366,
      "reward_std": 0.49956942200660703,
      "rewards/IngredientFormatReward/mean": 0.979479169845581,
      "rewards/IngredientFormatReward/std": 0.11220841258764266,
      "rewards/IngredientMatchReward/mean": 0.6905896306037903,
      "rewards/IngredientMatchReward/std": 0.27779283225536344,
      "rewards/IngredientQuantityMatchReward/mean": 0.6395256638526916,
      "rewards/IngredientQuantityMatchReward/std": 0.4275822460651398,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.4142127275466919,
      "step": 1440
    },
    {
      "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": 0.009375,
      "completions/max_length": 501.4,
      "completions/mean_length": 382.08125,
      "completions/min_length": 248.0,
      "epoch": 0.4019471488178025,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7125925421714783,
      "kl": 0.05212058457545936,
      "learning_rate": 6.97961666899179e-07,
      "loss": 0.002085070312023163,
      "reward": 2.974539279937744,
      "reward_std": 0.4162575304508209,
      "rewards/IngredientFormatReward/mean": 0.9879166603088378,
      "rewards/IngredientFormatReward/std": 0.08713518846780062,
      "rewards/IngredientMatchReward/mean": 0.6231901168823242,
      "rewards/IngredientMatchReward/std": 0.2897790253162384,
      "rewards/IngredientQuantityMatchReward/mean": 0.5634325981140137,
      "rewards/IngredientQuantityMatchReward/std": 0.4310961186885834,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3805993765592575,
      "step": 1445
    },
    {
      "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": 0.0140625,
      "completions/max_length": 509.8,
      "completions/mean_length": 387.584375,
      "completions/min_length": 254.2,
      "epoch": 0.40333796940194716,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.7150693535804749,
      "kl": 0.0496115677524358,
      "learning_rate": 6.958476673694888e-07,
      "loss": 0.0019844066351652145,
      "reward": 3.0862731456756594,
      "reward_std": 0.4122806191444397,
      "rewards/IngredientFormatReward/mean": 0.9861458301544189,
      "rewards/IngredientFormatReward/std": 0.10063027888536454,
      "rewards/IngredientMatchReward/mean": 0.6254792809486389,
      "rewards/IngredientMatchReward/std": 0.27667003870010376,
      "rewards/IngredientQuantityMatchReward/mean": 0.6402730107307434,
      "rewards/IngredientQuantityMatchReward/std": 0.4204639375209808,
      "rewards/TotalKcalExactMatchReward/mean": 0.834375,
      "rewards/TotalKcalExactMatchReward/std": 0.363431441783905,
      "step": 1450
    },
    {
      "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": 0.0140625,
      "completions/max_length": 512.8,
      "completions/mean_length": 393.0984375,
      "completions/min_length": 285.8,
      "epoch": 0.4047287899860918,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6466384530067444,
      "kl": 0.050380307948216796,
      "learning_rate": 6.93729524257438e-07,
      "loss": 0.002015003561973572,
      "reward": 3.1226918697357178,
      "reward_std": 0.43267813324928284,
      "rewards/IngredientFormatReward/mean": 0.9815885305404664,
      "rewards/IngredientFormatReward/std": 0.11584979891777039,
      "rewards/IngredientMatchReward/mean": 0.6187383890151977,
      "rewards/IngredientMatchReward/std": 0.2922286242246628,
      "rewards/IngredientQuantityMatchReward/mean": 0.6879899024963378,
      "rewards/IngredientQuantityMatchReward/std": 0.37570016384124755,
      "rewards/TotalKcalExactMatchReward/mean": 0.834375,
      "rewards/TotalKcalExactMatchReward/std": 0.35577985644340515,
      "step": 1455
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.3609375,
      "completions/min_length": 266.8,
      "epoch": 0.40611961057023643,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6766321659088135,
      "kl": 0.08121923571452498,
      "learning_rate": 6.916072823769398e-07,
      "loss": 0.003253092244267464,
      "reward": 2.9472359657287597,
      "reward_std": 0.44583975076675414,
      "rewards/IngredientFormatReward/mean": 0.9813020825386047,
      "rewards/IngredientFormatReward/std": 0.11078893411904574,
      "rewards/IngredientMatchReward/mean": 0.6114800453186036,
      "rewards/IngredientMatchReward/std": 0.2834586203098297,
      "rewards/IngredientQuantityMatchReward/mean": 0.5935163855552673,
      "rewards/IngredientQuantityMatchReward/std": 0.42633000016212463,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.4214301824569702,
      "step": 1460
    },
    {
      "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": 0.0140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 393.3234375,
      "completions/min_length": 279.2,
      "epoch": 0.40751043115438107,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6546911597251892,
      "kl": 0.04946425724774599,
      "learning_rate": 6.894809866286269e-07,
      "loss": 0.0019789854064583778,
      "reward": 3.0060822010040282,
      "reward_std": 0.42944018840789794,
      "rewards/IngredientFormatReward/mean": 0.9875,
      "rewards/IngredientFormatReward/std": 0.1100594773888588,
      "rewards/IngredientMatchReward/mean": 0.6255816102027894,
      "rewards/IngredientMatchReward/std": 0.2829381048679352,
      "rewards/IngredientQuantityMatchReward/mean": 0.646125602722168,
      "rewards/IngredientQuantityMatchReward/std": 0.40343931317329407,
      "rewards/TotalKcalExactMatchReward/mean": 0.746875,
      "rewards/TotalKcalExactMatchReward/std": 0.4242517352104187,
      "step": 1465
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.0,
      "completions/mean_length": 391.7578125,
      "completions/min_length": 267.0,
      "epoch": 0.4089012517385257,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.644096314907074,
      "kl": 0.04821961410343647,
      "learning_rate": 6.873506819988985e-07,
      "loss": 0.0019289355725049973,
      "reward": 3.0299394607543944,
      "reward_std": 0.38674944043159487,
      "rewards/IngredientFormatReward/mean": 0.9899999976158143,
      "rewards/IngredientFormatReward/std": 0.0847195789217949,
      "rewards/IngredientMatchReward/mean": 0.6211880207061767,
      "rewards/IngredientMatchReward/std": 0.27895829677581785,
      "rewards/IngredientQuantityMatchReward/mean": 0.6437513709068299,
      "rewards/IngredientQuantityMatchReward/std": 0.42142691612243655,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.41285876035690305,
      "step": 1470
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.0,
      "completions/mean_length": 387.296875,
      "completions/min_length": 282.8,
      "epoch": 0.4102920723226704,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6414361000061035,
      "kl": 0.05062154838815332,
      "learning_rate": 6.852164135589724e-07,
      "loss": 0.0020248919725418093,
      "reward": 3.091634464263916,
      "reward_std": 0.39072946906089784,
      "rewards/IngredientFormatReward/mean": 0.9922135353088379,
      "rewards/IngredientFormatReward/std": 0.062381112948060036,
      "rewards/IngredientMatchReward/mean": 0.6525762677192688,
      "rewards/IngredientMatchReward/std": 0.2883527398109436,
      "rewards/IngredientQuantityMatchReward/mean": 0.6530946910381317,
      "rewards/IngredientQuantityMatchReward/std": 0.39076480865478513,
      "rewards/TotalKcalExactMatchReward/mean": 0.79375,
      "rewards/TotalKcalExactMatchReward/std": 0.3955763101577759,
      "step": 1475
    },
    {
      "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": 0.0109375,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.8875,
      "completions/min_length": 266.4,
      "epoch": 0.41168289290681503,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6684796810150146,
      "kl": 0.052700078953057526,
      "learning_rate": 6.83078226463928e-07,
      "loss": 0.0021081626415252685,
      "reward": 3.0118268489837647,
      "reward_std": 0.41152234077453614,
      "rewards/IngredientFormatReward/mean": 0.9853869080543518,
      "rewards/IngredientFormatReward/std": 0.09761874377727509,
      "rewards/IngredientMatchReward/mean": 0.5979613065719604,
      "rewards/IngredientMatchReward/std": 0.2978090465068817,
      "rewards/IngredientQuantityMatchReward/mean": 0.6112910985946656,
      "rewards/IngredientQuantityMatchReward/std": 0.4339863657951355,
      "rewards/TotalKcalExactMatchReward/mean": 0.8171875,
      "rewards/TotalKcalExactMatchReward/std": 0.37414880394935607,
      "step": 1480
    },
    {
      "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": 0.00625,
      "completions/max_length": 511.8,
      "completions/mean_length": 386.5359375,
      "completions/min_length": 260.2,
      "epoch": 0.41307371349095967,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6525306105613708,
      "kl": 0.0471834511961788,
      "learning_rate": 6.809361659517527e-07,
      "loss": 0.0018876727670431137,
      "reward": 3.0948177337646485,
      "reward_std": 0.369168359041214,
      "rewards/IngredientFormatReward/mean": 0.990234375,
      "rewards/IngredientFormatReward/std": 0.08957751505076886,
      "rewards/IngredientMatchReward/mean": 0.6415643811225891,
      "rewards/IngredientMatchReward/std": 0.2889476329088211,
      "rewards/IngredientQuantityMatchReward/mean": 0.6364565491676331,
      "rewards/IngredientQuantityMatchReward/std": 0.4115687429904938,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.37568016052246095,
      "step": 1485
    },
    {
      "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": 0.0171875,
      "completions/max_length": 508.2,
      "completions/mean_length": 393.3828125,
      "completions/min_length": 255.2,
      "epoch": 0.4144645340751043,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.637024998664856,
      "kl": 0.04654967598617077,
      "learning_rate": 6.78790277342385e-07,
      "loss": 0.0018621478229761124,
      "reward": 3.1724345684051514,
      "reward_std": 0.40526561737060546,
      "rewards/IngredientFormatReward/mean": 0.9826022982597351,
      "rewards/IngredientFormatReward/std": 0.10656376946717501,
      "rewards/IngredientMatchReward/mean": 0.6608142852783203,
      "rewards/IngredientMatchReward/std": 0.28252187967300413,
      "rewards/IngredientQuantityMatchReward/mean": 0.7321430206298828,
      "rewards/IngredientQuantityMatchReward/std": 0.3806688904762268,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.3964962542057037,
      "step": 1490
    },
    {
      "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": 0.009375,
      "completions/max_length": 510.6,
      "completions/mean_length": 393.2796875,
      "completions/min_length": 271.4,
      "epoch": 0.41585535465924894,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.635031521320343,
      "kl": 0.048082446400076154,
      "learning_rate": 6.766406060367543e-07,
      "loss": 0.001923457160592079,
      "reward": 3.1764147758483885,
      "reward_std": 0.4066992163658142,
      "rewards/IngredientFormatReward/mean": 0.9915624976158142,
      "rewards/IngredientFormatReward/std": 0.06838876903057098,
      "rewards/IngredientMatchReward/mean": 0.6770944952964782,
      "rewards/IngredientMatchReward/std": 0.29843786358833313,
      "rewards/IngredientQuantityMatchReward/mean": 0.6796328067779541,
      "rewards/IngredientQuantityMatchReward/std": 0.3973805606365204,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.36534249782562256,
      "step": 1495
    },
    {
      "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": 0.0109375,
      "completions/max_length": 502.8,
      "completions/mean_length": 387.5390625,
      "completions/min_length": 267.0,
      "epoch": 0.4172461752433936,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.706847071647644,
      "kl": 0.04723797417245805,
      "learning_rate": 6.744871975158215e-07,
      "loss": 0.001889602467417717,
      "reward": 3.0415586471557616,
      "reward_std": 0.4428049921989441,
      "rewards/IngredientFormatReward/mean": 0.9885416746139526,
      "rewards/IngredientFormatReward/std": 0.0781378224492073,
      "rewards/IngredientMatchReward/mean": 0.6192667484283447,
      "rewards/IngredientMatchReward/std": 0.27974243760108947,
      "rewards/IngredientQuantityMatchReward/mean": 0.6540627002716064,
      "rewards/IngredientQuantityMatchReward/std": 0.42795388102531434,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.412675142288208,
      "step": 1500
    },
    {
      "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": 0.0109375,
      "completions/max_length": 502.4,
      "completions/mean_length": 386.4796875,
      "completions/min_length": 258.0,
      "epoch": 0.41863699582753827,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6618068814277649,
      "kl": 0.05870349365286529,
      "learning_rate": 6.723300973396166e-07,
      "loss": 0.0023463867604732514,
      "reward": 3.161744213104248,
      "reward_std": 0.42260637879371643,
      "rewards/IngredientFormatReward/mean": 0.9826562404632568,
      "rewards/IngredientFormatReward/std": 0.09131253361701966,
      "rewards/IngredientMatchReward/mean": 0.6721211671829224,
      "rewards/IngredientMatchReward/std": 0.3050087809562683,
      "rewards/IngredientQuantityMatchReward/mean": 0.6444667696952819,
      "rewards/IngredientQuantityMatchReward/std": 0.39916192293167113,
      "rewards/TotalKcalExactMatchReward/mean": 0.8625,
      "rewards/TotalKcalExactMatchReward/std": 0.34389914870262145,
      "step": 1505
    },
    {
      "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": 0.0015625,
      "completions/max_length": 501.0,
      "completions/mean_length": 391.0140625,
      "completions/min_length": 275.6,
      "epoch": 0.4200278164116829,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6646866202354431,
      "kl": 0.04706322541460395,
      "learning_rate": 6.701693511462743e-07,
      "loss": 0.0018825791776180267,
      "reward": 3.094019556045532,
      "reward_std": 0.3848333120346069,
      "rewards/IngredientFormatReward/mean": 0.9956249952316284,
      "rewards/IngredientFormatReward/std": 0.035187306255102156,
      "rewards/IngredientMatchReward/mean": 0.6332756876945496,
      "rewards/IngredientMatchReward/std": 0.28339311182498933,
      "rewards/IngredientQuantityMatchReward/mean": 0.6979313969612122,
      "rewards/IngredientQuantityMatchReward/std": 0.39676953554153443,
      "rewards/TotalKcalExactMatchReward/mean": 0.7671875,
      "rewards/TotalKcalExactMatchReward/std": 0.4147461235523224,
      "step": 1510
    },
    {
      "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": 0.0171875,
      "completions/max_length": 509.0,
      "completions/mean_length": 389.246875,
      "completions/min_length": 254.6,
      "epoch": 0.42141863699582754,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6115160584449768,
      "kl": 0.04909869590774178,
      "learning_rate": 6.680050046510687e-07,
      "loss": 0.0019641295075416565,
      "reward": 3.0600674152374268,
      "reward_std": 0.41310828924179077,
      "rewards/IngredientFormatReward/mean": 0.9803515672683716,
      "rewards/IngredientFormatReward/std": 0.10586707219481468,
      "rewards/IngredientMatchReward/mean": 0.6523542881011963,
      "rewards/IngredientMatchReward/std": 0.2938463926315308,
      "rewards/IngredientQuantityMatchReward/mean": 0.6664240956306458,
      "rewards/IngredientQuantityMatchReward/std": 0.40689346194267273,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.3891270637512207,
      "step": 1515
    },
    {
      "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": 0.009375,
      "completions/max_length": 506.0,
      "completions/mean_length": 389.6171875,
      "completions/min_length": 262.6,
      "epoch": 0.4228094575799722,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6689432859420776,
      "kl": 0.051034286990761756,
      "learning_rate": 6.658371036454464e-07,
      "loss": 0.002041558176279068,
      "reward": 3.0063760757446287,
      "reward_std": 0.4658251166343689,
      "rewards/IngredientFormatReward/mean": 0.9912500023841858,
      "rewards/IngredientFormatReward/std": 0.06588455513119698,
      "rewards/IngredientMatchReward/mean": 0.5811309456825257,
      "rewards/IngredientMatchReward/std": 0.2976264595985413,
      "rewards/IngredientQuantityMatchReward/mean": 0.6136826276779175,
      "rewards/IngredientQuantityMatchReward/std": 0.4315095841884613,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.3791858911514282,
      "step": 1520
    },
    {
      "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": 0.0109375,
      "completions/max_length": 498.4,
      "completions/mean_length": 391.5,
      "completions/min_length": 246.0,
      "epoch": 0.4242002781641168,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6664140224456787,
      "kl": 0.045389461284503343,
      "learning_rate": 6.636656939960569e-07,
      "loss": 0.0018157461658120156,
      "reward": 3.1146235942840574,
      "reward_std": 0.41344255208969116,
      "rewards/IngredientFormatReward/mean": 0.9910788655281066,
      "rewards/IngredientFormatReward/std": 0.05945278406143188,
      "rewards/IngredientMatchReward/mean": 0.6309400081634522,
      "rewards/IngredientMatchReward/std": 0.30454255938529967,
      "rewards/IngredientQuantityMatchReward/mean": 0.6738548040390014,
      "rewards/IngredientQuantityMatchReward/std": 0.39767482280731203,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.37096219062805175,
      "step": 1525
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.4,
      "completions/mean_length": 387.33125,
      "completions/min_length": 264.2,
      "epoch": 0.42559109874826145,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6901034712791443,
      "kl": 0.049771115835756066,
      "learning_rate": 6.614908216437831e-07,
      "loss": 0.0019908515736460687,
      "reward": 3.086984157562256,
      "reward_std": 0.46495330333709717,
      "rewards/IngredientFormatReward/mean": 0.9900111675262451,
      "rewards/IngredientFormatReward/std": 0.08576798439025879,
      "rewards/IngredientMatchReward/mean": 0.6765054702758789,
      "rewards/IngredientMatchReward/std": 0.27772613167762755,
      "rewards/IngredientQuantityMatchReward/mean": 0.6423426389694213,
      "rewards/IngredientQuantityMatchReward/std": 0.4157872200012207,
      "rewards/TotalKcalExactMatchReward/mean": 0.778125,
      "rewards/TotalKcalExactMatchReward/std": 0.4095184445381165,
      "step": 1530
    },
    {
      "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": 0.0203125,
      "completions/max_length": 511.0,
      "completions/mean_length": 400.621875,
      "completions/min_length": 261.8,
      "epoch": 0.42698191933240615,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6421478986740112,
      "kl": 0.04838813366368413,
      "learning_rate": 6.593125326027689e-07,
      "loss": 0.0019358258694410324,
      "reward": 2.994578552246094,
      "reward_std": 0.44342033863067626,
      "rewards/IngredientFormatReward/mean": 0.9785788774490356,
      "rewards/IngredientFormatReward/std": 0.12433337345719338,
      "rewards/IngredientMatchReward/mean": 0.5929954409599304,
      "rewards/IngredientMatchReward/std": 0.30798673033714297,
      "rewards/IngredientQuantityMatchReward/mean": 0.6261291861534118,
      "rewards/IngredientQuantityMatchReward/std": 0.43534895181655886,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.39741336107254027,
      "step": 1535
    },
    {
      "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": 0.00625,
      "completions/max_length": 512.0,
      "completions/mean_length": 394.778125,
      "completions/min_length": 273.6,
      "epoch": 0.4283727399165508,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6390155553817749,
      "kl": 0.04921357906423509,
      "learning_rate": 6.571308729594449e-07,
      "loss": 0.0019685806706547736,
      "reward": 3.1461612224578857,
      "reward_std": 0.39753777980804444,
      "rewards/IngredientFormatReward/mean": 0.9929166674613953,
      "rewards/IngredientFormatReward/std": 0.07146886438131332,
      "rewards/IngredientMatchReward/mean": 0.7030505895614624,
      "rewards/IngredientMatchReward/std": 0.2847248435020447,
      "rewards/IngredientQuantityMatchReward/mean": 0.6126938581466674,
      "rewards/IngredientQuantityMatchReward/std": 0.42437596917152404,
      "rewards/TotalKcalExactMatchReward/mean": 0.8375,
      "rewards/TotalKcalExactMatchReward/std": 0.3570310711860657,
      "step": 1540
    },
    {
      "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": 0.0171875,
      "completions/max_length": 511.4,
      "completions/mean_length": 390.5375,
      "completions/min_length": 265.0,
      "epoch": 0.4297635605006954,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6613978147506714,
      "kl": 0.05156162865459919,
      "learning_rate": 6.549458888715555e-07,
      "loss": 0.0020627494901418685,
      "reward": 3.0349883079528808,
      "reward_std": 0.4523595035076141,
      "rewards/IngredientFormatReward/mean": 0.9807142972946167,
      "rewards/IngredientFormatReward/std": 0.1297312021255493,
      "rewards/IngredientMatchReward/mean": 0.6188566565513611,
      "rewards/IngredientMatchReward/std": 0.28720327019691466,
      "rewards/IngredientQuantityMatchReward/mean": 0.6479173898696899,
      "rewards/IngredientQuantityMatchReward/std": 0.43776082396507265,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.3997353255748749,
      "step": 1545
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.8234375,
      "completions/min_length": 270.8,
      "epoch": 0.43115438108484005,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6673920154571533,
      "kl": 0.04912932915613055,
      "learning_rate": 6.527576265671794e-07,
      "loss": 0.0019652603194117548,
      "reward": 3.0716404914855957,
      "reward_std": 0.47040929794311526,
      "rewards/IngredientFormatReward/mean": 0.9731250047683716,
      "rewards/IngredientFormatReward/std": 0.1522574841976166,
      "rewards/IngredientMatchReward/mean": 0.6390248894691467,
      "rewards/IngredientMatchReward/std": 0.3047917664051056,
      "rewards/IngredientQuantityMatchReward/mean": 0.6938655853271485,
      "rewards/IngredientQuantityMatchReward/std": 0.39961657524108884,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.41862271428108216,
      "step": 1550
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 400.134375,
      "completions/min_length": 276.2,
      "epoch": 0.4325452016689847,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.653672993183136,
      "kl": 0.049115463299676775,
      "learning_rate": 6.505661323437544e-07,
      "loss": 0.001964777708053589,
      "reward": 3.160717725753784,
      "reward_std": 0.4397841155529022,
      "rewards/IngredientFormatReward/mean": 0.9830952405929565,
      "rewards/IngredientFormatReward/std": 0.12367947846651077,
      "rewards/IngredientMatchReward/mean": 0.6457465529441834,
      "rewards/IngredientMatchReward/std": 0.3012327909469604,
      "rewards/IngredientQuantityMatchReward/mean": 0.7271884560585022,
      "rewards/IngredientQuantityMatchReward/std": 0.38396180272102354,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.37747406363487246,
      "step": 1555
    },
    {
      "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": 0.0125,
      "completions/max_length": 509.8,
      "completions/mean_length": 397.1421875,
      "completions/min_length": 256.8,
      "epoch": 0.4339360222531293,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.617830753326416,
      "kl": 0.05066547570750117,
      "learning_rate": 6.483714525670955e-07,
      "loss": 0.0020269980654120446,
      "reward": 3.1229011535644533,
      "reward_std": 0.40159772634506224,
      "rewards/IngredientFormatReward/mean": 0.9884374976158142,
      "rewards/IngredientFormatReward/std": 0.09104880094528198,
      "rewards/IngredientMatchReward/mean": 0.6592550873756409,
      "rewards/IngredientMatchReward/std": 0.29240604639053347,
      "rewards/IngredientQuantityMatchReward/mean": 0.6814584970474243,
      "rewards/IngredientQuantityMatchReward/std": 0.3873124122619629,
      "rewards/TotalKcalExactMatchReward/mean": 0.79375,
      "rewards/TotalKcalExactMatchReward/std": 0.3895157277584076,
      "step": 1560
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.6,
      "completions/mean_length": 393.6171875,
      "completions/min_length": 235.4,
      "epoch": 0.43532684283727396,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6887730360031128,
      "kl": 0.048044915916398166,
      "learning_rate": 6.46173633670416e-07,
      "loss": 0.001921633630990982,
      "reward": 3.046199083328247,
      "reward_std": 0.40088788866996766,
      "rewards/IngredientFormatReward/mean": 0.9917931675910949,
      "rewards/IngredientFormatReward/std": 0.0687427580356598,
      "rewards/IngredientMatchReward/mean": 0.6264453172683716,
      "rewards/IngredientMatchReward/std": 0.30637386441230774,
      "rewards/IngredientQuantityMatchReward/mean": 0.6310855627059937,
      "rewards/IngredientQuantityMatchReward/std": 0.434447181224823,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.3885793924331665,
      "step": 1565
    },
    {
      "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": 0.025,
      "completions/max_length": 513.0,
      "completions/mean_length": 398.70625,
      "completions/min_length": 262.8,
      "epoch": 0.43671766342141866,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7046116590499878,
      "kl": 0.049918401753529906,
      "learning_rate": 6.43972722153343e-07,
      "loss": 0.0019969020038843157,
      "reward": 2.955770397186279,
      "reward_std": 0.44179911613464357,
      "rewards/IngredientFormatReward/mean": 0.9754910826683044,
      "rewards/IngredientFormatReward/std": 0.1414255291223526,
      "rewards/IngredientMatchReward/mean": 0.6252684831619263,
      "rewards/IngredientMatchReward/std": 0.31577845811843874,
      "rewards/IngredientQuantityMatchReward/mean": 0.6143857598304748,
      "rewards/IngredientQuantityMatchReward/std": 0.42812495231628417,
      "rewards/TotalKcalExactMatchReward/mean": 0.740625,
      "rewards/TotalKcalExactMatchReward/std": 0.4230214834213257,
      "step": 1570
    },
    {
      "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": 0.0078125,
      "completions/max_length": 511.4,
      "completions/mean_length": 396.0625,
      "completions/min_length": 255.2,
      "epoch": 0.4381084840055633,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6653915643692017,
      "kl": 0.07790568340569734,
      "learning_rate": 6.417687645809358e-07,
      "loss": 0.0031152090057730676,
      "reward": 3.0399940490722654,
      "reward_std": 0.4159983456134796,
      "rewards/IngredientFormatReward/mean": 0.9914843797683716,
      "rewards/IngredientFormatReward/std": 0.07876742035150527,
      "rewards/IngredientMatchReward/mean": 0.6409877419471741,
      "rewards/IngredientMatchReward/std": 0.2738662242889404,
      "rewards/IngredientQuantityMatchReward/mean": 0.6075219869613647,
      "rewards/IngredientQuantityMatchReward/std": 0.41345948576927183,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3890666842460632,
      "step": 1575
    },
    {
      "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": 0.0140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.325,
      "completions/min_length": 271.4,
      "epoch": 0.43949930458970793,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6310789585113525,
      "kl": 0.051219089329242705,
      "learning_rate": 6.395618075826987e-07,
      "loss": 0.002048930525779724,
      "reward": 3.0662962436676025,
      "reward_std": 0.4428023874759674,
      "rewards/IngredientFormatReward/mean": 0.9847916722297668,
      "rewards/IngredientFormatReward/std": 0.11453506052494049,
      "rewards/IngredientMatchReward/mean": 0.6471811294555664,
      "rewards/IngredientMatchReward/std": 0.306696617603302,
      "rewards/IngredientQuantityMatchReward/mean": 0.6358859181404114,
      "rewards/IngredientQuantityMatchReward/std": 0.41814287304878234,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.3955422639846802,
      "step": 1580
    },
    {
      "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": 0.0125,
      "completions/max_length": 506.6,
      "completions/mean_length": 389.2515625,
      "completions/min_length": 267.4,
      "epoch": 0.44089012517385257,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7244879007339478,
      "kl": 0.05323730851523578,
      "learning_rate": 6.373518978515957e-07,
      "loss": 0.0021291349083185198,
      "reward": 3.0412669658660887,
      "reward_std": 0.44353756308555603,
      "rewards/IngredientFormatReward/mean": 0.98671875,
      "rewards/IngredientFormatReward/std": 0.09624509960412979,
      "rewards/IngredientMatchReward/mean": 0.6482967615127564,
      "rewards/IngredientMatchReward/std": 0.27869592010974886,
      "rewards/IngredientQuantityMatchReward/mean": 0.6656264781951904,
      "rewards/IngredientQuantityMatchReward/std": 0.41459755301475526,
      "rewards/TotalKcalExactMatchReward/mean": 0.740625,
      "rewards/TotalKcalExactMatchReward/std": 0.4251006543636322,
      "step": 1585
    },
    {
      "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": 0.009375,
      "completions/max_length": 507.4,
      "completions/mean_length": 392.8515625,
      "completions/min_length": 272.8,
      "epoch": 0.4422809457579972,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6723690032958984,
      "kl": 0.048212344851344825,
      "learning_rate": 6.351390821430626e-07,
      "loss": 0.0019285336136817932,
      "reward": 3.0731667995452883,
      "reward_std": 0.4329515159130096,
      "rewards/IngredientFormatReward/mean": 0.99140625,
      "rewards/IngredientFormatReward/std": 0.055959947593510154,
      "rewards/IngredientMatchReward/mean": 0.6085950136184692,
      "rewards/IngredientMatchReward/std": 0.3112369000911713,
      "rewards/IngredientQuantityMatchReward/mean": 0.7184779644012451,
      "rewards/IngredientQuantityMatchReward/std": 0.376867213845253,
      "rewards/TotalKcalExactMatchReward/mean": 0.7546875,
      "rewards/TotalKcalExactMatchReward/std": 0.4265890300273895,
      "step": 1590
    },
    {
      "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": 0.00625,
      "completions/max_length": 503.2,
      "completions/mean_length": 393.8125,
      "completions/min_length": 281.4,
      "epoch": 0.44367176634214184,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.626299262046814,
      "kl": 0.046462705451995136,
      "learning_rate": 6.329234072740169e-07,
      "loss": 0.0018586689606308938,
      "reward": 3.0387376308441163,
      "reward_std": 0.38438517451286314,
      "rewards/IngredientFormatReward/mean": 0.9895573019981384,
      "rewards/IngredientFormatReward/std": 0.08768100291490555,
      "rewards/IngredientMatchReward/mean": 0.6520031332969666,
      "rewards/IngredientMatchReward/std": 0.29606057405471803,
      "rewards/IngredientQuantityMatchReward/mean": 0.659677255153656,
      "rewards/IngredientQuantityMatchReward/std": 0.41896759867668154,
      "rewards/TotalKcalExactMatchReward/mean": 0.7375,
      "rewards/TotalKcalExactMatchReward/std": 0.43825141787528993,
      "step": 1595
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.2,
      "completions/mean_length": 394.828125,
      "completions/min_length": 272.4,
      "epoch": 0.44506258692628653,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6427936553955078,
      "kl": 0.046727404044941065,
      "learning_rate": 6.307049201218683e-07,
      "loss": 0.001869230531156063,
      "reward": 3.151887559890747,
      "reward_std": 0.4053348183631897,
      "rewards/IngredientFormatReward/mean": 0.9919270753860474,
      "rewards/IngredientFormatReward/std": 0.055418914556503295,
      "rewards/IngredientMatchReward/mean": 0.6315675854682923,
      "rewards/IngredientMatchReward/std": 0.29659752249717714,
      "rewards/IngredientQuantityMatchReward/mean": 0.7205803394317627,
      "rewards/IngredientQuantityMatchReward/std": 0.3834276497364044,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.39012913703918456,
      "step": 1600
    },
    {
      "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": 0.0171875,
      "completions/max_length": 512.8,
      "completions/mean_length": 396.3390625,
      "completions/min_length": 266.4,
      "epoch": 0.44645340751043117,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6290833950042725,
      "kl": 0.047926258062943816,
      "learning_rate": 6.284836676235261e-07,
      "loss": 0.0019170723855495453,
      "reward": 3.0292068004608153,
      "reward_std": 0.40457791090011597,
      "rewards/IngredientFormatReward/mean": 0.9802083253860474,
      "rewards/IngredientFormatReward/std": 0.11273946780711412,
      "rewards/IngredientMatchReward/mean": 0.6371831655502319,
      "rewards/IngredientMatchReward/std": 0.29279580116271975,
      "rewards/IngredientQuantityMatchReward/mean": 0.6602527856826782,
      "rewards/IngredientQuantityMatchReward/std": 0.42310287356376647,
      "rewards/TotalKcalExactMatchReward/mean": 0.7515625,
      "rewards/TotalKcalExactMatchReward/std": 0.42987608909606934,
      "step": 1605
    },
    {
      "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": 0.009375,
      "completions/max_length": 511.8,
      "completions/mean_length": 390.19375,
      "completions/min_length": 262.6,
      "epoch": 0.4478442280945758,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6547244191169739,
      "kl": 0.04830891098827124,
      "learning_rate": 6.262596967744068e-07,
      "loss": 0.001932542771100998,
      "reward": 3.120107126235962,
      "reward_std": 0.38998834490776063,
      "rewards/IngredientFormatReward/mean": 0.9877306580543518,
      "rewards/IngredientFormatReward/std": 0.09338119924068451,
      "rewards/IngredientMatchReward/mean": 0.6428608655929565,
      "rewards/IngredientMatchReward/std": 0.30790755748748777,
      "rewards/IngredientQuantityMatchReward/mean": 0.6910781860351562,
      "rewards/IngredientQuantityMatchReward/std": 0.41570178270339964,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.35660522133111955,
      "step": 1610
    },
    {
      "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": 0.0046875,
      "completions/max_length": 507.4,
      "completions/mean_length": 391.8015625,
      "completions/min_length": 268.0,
      "epoch": 0.44923504867872044,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6382926106452942,
      "kl": 0.04987063794396818,
      "learning_rate": 6.240330546274393e-07,
      "loss": 0.0019953008741140366,
      "reward": 3.0715692043304443,
      "reward_std": 0.3866082727909088,
      "rewards/IngredientFormatReward/mean": 0.9904687404632568,
      "rewards/IngredientFormatReward/std": 0.07038679048418998,
      "rewards/IngredientMatchReward/mean": 0.6318895101547242,
      "rewards/IngredientMatchReward/std": 0.30490061044692995,
      "rewards/IngredientQuantityMatchReward/mean": 0.6007733345031738,
      "rewards/IngredientQuantityMatchReward/std": 0.41641207933425906,
      "rewards/TotalKcalExactMatchReward/mean": 0.8484375,
      "rewards/TotalKcalExactMatchReward/std": 0.3494331300258636,
      "step": 1615
    },
    {
      "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": 0.0125,
      "completions/max_length": 507.6,
      "completions/mean_length": 392.3875,
      "completions/min_length": 259.4,
      "epoch": 0.4506258692628651,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.654926598072052,
      "kl": 0.04706335519440472,
      "learning_rate": 6.218037882920697e-07,
      "loss": 0.001882902719080448,
      "reward": 3.0398138999938964,
      "reward_std": 0.463697224855423,
      "rewards/IngredientFormatReward/mean": 0.9875,
      "rewards/IngredientFormatReward/std": 0.08298950344324112,
      "rewards/IngredientMatchReward/mean": 0.6279631614685058,
      "rewards/IngredientMatchReward/std": 0.2736996293067932,
      "rewards/IngredientQuantityMatchReward/mean": 0.6321632027626037,
      "rewards/IngredientQuantityMatchReward/std": 0.4176937580108643,
      "rewards/TotalKcalExactMatchReward/mean": 0.7921875,
      "rewards/TotalKcalExactMatchReward/std": 0.4008431792259216,
      "step": 1620
    },
    {
      "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": 0.0171875,
      "completions/max_length": 510.2,
      "completions/mean_length": 397.5578125,
      "completions/min_length": 291.6,
      "epoch": 0.4520166898470097,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6824511289596558,
      "kl": 0.04578695949167013,
      "learning_rate": 6.195719449332643e-07,
      "loss": 0.001831657439470291,
      "reward": 3.0820634365081787,
      "reward_std": 0.3959769606590271,
      "rewards/IngredientFormatReward/mean": 0.9825520873069763,
      "rewards/IngredientFormatReward/std": 0.11164550334215165,
      "rewards/IngredientMatchReward/mean": 0.6557118058204651,
      "rewards/IngredientMatchReward/std": 0.28708990216255187,
      "rewards/IngredientQuantityMatchReward/mean": 0.6781745553016663,
      "rewards/IngredientQuantityMatchReward/std": 0.4067881166934967,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4194814503192902,
      "step": 1625
    },
    {
      "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": 0.0078125,
      "completions/max_length": 502.0,
      "completions/mean_length": 394.9453125,
      "completions/min_length": 275.8,
      "epoch": 0.4534075104311544,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7145374417304993,
      "kl": 0.04789355676621199,
      "learning_rate": 6.173375717705123e-07,
      "loss": 0.001915960945188999,
      "reward": 3.124998092651367,
      "reward_std": 0.40861119627952575,
      "rewards/IngredientFormatReward/mean": 0.9889583349227905,
      "rewards/IngredientFormatReward/std": 0.08167827948927879,
      "rewards/IngredientMatchReward/mean": 0.6187487602233886,
      "rewards/IngredientMatchReward/std": 0.2515649914741516,
      "rewards/IngredientQuantityMatchReward/mean": 0.729790985584259,
      "rewards/IngredientQuantityMatchReward/std": 0.3765805721282959,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.40658249855041506,
      "step": 1630
    },
    {
      "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": 0.0046875,
      "completions/max_length": 505.6,
      "completions/mean_length": 396.221875,
      "completions/min_length": 280.8,
      "epoch": 0.45479833101529904,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6588624119758606,
      "kl": 0.04649864921811968,
      "learning_rate": 6.151007160768263e-07,
      "loss": 0.0018601721152663232,
      "reward": 3.0500646591186524,
      "reward_std": 0.4239701569080353,
      "rewards/IngredientFormatReward/mean": 0.9918303489685059,
      "rewards/IngredientFormatReward/std": 0.05691017508506775,
      "rewards/IngredientMatchReward/mean": 0.6226277112960815,
      "rewards/IngredientMatchReward/std": 0.2716664791107178,
      "rewards/IngredientQuantityMatchReward/mean": 0.6199815273284912,
      "rewards/IngredientQuantityMatchReward/std": 0.407462352514267,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.3732294499874115,
      "step": 1635
    },
    {
      "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": 0.0140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 388.575,
      "completions/min_length": 264.2,
      "epoch": 0.4561891515994437,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6433203220367432,
      "kl": 0.05448652170598507,
      "learning_rate": 6.128614251777416e-07,
      "loss": 0.0021792571991682053,
      "reward": 3.0209889888763426,
      "reward_std": 0.41428587436676023,
      "rewards/IngredientFormatReward/mean": 0.9835137724876404,
      "rewards/IngredientFormatReward/std": 0.10485587120056153,
      "rewards/IngredientMatchReward/mean": 0.6340167284011841,
      "rewards/IngredientMatchReward/std": 0.2999426305294037,
      "rewards/IngredientQuantityMatchReward/mean": 0.6300209879875183,
      "rewards/IngredientQuantityMatchReward/std": 0.44520034790039065,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.41060879826545715,
      "step": 1640
    },
    {
      "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": 0.003125,
      "completions/max_length": 500.4,
      "completions/mean_length": 390.2140625,
      "completions/min_length": 270.2,
      "epoch": 0.4575799721835883,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6303852200508118,
      "kl": 0.050624200236052276,
      "learning_rate": 6.106197464503168e-07,
      "loss": 0.0020252134650945663,
      "reward": 3.140613651275635,
      "reward_std": 0.3557996809482574,
      "rewards/IngredientFormatReward/mean": 0.9965625047683716,
      "rewards/IngredientFormatReward/std": 0.03889087215065956,
      "rewards/IngredientMatchReward/mean": 0.6494605898857116,
      "rewards/IngredientMatchReward/std": 0.29842119812965395,
      "rewards/IngredientQuantityMatchReward/mean": 0.6992781519889831,
      "rewards/IngredientQuantityMatchReward/std": 0.3938143730163574,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.3920258104801178,
      "step": 1645
    },
    {
      "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": 0.0015625,
      "completions/max_length": 492.6,
      "completions/mean_length": 383.9453125,
      "completions/min_length": 259.4,
      "epoch": 0.45897079276773295,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7204416394233704,
      "kl": 0.049180832644924524,
      "learning_rate": 6.083757273221288e-07,
      "loss": 0.001967577263712883,
      "reward": 3.166611337661743,
      "reward_std": 0.3798985183238983,
      "rewards/IngredientFormatReward/mean": 0.9952566981315613,
      "rewards/IngredientFormatReward/std": 0.048683957941830155,
      "rewards/IngredientMatchReward/mean": 0.6497941613197327,
      "rewards/IngredientMatchReward/std": 0.28912633061409,
      "rewards/IngredientQuantityMatchReward/mean": 0.6762479662895202,
      "rewards/IngredientQuantityMatchReward/std": 0.4136974036693573,
      "rewards/TotalKcalExactMatchReward/mean": 0.8453125,
      "rewards/TotalKcalExactMatchReward/std": 0.35293395519256593,
      "step": 1650
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.4,
      "completions/mean_length": 389.878125,
      "completions/min_length": 259.0,
      "epoch": 0.4603616133518776,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6586219072341919,
      "kl": 0.04915962847881019,
      "learning_rate": 6.061294152702716e-07,
      "loss": 0.0019666045904159547,
      "reward": 3.1720734596252442,
      "reward_std": 0.39295960068702696,
      "rewards/IngredientFormatReward/mean": 0.9903497099876404,
      "rewards/IngredientFormatReward/std": 0.08478823751211166,
      "rewards/IngredientMatchReward/mean": 0.6524268507957458,
      "rewards/IngredientMatchReward/std": 0.2759318143129349,
      "rewards/IngredientQuantityMatchReward/mean": 0.6714844465255737,
      "rewards/IngredientQuantityMatchReward/std": 0.429459947347641,
      "rewards/TotalKcalExactMatchReward/mean": 0.8578125,
      "rewards/TotalKcalExactMatchReward/std": 0.3274064376950264,
      "step": 1655
    },
    {
      "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": 0.0046875,
      "completions/max_length": 504.6,
      "completions/mean_length": 389.371875,
      "completions/min_length": 278.0,
      "epoch": 0.4617524339360223,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6691665649414062,
      "kl": 0.04552193239796907,
      "learning_rate": 6.03880857820351e-07,
      "loss": 0.0018210854381322862,
      "reward": 3.1165382862091064,
      "reward_std": 0.39113470911979675,
      "rewards/IngredientFormatReward/mean": 0.9834821343421936,
      "rewards/IngredientFormatReward/std": 0.07514195144176483,
      "rewards/IngredientMatchReward/mean": 0.6446856260299683,
      "rewards/IngredientMatchReward/std": 0.2858420073986053,
      "rewards/IngredientQuantityMatchReward/mean": 0.6758704543113708,
      "rewards/IngredientQuantityMatchReward/std": 0.39811512231826784,
      "rewards/TotalKcalExactMatchReward/mean": 0.8125,
      "rewards/TotalKcalExactMatchReward/std": 0.3781652688980103,
      "step": 1660
    },
    {
      "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": 0.00625,
      "completions/max_length": 510.0,
      "completions/mean_length": 390.2328125,
      "completions/min_length": 271.8,
      "epoch": 0.4631432545201669,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7427563071250916,
      "kl": 0.047903071716427804,
      "learning_rate": 6.016301025454787e-07,
      "loss": 0.0019164852797985076,
      "reward": 3.088708591461182,
      "reward_std": 0.36983243227005,
      "rewards/IngredientFormatReward/mean": 0.9919270753860474,
      "rewards/IngredientFormatReward/std": 0.08860928863286972,
      "rewards/IngredientMatchReward/mean": 0.6738368034362793,
      "rewards/IngredientMatchReward/std": 0.2914969980716705,
      "rewards/IngredientQuantityMatchReward/mean": 0.633882200717926,
      "rewards/IngredientQuantityMatchReward/std": 0.4225495457649231,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.39090106785297396,
      "step": 1665
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.2,
      "completions/mean_length": 394.36875,
      "completions/min_length": 286.6,
      "epoch": 0.46453407510431155,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6977032423019409,
      "kl": 0.048976365476846695,
      "learning_rate": 5.99377197065266e-07,
      "loss": 0.001959040015935898,
      "reward": 3.0245996475219727,
      "reward_std": 0.39999868273735045,
      "rewards/IngredientFormatReward/mean": 0.991223955154419,
      "rewards/IngredientFormatReward/std": 0.06987991854548455,
      "rewards/IngredientMatchReward/mean": 0.6229538559913635,
      "rewards/IngredientMatchReward/std": 0.29490872621536257,
      "rewards/IngredientQuantityMatchReward/mean": 0.6401092886924744,
      "rewards/IngredientQuantityMatchReward/std": 0.42952102422714233,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.38621187806129453,
      "step": 1670
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.6,
      "completions/mean_length": 394.971875,
      "completions/min_length": 265.2,
      "epoch": 0.4659248956884562,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7027736306190491,
      "kl": 0.05804067454300821,
      "learning_rate": 5.971221890448175e-07,
      "loss": 0.002321036159992218,
      "reward": 3.1033143043518066,
      "reward_std": 0.39811227917671205,
      "rewards/IngredientFormatReward/mean": 0.9859375,
      "rewards/IngredientFormatReward/std": 0.10241568833589554,
      "rewards/IngredientMatchReward/mean": 0.6516257405281067,
      "rewards/IngredientMatchReward/std": 0.2847370356321335,
      "rewards/IngredientQuantityMatchReward/mean": 0.6501261115074157,
      "rewards/IngredientQuantityMatchReward/std": 0.4035037100315094,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.383716356754303,
      "step": 1675
    },
    {
      "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": 0.0078125,
      "completions/max_length": 505.8,
      "completions/mean_length": 394.65,
      "completions/min_length": 272.0,
      "epoch": 0.4673157162726008,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6062320470809937,
      "kl": 0.04617527783848345,
      "learning_rate": 5.948651261937202e-07,
      "loss": 0.0018472731113433837,
      "reward": 3.1725085735321046,
      "reward_std": 0.3940969526767731,
      "rewards/IngredientFormatReward/mean": 0.9842708349227905,
      "rewards/IngredientFormatReward/std": 0.09453187361359597,
      "rewards/IngredientMatchReward/mean": 0.6568180322647095,
      "rewards/IngredientMatchReward/std": 0.2938332736492157,
      "rewards/IngredientQuantityMatchReward/mean": 0.6704821944236755,
      "rewards/IngredientQuantityMatchReward/std": 0.41570605635643004,
      "rewards/TotalKcalExactMatchReward/mean": 0.8609375,
      "rewards/TotalKcalExactMatchReward/std": 0.3382291287183762,
      "step": 1680
    },
    {
      "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": 0.0125,
      "completions/max_length": 507.0,
      "completions/mean_length": 387.928125,
      "completions/min_length": 264.2,
      "epoch": 0.46870653685674546,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7044047117233276,
      "kl": 0.048976872209459545,
      "learning_rate": 5.926060562650365e-07,
      "loss": 0.001959339529275894,
      "reward": 2.9777952671051025,
      "reward_std": 0.4341627299785614,
      "rewards/IngredientFormatReward/mean": 0.983802080154419,
      "rewards/IngredientFormatReward/std": 0.10478676557540893,
      "rewards/IngredientMatchReward/mean": 0.6123015999794006,
      "rewards/IngredientMatchReward/std": 0.2814868152141571,
      "rewards/IngredientQuantityMatchReward/mean": 0.627004086971283,
      "rewards/IngredientQuantityMatchReward/std": 0.41599141955375674,
      "rewards/TotalKcalExactMatchReward/mean": 0.7546875,
      "rewards/TotalKcalExactMatchReward/std": 0.4197186291217804,
      "step": 1685
    },
    {
      "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": 0.0140625,
      "completions/max_length": 502.6,
      "completions/mean_length": 389.990625,
      "completions/min_length": 255.4,
      "epoch": 0.47009735744089015,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.644118070602417,
      "kl": 0.04798829292412847,
      "learning_rate": 5.903450270542924e-07,
      "loss": 0.0019195247441530228,
      "reward": 3.198506736755371,
      "reward_std": 0.41594612002372744,
      "rewards/IngredientFormatReward/mean": 0.986517870426178,
      "rewards/IngredientFormatReward/std": 0.09814060274511575,
      "rewards/IngredientMatchReward/mean": 0.6614857912063599,
      "rewards/IngredientMatchReward/std": 0.2818374246358871,
      "rewards/IngredientQuantityMatchReward/mean": 0.6973781704902648,
      "rewards/IngredientQuantityMatchReward/std": 0.40497069954872134,
      "rewards/TotalKcalExactMatchReward/mean": 0.853125,
      "rewards/TotalKcalExactMatchReward/std": 0.34283066987991334,
      "step": 1690
    },
    {
      "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": 0.0140625,
      "completions/max_length": 512.2,
      "completions/mean_length": 394.8625,
      "completions/min_length": 287.2,
      "epoch": 0.4714881780250348,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7088591456413269,
      "kl": 0.04952654310036451,
      "learning_rate": 5.880820863984671e-07,
      "loss": 0.001980846002697945,
      "reward": 2.948525238037109,
      "reward_std": 0.4502517580986023,
      "rewards/IngredientFormatReward/mean": 0.9842187523841858,
      "rewards/IngredientFormatReward/std": 0.08936193883419037,
      "rewards/IngredientMatchReward/mean": 0.5951304972171784,
      "rewards/IngredientMatchReward/std": 0.28045451641082764,
      "rewards/IngredientQuantityMatchReward/mean": 0.5848009943962097,
      "rewards/IngredientQuantityMatchReward/std": 0.43246378302574157,
      "rewards/TotalKcalExactMatchReward/mean": 0.784375,
      "rewards/TotalKcalExactMatchReward/std": 0.39880833625793455,
      "step": 1695
    },
    {
      "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": 0.0046875,
      "completions/max_length": 499.2,
      "completions/mean_length": 384.971875,
      "completions/min_length": 279.4,
      "epoch": 0.4728789986091794,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.6974106431007385,
      "kl": 0.048825663654133676,
      "learning_rate": 5.858172821749804e-07,
      "loss": 0.0019530490040779113,
      "reward": 3.026969051361084,
      "reward_std": 0.3304957687854767,
      "rewards/IngredientFormatReward/mean": 0.9939843773841858,
      "rewards/IngredientFormatReward/std": 0.03864100463688373,
      "rewards/IngredientMatchReward/mean": 0.5842813670635223,
      "rewards/IngredientMatchReward/std": 0.2920273244380951,
      "rewards/IngredientQuantityMatchReward/mean": 0.6471408128738403,
      "rewards/IngredientQuantityMatchReward/std": 0.41217195987701416,
      "rewards/TotalKcalExactMatchReward/mean": 0.8015625,
      "rewards/TotalKcalExactMatchReward/std": 0.3969415783882141,
      "step": 1700
    },
    {
      "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": 0.0078125,
      "completions/max_length": 502.8,
      "completions/mean_length": 386.540625,
      "completions/min_length": 265.0,
      "epoch": 0.47426981919332406,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6712289452552795,
      "kl": 0.05012807142920792,
      "learning_rate": 5.835506623006797e-07,
      "loss": 0.002004893682897091,
      "reward": 3.0048501968383787,
      "reward_std": 0.4382961392402649,
      "rewards/IngredientFormatReward/mean": 0.9930059671401977,
      "rewards/IngredientFormatReward/std": 0.051491785980761054,
      "rewards/IngredientMatchReward/mean": 0.6337772965431213,
      "rewards/IngredientMatchReward/std": 0.305075603723526,
      "rewards/IngredientQuantityMatchReward/mean": 0.6108795642852783,
      "rewards/IngredientQuantityMatchReward/std": 0.43121986389160155,
      "rewards/TotalKcalExactMatchReward/mean": 0.7671875,
      "rewards/TotalKcalExactMatchReward/std": 0.4157248854637146,
      "step": 1705
    },
    {
      "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": 0.0125,
      "completions/max_length": 499.2,
      "completions/mean_length": 387.99375,
      "completions/min_length": 263.2,
      "epoch": 0.4756606397774687,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6293528079986572,
      "kl": 0.048464122042059896,
      "learning_rate": 5.812822747308267e-07,
      "loss": 0.0019388705492019652,
      "reward": 3.175932788848877,
      "reward_std": 0.3702836275100708,
      "rewards/IngredientFormatReward/mean": 0.9876116037368774,
      "rewards/IngredientFormatReward/std": 0.07741203494369983,
      "rewards/IngredientMatchReward/mean": 0.659701144695282,
      "rewards/IngredientMatchReward/std": 0.29684895277023315,
      "rewards/IngredientQuantityMatchReward/mean": 0.7348701000213623,
      "rewards/IngredientQuantityMatchReward/std": 0.37621399760246277,
      "rewards/TotalKcalExactMatchReward/mean": 0.79375,
      "rewards/TotalKcalExactMatchReward/std": 0.397750324010849,
      "step": 1710
    },
    {
      "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": 0.0078125,
      "completions/max_length": 504.0,
      "completions/mean_length": 389.025,
      "completions/min_length": 238.4,
      "epoch": 0.47705146036161333,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6771330833435059,
      "kl": 0.049341540411114694,
      "learning_rate": 5.790121674580825e-07,
      "loss": 0.0019737355411052703,
      "reward": 2.981400489807129,
      "reward_std": 0.3796704918146133,
      "rewards/IngredientFormatReward/mean": 0.9787500023841857,
      "rewards/IngredientFormatReward/std": 0.10464536398649216,
      "rewards/IngredientMatchReward/mean": 0.6600198388099671,
      "rewards/IngredientMatchReward/std": 0.2891321390867233,
      "rewards/IngredientQuantityMatchReward/mean": 0.5941931962966919,
      "rewards/IngredientQuantityMatchReward/std": 0.40178399682044985,
      "rewards/TotalKcalExactMatchReward/mean": 0.7484375,
      "rewards/TotalKcalExactMatchReward/std": 0.4310473442077637,
      "step": 1715
    },
    {
      "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": 0.003125,
      "completions/max_length": 499.2,
      "completions/mean_length": 387.9921875,
      "completions/min_length": 253.6,
      "epoch": 0.47844228094575797,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6627811789512634,
      "kl": 0.04793977346271276,
      "learning_rate": 5.767403885114922e-07,
      "loss": 0.0019301587715744972,
      "reward": 3.085385274887085,
      "reward_std": 0.36830160617828367,
      "rewards/IngredientFormatReward/mean": 0.9941145777702332,
      "rewards/IngredientFormatReward/std": 0.04807176664471626,
      "rewards/IngredientMatchReward/mean": 0.6792026281356811,
      "rewards/IngredientMatchReward/std": 0.2884396731853485,
      "rewards/IngredientQuantityMatchReward/mean": 0.6370681166648865,
      "rewards/IngredientQuantityMatchReward/std": 0.4236335575580597,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.40007728040218354,
      "step": 1720
    },
    {
      "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": 0.0109375,
      "completions/max_length": 512.6,
      "completions/mean_length": 395.703125,
      "completions/min_length": 256.8,
      "epoch": 0.47983310152990266,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6980003118515015,
      "kl": 0.051507193874567744,
      "learning_rate": 5.744669859554688e-07,
      "loss": 0.0020601864904165267,
      "reward": 2.9863351345062257,
      "reward_std": 0.4324962854385376,
      "rewards/IngredientFormatReward/mean": 0.984587037563324,
      "rewards/IngredientFormatReward/std": 0.0999862365424633,
      "rewards/IngredientMatchReward/mean": 0.6219358801841736,
      "rewards/IngredientMatchReward/std": 0.29111440777778624,
      "rewards/IngredientQuantityMatchReward/mean": 0.643874716758728,
      "rewards/IngredientQuantityMatchReward/std": 0.41615907549858094,
      "rewards/TotalKcalExactMatchReward/mean": 0.7359375,
      "rewards/TotalKcalExactMatchReward/std": 0.43581517934799197,
      "step": 1725
    },
    {
      "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": 0.0140625,
      "completions/max_length": 507.4,
      "completions/mean_length": 393.6796875,
      "completions/min_length": 257.6,
      "epoch": 0.4812239221140473,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.652874231338501,
      "kl": 0.04963922812603414,
      "learning_rate": 5.721920078887763e-07,
      "loss": 0.0019854392856359484,
      "reward": 3.042120361328125,
      "reward_std": 0.3817890167236328,
      "rewards/IngredientFormatReward/mean": 0.9817125558853149,
      "rewards/IngredientFormatReward/std": 0.11481560580432415,
      "rewards/IngredientMatchReward/mean": 0.6558786153793335,
      "rewards/IngredientMatchReward/std": 0.2739634424448013,
      "rewards/IngredientQuantityMatchReward/mean": 0.6482792258262634,
      "rewards/IngredientQuantityMatchReward/std": 0.4263168752193451,
      "rewards/TotalKcalExactMatchReward/mean": 0.75625,
      "rewards/TotalKcalExactMatchReward/std": 0.41720872521400454,
      "step": 1730
    },
    {
      "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": 0.0046875,
      "completions/max_length": 503.8,
      "completions/mean_length": 395.4546875,
      "completions/min_length": 266.4,
      "epoch": 0.48261474269819193,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6124057769775391,
      "kl": 0.045404232246801256,
      "learning_rate": 5.699155024435123e-07,
      "loss": 0.0018159780651330949,
      "reward": 3.0864189147949217,
      "reward_std": 0.34279492497444153,
      "rewards/IngredientFormatReward/mean": 0.9932775378227234,
      "rewards/IngredientFormatReward/std": 0.05568651407957077,
      "rewards/IngredientMatchReward/mean": 0.6521255016326905,
      "rewards/IngredientMatchReward/std": 0.2828336536884308,
      "rewards/IngredientQuantityMatchReward/mean": 0.622265899181366,
      "rewards/IngredientQuantityMatchReward/std": 0.4201075792312622,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.38487215638160704,
      "step": 1735
    },
    {
      "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": 0.0109375,
      "completions/max_length": 512.8,
      "completions/mean_length": 394.9109375,
      "completions/min_length": 261.4,
      "epoch": 0.48400556328233657,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6815362572669983,
      "kl": 0.04877625331282616,
      "learning_rate": 5.676375177840886e-07,
      "loss": 0.0019514326006174088,
      "reward": 3.0187916278839113,
      "reward_std": 0.42291984558105467,
      "rewards/IngredientFormatReward/mean": 0.9903645753860474,
      "rewards/IngredientFormatReward/std": 0.08810437638312578,
      "rewards/IngredientMatchReward/mean": 0.6148133635520935,
      "rewards/IngredientMatchReward/std": 0.2883909821510315,
      "rewards/IngredientQuantityMatchReward/mean": 0.6151761889457703,
      "rewards/IngredientQuantityMatchReward/std": 0.4394866168498993,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.39994245767593384,
      "step": 1740
    },
    {
      "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": 0.01875,
      "completions/max_length": 501.6,
      "completions/mean_length": 390.396875,
      "completions/min_length": 268.4,
      "epoch": 0.4853963838664812,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6231136918067932,
      "kl": 0.047634016210213306,
      "learning_rate": 5.653581021062138e-07,
      "loss": 0.0019053995609283447,
      "reward": 3.1004568576812743,
      "reward_std": 0.4429864943027496,
      "rewards/IngredientFormatReward/mean": 0.9809375047683716,
      "rewards/IngredientFormatReward/std": 0.08808905854821206,
      "rewards/IngredientMatchReward/mean": 0.6266797065734864,
      "rewards/IngredientMatchReward/std": 0.27853635549545286,
      "rewards/IngredientQuantityMatchReward/mean": 0.6850271940231323,
      "rewards/IngredientQuantityMatchReward/std": 0.41117426156997683,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.3709612190723419,
      "step": 1745
    },
    {
      "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": 0.03125,
      "completions/max_length": 512.8,
      "completions/mean_length": 400.534375,
      "completions/min_length": 275.2,
      "epoch": 0.48678720445062584,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6302758455276489,
      "kl": 0.055369544448331,
      "learning_rate": 5.630773036358727e-07,
      "loss": 0.0022154580801725388,
      "reward": 3.1487196922302245,
      "reward_std": 0.43740108609199524,
      "rewards/IngredientFormatReward/mean": 0.9723288774490356,
      "rewards/IngredientFormatReward/std": 0.13584012035280466,
      "rewards/IngredientMatchReward/mean": 0.6218018412590027,
      "rewards/IngredientMatchReward/std": 0.29499236941337587,
      "rewards/IngredientQuantityMatchReward/mean": 0.715526533126831,
      "rewards/IngredientQuantityMatchReward/std": 0.3941978871822357,
      "rewards/TotalKcalExactMatchReward/mean": 0.8390625,
      "rewards/TotalKcalExactMatchReward/std": 0.3547331839799881,
      "step": 1750
    },
    {
      "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": 0.0046875,
      "completions/max_length": 502.8,
      "completions/mean_length": 393.821875,
      "completions/min_length": 278.6,
      "epoch": 0.48817802503477054,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.655293345451355,
      "kl": 0.046050224895589055,
      "learning_rate": 5.607951706283055e-07,
      "loss": 0.0018420252948999406,
      "reward": 3.1131840705871583,
      "reward_std": 0.39082067608833315,
      "rewards/IngredientFormatReward/mean": 0.9950520753860473,
      "rewards/IngredientFormatReward/std": 0.04335647970438004,
      "rewards/IngredientMatchReward/mean": 0.6560224533081055,
      "rewards/IngredientMatchReward/std": 0.2750636339187622,
      "rewards/IngredientQuantityMatchReward/mean": 0.6605471014976502,
      "rewards/IngredientQuantityMatchReward/std": 0.4154274582862854,
      "rewards/TotalKcalExactMatchReward/mean": 0.8015625,
      "rewards/TotalKcalExactMatchReward/std": 0.39623318910598754,
      "step": 1755
    },
    {
      "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": 0.0203125,
      "completions/max_length": 508.0,
      "completions/mean_length": 389.0453125,
      "completions/min_length": 257.4,
      "epoch": 0.48956884561891517,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7058917880058289,
      "kl": 0.05053422078490257,
      "learning_rate": 5.585117513669882e-07,
      "loss": 0.002021782286465168,
      "reward": 3.0528096675872805,
      "reward_std": 0.46687796115875246,
      "rewards/IngredientFormatReward/mean": 0.9771986722946167,
      "rewards/IngredientFormatReward/std": 0.12420394718647003,
      "rewards/IngredientMatchReward/mean": 0.6485894083976745,
      "rewards/IngredientMatchReward/std": 0.29609904885292054,
      "rewards/IngredientQuantityMatchReward/mean": 0.6395214915275573,
      "rewards/IngredientQuantityMatchReward/std": 0.4058083415031433,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.3928277313709259,
      "step": 1760
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.6,
      "completions/mean_length": 401.221875,
      "completions/min_length": 272.6,
      "epoch": 0.4909596662030598,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.659774899482727,
      "kl": 0.04758750724140555,
      "learning_rate": 5.562270941626098e-07,
      "loss": 0.001903420127928257,
      "reward": 3.078034543991089,
      "reward_std": 0.4054509222507477,
      "rewards/IngredientFormatReward/mean": 0.9913541555404664,
      "rewards/IngredientFormatReward/std": 0.06280860994011164,
      "rewards/IngredientMatchReward/mean": 0.6158724069595337,
      "rewards/IngredientMatchReward/std": 0.273968768119812,
      "rewards/IngredientQuantityMatchReward/mean": 0.6911204814910888,
      "rewards/IngredientQuantityMatchReward/std": 0.3966470003128052,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.4132148861885071,
      "step": 1765
    },
    {
      "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": 0.0125,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.8234375,
      "completions/min_length": 256.2,
      "epoch": 0.49235048678720444,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 4.437183380126953,
      "kl": 0.06853532809764147,
      "learning_rate": 5.539412473520507e-07,
      "loss": 0.0027404358610510827,
      "reward": 3.176833152770996,
      "reward_std": 0.3862785041332245,
      "rewards/IngredientFormatReward/mean": 0.9880208253860474,
      "rewards/IngredientFormatReward/std": 0.1034691423177719,
      "rewards/IngredientMatchReward/mean": 0.6355412840843201,
      "rewards/IngredientMatchReward/std": 0.2928665101528168,
      "rewards/IngredientQuantityMatchReward/mean": 0.6907709836959839,
      "rewards/IngredientQuantityMatchReward/std": 0.3815356969833374,
      "rewards/TotalKcalExactMatchReward/mean": 0.8625,
      "rewards/TotalKcalExactMatchReward/std": 0.3310780256986618,
      "step": 1770
    },
    {
      "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": 0.0203125,
      "completions/max_length": 505.8,
      "completions/mean_length": 399.0546875,
      "completions/min_length": 283.2,
      "epoch": 0.4937413073713491,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6291945576667786,
      "kl": 0.047819344280287626,
      "learning_rate": 5.516542592973603e-07,
      "loss": 0.0019129764288663865,
      "reward": 2.977151393890381,
      "reward_std": 0.47986037731170655,
      "rewards/IngredientFormatReward/mean": 0.9804017782211304,
      "rewards/IngredientFormatReward/std": 0.11696080565452575,
      "rewards/IngredientMatchReward/mean": 0.6131088972091675,
      "rewards/IngredientMatchReward/std": 0.3076077878475189,
      "rewards/IngredientQuantityMatchReward/mean": 0.635203194618225,
      "rewards/IngredientQuantityMatchReward/std": 0.4294419288635254,
      "rewards/TotalKcalExactMatchReward/mean": 0.7484375,
      "rewards/TotalKcalExactMatchReward/std": 0.41134634613990784,
      "step": 1775
    },
    {
      "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": 0.01875,
      "completions/max_length": 511.8,
      "completions/mean_length": 394.38125,
      "completions/min_length": 256.8,
      "epoch": 0.4951321279554937,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6367654204368591,
      "kl": 0.04918671539053321,
      "learning_rate": 5.49366178384733e-07,
      "loss": 0.0019677009433507918,
      "reward": 3.022269535064697,
      "reward_std": 0.42463927865028384,
      "rewards/IngredientFormatReward/mean": 0.9825520753860474,
      "rewards/IngredientFormatReward/std": 0.11480289716273546,
      "rewards/IngredientMatchReward/mean": 0.6272984862327575,
      "rewards/IngredientMatchReward/std": 0.3080913841724396,
      "rewards/IngredientQuantityMatchReward/mean": 0.6374189138412476,
      "rewards/IngredientQuantityMatchReward/std": 0.44004141092300414,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.4037832021713257,
      "step": 1780
    },
    {
      "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": 0.00625,
      "completions/max_length": 511.2,
      "completions/mean_length": 395.1953125,
      "completions/min_length": 271.6,
      "epoch": 0.4965229485396384,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6336233615875244,
      "kl": 0.05262123909778893,
      "learning_rate": 5.470770530234855e-07,
      "loss": 0.002105065435171127,
      "reward": 3.0902578353881838,
      "reward_std": 0.3941184222698212,
      "rewards/IngredientFormatReward/mean": 0.9919345259666443,
      "rewards/IngredientFormatReward/std": 0.0732752576470375,
      "rewards/IngredientMatchReward/mean": 0.644134521484375,
      "rewards/IngredientMatchReward/std": 0.282639217376709,
      "rewards/IngredientQuantityMatchReward/mean": 0.6432512640953064,
      "rewards/IngredientQuantityMatchReward/std": 0.41275890469551085,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.3679554224014282,
      "step": 1785
    },
    {
      "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": 0.009375,
      "completions/max_length": 505.0,
      "completions/mean_length": 400.9296875,
      "completions/min_length": 299.4,
      "epoch": 0.49791376912378305,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6558573842048645,
      "kl": 0.04567183582112193,
      "learning_rate": 5.447869316450318e-07,
      "loss": 0.0018267896026372909,
      "reward": 3.1581130981445313,
      "reward_std": 0.3598249554634094,
      "rewards/IngredientFormatReward/mean": 0.9941666841506958,
      "rewards/IngredientFormatReward/std": 0.04706086441874504,
      "rewards/IngredientMatchReward/mean": 0.6860336184501648,
      "rewards/IngredientMatchReward/std": 0.2729302167892456,
      "rewards/IngredientQuantityMatchReward/mean": 0.6732252895832062,
      "rewards/IngredientQuantityMatchReward/std": 0.3966924250125885,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.3842876791954041,
      "step": 1790
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.6,
      "completions/mean_length": 396.1359375,
      "completions/min_length": 287.4,
      "epoch": 0.4993045897079277,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.5723157525062561,
      "kl": 0.04732802438084036,
      "learning_rate": 5.424958627018586e-07,
      "loss": 0.001893388107419014,
      "reward": 3.118370294570923,
      "reward_std": 0.37871933579444883,
      "rewards/IngredientFormatReward/mean": 0.98984375,
      "rewards/IngredientFormatReward/std": 0.08600255399942398,
      "rewards/IngredientMatchReward/mean": 0.681352949142456,
      "rewards/IngredientMatchReward/std": 0.27151423394680024,
      "rewards/IngredientQuantityMatchReward/mean": 0.6377986788749694,
      "rewards/IngredientQuantityMatchReward/std": 0.4133506715297699,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.3859631597995758,
      "step": 1795
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.8,
      "completions/mean_length": 392.56875,
      "completions/min_length": 273.8,
      "epoch": 0.5006954102920723,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6640133261680603,
      "kl": 0.05193311013281345,
      "learning_rate": 5.402038946665011e-07,
      "loss": 0.0020771391689777376,
      "reward": 3.1144729137420653,
      "reward_std": 0.4295255720615387,
      "rewards/IngredientFormatReward/mean": 0.9897916555404663,
      "rewards/IngredientFormatReward/std": 0.07800195086747408,
      "rewards/IngredientMatchReward/mean": 0.6456776976585388,
      "rewards/IngredientMatchReward/std": 0.30382000207901,
      "rewards/IngredientQuantityMatchReward/mean": 0.6586910247802734,
      "rewards/IngredientQuantityMatchReward/std": 0.4264285624027252,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.3754447340965271,
      "step": 1800
    },
    {
      "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": 0.0203125,
      "completions/max_length": 510.8,
      "completions/mean_length": 396.7265625,
      "completions/min_length": 285.0,
      "epoch": 0.502086230876217,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.612971305847168,
      "kl": 0.052038407209329306,
      "learning_rate": 5.37911076030516e-07,
      "loss": 0.0020814632996916773,
      "reward": 2.9839890956878663,
      "reward_std": 0.4456018149852753,
      "rewards/IngredientFormatReward/mean": 0.9828125,
      "rewards/IngredientFormatReward/std": 0.10817223787307739,
      "rewards/IngredientMatchReward/mean": 0.5959939360618591,
      "rewards/IngredientMatchReward/std": 0.3077159345149994,
      "rewards/IngredientQuantityMatchReward/mean": 0.6067451357841491,
      "rewards/IngredientQuantityMatchReward/std": 0.41565820574760437,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.38725520074367525,
      "step": 1805
    },
    {
      "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": 0.009375,
      "completions/max_length": 506.2,
      "completions/mean_length": 395.128125,
      "completions/min_length": 252.2,
      "epoch": 0.5034770514603616,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6170816421508789,
      "kl": 0.04653078094124794,
      "learning_rate": 5.356174553034564e-07,
      "loss": 0.0018611878156661987,
      "reward": 3.077640676498413,
      "reward_std": 0.41207525730133054,
      "rewards/IngredientFormatReward/mean": 0.9891145944595336,
      "rewards/IngredientFormatReward/std": 0.08445280343294144,
      "rewards/IngredientMatchReward/mean": 0.6564167976379395,
      "rewards/IngredientMatchReward/std": 0.2918438494205475,
      "rewards/IngredientQuantityMatchReward/mean": 0.658671748638153,
      "rewards/IngredientQuantityMatchReward/std": 0.4138647198677063,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.4117767035961151,
      "step": 1810
    },
    {
      "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": 0.0171875,
      "completions/max_length": 512.0,
      "completions/mean_length": 400.0734375,
      "completions/min_length": 255.0,
      "epoch": 0.5048678720445062,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6442963480949402,
      "kl": 0.056444175005890426,
      "learning_rate": 5.333230810118461e-07,
      "loss": 0.0022573621943593027,
      "reward": 3.116063928604126,
      "reward_std": 0.4383660852909088,
      "rewards/IngredientFormatReward/mean": 0.9827827334403991,
      "rewards/IngredientFormatReward/std": 0.11086281388998032,
      "rewards/IngredientMatchReward/mean": 0.6368998050689697,
      "rewards/IngredientMatchReward/std": 0.2972823202610016,
      "rewards/IngredientQuantityMatchReward/mean": 0.6682563424110413,
      "rewards/IngredientQuantityMatchReward/std": 0.3951943576335907,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.37783130407333376,
      "step": 1815
    },
    {
      "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": 0.00625,
      "completions/max_length": 505.4,
      "completions/mean_length": 387.6359375,
      "completions/min_length": 251.8,
      "epoch": 0.5062586926286509,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6564891934394836,
      "kl": 0.05077703422866762,
      "learning_rate": 5.310280016981512e-07,
      "loss": 0.002030998095870018,
      "reward": 3.0581398963928224,
      "reward_std": 0.4139727771282196,
      "rewards/IngredientFormatReward/mean": 0.9908854246139527,
      "rewards/IngredientFormatReward/std": 0.07461793459951878,
      "rewards/IngredientMatchReward/mean": 0.6378273844718934,
      "rewards/IngredientMatchReward/std": 0.28663045167922974,
      "rewards/IngredientQuantityMatchReward/mean": 0.6044270992279053,
      "rewards/IngredientQuantityMatchReward/std": 0.42292520403862,
      "rewards/TotalKcalExactMatchReward/mean": 0.825,
      "rewards/TotalKcalExactMatchReward/std": 0.37856464385986327,
      "step": 1820
    },
    {
      "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": 0.021875,
      "completions/max_length": 512.6,
      "completions/mean_length": 394.4625,
      "completions/min_length": 240.6,
      "epoch": 0.5076495132127955,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 8.344937324523926,
      "kl": 0.08002238939516246,
      "learning_rate": 5.287322659197547e-07,
      "loss": 0.0032052285969257355,
      "reward": 2.9394122123718263,
      "reward_std": 0.4769141972064972,
      "rewards/IngredientFormatReward/mean": 0.9648660778999328,
      "rewards/IngredientFormatReward/std": 0.17725641131401063,
      "rewards/IngredientMatchReward/mean": 0.6225856423377991,
      "rewards/IngredientMatchReward/std": 0.29008139967918395,
      "rewards/IngredientQuantityMatchReward/mean": 0.6519605159759522,
      "rewards/IngredientQuantityMatchReward/std": 0.42122418284416197,
      "rewards/TotalKcalExactMatchReward/mean": 0.7,
      "rewards/TotalKcalExactMatchReward/std": 0.44271482825279235,
      "step": 1825
    },
    {
      "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": 0.01875,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.7015625,
      "completions/min_length": 277.0,
      "epoch": 0.5090403337969402,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6441317796707153,
      "kl": 0.05329477796331048,
      "learning_rate": 5.264359222479284e-07,
      "loss": 0.0021320804953575134,
      "reward": 2.9984073638916016,
      "reward_std": 0.3900154232978821,
      "rewards/IngredientFormatReward/mean": 0.9801413774490356,
      "rewards/IngredientFormatReward/std": 0.13333013355731965,
      "rewards/IngredientMatchReward/mean": 0.6038113951683044,
      "rewards/IngredientMatchReward/std": 0.2904983043670654,
      "rewards/IngredientQuantityMatchReward/mean": 0.5894545555114746,
      "rewards/IngredientQuantityMatchReward/std": 0.4164754033088684,
      "rewards/TotalKcalExactMatchReward/mean": 0.825,
      "rewards/TotalKcalExactMatchReward/std": 0.3774557411670685,
      "step": 1830
    },
    {
      "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": 0.0078125,
      "completions/max_length": 506.0,
      "completions/mean_length": 394.346875,
      "completions/min_length": 231.2,
      "epoch": 0.5104311543810849,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.571604311466217,
      "kl": 0.048023554123938085,
      "learning_rate": 5.241390192668053e-07,
      "loss": 0.0019212931394577027,
      "reward": 3.1547540187835694,
      "reward_std": 0.41128666400909425,
      "rewards/IngredientFormatReward/mean": 0.9815755128860474,
      "rewards/IngredientFormatReward/std": 0.1034016104415059,
      "rewards/IngredientMatchReward/mean": 0.662100088596344,
      "rewards/IngredientMatchReward/std": 0.2865631103515625,
      "rewards/IngredientQuantityMatchReward/mean": 0.7142034649848938,
      "rewards/IngredientQuantityMatchReward/std": 0.3982077479362488,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.39826048016548155,
      "step": 1835
    },
    {
      "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": 0.040625,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.8265625,
      "completions/min_length": 278.4,
      "epoch": 0.5118219749652295,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6243442893028259,
      "kl": 0.049475864064879714,
      "learning_rate": 5.218416055723517e-07,
      "loss": 0.0019794130697846414,
      "reward": 3.047327184677124,
      "reward_std": 0.42559523582458497,
      "rewards/IngredientFormatReward/mean": 0.9601413607597351,
      "rewards/IngredientFormatReward/std": 0.18492761999368668,
      "rewards/IngredientMatchReward/mean": 0.6405964970588685,
      "rewards/IngredientMatchReward/std": 0.2941777765750885,
      "rewards/IngredientQuantityMatchReward/mean": 0.6387767672538758,
      "rewards/IngredientQuantityMatchReward/std": 0.4147272765636444,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.3916107892990112,
      "step": 1840
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.4,
      "completions/mean_length": 398.18125,
      "completions/min_length": 289.2,
      "epoch": 0.5132127955493742,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6171333193778992,
      "kl": 0.048615764314308765,
      "learning_rate": 5.195437297713396e-07,
      "loss": 0.001944425329566002,
      "reward": 3.102466917037964,
      "reward_std": 0.4576135456562042,
      "rewards/IngredientFormatReward/mean": 0.9871726155281066,
      "rewards/IngredientFormatReward/std": 0.08229567054659129,
      "rewards/IngredientMatchReward/mean": 0.6418576717376709,
      "rewards/IngredientMatchReward/std": 0.2877155005931854,
      "rewards/IngredientQuantityMatchReward/mean": 0.61249920129776,
      "rewards/IngredientQuantityMatchReward/std": 0.41496188640594484,
      "rewards/TotalKcalExactMatchReward/mean": 0.8609375,
      "rewards/TotalKcalExactMatchReward/std": 0.3430387437343597,
      "step": 1845
    },
    {
      "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": 0.0015625,
      "completions/max_length": 500.4,
      "completions/mean_length": 393.8796875,
      "completions/min_length": 265.2,
      "epoch": 0.5146036161335188,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5736840963363647,
      "kl": 0.04830085132271052,
      "learning_rate": 5.172454404803176e-07,
      "loss": 0.0019320830702781676,
      "reward": 3.0672821044921874,
      "reward_std": 0.4006498992443085,
      "rewards/IngredientFormatReward/mean": 0.9946875095367431,
      "rewards/IngredientFormatReward/std": 0.04643735066056252,
      "rewards/IngredientMatchReward/mean": 0.635732901096344,
      "rewards/IngredientMatchReward/std": 0.29260424375534055,
      "rewards/IngredientQuantityMatchReward/mean": 0.6556117177009583,
      "rewards/IngredientQuantityMatchReward/std": 0.4205364763736725,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.4017998993396759,
      "step": 1850
    },
    {
      "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": 0.0203125,
      "completions/max_length": 509.4,
      "completions/mean_length": 396.253125,
      "completions/min_length": 264.8,
      "epoch": 0.5159944367176634,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6130569577217102,
      "kl": 0.049074625875800845,
      "learning_rate": 5.149467863245823e-07,
      "loss": 0.0019630441442131997,
      "reward": 3.0270339012145997,
      "reward_std": 0.4051498770713806,
      "rewards/IngredientFormatReward/mean": 0.9796875,
      "rewards/IngredientFormatReward/std": 0.08117598444223403,
      "rewards/IngredientMatchReward/mean": 0.627546489238739,
      "rewards/IngredientMatchReward/std": 0.27951192259788515,
      "rewards/IngredientQuantityMatchReward/mean": 0.6651124238967896,
      "rewards/IngredientQuantityMatchReward/std": 0.41736636161804197,
      "rewards/TotalKcalExactMatchReward/mean": 0.7546875,
      "rewards/TotalKcalExactMatchReward/std": 0.42597426772117614,
      "step": 1855
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.4453125,
      "completions/min_length": 284.8,
      "epoch": 0.5173852573018081,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6463583707809448,
      "kl": 0.045586379477754235,
      "learning_rate": 5.126478159371503e-07,
      "loss": 0.0018231194466352464,
      "reward": 3.189420700073242,
      "reward_std": 0.36946871876716614,
      "rewards/IngredientFormatReward/mean": 0.9851041674613953,
      "rewards/IngredientFormatReward/std": 0.09958937615156174,
      "rewards/IngredientMatchReward/mean": 0.6497030019760132,
      "rewards/IngredientMatchReward/std": 0.29318724274635316,
      "rewards/IngredientQuantityMatchReward/mean": 0.7218011021614075,
      "rewards/IngredientQuantityMatchReward/std": 0.36770058870315553,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.37152652740478515,
      "step": 1860
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 401.04375,
      "completions/min_length": 285.0,
      "epoch": 0.5187760778859527,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.65671706199646,
      "kl": 0.051340340590104463,
      "learning_rate": 5.103485779577284e-07,
      "loss": 0.002053655683994293,
      "reward": 3.0907777309417725,
      "reward_std": 0.4277846872806549,
      "rewards/IngredientFormatReward/mean": 0.9821614503860474,
      "rewards/IngredientFormatReward/std": 0.1232692077755928,
      "rewards/IngredientMatchReward/mean": 0.6434660196304322,
      "rewards/IngredientMatchReward/std": 0.29241865277290346,
      "rewards/IngredientQuantityMatchReward/mean": 0.6682752728462219,
      "rewards/IngredientQuantityMatchReward/std": 0.40513462424278257,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.4023512899875641,
      "step": 1865
    },
    {
      "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": 0.0265625,
      "completions/max_length": 513.0,
      "completions/mean_length": 388.9578125,
      "completions/min_length": 270.4,
      "epoch": 0.5201668984700973,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7248250246047974,
      "kl": 0.05152199740987271,
      "learning_rate": 5.08049121031685e-07,
      "loss": 0.002060762792825699,
      "reward": 3.028227710723877,
      "reward_std": 0.46579166650772097,
      "rewards/IngredientFormatReward/mean": 0.9721874952316284,
      "rewards/IngredientFormatReward/std": 0.15746290534734725,
      "rewards/IngredientMatchReward/mean": 0.6362196326255798,
      "rewards/IngredientMatchReward/std": 0.306441468000412,
      "rewards/IngredientQuantityMatchReward/mean": 0.6495080471038819,
      "rewards/IngredientQuantityMatchReward/std": 0.4341913163661957,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.41461523771286013,
      "step": 1870
    },
    {
      "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": 0.0078125,
      "completions/max_length": 506.6,
      "completions/mean_length": 395.9140625,
      "completions/min_length": 284.8,
      "epoch": 0.521557719054242,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.616625964641571,
      "kl": 0.04834558917209506,
      "learning_rate": 5.057494938090211e-07,
      "loss": 0.001933823898434639,
      "reward": 3.110442638397217,
      "reward_std": 0.41063175201416013,
      "rewards/IngredientFormatReward/mean": 0.9895089268684387,
      "rewards/IngredientFormatReward/std": 0.07405889481306076,
      "rewards/IngredientMatchReward/mean": 0.6785714387893677,
      "rewards/IngredientMatchReward/std": 0.2774973541498184,
      "rewards/IngredientQuantityMatchReward/mean": 0.6517373204231263,
      "rewards/IngredientQuantityMatchReward/std": 0.39638755321502683,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.39762794971466064,
      "step": 1875
    },
    {
      "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": 0.009375,
      "completions/max_length": 506.6,
      "completions/mean_length": 391.709375,
      "completions/min_length": 276.4,
      "epoch": 0.5229485396383866,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.665547251701355,
      "kl": 0.04781521300319582,
      "learning_rate": 5.034497449433401e-07,
      "loss": 0.0019125932827591895,
      "reward": 3.0387969493865965,
      "reward_std": 0.42612879276275634,
      "rewards/IngredientFormatReward/mean": 0.9903125047683716,
      "rewards/IngredientFormatReward/std": 0.07061336636543274,
      "rewards/IngredientMatchReward/mean": 0.6220145106315613,
      "rewards/IngredientMatchReward/std": 0.2822760850191116,
      "rewards/IngredientQuantityMatchReward/mean": 0.6280323743820191,
      "rewards/IngredientQuantityMatchReward/std": 0.4209755420684814,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.37942943572998045,
      "step": 1880
    },
    {
      "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": 0.00625,
      "completions/max_length": 511.8,
      "completions/mean_length": 385.7609375,
      "completions/min_length": 254.0,
      "epoch": 0.5243393602225312,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6137979030609131,
      "kl": 0.05282691507600248,
      "learning_rate": 5.011499230908194e-07,
      "loss": 0.002113470435142517,
      "reward": 3.118752431869507,
      "reward_std": 0.41371476650238037,
      "rewards/IngredientFormatReward/mean": 0.9895312547683716,
      "rewards/IngredientFormatReward/std": 0.08896933421492577,
      "rewards/IngredientMatchReward/mean": 0.646869421005249,
      "rewards/IngredientMatchReward/std": 0.285538986325264,
      "rewards/IngredientQuantityMatchReward/mean": 0.6636016845703125,
      "rewards/IngredientQuantityMatchReward/std": 0.4197295904159546,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.37991697788238527,
      "step": 1885
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.8,
      "completions/mean_length": 396.703125,
      "completions/min_length": 287.0,
      "epoch": 0.5257301808066759,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6123167276382446,
      "kl": 0.04784087296575308,
      "learning_rate": 4.988500769091808e-07,
      "loss": 0.0019137132912874222,
      "reward": 2.978975248336792,
      "reward_std": 0.41777234673500063,
      "rewards/IngredientFormatReward/mean": 0.9920833110809326,
      "rewards/IngredientFormatReward/std": 0.07661278918385506,
      "rewards/IngredientMatchReward/mean": 0.668877112865448,
      "rewards/IngredientMatchReward/std": 0.2592278867959976,
      "rewards/IngredientQuantityMatchReward/mean": 0.5773897886276245,
      "rewards/IngredientQuantityMatchReward/std": 0.4320713520050049,
      "rewards/TotalKcalExactMatchReward/mean": 0.740625,
      "rewards/TotalKcalExactMatchReward/std": 0.4287377059459686,
      "step": 1890
    },
    {
      "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": 0.0125,
      "completions/max_length": 510.0,
      "completions/mean_length": 394.1625,
      "completions/min_length": 252.8,
      "epoch": 0.5271210013908206,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6603817343711853,
      "kl": 0.04811024349182844,
      "learning_rate": 4.9655025505666e-07,
      "loss": 0.0019248005002737046,
      "reward": 3.117777681350708,
      "reward_std": 0.42486504316329954,
      "rewards/IngredientFormatReward/mean": 0.9875,
      "rewards/IngredientFormatReward/std": 0.08148004412651062,
      "rewards/IngredientMatchReward/mean": 0.6686545133590698,
      "rewards/IngredientMatchReward/std": 0.2971228897571564,
      "rewards/IngredientQuantityMatchReward/mean": 0.6897481679916382,
      "rewards/IngredientQuantityMatchReward/std": 0.39132343530654906,
      "rewards/TotalKcalExactMatchReward/mean": 0.771875,
      "rewards/TotalKcalExactMatchReward/std": 0.4117573142051697,
      "step": 1895
    },
    {
      "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": 0.0046875,
      "completions/max_length": 499.6,
      "completions/mean_length": 390.7015625,
      "completions/min_length": 272.6,
      "epoch": 0.5285118219749653,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6234140992164612,
      "kl": 0.05170197752304375,
      "learning_rate": 4.942505061909789e-07,
      "loss": 0.0020677197724580767,
      "reward": 3.123731803894043,
      "reward_std": 0.38712977766990664,
      "rewards/IngredientFormatReward/mean": 0.992968738079071,
      "rewards/IngredientFormatReward/std": 0.05120236147195101,
      "rewards/IngredientMatchReward/mean": 0.6404768109321595,
      "rewards/IngredientMatchReward/std": 0.2947970926761627,
      "rewards/IngredientQuantityMatchReward/mean": 0.6902862906455993,
      "rewards/IngredientQuantityMatchReward/std": 0.41358540654182435,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.39628866314888,
      "step": 1900
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.00625,
      "completions/min_length": 273.0,
      "epoch": 0.5299026425591099,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6213961839675903,
      "kl": 0.051998325670138,
      "learning_rate": 4.919508789683148e-07,
      "loss": 0.002079806849360466,
      "reward": 2.9941901683807375,
      "reward_std": 0.44365312457084655,
      "rewards/IngredientFormatReward/mean": 0.98125,
      "rewards/IngredientFormatReward/std": 0.12953428775072098,
      "rewards/IngredientMatchReward/mean": 0.6144717335700989,
      "rewards/IngredientMatchReward/std": 0.27872682809829713,
      "rewards/IngredientQuantityMatchReward/mean": 0.6562810420989991,
      "rewards/IngredientQuantityMatchReward/std": 0.42560343742370604,
      "rewards/TotalKcalExactMatchReward/mean": 0.7421875,
      "rewards/TotalKcalExactMatchReward/std": 0.4294995665550232,
      "step": 1905
    },
    {
      "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": 0.0109375,
      "completions/max_length": 505.8,
      "completions/mean_length": 397.371875,
      "completions/min_length": 258.6,
      "epoch": 0.5312934631432545,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6720630526542664,
      "kl": 0.051461770036257806,
      "learning_rate": 4.896514220422716e-07,
      "loss": 0.0020584441721439362,
      "reward": 3.0832706928253173,
      "reward_std": 0.43686205744743345,
      "rewards/IngredientFormatReward/mean": 0.9894661426544189,
      "rewards/IngredientFormatReward/std": 0.07865497097373009,
      "rewards/IngredientMatchReward/mean": 0.6534127116203308,
      "rewards/IngredientMatchReward/std": 0.2737223029136658,
      "rewards/IngredientQuantityMatchReward/mean": 0.6482043623924255,
      "rewards/IngredientQuantityMatchReward/std": 0.393373167514801,
      "rewards/TotalKcalExactMatchReward/mean": 0.7921875,
      "rewards/TotalKcalExactMatchReward/std": 0.39381408095359804,
      "step": 1910
    },
    {
      "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": 0.00625,
      "completions/max_length": 505.8,
      "completions/mean_length": 390.590625,
      "completions/min_length": 258.4,
      "epoch": 0.5326842837273992,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6541943550109863,
      "kl": 0.04684328753501177,
      "learning_rate": 4.873521840628498e-07,
      "loss": 0.0018735647201538086,
      "reward": 3.1317606925964356,
      "reward_std": 0.3644884526729584,
      "rewards/IngredientFormatReward/mean": 0.9934895753860473,
      "rewards/IngredientFormatReward/std": 0.052749037928879264,
      "rewards/IngredientMatchReward/mean": 0.6564211845397949,
      "rewards/IngredientMatchReward/std": 0.2947027921676636,
      "rewards/IngredientQuantityMatchReward/mean": 0.6568499028682708,
      "rewards/IngredientQuantityMatchReward/std": 0.4014035701751709,
      "rewards/TotalKcalExactMatchReward/mean": 0.825,
      "rewards/TotalKcalExactMatchReward/std": 0.37870767116546633,
      "step": 1915
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.8,
      "completions/mean_length": 387.7453125,
      "completions/min_length": 261.0,
      "epoch": 0.5340751043115438,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6302548050880432,
      "kl": 0.04872983023524284,
      "learning_rate": 4.850532136754177e-07,
      "loss": 0.0019491085782647132,
      "reward": 3.096049499511719,
      "reward_std": 0.3624439001083374,
      "rewards/IngredientFormatReward/mean": 0.9915624976158142,
      "rewards/IngredientFormatReward/std": 0.06838876903057098,
      "rewards/IngredientMatchReward/mean": 0.6316896080970764,
      "rewards/IngredientMatchReward/std": 0.29681622982025146,
      "rewards/IngredientQuantityMatchReward/mean": 0.6462348580360413,
      "rewards/IngredientQuantityMatchReward/std": 0.40400696396827696,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.3677105069160461,
      "step": 1920
    },
    {
      "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": 0.0171875,
      "completions/max_length": 504.8,
      "completions/mean_length": 392.2953125,
      "completions/min_length": 272.6,
      "epoch": 0.5354659248956884,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7258259654045105,
      "kl": 0.05029868185520172,
      "learning_rate": 4.827545595196824e-07,
      "loss": 0.002012159302830696,
      "reward": 3.0869573593139648,
      "reward_std": 0.3494667589664459,
      "rewards/IngredientFormatReward/mean": 0.9827585577964782,
      "rewards/IngredientFormatReward/std": 0.09563635978847743,
      "rewards/IngredientMatchReward/mean": 0.6535475254058838,
      "rewards/IngredientMatchReward/std": 0.29701568484306334,
      "rewards/IngredientQuantityMatchReward/mean": 0.6256512045860291,
      "rewards/IngredientQuantityMatchReward/std": 0.43892085552215576,
      "rewards/TotalKcalExactMatchReward/mean": 0.825,
      "rewards/TotalKcalExactMatchReward/std": 0.37521823644638064,
      "step": 1925
    },
    {
      "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": 0.0078125,
      "completions/max_length": 496.2,
      "completions/mean_length": 391.184375,
      "completions/min_length": 274.4,
      "epoch": 0.5368567454798331,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6211183071136475,
      "kl": 0.04746301844716072,
      "learning_rate": 4.804562702286604e-07,
      "loss": 0.001898600347340107,
      "reward": 3.1789096355438233,
      "reward_std": 0.349788224697113,
      "rewards/IngredientFormatReward/mean": 0.99296875,
      "rewards/IngredientFormatReward/std": 0.05298392586410046,
      "rewards/IngredientMatchReward/mean": 0.6501227736473083,
      "rewards/IngredientMatchReward/std": 0.2874899566173553,
      "rewards/IngredientQuantityMatchReward/mean": 0.6701931238174439,
      "rewards/IngredientQuantityMatchReward/std": 0.4015944302082062,
      "rewards/TotalKcalExactMatchReward/mean": 0.865625,
      "rewards/TotalKcalExactMatchReward/std": 0.3344068914651871,
      "step": 1930
    },
    {
      "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": 0.009375,
      "completions/max_length": 508.8,
      "completions/mean_length": 396.7921875,
      "completions/min_length": 276.6,
      "epoch": 0.5382475660639777,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6513813734054565,
      "kl": 0.04723638026043773,
      "learning_rate": 4.781583944276484e-07,
      "loss": 0.001889600232243538,
      "reward": 3.0589586734771728,
      "reward_std": 0.3765470445156097,
      "rewards/IngredientFormatReward/mean": 0.9848958373069763,
      "rewards/IngredientFormatReward/std": 0.10098038241267204,
      "rewards/IngredientMatchReward/mean": 0.6359009265899658,
      "rewards/IngredientMatchReward/std": 0.2934915781021118,
      "rewards/IngredientQuantityMatchReward/mean": 0.6553495287895202,
      "rewards/IngredientQuantityMatchReward/std": 0.407476407289505,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4078086793422699,
      "step": 1935
    },
    {
      "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": 0.015625,
      "completions/max_length": 512.4,
      "completions/mean_length": 396.7640625,
      "completions/min_length": 264.8,
      "epoch": 0.5396383866481224,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6842653155326843,
      "kl": 0.05088149183429778,
      "learning_rate": 4.7586098073319477e-07,
      "loss": 0.002035403996706009,
      "reward": 3.08815450668335,
      "reward_std": 0.4408495545387268,
      "rewards/IngredientFormatReward/mean": 0.9842559576034546,
      "rewards/IngredientFormatReward/std": 0.11024485677480697,
      "rewards/IngredientMatchReward/mean": 0.6107713341712951,
      "rewards/IngredientMatchReward/std": 0.26985863745212557,
      "rewards/IngredientQuantityMatchReward/mean": 0.6790647387504578,
      "rewards/IngredientQuantityMatchReward/std": 0.40906696319580077,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.38226786255836487,
      "step": 1940
    },
    {
      "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": 0.00625,
      "completions/max_length": 508.0,
      "completions/mean_length": 387.1421875,
      "completions/min_length": 241.0,
      "epoch": 0.541029207232267,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6372060775756836,
      "kl": 0.051733718090690675,
      "learning_rate": 4.735640777520716e-07,
      "loss": 0.0020697489380836487,
      "reward": 3.1369984626770018,
      "reward_std": 0.41252625584602354,
      "rewards/IngredientFormatReward/mean": 0.9944010376930237,
      "rewards/IngredientFormatReward/std": 0.051148696616292,
      "rewards/IngredientMatchReward/mean": 0.6390866875648499,
      "rewards/IngredientMatchReward/std": 0.29094711542129514,
      "rewards/IngredientQuantityMatchReward/mean": 0.6597607374191284,
      "rewards/IngredientQuantityMatchReward/std": 0.40805192589759826,
      "rewards/TotalKcalExactMatchReward/mean": 0.84375,
      "rewards/TotalKcalExactMatchReward/std": 0.3470917284488678,
      "step": 1945
    },
    {
      "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": 0.0109375,
      "completions/max_length": 504.8,
      "completions/mean_length": 390.95625,
      "completions/min_length": 248.2,
      "epoch": 0.5424200278164116,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 16.270816802978516,
      "kl": 0.11754101505503059,
      "learning_rate": 4.7126773408024536e-07,
      "loss": 0.004704966396093369,
      "reward": 3.0708155155181887,
      "reward_std": 0.3849705934524536,
      "rewards/IngredientFormatReward/mean": 0.9884895801544189,
      "rewards/IngredientFormatReward/std": 0.08361713178455829,
      "rewards/IngredientMatchReward/mean": 0.6275595426559448,
      "rewards/IngredientMatchReward/std": 0.30697761178016664,
      "rewards/IngredientQuantityMatchReward/mean": 0.6485163748264313,
      "rewards/IngredientQuantityMatchReward/std": 0.41862033009529115,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.39109028577804567,
      "step": 1950
    },
    {
      "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": 0.009375,
      "completions/max_length": 500.2,
      "completions/mean_length": 396.546875,
      "completions/min_length": 267.6,
      "epoch": 0.5438108484005564,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 26.406909942626953,
      "kl": 0.10048759747296572,
      "learning_rate": 4.689719983018488e-07,
      "loss": 0.004031499475240707,
      "reward": 3.0238998413085936,
      "reward_std": 0.37822062373161314,
      "rewards/IngredientFormatReward/mean": 0.9881770849227905,
      "rewards/IngredientFormatReward/std": 0.07832716181874275,
      "rewards/IngredientMatchReward/mean": 0.6181553840637207,
      "rewards/IngredientMatchReward/std": 0.29799574613571167,
      "rewards/IngredientQuantityMatchReward/mean": 0.5863174438476563,
      "rewards/IngredientQuantityMatchReward/std": 0.4404876172542572,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.3680418610572815,
      "step": 1955
    },
    {
      "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": 0.0234375,
      "completions/max_length": 511.0,
      "completions/mean_length": 401.3109375,
      "completions/min_length": 271.8,
      "epoch": 0.545201668984701,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6423550248146057,
      "kl": 0.047905878722667695,
      "learning_rate": 4.666769189881539e-07,
      "loss": 0.0019165009260177612,
      "reward": 3.1259632110595703,
      "reward_std": 0.46272686719894407,
      "rewards/IngredientFormatReward/mean": 0.9774181604385376,
      "rewards/IngredientFormatReward/std": 0.1268059030175209,
      "rewards/IngredientMatchReward/mean": 0.6502027630805969,
      "rewards/IngredientMatchReward/std": 0.29270836114883425,
      "rewards/IngredientQuantityMatchReward/mean": 0.6874048829078674,
      "rewards/IngredientQuantityMatchReward/std": 0.40191797018051145,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.39044166207313535,
      "step": 1960
    },
    {
      "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": 0.0046875,
      "completions/max_length": 497.2,
      "completions/mean_length": 383.03125,
      "completions/min_length": 239.2,
      "epoch": 0.5465924895688457,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7022526860237122,
      "kl": 0.051917689852416515,
      "learning_rate": 4.6438254469654346e-07,
      "loss": 0.0020762814208865167,
      "reward": 3.08801589012146,
      "reward_std": 0.39448719620704653,
      "rewards/IngredientFormatReward/mean": 0.9934895753860473,
      "rewards/IngredientFormatReward/std": 0.05592806488275528,
      "rewards/IngredientMatchReward/mean": 0.6122198224067688,
      "rewards/IngredientMatchReward/std": 0.29717632532119753,
      "rewards/IngredientQuantityMatchReward/mean": 0.6713690161705017,
      "rewards/IngredientQuantityMatchReward/std": 0.4097886860370636,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.3882238745689392,
      "step": 1965
    },
    {
      "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": 0.0125,
      "completions/max_length": 512.8,
      "completions/mean_length": 390.5203125,
      "completions/min_length": 262.6,
      "epoch": 0.5479833101529903,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.702995777130127,
      "kl": 0.052073144866153595,
      "learning_rate": 4.6208892396948415e-07,
      "loss": 0.0020829498767852782,
      "reward": 3.050513982772827,
      "reward_std": 0.42421574592590333,
      "rewards/IngredientFormatReward/mean": 0.9853236675262451,
      "rewards/IngredientFormatReward/std": 0.10791139267385005,
      "rewards/IngredientMatchReward/mean": 0.6532893180847168,
      "rewards/IngredientMatchReward/std": 0.28823532462120055,
      "rewards/IngredientQuantityMatchReward/mean": 0.6759635448455811,
      "rewards/IngredientQuantityMatchReward/std": 0.41645088195800783,
      "rewards/TotalKcalExactMatchReward/mean": 0.7359375,
      "rewards/TotalKcalExactMatchReward/std": 0.40241574943065644,
      "step": 1970
    },
    {
      "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": 0.0078125,
      "completions/max_length": 511.6,
      "completions/mean_length": 391.7546875,
      "completions/min_length": 266.2,
      "epoch": 0.5493741307371349,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.5921576619148254,
      "kl": 0.04812497249804437,
      "learning_rate": 4.59796105333499e-07,
      "loss": 0.0019246641546487807,
      "reward": 3.038289928436279,
      "reward_std": 0.3981735408306122,
      "rewards/IngredientFormatReward/mean": 0.9900000095367432,
      "rewards/IngredientFormatReward/std": 0.07664789929986,
      "rewards/IngredientMatchReward/mean": 0.625024163722992,
      "rewards/IngredientMatchReward/std": 0.29402685165405273,
      "rewards/IngredientQuantityMatchReward/mean": 0.6451407432556152,
      "rewards/IngredientQuantityMatchReward/std": 0.42612059116363527,
      "rewards/TotalKcalExactMatchReward/mean": 0.778125,
      "rewards/TotalKcalExactMatchReward/std": 0.40204805731773374,
      "step": 1975
    },
    {
      "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": 0.01875,
      "completions/max_length": 508.8,
      "completions/mean_length": 402.7,
      "completions/min_length": 276.0,
      "epoch": 0.5507649513212796,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6542710661888123,
      "kl": 0.04761419745627791,
      "learning_rate": 4.5750413729814134e-07,
      "loss": 0.0019045617431402207,
      "reward": 3.0831342220306395,
      "reward_std": 0.4208559811115265,
      "rewards/IngredientFormatReward/mean": 0.9742559432983399,
      "rewards/IngredientFormatReward/std": 0.12943890187889337,
      "rewards/IngredientMatchReward/mean": 0.654008400440216,
      "rewards/IngredientMatchReward/std": 0.2880619168281555,
      "rewards/IngredientQuantityMatchReward/mean": 0.6736199378967285,
      "rewards/IngredientQuantityMatchReward/std": 0.37951394319534304,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.4055814027786255,
      "step": 1980
    },
    {
      "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": 0.0171875,
      "completions/max_length": 511.8,
      "completions/mean_length": 396.68125,
      "completions/min_length": 259.6,
      "epoch": 0.5521557719054242,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6633925437927246,
      "kl": 0.09954718144144863,
      "learning_rate": 4.552130683549682e-07,
      "loss": 0.003990843892097473,
      "reward": 3.1635501384735107,
      "reward_std": 0.4512316048145294,
      "rewards/IngredientFormatReward/mean": 0.9836718797683716,
      "rewards/IngredientFormatReward/std": 0.10353007614612579,
      "rewards/IngredientMatchReward/mean": 0.7020808815956116,
      "rewards/IngredientMatchReward/std": 0.2649115890264511,
      "rewards/IngredientQuantityMatchReward/mean": 0.6449849724769592,
      "rewards/IngredientQuantityMatchReward/std": 0.42183072566986085,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.3696657299995422,
      "step": 1985
    },
    {
      "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": 0.0109375,
      "completions/max_length": 512.8,
      "completions/mean_length": 393.4203125,
      "completions/min_length": 262.8,
      "epoch": 0.5535465924895688,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.699221134185791,
      "kl": 0.05136008146218955,
      "learning_rate": 4.529229469765147e-07,
      "loss": 0.002054634690284729,
      "reward": 3.000317668914795,
      "reward_std": 0.4171613872051239,
      "rewards/IngredientFormatReward/mean": 0.983411455154419,
      "rewards/IngredientFormatReward/std": 0.1190970316529274,
      "rewards/IngredientMatchReward/mean": 0.6175824522972106,
      "rewards/IngredientMatchReward/std": 0.2949405133724213,
      "rewards/IngredientQuantityMatchReward/mean": 0.6352612376213074,
      "rewards/IngredientQuantityMatchReward/std": 0.4082452058792114,
      "rewards/TotalKcalExactMatchReward/mean": 0.7640625,
      "rewards/TotalKcalExactMatchReward/std": 0.40400839447975156,
      "step": 1990
    },
    {
      "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": 0.0046875,
      "completions/max_length": 500.6,
      "completions/mean_length": 394.9859375,
      "completions/min_length": 268.6,
      "epoch": 0.5549374130737135,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 1.7067408561706543,
      "kl": 0.05321628218516707,
      "learning_rate": 4.50633821615267e-07,
      "loss": 0.0021282849833369256,
      "reward": 3.1403940677642823,
      "reward_std": 0.3678104758262634,
      "rewards/IngredientFormatReward/mean": 0.9927938938140869,
      "rewards/IngredientFormatReward/std": 0.05982886739075184,
      "rewards/IngredientMatchReward/mean": 0.6677821159362793,
      "rewards/IngredientMatchReward/std": 0.2782954514026642,
      "rewards/IngredientQuantityMatchReward/mean": 0.6579430580139161,
      "rewards/IngredientQuantityMatchReward/std": 0.40156732201576234,
      "rewards/TotalKcalExactMatchReward/mean": 0.821875,
      "rewards/TotalKcalExactMatchReward/std": 0.3716013550758362,
      "step": 1995
    },
    {
      "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": 0.01875,
      "completions/max_length": 509.6,
      "completions/mean_length": 391.5359375,
      "completions/min_length": 244.0,
      "epoch": 0.5563282336578581,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6526069045066833,
      "kl": 0.04948800450656563,
      "learning_rate": 4.4834574070263974e-07,
      "loss": 0.001979481242597103,
      "reward": 3.083600330352783,
      "reward_std": 0.4555691659450531,
      "rewards/IngredientFormatReward/mean": 0.9757143020629883,
      "rewards/IngredientFormatReward/std": 0.13633684441447258,
      "rewards/IngredientMatchReward/mean": 0.6698400378227234,
      "rewards/IngredientMatchReward/std": 0.29061307311058043,
      "rewards/IngredientQuantityMatchReward/mean": 0.6161710500717164,
      "rewards/IngredientQuantityMatchReward/std": 0.4445948779582977,
      "rewards/TotalKcalExactMatchReward/mean": 0.821875,
      "rewards/TotalKcalExactMatchReward/std": 0.3704223304986954,
      "step": 2000
    },
    {
      "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": 0.0046875,
      "completions/max_length": 499.6,
      "completions/mean_length": 397.13125,
      "completions/min_length": 281.2,
      "epoch": 0.5577190542420027,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.640072226524353,
      "kl": 0.05212169520091266,
      "learning_rate": 4.460587526479492e-07,
      "loss": 0.002084844559431076,
      "reward": 3.1748961925506594,
      "reward_std": 0.3355372905731201,
      "rewards/IngredientFormatReward/mean": 0.9914843797683716,
      "rewards/IngredientFormatReward/std": 0.05400210823863745,
      "rewards/IngredientMatchReward/mean": 0.6208835721015931,
      "rewards/IngredientMatchReward/std": 0.27205938696861265,
      "rewards/IngredientQuantityMatchReward/mean": 0.7047157406806945,
      "rewards/IngredientQuantityMatchReward/std": 0.3964652955532074,
      "rewards/TotalKcalExactMatchReward/mean": 0.8578125,
      "rewards/TotalKcalExactMatchReward/std": 0.344100821018219,
      "step": 2005
    },
    {
      "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": 0.0203125,
      "completions/max_length": 509.0,
      "completions/mean_length": 394.5671875,
      "completions/min_length": 266.0,
      "epoch": 0.5591098748261474,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.5653246641159058,
      "kl": 0.04876701077446342,
      "learning_rate": 4.4377290583739023e-07,
      "loss": 0.0019505947828292846,
      "reward": 3.094419002532959,
      "reward_std": 0.43140562176704406,
      "rewards/IngredientFormatReward/mean": 0.9779278397560119,
      "rewards/IngredientFormatReward/std": 0.13397145569324492,
      "rewards/IngredientMatchReward/mean": 0.6553794503211975,
      "rewards/IngredientMatchReward/std": 0.30059259533882143,
      "rewards/IngredientQuantityMatchReward/mean": 0.6423617362976074,
      "rewards/IngredientQuantityMatchReward/std": 0.4208782434463501,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.36920258700847625,
      "step": 2010
    },
    {
      "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": 0.015625,
      "completions/max_length": 506.2,
      "completions/mean_length": 387.225,
      "completions/min_length": 254.6,
      "epoch": 0.5605006954102921,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7243109941482544,
      "kl": 0.0509596262825653,
      "learning_rate": 4.414882486330118e-07,
      "loss": 0.0020386766642332076,
      "reward": 2.980712413787842,
      "reward_std": 0.4302122414112091,
      "rewards/IngredientFormatReward/mean": 0.9727083325386048,
      "rewards/IngredientFormatReward/std": 0.12047353163361549,
      "rewards/IngredientMatchReward/mean": 0.6474566102027893,
      "rewards/IngredientMatchReward/std": 0.28137449622154237,
      "rewards/IngredientQuantityMatchReward/mean": 0.5964849531650543,
      "rewards/IngredientQuantityMatchReward/std": 0.43391476273536683,
      "rewards/TotalKcalExactMatchReward/mean": 0.7640625,
      "rewards/TotalKcalExactMatchReward/std": 0.41384188532829286,
      "step": 2015
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.8,
      "completions/mean_length": 388.784375,
      "completions/min_length": 259.4,
      "epoch": 0.5618915159944368,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7031833529472351,
      "kl": 0.04954910618253052,
      "learning_rate": 4.392048293716945e-07,
      "loss": 0.0019819369539618493,
      "reward": 3.0167937755584715,
      "reward_std": 0.42304224967956544,
      "rewards/IngredientFormatReward/mean": 0.9863020658493042,
      "rewards/IngredientFormatReward/std": 0.09604550898075104,
      "rewards/IngredientMatchReward/mean": 0.6214434504508972,
      "rewards/IngredientMatchReward/std": 0.30990185141563414,
      "rewards/IngredientQuantityMatchReward/mean": 0.6402982592582702,
      "rewards/IngredientQuantityMatchReward/std": 0.42372643351554873,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.42207556366920473,
      "step": 2020
    },
    {
      "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": 0.0296875,
      "completions/max_length": 512.8,
      "completions/mean_length": 395.628125,
      "completions/min_length": 260.6,
      "epoch": 0.5632823365785814,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5899977684020996,
      "kl": 0.05259674787521362,
      "learning_rate": 4.3692269636412735e-07,
      "loss": 0.0021039582788944244,
      "reward": 3.0533965110778807,
      "reward_std": 0.44099304676055906,
      "rewards/IngredientFormatReward/mean": 0.9689062356948852,
      "rewards/IngredientFormatReward/std": 0.15842351466417312,
      "rewards/IngredientMatchReward/mean": 0.663708484172821,
      "rewards/IngredientMatchReward/std": 0.2866365730762482,
      "rewards/IngredientQuantityMatchReward/mean": 0.6770317912101745,
      "rewards/IngredientQuantityMatchReward/std": 0.40612250566482544,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.4267735958099365,
      "step": 2025
    },
    {
      "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": 0.0046875,
      "completions/max_length": 502.8,
      "completions/mean_length": 399.5,
      "completions/min_length": 294.4,
      "epoch": 0.564673157162726,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6661276817321777,
      "kl": 0.0543745540548116,
      "learning_rate": 4.3464189789378623e-07,
      "loss": 0.0021728880703449248,
      "reward": 3.174680805206299,
      "reward_std": 0.3439257264137268,
      "rewards/IngredientFormatReward/mean": 0.9936979055404663,
      "rewards/IngredientFormatReward/std": 0.05086542461067438,
      "rewards/IngredientMatchReward/mean": 0.6711421132087707,
      "rewards/IngredientMatchReward/std": 0.2759589582681656,
      "rewards/IngredientQuantityMatchReward/mean": 0.7098408579826355,
      "rewards/IngredientQuantityMatchReward/std": 0.38981047868728635,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.39488060474395753,
      "step": 2030
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.2,
      "completions/mean_length": 387.4421875,
      "completions/min_length": 256.0,
      "epoch": 0.5660639777468707,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7035443782806396,
      "kl": 0.05331933759152889,
      "learning_rate": 4.3236248221591155e-07,
      "loss": 0.0021326553076505663,
      "reward": 3.0655640602111816,
      "reward_std": 0.43879054188728334,
      "rewards/IngredientFormatReward/mean": 0.9842187523841858,
      "rewards/IngredientFormatReward/std": 0.0974664755165577,
      "rewards/IngredientMatchReward/mean": 0.6237977385520935,
      "rewards/IngredientMatchReward/std": 0.26321753561496736,
      "rewards/IngredientQuantityMatchReward/mean": 0.6575475811958313,
      "rewards/IngredientQuantityMatchReward/std": 0.4133492708206177,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.39513932466506957,
      "step": 2035
    },
    {
      "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": 0.0125,
      "completions/max_length": 511.0,
      "completions/mean_length": 394.975,
      "completions/min_length": 254.4,
      "epoch": 0.5674547983310153,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.7625192403793335,
      "kl": 0.051801764173433185,
      "learning_rate": 4.300844975564878e-07,
      "loss": 0.0020724233239889146,
      "reward": 3.065888023376465,
      "reward_std": 0.40156723856925963,
      "rewards/IngredientFormatReward/mean": 0.9851637005805969,
      "rewards/IngredientFormatReward/std": 0.09742565155029297,
      "rewards/IngredientMatchReward/mean": 0.6349423289299011,
      "rewards/IngredientMatchReward/std": 0.28503390550613406,
      "rewards/IngredientQuantityMatchReward/mean": 0.6426570415496826,
      "rewards/IngredientQuantityMatchReward/std": 0.42421199679374694,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.3835489392280579,
      "step": 2040
    },
    {
      "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": 0.00625,
      "completions/max_length": 499.6,
      "completions/mean_length": 390.475,
      "completions/min_length": 267.0,
      "epoch": 0.56884561891516,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6383020281791687,
      "kl": 0.04932130335364491,
      "learning_rate": 4.278079921112235e-07,
      "loss": 0.0019727183505892755,
      "reward": 3.1172127723693848,
      "reward_std": 0.3688182234764099,
      "rewards/IngredientFormatReward/mean": 0.9934895753860473,
      "rewards/IngredientFormatReward/std": 0.04815645068883896,
      "rewards/IngredientMatchReward/mean": 0.6810300469398498,
      "rewards/IngredientMatchReward/std": 0.28476625084877016,
      "rewards/IngredientQuantityMatchReward/mean": 0.6755056738853454,
      "rewards/IngredientQuantityMatchReward/std": 0.40387414693832396,
      "rewards/TotalKcalExactMatchReward/mean": 0.7671875,
      "rewards/TotalKcalExactMatchReward/std": 0.41967945694923403,
      "step": 2045
    },
    {
      "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": 0.0171875,
      "completions/max_length": 509.6,
      "completions/mean_length": 394.609375,
      "completions/min_length": 254.4,
      "epoch": 0.5702364394993046,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6549330353736877,
      "kl": 0.05572653063572943,
      "learning_rate": 4.2553301404453117e-07,
      "loss": 0.0022291053086519242,
      "reward": 2.9725974559783936,
      "reward_std": 0.4430975794792175,
      "rewards/IngredientFormatReward/mean": 0.9778125047683716,
      "rewards/IngredientFormatReward/std": 0.1242400899529457,
      "rewards/IngredientMatchReward/mean": 0.5863467276096344,
      "rewards/IngredientMatchReward/std": 0.2854700267314911,
      "rewards/IngredientQuantityMatchReward/mean": 0.5896882534027099,
      "rewards/IngredientQuantityMatchReward/std": 0.43062989711761473,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.37587525844573977,
      "step": 2050
    },
    {
      "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": 0.00625,
      "completions/max_length": 508.2,
      "completions/mean_length": 393.278125,
      "completions/min_length": 278.6,
      "epoch": 0.5716272600834492,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6046156883239746,
      "kl": 0.0488880162127316,
      "learning_rate": 4.232596114885078e-07,
      "loss": 0.0019555903971195223,
      "reward": 3.081752300262451,
      "reward_std": 0.37385761737823486,
      "rewards/IngredientFormatReward/mean": 0.9907775282859802,
      "rewards/IngredientFormatReward/std": 0.07260209694504738,
      "rewards/IngredientMatchReward/mean": 0.5988665580749511,
      "rewards/IngredientMatchReward/std": 0.2849681168794632,
      "rewards/IngredientQuantityMatchReward/mean": 0.679608142375946,
      "rewards/IngredientQuantityMatchReward/std": 0.4193470299243927,
      "rewards/TotalKcalExactMatchReward/mean": 0.8125,
      "rewards/TotalKcalExactMatchReward/std": 0.3588409751653671,
      "step": 2055
    },
    {
      "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": 0.0125,
      "completions/max_length": 503.8,
      "completions/mean_length": 385.078125,
      "completions/min_length": 271.2,
      "epoch": 0.5730180806675939,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6553982496261597,
      "kl": 0.05303655038587749,
      "learning_rate": 4.209878325419175e-07,
      "loss": 0.0021214175969362257,
      "reward": 2.986969470977783,
      "reward_std": 0.4297593176364899,
      "rewards/IngredientFormatReward/mean": 0.9858854174613952,
      "rewards/IngredientFormatReward/std": 0.07832327783107758,
      "rewards/IngredientMatchReward/mean": 0.6215869665145874,
      "rewards/IngredientMatchReward/std": 0.3030324518680573,
      "rewards/IngredientQuantityMatchReward/mean": 0.6326221823692322,
      "rewards/IngredientQuantityMatchReward/std": 0.4269497454166412,
      "rewards/TotalKcalExactMatchReward/mean": 0.746875,
      "rewards/TotalKcalExactMatchReward/std": 0.4260698974132538,
      "step": 2060
    },
    {
      "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": 0.0140625,
      "completions/max_length": 508.8,
      "completions/mean_length": 398.115625,
      "completions/min_length": 281.0,
      "epoch": 0.5744089012517385,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6358456015586853,
      "kl": 0.04783306939061731,
      "learning_rate": 4.187177252691733e-07,
      "loss": 0.0019132602959871293,
      "reward": 2.9515925884246825,
      "reward_std": 0.40451722145080565,
      "rewards/IngredientFormatReward/mean": 0.9834486603736877,
      "rewards/IngredientFormatReward/std": 0.09680206961929798,
      "rewards/IngredientMatchReward/mean": 0.6217875838279724,
      "rewards/IngredientMatchReward/std": 0.2716970294713974,
      "rewards/IngredientQuantityMatchReward/mean": 0.5947938621044159,
      "rewards/IngredientQuantityMatchReward/std": 0.42498272061347964,
      "rewards/TotalKcalExactMatchReward/mean": 0.7515625,
      "rewards/TotalKcalExactMatchReward/std": 0.42524816393852233,
      "step": 2065
    },
    {
      "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": 0.0109375,
      "completions/max_length": 488.6,
      "completions/mean_length": 386.153125,
      "completions/min_length": 233.2,
      "epoch": 0.5757997218358831,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6399543881416321,
      "kl": 0.04871748448349535,
      "learning_rate": 4.164493376993204e-07,
      "loss": 0.0019484382122755052,
      "reward": 3.167618465423584,
      "reward_std": 0.39762314558029177,
      "rewards/IngredientFormatReward/mean": 0.9763392925262451,
      "rewards/IngredientFormatReward/std": 0.08031712025403977,
      "rewards/IngredientMatchReward/mean": 0.6643613576889038,
      "rewards/IngredientMatchReward/std": 0.27025269269943236,
      "rewards/IngredientQuantityMatchReward/mean": 0.689417815208435,
      "rewards/IngredientQuantityMatchReward/std": 0.3921951472759247,
      "rewards/TotalKcalExactMatchReward/mean": 0.8375,
      "rewards/TotalKcalExactMatchReward/std": 0.3567990481853485,
      "step": 2070
    },
    {
      "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": 0.0109375,
      "completions/max_length": 510.0,
      "completions/mean_length": 392.71875,
      "completions/min_length": 275.6,
      "epoch": 0.5771905424200278,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6949378252029419,
      "kl": 0.04844976761378348,
      "learning_rate": 4.141827178250197e-07,
      "loss": 0.001937885209918022,
      "reward": 3.137814426422119,
      "reward_std": 0.3677864849567413,
      "rewards/IngredientFormatReward/mean": 0.9881119847297668,
      "rewards/IngredientFormatReward/std": 0.09051166772842408,
      "rewards/IngredientMatchReward/mean": 0.6526115655899047,
      "rewards/IngredientMatchReward/std": 0.2787861585617065,
      "rewards/IngredientQuantityMatchReward/mean": 0.6767783999443054,
      "rewards/IngredientQuantityMatchReward/std": 0.3985203385353088,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.37351303100585936,
      "step": 2075
    },
    {
      "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": 0.0171875,
      "completions/max_length": 507.4,
      "completions/mean_length": 393.246875,
      "completions/min_length": 253.8,
      "epoch": 0.5785813630041725,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.603931188583374,
      "kl": 0.04737849328666925,
      "learning_rate": 4.1191791360153283e-07,
      "loss": 0.0018953494727611542,
      "reward": 3.06357889175415,
      "reward_std": 0.4036316812038422,
      "rewards/IngredientFormatReward/mean": 0.9819791555404663,
      "rewards/IngredientFormatReward/std": 0.09684197474271058,
      "rewards/IngredientMatchReward/mean": 0.6268632292747498,
      "rewards/IngredientMatchReward/std": 0.2809248685836792,
      "rewards/IngredientQuantityMatchReward/mean": 0.6234864950180053,
      "rewards/IngredientQuantityMatchReward/std": 0.4336868166923523,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.3656547605991364,
      "step": 2080
    },
    {
      "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": 0.0078125,
      "completions/max_length": 504.0,
      "completions/mean_length": 386.89375,
      "completions/min_length": 262.4,
      "epoch": 0.5799721835883171,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6973137855529785,
      "kl": 0.048510912759229544,
      "learning_rate": 4.0965497294570736e-07,
      "loss": 0.0019405171275138854,
      "reward": 3.129045104980469,
      "reward_std": 0.4232663154602051,
      "rewards/IngredientFormatReward/mean": 0.9909635543823242,
      "rewards/IngredientFormatReward/std": 0.07800297439098358,
      "rewards/IngredientMatchReward/mean": 0.65011967420578,
      "rewards/IngredientMatchReward/std": 0.2852735579013824,
      "rewards/IngredientQuantityMatchReward/mean": 0.6926494479179383,
      "rewards/IngredientQuantityMatchReward/std": 0.39057775139808654,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.4028699815273285,
      "step": 2085
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.0,
      "completions/mean_length": 389.546875,
      "completions/min_length": 274.8,
      "epoch": 0.5813630041724618,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6442022919654846,
      "kl": 0.04921837141737342,
      "learning_rate": 4.073939437349636e-07,
      "loss": 0.001969057321548462,
      "reward": 2.978153944015503,
      "reward_std": 0.43001670241355894,
      "rewards/IngredientFormatReward/mean": 0.9893489599227905,
      "rewards/IngredientFormatReward/std": 0.08917459473013878,
      "rewards/IngredientMatchReward/mean": 0.5978354513645172,
      "rewards/IngredientMatchReward/std": 0.27598640620708464,
      "rewards/IngredientQuantityMatchReward/mean": 0.6284695625305176,
      "rewards/IngredientQuantityMatchReward/std": 0.4286845028400421,
      "rewards/TotalKcalExactMatchReward/mean": 0.7625,
      "rewards/TotalKcalExactMatchReward/std": 0.42297797799110415,
      "step": 2090
    },
    {
      "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": 0.0140625,
      "completions/max_length": 505.2,
      "completions/mean_length": 390.3421875,
      "completions/min_length": 277.2,
      "epoch": 0.5827538247566064,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.627018392086029,
      "kl": 0.05088339238427579,
      "learning_rate": 4.0513487380627983e-07,
      "loss": 0.0020353712141513826,
      "reward": 3.052702856063843,
      "reward_std": 0.4378420948982239,
      "rewards/IngredientFormatReward/mean": 0.9756064057350159,
      "rewards/IngredientFormatReward/std": 0.13277253992855548,
      "rewards/IngredientMatchReward/mean": 0.6264713406562805,
      "rewards/IngredientMatchReward/std": 0.2919108152389526,
      "rewards/IngredientQuantityMatchReward/mean": 0.6475001513957978,
      "rewards/IngredientQuantityMatchReward/std": 0.40529059767723086,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.38602237701416015,
      "step": 2095
    },
    {
      "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": 0.0046875,
      "completions/max_length": 502.0,
      "completions/mean_length": 385.890625,
      "completions/min_length": 241.6,
      "epoch": 0.5841446453407511,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6189238429069519,
      "kl": 0.050543641904369,
      "learning_rate": 4.028778109551826e-07,
      "loss": 0.002021963521838188,
      "reward": 3.0809757709503174,
      "reward_std": 0.34558130502700807,
      "rewards/IngredientFormatReward/mean": 0.993359375,
      "rewards/IngredientFormatReward/std": 0.06467613540589809,
      "rewards/IngredientMatchReward/mean": 0.6460584044456482,
      "rewards/IngredientMatchReward/std": 0.29698742628097535,
      "rewards/IngredientQuantityMatchReward/mean": 0.6228079795837402,
      "rewards/IngredientQuantityMatchReward/std": 0.423473596572876,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.3815928041934967,
      "step": 2100
    },
    {
      "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": 0.009375,
      "completions/max_length": 499.0,
      "completions/mean_length": 382.5234375,
      "completions/min_length": 252.6,
      "epoch": 0.5855354659248957,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6478217244148254,
      "kl": 0.049496779195033015,
      "learning_rate": 4.006228029347339e-07,
      "loss": 0.0019800461828708648,
      "reward": 3.1182824611663817,
      "reward_std": 0.32764893770217896,
      "rewards/IngredientFormatReward/mean": 0.9918229103088378,
      "rewards/IngredientFormatReward/std": 0.06296356171369552,
      "rewards/IngredientMatchReward/mean": 0.6453193426132202,
      "rewards/IngredientMatchReward/std": 0.3042366325855255,
      "rewards/IngredientQuantityMatchReward/mean": 0.6686402320861816,
      "rewards/IngredientQuantityMatchReward/std": 0.4119422197341919,
      "rewards/TotalKcalExactMatchReward/mean": 0.8125,
      "rewards/TotalKcalExactMatchReward/std": 0.3838110685348511,
      "step": 2105
    },
    {
      "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": 0.0140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 390.55625,
      "completions/min_length": 255.6,
      "epoch": 0.5869262865090403,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6478376984596252,
      "kl": 0.048133849655278026,
      "learning_rate": 3.983698974545215e-07,
      "loss": 0.0019253019243478775,
      "reward": 3.0814141750335695,
      "reward_std": 0.41355874538421633,
      "rewards/IngredientFormatReward/mean": 0.9811160683631897,
      "rewards/IngredientFormatReward/std": 0.12442523241043091,
      "rewards/IngredientMatchReward/mean": 0.675525176525116,
      "rewards/IngredientMatchReward/std": 0.2912832021713257,
      "rewards/IngredientQuantityMatchReward/mean": 0.6716479063034058,
      "rewards/IngredientQuantityMatchReward/std": 0.39944403171539306,
      "rewards/TotalKcalExactMatchReward/mean": 0.753125,
      "rewards/TotalKcalExactMatchReward/std": 0.41150047183036803,
      "step": 2110
    },
    {
      "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": 0.003125,
      "completions/max_length": 501.2,
      "completions/mean_length": 390.93125,
      "completions/min_length": 275.2,
      "epoch": 0.588317107093185,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6301591396331787,
      "kl": 0.05105856913141906,
      "learning_rate": 3.9611914217964916e-07,
      "loss": 0.0020423345267772675,
      "reward": 3.1168880462646484,
      "reward_std": 0.3775189995765686,
      "rewards/IngredientFormatReward/mean": 0.9966145753860474,
      "rewards/IngredientFormatReward/std": 0.035576280951499936,
      "rewards/IngredientMatchReward/mean": 0.6276953220367432,
      "rewards/IngredientMatchReward/std": 0.2857614278793335,
      "rewards/IngredientQuantityMatchReward/mean": 0.691015648841858,
      "rewards/IngredientQuantityMatchReward/std": 0.4060320556163788,
      "rewards/TotalKcalExactMatchReward/mean": 0.8015625,
      "rewards/TotalKcalExactMatchReward/std": 0.3695449620485306,
      "step": 2115
    },
    {
      "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": 0.0078125,
      "completions/max_length": 511.4,
      "completions/mean_length": 393.4765625,
      "completions/min_length": 262.2,
      "epoch": 0.5897079276773296,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6236531734466553,
      "kl": 0.0492264065425843,
      "learning_rate": 3.9387058472972843e-07,
      "loss": 0.001969209872186184,
      "reward": 3.0744152545928953,
      "reward_std": 0.39720383286476135,
      "rewards/IngredientFormatReward/mean": 0.9910416722297668,
      "rewards/IngredientFormatReward/std": 0.06891115196049213,
      "rewards/IngredientMatchReward/mean": 0.6055537104606629,
      "rewards/IngredientMatchReward/std": 0.30302745699882505,
      "rewards/IngredientQuantityMatchReward/mean": 0.6168824195861816,
      "rewards/IngredientQuantityMatchReward/std": 0.4277641594409943,
      "rewards/TotalKcalExactMatchReward/mean": 0.8609375,
      "rewards/TotalKcalExactMatchReward/std": 0.3140183940529823,
      "step": 2120
    },
    {
      "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": 0.00625,
      "completions/max_length": 501.6,
      "completions/mean_length": 385.7359375,
      "completions/min_length": 263.2,
      "epoch": 0.5910987482614742,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6744566559791565,
      "kl": 0.04763327166438103,
      "learning_rate": 3.916242726778712e-07,
      "loss": 0.0019053336232900619,
      "reward": 3.2508696556091308,
      "reward_std": 0.37193950414657595,
      "rewards/IngredientFormatReward/mean": 0.9950520753860473,
      "rewards/IngredientFormatReward/std": 0.0532539501786232,
      "rewards/IngredientMatchReward/mean": 0.7194382548332214,
      "rewards/IngredientMatchReward/std": 0.2809447020292282,
      "rewards/IngredientQuantityMatchReward/mean": 0.7191918611526489,
      "rewards/IngredientQuantityMatchReward/std": 0.39092591404914856,
      "rewards/TotalKcalExactMatchReward/mean": 0.8171875,
      "rewards/TotalKcalExactMatchReward/std": 0.3763358473777771,
      "step": 2125
    },
    {
      "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": 0.0078125,
      "completions/max_length": 506.8,
      "completions/mean_length": 389.2796875,
      "completions/min_length": 261.0,
      "epoch": 0.5924895688456189,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6765151619911194,
      "kl": 0.04687405494041741,
      "learning_rate": 3.893802535496834e-07,
      "loss": 0.0018747389316558839,
      "reward": 3.061327362060547,
      "reward_std": 0.3514033019542694,
      "rewards/IngredientFormatReward/mean": 0.9900520801544189,
      "rewards/IngredientFormatReward/std": 0.08974247500300407,
      "rewards/IngredientMatchReward/mean": 0.5992509841918945,
      "rewards/IngredientMatchReward/std": 0.31761006712913514,
      "rewards/IngredientQuantityMatchReward/mean": 0.6188993215560913,
      "rewards/IngredientQuantityMatchReward/std": 0.41553970575332644,
      "rewards/TotalKcalExactMatchReward/mean": 0.853125,
      "rewards/TotalKcalExactMatchReward/std": 0.34410153329372406,
      "step": 2130
    },
    {
      "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": 0.00625,
      "completions/max_length": 510.2,
      "completions/mean_length": 390.790625,
      "completions/min_length": 295.4,
      "epoch": 0.5938803894297635,
      "frac_reward_zero_std": 0.1375,
      "grad_norm": 0.5965428352355957,
      "kl": 0.05046862023882568,
      "learning_rate": 3.871385748222583e-07,
      "loss": 0.0020183827728033066,
      "reward": 3.207280683517456,
      "reward_std": 0.32688103020191195,
      "rewards/IngredientFormatReward/mean": 0.9804427146911621,
      "rewards/IngredientFormatReward/std": 0.11501296181231738,
      "rewards/IngredientMatchReward/mean": 0.6385406732559205,
      "rewards/IngredientMatchReward/std": 0.2917583525180817,
      "rewards/IngredientQuantityMatchReward/mean": 0.7101723670959472,
      "rewards/IngredientQuantityMatchReward/std": 0.3938800930976868,
      "rewards/TotalKcalExactMatchReward/mean": 0.878125,
      "rewards/TotalKcalExactMatchReward/std": 0.3060554936528206,
      "step": 2135
    },
    {
      "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": 0.009375,
      "completions/max_length": 508.6,
      "completions/mean_length": 390.0828125,
      "completions/min_length": 251.4,
      "epoch": 0.5952712100139083,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6337137222290039,
      "kl": 0.05112361330538988,
      "learning_rate": 3.848992839231737e-07,
      "loss": 0.0020450735464692116,
      "reward": 3.1612887382507324,
      "reward_std": 0.41824568510055543,
      "rewards/IngredientFormatReward/mean": 0.9850520730018616,
      "rewards/IngredientFormatReward/std": 0.09320681095123291,
      "rewards/IngredientMatchReward/mean": 0.6647279262542725,
      "rewards/IngredientMatchReward/std": 0.27302170395851133,
      "rewards/IngredientQuantityMatchReward/mean": 0.6865087270736694,
      "rewards/IngredientQuantityMatchReward/std": 0.3848438024520874,
      "rewards/TotalKcalExactMatchReward/mean": 0.825,
      "rewards/TotalKcalExactMatchReward/std": 0.35974394381046293,
      "step": 2140
    },
    {
      "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": 0.00625,
      "completions/max_length": 507.4,
      "completions/mean_length": 388.5140625,
      "completions/min_length": 274.8,
      "epoch": 0.5966620305980529,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.556304395198822,
      "kl": 0.050310481898486616,
      "learning_rate": 3.826624282294875e-07,
      "loss": 0.002012613415718079,
      "reward": 3.023041009902954,
      "reward_std": 0.3705463230609894,
      "rewards/IngredientFormatReward/mean": 0.9921168088912964,
      "rewards/IngredientFormatReward/std": 0.06639720574021339,
      "rewards/IngredientMatchReward/mean": 0.6476860284805298,
      "rewards/IngredientMatchReward/std": 0.27975512146949766,
      "rewards/IngredientQuantityMatchReward/mean": 0.6629256725311279,
      "rewards/IngredientQuantityMatchReward/std": 0.3901920020580292,
      "rewards/TotalKcalExactMatchReward/mean": 0.7203125,
      "rewards/TotalKcalExactMatchReward/std": 0.4402440369129181,
      "step": 2145
    },
    {
      "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": 0.0078125,
      "completions/max_length": 505.0,
      "completions/mean_length": 385.0921875,
      "completions/min_length": 243.8,
      "epoch": 0.5980528511821975,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.597395122051239,
      "kl": 0.05190782949794084,
      "learning_rate": 3.804280550667356e-07,
      "loss": 0.0020766373723745345,
      "reward": 3.165365695953369,
      "reward_std": 0.35970969796180724,
      "rewards/IngredientFormatReward/mean": 0.9876413464546203,
      "rewards/IngredientFormatReward/std": 0.09138427898287774,
      "rewards/IngredientMatchReward/mean": 0.6722779870033264,
      "rewards/IngredientMatchReward/std": 0.266836553812027,
      "rewards/IngredientQuantityMatchReward/mean": 0.6866963863372803,
      "rewards/IngredientQuantityMatchReward/std": 0.3877449631690979,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.37181003093719484,
      "step": 2150
    },
    {
      "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": 0.0046875,
      "completions/max_length": 499.4,
      "completions/mean_length": 389.1734375,
      "completions/min_length": 274.8,
      "epoch": 0.5994436717663422,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6142936944961548,
      "kl": 0.049102706369012594,
      "learning_rate": 3.7819621170793034e-07,
      "loss": 0.0019642310217022896,
      "reward": 3.1915061473846436,
      "reward_std": 0.3775101125240326,
      "rewards/IngredientFormatReward/mean": 0.9940624952316284,
      "rewards/IngredientFormatReward/std": 0.04380592703819275,
      "rewards/IngredientMatchReward/mean": 0.6725663423538208,
      "rewards/IngredientMatchReward/std": 0.2846332907676697,
      "rewards/IngredientQuantityMatchReward/mean": 0.6967523574829102,
      "rewards/IngredientQuantityMatchReward/std": 0.4011437356472015,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3704190969467163,
      "step": 2155
    },
    {
      "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": 0.0109375,
      "completions/max_length": 507.8,
      "completions/mean_length": 387.859375,
      "completions/min_length": 271.0,
      "epoch": 0.6008344923504868,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.601314902305603,
      "kl": 0.050125482678413394,
      "learning_rate": 3.759669453725607e-07,
      "loss": 0.0020050119608640673,
      "reward": 3.091590166091919,
      "reward_std": 0.42018883228302,
      "rewards/IngredientFormatReward/mean": 0.9830989599227905,
      "rewards/IngredientFormatReward/std": 0.10198934972286225,
      "rewards/IngredientMatchReward/mean": 0.6450570344924926,
      "rewards/IngredientMatchReward/std": 0.2986615300178528,
      "rewards/IngredientQuantityMatchReward/mean": 0.6493716359138488,
      "rewards/IngredientQuantityMatchReward/std": 0.4103902757167816,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.37561099529266356,
      "step": 2160
    },
    {
      "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": 0.0046875,
      "completions/max_length": 497.6,
      "completions/mean_length": 387.8203125,
      "completions/min_length": 259.8,
      "epoch": 0.6022253129346314,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6496283411979675,
      "kl": 0.04889493566006422,
      "learning_rate": 3.7374030322559314e-07,
      "loss": 0.001956149935722351,
      "reward": 3.112612819671631,
      "reward_std": 0.3562726080417633,
      "rewards/IngredientFormatReward/mean": 0.9957849740982055,
      "rewards/IngredientFormatReward/std": 0.03390633724629879,
      "rewards/IngredientMatchReward/mean": 0.6423406958580017,
      "rewards/IngredientMatchReward/std": 0.2840432107448578,
      "rewards/IngredientQuantityMatchReward/mean": 0.6744871377944947,
      "rewards/IngredientQuantityMatchReward/std": 0.39336987137794494,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3893772900104523,
      "step": 2165
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.4,
      "completions/mean_length": 392.259375,
      "completions/min_length": 272.0,
      "epoch": 0.6036161335187761,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6597304344177246,
      "kl": 1.401045136200264,
      "learning_rate": 3.71516332376474e-07,
      "loss": 0.055953198671340944,
      "reward": 3.019261026382446,
      "reward_std": 0.4189970850944519,
      "rewards/IngredientFormatReward/mean": 0.977146577835083,
      "rewards/IngredientFormatReward/std": 0.11853427737951279,
      "rewards/IngredientMatchReward/mean": 0.6084200739860535,
      "rewards/IngredientMatchReward/std": 0.29643736481666566,
      "rewards/IngredientQuantityMatchReward/mean": 0.6790068507194519,
      "rewards/IngredientQuantityMatchReward/std": 0.40717952251434325,
      "rewards/TotalKcalExactMatchReward/mean": 0.7546875,
      "rewards/TotalKcalExactMatchReward/std": 0.4276392340660095,
      "step": 2170
    },
    {
      "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": 0.0140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.103125,
      "completions/min_length": 248.4,
      "epoch": 0.6050069541029207,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6310843229293823,
      "kl": 0.04990187431685626,
      "learning_rate": 3.692950798781318e-07,
      "loss": 0.001995875313878059,
      "reward": 3.0463210582733153,
      "reward_std": 0.39750961065292356,
      "rewards/IngredientFormatReward/mean": 0.9845368385314941,
      "rewards/IngredientFormatReward/std": 0.11547477096319199,
      "rewards/IngredientMatchReward/mean": 0.6253199338912964,
      "rewards/IngredientMatchReward/std": 0.2806160271167755,
      "rewards/IngredientQuantityMatchReward/mean": 0.6724017977714538,
      "rewards/IngredientQuantityMatchReward/std": 0.40632015466690063,
      "rewards/TotalKcalExactMatchReward/mean": 0.7640625,
      "rewards/TotalKcalExactMatchReward/std": 0.4197130024433136,
      "step": 2175
    },
    {
      "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": 0.0078125,
      "completions/max_length": 510.2,
      "completions/mean_length": 388.1296875,
      "completions/min_length": 266.0,
      "epoch": 0.6063977746870653,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.706554651260376,
      "kl": 0.052620567847043276,
      "learning_rate": 3.670765927259831e-07,
      "loss": 0.0021052807569503784,
      "reward": 3.153690004348755,
      "reward_std": 0.41479485630989077,
      "rewards/IngredientFormatReward/mean": 0.9915085434913635,
      "rewards/IngredientFormatReward/std": 0.06617667004466057,
      "rewards/IngredientMatchReward/mean": 0.6520244598388671,
      "rewards/IngredientMatchReward/std": 0.2850645899772644,
      "rewards/IngredientQuantityMatchReward/mean": 0.7070319652557373,
      "rewards/IngredientQuantityMatchReward/std": 0.4136449575424194,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.3873530626296997,
      "step": 2180
    },
    {
      "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": 0.009375,
      "completions/max_length": 510.6,
      "completions/mean_length": 397.1109375,
      "completions/min_length": 283.8,
      "epoch": 0.60778859527121,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6182182431221008,
      "kl": 0.05071941772475839,
      "learning_rate": 3.648609178569374e-07,
      "loss": 0.0020290974527597426,
      "reward": 3.0801242351531983,
      "reward_std": 0.4037054359912872,
      "rewards/IngredientFormatReward/mean": 0.9916145920753479,
      "rewards/IngredientFormatReward/std": 0.06768649965524673,
      "rewards/IngredientMatchReward/mean": 0.6478013277053833,
      "rewards/IngredientMatchReward/std": 0.29819480180740354,
      "rewards/IngredientQuantityMatchReward/mean": 0.684458315372467,
      "rewards/IngredientQuantityMatchReward/std": 0.38002619743347166,
      "rewards/TotalKcalExactMatchReward/mean": 0.75625,
      "rewards/TotalKcalExactMatchReward/std": 0.41520002484321594,
      "step": 2185
    },
    {
      "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": 0.0015625,
      "completions/max_length": 498.8,
      "completions/mean_length": 386.0421875,
      "completions/min_length": 272.4,
      "epoch": 0.6091794158553546,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6383892893791199,
      "kl": 0.04978294498287141,
      "learning_rate": 3.626481021484045e-07,
      "loss": 0.0019914690405130387,
      "reward": 3.1687289237976075,
      "reward_std": 0.34659733176231383,
      "rewards/IngredientFormatReward/mean": 0.9972395896911621,
      "rewards/IngredientFormatReward/std": 0.02697129435837269,
      "rewards/IngredientMatchReward/mean": 0.6780971050262451,
      "rewards/IngredientMatchReward/std": 0.2776097983121872,
      "rewards/IngredientQuantityMatchReward/mean": 0.710579776763916,
      "rewards/IngredientQuantityMatchReward/std": 0.3856403946876526,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4049029767513275,
      "step": 2190
    },
    {
      "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": 0.0109375,
      "completions/max_length": 502.8,
      "completions/mean_length": 390.121875,
      "completions/min_length": 266.0,
      "epoch": 0.6105702364394993,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.650026798248291,
      "kl": 0.05147871123626828,
      "learning_rate": 3.6043819241730145e-07,
      "loss": 0.0020590342581272127,
      "reward": 3.158277082443237,
      "reward_std": 0.3794547736644745,
      "rewards/IngredientFormatReward/mean": 0.9897916555404663,
      "rewards/IngredientFormatReward/std": 0.0673672953620553,
      "rewards/IngredientMatchReward/mean": 0.6765773892402649,
      "rewards/IngredientMatchReward/std": 0.280401399731636,
      "rewards/IngredientQuantityMatchReward/mean": 0.660658085346222,
      "rewards/IngredientQuantityMatchReward/std": 0.406155651807785,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.3482891798019409,
      "step": 2195
    },
    {
      "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": 0.0109375,
      "completions/max_length": 508.4,
      "completions/mean_length": 392.6640625,
      "completions/min_length": 270.2,
      "epoch": 0.6119610570236439,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6922538876533508,
      "kl": 0.051177499582991004,
      "learning_rate": 3.5823123541906424e-07,
      "loss": 0.00204671174287796,
      "reward": 3.038310670852661,
      "reward_std": 0.4538100779056549,
      "rewards/IngredientFormatReward/mean": 0.9873809576034546,
      "rewards/IngredientFormatReward/std": 0.10423120856285095,
      "rewards/IngredientMatchReward/mean": 0.6075683832168579,
      "rewards/IngredientMatchReward/std": 0.2966679573059082,
      "rewards/IngredientQuantityMatchReward/mean": 0.6402364134788513,
      "rewards/IngredientQuantityMatchReward/std": 0.4361319363117218,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.39142870903015137,
      "step": 2200
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.6,
      "completions/mean_length": 394.4796875,
      "completions/min_length": 276.0,
      "epoch": 0.6133518776077886,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6373947262763977,
      "kl": 0.04903403371572494,
      "learning_rate": 3.560272778466569e-07,
      "loss": 0.001961347088217735,
      "reward": 3.1263299942016602,
      "reward_std": 0.38947783708572387,
      "rewards/IngredientFormatReward/mean": 0.9933854103088379,
      "rewards/IngredientFormatReward/std": 0.04924058765172958,
      "rewards/IngredientMatchReward/mean": 0.6449187755584717,
      "rewards/IngredientMatchReward/std": 0.27405828833580015,
      "rewards/IngredientQuantityMatchReward/mean": 0.6349007368087769,
      "rewards/IngredientQuantityMatchReward/std": 0.41207821369171144,
      "rewards/TotalKcalExactMatchReward/mean": 0.853125,
      "rewards/TotalKcalExactMatchReward/std": 0.35279104113578796,
      "step": 2205
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.2,
      "completions/mean_length": 389.8140625,
      "completions/min_length": 267.2,
      "epoch": 0.6147426981919333,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6941333413124084,
      "kl": 0.20643644286319612,
      "learning_rate": 3.5382636632958405e-07,
      "loss": 0.008257143199443817,
      "reward": 3.016005277633667,
      "reward_std": 0.4019562005996704,
      "rewards/IngredientFormatReward/mean": 0.9874702215194702,
      "rewards/IngredientFormatReward/std": 0.08696230705827475,
      "rewards/IngredientMatchReward/mean": 0.6278738975524902,
      "rewards/IngredientMatchReward/std": 0.293187814950943,
      "rewards/IngredientQuantityMatchReward/mean": 0.6319111466407776,
      "rewards/IngredientQuantityMatchReward/std": 0.43076640367507935,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.41989484429359436,
      "step": 2210
    },
    {
      "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": 0.00625,
      "completions/max_length": 508.6,
      "completions/mean_length": 393.309375,
      "completions/min_length": 265.8,
      "epoch": 0.6161335187760779,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6270684003829956,
      "kl": 0.04865258978679776,
      "learning_rate": 3.516285474329044e-07,
      "loss": 0.0019460489973425865,
      "reward": 3.1262417793273927,
      "reward_std": 0.3906321346759796,
      "rewards/IngredientFormatReward/mean": 0.99453125,
      "rewards/IngredientFormatReward/std": 0.05367666333913803,
      "rewards/IngredientMatchReward/mean": 0.6526476263999939,
      "rewards/IngredientMatchReward/std": 0.27952222228050233,
      "rewards/IngredientQuantityMatchReward/mean": 0.6915629744529724,
      "rewards/IngredientQuantityMatchReward/std": 0.400864040851593,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.3981053829193115,
      "step": 2215
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.8,
      "completions/mean_length": 395.0640625,
      "completions/min_length": 248.2,
      "epoch": 0.6175243393602226,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.5712026953697205,
      "kl": 0.05006823530420661,
      "learning_rate": 3.4943386765624563e-07,
      "loss": 0.002002709172666073,
      "reward": 3.077920436859131,
      "reward_std": 0.392623233795166,
      "rewards/IngredientFormatReward/mean": 0.9903906106948852,
      "rewards/IngredientFormatReward/std": 0.07571376450359821,
      "rewards/IngredientMatchReward/mean": 0.647243320941925,
      "rewards/IngredientMatchReward/std": 0.284138685464859,
      "rewards/IngredientQuantityMatchReward/mean": 0.6262240290641785,
      "rewards/IngredientQuantityMatchReward/std": 0.4356621503829956,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3902096688747406,
      "step": 2220
    },
    {
      "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": 0.003125,
      "completions/max_length": 497.8,
      "completions/mean_length": 387.2,
      "completions/min_length": 234.8,
      "epoch": 0.6189151599443672,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.6324188113212585,
      "kl": 0.046689180727116766,
      "learning_rate": 3.472423734328204e-07,
      "loss": 0.001867634803056717,
      "reward": 3.09603590965271,
      "reward_std": 0.35402976274490355,
      "rewards/IngredientFormatReward/mean": 0.9944270849227905,
      "rewards/IngredientFormatReward/std": 0.05454032719135284,
      "rewards/IngredientMatchReward/mean": 0.6432204723358155,
      "rewards/IngredientMatchReward/std": 0.28954875469207764,
      "rewards/IngredientQuantityMatchReward/mean": 0.6849508166313172,
      "rewards/IngredientQuantityMatchReward/std": 0.3964232563972473,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.4028464019298553,
      "step": 2225
    },
    {
      "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": 0.0234375,
      "completions/max_length": 506.8,
      "completions/mean_length": 384.578125,
      "completions/min_length": 252.4,
      "epoch": 0.6203059805285118,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6558968424797058,
      "kl": 0.053798887552693485,
      "learning_rate": 3.450541111284447e-07,
      "loss": 0.0021519482135772703,
      "reward": 3.114434099197388,
      "reward_std": 0.43681026697158815,
      "rewards/IngredientFormatReward/mean": 0.9728868842124939,
      "rewards/IngredientFormatReward/std": 0.12590492386370897,
      "rewards/IngredientMatchReward/mean": 0.6601122260093689,
      "rewards/IngredientMatchReward/std": 0.28873300552368164,
      "rewards/IngredientQuantityMatchReward/mean": 0.7111225247383117,
      "rewards/IngredientQuantityMatchReward/std": 0.41141672134399415,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.38224449157714846,
      "step": 2230
    },
    {
      "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": 0.0125,
      "completions/max_length": 513.0,
      "completions/mean_length": 391.4203125,
      "completions/min_length": 262.8,
      "epoch": 0.6216968011126565,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6426628828048706,
      "kl": 0.0496185167459771,
      "learning_rate": 3.4286912704055505e-07,
      "loss": 0.00198509581387043,
      "reward": 3.095783233642578,
      "reward_std": 0.39932904839515687,
      "rewards/IngredientFormatReward/mean": 0.9864955544471741,
      "rewards/IngredientFormatReward/std": 0.10071748532354832,
      "rewards/IngredientMatchReward/mean": 0.6558928847312927,
      "rewards/IngredientMatchReward/std": 0.2895587205886841,
      "rewards/IngredientQuantityMatchReward/mean": 0.6299573540687561,
      "rewards/IngredientQuantityMatchReward/std": 0.4130510151386261,
      "rewards/TotalKcalExactMatchReward/mean": 0.8234375,
      "rewards/TotalKcalExactMatchReward/std": 0.3680810809135437,
      "step": 2235
    },
    {
      "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": 0.0171875,
      "completions/max_length": 499.0,
      "completions/mean_length": 393.0328125,
      "completions/min_length": 260.4,
      "epoch": 0.6230876216968011,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6363694071769714,
      "kl": 0.04835720381233841,
      "learning_rate": 3.406874673972312e-07,
      "loss": 0.0019343402236700058,
      "reward": 3.0554323196411133,
      "reward_std": 0.39848679304122925,
      "rewards/IngredientFormatReward/mean": 0.9841145753860474,
      "rewards/IngredientFormatReward/std": 0.08837253898382187,
      "rewards/IngredientMatchReward/mean": 0.6605754017829895,
      "rewards/IngredientMatchReward/std": 0.2916788816452026,
      "rewards/IngredientQuantityMatchReward/mean": 0.6326173186302185,
      "rewards/IngredientQuantityMatchReward/std": 0.40582574605941774,
      "rewards/TotalKcalExactMatchReward/mean": 0.778125,
      "rewards/TotalKcalExactMatchReward/std": 0.3950036108493805,
      "step": 2240
    },
    {
      "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": 0.0171875,
      "completions/max_length": 506.2,
      "completions/mean_length": 394.8921875,
      "completions/min_length": 286.6,
      "epoch": 0.6244784422809457,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5934588313102722,
      "kl": 0.05474424082785845,
      "learning_rate": 3.385091783562168e-07,
      "loss": 0.002190130949020386,
      "reward": 3.085518455505371,
      "reward_std": 0.3905796468257904,
      "rewards/IngredientFormatReward/mean": 0.9823958396911621,
      "rewards/IngredientFormatReward/std": 0.11776968874037266,
      "rewards/IngredientMatchReward/mean": 0.6468719124794007,
      "rewards/IngredientMatchReward/std": 0.29382801055908203,
      "rewards/IngredientQuantityMatchReward/mean": 0.6828132152557373,
      "rewards/IngredientQuantityMatchReward/std": 0.4147923529148102,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.4130185544490814,
      "step": 2245
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.2,
      "completions/mean_length": 391.8046875,
      "completions/min_length": 259.8,
      "epoch": 0.6258692628650904,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7126057744026184,
      "kl": 0.051664074324071406,
      "learning_rate": 3.363343060039431e-07,
      "loss": 0.0020667586475610735,
      "reward": 2.926431131362915,
      "reward_std": 0.4142996668815613,
      "rewards/IngredientFormatReward/mean": 0.9811979055404663,
      "rewards/IngredientFormatReward/std": 0.12898623794317246,
      "rewards/IngredientMatchReward/mean": 0.621659231185913,
      "rewards/IngredientMatchReward/std": 0.29404624104499816,
      "rewards/IngredientQuantityMatchReward/mean": 0.5407614946365357,
      "rewards/IngredientQuantityMatchReward/std": 0.43060929179191587,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3951663553714752,
      "step": 2250
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.2,
      "completions/mean_length": 386.009375,
      "completions/min_length": 257.4,
      "epoch": 0.627260083449235,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.6304970383644104,
      "kl": 0.049507897021248935,
      "learning_rate": 3.3416289635455364e-07,
      "loss": 0.001980157569050789,
      "reward": 3.180800724029541,
      "reward_std": 0.40125504732131956,
      "rewards/IngredientFormatReward/mean": 0.9903125047683716,
      "rewards/IngredientFormatReward/std": 0.08373235017061234,
      "rewards/IngredientMatchReward/mean": 0.6888733983039856,
      "rewards/IngredientMatchReward/std": 0.2980964004993439,
      "rewards/IngredientQuantityMatchReward/mean": 0.6859898090362548,
      "rewards/IngredientQuantityMatchReward/std": 0.4196448683738708,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.3790489435195923,
      "step": 2255
    },
    {
      "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": 0.00625,
      "completions/max_length": 493.0,
      "completions/mean_length": 389.01875,
      "completions/min_length": 274.0,
      "epoch": 0.6286509040333796,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.5745730996131897,
      "kl": 0.052322480361908674,
      "learning_rate": 3.3199499534893126e-07,
      "loss": 0.0020928554236888885,
      "reward": 3.0668933391571045,
      "reward_std": 0.37459617853164673,
      "rewards/IngredientFormatReward/mean": 0.9922395944595337,
      "rewards/IngredientFormatReward/std": 0.054567881673574445,
      "rewards/IngredientMatchReward/mean": 0.672517991065979,
      "rewards/IngredientMatchReward/std": 0.2672812879085541,
      "rewards/IngredientQuantityMatchReward/mean": 0.6583857655525207,
      "rewards/IngredientQuantityMatchReward/std": 0.4205239474773407,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.43603759407997134,
      "step": 2260
    },
    {
      "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": 0.0046875,
      "completions/max_length": 500.0,
      "completions/mean_length": 394.8109375,
      "completions/min_length": 271.0,
      "epoch": 0.6300417246175244,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6598750352859497,
      "kl": 0.06744713457301259,
      "learning_rate": 3.298306488537257e-07,
      "loss": 0.0026970580220222473,
      "reward": 3.1390488147735596,
      "reward_std": 0.36806185245513917,
      "rewards/IngredientFormatReward/mean": 0.9928125143051147,
      "rewards/IngredientFormatReward/std": 0.06443778797984123,
      "rewards/IngredientMatchReward/mean": 0.6801209092140198,
      "rewards/IngredientMatchReward/std": 0.26344968378543854,
      "rewards/IngredientQuantityMatchReward/mean": 0.659865391254425,
      "rewards/IngredientQuantityMatchReward/std": 0.40859387516975404,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3911212682723999,
      "step": 2265
    },
    {
      "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": 0.00625,
      "completions/max_length": 501.2,
      "completions/mean_length": 391.24375,
      "completions/min_length": 270.8,
      "epoch": 0.631432545201669,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7604984641075134,
      "kl": 0.06415583626367152,
      "learning_rate": 3.276699026603835e-07,
      "loss": 0.00256730355322361,
      "reward": 3.187400960922241,
      "reward_std": 0.38949325680732727,
      "rewards/IngredientFormatReward/mean": 0.9868749976158142,
      "rewards/IngredientFormatReward/std": 0.09232946932315826,
      "rewards/IngredientMatchReward/mean": 0.6554042816162109,
      "rewards/IngredientMatchReward/std": 0.2981587827205658,
      "rewards/IngredientQuantityMatchReward/mean": 0.6841842651367187,
      "rewards/IngredientQuantityMatchReward/std": 0.3940977334976196,
      "rewards/TotalKcalExactMatchReward/mean": 0.8609375,
      "rewards/TotalKcalExactMatchReward/std": 0.3312438249588013,
      "step": 2270
    },
    {
      "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": 0.0125,
      "completions/max_length": 506.4,
      "completions/mean_length": 397.3140625,
      "completions/min_length": 277.4,
      "epoch": 0.6328233657858137,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6403272747993469,
      "kl": 0.0502887899056077,
      "learning_rate": 3.2551280248417855e-07,
      "loss": 0.0020116424188017846,
      "reward": 3.098349761962891,
      "reward_std": 0.4067712306976318,
      "rewards/IngredientFormatReward/mean": 0.9885267853736878,
      "rewards/IngredientFormatReward/std": 0.09110912680625916,
      "rewards/IngredientMatchReward/mean": 0.6192803025245667,
      "rewards/IngredientMatchReward/std": 0.28211758434772494,
      "rewards/IngredientQuantityMatchReward/mean": 0.6592926383018494,
      "rewards/IngredientQuantityMatchReward/std": 0.4010209202766418,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.3651015400886536,
      "step": 2275
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.2,
      "completions/mean_length": 395.259375,
      "completions/min_length": 273.6,
      "epoch": 0.6342141863699583,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6547976732254028,
      "kl": 0.04664964359253645,
      "learning_rate": 3.2335939396324574e-07,
      "loss": 0.001865985430777073,
      "reward": 3.156012487411499,
      "reward_std": 0.36845679879188536,
      "rewards/IngredientFormatReward/mean": 0.9910937547683716,
      "rewards/IngredientFormatReward/std": 0.07213107720017434,
      "rewards/IngredientMatchReward/mean": 0.673369300365448,
      "rewards/IngredientMatchReward/std": 0.29961636662483215,
      "rewards/IngredientQuantityMatchReward/mean": 0.6462370038032532,
      "rewards/IngredientQuantityMatchReward/std": 0.4035289347171783,
      "rewards/TotalKcalExactMatchReward/mean": 0.8453125,
      "rewards/TotalKcalExactMatchReward/std": 0.33718748241662977,
      "step": 2280
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.4,
      "completions/mean_length": 402.1625,
      "completions/min_length": 282.4,
      "epoch": 0.6356050069541029,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6613708734512329,
      "kl": 0.04664005022495985,
      "learning_rate": 3.21209722657615e-07,
      "loss": 0.0018660373985767364,
      "reward": 3.0435181617736817,
      "reward_std": 0.3617978811264038,
      "rewards/IngredientFormatReward/mean": 0.9830654621124267,
      "rewards/IngredientFormatReward/std": 0.09950861930847169,
      "rewards/IngredientMatchReward/mean": 0.6392280459403992,
      "rewards/IngredientMatchReward/std": 0.2842461735010147,
      "rewards/IngredientQuantityMatchReward/mean": 0.6149746000766754,
      "rewards/IngredientQuantityMatchReward/std": 0.4098484456539154,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.384172785282135,
      "step": 2285
    },
    {
      "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": 0.003125,
      "completions/max_length": 504.2,
      "completions/mean_length": 388.5296875,
      "completions/min_length": 260.8,
      "epoch": 0.6369958275382476,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.5699143409729004,
      "kl": 0.04985083430074155,
      "learning_rate": 3.1906383404824735e-07,
      "loss": 0.001994183287024498,
      "reward": 3.0484458923339846,
      "reward_std": 0.372654390335083,
      "rewards/IngredientFormatReward/mean": 0.9962500095367431,
      "rewards/IngredientFormatReward/std": 0.03921364024281502,
      "rewards/IngredientMatchReward/mean": 0.6028466105461121,
      "rewards/IngredientMatchReward/std": 0.28491021394729615,
      "rewards/IngredientQuantityMatchReward/mean": 0.6618492722511291,
      "rewards/IngredientQuantityMatchReward/std": 0.41888644695281985,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.3779851108789444,
      "step": 2290
    },
    {
      "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": 0.0109375,
      "completions/max_length": 507.6,
      "completions/mean_length": 394.1171875,
      "completions/min_length": 288.0,
      "epoch": 0.6383866481223922,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5812567472457886,
      "kl": 0.048090280825272204,
      "learning_rate": 3.169217735360721e-07,
      "loss": 0.001923968270421028,
      "reward": 3.0279534339904783,
      "reward_std": 0.3708513677120209,
      "rewards/IngredientFormatReward/mean": 0.9885416507720948,
      "rewards/IngredientFormatReward/std": 0.09087646007537842,
      "rewards/IngredientMatchReward/mean": 0.6266480684280396,
      "rewards/IngredientMatchReward/std": 0.2966554284095764,
      "rewards/IngredientQuantityMatchReward/mean": 0.6346386492252349,
      "rewards/IngredientQuantityMatchReward/std": 0.41230267882347105,
      "rewards/TotalKcalExactMatchReward/mean": 0.778125,
      "rewards/TotalKcalExactMatchReward/std": 0.4041856527328491,
      "step": 2295
    },
    {
      "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": 0.0046875,
      "completions/max_length": 499.6,
      "completions/mean_length": 385.1796875,
      "completions/min_length": 244.2,
      "epoch": 0.6397774687065368,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6119884252548218,
      "kl": 0.05169716072268784,
      "learning_rate": 3.147835864410276e-07,
      "loss": 0.002067873626947403,
      "reward": 3.1669593334197996,
      "reward_std": 0.4164513349533081,
      "rewards/IngredientFormatReward/mean": 0.985141372680664,
      "rewards/IngredientFormatReward/std": 0.10293094590306281,
      "rewards/IngredientMatchReward/mean": 0.6206455707550049,
      "rewards/IngredientMatchReward/std": 0.29413315653800964,
      "rewards/IngredientQuantityMatchReward/mean": 0.7252350091934204,
      "rewards/IngredientQuantityMatchReward/std": 0.39733303189277647,
      "rewards/TotalKcalExactMatchReward/mean": 0.8359375,
      "rewards/TotalKcalExactMatchReward/std": 0.35827775597572326,
      "step": 2300
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.4,
      "completions/mean_length": 392.31875,
      "completions/min_length": 252.2,
      "epoch": 0.6411682892906815,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6846728920936584,
      "kl": 0.04950613779947162,
      "learning_rate": 3.126493180011014e-07,
      "loss": 0.001980242505669594,
      "reward": 3.1633121967315674,
      "reward_std": 0.35973186790943146,
      "rewards/IngredientFormatReward/mean": 0.9850520730018616,
      "rewards/IngredientFormatReward/std": 0.10152259021997452,
      "rewards/IngredientMatchReward/mean": 0.6418793201446533,
      "rewards/IngredientMatchReward/std": 0.27253933548927306,
      "rewards/IngredientQuantityMatchReward/mean": 0.7113808035850525,
      "rewards/IngredientQuantityMatchReward/std": 0.40159483551979064,
      "rewards/TotalKcalExactMatchReward/mean": 0.825,
      "rewards/TotalKcalExactMatchReward/std": 0.36822456419467925,
      "step": 2305
    },
    {
      "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": 0.0203125,
      "completions/max_length": 508.2,
      "completions/mean_length": 397.2546875,
      "completions/min_length": 294.4,
      "epoch": 0.6425591098748261,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6922708749771118,
      "kl": 0.053410388017073276,
      "learning_rate": 3.105190133713733e-07,
      "loss": 0.002136606350541115,
      "reward": 3.185721254348755,
      "reward_std": 0.43321176767349245,
      "rewards/IngredientFormatReward/mean": 0.9797656297683716,
      "rewards/IngredientFormatReward/std": 0.11258613169193268,
      "rewards/IngredientMatchReward/mean": 0.6650626420974731,
      "rewards/IngredientMatchReward/std": 0.28691638112068174,
      "rewards/IngredientQuantityMatchReward/mean": 0.7252679705619812,
      "rewards/IngredientQuantityMatchReward/std": 0.37057950496673586,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.38518795371055603,
      "step": 2310
    },
    {
      "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": 0.0171875,
      "completions/max_length": 510.6,
      "completions/mean_length": 394.528125,
      "completions/min_length": 274.8,
      "epoch": 0.6439499304589708,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6853731274604797,
      "kl": 0.04699210901744664,
      "learning_rate": 3.0839271762306005e-07,
      "loss": 0.00188013669103384,
      "reward": 3.0963015079498293,
      "reward_std": 0.37388708591461184,
      "rewards/IngredientFormatReward/mean": 0.985156238079071,
      "rewards/IngredientFormatReward/std": 0.09949494898319244,
      "rewards/IngredientMatchReward/mean": 0.6302870869636535,
      "rewards/IngredientMatchReward/std": 0.27550233602523805,
      "rewards/IngredientQuantityMatchReward/mean": 0.6964831590652466,
      "rewards/IngredientQuantityMatchReward/std": 0.4114985764026642,
      "rewards/TotalKcalExactMatchReward/mean": 0.784375,
      "rewards/TotalKcalExactMatchReward/std": 0.41028077006340025,
      "step": 2315
    },
    {
      "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": 497.4,
      "completions/mean_length": 389.7984375,
      "completions/min_length": 265.0,
      "epoch": 0.6453407510431154,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6921623349189758,
      "kl": 0.046600739704445004,
      "learning_rate": 3.0627047574256216e-07,
      "loss": 0.0018642637878656387,
      "reward": 3.076311540603638,
      "reward_std": 0.3626805365085602,
      "rewards/IngredientFormatReward/mean": 0.996666669845581,
      "rewards/IngredientFormatReward/std": 0.02143365629017353,
      "rewards/IngredientMatchReward/mean": 0.6558581352233886,
      "rewards/IngredientMatchReward/std": 0.28329201638698576,
      "rewards/IngredientQuantityMatchReward/mean": 0.6050367593765259,
      "rewards/IngredientQuantityMatchReward/std": 0.4378706097602844,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.3791636288166046,
      "step": 2320
    },
    {
      "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": 0.0203125,
      "completions/max_length": 508.6,
      "completions/mean_length": 396.7640625,
      "completions/min_length": 269.4,
      "epoch": 0.6467315716272601,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.5751472115516663,
      "kl": 0.04943703303579241,
      "learning_rate": 3.041523326305112e-07,
      "loss": 0.001977463066577911,
      "reward": 3.194409990310669,
      "reward_std": 0.4305049300193787,
      "rewards/IngredientFormatReward/mean": 0.9797916531562805,
      "rewards/IngredientFormatReward/std": 0.12461420521140099,
      "rewards/IngredientMatchReward/mean": 0.6630220651626587,
      "rewards/IngredientMatchReward/std": 0.2750160127878189,
      "rewards/IngredientQuantityMatchReward/mean": 0.6890962600708008,
      "rewards/IngredientQuantityMatchReward/std": 0.385354483127594,
      "rewards/TotalKcalExactMatchReward/mean": 0.8625,
      "rewards/TotalKcalExactMatchReward/std": 0.3144735336303711,
      "step": 2325
    },
    {
      "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": 0.0203125,
      "completions/max_length": 513.0,
      "completions/mean_length": 393.4203125,
      "completions/min_length": 266.4,
      "epoch": 0.6481223922114048,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6151979565620422,
      "kl": 0.047250632150098684,
      "learning_rate": 3.020383331008212e-07,
      "loss": 0.0018899183720350265,
      "reward": 2.952303075790405,
      "reward_std": 0.4416388809680939,
      "rewards/IngredientFormatReward/mean": 0.9795163750648499,
      "rewards/IngredientFormatReward/std": 0.13366409242153168,
      "rewards/IngredientMatchReward/mean": 0.6384114503860474,
      "rewards/IngredientMatchReward/std": 0.29739701747894287,
      "rewards/IngredientQuantityMatchReward/mean": 0.5906252264976501,
      "rewards/IngredientQuantityMatchReward/std": 0.4250226616859436,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.40911481976509095,
      "step": 2330
    },
    {
      "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": 0.0109375,
      "completions/max_length": 507.4,
      "completions/mean_length": 392.5203125,
      "completions/min_length": 262.2,
      "epoch": 0.6495132127955494,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6049847602844238,
      "kl": 0.04755361638963222,
      "learning_rate": 2.9992852187973874e-07,
      "loss": 0.001902458816766739,
      "reward": 3.0733123302459715,
      "reward_std": 0.3850603997707367,
      "rewards/IngredientFormatReward/mean": 0.9869791746139527,
      "rewards/IngredientFormatReward/std": 0.08390166461467743,
      "rewards/IngredientMatchReward/mean": 0.610430920124054,
      "rewards/IngredientMatchReward/std": 0.29726721048355104,
      "rewards/IngredientQuantityMatchReward/mean": 0.641527247428894,
      "rewards/IngredientQuantityMatchReward/std": 0.4174162566661835,
      "rewards/TotalKcalExactMatchReward/mean": 0.834375,
      "rewards/TotalKcalExactMatchReward/std": 0.3625717520713806,
      "step": 2335
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.8,
      "completions/mean_length": 391.6453125,
      "completions/min_length": 270.2,
      "epoch": 0.650904033379694,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7412999272346497,
      "kl": 0.05683932460378856,
      "learning_rate": 2.9782294360489824e-07,
      "loss": 0.0022735811769962313,
      "reward": 3.0667440414428713,
      "reward_std": 0.4293572545051575,
      "rewards/IngredientFormatReward/mean": 0.9919270753860474,
      "rewards/IngredientFormatReward/std": 0.07807534784078599,
      "rewards/IngredientMatchReward/mean": 0.6268408894538879,
      "rewards/IngredientMatchReward/std": 0.29498333036899566,
      "rewards/IngredientQuantityMatchReward/mean": 0.6745386362075806,
      "rewards/IngredientQuantityMatchReward/std": 0.4114294946193695,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.4150561511516571,
      "step": 2340
    },
    {
      "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": 0.00625,
      "completions/max_length": 505.8,
      "completions/mean_length": 394.3515625,
      "completions/min_length": 272.6,
      "epoch": 0.6522948539638387,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6499749422073364,
      "kl": 0.04816399768460542,
      "learning_rate": 2.957216428243772e-07,
      "loss": 0.001926695369184017,
      "reward": 2.9874334812164305,
      "reward_std": 0.44137861132621764,
      "rewards/IngredientFormatReward/mean": 0.9909077405929565,
      "rewards/IngredientFormatReward/std": 0.07433488909155131,
      "rewards/IngredientMatchReward/mean": 0.6615043878555298,
      "rewards/IngredientMatchReward/std": 0.2838159501552582,
      "rewards/IngredientQuantityMatchReward/mean": 0.6256462693214416,
      "rewards/IngredientQuantityMatchReward/std": 0.43513540625572206,
      "rewards/TotalKcalExactMatchReward/mean": 0.709375,
      "rewards/TotalKcalExactMatchReward/std": 0.4434183180332184,
      "step": 2345
    },
    {
      "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": 0.00625,
      "completions/max_length": 505.0,
      "completions/mean_length": 392.26875,
      "completions/min_length": 270.8,
      "epoch": 0.6536856745479833,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6249247193336487,
      "kl": 0.04946722569875419,
      "learning_rate": 2.936246639957529e-07,
      "loss": 0.001978399232029915,
      "reward": 3.093481254577637,
      "reward_std": 0.4033039391040802,
      "rewards/IngredientFormatReward/mean": 0.9912500023841858,
      "rewards/IngredientFormatReward/std": 0.06858442425727844,
      "rewards/IngredientMatchReward/mean": 0.6428137421607971,
      "rewards/IngredientMatchReward/std": 0.26454198360443115,
      "rewards/IngredientQuantityMatchReward/mean": 0.6312925815582275,
      "rewards/IngredientQuantityMatchReward/std": 0.4156554341316223,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.37494794726371766,
      "step": 2350
    },
    {
      "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": 0.0140625,
      "completions/max_length": 503.4,
      "completions/mean_length": 389.2296875,
      "completions/min_length": 238.6,
      "epoch": 0.655076495132128,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6602175235748291,
      "kl": 0.050524069415405395,
      "learning_rate": 2.915320514851627e-07,
      "loss": 0.002020845375955105,
      "reward": 3.121209001541138,
      "reward_std": 0.4390512526035309,
      "rewards/IngredientFormatReward/mean": 0.9868749976158142,
      "rewards/IngredientFormatReward/std": 0.08971829488873481,
      "rewards/IngredientMatchReward/mean": 0.6932979822158813,
      "rewards/IngredientMatchReward/std": 0.27488317489624026,
      "rewards/IngredientQuantityMatchReward/mean": 0.6410360217094422,
      "rewards/IngredientQuantityMatchReward/std": 0.41615588665008546,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3990713179111481,
      "step": 2355
    },
    {
      "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": 0.0015625,
      "completions/max_length": 495.8,
      "completions/mean_length": 393.64375,
      "completions/min_length": 273.0,
      "epoch": 0.6564673157162726,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5934226512908936,
      "kl": 0.04695138649549335,
      "learning_rate": 2.8944384956636436e-07,
      "loss": 0.0018780270591378212,
      "reward": 3.1596553325653076,
      "reward_std": 0.3678014278411865,
      "rewards/IngredientFormatReward/mean": 0.9976041674613952,
      "rewards/IngredientFormatReward/std": 0.02309950739145279,
      "rewards/IngredientMatchReward/mean": 0.6859988927841186,
      "rewards/IngredientMatchReward/std": 0.2769916921854019,
      "rewards/IngredientQuantityMatchReward/mean": 0.6854272723197937,
      "rewards/IngredientQuantityMatchReward/std": 0.40260618925094604,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.38472620248794553,
      "step": 2360
    },
    {
      "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": 0.00625,
      "completions/max_length": 505.8,
      "completions/mean_length": 386.9125,
      "completions/min_length": 261.0,
      "epoch": 0.6578581363004172,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6448837518692017,
      "kl": 0.04659861430991441,
      "learning_rate": 2.873601024198011e-07,
      "loss": 0.001864098198711872,
      "reward": 3.1337858200073243,
      "reward_std": 0.336954790353775,
      "rewards/IngredientFormatReward/mean": 0.99375,
      "rewards/IngredientFormatReward/std": 0.06025671809911728,
      "rewards/IngredientMatchReward/mean": 0.6501909971237183,
      "rewards/IngredientMatchReward/std": 0.29020614326000216,
      "rewards/IngredientQuantityMatchReward/mean": 0.6664073586463928,
      "rewards/IngredientQuantityMatchReward/std": 0.41391525864601136,
      "rewards/TotalKcalExactMatchReward/mean": 0.8234375,
      "rewards/TotalKcalExactMatchReward/std": 0.38152927756309507,
      "step": 2365
    },
    {
      "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": 0.0109375,
      "completions/max_length": 505.6,
      "completions/mean_length": 389.090625,
      "completions/min_length": 271.4,
      "epoch": 0.6592489568845619,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6337307691574097,
      "kl": 0.10345027199946344,
      "learning_rate": 2.8528085413166524e-07,
      "loss": 0.004148460924625397,
      "reward": 3.137276220321655,
      "reward_std": 0.42400845885276794,
      "rewards/IngredientFormatReward/mean": 0.9884374976158142,
      "rewards/IngredientFormatReward/std": 0.09104880094528198,
      "rewards/IngredientMatchReward/mean": 0.6613380670547485,
      "rewards/IngredientMatchReward/std": 0.2888974130153656,
      "rewards/IngredientQuantityMatchReward/mean": 0.6265631735324859,
      "rewards/IngredientQuantityMatchReward/std": 0.41725427508354185,
      "rewards/TotalKcalExactMatchReward/mean": 0.8609375,
      "rewards/TotalKcalExactMatchReward/std": 0.3248093917965889,
      "step": 2370
    },
    {
      "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": 0.015625,
      "completions/max_length": 507.2,
      "completions/mean_length": 393.634375,
      "completions/min_length": 273.2,
      "epoch": 0.6606397774687065,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6260727047920227,
      "kl": 0.04925237852148712,
      "learning_rate": 2.832061486929656e-07,
      "loss": 0.0019701194018125535,
      "reward": 3.1139048099517823,
      "reward_std": 0.41956409215927126,
      "rewards/IngredientFormatReward/mean": 0.9828125,
      "rewards/IngredientFormatReward/std": 0.09825026765465736,
      "rewards/IngredientMatchReward/mean": 0.6576175332069397,
      "rewards/IngredientMatchReward/std": 0.3032703876495361,
      "rewards/IngredientQuantityMatchReward/mean": 0.6437873005867004,
      "rewards/IngredientQuantityMatchReward/std": 0.43343395590782163,
      "rewards/TotalKcalExactMatchReward/mean": 0.8296875,
      "rewards/TotalKcalExactMatchReward/std": 0.3699494957923889,
      "step": 2375
    },
    {
      "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": 0.015625,
      "completions/max_length": 510.0,
      "completions/mean_length": 393.4859375,
      "completions/min_length": 253.0,
      "epoch": 0.6620305980528511,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.6546136736869812,
      "kl": 0.04857088509015739,
      "learning_rate": 2.8113602999859756e-07,
      "loss": 0.0019431136548519134,
      "reward": 3.0654694080352782,
      "reward_std": 0.38434099555015566,
      "rewards/IngredientFormatReward/mean": 0.984375,
      "rewards/IngredientFormatReward/std": 0.10262673646211624,
      "rewards/IngredientMatchReward/mean": 0.6572916746139527,
      "rewards/IngredientMatchReward/std": 0.2996719777584076,
      "rewards/IngredientQuantityMatchReward/mean": 0.6441153109073638,
      "rewards/IngredientQuantityMatchReward/std": 0.4026394784450531,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.4011474013328552,
      "step": 2380
    },
    {
      "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": 0.009375,
      "completions/max_length": 508.2,
      "completions/mean_length": 397.2328125,
      "completions/min_length": 286.4,
      "epoch": 0.6634214186369958,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7171292304992676,
      "kl": 0.04705277043394744,
      "learning_rate": 2.790705418464141e-07,
      "loss": 0.0018818611279129982,
      "reward": 3.167479467391968,
      "reward_std": 0.40963455438613894,
      "rewards/IngredientFormatReward/mean": 0.9888727784156799,
      "rewards/IngredientFormatReward/std": 0.07853043712675571,
      "rewards/IngredientMatchReward/mean": 0.7033141136169434,
      "rewards/IngredientMatchReward/std": 0.2901200741529465,
      "rewards/IngredientQuantityMatchReward/mean": 0.7034175992012024,
      "rewards/IngredientQuantityMatchReward/std": 0.39401066303253174,
      "rewards/TotalKcalExactMatchReward/mean": 0.771875,
      "rewards/TotalKcalExactMatchReward/std": 0.4138317406177521,
      "step": 2385
    },
    {
      "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": 0.0109375,
      "completions/max_length": 508.0,
      "completions/mean_length": 391.1328125,
      "completions/min_length": 260.8,
      "epoch": 0.6648122392211405,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7062774896621704,
      "kl": 0.0725515817059204,
      "learning_rate": 2.770097279362986e-07,
      "loss": 0.0029019616544246674,
      "reward": 3.0785242557525634,
      "reward_std": 0.4356281340122223,
      "rewards/IngredientFormatReward/mean": 0.9871577501296998,
      "rewards/IngredientFormatReward/std": 0.08785807080566883,
      "rewards/IngredientMatchReward/mean": 0.6512655138969421,
      "rewards/IngredientMatchReward/std": 0.2800576239824295,
      "rewards/IngredientQuantityMatchReward/mean": 0.6526010870933533,
      "rewards/IngredientQuantityMatchReward/std": 0.40168362855911255,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.39667311906814573,
      "step": 2390
    },
    {
      "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": 0.0125,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.40625,
      "completions/min_length": 275.8,
      "epoch": 0.6662030598052852,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6249176263809204,
      "kl": 0.047266633994877336,
      "learning_rate": 2.749536318692412e-07,
      "loss": 0.0018904266878962516,
      "reward": 3.128342056274414,
      "reward_std": 0.40333980321884155,
      "rewards/IngredientFormatReward/mean": 0.9869791626930237,
      "rewards/IngredientFormatReward/std": 0.10851280093193054,
      "rewards/IngredientMatchReward/mean": 0.6634164214134216,
      "rewards/IngredientMatchReward/std": 0.2770405501127243,
      "rewards/IngredientQuantityMatchReward/mean": 0.6513839364051819,
      "rewards/IngredientQuantityMatchReward/std": 0.40092127323150634,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.37440555095672606,
      "step": 2395
    },
    {
      "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": 0.0171875,
      "completions/max_length": 511.6,
      "completions/mean_length": 398.01875,
      "completions/min_length": 271.0,
      "epoch": 0.6675938803894298,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6117978692054749,
      "kl": 0.04874067837372422,
      "learning_rate": 2.7290229714641545e-07,
      "loss": 0.0019499026238918304,
      "reward": 3.1176537036895753,
      "reward_std": 0.39257633686065674,
      "rewards/IngredientFormatReward/mean": 0.9817708253860473,
      "rewards/IngredientFormatReward/std": 0.1140602245926857,
      "rewards/IngredientMatchReward/mean": 0.6361191749572754,
      "rewards/IngredientMatchReward/std": 0.293190997838974,
      "rewards/IngredientQuantityMatchReward/mean": 0.6950762033462524,
      "rewards/IngredientQuantityMatchReward/std": 0.3991281509399414,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.3917708516120911,
      "step": 2400
    },
    {
      "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": 0.0078125,
      "completions/max_length": 506.8,
      "completions/mean_length": 395.8390625,
      "completions/min_length": 275.2,
      "epoch": 0.6689847009735744,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6851879358291626,
      "kl": 0.04812838714569807,
      "learning_rate": 2.708557671682585e-07,
      "loss": 0.0019251177087426185,
      "reward": 3.0077756881713866,
      "reward_std": 0.3903347373008728,
      "rewards/IngredientFormatReward/mean": 0.9932142972946167,
      "rewards/IngredientFormatReward/std": 0.060679107904434204,
      "rewards/IngredientMatchReward/mean": 0.6542746782302856,
      "rewards/IngredientMatchReward/std": 0.2864963531494141,
      "rewards/IngredientQuantityMatchReward/mean": 0.6087242603302002,
      "rewards/IngredientQuantityMatchReward/std": 0.43384186625480653,
      "rewards/TotalKcalExactMatchReward/mean": 0.7515625,
      "rewards/TotalKcalExactMatchReward/std": 0.4268688678741455,
      "step": 2405
    },
    {
      "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": 0.0109375,
      "completions/max_length": 504.8,
      "completions/mean_length": 394.96875,
      "completions/min_length": 269.2,
      "epoch": 0.6703755215577191,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.651664674282074,
      "kl": 0.05091009079478681,
      "learning_rate": 2.6881408523355296e-07,
      "loss": 0.0020366348326206207,
      "reward": 3.191921615600586,
      "reward_std": 0.40628759264945985,
      "rewards/IngredientFormatReward/mean": 0.9890625,
      "rewards/IngredientFormatReward/std": 0.07843081802129745,
      "rewards/IngredientMatchReward/mean": 0.7092541575431823,
      "rewards/IngredientMatchReward/std": 0.2892185389995575,
      "rewards/IngredientQuantityMatchReward/mean": 0.6748549461364746,
      "rewards/IngredientQuantityMatchReward/std": 0.4222606658935547,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.374184775352478,
      "step": 2410
    },
    {
      "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": 0.003125,
      "completions/max_length": 502.4,
      "completions/mean_length": 387.9296875,
      "completions/min_length": 265.2,
      "epoch": 0.6717663421418637,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6594770550727844,
      "kl": 0.04690086429473013,
      "learning_rate": 2.667772945385096e-07,
      "loss": 0.0018760446459054948,
      "reward": 3.105556106567383,
      "reward_std": 0.34251810908317565,
      "rewards/IngredientFormatReward/mean": 0.9966145753860474,
      "rewards/IngredientFormatReward/std": 0.035576280951499936,
      "rewards/IngredientMatchReward/mean": 0.6220870614051819,
      "rewards/IngredientMatchReward/std": 0.2938617765903473,
      "rewards/IngredientQuantityMatchReward/mean": 0.6837294578552247,
      "rewards/IngredientQuantityMatchReward/std": 0.4037123382091522,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.3925239682197571,
      "step": 2415
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.6,
      "completions/mean_length": 392.546875,
      "completions/min_length": 273.0,
      "epoch": 0.6731571627260083,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6033135056495667,
      "kl": 0.04802413797006011,
      "learning_rate": 2.647454381758557e-07,
      "loss": 0.0019207324832677842,
      "reward": 3.080314111709595,
      "reward_std": 0.4404604434967041,
      "rewards/IngredientFormatReward/mean": 0.9889583230018616,
      "rewards/IngredientFormatReward/std": 0.0812400221824646,
      "rewards/IngredientMatchReward/mean": 0.6385317444801331,
      "rewards/IngredientMatchReward/std": 0.288109216094017,
      "rewards/IngredientQuantityMatchReward/mean": 0.6762615799903869,
      "rewards/IngredientQuantityMatchReward/std": 0.40315200090408326,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4032036066055298,
      "step": 2420
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.6265625,
      "completions/min_length": 265.2,
      "epoch": 0.674547983310153,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6370003819465637,
      "kl": 0.04624895497690886,
      "learning_rate": 2.627185591339212e-07,
      "loss": 0.0018500026315450668,
      "reward": 3.1716074466705324,
      "reward_std": 0.39525836110115053,
      "rewards/IngredientFormatReward/mean": 0.98515625,
      "rewards/IngredientFormatReward/std": 0.11617832779884338,
      "rewards/IngredientMatchReward/mean": 0.6576748609542846,
      "rewards/IngredientMatchReward/std": 0.2919008910655975,
      "rewards/IngredientQuantityMatchReward/mean": 0.6819014191627503,
      "rewards/IngredientQuantityMatchReward/std": 0.40078869462013245,
      "rewards/TotalKcalExactMatchReward/mean": 0.846875,
      "rewards/TotalKcalExactMatchReward/std": 0.35469257831573486,
      "step": 2425
    },
    {
      "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": 0.0203125,
      "completions/max_length": 505.6,
      "completions/mean_length": 386.7359375,
      "completions/min_length": 245.2,
      "epoch": 0.6759388038942976,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6779886484146118,
      "kl": 0.04996294579468667,
      "learning_rate": 2.606967002957303e-07,
      "loss": 0.0019983302801847456,
      "reward": 3.0284729957580567,
      "reward_std": 0.43177472352981566,
      "rewards/IngredientFormatReward/mean": 0.9650520801544189,
      "rewards/IngredientFormatReward/std": 0.15353514421731235,
      "rewards/IngredientMatchReward/mean": 0.6522922873497009,
      "rewards/IngredientMatchReward/std": 0.31233987808227537,
      "rewards/IngredientQuantityMatchReward/mean": 0.6408162117004395,
      "rewards/IngredientQuantityMatchReward/std": 0.410346245765686,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.4039487600326538,
      "step": 2430
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.0,
      "completions/mean_length": 392.6703125,
      "completions/min_length": 281.4,
      "epoch": 0.6773296244784422,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6478075981140137,
      "kl": 0.04866862255148589,
      "learning_rate": 2.5867990443809365e-07,
      "loss": 0.0019466172903776168,
      "reward": 3.102987194061279,
      "reward_std": 0.44889168739318847,
      "rewards/IngredientFormatReward/mean": 0.9877343773841858,
      "rewards/IngredientFormatReward/std": 0.08591293394565583,
      "rewards/IngredientMatchReward/mean": 0.6462679743766785,
      "rewards/IngredientMatchReward/std": 0.2983993053436279,
      "rewards/IngredientQuantityMatchReward/mean": 0.6330472826957703,
      "rewards/IngredientQuantityMatchReward/std": 0.42972511053085327,
      "rewards/TotalKcalExactMatchReward/mean": 0.8359375,
      "rewards/TotalKcalExactMatchReward/std": 0.3645253419876099,
      "step": 2435
    },
    {
      "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": 0.00625,
      "completions/max_length": 510.2,
      "completions/mean_length": 389.6546875,
      "completions/min_length": 233.2,
      "epoch": 0.6787204450625869,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6969110369682312,
      "kl": 0.0487577939638868,
      "learning_rate": 2.5666821423070384e-07,
      "loss": 0.001950256898999214,
      "reward": 3.03529806137085,
      "reward_std": 0.39315775632858274,
      "rewards/IngredientFormatReward/mean": 0.9934375047683716,
      "rewards/IngredientFormatReward/std": 0.0710334450006485,
      "rewards/IngredientMatchReward/mean": 0.643232274055481,
      "rewards/IngredientMatchReward/std": 0.27578197717666625,
      "rewards/IngredientQuantityMatchReward/mean": 0.600190806388855,
      "rewards/IngredientQuantityMatchReward/std": 0.4328667461872101,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.39933934807777405,
      "step": 2440
    },
    {
      "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": 0.003125,
      "completions/max_length": 496.2,
      "completions/mean_length": 395.3625,
      "completions/min_length": 277.2,
      "epoch": 0.6801112656467315,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.629336953163147,
      "kl": 0.049779637204483154,
      "learning_rate": 2.5466167223523207e-07,
      "loss": 0.0019913079217076302,
      "reward": 3.1274917125701904,
      "reward_std": 0.35138294100761414,
      "rewards/IngredientFormatReward/mean": 0.9953348159790039,
      "rewards/IngredientFormatReward/std": 0.04012070763856172,
      "rewards/IngredientMatchReward/mean": 0.6890367865562439,
      "rewards/IngredientMatchReward/std": 0.2609582722187042,
      "rewards/IngredientQuantityMatchReward/mean": 0.6618700504302979,
      "rewards/IngredientQuantityMatchReward/std": 0.40753733515739443,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.4101200461387634,
      "step": 2445
    },
    {
      "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": 0.0140625,
      "completions/max_length": 509.4,
      "completions/mean_length": 396.7015625,
      "completions/min_length": 273.4,
      "epoch": 0.6815020862308763,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6589193344116211,
      "kl": 0.0490143105853349,
      "learning_rate": 2.5266032090442857e-07,
      "loss": 0.001960846781730652,
      "reward": 3.1061669826507567,
      "reward_std": 0.4119180262088776,
      "rewards/IngredientFormatReward/mean": 0.9869642972946167,
      "rewards/IngredientFormatReward/std": 0.0859108954668045,
      "rewards/IngredientMatchReward/mean": 0.6312419414520264,
      "rewards/IngredientMatchReward/std": 0.2935998737812042,
      "rewards/IngredientQuantityMatchReward/mean": 0.7145232081413269,
      "rewards/IngredientQuantityMatchReward/std": 0.38715381026268003,
      "rewards/TotalKcalExactMatchReward/mean": 0.7734375,
      "rewards/TotalKcalExactMatchReward/std": 0.41800680160522463,
      "step": 2450
    },
    {
      "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": 0.025,
      "completions/max_length": 509.8,
      "completions/mean_length": 401.896875,
      "completions/min_length": 264.4,
      "epoch": 0.6828929068150209,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 5.397259712219238,
      "kl": 0.0734444867586717,
      "learning_rate": 2.50664202581223e-07,
      "loss": 0.0029396936297416687,
      "reward": 2.9933862686157227,
      "reward_std": 0.4221198558807373,
      "rewards/IngredientFormatReward/mean": 0.9723698019981384,
      "rewards/IngredientFormatReward/std": 0.1247881144285202,
      "rewards/IngredientMatchReward/mean": 0.6287630200386047,
      "rewards/IngredientMatchReward/std": 0.29730098247528075,
      "rewards/IngredientQuantityMatchReward/mean": 0.5828785061836242,
      "rewards/IngredientQuantityMatchReward/std": 0.4397486746311188,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.37806967496871946,
      "step": 2455
    },
    {
      "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": 0.0125,
      "completions/max_length": 510.0,
      "completions/mean_length": 394.15625,
      "completions/min_length": 247.4,
      "epoch": 0.6842837273991655,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6566894054412842,
      "kl": 0.07639441229403018,
      "learning_rate": 2.4867335949782977e-07,
      "loss": 0.003058093413710594,
      "reward": 3.0628830909729006,
      "reward_std": 0.42465184926986693,
      "rewards/IngredientFormatReward/mean": 0.9839192509651185,
      "rewards/IngredientFormatReward/std": 0.11864397823810577,
      "rewards/IngredientMatchReward/mean": 0.6475834965705871,
      "rewards/IngredientMatchReward/std": 0.2913681983947754,
      "rewards/IngredientQuantityMatchReward/mean": 0.6204427242279053,
      "rewards/IngredientQuantityMatchReward/std": 0.4370134472846985,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.38879077434539794,
      "step": 2460
    },
    {
      "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": 0.015625,
      "completions/max_length": 512.8,
      "completions/mean_length": 396.884375,
      "completions/min_length": 264.0,
      "epoch": 0.6856745479833102,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6536556482315063,
      "kl": 0.045558814378455284,
      "learning_rate": 2.4668783377485406e-07,
      "loss": 0.0018225433304905891,
      "reward": 3.0459858417510985,
      "reward_std": 0.43674235939979555,
      "rewards/IngredientFormatReward/mean": 0.9867038726806641,
      "rewards/IngredientFormatReward/std": 0.09818108770996332,
      "rewards/IngredientMatchReward/mean": 0.6260255217552185,
      "rewards/IngredientMatchReward/std": 0.2841141104698181,
      "rewards/IngredientQuantityMatchReward/mean": 0.6441938519477844,
      "rewards/IngredientQuantityMatchReward/std": 0.4117276132106781,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.399785590171814,
      "step": 2465
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.6,
      "completions/mean_length": 398.1921875,
      "completions/min_length": 265.0,
      "epoch": 0.6870653685674548,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6002658605575562,
      "kl": 0.047888010111637416,
      "learning_rate": 2.4470766742040107e-07,
      "loss": 0.0019158570095896721,
      "reward": 3.166787576675415,
      "reward_std": 0.38818521797657013,
      "rewards/IngredientFormatReward/mean": 0.9827827453613281,
      "rewards/IngredientFormatReward/std": 0.11082592085003853,
      "rewards/IngredientMatchReward/mean": 0.64622642993927,
      "rewards/IngredientMatchReward/std": 0.28593012094497683,
      "rewards/IngredientQuantityMatchReward/mean": 0.6784033298492431,
      "rewards/IngredientQuantityMatchReward/std": 0.41570827960968015,
      "rewards/TotalKcalExactMatchReward/mean": 0.859375,
      "rewards/TotalKcalExactMatchReward/std": 0.3468464195728302,
      "step": 2470
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.6,
      "completions/mean_length": 394.9578125,
      "completions/min_length": 263.6,
      "epoch": 0.6884561891515995,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6655333638191223,
      "kl": 0.05003447597846389,
      "learning_rate": 2.427329023291864e-07,
      "loss": 0.002001406252384186,
      "reward": 3.0961260318756105,
      "reward_std": 0.4364780128002167,
      "rewards/IngredientFormatReward/mean": 0.9879687428474426,
      "rewards/IngredientFormatReward/std": 0.0807617649435997,
      "rewards/IngredientMatchReward/mean": 0.6726500511169433,
      "rewards/IngredientMatchReward/std": 0.28994928300380707,
      "rewards/IngredientQuantityMatchReward/mean": 0.6292572021484375,
      "rewards/IngredientQuantityMatchReward/std": 0.43481956124305726,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.39507427215576174,
      "step": 2475
    },
    {
      "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": 0.0171875,
      "completions/max_length": 512.6,
      "completions/mean_length": 394.7015625,
      "completions/min_length": 247.6,
      "epoch": 0.6898470097357441,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6104139685630798,
      "kl": 0.05500267161987722,
      "learning_rate": 2.4076358028165053e-07,
      "loss": 0.002200280874967575,
      "reward": 3.086128854751587,
      "reward_std": 0.42833173274993896,
      "rewards/IngredientFormatReward/mean": 0.9812053442001343,
      "rewards/IngredientFormatReward/std": 0.11338990479707718,
      "rewards/IngredientMatchReward/mean": 0.6116282224655152,
      "rewards/IngredientMatchReward/std": 0.2804492115974426,
      "rewards/IngredientQuantityMatchReward/mean": 0.6339202046394348,
      "rewards/IngredientQuantityMatchReward/std": 0.4178707242012024,
      "rewards/TotalKcalExactMatchReward/mean": 0.859375,
      "rewards/TotalKcalExactMatchReward/std": 0.32549999952316283,
      "step": 2480
    },
    {
      "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": 0.0078125,
      "completions/max_length": 511.4,
      "completions/mean_length": 389.6890625,
      "completions/min_length": 294.0,
      "epoch": 0.6912378303198887,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6676151752471924,
      "kl": 0.048110749153420326,
      "learning_rate": 2.387997429430746e-07,
      "loss": 0.001924731768667698,
      "reward": 3.2237563610076903,
      "reward_std": 0.36057083010673524,
      "rewards/IngredientFormatReward/mean": 0.9906845331192017,
      "rewards/IngredientFormatReward/std": 0.08420297093689441,
      "rewards/IngredientMatchReward/mean": 0.7073366641998291,
      "rewards/IngredientMatchReward/std": 0.2901332199573517,
      "rewards/IngredientQuantityMatchReward/mean": 0.7022976398468017,
      "rewards/IngredientQuantityMatchReward/std": 0.3529376655817032,
      "rewards/TotalKcalExactMatchReward/mean": 0.8234375,
      "rewards/TotalKcalExactMatchReward/std": 0.3717910945415497,
      "step": 2485
    },
    {
      "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": 0.0046875,
      "completions/max_length": 497.2,
      "completions/mean_length": 389.11875,
      "completions/min_length": 252.8,
      "epoch": 0.6926286509040334,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6633597612380981,
      "kl": 0.048306941101327536,
      "learning_rate": 2.3684143186269885e-07,
      "loss": 0.0019321149215102197,
      "reward": 3.178741788864136,
      "reward_std": 0.377359014749527,
      "rewards/IngredientFormatReward/mean": 0.9950000047683716,
      "rewards/IngredientFormatReward/std": 0.04611458256840706,
      "rewards/IngredientMatchReward/mean": 0.7043855547904968,
      "rewards/IngredientMatchReward/std": 0.2860237658023834,
      "rewards/IngredientQuantityMatchReward/mean": 0.6965437650680542,
      "rewards/IngredientQuantityMatchReward/std": 0.39232956767082217,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.40132089853286745,
      "step": 2490
    },
    {
      "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": 0.0015625,
      "completions/max_length": 498.2,
      "completions/mean_length": 391.478125,
      "completions/min_length": 264.6,
      "epoch": 0.694019471488178,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7364270687103271,
      "kl": 0.04957497245632112,
      "learning_rate": 2.3488868847284292e-07,
      "loss": 0.0019831184297800066,
      "reward": 3.1426888942718505,
      "reward_std": 0.3652894914150238,
      "rewards/IngredientFormatReward/mean": 0.996932053565979,
      "rewards/IngredientFormatReward/std": 0.02726769857108593,
      "rewards/IngredientMatchReward/mean": 0.6408079981803894,
      "rewards/IngredientMatchReward/std": 0.2871991991996765,
      "rewards/IngredientQuantityMatchReward/mean": 0.6596363306045532,
      "rewards/IngredientQuantityMatchReward/std": 0.41957661509513855,
      "rewards/TotalKcalExactMatchReward/mean": 0.8453125,
      "rewards/TotalKcalExactMatchReward/std": 0.3346115857362747,
      "step": 2495
    },
    {
      "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": 0.0078125,
      "completions/max_length": 504.2,
      "completions/mean_length": 394.440625,
      "completions/min_length": 263.6,
      "epoch": 0.6954102920723226,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.664070725440979,
      "kl": 0.05223251197021454,
      "learning_rate": 2.3294155408803067e-07,
      "loss": 0.0020896628499031066,
      "reward": 3.0636887550354004,
      "reward_std": 0.3960974931716919,
      "rewards/IngredientFormatReward/mean": 0.9897916555404663,
      "rewards/IngredientFormatReward/std": 0.08864808697253465,
      "rewards/IngredientMatchReward/mean": 0.6588827013969422,
      "rewards/IngredientMatchReward/std": 0.3019312024116516,
      "rewards/IngredientQuantityMatchReward/mean": 0.585326862335205,
      "rewards/IngredientQuantityMatchReward/std": 0.4312814950942993,
      "rewards/TotalKcalExactMatchReward/mean": 0.8296875,
      "rewards/TotalKcalExactMatchReward/std": 0.3555891036987305,
      "step": 2500
    },
    {
      "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": 0.028125,
      "completions/max_length": 511.6,
      "completions/mean_length": 399.46875,
      "completions/min_length": 279.0,
      "epoch": 0.6968011126564673,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6669741868972778,
      "kl": 0.050500425044447184,
      "learning_rate": 2.3100006990411474e-07,
      "loss": 0.0020199045538902283,
      "reward": 3.1506638526916504,
      "reward_std": 0.468947970867157,
      "rewards/IngredientFormatReward/mean": 0.9685267925262451,
      "rewards/IngredientFormatReward/std": 0.1518846545368433,
      "rewards/IngredientMatchReward/mean": 0.6512828588485717,
      "rewards/IngredientMatchReward/std": 0.2848780870437622,
      "rewards/IngredientQuantityMatchReward/mean": 0.6996042728424072,
      "rewards/IngredientQuantityMatchReward/std": 0.4034592926502228,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.34879711270332336,
      "step": 2505
    },
    {
      "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": 0.015625,
      "completions/max_length": 508.2,
      "completions/mean_length": 392.584375,
      "completions/min_length": 253.6,
      "epoch": 0.6981919332406119,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7342477440834045,
      "kl": 0.05032761418260634,
      "learning_rate": 2.2906427699740627e-07,
      "loss": 0.0020130135118961332,
      "reward": 3.1510085582733156,
      "reward_std": 0.439526242017746,
      "rewards/IngredientFormatReward/mean": 0.9832812547683716,
      "rewards/IngredientFormatReward/std": 0.11577541343867778,
      "rewards/IngredientMatchReward/mean": 0.668173360824585,
      "rewards/IngredientMatchReward/std": 0.2981277287006378,
      "rewards/IngredientQuantityMatchReward/mean": 0.6729914307594299,
      "rewards/IngredientQuantityMatchReward/std": 0.42339866757392886,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.37619284987449647,
      "step": 2510
    },
    {
      "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": 0.03125,
      "completions/max_length": 513.0,
      "completions/mean_length": 398.271875,
      "completions/min_length": 269.4,
      "epoch": 0.6995827538247567,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6291016340255737,
      "kl": 0.045997858187183736,
      "learning_rate": 2.271342163238041e-07,
      "loss": 0.0018398519605398178,
      "reward": 3.1013444900512694,
      "reward_std": 0.43752923607826233,
      "rewards/IngredientFormatReward/mean": 0.96875,
      "rewards/IngredientFormatReward/std": 0.16352393478155136,
      "rewards/IngredientMatchReward/mean": 0.6443898916244507,
      "rewards/IngredientMatchReward/std": 0.3039585888385773,
      "rewards/IngredientQuantityMatchReward/mean": 0.6710170984268189,
      "rewards/IngredientQuantityMatchReward/std": 0.42277750968933103,
      "rewards/TotalKcalExactMatchReward/mean": 0.8171875,
      "rewards/TotalKcalExactMatchReward/std": 0.3669287174940109,
      "step": 2515
    },
    {
      "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": 0.0109375,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.303125,
      "completions/min_length": 234.8,
      "epoch": 0.7009735744089013,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 4.0983734130859375,
      "kl": 0.06823289862368256,
      "learning_rate": 2.2520992871792977e-07,
      "loss": 0.0027291785925626756,
      "reward": 2.946266031265259,
      "reward_std": 0.4240442216396332,
      "rewards/IngredientFormatReward/mean": 0.9885416507720948,
      "rewards/IngredientFormatReward/std": 0.10327765196561814,
      "rewards/IngredientMatchReward/mean": 0.6329321742057801,
      "rewards/IngredientMatchReward/std": 0.2933188438415527,
      "rewards/IngredientQuantityMatchReward/mean": 0.6232297539710998,
      "rewards/IngredientQuantityMatchReward/std": 0.44109280705451964,
      "rewards/TotalKcalExactMatchReward/mean": 0.7015625,
      "rewards/TotalKcalExactMatchReward/std": 0.4390376150608063,
      "step": 2520
    },
    {
      "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": 0.0171875,
      "completions/max_length": 503.8,
      "completions/mean_length": 391.29375,
      "completions/min_length": 262.6,
      "epoch": 0.7023643949930459,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6922910213470459,
      "kl": 0.05018776603974402,
      "learning_rate": 2.2329145489226304e-07,
      "loss": 0.0020075805485248567,
      "reward": 3.0498045921325683,
      "reward_std": 0.4153340578079224,
      "rewards/IngredientFormatReward/mean": 0.9825595259666443,
      "rewards/IngredientFormatReward/std": 0.0890352837741375,
      "rewards/IngredientMatchReward/mean": 0.6578689217567444,
      "rewards/IngredientMatchReward/std": 0.2839093506336212,
      "rewards/IngredientQuantityMatchReward/mean": 0.6328136444091796,
      "rewards/IngredientQuantityMatchReward/std": 0.40785926580429077,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4117879748344421,
      "step": 2525
    },
    {
      "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": 0.0078125,
      "completions/max_length": 506.8,
      "completions/mean_length": 394.7578125,
      "completions/min_length": 260.6,
      "epoch": 0.7037552155771906,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6483815908432007,
      "kl": 0.04718795008957386,
      "learning_rate": 2.2137883543628044e-07,
      "loss": 0.0018875142559409142,
      "reward": 3.110764265060425,
      "reward_std": 0.40262900590896605,
      "rewards/IngredientFormatReward/mean": 0.9931249976158142,
      "rewards/IngredientFormatReward/std": 0.06067222654819489,
      "rewards/IngredientMatchReward/mean": 0.671670389175415,
      "rewards/IngredientMatchReward/std": 0.2906846523284912,
      "rewards/IngredientQuantityMatchReward/mean": 0.6490937829017639,
      "rewards/IngredientQuantityMatchReward/std": 0.4294052541255951,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.3973664820194244,
      "step": 2530
    },
    {
      "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": 0.01875,
      "completions/max_length": 511.6,
      "completions/mean_length": 395.5609375,
      "completions/min_length": 270.4,
      "epoch": 0.7051460361613352,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6322222948074341,
      "kl": 0.047202359139919284,
      "learning_rate": 2.1947211081559664e-07,
      "loss": 0.0018881641328334809,
      "reward": 3.0978555202484133,
      "reward_std": 0.4240407347679138,
      "rewards/IngredientFormatReward/mean": 0.979281997680664,
      "rewards/IngredientFormatReward/std": 0.1245550911873579,
      "rewards/IngredientMatchReward/mean": 0.6488405346870423,
      "rewards/IngredientMatchReward/std": 0.29211672842502595,
      "rewards/IngredientQuantityMatchReward/mean": 0.660357940196991,
      "rewards/IngredientQuantityMatchReward/std": 0.41334924697875974,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.392572158575058,
      "step": 2535
    },
    {
      "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": 0.0125,
      "completions/max_length": 510.6,
      "completions/mean_length": 391.8828125,
      "completions/min_length": 262.8,
      "epoch": 0.7065368567454798,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6519886255264282,
      "kl": 0.060429237270727756,
      "learning_rate": 2.1757132137110823e-07,
      "loss": 0.002402695082128048,
      "reward": 3.05753607749939,
      "reward_std": 0.4117233157157898,
      "rewards/IngredientFormatReward/mean": 0.9845312595367431,
      "rewards/IngredientFormatReward/std": 0.10385205522179604,
      "rewards/IngredientMatchReward/mean": 0.6303422689437866,
      "rewards/IngredientMatchReward/std": 0.2817304372787476,
      "rewards/IngredientQuantityMatchReward/mean": 0.6317250967025757,
      "rewards/IngredientQuantityMatchReward/std": 0.4206810534000397,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.3870365977287292,
      "step": 2540
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.6,
      "completions/mean_length": 393.0890625,
      "completions/min_length": 270.6,
      "epoch": 0.7079276773296245,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.5975901484489441,
      "kl": 0.05394169162027538,
      "learning_rate": 2.156765073181404e-07,
      "loss": 0.0021578073501586914,
      "reward": 2.9798144340515136,
      "reward_std": 0.37361904978752136,
      "rewards/IngredientFormatReward/mean": 0.9895572900772095,
      "rewards/IngredientFormatReward/std": 0.07822814472019672,
      "rewards/IngredientMatchReward/mean": 0.5983500719070435,
      "rewards/IngredientMatchReward/std": 0.30022152662277224,
      "rewards/IngredientQuantityMatchReward/mean": 0.6169070601463318,
      "rewards/IngredientQuantityMatchReward/std": 0.44377334117889405,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.4066710531711578,
      "step": 2545
    },
    {
      "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": 0.01875,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.8390625,
      "completions/min_length": 242.4,
      "epoch": 0.7093184979137691,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6239141225814819,
      "kl": 0.046976763568818566,
      "learning_rate": 2.1378770874559605e-07,
      "loss": 0.001879183202981949,
      "reward": 3.0559041023254396,
      "reward_std": 0.4519073784351349,
      "rewards/IngredientFormatReward/mean": 0.9795312404632568,
      "rewards/IngredientFormatReward/std": 0.13446595668792724,
      "rewards/IngredientMatchReward/mean": 0.6565352320671082,
      "rewards/IngredientMatchReward/std": 0.2887867510318756,
      "rewards/IngredientQuantityMatchReward/mean": 0.6714001893997192,
      "rewards/IngredientQuantityMatchReward/std": 0.40756980776786805,
      "rewards/TotalKcalExactMatchReward/mean": 0.7484375,
      "rewards/TotalKcalExactMatchReward/std": 0.4154438376426697,
      "step": 2550
    },
    {
      "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": 0.0203125,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.01875,
      "completions/min_length": 279.0,
      "epoch": 0.7107093184979137,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6433897614479065,
      "kl": 0.04894828526303172,
      "learning_rate": 2.119049656151069e-07,
      "loss": 0.001957985386252403,
      "reward": 3.025065851211548,
      "reward_std": 0.4066974937915802,
      "rewards/IngredientFormatReward/mean": 0.980859375,
      "rewards/IngredientFormatReward/std": 0.13396839499473573,
      "rewards/IngredientMatchReward/mean": 0.6333761096000672,
      "rewards/IngredientMatchReward/std": 0.2985145926475525,
      "rewards/IngredientQuantityMatchReward/mean": 0.6202053546905517,
      "rewards/IngredientQuantityMatchReward/std": 0.42285839915275575,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.4044800400733948,
      "step": 2555
    },
    {
      "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": 0.0046875,
      "completions/max_length": 494.8,
      "completions/mean_length": 383.159375,
      "completions/min_length": 244.6,
      "epoch": 0.7121001390820584,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6727476119995117,
      "kl": 0.04658746891655028,
      "learning_rate": 2.100283177601892e-07,
      "loss": 0.0018635163083672523,
      "reward": 3.1245415687561033,
      "reward_std": 0.40727477669715884,
      "rewards/IngredientFormatReward/mean": 0.99375,
      "rewards/IngredientFormatReward/std": 0.06025671809911728,
      "rewards/IngredientMatchReward/mean": 0.6800031065940857,
      "rewards/IngredientMatchReward/std": 0.27216280102729795,
      "rewards/IngredientQuantityMatchReward/mean": 0.64922593832016,
      "rewards/IngredientQuantityMatchReward/std": 0.4209359347820282,
      "rewards/TotalKcalExactMatchReward/mean": 0.8015625,
      "rewards/TotalKcalExactMatchReward/std": 0.3881435811519623,
      "step": 2560
    },
    {
      "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": 0.0125,
      "completions/max_length": 511.8,
      "completions/mean_length": 394.890625,
      "completions/min_length": 268.8,
      "epoch": 0.713490959666203,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6577991843223572,
      "kl": 0.04966968102380633,
      "learning_rate": 2.081578048854007e-07,
      "loss": 0.001986708864569664,
      "reward": 3.146220636367798,
      "reward_std": 0.397566819190979,
      "rewards/IngredientFormatReward/mean": 0.9839787960052491,
      "rewards/IngredientFormatReward/std": 0.10710346139967442,
      "rewards/IngredientMatchReward/mean": 0.6619359254837036,
      "rewards/IngredientMatchReward/std": 0.27924045324325564,
      "rewards/IngredientQuantityMatchReward/mean": 0.6721809267997741,
      "rewards/IngredientQuantityMatchReward/std": 0.3993909597396851,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.36588002145290377,
      "step": 2565
    },
    {
      "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": 0.015625,
      "completions/max_length": 510.0,
      "completions/mean_length": 395.153125,
      "completions/min_length": 275.8,
      "epoch": 0.7148817802503477,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6442493200302124,
      "kl": 0.045088665070943536,
      "learning_rate": 2.0629346656549995e-07,
      "loss": 0.001803700253367424,
      "reward": 3.17347731590271,
      "reward_std": 0.41533090472221373,
      "rewards/IngredientFormatReward/mean": 0.9840625047683715,
      "rewards/IngredientFormatReward/std": 0.10534819960594177,
      "rewards/IngredientMatchReward/mean": 0.6939707398414612,
      "rewards/IngredientMatchReward/std": 0.2825161337852478,
      "rewards/IngredientQuantityMatchReward/mean": 0.7095066428184509,
      "rewards/IngredientQuantityMatchReward/std": 0.38486247062683104,
      "rewards/TotalKcalExactMatchReward/mean": 0.7859375,
      "rewards/TotalKcalExactMatchReward/std": 0.40840469002723695,
      "step": 2570
    },
    {
      "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": 0.015625,
      "completions/max_length": 513.0,
      "completions/mean_length": 397.44375,
      "completions/min_length": 277.8,
      "epoch": 0.7162726008344924,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7509127259254456,
      "kl": 0.046041283709928396,
      "learning_rate": 2.0443534224460907e-07,
      "loss": 0.001841634139418602,
      "reward": 3.1562008380889894,
      "reward_std": 0.42731393575668336,
      "rewards/IngredientFormatReward/mean": 0.9847916603088379,
      "rewards/IngredientFormatReward/std": 0.11797711700201034,
      "rewards/IngredientMatchReward/mean": 0.650058913230896,
      "rewards/IngredientMatchReward/std": 0.28965330123901367,
      "rewards/IngredientQuantityMatchReward/mean": 0.7229127407073974,
      "rewards/IngredientQuantityMatchReward/std": 0.3845226109027863,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.3828358560800552,
      "step": 2575
    },
    {
      "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": 0.0109375,
      "completions/max_length": 513.0,
      "completions/mean_length": 395.8515625,
      "completions/min_length": 267.8,
      "epoch": 0.717663421418637,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6616153120994568,
      "kl": 0.049780993908643725,
      "learning_rate": 2.025834712353801e-07,
      "loss": 0.0019911348819732664,
      "reward": 3.0544140338897705,
      "reward_std": 0.39741013646125795,
      "rewards/IngredientFormatReward/mean": 0.9822098255157471,
      "rewards/IngredientFormatReward/std": 0.12083162069320678,
      "rewards/IngredientMatchReward/mean": 0.6724361538887024,
      "rewards/IngredientMatchReward/std": 0.28971803188323975,
      "rewards/IngredientQuantityMatchReward/mean": 0.6857055306434632,
      "rewards/IngredientQuantityMatchReward/std": 0.3946858704090118,
      "rewards/TotalKcalExactMatchReward/mean": 0.7140625,
      "rewards/TotalKcalExactMatchReward/std": 0.4354426920413971,
      "step": 2580
    },
    {
      "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": 0.00625,
      "completions/max_length": 505.2,
      "completions/mean_length": 393.409375,
      "completions/min_length": 230.8,
      "epoch": 0.7190542420027817,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6178598403930664,
      "kl": 0.048539007431827486,
      "learning_rate": 2.0073789271816243e-07,
      "loss": 0.0019415821880102158,
      "reward": 3.0660136222839354,
      "reward_std": 0.3811270177364349,
      "rewards/IngredientFormatReward/mean": 0.991796875,
      "rewards/IngredientFormatReward/std": 0.05703234635293484,
      "rewards/IngredientMatchReward/mean": 0.6717540979385376,
      "rewards/IngredientMatchReward/std": 0.28819016814231874,
      "rewards/IngredientQuantityMatchReward/mean": 0.6602752327919006,
      "rewards/IngredientQuantityMatchReward/std": 0.40727475881576536,
      "rewards/TotalKcalExactMatchReward/mean": 0.7421875,
      "rewards/TotalKcalExactMatchReward/std": 0.4277864396572113,
      "step": 2585
    },
    {
      "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": 0.003125,
      "completions/max_length": 504.2,
      "completions/mean_length": 392.73125,
      "completions/min_length": 267.8,
      "epoch": 0.7204450625869263,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6725686192512512,
      "kl": 0.04864217657595873,
      "learning_rate": 1.9889864574017428e-07,
      "loss": 0.001945541426539421,
      "reward": 3.1187971591949464,
      "reward_std": 0.3799791693687439,
      "rewards/IngredientFormatReward/mean": 0.9953645825386047,
      "rewards/IngredientFormatReward/std": 0.04649723395705223,
      "rewards/IngredientMatchReward/mean": 0.6216610789299011,
      "rewards/IngredientMatchReward/std": 0.2948246955871582,
      "rewards/IngredientQuantityMatchReward/mean": 0.6548964500427246,
      "rewards/IngredientQuantityMatchReward/std": 0.41355618834495544,
      "rewards/TotalKcalExactMatchReward/mean": 0.846875,
      "rewards/TotalKcalExactMatchReward/std": 0.34510646760463715,
      "step": 2590
    },
    {
      "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": 0.0125,
      "completions/max_length": 512.8,
      "completions/mean_length": 394.4046875,
      "completions/min_length": 270.8,
      "epoch": 0.721835883171071,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6282512545585632,
      "kl": 0.046746038901619616,
      "learning_rate": 1.9706576921467627e-07,
      "loss": 0.0018698884174227714,
      "reward": 3.0880001544952393,
      "reward_std": 0.4526852428913116,
      "rewards/IngredientFormatReward/mean": 0.9884895801544189,
      "rewards/IngredientFormatReward/std": 0.09109702110290527,
      "rewards/IngredientMatchReward/mean": 0.6641840219497681,
      "rewards/IngredientMatchReward/std": 0.289777010679245,
      "rewards/IngredientQuantityMatchReward/mean": 0.6540765881538391,
      "rewards/IngredientQuantityMatchReward/std": 0.4062558710575104,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.4044061481952667,
      "step": 2595
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.0,
      "completions/mean_length": 390.415625,
      "completions/min_length": 253.6,
      "epoch": 0.7232267037552156,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6331611275672913,
      "kl": 0.04611538765020669,
      "learning_rate": 1.9523930192014838e-07,
      "loss": 0.0018447006121277809,
      "reward": 3.1182607650756835,
      "reward_std": 0.4262834906578064,
      "rewards/IngredientFormatReward/mean": 0.9869791507720947,
      "rewards/IngredientFormatReward/std": 0.10090548861771823,
      "rewards/IngredientMatchReward/mean": 0.6147550702095032,
      "rewards/IngredientMatchReward/std": 0.285012686252594,
      "rewards/IngredientQuantityMatchReward/mean": 0.7415265083312989,
      "rewards/IngredientQuantityMatchReward/std": 0.3865606963634491,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.40770527720451355,
      "step": 2600
    },
    {
      "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": 0.0078125,
      "completions/max_length": 505.8,
      "completions/mean_length": 387.396875,
      "completions/min_length": 236.4,
      "epoch": 0.7246175243393602,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6415396332740784,
      "kl": 0.05292570195160806,
      "learning_rate": 1.9341928249946937e-07,
      "loss": 0.0021176621317863463,
      "reward": 3.027360534667969,
      "reward_std": 0.39936086535453796,
      "rewards/IngredientFormatReward/mean": 0.9881770730018615,
      "rewards/IngredientFormatReward/std": 0.09301825910806656,
      "rewards/IngredientMatchReward/mean": 0.6336619615554809,
      "rewards/IngredientMatchReward/std": 0.2789487838745117,
      "rewards/IngredientQuantityMatchReward/mean": 0.6789589166641236,
      "rewards/IngredientQuantityMatchReward/std": 0.4066279888153076,
      "rewards/TotalKcalExactMatchReward/mean": 0.7265625,
      "rewards/TotalKcalExactMatchReward/std": 0.43816269040107725,
      "step": 2605
    },
    {
      "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": 0.0125,
      "completions/max_length": 507.6,
      "completions/mean_length": 389.815625,
      "completions/min_length": 249.6,
      "epoch": 0.7260083449235049,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6317312717437744,
      "kl": 0.05009318320080638,
      "learning_rate": 1.9160574945909934e-07,
      "loss": 0.0020037142559885977,
      "reward": 3.151788091659546,
      "reward_std": 0.3983535706996918,
      "rewards/IngredientFormatReward/mean": 0.98302081823349,
      "rewards/IngredientFormatReward/std": 0.09761920589953661,
      "rewards/IngredientMatchReward/mean": 0.6275359749794006,
      "rewards/IngredientMatchReward/std": 0.28289095759391786,
      "rewards/IngredientQuantityMatchReward/mean": 0.6756062507629395,
      "rewards/IngredientQuantityMatchReward/std": 0.4136093854904175,
      "rewards/TotalKcalExactMatchReward/mean": 0.865625,
      "rewards/TotalKcalExactMatchReward/std": 0.3261595815420151,
      "step": 2610
    },
    {
      "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": 0.0046875,
      "completions/max_length": 502.6,
      "completions/mean_length": 387.7125,
      "completions/min_length": 240.0,
      "epoch": 0.7273991655076495,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6524665355682373,
      "kl": 0.049897806439548734,
      "learning_rate": 1.8979874116826434e-07,
      "loss": 0.0019955608993768694,
      "reward": 3.136799764633179,
      "reward_std": 0.4250612795352936,
      "rewards/IngredientFormatReward/mean": 0.9895833253860473,
      "rewards/IngredientFormatReward/std": 0.08923212587833404,
      "rewards/IngredientMatchReward/mean": 0.6881510496139527,
      "rewards/IngredientMatchReward/std": 0.2866966396570206,
      "rewards/IngredientQuantityMatchReward/mean": 0.6512529373168945,
      "rewards/IngredientQuantityMatchReward/std": 0.3973442554473877,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.3602495089173317,
      "step": 2615
    },
    {
      "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": 0.00625,
      "completions/max_length": 502.2,
      "completions/mean_length": 387.3828125,
      "completions/min_length": 239.4,
      "epoch": 0.7287899860917941,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.611609160900116,
      "kl": 0.04605715712532401,
      "learning_rate": 1.8799829585814626e-07,
      "loss": 0.0018422754481434822,
      "reward": 3.1545867919921875,
      "reward_std": 0.34553852677345276,
      "rewards/IngredientFormatReward/mean": 0.9854166507720947,
      "rewards/IngredientFormatReward/std": 0.08996033240109683,
      "rewards/IngredientMatchReward/mean": 0.6572426676750183,
      "rewards/IngredientMatchReward/std": 0.2847525715827942,
      "rewards/IngredientQuantityMatchReward/mean": 0.7134899377822876,
      "rewards/IngredientQuantityMatchReward/std": 0.3990268886089325,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.39305478930473325,
      "step": 2620
    },
    {
      "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": 0.003125,
      "completions/max_length": 498.8,
      "completions/mean_length": 389.04375,
      "completions/min_length": 270.8,
      "epoch": 0.7301808066759388,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6962397694587708,
      "kl": 0.04859261708334088,
      "learning_rate": 1.8620445162107202e-07,
      "loss": 0.0019435137510299683,
      "reward": 3.1344559669494627,
      "reward_std": 0.43296399116516116,
      "rewards/IngredientFormatReward/mean": 0.9953125,
      "rewards/IngredientFormatReward/std": 0.036588290333747865,
      "rewards/IngredientMatchReward/mean": 0.657600736618042,
      "rewards/IngredientMatchReward/std": 0.2820374250411987,
      "rewards/IngredientQuantityMatchReward/mean": 0.6909177303314209,
      "rewards/IngredientQuantityMatchReward/std": 0.4004930078983307,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.39993174076080323,
      "step": 2625
    },
    {
      "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": 0.0109375,
      "completions/max_length": 505.4,
      "completions/mean_length": 393.1640625,
      "completions/min_length": 269.6,
      "epoch": 0.7315716272600834,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7534053325653076,
      "kl": 0.05059192152693868,
      "learning_rate": 1.84417246409709e-07,
      "loss": 0.0020237032324075697,
      "reward": 3.184044027328491,
      "reward_std": 0.4079024374485016,
      "rewards/IngredientFormatReward/mean": 0.9871875047683716,
      "rewards/IngredientFormatReward/std": 0.09551474601030349,
      "rewards/IngredientMatchReward/mean": 0.6866220116615296,
      "rewards/IngredientMatchReward/std": 0.28260781764984133,
      "rewards/IngredientQuantityMatchReward/mean": 0.7149220108985901,
      "rewards/IngredientQuantityMatchReward/std": 0.39453916549682616,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.3818435907363892,
      "step": 2630
    },
    {
      "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": 0.0171875,
      "completions/max_length": 513.0,
      "completions/mean_length": 394.4578125,
      "completions/min_length": 267.6,
      "epoch": 0.7329624478442281,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.5793359279632568,
      "kl": 0.0508285547606647,
      "learning_rate": 1.826367180362612e-07,
      "loss": 0.0020331652835011483,
      "reward": 3.0561100482940673,
      "reward_std": 0.4060769438743591,
      "rewards/IngredientFormatReward/mean": 0.9804538607597351,
      "rewards/IngredientFormatReward/std": 0.1272687464952469,
      "rewards/IngredientMatchReward/mean": 0.6044321060180664,
      "rewards/IngredientMatchReward/std": 0.2949958324432373,
      "rewards/IngredientQuantityMatchReward/mean": 0.67903653383255,
      "rewards/IngredientQuantityMatchReward/std": 0.42194249033927916,
      "rewards/TotalKcalExactMatchReward/mean": 0.7921875,
      "rewards/TotalKcalExactMatchReward/std": 0.40656169056892394,
      "step": 2635
    },
    {
      "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": 0.0078125,
      "completions/max_length": 500.6,
      "completions/mean_length": 390.5515625,
      "completions/min_length": 281.2,
      "epoch": 0.7343532684283728,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6770186424255371,
      "kl": 0.047754300339147446,
      "learning_rate": 1.8086290417166977e-07,
      "loss": 0.0018999837338924408,
      "reward": 3.0642573833465576,
      "reward_std": 0.3867913544178009,
      "rewards/IngredientFormatReward/mean": 0.9903497099876404,
      "rewards/IngredientFormatReward/std": 0.07810738943517208,
      "rewards/IngredientMatchReward/mean": 0.638651430606842,
      "rewards/IngredientMatchReward/std": 0.29393835067749025,
      "rewards/IngredientQuantityMatchReward/mean": 0.6524439454078674,
      "rewards/IngredientQuantityMatchReward/std": 0.4144913673400879,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4063707053661346,
      "step": 2640
    },
    {
      "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": 0.0046875,
      "completions/max_length": 496.0,
      "completions/mean_length": 393.4171875,
      "completions/min_length": 285.4,
      "epoch": 0.7357440890125174,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7029209136962891,
      "kl": 0.04800816969946027,
      "learning_rate": 1.790958423448159e-07,
      "loss": 0.0019205402582883834,
      "reward": 3.187349796295166,
      "reward_std": 0.3639358103275299,
      "rewards/IngredientFormatReward/mean": 0.9950000047683716,
      "rewards/IngredientFormatReward/std": 0.04290181696414948,
      "rewards/IngredientMatchReward/mean": 0.60570809841156,
      "rewards/IngredientMatchReward/std": 0.29033520817756653,
      "rewards/IngredientQuantityMatchReward/mean": 0.6882042646408081,
      "rewards/IngredientQuantityMatchReward/std": 0.4012244701385498,
      "rewards/TotalKcalExactMatchReward/mean": 0.8984375,
      "rewards/TotalKcalExactMatchReward/std": 0.29061916172504426,
      "step": 2645
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.6,
      "completions/mean_length": 388.65,
      "completions/min_length": 230.4,
      "epoch": 0.7371349095966621,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6581112742424011,
      "kl": 0.047632145672105254,
      "learning_rate": 1.7733556994172715e-07,
      "loss": 0.0019055701792240142,
      "reward": 2.9977715492248533,
      "reward_std": 0.3989086389541626,
      "rewards/IngredientFormatReward/mean": 0.9749330282211304,
      "rewards/IngredientFormatReward/std": 0.12201923131942749,
      "rewards/IngredientMatchReward/mean": 0.6350185871124268,
      "rewards/IngredientMatchReward/std": 0.2913230359554291,
      "rewards/IngredientQuantityMatchReward/mean": 0.6721949458122254,
      "rewards/IngredientQuantityMatchReward/std": 0.4181333541870117,
      "rewards/TotalKcalExactMatchReward/mean": 0.715625,
      "rewards/TotalKcalExactMatchReward/std": 0.4446301102638245,
      "step": 2650
    },
    {
      "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": 0.0140625,
      "completions/max_length": 512.4,
      "completions/mean_length": 387.428125,
      "completions/min_length": 267.8,
      "epoch": 0.7385257301808067,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.7104362845420837,
      "kl": 0.046882928302511576,
      "learning_rate": 1.755821242047853e-07,
      "loss": 0.0018756043165922165,
      "reward": 3.117317819595337,
      "reward_std": 0.3763087272644043,
      "rewards/IngredientFormatReward/mean": 0.9822916746139526,
      "rewards/IngredientFormatReward/std": 0.12619206607341765,
      "rewards/IngredientMatchReward/mean": 0.67864009141922,
      "rewards/IngredientMatchReward/std": 0.2986286699771881,
      "rewards/IngredientQuantityMatchReward/mean": 0.6751359701156616,
      "rewards/IngredientQuantityMatchReward/std": 0.4090684413909912,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.41046456098556516,
      "step": 2655
    },
    {
      "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": 0.0125,
      "completions/max_length": 509.6,
      "completions/mean_length": 390.596875,
      "completions/min_length": 269.0,
      "epoch": 0.7399165507649513,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7171072363853455,
      "kl": 0.048218688229098916,
      "learning_rate": 1.7383554223193975e-07,
      "loss": 0.001929207518696785,
      "reward": 3.160565423965454,
      "reward_std": 0.3804220199584961,
      "rewards/IngredientFormatReward/mean": 0.9854557156562805,
      "rewards/IngredientFormatReward/std": 0.09512746483087539,
      "rewards/IngredientMatchReward/mean": 0.6397488951683045,
      "rewards/IngredientMatchReward/std": 0.3007284939289093,
      "rewards/IngredientQuantityMatchReward/mean": 0.7212984204292298,
      "rewards/IngredientQuantityMatchReward/std": 0.37817172408103944,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3665972471237183,
      "step": 2660
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.8,
      "completions/mean_length": 392.85,
      "completions/min_length": 271.8,
      "epoch": 0.741307371349096,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6789398789405823,
      "kl": 0.053108946816064415,
      "learning_rate": 1.7209586097592188e-07,
      "loss": 0.0021249458193778993,
      "reward": 3.0680684566497805,
      "reward_std": 0.3909214198589325,
      "rewards/IngredientFormatReward/mean": 0.9868749618530274,
      "rewards/IngredientFormatReward/std": 0.09705453515052795,
      "rewards/IngredientMatchReward/mean": 0.6739999413490295,
      "rewards/IngredientMatchReward/std": 0.27780568301677705,
      "rewards/IngredientQuantityMatchReward/mean": 0.6884435415267944,
      "rewards/IngredientQuantityMatchReward/std": 0.40112354755401614,
      "rewards/TotalKcalExactMatchReward/mean": 0.71875,
      "rewards/TotalKcalExactMatchReward/std": 0.44095770716667176,
      "step": 2665
    },
    {
      "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": 0.0046875,
      "completions/max_length": 504.6,
      "completions/mean_length": 394.878125,
      "completions/min_length": 276.6,
      "epoch": 0.7426981919332406,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6982694268226624,
      "kl": 0.04645406629424542,
      "learning_rate": 1.7036311724346358e-07,
      "loss": 0.0018581092357635499,
      "reward": 3.106524705886841,
      "reward_std": 0.3474747657775879,
      "rewards/IngredientFormatReward/mean": 0.9910416722297668,
      "rewards/IngredientFormatReward/std": 0.04977382682263851,
      "rewards/IngredientMatchReward/mean": 0.6606696367263794,
      "rewards/IngredientMatchReward/std": 0.25854469537734986,
      "rewards/IngredientQuantityMatchReward/mean": 0.7063759326934814,
      "rewards/IngredientQuantityMatchReward/std": 0.39123660922050474,
      "rewards/TotalKcalExactMatchReward/mean": 0.7484375,
      "rewards/TotalKcalExactMatchReward/std": 0.42715450525283816,
      "step": 2670
    },
    {
      "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": 0.0140625,
      "completions/max_length": 505.0,
      "completions/mean_length": 386.7671875,
      "completions/min_length": 250.2,
      "epoch": 0.7440890125173852,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6369605660438538,
      "kl": 0.048309245309792456,
      "learning_rate": 1.686373476945182e-07,
      "loss": 0.00193245280534029,
      "reward": 3.176955461502075,
      "reward_std": 0.39188035726547243,
      "rewards/IngredientFormatReward/mean": 0.9866666555404663,
      "rewards/IngredientFormatReward/std": 0.08516232818365096,
      "rewards/IngredientMatchReward/mean": 0.6687934160232544,
      "rewards/IngredientMatchReward/std": 0.28387183248996734,
      "rewards/IngredientQuantityMatchReward/mean": 0.6886829137802124,
      "rewards/IngredientQuantityMatchReward/std": 0.3933670938014984,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.36141787767410277,
      "step": 2675
    },
    {
      "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": 0.0125,
      "completions/max_length": 509.8,
      "completions/mean_length": 393.225,
      "completions/min_length": 255.0,
      "epoch": 0.7454798331015299,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6349705457687378,
      "kl": 0.051582214632071556,
      "learning_rate": 1.6691858884148525e-07,
      "loss": 0.0020633328706026076,
      "reward": 3.123720407485962,
      "reward_std": 0.38062790036201477,
      "rewards/IngredientFormatReward/mean": 0.9845535635948182,
      "rewards/IngredientFormatReward/std": 0.0967326819896698,
      "rewards/IngredientMatchReward/mean": 0.6521354436874389,
      "rewards/IngredientMatchReward/std": 0.28357810974121095,
      "rewards/IngredientQuantityMatchReward/mean": 0.7073439955711365,
      "rewards/IngredientQuantityMatchReward/std": 0.3868045091629028,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.40714595913887025,
      "step": 2680
    },
    {
      "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": 0.0125,
      "completions/max_length": 499.2,
      "completions/mean_length": 392.975,
      "completions/min_length": 281.0,
      "epoch": 0.7468706536856745,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6418431401252747,
      "kl": 0.04696134240366519,
      "learning_rate": 1.6520687704843761e-07,
      "loss": 0.001878587156534195,
      "reward": 3.185035562515259,
      "reward_std": 0.4062525689601898,
      "rewards/IngredientFormatReward/mean": 0.9868750095367431,
      "rewards/IngredientFormatReward/std": 0.07441277205944061,
      "rewards/IngredientMatchReward/mean": 0.6845523357391358,
      "rewards/IngredientMatchReward/std": 0.2672650098800659,
      "rewards/IngredientQuantityMatchReward/mean": 0.7026707291603088,
      "rewards/IngredientQuantityMatchReward/std": 0.41382738947868347,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.38031128644943235,
      "step": 2685
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.0,
      "completions/mean_length": 392.071875,
      "completions/min_length": 270.2,
      "epoch": 0.7482614742698191,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6632663607597351,
      "kl": 0.04900345194619149,
      "learning_rate": 1.6350224853035266e-07,
      "loss": 0.0019601194187998773,
      "reward": 3.096248960494995,
      "reward_std": 0.3981503129005432,
      "rewards/IngredientFormatReward/mean": 0.9906473159790039,
      "rewards/IngredientFormatReward/std": 0.07666670083999634,
      "rewards/IngredientMatchReward/mean": 0.6439918160438538,
      "rewards/IngredientMatchReward/std": 0.2905239939689636,
      "rewards/IngredientQuantityMatchReward/mean": 0.6991097688674927,
      "rewards/IngredientQuantityMatchReward/std": 0.41301283836364744,
      "rewards/TotalKcalExactMatchReward/mean": 0.7625,
      "rewards/TotalKcalExactMatchReward/std": 0.3906921327114105,
      "step": 2690
    },
    {
      "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": 0.003125,
      "completions/max_length": 487.2,
      "completions/mean_length": 384.54375,
      "completions/min_length": 256.6,
      "epoch": 0.7496522948539638,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6603626608848572,
      "kl": 0.046728350967168805,
      "learning_rate": 1.6180473935234508e-07,
      "loss": 0.001869351789355278,
      "reward": 3.1564847946166994,
      "reward_std": 0.4320804297924042,
      "rewards/IngredientFormatReward/mean": 0.9943750143051148,
      "rewards/IngredientFormatReward/std": 0.057214077562093735,
      "rewards/IngredientMatchReward/mean": 0.663671863079071,
      "rewards/IngredientMatchReward/std": 0.287596070766449,
      "rewards/IngredientQuantityMatchReward/mean": 0.6671879291534424,
      "rewards/IngredientQuantityMatchReward/std": 0.4132254719734192,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.367131644487381,
      "step": 2695
    },
    {
      "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": 0.009375,
      "completions/max_length": 512.6,
      "completions/mean_length": 393.4859375,
      "completions/min_length": 254.0,
      "epoch": 0.7510431154381085,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6157970428466797,
      "kl": 0.04571370454505086,
      "learning_rate": 1.6011438542890481e-07,
      "loss": 0.0018287433311343193,
      "reward": 3.0726105690002443,
      "reward_std": 0.3633168041706085,
      "rewards/IngredientFormatReward/mean": 0.9866666674613953,
      "rewards/IngredientFormatReward/std": 0.10891215056180954,
      "rewards/IngredientMatchReward/mean": 0.6230196475982666,
      "rewards/IngredientMatchReward/std": 0.29381694197654723,
      "rewards/IngredientQuantityMatchReward/mean": 0.6301117181777954,
      "rewards/IngredientQuantityMatchReward/std": 0.4268217861652374,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.3544869065284729,
      "step": 2700
    },
    {
      "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": 0.0078125,
      "completions/max_length": 502.0,
      "completions/mean_length": 390.8546875,
      "completions/min_length": 276.2,
      "epoch": 0.7524339360222532,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.6334781646728516,
      "kl": 0.04829642854165286,
      "learning_rate": 1.584312225231373e-07,
      "loss": 0.0019318748265504837,
      "reward": 3.1972563743591307,
      "reward_std": 0.3812137722969055,
      "rewards/IngredientFormatReward/mean": 0.9905580282211304,
      "rewards/IngredientFormatReward/std": 0.06447032019495964,
      "rewards/IngredientMatchReward/mean": 0.6358643293380737,
      "rewards/IngredientMatchReward/std": 0.29064218103885653,
      "rewards/IngredientQuantityMatchReward/mean": 0.6880215525627136,
      "rewards/IngredientQuantityMatchReward/std": 0.41757785677909853,
      "rewards/TotalKcalExactMatchReward/mean": 0.8828125,
      "rewards/TotalKcalExactMatchReward/std": 0.2939798399806023,
      "step": 2705
    },
    {
      "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": 0.0046875,
      "completions/max_length": 508.6,
      "completions/mean_length": 387.80625,
      "completions/min_length": 260.2,
      "epoch": 0.7538247566063978,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.662911057472229,
      "kl": 0.04823250975459814,
      "learning_rate": 1.5675528624600598e-07,
      "loss": 0.0019293729215860366,
      "reward": 3.137434196472168,
      "reward_std": 0.4250971019268036,
      "rewards/IngredientFormatReward/mean": 0.9915624976158142,
      "rewards/IngredientFormatReward/std": 0.06401846744120121,
      "rewards/IngredientMatchReward/mean": 0.6299597263336182,
      "rewards/IngredientMatchReward/std": 0.2888274848461151,
      "rewards/IngredientQuantityMatchReward/mean": 0.6893495202064515,
      "rewards/IngredientQuantityMatchReward/std": 0.39977465867996215,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.3759679079055786,
      "step": 2710
    },
    {
      "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": 0.0046875,
      "completions/max_length": 505.2,
      "completions/mean_length": 386.7421875,
      "completions/min_length": 259.0,
      "epoch": 0.7552155771905424,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6722151637077332,
      "kl": 0.046527455444447696,
      "learning_rate": 1.5508661205557898e-07,
      "loss": 0.0018611792474985123,
      "reward": 3.195002794265747,
      "reward_std": 0.3524541914463043,
      "rewards/IngredientFormatReward/mean": 0.9923214316368103,
      "rewards/IngredientFormatReward/std": 0.06965172737836837,
      "rewards/IngredientMatchReward/mean": 0.6588944554328918,
      "rewards/IngredientMatchReward/std": 0.2890048623085022,
      "rewards/IngredientQuantityMatchReward/mean": 0.7062869071960449,
      "rewards/IngredientQuantityMatchReward/std": 0.41538079380989074,
      "rewards/TotalKcalExactMatchReward/mean": 0.8375,
      "rewards/TotalKcalExactMatchReward/std": 0.3644015729427338,
      "step": 2715
    },
    {
      "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": 0.003125,
      "completions/max_length": 503.6,
      "completions/mean_length": 388.00625,
      "completions/min_length": 251.0,
      "epoch": 0.7566063977746871,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7072897553443909,
      "kl": 0.043907552887685594,
      "learning_rate": 1.534252352562797e-07,
      "loss": 0.0017563527449965476,
      "reward": 3.0336654663085936,
      "reward_std": 0.37396565079689026,
      "rewards/IngredientFormatReward/mean": 0.996875,
      "rewards/IngredientFormatReward/std": 0.03535533845424652,
      "rewards/IngredientMatchReward/mean": 0.6149045109748841,
      "rewards/IngredientMatchReward/std": 0.30236271023750305,
      "rewards/IngredientQuantityMatchReward/mean": 0.6406359076499939,
      "rewards/IngredientQuantityMatchReward/std": 0.42598314881324767,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.38697641491889956,
      "step": 2720
    },
    {
      "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": 0.009375,
      "completions/max_length": 511.8,
      "completions/mean_length": 391.503125,
      "completions/min_length": 258.2,
      "epoch": 0.7579972183588317,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6333484053611755,
      "kl": 0.04943839656189084,
      "learning_rate": 1.5177119099813922e-07,
      "loss": 0.0019777391105890276,
      "reward": 3.254866600036621,
      "reward_std": 0.36433430314064025,
      "rewards/IngredientFormatReward/mean": 0.9916666626930237,
      "rewards/IngredientFormatReward/std": 0.06988214328885078,
      "rewards/IngredientMatchReward/mean": 0.716341781616211,
      "rewards/IngredientMatchReward/std": 0.2796147793531418,
      "rewards/IngredientQuantityMatchReward/mean": 0.7327957391738892,
      "rewards/IngredientQuantityMatchReward/std": 0.37077197432518005,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.386597740650177,
      "step": 2725
    },
    {
      "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": 0.0046875,
      "completions/max_length": 504.8,
      "completions/mean_length": 385.6921875,
      "completions/min_length": 246.0,
      "epoch": 0.7593880389429764,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.669300377368927,
      "kl": 0.04992357234004885,
      "learning_rate": 1.5012451427605295e-07,
      "loss": 0.0019968684762716295,
      "reward": 3.1358933448791504,
      "reward_std": 0.3479516267776489,
      "rewards/IngredientFormatReward/mean": 0.9936979055404663,
      "rewards/IngredientFormatReward/std": 0.057158236391842365,
      "rewards/IngredientMatchReward/mean": 0.6550303936004639,
      "rewards/IngredientMatchReward/std": 0.2722333997488022,
      "rewards/IngredientQuantityMatchReward/mean": 0.659040093421936,
      "rewards/IngredientQuantityMatchReward/std": 0.4142662465572357,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3717784404754639,
      "step": 2730
    },
    {
      "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": 0.0109375,
      "completions/max_length": 508.6,
      "completions/mean_length": 389.209375,
      "completions/min_length": 242.4,
      "epoch": 0.760778859527121,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6446650624275208,
      "kl": 0.04634713255800307,
      "learning_rate": 1.4848523992903978e-07,
      "loss": 0.0018548401072621345,
      "reward": 3.045489501953125,
      "reward_std": 0.43291773796081545,
      "rewards/IngredientFormatReward/mean": 0.9876562476158142,
      "rewards/IngredientFormatReward/std": 0.09445627927780151,
      "rewards/IngredientMatchReward/mean": 0.6676263451576233,
      "rewards/IngredientMatchReward/std": 0.2896029084920883,
      "rewards/IngredientQuantityMatchReward/mean": 0.6136444568634033,
      "rewards/IngredientQuantityMatchReward/std": 0.4410985291004181,
      "rewards/TotalKcalExactMatchReward/mean": 0.7765625,
      "rewards/TotalKcalExactMatchReward/std": 0.4121951341629028,
      "step": 2735
    },
    {
      "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": 0.00625,
      "completions/max_length": 507.2,
      "completions/mean_length": 395.109375,
      "completions/min_length": 274.0,
      "epoch": 0.7621696801112656,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6219557523727417,
      "kl": 0.045795883238315585,
      "learning_rate": 1.468534026395056e-07,
      "loss": 0.0018320389091968537,
      "reward": 3.003160810470581,
      "reward_std": 0.3470380902290344,
      "rewards/IngredientFormatReward/mean": 0.9934375047683716,
      "rewards/IngredientFormatReward/std": 0.0515897773206234,
      "rewards/IngredientMatchReward/mean": 0.6431274652481079,
      "rewards/IngredientMatchReward/std": 0.2938917726278305,
      "rewards/IngredientQuantityMatchReward/mean": 0.5962833166122437,
      "rewards/IngredientQuantityMatchReward/std": 0.43143598437309266,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.4162842929363251,
      "step": 2740
    },
    {
      "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": 0.0078125,
      "completions/max_length": 511.2,
      "completions/mean_length": 387.2671875,
      "completions/min_length": 268.6,
      "epoch": 0.7635605006954103,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6196995377540588,
      "kl": 0.046195710496976974,
      "learning_rate": 1.4522903693250904e-07,
      "loss": 0.0018479309976100922,
      "reward": 3.1460134029388427,
      "reward_std": 0.4049987256526947,
      "rewards/IngredientFormatReward/mean": 0.9915364503860473,
      "rewards/IngredientFormatReward/std": 0.07865389734506607,
      "rewards/IngredientMatchReward/mean": 0.6382266998291015,
      "rewards/IngredientMatchReward/std": 0.29503755569458007,
      "rewards/IngredientQuantityMatchReward/mean": 0.7006252765655517,
      "rewards/IngredientQuantityMatchReward/std": 0.3964582920074463,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.3763843536376953,
      "step": 2745
    },
    {
      "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": 0.003125,
      "completions/max_length": 502.8,
      "completions/mean_length": 386.3515625,
      "completions/min_length": 249.8,
      "epoch": 0.7649513212795549,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6867331862449646,
      "kl": 0.04726094771176577,
      "learning_rate": 1.4361217717503143e-07,
      "loss": 0.0018905647099018096,
      "reward": 3.0409469604492188,
      "reward_std": 0.36254254579544065,
      "rewards/IngredientFormatReward/mean": 0.996875,
      "rewards/IngredientFormatReward/std": 0.02490137964487076,
      "rewards/IngredientMatchReward/mean": 0.5823530673980712,
      "rewards/IngredientMatchReward/std": 0.313857102394104,
      "rewards/IngredientQuantityMatchReward/mean": 0.6867189168930053,
      "rewards/IngredientQuantityMatchReward/std": 0.4137806832790375,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.41760230660438535,
      "step": 2750
    },
    {
      "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": 0.0046875,
      "completions/max_length": 504.2,
      "completions/mean_length": 400.6734375,
      "completions/min_length": 285.4,
      "epoch": 0.7663421418636995,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6223328113555908,
      "kl": 0.048341109650209546,
      "learning_rate": 1.4200285757524894e-07,
      "loss": 0.0019336594268679619,
      "reward": 3.2103977680206297,
      "reward_std": 0.3802697777748108,
      "rewards/IngredientFormatReward/mean": 0.9953125,
      "rewards/IngredientFormatReward/std": 0.04257904887199402,
      "rewards/IngredientMatchReward/mean": 0.6562047481536866,
      "rewards/IngredientMatchReward/std": 0.2683935105800629,
      "rewards/IngredientQuantityMatchReward/mean": 0.7104430794715881,
      "rewards/IngredientQuantityMatchReward/std": 0.39671446681022643,
      "rewards/TotalKcalExactMatchReward/mean": 0.8484375,
      "rewards/TotalKcalExactMatchReward/std": 0.35886216163635254,
      "step": 2755
    },
    {
      "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": 0.0015625,
      "completions/max_length": 500.0,
      "completions/mean_length": 388.2171875,
      "completions/min_length": 250.4,
      "epoch": 0.7677329624478443,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6476879715919495,
      "kl": 0.04689363501966,
      "learning_rate": 1.4040111218180966e-07,
      "loss": 0.0018757812678813935,
      "reward": 3.1677744388580322,
      "reward_std": 0.34500511884689333,
      "rewards/IngredientFormatReward/mean": 0.9977306485176086,
      "rewards/IngredientFormatReward/std": 0.021293053217232228,
      "rewards/IngredientMatchReward/mean": 0.6576048016548157,
      "rewards/IngredientMatchReward/std": 0.27767792642116546,
      "rewards/IngredientQuantityMatchReward/mean": 0.6718139767646789,
      "rewards/IngredientQuantityMatchReward/std": 0.40898163318634034,
      "rewards/TotalKcalExactMatchReward/mean": 0.840625,
      "rewards/TotalKcalExactMatchReward/std": 0.3573881030082703,
      "step": 2760
    },
    {
      "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": 0.003125,
      "completions/max_length": 500.2,
      "completions/mean_length": 386.4,
      "completions/min_length": 268.8,
      "epoch": 0.7691237830319889,
      "frac_reward_zero_std": 0.15,
      "grad_norm": 0.6938906908035278,
      "kl": 0.046820966992527246,
      "learning_rate": 1.3880697488311326e-07,
      "loss": 0.0018731381744146347,
      "reward": 3.158130073547363,
      "reward_std": 0.3146629989147186,
      "rewards/IngredientFormatReward/mean": 0.9966257452964783,
      "rewards/IngredientFormatReward/std": 0.03063418995589018,
      "rewards/IngredientMatchReward/mean": 0.658064866065979,
      "rewards/IngredientMatchReward/std": 0.2844700157642365,
      "rewards/IngredientQuantityMatchReward/mean": 0.6815645098686218,
      "rewards/IngredientQuantityMatchReward/std": 0.40854601860046386,
      "rewards/TotalKcalExactMatchReward/mean": 0.821875,
      "rewards/TotalKcalExactMatchReward/std": 0.33308517932891846,
      "step": 2765
    },
    {
      "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": 0.0015625,
      "completions/max_length": 507.6,
      "completions/mean_length": 389.0953125,
      "completions/min_length": 279.4,
      "epoch": 0.7705146036161336,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.6388007402420044,
      "kl": 0.04836219367571175,
      "learning_rate": 1.3722047940659326e-07,
      "loss": 0.0019346587359905243,
      "reward": 3.077744960784912,
      "reward_std": 0.3748651802539825,
      "rewards/IngredientFormatReward/mean": 0.9966145753860474,
      "rewards/IngredientFormatReward/std": 0.03830161709338427,
      "rewards/IngredientMatchReward/mean": 0.6705307602882385,
      "rewards/IngredientMatchReward/std": 0.25971758365631104,
      "rewards/IngredientQuantityMatchReward/mean": 0.6543496251106262,
      "rewards/IngredientQuantityMatchReward/std": 0.4099151134490967,
      "rewards/TotalKcalExactMatchReward/mean": 0.75625,
      "rewards/TotalKcalExactMatchReward/std": 0.4248959481716156,
      "step": 2770
    },
    {
      "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": 0.0078125,
      "completions/max_length": 496.8,
      "completions/mean_length": 384.6,
      "completions/min_length": 256.0,
      "epoch": 0.7719054242002782,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7773398160934448,
      "kl": 0.05296712294220925,
      "learning_rate": 1.356416593180036e-07,
      "loss": 0.0021185621619224547,
      "reward": 3.0785492420196534,
      "reward_std": 0.3721375286579132,
      "rewards/IngredientFormatReward/mean": 0.9884579658508301,
      "rewards/IngredientFormatReward/std": 0.06780673321336508,
      "rewards/IngredientMatchReward/mean": 0.6230674982070923,
      "rewards/IngredientMatchReward/std": 0.2801555782556534,
      "rewards/IngredientQuantityMatchReward/mean": 0.6560862421989441,
      "rewards/IngredientQuantityMatchReward/std": 0.39724615812301634,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.38468171954154967,
      "step": 2775
    },
    {
      "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": 0.0109375,
      "completions/max_length": 506.8,
      "completions/mean_length": 390.98125,
      "completions/min_length": 266.4,
      "epoch": 0.7732962447844228,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6303114891052246,
      "kl": 0.048468559980392456,
      "learning_rate": 1.3407054802070923e-07,
      "loss": 0.0019390463829040527,
      "reward": 3.252069044113159,
      "reward_std": 0.3501974791288376,
      "rewards/IngredientFormatReward/mean": 0.983392858505249,
      "rewards/IngredientFormatReward/std": 0.09273147732019424,
      "rewards/IngredientMatchReward/mean": 0.7026979684829712,
      "rewards/IngredientMatchReward/std": 0.2675772845745087,
      "rewards/IngredientQuantityMatchReward/mean": 0.7081657171249389,
      "rewards/IngredientQuantityMatchReward/std": 0.37891228795051574,
      "rewards/TotalKcalExactMatchReward/mean": 0.8578125,
      "rewards/TotalKcalExactMatchReward/std": 0.34481988549232484,
      "step": 2780
    },
    {
      "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": 0.009375,
      "completions/max_length": 513.0,
      "completions/mean_length": 393.4015625,
      "completions/min_length": 269.6,
      "epoch": 0.7746870653685675,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7151004076004028,
      "kl": 0.04872276636306196,
      "learning_rate": 1.3250717875497864e-07,
      "loss": 0.0019489120692014693,
      "reward": 3.0865904331207275,
      "reward_std": 0.42741536498069765,
      "rewards/IngredientFormatReward/mean": 0.9932663679122925,
      "rewards/IngredientFormatReward/std": 0.0710913971066475,
      "rewards/IngredientMatchReward/mean": 0.6273127555847168,
      "rewards/IngredientMatchReward/std": 0.29151912927627566,
      "rewards/IngredientQuantityMatchReward/mean": 0.6535112500190735,
      "rewards/IngredientQuantityMatchReward/std": 0.4204251229763031,
      "rewards/TotalKcalExactMatchReward/mean": 0.8125,
      "rewards/TotalKcalExactMatchReward/std": 0.38339470624923705,
      "step": 2785
    },
    {
      "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": 0.00625,
      "completions/max_length": 510.8,
      "completions/mean_length": 385.0453125,
      "completions/min_length": 271.8,
      "epoch": 0.7760778859527121,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6237897276878357,
      "kl": 0.04654221590608358,
      "learning_rate": 1.3095158459728088e-07,
      "loss": 0.0018617745488882064,
      "reward": 3.0792617797851562,
      "reward_std": 0.37338990569114683,
      "rewards/IngredientFormatReward/mean": 0.993359375,
      "rewards/IngredientFormatReward/std": 0.07122094184160233,
      "rewards/IngredientMatchReward/mean": 0.680286455154419,
      "rewards/IngredientMatchReward/std": 0.2871041715145111,
      "rewards/IngredientQuantityMatchReward/mean": 0.6446785092353821,
      "rewards/IngredientQuantityMatchReward/std": 0.39799618124961855,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.4222764551639557,
      "step": 2790
    },
    {
      "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": 0.0109375,
      "completions/max_length": 504.4,
      "completions/mean_length": 385.3328125,
      "completions/min_length": 267.0,
      "epoch": 0.7774687065368567,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6749888062477112,
      "kl": 0.06970281926915049,
      "learning_rate": 1.2940379845958588e-07,
      "loss": 0.0027874993160367013,
      "reward": 3.0723299026489257,
      "reward_std": 0.42997482419013977,
      "rewards/IngredientFormatReward/mean": 0.9851785659790039,
      "rewards/IngredientFormatReward/std": 0.10947459004819393,
      "rewards/IngredientMatchReward/mean": 0.621602201461792,
      "rewards/IngredientMatchReward/std": 0.2980756342411041,
      "rewards/IngredientQuantityMatchReward/mean": 0.6827366828918457,
      "rewards/IngredientQuantityMatchReward/std": 0.42669169306755067,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.39410466253757476,
      "step": 2795
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.6,
      "completions/mean_length": 390.4203125,
      "completions/min_length": 274.2,
      "epoch": 0.7788595271210014,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6576166152954102,
      "kl": 0.04848268665373325,
      "learning_rate": 1.278638530886677e-07,
      "loss": 0.00193945299834013,
      "reward": 3.1979696273803713,
      "reward_std": 0.3592884033918381,
      "rewards/IngredientFormatReward/mean": 0.9953125,
      "rewards/IngredientFormatReward/std": 0.04257904887199402,
      "rewards/IngredientMatchReward/mean": 0.6500452756881714,
      "rewards/IngredientMatchReward/std": 0.2857871323823929,
      "rewards/IngredientQuantityMatchReward/mean": 0.6838618040084838,
      "rewards/IngredientQuantityMatchReward/std": 0.4050181329250336,
      "rewards/TotalKcalExactMatchReward/mean": 0.86875,
      "rewards/TotalKcalExactMatchReward/std": 0.31754682660102845,
      "step": 2800
    },
    {
      "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": 0.009375,
      "completions/max_length": 506.2,
      "completions/mean_length": 387.875,
      "completions/min_length": 247.4,
      "epoch": 0.780250347705146,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7045865654945374,
      "kl": 0.045610047527588904,
      "learning_rate": 1.2633178106541214e-07,
      "loss": 0.0018244288861751557,
      "reward": 3.0865010738372805,
      "reward_std": 0.40377230048179624,
      "rewards/IngredientFormatReward/mean": 0.9901041626930237,
      "rewards/IngredientFormatReward/std": 0.05828761123120785,
      "rewards/IngredientMatchReward/mean": 0.6850093126296997,
      "rewards/IngredientMatchReward/std": 0.2759521782398224,
      "rewards/IngredientQuantityMatchReward/mean": 0.623887574672699,
      "rewards/IngredientQuantityMatchReward/std": 0.410383015871048,
      "rewards/TotalKcalExactMatchReward/mean": 0.7875,
      "rewards/TotalKcalExactMatchReward/std": 0.406889545917511,
      "step": 2805
    },
    {
      "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": 0.003125,
      "completions/max_length": 488.6,
      "completions/mean_length": 382.325,
      "completions/min_length": 264.8,
      "epoch": 0.7816411682892906,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7042024731636047,
      "kl": 0.04793581950943917,
      "learning_rate": 1.2480761480412738e-07,
      "loss": 0.0019172405824065208,
      "reward": 3.1829829692840574,
      "reward_std": 0.35909828543663025,
      "rewards/IngredientFormatReward/mean": 0.9947916746139527,
      "rewards/IngredientFormatReward/std": 0.05394516885280609,
      "rewards/IngredientMatchReward/mean": 0.6396100044250488,
      "rewards/IngredientMatchReward/std": 0.29437714219093325,
      "rewards/IngredientQuantityMatchReward/mean": 0.7001438021659852,
      "rewards/IngredientQuantityMatchReward/std": 0.4026173710823059,
      "rewards/TotalKcalExactMatchReward/mean": 0.8484375,
      "rewards/TotalKcalExactMatchReward/std": 0.3493222385644913,
      "step": 2810
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.2,
      "completions/mean_length": 388.528125,
      "completions/min_length": 253.4,
      "epoch": 0.7830319888734353,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6657385230064392,
      "kl": 0.049304631818085906,
      "learning_rate": 1.2329138655185733e-07,
      "loss": 0.0019720766693353655,
      "reward": 3.040740966796875,
      "reward_std": 0.4243003189563751,
      "rewards/IngredientFormatReward/mean": 0.9796316981315613,
      "rewards/IngredientFormatReward/std": 0.10241263806819915,
      "rewards/IngredientMatchReward/mean": 0.6618905305862427,
      "rewards/IngredientMatchReward/std": 0.31010615825653076,
      "rewards/IngredientQuantityMatchReward/mean": 0.6554687976837158,
      "rewards/IngredientQuantityMatchReward/std": 0.43676078915596006,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.4280146896839142,
      "step": 2815
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.4,
      "completions/mean_length": 390.76875,
      "completions/min_length": 253.0,
      "epoch": 0.7844228094575799,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6594431400299072,
      "kl": 0.054166854242794216,
      "learning_rate": 1.2178312838770115e-07,
      "loss": 0.002167240343987942,
      "reward": 3.222614049911499,
      "reward_std": 0.3778720319271088,
      "rewards/IngredientFormatReward/mean": 0.9932291746139527,
      "rewards/IngredientFormatReward/std": 0.07162283807992935,
      "rewards/IngredientMatchReward/mean": 0.6805189847946167,
      "rewards/IngredientMatchReward/std": 0.280741947889328,
      "rewards/IngredientQuantityMatchReward/mean": 0.6848033666610718,
      "rewards/IngredientQuantityMatchReward/std": 0.38373754024505613,
      "rewards/TotalKcalExactMatchReward/mean": 0.8640625,
      "rewards/TotalKcalExactMatchReward/std": 0.2921728253364563,
      "step": 2820
    },
    {
      "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": 0.009375,
      "completions/max_length": 502.8,
      "completions/mean_length": 395.3796875,
      "completions/min_length": 272.4,
      "epoch": 0.7858136300417247,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6570203304290771,
      "kl": 0.04562732768245041,
      "learning_rate": 1.2028287222213286e-07,
      "loss": 0.0018250590190291404,
      "reward": 3.0641520023345947,
      "reward_std": 0.3975339472293854,
      "rewards/IngredientFormatReward/mean": 0.989661455154419,
      "rewards/IngredientFormatReward/std": 0.06459243036806583,
      "rewards/IngredientMatchReward/mean": 0.6596819043159485,
      "rewards/IngredientMatchReward/std": 0.2889078378677368,
      "rewards/IngredientQuantityMatchReward/mean": 0.6054336071014405,
      "rewards/IngredientQuantityMatchReward/std": 0.4233911633491516,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.3768185257911682,
      "step": 2825
    },
    {
      "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": 0.003125,
      "completions/max_length": 497.8,
      "completions/mean_length": 388.7671875,
      "completions/min_length": 262.2,
      "epoch": 0.7872044506258693,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 2.9295973777770996,
      "kl": 0.06034769716206938,
      "learning_rate": 1.1879064979632702e-07,
      "loss": 0.0024134255945682526,
      "reward": 2.941163158416748,
      "reward_std": 0.3750101804733276,
      "rewards/IngredientFormatReward/mean": 0.9932291746139527,
      "rewards/IngredientFormatReward/std": 0.04726078063249588,
      "rewards/IngredientMatchReward/mean": 0.5866058945655823,
      "rewards/IngredientMatchReward/std": 0.3109011113643646,
      "rewards/IngredientQuantityMatchReward/mean": 0.6300781726837158,
      "rewards/IngredientQuantityMatchReward/std": 0.42629345059394835,
      "rewards/TotalKcalExactMatchReward/mean": 0.73125,
      "rewards/TotalKcalExactMatchReward/std": 0.43264933228492736,
      "step": 2830
    },
    {
      "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": 0.00625,
      "completions/max_length": 504.4,
      "completions/mean_length": 386.409375,
      "completions/min_length": 260.8,
      "epoch": 0.7885952712100139,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6922607421875,
      "kl": 0.04646100748796016,
      "learning_rate": 1.1730649268148662e-07,
      "loss": 0.0018586434423923492,
      "reward": 2.9923593521118166,
      "reward_std": 0.43100237250328066,
      "rewards/IngredientFormatReward/mean": 0.9923437356948852,
      "rewards/IngredientFormatReward/std": 0.05373183339834213,
      "rewards/IngredientMatchReward/mean": 0.6438275218009949,
      "rewards/IngredientMatchReward/std": 0.27490496039390566,
      "rewards/IngredientQuantityMatchReward/mean": 0.6374380946159363,
      "rewards/IngredientQuantityMatchReward/std": 0.41457298398017883,
      "rewards/TotalKcalExactMatchReward/mean": 0.71875,
      "rewards/TotalKcalExactMatchReward/std": 0.4473869621753693,
      "step": 2835
    },
    {
      "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": 0.009375,
      "completions/max_length": 505.8,
      "completions/mean_length": 385.346875,
      "completions/min_length": 264.2,
      "epoch": 0.7899860917941586,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5764864683151245,
      "kl": 0.06799533735029399,
      "learning_rate": 1.1583043227817607e-07,
      "loss": 0.0027182653546333315,
      "reward": 3.036129665374756,
      "reward_std": 0.4375645399093628,
      "rewards/IngredientFormatReward/mean": 0.985531997680664,
      "rewards/IngredientFormatReward/std": 0.0772362755611539,
      "rewards/IngredientMatchReward/mean": 0.6238765120506287,
      "rewards/IngredientMatchReward/std": 0.3162325620651245,
      "rewards/IngredientQuantityMatchReward/mean": 0.6470335960388184,
      "rewards/IngredientQuantityMatchReward/std": 0.4036581993103027,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.41052433252334597,
      "step": 2840
    },
    {
      "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": 0.0046875,
      "completions/max_length": 501.6,
      "completions/mean_length": 382.0109375,
      "completions/min_length": 247.8,
      "epoch": 0.7913769123783032,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6719501614570618,
      "kl": 0.046000178135000167,
      "learning_rate": 1.1436249981565576e-07,
      "loss": 0.0018522294238209724,
      "reward": 3.1666172981262206,
      "reward_std": 0.3686310827732086,
      "rewards/IngredientFormatReward/mean": 0.9954166531562805,
      "rewards/IngredientFormatReward/std": 0.047038369067013266,
      "rewards/IngredientMatchReward/mean": 0.6741952061653137,
      "rewards/IngredientMatchReward/std": 0.2847984731197357,
      "rewards/IngredientQuantityMatchReward/mean": 0.7157554388046264,
      "rewards/IngredientQuantityMatchReward/std": 0.4073028028011322,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.40991033911705016,
      "step": 2845
    },
    {
      "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": 0.003125,
      "completions/max_length": 503.0,
      "completions/mean_length": 388.4515625,
      "completions/min_length": 243.4,
      "epoch": 0.7927677329624478,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.672032356262207,
      "kl": 0.04769486370496452,
      "learning_rate": 1.1290272635122256e-07,
      "loss": 0.0019078360870480537,
      "reward": 2.9494760036468506,
      "reward_std": 0.3965733826160431,
      "rewards/IngredientFormatReward/mean": 0.9865625023841857,
      "rewards/IngredientFormatReward/std": 0.08309870734810829,
      "rewards/IngredientMatchReward/mean": 0.5798406481742859,
      "rewards/IngredientMatchReward/std": 0.2907747864723206,
      "rewards/IngredientQuantityMatchReward/mean": 0.6393229246139527,
      "rewards/IngredientQuantityMatchReward/std": 0.42374877333641053,
      "rewards/TotalKcalExactMatchReward/mean": 0.74375,
      "rewards/TotalKcalExactMatchReward/std": 0.41585314869880674,
      "step": 2850
    },
    {
      "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": 0.0078125,
      "completions/max_length": 503.6,
      "completions/mean_length": 388.7546875,
      "completions/min_length": 252.8,
      "epoch": 0.7941585535465925,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6961467266082764,
      "kl": 0.048885095841251315,
      "learning_rate": 1.114511427695512e-07,
      "loss": 0.0019551247358322145,
      "reward": 3.0520843505859374,
      "reward_std": 0.45254541039466856,
      "rewards/IngredientFormatReward/mean": 0.9900892972946167,
      "rewards/IngredientFormatReward/std": 0.07334856688976288,
      "rewards/IngredientMatchReward/mean": 0.6228714346885681,
      "rewards/IngredientMatchReward/std": 0.2974251747131348,
      "rewards/IngredientQuantityMatchReward/mean": 0.6313111305236816,
      "rewards/IngredientQuantityMatchReward/std": 0.42033830285072327,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.39325725436210635,
      "step": 2855
    },
    {
      "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": 0.009375,
      "completions/max_length": 504.8,
      "completions/mean_length": 387.7484375,
      "completions/min_length": 264.2,
      "epoch": 0.7955493741307371,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.7616549134254456,
      "kl": 0.04823335437104106,
      "learning_rate": 1.1000777978204211e-07,
      "loss": 0.0019295023754239081,
      "reward": 3.0298017501831054,
      "reward_std": 0.38049039244651794,
      "rewards/IngredientFormatReward/mean": 0.9914434432983399,
      "rewards/IngredientFormatReward/std": 0.0785360500216484,
      "rewards/IngredientMatchReward/mean": 0.6197124242782592,
      "rewards/IngredientMatchReward/std": 0.2868492126464844,
      "rewards/IngredientQuantityMatchReward/mean": 0.6389583468437194,
      "rewards/IngredientQuantityMatchReward/std": 0.42402172088623047,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.40851550102233886,
      "step": 2860
    },
    {
      "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": 0.015625,
      "completions/max_length": 505.8,
      "completions/mean_length": 393.36875,
      "completions/min_length": 271.6,
      "epoch": 0.7969401947148818,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.647112250328064,
      "kl": 0.045586750074289736,
      "learning_rate": 1.0857266792617121e-07,
      "loss": 0.0018235519528388977,
      "reward": 3.1301424980163572,
      "reward_std": 0.36510835886001586,
      "rewards/IngredientFormatReward/mean": 0.9830729126930237,
      "rewards/IngredientFormatReward/std": 0.09126228243112564,
      "rewards/IngredientMatchReward/mean": 0.6374433040618896,
      "rewards/IngredientMatchReward/std": 0.30091888308525083,
      "rewards/IngredientQuantityMatchReward/mean": 0.6236887693405151,
      "rewards/IngredientQuantityMatchReward/std": 0.42500794529914854,
      "rewards/TotalKcalExactMatchReward/mean": 0.8859375,
      "rewards/TotalKcalExactMatchReward/std": 0.30419874787330625,
      "step": 2865
    },
    {
      "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": 0.0015625,
      "completions/max_length": 494.6,
      "completions/mean_length": 386.471875,
      "completions/min_length": 238.8,
      "epoch": 0.7983310152990264,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.665838897228241,
      "kl": 0.0487892878241837,
      "learning_rate": 1.071458375648438e-07,
      "loss": 0.001963186264038086,
      "reward": 3.08637056350708,
      "reward_std": 0.36597791910171507,
      "rewards/IngredientFormatReward/mean": 0.9940885305404663,
      "rewards/IngredientFormatReward/std": 0.06024602744728327,
      "rewards/IngredientMatchReward/mean": 0.6441040515899659,
      "rewards/IngredientMatchReward/std": 0.2958334803581238,
      "rewards/IngredientQuantityMatchReward/mean": 0.6200530052185058,
      "rewards/IngredientQuantityMatchReward/std": 0.4220093309879303,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.33209551125764847,
      "step": 2870
    },
    {
      "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": 0.021875,
      "completions/max_length": 511.0,
      "completions/mean_length": 393.8234375,
      "completions/min_length": 266.0,
      "epoch": 0.799721835883171,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6649406552314758,
      "kl": 0.04649213703814894,
      "learning_rate": 1.0572731888575209e-07,
      "loss": 0.0018598034977912903,
      "reward": 3.0285863399505617,
      "reward_std": 0.40862372517585754,
      "rewards/IngredientFormatReward/mean": 0.9753236532211303,
      "rewards/IngredientFormatReward/std": 0.1264018550515175,
      "rewards/IngredientMatchReward/mean": 0.6302784562110901,
      "rewards/IngredientMatchReward/std": 0.30015496611595155,
      "rewards/IngredientQuantityMatchReward/mean": 0.65423424243927,
      "rewards/IngredientQuantityMatchReward/std": 0.4166111648082733,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.4195781469345093,
      "step": 2875
    },
    {
      "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": 0.0109375,
      "completions/max_length": 506.0,
      "completions/mean_length": 389.603125,
      "completions/min_length": 238.4,
      "epoch": 0.8011126564673157,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6201867461204529,
      "kl": 0.046519194776192305,
      "learning_rate": 1.043171419007367e-07,
      "loss": 0.0018607662990689279,
      "reward": 3.052107810974121,
      "reward_std": 0.4012625515460968,
      "rewards/IngredientFormatReward/mean": 0.9785788774490356,
      "rewards/IngredientFormatReward/std": 0.0843801498413086,
      "rewards/IngredientMatchReward/mean": 0.6381122231483459,
      "rewards/IngredientMatchReward/std": 0.30227981209754945,
      "rewards/IngredientQuantityMatchReward/mean": 0.6541667699813842,
      "rewards/IngredientQuantityMatchReward/std": 0.43643729090690614,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.40987553000450133,
      "step": 2880
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.2,
      "completions/mean_length": 390.0890625,
      "completions/min_length": 273.0,
      "epoch": 0.8025034770514604,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5778425335884094,
      "kl": 0.04776061798911542,
      "learning_rate": 1.0291533644515166e-07,
      "loss": 0.0019105035811662675,
      "reward": 3.0733968734741213,
      "reward_std": 0.42165454030036925,
      "rewards/IngredientFormatReward/mean": 0.987287950515747,
      "rewards/IngredientFormatReward/std": 0.0955145888030529,
      "rewards/IngredientMatchReward/mean": 0.6008587718009949,
      "rewards/IngredientMatchReward/std": 0.28134148716926577,
      "rewards/IngredientQuantityMatchReward/mean": 0.6790001511573791,
      "rewards/IngredientQuantityMatchReward/std": 0.4184362173080444,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3811137080192566,
      "step": 2885
    },
    {
      "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": 0.0078125,
      "completions/max_length": 512.4,
      "completions/mean_length": 389.5046875,
      "completions/min_length": 259.0,
      "epoch": 0.803894297635605,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7275820374488831,
      "kl": 0.048515303037129344,
      "learning_rate": 1.0152193217723315e-07,
      "loss": 0.0019406870007514954,
      "reward": 3.0267982006073,
      "reward_std": 0.399430650472641,
      "rewards/IngredientFormatReward/mean": 0.9880431652069092,
      "rewards/IngredientFormatReward/std": 0.09137556552886963,
      "rewards/IngredientMatchReward/mean": 0.6249008059501648,
      "rewards/IngredientMatchReward/std": 0.29404258728027344,
      "rewards/IngredientQuantityMatchReward/mean": 0.6091667413711548,
      "rewards/IngredientQuantityMatchReward/std": 0.4240457832813263,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.3877830386161804,
      "step": 2890
    },
    {
      "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": 0.0078125,
      "completions/max_length": 506.2,
      "completions/mean_length": 394.7546875,
      "completions/min_length": 278.8,
      "epoch": 0.8052851182197497,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6058497428894043,
      "kl": 0.046461883233860135,
      "learning_rate": 1.0013695857747172e-07,
      "loss": 0.0018587440252304077,
      "reward": 3.025787353515625,
      "reward_std": 0.36919748187065127,
      "rewards/IngredientFormatReward/mean": 0.9911681652069092,
      "rewards/IngredientFormatReward/std": 0.07216834891587495,
      "rewards/IngredientMatchReward/mean": 0.623315978050232,
      "rewards/IngredientMatchReward/std": 0.2807840764522552,
      "rewards/IngredientQuantityMatchReward/mean": 0.622240686416626,
      "rewards/IngredientQuantityMatchReward/std": 0.42640805840492246,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.3843976229429245,
      "step": 2895
    },
    {
      "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": 0.0125,
      "completions/max_length": 502.2,
      "completions/mean_length": 388.85625,
      "completions/min_length": 271.2,
      "epoch": 0.8066759388038943,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 113.02948760986328,
      "kl": 0.6814675867091864,
      "learning_rate": 9.876044494798897e-08,
      "loss": 0.027234816551208497,
      "reward": 3.025029420852661,
      "reward_std": 0.386387437582016,
      "rewards/IngredientFormatReward/mean": 0.990833330154419,
      "rewards/IngredientFormatReward/std": 0.0677767276763916,
      "rewards/IngredientMatchReward/mean": 0.6841958165168762,
      "rewards/IngredientMatchReward/std": 0.29024302363395693,
      "rewards/IngredientQuantityMatchReward/mean": 0.6140627264976501,
      "rewards/IngredientQuantityMatchReward/std": 0.4360509991645813,
      "rewards/TotalKcalExactMatchReward/mean": 0.7359375,
      "rewards/TotalKcalExactMatchReward/std": 0.4356089770793915,
      "step": 2900
    },
    {
      "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": 0.0046875,
      "completions/max_length": 503.8,
      "completions/mean_length": 393.8546875,
      "completions/min_length": 285.6,
      "epoch": 0.808066759388039,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.6261676549911499,
      "kl": 0.04418616164475679,
      "learning_rate": 9.73924204119178e-08,
      "loss": 0.001767529733479023,
      "reward": 3.1210771560668946,
      "reward_std": 0.36356693506240845,
      "rewards/IngredientFormatReward/mean": 0.9943750023841857,
      "rewards/IngredientFormatReward/std": 0.04674905762076378,
      "rewards/IngredientMatchReward/mean": 0.6692689657211304,
      "rewards/IngredientMatchReward/std": 0.2812860310077667,
      "rewards/IngredientQuantityMatchReward/mean": 0.6855581760406494,
      "rewards/IngredientQuantityMatchReward/std": 0.3925300002098083,
      "rewards/TotalKcalExactMatchReward/mean": 0.771875,
      "rewards/TotalKcalExactMatchReward/std": 0.4036038815975189,
      "step": 2905
    },
    {
      "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": 0.0046875,
      "completions/max_length": 491.2,
      "completions/mean_length": 379.5515625,
      "completions/min_length": 250.4,
      "epoch": 0.8094575799721836,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7257301807403564,
      "kl": 0.04759784871712327,
      "learning_rate": 9.603291391278568e-08,
      "loss": 0.0019040904939174653,
      "reward": 3.122410774230957,
      "reward_std": 0.39100813269615176,
      "rewards/IngredientFormatReward/mean": 0.9928645849227905,
      "rewards/IngredientFormatReward/std": 0.0445655882358551,
      "rewards/IngredientMatchReward/mean": 0.6938579082489014,
      "rewards/IngredientMatchReward/std": 0.28338626623153684,
      "rewards/IngredientQuantityMatchReward/mean": 0.6185007572174073,
      "rewards/IngredientQuantityMatchReward/std": 0.4316678524017334,
      "rewards/TotalKcalExactMatchReward/mean": 0.8171875,
      "rewards/TotalKcalExactMatchReward/std": 0.37480711936950684,
      "step": 2910
    },
    {
      "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": 0.0109375,
      "completions/max_length": 503.8,
      "completions/mean_length": 389.6546875,
      "completions/min_length": 239.8,
      "epoch": 0.8108484005563282,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.612226128578186,
      "kl": 0.0521865741815418,
      "learning_rate": 9.46819542139023e-08,
      "loss": 0.0020877527073025703,
      "reward": 3.0481145858764647,
      "reward_std": 0.3714823067188263,
      "rewards/IngredientFormatReward/mean": 0.9858705282211304,
      "rewards/IngredientFormatReward/std": 0.08456218093633652,
      "rewards/IngredientMatchReward/mean": 0.6272296667098999,
      "rewards/IngredientMatchReward/std": 0.2885073244571686,
      "rewards/IngredientQuantityMatchReward/mean": 0.6459518194198608,
      "rewards/IngredientQuantityMatchReward/std": 0.42370365262031556,
      "rewards/TotalKcalExactMatchReward/mean": 0.7890625,
      "rewards/TotalKcalExactMatchReward/std": 0.4039949357509613,
      "step": 2915
    },
    {
      "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": 0.003125,
      "completions/max_length": 491.8,
      "completions/mean_length": 386.0296875,
      "completions/min_length": 248.4,
      "epoch": 0.8122392211404729,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6456689834594727,
      "kl": 0.048775306739844385,
      "learning_rate": 9.333956989775149e-08,
      "loss": 0.0019510995596647262,
      "reward": 3.202948474884033,
      "reward_std": 0.3436084806919098,
      "rewards/IngredientFormatReward/mean": 0.9966145753860474,
      "rewards/IngredientFormatReward/std": 0.0278476582840085,
      "rewards/IngredientMatchReward/mean": 0.6758649468421936,
      "rewards/IngredientMatchReward/std": 0.295900821685791,
      "rewards/IngredientQuantityMatchReward/mean": 0.7242189645767212,
      "rewards/IngredientQuantityMatchReward/std": 0.4036065459251404,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.37954323291778563,
      "step": 2920
    },
    {
      "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": 0.0109375,
      "completions/max_length": 513.0,
      "completions/mean_length": 389.5921875,
      "completions/min_length": 256.2,
      "epoch": 0.8136300417246175,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7192108035087585,
      "kl": 0.051587856956757605,
      "learning_rate": 9.200578936538628e-08,
      "loss": 0.002063538506627083,
      "reward": 3.1155295848846434,
      "reward_std": 0.39563546180725095,
      "rewards/IngredientFormatReward/mean": 0.9871875047683716,
      "rewards/IngredientFormatReward/std": 0.1086337298154831,
      "rewards/IngredientMatchReward/mean": 0.6720517158508301,
      "rewards/IngredientMatchReward/std": 0.304996919631958,
      "rewards/IngredientQuantityMatchReward/mean": 0.6562903761863709,
      "rewards/IngredientQuantityMatchReward/std": 0.40350993871688845,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3642785012722015,
      "step": 2925
    },
    {
      "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": 0.009375,
      "completions/max_length": 505.6,
      "completions/mean_length": 392.51875,
      "completions/min_length": 263.0,
      "epoch": 0.8150208623087621,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6309117674827576,
      "kl": 0.04693901997525245,
      "learning_rate": 9.068064083582789e-08,
      "loss": 0.0018776778131723403,
      "reward": 3.034470796585083,
      "reward_std": 0.4299850046634674,
      "rewards/IngredientFormatReward/mean": 0.9903125047683716,
      "rewards/IngredientFormatReward/std": 0.07649115696549416,
      "rewards/IngredientMatchReward/mean": 0.6331597208976746,
      "rewards/IngredientMatchReward/std": 0.28655193746089935,
      "rewards/IngredientQuantityMatchReward/mean": 0.63599853515625,
      "rewards/IngredientQuantityMatchReward/std": 0.4148427963256836,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.41708130240440366,
      "step": 2930
    },
    {
      "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": 0.0125,
      "completions/max_length": 512.6,
      "completions/mean_length": 383.5734375,
      "completions/min_length": 258.6,
      "epoch": 0.8164116828929068,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6402202844619751,
      "kl": 0.04907006416469813,
      "learning_rate": 8.93641523454688e-08,
      "loss": 0.001963062025606632,
      "reward": 3.087191581726074,
      "reward_std": 0.389241361618042,
      "rewards/IngredientFormatReward/mean": 0.9887500047683716,
      "rewards/IngredientFormatReward/std": 0.07457910180091858,
      "rewards/IngredientMatchReward/mean": 0.6492887973785401,
      "rewards/IngredientMatchReward/std": 0.30371943712234495,
      "rewards/IngredientQuantityMatchReward/mean": 0.6538402795791626,
      "rewards/IngredientQuantityMatchReward/std": 0.39292051196098327,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.38929224014282227,
      "step": 2935
    },
    {
      "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": 0.003125,
      "completions/max_length": 497.6,
      "completions/mean_length": 393.7984375,
      "completions/min_length": 254.8,
      "epoch": 0.8178025034770514,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6798999309539795,
      "kl": 0.04532430283725262,
      "learning_rate": 8.805635174747961e-08,
      "loss": 0.0018129955977201461,
      "reward": 3.1188385486602783,
      "reward_std": 0.40057496428489686,
      "rewards/IngredientFormatReward/mean": 0.995312488079071,
      "rewards/IngredientFormatReward/std": 0.03757027108222246,
      "rewards/IngredientMatchReward/mean": 0.651212191581726,
      "rewards/IngredientMatchReward/std": 0.27968040108680725,
      "rewards/IngredientQuantityMatchReward/mean": 0.6723138213157653,
      "rewards/IngredientQuantityMatchReward/std": 0.4165939033031464,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3970761299133301,
      "step": 2940
    },
    {
      "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": 0.0125,
      "completions/max_length": 495.8,
      "completions/mean_length": 390.9703125,
      "completions/min_length": 279.4,
      "epoch": 0.8191933240611962,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7016852498054504,
      "kl": 0.04716307336930185,
      "learning_rate": 8.675726671121968e-08,
      "loss": 0.0018864812329411507,
      "reward": 3.167627954483032,
      "reward_std": 0.39934841394424436,
      "rewards/IngredientFormatReward/mean": 0.9858463525772094,
      "rewards/IngredientFormatReward/std": 0.06388909518718719,
      "rewards/IngredientMatchReward/mean": 0.6783639073371888,
      "rewards/IngredientMatchReward/std": 0.2869694232940674,
      "rewards/IngredientQuantityMatchReward/mean": 0.6706052541732788,
      "rewards/IngredientQuantityMatchReward/std": 0.4015600621700287,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.3679316818714142,
      "step": 2945
    },
    {
      "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": 0.0046875,
      "completions/max_length": 503.2,
      "completions/mean_length": 387.15625,
      "completions/min_length": 278.6,
      "epoch": 0.8205841446453408,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6271607875823975,
      "kl": 0.04509587762877345,
      "learning_rate": 8.546692472165195e-08,
      "loss": 0.0018041588366031647,
      "reward": 3.275380325317383,
      "reward_std": 0.3708919048309326,
      "rewards/IngredientFormatReward/mean": 0.9959895730018615,
      "rewards/IngredientFormatReward/std": 0.04264734834432602,
      "rewards/IngredientMatchReward/mean": 0.709428322315216,
      "rewards/IngredientMatchReward/std": 0.2649887174367905,
      "rewards/IngredientQuantityMatchReward/mean": 0.7152748942375183,
      "rewards/IngredientQuantityMatchReward/std": 0.4001452922821045,
      "rewards/TotalKcalExactMatchReward/mean": 0.8546875,
      "rewards/TotalKcalExactMatchReward/std": 0.35101463794708254,
      "step": 2950
    },
    {
      "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": 0.0140625,
      "completions/max_length": 506.4,
      "completions/mean_length": 393.584375,
      "completions/min_length": 252.8,
      "epoch": 0.8219749652294854,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6617047190666199,
      "kl": 0.05037928493693471,
      "learning_rate": 8.418535307876057e-08,
      "loss": 0.0020150141790509224,
      "reward": 3.1745755672454834,
      "reward_std": 0.39280285835266116,
      "rewards/IngredientFormatReward/mean": 0.9833072900772095,
      "rewards/IngredientFormatReward/std": 0.10187680423259735,
      "rewards/IngredientMatchReward/mean": 0.6803211688995361,
      "rewards/IngredientMatchReward/std": 0.29972589910030367,
      "rewards/IngredientQuantityMatchReward/mean": 0.704697048664093,
      "rewards/IngredientQuantityMatchReward/std": 0.39612128734588625,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.39199302196502683,
      "step": 2955
    },
    {
      "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": 0.0046875,
      "completions/max_length": 503.0,
      "completions/mean_length": 389.19375,
      "completions/min_length": 266.0,
      "epoch": 0.8233657858136301,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6642850637435913,
      "kl": 0.04720083279535174,
      "learning_rate": 8.291257889697484e-08,
      "loss": 0.001888134703040123,
      "reward": 3.070013332366943,
      "reward_std": 0.41594597697257996,
      "rewards/IngredientFormatReward/mean": 0.9950000047683716,
      "rewards/IngredientFormatReward/std": 0.04611458256840706,
      "rewards/IngredientMatchReward/mean": 0.6542714476585388,
      "rewards/IngredientMatchReward/std": 0.25903120040893557,
      "rewards/IngredientQuantityMatchReward/mean": 0.6598044037818909,
      "rewards/IngredientQuantityMatchReward/std": 0.42194475531578063,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.4252581536769867,
      "step": 2960
    },
    {
      "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": 0.003125,
      "completions/max_length": 498.4,
      "completions/mean_length": 389.225,
      "completions/min_length": 275.6,
      "epoch": 0.8247566063977747,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6487119197845459,
      "kl": 0.045345666841603814,
      "learning_rate": 8.164862910459419e-08,
      "loss": 0.0018140103667974472,
      "reward": 3.168592071533203,
      "reward_std": 0.3895823538303375,
      "rewards/IngredientFormatReward/mean": 0.9959895730018615,
      "rewards/IngredientFormatReward/std": 0.04264734834432602,
      "rewards/IngredientMatchReward/mean": 0.6649200201034546,
      "rewards/IngredientMatchReward/std": 0.29957814812660216,
      "rewards/IngredientQuantityMatchReward/mean": 0.6608074545860291,
      "rewards/IngredientQuantityMatchReward/std": 0.42049002051353457,
      "rewards/TotalKcalExactMatchReward/mean": 0.846875,
      "rewards/TotalKcalExactMatchReward/std": 0.35720263719558715,
      "step": 2965
    },
    {
      "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": 0.0046875,
      "completions/max_length": 495.2,
      "completions/mean_length": 385.578125,
      "completions/min_length": 256.4,
      "epoch": 0.8261474269819193,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.653709888458252,
      "kl": 0.05087601642590016,
      "learning_rate": 8.039353044321918e-08,
      "loss": 0.002034936100244522,
      "reward": 3.047412872314453,
      "reward_std": 0.4022683262825012,
      "rewards/IngredientFormatReward/mean": 0.9851041674613953,
      "rewards/IngredientFormatReward/std": 0.08566536605358124,
      "rewards/IngredientMatchReward/mean": 0.6148604869842529,
      "rewards/IngredientMatchReward/std": 0.28956696689128875,
      "rewards/IngredientQuantityMatchReward/mean": 0.6208856463432312,
      "rewards/IngredientQuantityMatchReward/std": 0.42820629477500916,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.36872145533561707,
      "step": 2970
    },
    {
      "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": 0.009375,
      "completions/max_length": 508.4,
      "completions/mean_length": 391.1875,
      "completions/min_length": 247.8,
      "epoch": 0.827538247566064,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6367550492286682,
      "kl": 0.04671785989776254,
      "learning_rate": 7.914730946718507e-08,
      "loss": 0.0018687203526496887,
      "reward": 3.100764036178589,
      "reward_std": 0.4259964883327484,
      "rewards/IngredientFormatReward/mean": 0.9904017925262452,
      "rewards/IngredientFormatReward/std": 0.0754810044541955,
      "rewards/IngredientMatchReward/mean": 0.6349913835525512,
      "rewards/IngredientMatchReward/std": 0.2922417223453522,
      "rewards/IngredientQuantityMatchReward/mean": 0.6753708362579346,
      "rewards/IngredientQuantityMatchReward/std": 0.4144640028476715,
      "rewards/TotalKcalExactMatchReward/mean": 0.8,
      "rewards/TotalKcalExactMatchReward/std": 0.3949720025062561,
      "step": 2975
    },
    {
      "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": 0.009375,
      "completions/max_length": 507.6,
      "completions/mean_length": 391.1828125,
      "completions/min_length": 265.6,
      "epoch": 0.8289290681502086,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.5961768627166748,
      "kl": 0.05127133592031896,
      "learning_rate": 7.790999254300079e-08,
      "loss": 0.0020504213869571688,
      "reward": 3.1191386222839355,
      "reward_std": 0.3484513610601425,
      "rewards/IngredientFormatReward/mean": 0.9890625,
      "rewards/IngredientFormatReward/std": 0.08017933368682861,
      "rewards/IngredientMatchReward/mean": 0.6748840570449829,
      "rewards/IngredientMatchReward/std": 0.2862056910991669,
      "rewards/IngredientQuantityMatchReward/mean": 0.5989420533180236,
      "rewards/IngredientQuantityMatchReward/std": 0.42672608494758607,
      "rewards/TotalKcalExactMatchReward/mean": 0.85625,
      "rewards/TotalKcalExactMatchReward/std": 0.3235033631324768,
      "step": 2980
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.6,
      "completions/mean_length": 386.3671875,
      "completions/min_length": 248.2,
      "epoch": 0.8303198887343533,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.655218243598938,
      "kl": 0.048838711553253235,
      "learning_rate": 7.668160584879068e-08,
      "loss": 0.001953764446079731,
      "reward": 3.109603023529053,
      "reward_std": 0.3838211238384247,
      "rewards/IngredientFormatReward/mean": 0.9916865229606628,
      "rewards/IngredientFormatReward/std": 0.0682242352515459,
      "rewards/IngredientMatchReward/mean": 0.6327531456947326,
      "rewards/IngredientMatchReward/std": 0.2947903752326965,
      "rewards/IngredientQuantityMatchReward/mean": 0.6726634740829468,
      "rewards/IngredientQuantityMatchReward/std": 0.4099680960178375,
      "rewards/TotalKcalExactMatchReward/mean": 0.8125,
      "rewards/TotalKcalExactMatchReward/std": 0.38938270807266234,
      "step": 2985
    },
    {
      "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": 0.0046875,
      "completions/max_length": 507.2,
      "completions/mean_length": 391.321875,
      "completions/min_length": 272.0,
      "epoch": 0.8317107093184979,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.5529534816741943,
      "kl": 0.042588020046241584,
      "learning_rate": 7.546217537374072e-08,
      "loss": 0.0017037199810147285,
      "reward": 3.1710736751556396,
      "reward_std": 0.33150983452796934,
      "rewards/IngredientFormatReward/mean": 0.9946056485176087,
      "rewards/IngredientFormatReward/std": 0.03346139807254076,
      "rewards/IngredientMatchReward/mean": 0.6643967270851135,
      "rewards/IngredientMatchReward/std": 0.29930198192596436,
      "rewards/IngredientQuantityMatchReward/mean": 0.6495713591575623,
      "rewards/IngredientQuantityMatchReward/std": 0.4249771058559418,
      "rewards/TotalKcalExactMatchReward/mean": 0.8625,
      "rewards/TotalKcalExactMatchReward/std": 0.33672092854976654,
      "step": 2990
    },
    {
      "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": 0.0046875,
      "completions/max_length": 504.6,
      "completions/mean_length": 389.7640625,
      "completions/min_length": 259.6,
      "epoch": 0.8331015299026425,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6654627919197083,
      "kl": 0.0474369776668027,
      "learning_rate": 7.42517269175485e-08,
      "loss": 0.001897449791431427,
      "reward": 3.0309982299804688,
      "reward_std": 0.39750537276268005,
      "rewards/IngredientFormatReward/mean": 0.9892187476158142,
      "rewards/IngredientFormatReward/std": 0.08536323457956314,
      "rewards/IngredientMatchReward/mean": 0.6950334906578064,
      "rewards/IngredientMatchReward/std": 0.29454014301300047,
      "rewards/IngredientQuantityMatchReward/mean": 0.5983084917068482,
      "rewards/IngredientQuantityMatchReward/std": 0.42347745299339296,
      "rewards/TotalKcalExactMatchReward/mean": 0.7484375,
      "rewards/TotalKcalExactMatchReward/std": 0.418589586019516,
      "step": 2995
    },
    {
      "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": 0.0078125,
      "completions/max_length": 502.4,
      "completions/mean_length": 382.84375,
      "completions/min_length": 256.4,
      "epoch": 0.8344923504867872,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 7.202028751373291,
      "kl": 0.06541607140097767,
      "learning_rate": 7.305028608987762e-08,
      "loss": 0.0026157567277550696,
      "reward": 3.107814598083496,
      "reward_std": 0.3948281526565552,
      "rewards/IngredientFormatReward/mean": 0.9901041746139526,
      "rewards/IngredientFormatReward/std": 0.0734422504901886,
      "rewards/IngredientMatchReward/mean": 0.6643247961997986,
      "rewards/IngredientMatchReward/std": 0.27318963408470154,
      "rewards/IngredientQuantityMatchReward/mean": 0.6783856272697448,
      "rewards/IngredientQuantityMatchReward/std": 0.4000774621963501,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.40691637992858887,
      "step": 3000
    },
    {
      "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": 0.0015625,
      "completions/max_length": 495.2,
      "completions/mean_length": 390.521875,
      "completions/min_length": 248.8,
      "epoch": 0.8358831710709318,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6431273818016052,
      "kl": 0.04963696929626167,
      "learning_rate": 7.185787830981571e-08,
      "loss": 0.001985309273004532,
      "reward": 3.116542387008667,
      "reward_std": 0.38055387139320374,
      "rewards/IngredientFormatReward/mean": 0.9975353479385376,
      "rewards/IngredientFormatReward/std": 0.024885546416044235,
      "rewards/IngredientMatchReward/mean": 0.6366834044456482,
      "rewards/IngredientMatchReward/std": 0.2909588277339935,
      "rewards/IngredientQuantityMatchReward/mean": 0.6620110869407654,
      "rewards/IngredientQuantityMatchReward/std": 0.3996714651584625,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.368584543466568,
      "step": 3005
    },
    {
      "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": 0.009375,
      "completions/max_length": 508.2,
      "completions/mean_length": 392.6125,
      "completions/min_length": 277.6,
      "epoch": 0.8372739916550765,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6810948252677917,
      "kl": 0.0459325288888067,
      "learning_rate": 7.067452880533664e-08,
      "loss": 0.0018374029546976089,
      "reward": 3.079831838607788,
      "reward_std": 0.40001912117004396,
      "rewards/IngredientFormatReward/mean": 0.9881584882736206,
      "rewards/IngredientFormatReward/std": 0.08064654730260372,
      "rewards/IngredientMatchReward/mean": 0.6321608543395996,
      "rewards/IngredientMatchReward/std": 0.30199761390686036,
      "rewards/IngredientQuantityMatchReward/mean": 0.6610750675201416,
      "rewards/IngredientQuantityMatchReward/std": 0.4067903161048889,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.38732614517211916,
      "step": 3010
    },
    {
      "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": 0.0015625,
      "completions/max_length": 502.4,
      "completions/mean_length": 389.1453125,
      "completions/min_length": 269.0,
      "epoch": 0.8386648122392212,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6376404762268066,
      "kl": 0.05073229791596532,
      "learning_rate": 6.950026261276698e-08,
      "loss": 0.002029072307050228,
      "reward": 3.1437541007995606,
      "reward_std": 0.35086989402770996,
      "rewards/IngredientFormatReward/mean": 0.99469496011734,
      "rewards/IngredientFormatReward/std": 0.04523397646844387,
      "rewards/IngredientMatchReward/mean": 0.6596317172050477,
      "rewards/IngredientMatchReward/std": 0.28066009283065796,
      "rewards/IngredientQuantityMatchReward/mean": 0.6753650188446045,
      "rewards/IngredientQuantityMatchReward/std": 0.40645899176597594,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3786509335041046,
      "step": 3015
    },
    {
      "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": 0.0078125,
      "completions/max_length": 505.4,
      "completions/mean_length": 397.4359375,
      "completions/min_length": 298.6,
      "epoch": 0.8400556328233658,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6552076935768127,
      "kl": 0.04711475428193808,
      "learning_rate": 6.833510457625585e-08,
      "loss": 0.0018848728388547898,
      "reward": 3.2228328227996825,
      "reward_std": 0.37228912115097046,
      "rewards/IngredientFormatReward/mean": 0.9916666626930237,
      "rewards/IngredientFormatReward/std": 0.06616733074188233,
      "rewards/IngredientMatchReward/mean": 0.6734991908073426,
      "rewards/IngredientMatchReward/std": 0.29104630947113036,
      "rewards/IngredientQuantityMatchReward/mean": 0.7248544454574585,
      "rewards/IngredientQuantityMatchReward/std": 0.3884549975395203,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.35898853838443756,
      "step": 3020
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.2,
      "completions/mean_length": 391.75625,
      "completions/min_length": 274.8,
      "epoch": 0.8414464534075105,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6299717426300049,
      "kl": 0.04465638347901404,
      "learning_rate": 6.717907934724982e-08,
      "loss": 0.0017864655703306199,
      "reward": 3.200212860107422,
      "reward_std": 0.3218538582324982,
      "rewards/IngredientFormatReward/mean": 0.9934895753860473,
      "rewards/IngredientFormatReward/std": 0.052749037928879264,
      "rewards/IngredientMatchReward/mean": 0.675394332408905,
      "rewards/IngredientMatchReward/std": 0.2600559026002884,
      "rewards/IngredientQuantityMatchReward/mean": 0.7000789642333984,
      "rewards/IngredientQuantityMatchReward/std": 0.3908191204071045,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.37399315237998965,
      "step": 3025
    },
    {
      "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": 0.0078125,
      "completions/max_length": 508.0,
      "completions/mean_length": 395.0703125,
      "completions/min_length": 292.4,
      "epoch": 0.8428372739916551,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6462646126747131,
      "kl": 0.04630208441521973,
      "learning_rate": 6.603221138397102e-08,
      "loss": 0.0018521670252084732,
      "reward": 3.1141008377075194,
      "reward_std": 0.43531657457351686,
      "rewards/IngredientFormatReward/mean": 0.9883407711982727,
      "rewards/IngredientFormatReward/std": 0.08107312843203544,
      "rewards/IngredientMatchReward/mean": 0.6730586767196656,
      "rewards/IngredientMatchReward/std": 0.2721665918827057,
      "rewards/IngredientQuantityMatchReward/mean": 0.6464512586593628,
      "rewards/IngredientQuantityMatchReward/std": 0.40384441614151,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3568021148443222,
      "step": 3030
    },
    {
      "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": 0.00625,
      "completions/max_length": 499.6,
      "completions/mean_length": 393.7890625,
      "completions/min_length": 278.2,
      "epoch": 0.8442280945757997,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6232410073280334,
      "kl": 0.046137299528345464,
      "learning_rate": 6.489452495089959e-08,
      "loss": 0.0018458150327205658,
      "reward": 3.069223690032959,
      "reward_std": 0.38708738684654237,
      "rewards/IngredientFormatReward/mean": 0.9917299151420593,
      "rewards/IngredientFormatReward/std": 0.048492889106273654,
      "rewards/IngredientMatchReward/mean": 0.6515817403793335,
      "rewards/IngredientMatchReward/std": 0.2845022022724152,
      "rewards/IngredientQuantityMatchReward/mean": 0.6180995345115662,
      "rewards/IngredientQuantityMatchReward/std": 0.41500294804573057,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.38249378800392153,
      "step": 3035
    },
    {
      "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": 0.01875,
      "completions/max_length": 509.8,
      "completions/mean_length": 392.3171875,
      "completions/min_length": 252.6,
      "epoch": 0.8456189151599444,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 8.307910919189453,
      "kl": 0.07405607257969678,
      "learning_rate": 6.376604411826069e-08,
      "loss": 0.002963574230670929,
      "reward": 3.0168015003204345,
      "reward_std": 0.45492454767227175,
      "rewards/IngredientFormatReward/mean": 0.9760770082473755,
      "rewards/IngredientFormatReward/std": 0.13225289061665535,
      "rewards/IngredientMatchReward/mean": 0.6269741415977478,
      "rewards/IngredientMatchReward/std": 0.3054593026638031,
      "rewards/IngredientQuantityMatchReward/mean": 0.6465628862380981,
      "rewards/IngredientQuantityMatchReward/std": 0.43766944408416747,
      "rewards/TotalKcalExactMatchReward/mean": 0.7671875,
      "rewards/TotalKcalExactMatchReward/std": 0.4169674217700958,
      "step": 3040
    },
    {
      "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": 0.0125,
      "completions/max_length": 507.4,
      "completions/mean_length": 396.1921875,
      "completions/min_length": 259.2,
      "epoch": 0.847009735744089,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.7011051774024963,
      "kl": 0.05020157103426755,
      "learning_rate": 6.264679276151485e-08,
      "loss": 0.0020083660259842873,
      "reward": 3.1013489723205567,
      "reward_std": 0.36317563652992246,
      "rewards/IngredientFormatReward/mean": 0.9876562356948853,
      "rewards/IngredientFormatReward/std": 0.08381983656436205,
      "rewards/IngredientMatchReward/mean": 0.6747749328613282,
      "rewards/IngredientMatchReward/std": 0.2843883991241455,
      "rewards/IngredientQuantityMatchReward/mean": 0.6686052441596985,
      "rewards/IngredientQuantityMatchReward/std": 0.38197819590568544,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.39580188393592836,
      "step": 3045
    },
    {
      "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": 0.003125,
      "completions/max_length": 508.2,
      "completions/mean_length": 393.009375,
      "completions/min_length": 265.0,
      "epoch": 0.8484005563282336,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6340299844741821,
      "kl": 0.04501628072466701,
      "learning_rate": 6.153679456085343e-08,
      "loss": 0.0018011130392551421,
      "reward": 3.1332841396331785,
      "reward_std": 0.33361607491970063,
      "rewards/IngredientFormatReward/mean": 0.9887760400772094,
      "rewards/IngredientFormatReward/std": 0.06544025391340255,
      "rewards/IngredientMatchReward/mean": 0.6738690495491028,
      "rewards/IngredientMatchReward/std": 0.2672120571136475,
      "rewards/IngredientQuantityMatchReward/mean": 0.6628265261650086,
      "rewards/IngredientQuantityMatchReward/std": 0.3794059336185455,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.3943259060382843,
      "step": 3050
    },
    {
      "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": 0.0046875,
      "completions/max_length": 501.2,
      "completions/mean_length": 394.6828125,
      "completions/min_length": 273.0,
      "epoch": 0.8497913769123783,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6601257920265198,
      "kl": 0.04646438907366246,
      "learning_rate": 6.043607300069653e-08,
      "loss": 0.0018583284690976143,
      "reward": 3.096318817138672,
      "reward_std": 0.3890433728694916,
      "rewards/IngredientFormatReward/mean": 0.9942968726158142,
      "rewards/IngredientFormatReward/std": 0.051978741958737376,
      "rewards/IngredientMatchReward/mean": 0.6110546827316284,
      "rewards/IngredientMatchReward/std": 0.31328256130218507,
      "rewards/IngredientQuantityMatchReward/mean": 0.6878422737121582,
      "rewards/IngredientQuantityMatchReward/std": 0.4182025969028473,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.3834996998310089,
      "step": 3055
    },
    {
      "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": 490.2,
      "completions/mean_length": 382.29375,
      "completions/min_length": 262.2,
      "epoch": 0.8511821974965229,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6450671553611755,
      "kl": 0.047940224315971135,
      "learning_rate": 5.934465136919736e-08,
      "loss": 0.001917634904384613,
      "reward": 3.150379753112793,
      "reward_std": 0.35938249826431273,
      "rewards/IngredientFormatReward/mean": 0.9995814800262451,
      "rewards/IngredientFormatReward/std": 0.0033425264060497283,
      "rewards/IngredientMatchReward/mean": 0.6239750862121582,
      "rewards/IngredientMatchReward/std": 0.2981367170810699,
      "rewards/IngredientQuantityMatchReward/mean": 0.6846357345581054,
      "rewards/IngredientQuantityMatchReward/std": 0.4151392221450806,
      "rewards/TotalKcalExactMatchReward/mean": 0.8421875,
      "rewards/TotalKcalExactMatchReward/std": 0.35220016837120055,
      "step": 3060
    },
    {
      "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": 0.021875,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.1671875,
      "completions/min_length": 276.2,
      "epoch": 0.8525730180806675,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6400882005691528,
      "kl": 0.046734277112409475,
      "learning_rate": 5.826255275774861e-08,
      "loss": 0.0018693899735808372,
      "reward": 3.0946020126342773,
      "reward_std": 0.41802999973297117,
      "rewards/IngredientFormatReward/mean": 0.9781845331192016,
      "rewards/IngredientFormatReward/std": 0.12996118217706681,
      "rewards/IngredientMatchReward/mean": 0.6358180522918702,
      "rewards/IngredientMatchReward/std": 0.3087162494659424,
      "rewards/IngredientQuantityMatchReward/mean": 0.6821619391441345,
      "rewards/IngredientQuantityMatchReward/std": 0.41959701776504515,
      "rewards/TotalKcalExactMatchReward/mean": 0.7984375,
      "rewards/TotalKcalExactMatchReward/std": 0.39511425495147706,
      "step": 3065
    },
    {
      "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": 0.0078125,
      "completions/max_length": 513.0,
      "completions/mean_length": 390.409375,
      "completions/min_length": 266.6,
      "epoch": 0.8539638386648123,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6598068475723267,
      "kl": 0.05595761397853494,
      "learning_rate": 5.718980006049445e-08,
      "loss": 0.0022384027019143105,
      "reward": 3.1849377155303955,
      "reward_std": 0.4088379919528961,
      "rewards/IngredientFormatReward/mean": 0.991391372680664,
      "rewards/IngredientFormatReward/std": 0.08908663541078568,
      "rewards/IngredientMatchReward/mean": 0.6616598486900329,
      "rewards/IngredientMatchReward/std": 0.2799966514110565,
      "rewards/IngredientQuantityMatchReward/mean": 0.6928239345550538,
      "rewards/IngredientQuantityMatchReward/std": 0.4054431617259979,
      "rewards/TotalKcalExactMatchReward/mean": 0.8390625,
      "rewards/TotalKcalExactMatchReward/std": 0.3501300573348999,
      "step": 3070
    },
    {
      "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": 0.01875,
      "completions/max_length": 508.4,
      "completions/mean_length": 397.04375,
      "completions/min_length": 263.8,
      "epoch": 0.8553546592489569,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6476394534111023,
      "kl": 0.04536970742046833,
      "learning_rate": 5.6126415973845875e-08,
      "loss": 0.00181522686034441,
      "reward": 3.127425193786621,
      "reward_std": 0.4319356918334961,
      "rewards/IngredientFormatReward/mean": 0.9797916650772095,
      "rewards/IngredientFormatReward/std": 0.11502126976847649,
      "rewards/IngredientMatchReward/mean": 0.6737530946731567,
      "rewards/IngredientMatchReward/std": 0.2894026696681976,
      "rewards/IngredientQuantityMatchReward/mean": 0.6629429399967194,
      "rewards/IngredientQuantityMatchReward/std": 0.403452605009079,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.3918249785900116,
      "step": 3075
    },
    {
      "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": 0.0046875,
      "completions/max_length": 506.4,
      "completions/mean_length": 394.4671875,
      "completions/min_length": 275.0,
      "epoch": 0.8567454798331016,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6560621857643127,
      "kl": 0.050239683431573215,
      "learning_rate": 5.5072422996000626e-08,
      "loss": 0.002009827084839344,
      "reward": 3.2230833530426026,
      "reward_std": 0.3705843985080719,
      "rewards/IngredientFormatReward/mean": 0.9926041603088379,
      "rewards/IngredientFormatReward/std": 0.05480610430240631,
      "rewards/IngredientMatchReward/mean": 0.6874944806098938,
      "rewards/IngredientMatchReward/std": 0.28081194758415223,
      "rewards/IngredientQuantityMatchReward/mean": 0.6945472121238708,
      "rewards/IngredientQuantityMatchReward/std": 0.39099807739257814,
      "rewards/TotalKcalExactMatchReward/mean": 0.8484375,
      "rewards/TotalKcalExactMatchReward/std": 0.35526342391967775,
      "step": 3080
    },
    {
      "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": 0.003125,
      "completions/max_length": 509.4,
      "completions/mean_length": 391.5140625,
      "completions/min_length": 271.8,
      "epoch": 0.8581363004172462,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 1.5736770629882812,
      "kl": 0.04943891023285687,
      "learning_rate": 5.40278434264671e-08,
      "loss": 0.001977892965078354,
      "reward": 3.0344571590423586,
      "reward_std": 0.34014609456062317,
      "rewards/IngredientFormatReward/mean": 0.9945535778999328,
      "rewards/IngredientFormatReward/std": 0.041323404759168625,
      "rewards/IngredientMatchReward/mean": 0.615483021736145,
      "rewards/IngredientMatchReward/std": 0.27883716523647306,
      "rewards/IngredientQuantityMatchReward/mean": 0.6556705832481384,
      "rewards/IngredientQuantityMatchReward/std": 0.41032910346984863,
      "rewards/TotalKcalExactMatchReward/mean": 0.76875,
      "rewards/TotalKcalExactMatchReward/std": 0.42164164781570435,
      "step": 3085
    },
    {
      "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": 0.003125,
      "completions/max_length": 495.2,
      "completions/mean_length": 386.803125,
      "completions/min_length": 246.8,
      "epoch": 0.8595271210013908,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.8206337690353394,
      "kl": 0.051253604143857955,
      "learning_rate": 5.299269936559275e-08,
      "loss": 0.0020507143810391424,
      "reward": 3.12697377204895,
      "reward_std": 0.40267518162727356,
      "rewards/IngredientFormatReward/mean": 0.9957291603088378,
      "rewards/IngredientFormatReward/std": 0.03133343979716301,
      "rewards/IngredientMatchReward/mean": 0.6811867475509643,
      "rewards/IngredientMatchReward/std": 0.2661647886037827,
      "rewards/IngredientQuantityMatchReward/mean": 0.6641203999519348,
      "rewards/IngredientQuantityMatchReward/std": 0.40845090746879575,
      "rewards/TotalKcalExactMatchReward/mean": 0.7859375,
      "rewards/TotalKcalExactMatchReward/std": 0.40290495157241824,
      "step": 3090
    },
    {
      "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": 0.015625,
      "completions/max_length": 506.4,
      "completions/mean_length": 394.28125,
      "completions/min_length": 265.0,
      "epoch": 0.8609179415855355,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7553901672363281,
      "kl": 0.06073210455942899,
      "learning_rate": 5.196701271409615e-08,
      "loss": 0.0024279464036226273,
      "reward": 3.1119028091430665,
      "reward_std": 0.47098749279975893,
      "rewards/IngredientFormatReward/mean": 0.9827256798744202,
      "rewards/IngredientFormatReward/std": 0.09406303185969592,
      "rewards/IngredientMatchReward/mean": 0.6565879464149476,
      "rewards/IngredientMatchReward/std": 0.28705950975418093,
      "rewards/IngredientQuantityMatchReward/mean": 0.6757142424583436,
      "rewards/IngredientQuantityMatchReward/std": 0.4014221131801605,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.39542931914329527,
      "step": 3095
    },
    {
      "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": 0.0125,
      "completions/max_length": 511.6,
      "completions/mean_length": 398.4203125,
      "completions/min_length": 274.6,
      "epoch": 0.8623087621696801,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7021408081054688,
      "kl": 0.04793826362583786,
      "learning_rate": 5.095080517260397e-08,
      "loss": 0.0019172437489032746,
      "reward": 3.079103422164917,
      "reward_std": 0.41487077474594114,
      "rewards/IngredientFormatReward/mean": 0.9877511143684388,
      "rewards/IngredientFormatReward/std": 0.0843487460166216,
      "rewards/IngredientMatchReward/mean": 0.6174058675765991,
      "rewards/IngredientMatchReward/std": 0.2904114544391632,
      "rewards/IngredientQuantityMatchReward/mean": 0.6380090236663818,
      "rewards/IngredientQuantityMatchReward/std": 0.4309688091278076,
      "rewards/TotalKcalExactMatchReward/mean": 0.8359375,
      "rewards/TotalKcalExactMatchReward/std": 0.3543159782886505,
      "step": 3100
    },
    {
      "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": 0.009375,
      "completions/max_length": 513.0,
      "completions/mean_length": 390.659375,
      "completions/min_length": 273.8,
      "epoch": 0.8636995827538247,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6427891254425049,
      "kl": 0.04723424823023379,
      "learning_rate": 4.994409824119189e-08,
      "loss": 0.001889539510011673,
      "reward": 3.184514856338501,
      "reward_std": 0.43525434732437135,
      "rewards/IngredientFormatReward/mean": 0.991796875,
      "rewards/IngredientFormatReward/std": 0.08889861106872558,
      "rewards/IngredientMatchReward/mean": 0.6531863808631897,
      "rewards/IngredientMatchReward/std": 0.28255691528320315,
      "rewards/IngredientQuantityMatchReward/mean": 0.7442190766334533,
      "rewards/IngredientQuantityMatchReward/std": 0.37337024211883546,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.39657433032989503,
      "step": 3105
    },
    {
      "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": 0.003125,
      "completions/max_length": 505.0,
      "completions/mean_length": 389.3578125,
      "completions/min_length": 267.2,
      "epoch": 0.8650904033379694,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6806484460830688,
      "kl": 0.23146929705981165,
      "learning_rate": 4.8946913218929466e-08,
      "loss": 0.009275725483894348,
      "reward": 3.085928964614868,
      "reward_std": 0.37679257392883303,
      "rewards/IngredientFormatReward/mean": 0.9950744032859802,
      "rewards/IngredientFormatReward/std": 0.04088784530758858,
      "rewards/IngredientMatchReward/mean": 0.6328473806381225,
      "rewards/IngredientMatchReward/std": 0.3016449213027954,
      "rewards/IngredientQuantityMatchReward/mean": 0.6376946866512299,
      "rewards/IngredientQuantityMatchReward/std": 0.42166988253593446,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.3804445922374725,
      "step": 3110
    },
    {
      "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": 0.01875,
      "completions/max_length": 506.6,
      "completions/mean_length": 387.0,
      "completions/min_length": 239.4,
      "epoch": 0.866481223922114,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6415086984634399,
      "kl": 0.07196273931767791,
      "learning_rate": 4.7959271203429405e-08,
      "loss": 0.002879709005355835,
      "reward": 3.1648409843444822,
      "reward_std": 0.3970713198184967,
      "rewards/IngredientFormatReward/mean": 0.979895830154419,
      "rewards/IngredientFormatReward/std": 0.10437989309430122,
      "rewards/IngredientMatchReward/mean": 0.6672836184501648,
      "rewards/IngredientMatchReward/std": 0.2711143881082535,
      "rewards/IngredientQuantityMatchReward/mean": 0.6801616072654724,
      "rewards/IngredientQuantityMatchReward/std": 0.40835112929344175,
      "rewards/TotalKcalExactMatchReward/mean": 0.8375,
      "rewards/TotalKcalExactMatchReward/std": 0.3565270006656647,
      "step": 3115
    },
    {
      "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": 0.0015625,
      "completions/max_length": 504.2,
      "completions/mean_length": 387.425,
      "completions/min_length": 270.8,
      "epoch": 0.8678720445062587,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6656712293624878,
      "kl": 0.046086785243824126,
      "learning_rate": 4.6981193090401604e-08,
      "loss": 0.0018435075879096984,
      "reward": 3.188760280609131,
      "reward_std": 0.400802755355835,
      "rewards/IngredientFormatReward/mean": 0.9972656011581421,
      "rewards/IngredientFormatReward/std": 0.02613703105598688,
      "rewards/IngredientMatchReward/mean": 0.6525930047035218,
      "rewards/IngredientMatchReward/std": 0.290238493680954,
      "rewards/IngredientQuantityMatchReward/mean": 0.701401686668396,
      "rewards/IngredientQuantityMatchReward/std": 0.39277478456497195,
      "rewards/TotalKcalExactMatchReward/mean": 0.8375,
      "rewards/TotalKcalExactMatchReward/std": 0.36130423545837403,
      "step": 3120
    },
    {
      "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": 0.0078125,
      "completions/max_length": 497.0,
      "completions/mean_length": 390.6546875,
      "completions/min_length": 247.8,
      "epoch": 0.8692628650904033,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6652013659477234,
      "kl": 0.04941854891367257,
      "learning_rate": 4.601269957321091e-08,
      "loss": 0.00197696965187788,
      "reward": 3.116376447677612,
      "reward_std": 0.41209186911582946,
      "rewards/IngredientFormatReward/mean": 0.9899330377578736,
      "rewards/IngredientFormatReward/std": 0.053738721460103986,
      "rewards/IngredientMatchReward/mean": 0.6498803496360779,
      "rewards/IngredientMatchReward/std": 0.2805606722831726,
      "rewards/IngredientQuantityMatchReward/mean": 0.670313024520874,
      "rewards/IngredientQuantityMatchReward/std": 0.41227946877479554,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.378220134973526,
      "step": 3125
    },
    {
      "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": 0.0140625,
      "completions/max_length": 510.2,
      "completions/mean_length": 391.4265625,
      "completions/min_length": 255.4,
      "epoch": 0.8706536856745479,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7125198245048523,
      "kl": 0.04970926723908633,
      "learning_rate": 4.5053811142439056e-08,
      "loss": 0.0019885823130607605,
      "reward": 2.987143039703369,
      "reward_std": 0.42771296501159667,
      "rewards/IngredientFormatReward/mean": 0.987291669845581,
      "rewards/IngredientFormatReward/std": 0.082947389036417,
      "rewards/IngredientMatchReward/mean": 0.6242001533508301,
      "rewards/IngredientMatchReward/std": 0.28825379014015196,
      "rewards/IngredientQuantityMatchReward/mean": 0.6147137761116028,
      "rewards/IngredientQuantityMatchReward/std": 0.42187526226043703,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.395613294839859,
      "step": 3130
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.6,
      "completions/mean_length": 387.49375,
      "completions/min_length": 264.4,
      "epoch": 0.8720445062586927,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6735195517539978,
      "kl": 0.048319844016805294,
      "learning_rate": 4.410454808545144e-08,
      "loss": 0.0019330214709043503,
      "reward": 3.125268268585205,
      "reward_std": 0.42156311869621277,
      "rewards/IngredientFormatReward/mean": 0.9879371166229248,
      "rewards/IngredientFormatReward/std": 0.09172794967889786,
      "rewards/IngredientMatchReward/mean": 0.6832422733306884,
      "rewards/IngredientMatchReward/std": 0.283637547492981,
      "rewards/IngredientQuantityMatchReward/mean": 0.652526342868805,
      "rewards/IngredientQuantityMatchReward/std": 0.4098579943180084,
      "rewards/TotalKcalExactMatchReward/mean": 0.8015625,
      "rewards/TotalKcalExactMatchReward/std": 0.3839901268482208,
      "step": 3135
    },
    {
      "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": 0.009375,
      "completions/max_length": 504.6,
      "completions/mean_length": 387.3328125,
      "completions/min_length": 260.6,
      "epoch": 0.8734353268428373,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6666101813316345,
      "kl": 0.04869773075915873,
      "learning_rate": 4.316493048596786e-08,
      "loss": 0.0019479818642139436,
      "reward": 3.1016587257385253,
      "reward_std": 0.4278287351131439,
      "rewards/IngredientFormatReward/mean": 0.9838541746139526,
      "rewards/IngredientFormatReward/std": 0.08848539292812348,
      "rewards/IngredientMatchReward/mean": 0.6473877549171447,
      "rewards/IngredientMatchReward/std": 0.3035359501838684,
      "rewards/IngredientQuantityMatchReward/mean": 0.6501043200492859,
      "rewards/IngredientQuantityMatchReward/std": 0.4268841803073883,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.37457784414291384,
      "step": 3140
    },
    {
      "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": 0.003125,
      "completions/max_length": 493.6,
      "completions/mean_length": 387.4078125,
      "completions/min_length": 255.6,
      "epoch": 0.874826147426982,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6627177596092224,
      "kl": 0.04510366797912866,
      "learning_rate": 4.2234978223637365e-08,
      "loss": 0.0018041422590613365,
      "reward": 3.216240119934082,
      "reward_std": 0.36082149744033815,
      "rewards/IngredientFormatReward/mean": 0.9941145896911621,
      "rewards/IngredientFormatReward/std": 0.05023606196045875,
      "rewards/IngredientMatchReward/mean": 0.674437654018402,
      "rewards/IngredientMatchReward/std": 0.28741170167922975,
      "rewards/IngredientQuantityMatchReward/mean": 0.7445629119873047,
      "rewards/IngredientQuantityMatchReward/std": 0.37724101543426514,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.39241496920585633,
      "step": 3145
    },
    {
      "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": 0.0015625,
      "completions/max_length": 490.4,
      "completions/mean_length": 392.046875,
      "completions/min_length": 276.2,
      "epoch": 0.8762169680111266,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.609026312828064,
      "kl": 0.046659936918877064,
      "learning_rate": 4.1314710973617994e-08,
      "loss": 0.001866631954908371,
      "reward": 3.122904109954834,
      "reward_std": 0.34099590182304385,
      "rewards/IngredientFormatReward/mean": 0.9961904764175415,
      "rewards/IngredientFormatReward/std": 0.03287437073886394,
      "rewards/IngredientMatchReward/mean": 0.6628584146499634,
      "rewards/IngredientMatchReward/std": 0.28121124804019926,
      "rewards/IngredientQuantityMatchReward/mean": 0.6841677546501159,
      "rewards/IngredientQuantityMatchReward/std": 0.39395946860313413,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.4074713945388794,
      "step": 3150
    },
    {
      "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": 0.0125,
      "completions/max_length": 509.8,
      "completions/mean_length": 390.603125,
      "completions/min_length": 265.4,
      "epoch": 0.8776077885952712,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7134307622909546,
      "kl": 0.04822455393150449,
      "learning_rate": 4.040414820616006e-08,
      "loss": 0.0019289350137114524,
      "reward": 3.003550624847412,
      "reward_std": 0.39522513151168825,
      "rewards/IngredientFormatReward/mean": 0.985182273387909,
      "rewards/IngredientFormatReward/std": 0.08801141232252122,
      "rewards/IngredientMatchReward/mean": 0.6495293974876404,
      "rewards/IngredientMatchReward/std": 0.2960770189762115,
      "rewards/IngredientQuantityMatchReward/mean": 0.6282139658927918,
      "rewards/IngredientQuantityMatchReward/std": 0.4116261124610901,
      "rewards/TotalKcalExactMatchReward/mean": 0.740625,
      "rewards/TotalKcalExactMatchReward/std": 0.43001351356506345,
      "step": 3155
    },
    {
      "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": 0.0125,
      "completions/max_length": 506.4,
      "completions/mean_length": 393.2734375,
      "completions/min_length": 262.8,
      "epoch": 0.8789986091794159,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6307358741760254,
      "kl": 0.04441832688171417,
      "learning_rate": 3.950330918619488e-08,
      "loss": 0.0017767375335097312,
      "reward": 3.0464088916778564,
      "reward_std": 0.45417346954345705,
      "rewards/IngredientFormatReward/mean": 0.9859703660011292,
      "rewards/IngredientFormatReward/std": 0.10039616972208024,
      "rewards/IngredientMatchReward/mean": 0.6358823299407959,
      "rewards/IngredientMatchReward/std": 0.2765005111694336,
      "rewards/IngredientQuantityMatchReward/mean": 0.6792436122894288,
      "rewards/IngredientQuantityMatchReward/std": 0.39857603907585143,
      "rewards/TotalKcalExactMatchReward/mean": 0.7453125,
      "rewards/TotalKcalExactMatchReward/std": 0.4332336366176605,
      "step": 3160
    },
    {
      "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": 0.0078125,
      "completions/max_length": 504.6,
      "completions/mean_length": 391.35,
      "completions/min_length": 274.2,
      "epoch": 0.8803894297635605,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6311498880386353,
      "kl": 0.04784476771019399,
      "learning_rate": 3.861221297292655e-08,
      "loss": 0.0019135732203722,
      "reward": 3.1883824348449705,
      "reward_std": 0.36939600110054016,
      "rewards/IngredientFormatReward/mean": 0.9859598278999329,
      "rewards/IngredientFormatReward/std": 0.10982539653778076,
      "rewards/IngredientMatchReward/mean": 0.6296354055404663,
      "rewards/IngredientMatchReward/std": 0.2733914017677307,
      "rewards/IngredientQuantityMatchReward/mean": 0.7274746179580689,
      "rewards/IngredientQuantityMatchReward/std": 0.37859617471694945,
      "rewards/TotalKcalExactMatchReward/mean": 0.8453125,
      "rewards/TotalKcalExactMatchReward/std": 0.35329467952251437,
      "step": 3165
    },
    {
      "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": 0.0125,
      "completions/max_length": 513.0,
      "completions/mean_length": 390.471875,
      "completions/min_length": 248.6,
      "epoch": 0.8817802503477051,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.5605978965759277,
      "kl": 0.047909293626435104,
      "learning_rate": 3.7730878419429066e-08,
      "loss": 0.0019164547324180604,
      "reward": 3.0932684421539305,
      "reward_std": 0.42589526176452636,
      "rewards/IngredientFormatReward/mean": 0.9879092216491699,
      "rewards/IngredientFormatReward/std": 0.10345041304826737,
      "rewards/IngredientMatchReward/mean": 0.6832546114921569,
      "rewards/IngredientMatchReward/std": 0.2867938756942749,
      "rewards/IngredientQuantityMatchReward/mean": 0.6080420970916748,
      "rewards/IngredientQuantityMatchReward/std": 0.4464531302452087,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3797110438346863,
      "step": 3170
    },
    {
      "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": 0.00625,
      "completions/max_length": 510.8,
      "completions/mean_length": 392.5828125,
      "completions/min_length": 275.8,
      "epoch": 0.8831710709318498,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6873921751976013,
      "kl": 0.04381905437912792,
      "learning_rate": 3.685932417224702e-08,
      "loss": 0.0017528809607028962,
      "reward": 3.072804307937622,
      "reward_std": 0.4026482582092285,
      "rewards/IngredientFormatReward/mean": 0.99140625,
      "rewards/IngredientFormatReward/std": 0.07906749099493027,
      "rewards/IngredientMatchReward/mean": 0.6443638324737548,
      "rewards/IngredientMatchReward/std": 0.28569428324699403,
      "rewards/IngredientQuantityMatchReward/mean": 0.6745341658592224,
      "rewards/IngredientQuantityMatchReward/std": 0.4039541006088257,
      "rewards/TotalKcalExactMatchReward/mean": 0.7625,
      "rewards/TotalKcalExactMatchReward/std": 0.41969347596168516,
      "step": 3175
    },
    {
      "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": 0.0140625,
      "completions/max_length": 508.6,
      "completions/mean_length": 389.384375,
      "completions/min_length": 250.6,
      "epoch": 0.8845618915159944,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7061554193496704,
      "kl": 0.04746296023949981,
      "learning_rate": 3.599756867100184e-08,
      "loss": 0.0018984757363796235,
      "reward": 3.1786093711853027,
      "reward_std": 0.42314148545265196,
      "rewards/IngredientFormatReward/mean": 0.9849218845367431,
      "rewards/IngredientFormatReward/std": 0.10753473863005639,
      "rewards/IngredientMatchReward/mean": 0.7073313474655152,
      "rewards/IngredientMatchReward/std": 0.26829246878623964,
      "rewards/IngredientQuantityMatchReward/mean": 0.6582311034202576,
      "rewards/IngredientQuantityMatchReward/std": 0.3875416398048401,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3658925950527191,
      "step": 3180
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.8,
      "completions/mean_length": 391.9296875,
      "completions/min_length": 267.0,
      "epoch": 0.885952712100139,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6179981231689453,
      "kl": 0.04594576419331133,
      "learning_rate": 3.5145630148000985e-08,
      "loss": 0.0018379855901002884,
      "reward": 3.1979689598083496,
      "reward_std": 0.3995221495628357,
      "rewards/IngredientFormatReward/mean": 0.9856621861457825,
      "rewards/IngredientFormatReward/std": 0.10145160276442766,
      "rewards/IngredientMatchReward/mean": 0.6657750487327576,
      "rewards/IngredientMatchReward/std": 0.28745636343955994,
      "rewards/IngredientQuantityMatchReward/mean": 0.6980942010879516,
      "rewards/IngredientQuantityMatchReward/std": 0.4006811559200287,
      "rewards/TotalKcalExactMatchReward/mean": 0.8484375,
      "rewards/TotalKcalExactMatchReward/std": 0.3429839700460434,
      "step": 3185
    },
    {
      "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": 0.0125,
      "completions/max_length": 503.4,
      "completions/mean_length": 397.546875,
      "completions/min_length": 297.6,
      "epoch": 0.8873435326842837,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6583110094070435,
      "kl": 0.2816643947735429,
      "learning_rate": 3.430352662785247e-08,
      "loss": 0.011268769949674606,
      "reward": 3.1793424606323244,
      "reward_std": 0.41445272564888,
      "rewards/IngredientFormatReward/mean": 0.984973955154419,
      "rewards/IngredientFormatReward/std": 0.09121790900826454,
      "rewards/IngredientMatchReward/mean": 0.6691094040870667,
      "rewards/IngredientMatchReward/std": 0.2692943513393402,
      "rewards/IngredientQuantityMatchReward/mean": 0.6674466133117676,
      "rewards/IngredientQuantityMatchReward/std": 0.3967915832996368,
      "rewards/TotalKcalExactMatchReward/mean": 0.8578125,
      "rewards/TotalKcalExactMatchReward/std": 0.34571763277053835,
      "step": 3190
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.6,
      "completions/mean_length": 389.496875,
      "completions/min_length": 263.6,
      "epoch": 0.8887343532684284,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6590292453765869,
      "kl": 0.05054741660133004,
      "learning_rate": 3.347127592708343e-08,
      "loss": 0.0020219020545482634,
      "reward": 3.209326696395874,
      "reward_std": 0.3645615816116333,
      "rewards/IngredientFormatReward/mean": 0.9900520801544189,
      "rewards/IngredientFormatReward/std": 0.08570180833339691,
      "rewards/IngredientMatchReward/mean": 0.6802461624145508,
      "rewards/IngredientMatchReward/std": 0.2965859532356262,
      "rewards/IngredientQuantityMatchReward/mean": 0.7109034299850464,
      "rewards/IngredientQuantityMatchReward/std": 0.38022114634513854,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3564129650592804,
      "step": 3195
    },
    {
      "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": 0.0125,
      "completions/max_length": 511.8,
      "completions/mean_length": 388.940625,
      "completions/min_length": 272.0,
      "epoch": 0.8901251738525731,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7058317065238953,
      "kl": 0.047892588051036,
      "learning_rate": 3.264889565376339e-08,
      "loss": 0.0019158922135829926,
      "reward": 3.0706414222717284,
      "reward_std": 0.4435698390007019,
      "rewards/IngredientFormatReward/mean": 0.9884374976158142,
      "rewards/IngredientFormatReward/std": 0.08515960946679116,
      "rewards/IngredientMatchReward/mean": 0.6230040967464447,
      "rewards/IngredientMatchReward/std": 0.2796986043453217,
      "rewards/IngredientQuantityMatchReward/mean": 0.6310748338699341,
      "rewards/IngredientQuantityMatchReward/std": 0.4169824719429016,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3632709324359894,
      "step": 3200
    },
    {
      "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": 0.009375,
      "completions/max_length": 506.2,
      "completions/mean_length": 388.6578125,
      "completions/min_length": 248.2,
      "epoch": 0.8915159944367177,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6913332343101501,
      "kl": 0.05210544446017593,
      "learning_rate": 3.183640320713138e-08,
      "loss": 0.0020846392959356307,
      "reward": 3.0734073162078857,
      "reward_std": 0.37518334984779356,
      "rewards/IngredientFormatReward/mean": 0.9895312547683716,
      "rewards/IngredientFormatReward/std": 0.0965784803032875,
      "rewards/IngredientMatchReward/mean": 0.6489539980888367,
      "rewards/IngredientMatchReward/std": 0.2867078483104706,
      "rewards/IngredientQuantityMatchReward/mean": 0.6114845871925354,
      "rewards/IngredientQuantityMatchReward/std": 0.43328599333763124,
      "rewards/TotalKcalExactMatchReward/mean": 0.8234375,
      "rewards/TotalKcalExactMatchReward/std": 0.37979313135147097,
      "step": 3205
    },
    {
      "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": 0.0078125,
      "completions/max_length": 509.0,
      "completions/mean_length": 395.76875,
      "completions/min_length": 270.6,
      "epoch": 0.8929068150208623,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6467673182487488,
      "kl": 0.047686185827478765,
      "learning_rate": 3.103381577722813e-08,
      "loss": 0.0019076012074947357,
      "reward": 3.168344497680664,
      "reward_std": 0.37602880597114563,
      "rewards/IngredientFormatReward/mean": 0.9888020753860474,
      "rewards/IngredientFormatReward/std": 0.0813770966604352,
      "rewards/IngredientMatchReward/mean": 0.6629953980445862,
      "rewards/IngredientMatchReward/std": 0.28421200513839723,
      "rewards/IngredientQuantityMatchReward/mean": 0.7024844765663147,
      "rewards/IngredientQuantityMatchReward/std": 0.39068577289581297,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3665438830852509,
      "step": 3210
    },
    {
      "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": 0.00625,
      "completions/max_length": 503.0,
      "completions/mean_length": 394.2890625,
      "completions/min_length": 269.0,
      "epoch": 0.894297635605007,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.644658625125885,
      "kl": 0.044069158774800596,
      "learning_rate": 3.024115034453223e-08,
      "loss": 0.0017627470195293426,
      "reward": 3.1621095657348635,
      "reward_std": 0.38854520320892333,
      "rewards/IngredientFormatReward/mean": 0.9934375047683716,
      "rewards/IngredientFormatReward/std": 0.053338292986154556,
      "rewards/IngredientMatchReward/mean": 0.6316596031188965,
      "rewards/IngredientMatchReward/std": 0.2820505201816559,
      "rewards/IngredientQuantityMatchReward/mean": 0.6838874697685242,
      "rewards/IngredientQuantityMatchReward/std": 0.4012498319149017,
      "rewards/TotalKcalExactMatchReward/mean": 0.853125,
      "rewards/TotalKcalExactMatchReward/std": 0.3469159841537476,
      "step": 3215
    },
    {
      "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": 0.0046875,
      "completions/max_length": 499.6,
      "completions/mean_length": 385.9265625,
      "completions/min_length": 242.8,
      "epoch": 0.8956884561891516,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.6403445601463318,
      "kl": 0.049373555136844514,
      "learning_rate": 2.945842367960083e-08,
      "loss": 0.0019750598818063735,
      "reward": 3.1126923084259035,
      "reward_std": 0.3951055645942688,
      "rewards/IngredientFormatReward/mean": 0.9936458230018616,
      "rewards/IngredientFormatReward/std": 0.04035495463758707,
      "rewards/IngredientMatchReward/mean": 0.6792900562286377,
      "rewards/IngredientMatchReward/std": 0.27618914246559145,
      "rewards/IngredientQuantityMatchReward/mean": 0.6788189053535462,
      "rewards/IngredientQuantityMatchReward/std": 0.41655081510543823,
      "rewards/TotalKcalExactMatchReward/mean": 0.7609375,
      "rewards/TotalKcalExactMatchReward/std": 0.4172710537910461,
      "step": 3220
    },
    {
      "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": 0.009375,
      "completions/max_length": 507.4,
      "completions/mean_length": 390.265625,
      "completions/min_length": 257.2,
      "epoch": 0.8970792767732962,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.717547595500946,
      "kl": 0.04932956439442933,
      "learning_rate": 2.8685652342714862e-08,
      "loss": 0.0019732709974050523,
      "reward": 3.148496913909912,
      "reward_std": 0.35338476300239563,
      "rewards/IngredientFormatReward/mean": 0.9878236651420593,
      "rewards/IngredientFormatReward/std": 0.07924887351691723,
      "rewards/IngredientMatchReward/mean": 0.6488428235054016,
      "rewards/IngredientMatchReward/std": 0.2869942843914032,
      "rewards/IngredientQuantityMatchReward/mean": 0.7008928775787353,
      "rewards/IngredientQuantityMatchReward/std": 0.3995750486850739,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.39261109232902525,
      "step": 3225
    },
    {
      "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": 0.0109375,
      "completions/max_length": 506.4,
      "completions/mean_length": 392.1296875,
      "completions/min_length": 259.8,
      "epoch": 0.8984700973574409,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6339384913444519,
      "kl": 0.04682166455313563,
      "learning_rate": 2.7922852683528897e-08,
      "loss": 0.0018730312585830688,
      "reward": 3.091051721572876,
      "reward_std": 0.3852741241455078,
      "rewards/IngredientFormatReward/mean": 0.983017110824585,
      "rewards/IngredientFormatReward/std": 0.09547805488109588,
      "rewards/IngredientMatchReward/mean": 0.6440420389175415,
      "rewards/IngredientMatchReward/std": 0.29824180603027345,
      "rewards/IngredientQuantityMatchReward/mean": 0.6530550360679627,
      "rewards/IngredientQuantityMatchReward/std": 0.4097617506980896,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.36531429886817934,
      "step": 3230
    },
    {
      "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": 0.009375,
      "completions/max_length": 510.8,
      "completions/mean_length": 393.89375,
      "completions/min_length": 267.8,
      "epoch": 0.8998609179415855,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.5954923629760742,
      "kl": 0.04409891550894827,
      "learning_rate": 2.7170040840724716e-08,
      "loss": 0.0017639545723795891,
      "reward": 3.036423349380493,
      "reward_std": 0.4317321002483368,
      "rewards/IngredientFormatReward/mean": 0.9889583349227905,
      "rewards/IngredientFormatReward/std": 0.08621723800897599,
      "rewards/IngredientMatchReward/mean": 0.6406045317649841,
      "rewards/IngredientMatchReward/std": 0.2745776265859604,
      "rewards/IngredientQuantityMatchReward/mean": 0.5740479826927185,
      "rewards/IngredientQuantityMatchReward/std": 0.43019055128097533,
      "rewards/TotalKcalExactMatchReward/mean": 0.8328125,
      "rewards/TotalKcalExactMatchReward/std": 0.3640967130661011,
      "step": 3235
    },
    {
      "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": 0.0078125,
      "completions/max_length": 512.2,
      "completions/mean_length": 392.6125,
      "completions/min_length": 269.6,
      "epoch": 0.9012517385257302,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6983107328414917,
      "kl": 0.04888640344142914,
      "learning_rate": 2.6427232741670357e-08,
      "loss": 0.0019554536789655685,
      "reward": 3.089215040206909,
      "reward_std": 0.4097185730934143,
      "rewards/IngredientFormatReward/mean": 0.9904798984527587,
      "rewards/IngredientFormatReward/std": 0.07944944500923157,
      "rewards/IngredientMatchReward/mean": 0.6514082670211792,
      "rewards/IngredientMatchReward/std": 0.28968934416770936,
      "rewards/IngredientQuantityMatchReward/mean": 0.6317018032073974,
      "rewards/IngredientQuantityMatchReward/std": 0.42319734692573546,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.38177411556243895,
      "step": 3240
    },
    {
      "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": 0.00625,
      "completions/max_length": 509.4,
      "completions/mean_length": 394.903125,
      "completions/min_length": 273.4,
      "epoch": 0.9026425591098748,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.5831993222236633,
      "kl": 0.04227819531224668,
      "learning_rate": 2.5694444102082936e-08,
      "loss": 0.0016911562532186508,
      "reward": 3.1611047267913817,
      "reward_std": 0.3636060357093811,
      "rewards/IngredientFormatReward/mean": 0.99296875,
      "rewards/IngredientFormatReward/std": 0.06629023440182209,
      "rewards/IngredientMatchReward/mean": 0.6555952429771423,
      "rewards/IngredientMatchReward/std": 0.2678823709487915,
      "rewards/IngredientQuantityMatchReward/mean": 0.7016033172607422,
      "rewards/IngredientQuantityMatchReward/std": 0.388336056470871,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.3912253439426422,
      "step": 3245
    },
    {
      "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": 0.0046875,
      "completions/max_length": 507.4,
      "completions/mean_length": 392.5296875,
      "completions/min_length": 274.6,
      "epoch": 0.9040333796940194,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.641535758972168,
      "kl": 0.0460631252033636,
      "learning_rate": 2.49716904256963e-08,
      "loss": 0.0018425412476062776,
      "reward": 3.134316158294678,
      "reward_std": 0.3648345053195953,
      "rewards/IngredientFormatReward/mean": 0.995892858505249,
      "rewards/IngredientFormatReward/std": 0.03714258074760437,
      "rewards/IngredientMatchReward/mean": 0.6300421595573426,
      "rewards/IngredientMatchReward/std": 0.2943162739276886,
      "rewards/IngredientQuantityMatchReward/mean": 0.6911936283111573,
      "rewards/IngredientQuantityMatchReward/std": 0.39555092453956603,
      "rewards/TotalKcalExactMatchReward/mean": 0.8171875,
      "rewards/TotalKcalExactMatchReward/std": 0.3763595879077911,
      "step": 3250
    },
    {
      "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": 0.01875,
      "completions/max_length": 507.8,
      "completions/mean_length": 392.590625,
      "completions/min_length": 259.0,
      "epoch": 0.9054242002781642,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6454972624778748,
      "kl": 0.04925240727607161,
      "learning_rate": 2.425898700393253e-08,
      "loss": 0.0019701555371284484,
      "reward": 2.9525651931762695,
      "reward_std": 0.43930028676986693,
      "rewards/IngredientFormatReward/mean": 0.9813281297683716,
      "rewards/IngredientFormatReward/std": 0.11844980865716934,
      "rewards/IngredientMatchReward/mean": 0.5760538458824158,
      "rewards/IngredientMatchReward/std": 0.29196472764015197,
      "rewards/IngredientQuantityMatchReward/mean": 0.6451832890510559,
      "rewards/IngredientQuantityMatchReward/std": 0.42400979399681094,
      "rewards/TotalKcalExactMatchReward/mean": 0.75,
      "rewards/TotalKcalExactMatchReward/std": 0.4188802659511566,
      "step": 3255
    },
    {
      "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": 0.0078125,
      "completions/max_length": 503.4,
      "completions/mean_length": 389.4859375,
      "completions/min_length": 275.2,
      "epoch": 0.9068150208623088,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6864833235740662,
      "kl": 0.045732753025367855,
      "learning_rate": 2.355634891557906e-08,
      "loss": 0.0018294647336006165,
      "reward": 3.015725040435791,
      "reward_std": 0.40643337965011594,
      "rewards/IngredientFormatReward/mean": 0.9921875,
      "rewards/IngredientFormatReward/std": 0.05261292904615402,
      "rewards/IngredientMatchReward/mean": 0.6334901928901673,
      "rewards/IngredientMatchReward/std": 0.3036071300506592,
      "rewards/IngredientQuantityMatchReward/mean": 0.5759848117828369,
      "rewards/IngredientQuantityMatchReward/std": 0.4259013533592224,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3635615885257721,
      "step": 3260
    },
    {
      "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": 0.0078125,
      "completions/max_length": 503.6,
      "completions/mean_length": 388.340625,
      "completions/min_length": 252.6,
      "epoch": 0.9082058414464534,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7106591463088989,
      "kl": 0.04727880118880421,
      "learning_rate": 2.2863791026469235e-08,
      "loss": 0.0018911713734269143,
      "reward": 3.2334872245788575,
      "reward_std": 0.37439358830451963,
      "rewards/IngredientFormatReward/mean": 0.9897135257720947,
      "rewards/IngredientFormatReward/std": 0.05642170626670122,
      "rewards/IngredientMatchReward/mean": 0.653626000881195,
      "rewards/IngredientMatchReward/std": 0.2840902924537659,
      "rewards/IngredientQuantityMatchReward/mean": 0.7245226621627807,
      "rewards/IngredientQuantityMatchReward/std": 0.38535062670707704,
      "rewards/TotalKcalExactMatchReward/mean": 0.865625,
      "rewards/TotalKcalExactMatchReward/std": 0.32322187423706056,
      "step": 3265
    },
    {
      "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": 0.015625,
      "completions/max_length": 511.4,
      "completions/mean_length": 393.809375,
      "completions/min_length": 271.6,
      "epoch": 0.9095966620305981,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.626308023929596,
      "kl": 0.047429022891446945,
      "learning_rate": 2.2181327989167998e-08,
      "loss": 0.0018970862030982972,
      "reward": 3.040598201751709,
      "reward_std": 0.42209060192108155,
      "rewards/IngredientFormatReward/mean": 0.9875,
      "rewards/IngredientFormatReward/std": 0.08148004412651062,
      "rewards/IngredientMatchReward/mean": 0.6066759705543519,
      "rewards/IngredientMatchReward/std": 0.3146925806999207,
      "rewards/IngredientQuantityMatchReward/mean": 0.637047290802002,
      "rewards/IngredientQuantityMatchReward/std": 0.4168846786022186,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.38911654353141784,
      "step": 3270
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.4,
      "completions/mean_length": 393.825,
      "completions/min_length": 284.8,
      "epoch": 0.9109874826147427,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.5958412289619446,
      "kl": 0.049182718200609085,
      "learning_rate": 2.1508974242661628e-08,
      "loss": 0.0019672702997922896,
      "reward": 3.096751403808594,
      "reward_std": 0.3994257628917694,
      "rewards/IngredientFormatReward/mean": 0.9866145730018616,
      "rewards/IngredientFormatReward/std": 0.0923412561416626,
      "rewards/IngredientMatchReward/mean": 0.6960999369621277,
      "rewards/IngredientMatchReward/std": 0.27538976073265076,
      "rewards/IngredientQuantityMatchReward/mean": 0.6546619057655334,
      "rewards/IngredientQuantityMatchReward/std": 0.37647982835769656,
      "rewards/TotalKcalExactMatchReward/mean": 0.759375,
      "rewards/TotalKcalExactMatchReward/std": 0.41687182784080506,
      "step": 3275
    },
    {
      "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": 0.00625,
      "completions/max_length": 503.8,
      "completions/mean_length": 387.3296875,
      "completions/min_length": 251.0,
      "epoch": 0.9123783031988874,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6608989238739014,
      "kl": 0.04685659667011351,
      "learning_rate": 2.084674401205261e-08,
      "loss": 0.0018743416294455528,
      "reward": 3.0759621143341063,
      "reward_std": 0.3722398400306702,
      "rewards/IngredientFormatReward/mean": 0.9926562428474426,
      "rewards/IngredientFormatReward/std": 0.05678885132074356,
      "rewards/IngredientMatchReward/mean": 0.6163002371788024,
      "rewards/IngredientMatchReward/std": 0.2801432520151138,
      "rewards/IngredientQuantityMatchReward/mean": 0.6420055627822876,
      "rewards/IngredientQuantityMatchReward/std": 0.435650622844696,
      "rewards/TotalKcalExactMatchReward/mean": 0.825,
      "rewards/TotalKcalExactMatchReward/std": 0.35427531599998474,
      "step": 3280
    },
    {
      "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": 0.003125,
      "completions/max_length": 496.4,
      "completions/mean_length": 388.78125,
      "completions/min_length": 271.8,
      "epoch": 0.913769123783032,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6058049201965332,
      "kl": 0.04507222096435726,
      "learning_rate": 2.019465130825837e-08,
      "loss": 0.001802671328186989,
      "reward": 3.2701576232910154,
      "reward_std": 0.345516049861908,
      "rewards/IngredientFormatReward/mean": 0.9930729150772095,
      "rewards/IngredientFormatReward/std": 0.048128366470336914,
      "rewards/IngredientMatchReward/mean": 0.664314353466034,
      "rewards/IngredientMatchReward/std": 0.2860167622566223,
      "rewards/IngredientQuantityMatchReward/mean": 0.7299577355384826,
      "rewards/IngredientQuantityMatchReward/std": 0.3946863770484924,
      "rewards/TotalKcalExactMatchReward/mean": 0.8828125,
      "rewards/TotalKcalExactMatchReward/std": 0.31640257239341735,
      "step": 3285
    },
    {
      "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": 0.00625,
      "completions/max_length": 507.8,
      "completions/mean_length": 396.428125,
      "completions/min_length": 276.4,
      "epoch": 0.9151599443671766,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6639472246170044,
      "kl": 0.0454582569655031,
      "learning_rate": 1.9552709927715072e-08,
      "loss": 0.0018183968961238862,
      "reward": 3.0982945442199705,
      "reward_std": 0.3722270905971527,
      "rewards/IngredientFormatReward/mean": 0.9878385543823243,
      "rewards/IngredientFormatReward/std": 0.07973492741584778,
      "rewards/IngredientMatchReward/mean": 0.6462369799613953,
      "rewards/IngredientMatchReward/std": 0.289064759016037,
      "rewards/IngredientQuantityMatchReward/mean": 0.6142190217971801,
      "rewards/IngredientQuantityMatchReward/std": 0.42048256993293764,
      "rewards/TotalKcalExactMatchReward/mean": 0.85,
      "rewards/TotalKcalExactMatchReward/std": 0.32386162132024765,
      "step": 3290
    },
    {
      "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": 0.015625,
      "completions/max_length": 512.0,
      "completions/mean_length": 392.221875,
      "completions/min_length": 265.8,
      "epoch": 0.9165507649513213,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.591355562210083,
      "kl": 0.04644181546755135,
      "learning_rate": 1.8920933452085396e-08,
      "loss": 0.0018579160794615745,
      "reward": 2.946201133728027,
      "reward_std": 0.4095645725727081,
      "rewards/IngredientFormatReward/mean": 0.9810007452964783,
      "rewards/IngredientFormatReward/std": 0.11646474674344062,
      "rewards/IngredientMatchReward/mean": 0.6444252252578735,
      "rewards/IngredientMatchReward/std": 0.3023004472255707,
      "rewards/IngredientQuantityMatchReward/mean": 0.6270251989364624,
      "rewards/IngredientQuantityMatchReward/std": 0.4267409145832062,
      "rewards/TotalKcalExactMatchReward/mean": 0.69375,
      "rewards/TotalKcalExactMatchReward/std": 0.44755579829216,
      "step": 3295
    },
    {
      "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": 0.009375,
      "completions/max_length": 502.8,
      "completions/mean_length": 391.75625,
      "completions/min_length": 262.0,
      "epoch": 0.9179415855354659,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6602470874786377,
      "kl": 0.04800860662944615,
      "learning_rate": 1.829933524797156e-08,
      "loss": 0.0019200056791305541,
      "reward": 3.1091113567352293,
      "reward_std": 0.396321576833725,
      "rewards/IngredientFormatReward/mean": 0.9884375095367431,
      "rewards/IngredientFormatReward/std": 0.08126041516661645,
      "rewards/IngredientMatchReward/mean": 0.6062586784362793,
      "rewards/IngredientMatchReward/std": 0.26619652807712557,
      "rewards/IngredientQuantityMatchReward/mean": 0.6925402522087097,
      "rewards/IngredientQuantityMatchReward/std": 0.39906328320503237,
      "rewards/TotalKcalExactMatchReward/mean": 0.821875,
      "rewards/TotalKcalExactMatchReward/std": 0.37994712591171265,
      "step": 3300
    },
    {
      "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": 0.00625,
      "completions/max_length": 506.2,
      "completions/mean_length": 392.1375,
      "completions/min_length": 275.0,
      "epoch": 0.9193324061196105,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6438077688217163,
      "kl": 0.05436852378770709,
      "learning_rate": 1.768792846663242e-08,
      "loss": 0.002175022102892399,
      "reward": 3.1007835388183596,
      "reward_std": 0.3859154999256134,
      "rewards/IngredientFormatReward/mean": 0.9934895753860473,
      "rewards/IngredientFormatReward/std": 0.0604776605963707,
      "rewards/IngredientMatchReward/mean": 0.6451841711997985,
      "rewards/IngredientMatchReward/std": 0.2823725491762161,
      "rewards/IngredientQuantityMatchReward/mean": 0.6699223518371582,
      "rewards/IngredientQuantityMatchReward/std": 0.41072106957435606,
      "rewards/TotalKcalExactMatchReward/mean": 0.7921875,
      "rewards/TotalKcalExactMatchReward/std": 0.4017234563827515,
      "step": 3305
    },
    {
      "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": 0.00625,
      "completions/max_length": 507.0,
      "completions/mean_length": 386.634375,
      "completions/min_length": 266.6,
      "epoch": 0.9207232267037552,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 1.0080318450927734,
      "kl": 0.05075234514661133,
      "learning_rate": 1.708672604370509e-08,
      "loss": 0.002030384540557861,
      "reward": 2.9925003051757812,
      "reward_std": 0.419496089220047,
      "rewards/IngredientFormatReward/mean": 0.9936458349227906,
      "rewards/IngredientFormatReward/std": 0.05940539613366127,
      "rewards/IngredientMatchReward/mean": 0.6513331174850464,
      "rewards/IngredientMatchReward/std": 0.30027254223823546,
      "rewards/IngredientQuantityMatchReward/mean": 0.5662714242935181,
      "rewards/IngredientQuantityMatchReward/std": 0.43809983134269714,
      "rewards/TotalKcalExactMatchReward/mean": 0.78125,
      "rewards/TotalKcalExactMatchReward/std": 0.401913982629776,
      "step": 3310
    },
    {
      "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": 0.00625,
      "completions/max_length": 495.2,
      "completions/mean_length": 390.7421875,
      "completions/min_length": 254.4,
      "epoch": 0.9221140472878998,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6824638247489929,
      "kl": 0.04954786649905145,
      "learning_rate": 1.6495740698931282e-08,
      "loss": 0.0019818780943751334,
      "reward": 3.1271950244903564,
      "reward_std": 0.41434894800186156,
      "rewards/IngredientFormatReward/mean": 0.9919791698455811,
      "rewards/IngredientFormatReward/std": 0.05760362073779106,
      "rewards/IngredientMatchReward/mean": 0.6849448323249817,
      "rewards/IngredientMatchReward/std": 0.27029902338981626,
      "rewards/IngredientQuantityMatchReward/mean": 0.6799584984779358,
      "rewards/IngredientQuantityMatchReward/std": 0.3997422754764557,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.41917206048965455,
      "step": 3315
    },
    {
      "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": 0.0109375,
      "completions/max_length": 506.8,
      "completions/mean_length": 393.246875,
      "completions/min_length": 258.2,
      "epoch": 0.9235048678720446,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6570489406585693,
      "kl": 0.09846211015246809,
      "learning_rate": 1.5914984935888277e-08,
      "loss": 0.0039469413459300995,
      "reward": 3.0435446739196776,
      "reward_std": 0.44498496055603026,
      "rewards/IngredientFormatReward/mean": 0.9882663726806641,
      "rewards/IngredientFormatReward/std": 0.07785128802061081,
      "rewards/IngredientMatchReward/mean": 0.6336315870285034,
      "rewards/IngredientMatchReward/std": 0.3072134733200073,
      "rewards/IngredientQuantityMatchReward/mean": 0.6185217320919036,
      "rewards/IngredientQuantityMatchReward/std": 0.417307448387146,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.37303715348243716,
      "step": 3320
    },
    {
      "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": 0.00625,
      "completions/max_length": 507.4,
      "completions/mean_length": 397.1890625,
      "completions/min_length": 289.6,
      "epoch": 0.9248956884561892,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.7150806188583374,
      "kl": 0.046813591942191125,
      "learning_rate": 1.5344471041724482e-08,
      "loss": 0.0018727334216237068,
      "reward": 3.0286718368530274,
      "reward_std": 0.37150052189826965,
      "rewards/IngredientFormatReward/mean": 0.9911458253860473,
      "rewards/IngredientFormatReward/std": 0.06259849891066552,
      "rewards/IngredientMatchReward/mean": 0.6338678121566772,
      "rewards/IngredientMatchReward/std": 0.29035924673080443,
      "rewards/IngredientQuantityMatchReward/mean": 0.6130332827568055,
      "rewards/IngredientQuantityMatchReward/std": 0.4491858184337616,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.39557355642318726,
      "step": 3325
    },
    {
      "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": 0.0109375,
      "completions/max_length": 499.4,
      "completions/mean_length": 387.3109375,
      "completions/min_length": 264.4,
      "epoch": 0.9262865090403338,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6601356267929077,
      "kl": 0.0573459361679852,
      "learning_rate": 1.4784211086899146e-08,
      "loss": 0.0022935055196285246,
      "reward": 3.205912399291992,
      "reward_std": 0.40079067945480346,
      "rewards/IngredientFormatReward/mean": 0.9884375095367431,
      "rewards/IngredientFormatReward/std": 0.08228911980986595,
      "rewards/IngredientMatchReward/mean": 0.6975455045700073,
      "rewards/IngredientMatchReward/std": 0.2682806015014648,
      "rewards/IngredientQuantityMatchReward/mean": 0.6918044209480285,
      "rewards/IngredientQuantityMatchReward/std": 0.39961135387420654,
      "rewards/TotalKcalExactMatchReward/mean": 0.828125,
      "rewards/TotalKcalExactMatchReward/std": 0.3684461772441864,
      "step": 3330
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.8,
      "completions/mean_length": 391.8828125,
      "completions/min_length": 260.2,
      "epoch": 0.9276773296244785,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 3.000927209854126,
      "kl": 0.05884796802420169,
      "learning_rate": 1.4234216924927377e-08,
      "loss": 0.0023534022271633147,
      "reward": 3.089488124847412,
      "reward_std": 0.4574960291385651,
      "rewards/IngredientFormatReward/mean": 0.9880059480667114,
      "rewards/IngredientFormatReward/std": 0.09700787290930749,
      "rewards/IngredientMatchReward/mean": 0.6292633891105652,
      "rewards/IngredientMatchReward/std": 0.2786643743515015,
      "rewards/IngredientQuantityMatchReward/mean": 0.6644062876701355,
      "rewards/IngredientQuantityMatchReward/std": 0.40214182138442994,
      "rewards/TotalKcalExactMatchReward/mean": 0.8078125,
      "rewards/TotalKcalExactMatchReward/std": 0.36265517473220826,
      "step": 3335
    },
    {
      "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": 0.0140625,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.290625,
      "completions/min_length": 271.6,
      "epoch": 0.9290681502086231,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6722849011421204,
      "kl": 0.04828175606671721,
      "learning_rate": 1.3694500192128977e-08,
      "loss": 0.0019316310063004494,
      "reward": 2.968636226654053,
      "reward_std": 0.45129722356796265,
      "rewards/IngredientFormatReward/mean": 0.9836458206176758,
      "rewards/IngredientFormatReward/std": 0.11726146638393402,
      "rewards/IngredientMatchReward/mean": 0.6276495575904846,
      "rewards/IngredientMatchReward/std": 0.28815919160842896,
      "rewards/IngredientQuantityMatchReward/mean": 0.6167159080505371,
      "rewards/IngredientQuantityMatchReward/std": 0.417825722694397,
      "rewards/TotalKcalExactMatchReward/mean": 0.740625,
      "rewards/TotalKcalExactMatchReward/std": 0.4330654442310333,
      "step": 3340
    },
    {
      "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": 0.0125,
      "completions/max_length": 511.8,
      "completions/mean_length": 394.0671875,
      "completions/min_length": 278.8,
      "epoch": 0.9304589707927677,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6125572323799133,
      "kl": 0.04887234810739756,
      "learning_rate": 1.3165072307382619e-08,
      "loss": 0.0019550468772649766,
      "reward": 3.2130596160888674,
      "reward_std": 0.39644322991371156,
      "rewards/IngredientFormatReward/mean": 0.9825446486473084,
      "rewards/IngredientFormatReward/std": 0.1088470995426178,
      "rewards/IngredientMatchReward/mean": 0.6876734018325805,
      "rewards/IngredientMatchReward/std": 0.29353204369544983,
      "rewards/IngredientQuantityMatchReward/mean": 0.6959666013717651,
      "rewards/IngredientQuantityMatchReward/std": 0.3904150426387787,
      "rewards/TotalKcalExactMatchReward/mean": 0.846875,
      "rewards/TotalKcalExactMatchReward/std": 0.33690011501312256,
      "step": 3345
    },
    {
      "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": 0.003125,
      "completions/max_length": 506.2,
      "completions/mean_length": 394.6640625,
      "completions/min_length": 282.6,
      "epoch": 0.9318497913769124,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.623755693435669,
      "kl": 0.045136950956657526,
      "learning_rate": 1.2645944471883995e-08,
      "loss": 0.0018055334687232972,
      "reward": 3.077497720718384,
      "reward_std": 0.3821293354034424,
      "rewards/IngredientFormatReward/mean": 0.9941257476806641,
      "rewards/IngredientFormatReward/std": 0.05168246813118458,
      "rewards/IngredientMatchReward/mean": 0.6398034691810608,
      "rewards/IngredientMatchReward/std": 0.2708088785409927,
      "rewards/IngredientQuantityMatchReward/mean": 0.6232560276985168,
      "rewards/IngredientQuantityMatchReward/std": 0.4233528017997742,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.3787771940231323,
      "step": 3350
    },
    {
      "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": 0.0125,
      "completions/max_length": 507.4,
      "completions/mean_length": 397.6359375,
      "completions/min_length": 283.0,
      "epoch": 0.933240611961057,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.615950345993042,
      "kl": 0.042504592309705914,
      "learning_rate": 1.2137127668908732e-08,
      "loss": 0.0017003992572426796,
      "reward": 3.1166717052459716,
      "reward_std": 0.3788813233375549,
      "rewards/IngredientFormatReward/mean": 0.9865625023841857,
      "rewards/IngredientFormatReward/std": 0.08676649853587151,
      "rewards/IngredientMatchReward/mean": 0.6693539142608642,
      "rewards/IngredientMatchReward/std": 0.283643913269043,
      "rewards/IngredientQuantityMatchReward/mean": 0.6466927766799927,
      "rewards/IngredientQuantityMatchReward/std": 0.42546847462654114,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3705523431301117,
      "step": 3355
    },
    {
      "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": 0.0125,
      "completions/max_length": 504.2,
      "completions/mean_length": 395.6390625,
      "completions/min_length": 281.0,
      "epoch": 0.9346314325452016,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.5828317403793335,
      "kl": 0.044918736536055805,
      "learning_rate": 1.163863266358045e-08,
      "loss": 0.001797039434313774,
      "reward": 3.166773557662964,
      "reward_std": 0.367119961977005,
      "rewards/IngredientFormatReward/mean": 0.985364580154419,
      "rewards/IngredientFormatReward/std": 0.08733292371034622,
      "rewards/IngredientMatchReward/mean": 0.6989391088485718,
      "rewards/IngredientMatchReward/std": 0.2778316617012024,
      "rewards/IngredientQuantityMatchReward/mean": 0.6699697375297546,
      "rewards/IngredientQuantityMatchReward/std": 0.39433544874191284,
      "rewards/TotalKcalExactMatchReward/mean": 0.8125,
      "rewards/TotalKcalExactMatchReward/std": 0.388651567697525,
      "step": 3360
    },
    {
      "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": 0.0140625,
      "completions/max_length": 501.0,
      "completions/mean_length": 390.125,
      "completions/min_length": 260.2,
      "epoch": 0.9360222531293463,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.627528190612793,
      "kl": 0.0473182033514604,
      "learning_rate": 1.1150470002642686e-08,
      "loss": 0.0018929582089185714,
      "reward": 2.9560650825500487,
      "reward_std": 0.39082145094871523,
      "rewards/IngredientFormatReward/mean": 0.9876562595367432,
      "rewards/IngredientFormatReward/std": 0.058681896328926085,
      "rewards/IngredientMatchReward/mean": 0.6209102034568786,
      "rewards/IngredientMatchReward/std": 0.2951041400432587,
      "rewards/IngredientQuantityMatchReward/mean": 0.6381236493587494,
      "rewards/IngredientQuantityMatchReward/std": 0.4160947024822235,
      "rewards/TotalKcalExactMatchReward/mean": 0.709375,
      "rewards/TotalKcalExactMatchReward/std": 0.45090446472167967,
      "step": 3365
    },
    {
      "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": 0.00625,
      "completions/max_length": 512.2,
      "completions/mean_length": 386.2125,
      "completions/min_length": 261.6,
      "epoch": 0.9374130737134909,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6412569284439087,
      "kl": 0.04862490389496088,
      "learning_rate": 1.0672650014235729e-08,
      "loss": 0.0019449492916464805,
      "reward": 3.171107292175293,
      "reward_std": 0.34414549469947814,
      "rewards/IngredientFormatReward/mean": 0.995520830154419,
      "rewards/IngredientFormatReward/std": 0.0420041311532259,
      "rewards/IngredientMatchReward/mean": 0.6497817516326905,
      "rewards/IngredientMatchReward/std": 0.2970588386058807,
      "rewards/IngredientQuantityMatchReward/mean": 0.7070546865463256,
      "rewards/IngredientQuantityMatchReward/std": 0.38046914935112,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.37668890953063966,
      "step": 3370
    },
    {
      "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": 0.009375,
      "completions/max_length": 513.0,
      "completions/mean_length": 396.796875,
      "completions/min_length": 276.8,
      "epoch": 0.9388038942976356,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.5388448238372803,
      "kl": 0.04376723191235214,
      "learning_rate": 1.0205182807678181e-08,
      "loss": 0.0017541015520691871,
      "reward": 3.140538787841797,
      "reward_std": 0.353511369228363,
      "rewards/IngredientFormatReward/mean": 0.9919270753860474,
      "rewards/IngredientFormatReward/std": 0.08860928863286972,
      "rewards/IngredientMatchReward/mean": 0.6582719564437867,
      "rewards/IngredientMatchReward/std": 0.256216949224472,
      "rewards/IngredientQuantityMatchReward/mean": 0.7153397917747497,
      "rewards/IngredientQuantityMatchReward/std": 0.3973944127559662,
      "rewards/TotalKcalExactMatchReward/mean": 0.775,
      "rewards/TotalKcalExactMatchReward/std": 0.41154061555862426,
      "step": 3375
    },
    {
      "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": 0.00625,
      "completions/max_length": 495.2,
      "completions/mean_length": 392.421875,
      "completions/min_length": 257.2,
      "epoch": 0.9401947148817803,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.660618245601654,
      "kl": 0.05038860326167196,
      "learning_rate": 9.748078273253136e-09,
      "loss": 0.002015620470046997,
      "reward": 3.008139133453369,
      "reward_std": 0.4135745406150818,
      "rewards/IngredientFormatReward/mean": 0.9920312523841858,
      "rewards/IngredientFormatReward/std": 0.05728701539337635,
      "rewards/IngredientMatchReward/mean": 0.6056801795959472,
      "rewards/IngredientMatchReward/std": 0.3066266417503357,
      "rewards/IngredientQuantityMatchReward/mean": 0.624490213394165,
      "rewards/IngredientQuantityMatchReward/std": 0.42028595209121705,
      "rewards/TotalKcalExactMatchReward/mean": 0.7859375,
      "rewards/TotalKcalExactMatchReward/std": 0.4056543827056885,
      "step": 3380
    },
    {
      "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": 0.01875,
      "completions/max_length": 510.4,
      "completions/mean_length": 389.103125,
      "completions/min_length": 268.6,
      "epoch": 0.9415855354659249,
      "frac_reward_zero_std": 0.0125,
      "grad_norm": 0.7143226861953735,
      "kl": 0.05210619480349123,
      "learning_rate": 9.30134608199884e-09,
      "loss": 0.0020843889564275742,
      "reward": 3.1568170547485352,
      "reward_std": 0.439932119846344,
      "rewards/IngredientFormatReward/mean": 0.9773883819580078,
      "rewards/IngredientFormatReward/std": 0.12660152986645698,
      "rewards/IngredientMatchReward/mean": 0.6738641619682312,
      "rewards/IngredientMatchReward/std": 0.2843153178691864,
      "rewards/IngredientQuantityMatchReward/mean": 0.6915020108222961,
      "rewards/IngredientQuantityMatchReward/std": 0.40528308153152465,
      "rewards/TotalKcalExactMatchReward/mean": 0.8140625,
      "rewards/TotalKcalExactMatchReward/std": 0.3795959770679474,
      "step": 3385
    },
    {
      "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": 0.0109375,
      "completions/max_length": 506.6,
      "completions/mean_length": 391.2890625,
      "completions/min_length": 265.2,
      "epoch": 0.9429763560500696,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6121615767478943,
      "kl": 0.04772463580593467,
      "learning_rate": 8.864995685504251e-09,
      "loss": 0.0019091980531811715,
      "reward": 2.93808856010437,
      "reward_std": 0.39786078333854674,
      "rewards/IngredientFormatReward/mean": 0.9845312595367431,
      "rewards/IngredientFormatReward/std": 0.0779263861477375,
      "rewards/IngredientMatchReward/mean": 0.5765872240066529,
      "rewards/IngredientMatchReward/std": 0.29070484042167666,
      "rewards/IngredientQuantityMatchReward/mean": 0.5816575646400451,
      "rewards/IngredientQuantityMatchReward/std": 0.42740336060523987,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.39443774819374083,
      "step": 3390
    },
    {
      "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": 0.003125,
      "completions/max_length": 495.2,
      "completions/mean_length": 387.0484375,
      "completions/min_length": 239.2,
      "epoch": 0.9443671766342142,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.7277557849884033,
      "kl": 0.047656627488322556,
      "learning_rate": 8.439036315708691e-09,
      "loss": 0.001906503364443779,
      "reward": 3.1081774711608885,
      "reward_std": 0.3849943280220032,
      "rewards/IngredientFormatReward/mean": 0.9951562523841858,
      "rewards/IngredientFormatReward/std": 0.039984484761953355,
      "rewards/IngredientMatchReward/mean": 0.6685410618782044,
      "rewards/IngredientMatchReward/std": 0.2922262877225876,
      "rewards/IngredientQuantityMatchReward/mean": 0.6538552045822144,
      "rewards/IngredientQuantityMatchReward/std": 0.4168520152568817,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.3908423066139221,
      "step": 3395
    },
    {
      "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": 0.0109375,
      "completions/max_length": 509.4,
      "completions/mean_length": 391.184375,
      "completions/min_length": 278.2,
      "epoch": 0.9457579972183588,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6436322331428528,
      "kl": 0.04723382429219782,
      "learning_rate": 8.023476984706956e-09,
      "loss": 0.001889285072684288,
      "reward": 3.161758041381836,
      "reward_std": 0.3478512942790985,
      "rewards/IngredientFormatReward/mean": 0.9847854614257813,
      "rewards/IngredientFormatReward/std": 0.0928487703204155,
      "rewards/IngredientMatchReward/mean": 0.635439395904541,
      "rewards/IngredientMatchReward/std": 0.27520969808101653,
      "rewards/IngredientQuantityMatchReward/mean": 0.7024707198143005,
      "rewards/IngredientQuantityMatchReward/std": 0.3910308003425598,
      "rewards/TotalKcalExactMatchReward/mean": 0.8390625,
      "rewards/TotalKcalExactMatchReward/std": 0.3589689821004868,
      "step": 3400
    },
    {
      "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": 0.003125,
      "completions/max_length": 507.6,
      "completions/mean_length": 386.3875,
      "completions/min_length": 266.6,
      "epoch": 0.9471488178025035,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7100093364715576,
      "kl": 0.049441801151260734,
      "learning_rate": 7.618326484558401e-09,
      "loss": 0.001977815106511116,
      "reward": 3.023116683959961,
      "reward_std": 0.3968947231769562,
      "rewards/IngredientFormatReward/mean": 0.9973437547683716,
      "rewards/IngredientFormatReward/std": 0.023174207657575607,
      "rewards/IngredientMatchReward/mean": 0.6149442076683045,
      "rewards/IngredientMatchReward/std": 0.2917324811220169,
      "rewards/IngredientQuantityMatchReward/mean": 0.6452036738395691,
      "rewards/IngredientQuantityMatchReward/std": 0.41264580488204955,
      "rewards/TotalKcalExactMatchReward/mean": 0.765625,
      "rewards/TotalKcalExactMatchReward/std": 0.416858971118927,
      "step": 3405
    },
    {
      "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": 0.003125,
      "completions/max_length": 497.8,
      "completions/mean_length": 390.4578125,
      "completions/min_length": 275.0,
      "epoch": 0.9485396383866481,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6251751780509949,
      "kl": 0.046490383613854644,
      "learning_rate": 7.2235933871008795e-09,
      "loss": 0.0018598001450300216,
      "reward": 3.1199548721313475,
      "reward_std": 0.355280739068985,
      "rewards/IngredientFormatReward/mean": 0.994527530670166,
      "rewards/IngredientFormatReward/std": 0.040901456214487555,
      "rewards/IngredientMatchReward/mean": 0.6296986818313599,
      "rewards/IngredientMatchReward/std": 0.2927346646785736,
      "rewards/IngredientQuantityMatchReward/mean": 0.6769786238670349,
      "rewards/IngredientQuantityMatchReward/std": 0.4171420454978943,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.3693612217903137,
      "step": 3410
    },
    {
      "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": 0.0125,
      "completions/max_length": 508.6,
      "completions/mean_length": 392.115625,
      "completions/min_length": 272.2,
      "epoch": 0.9499304589707928,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6708676815032959,
      "kl": 0.047037134994752705,
      "learning_rate": 6.839286043769654e-09,
      "loss": 0.0018814524635672569,
      "reward": 3.0176434993743895,
      "reward_std": 0.4097278594970703,
      "rewards/IngredientFormatReward/mean": 0.9863913774490356,
      "rewards/IngredientFormatReward/std": 0.07142920605838299,
      "rewards/IngredientMatchReward/mean": 0.5660413384437561,
      "rewards/IngredientMatchReward/std": 0.2902086913585663,
      "rewards/IngredientQuantityMatchReward/mean": 0.6245858073234558,
      "rewards/IngredientQuantityMatchReward/std": 0.429430627822876,
      "rewards/TotalKcalExactMatchReward/mean": 0.840625,
      "rewards/TotalKcalExactMatchReward/std": 0.3561762750148773,
      "step": 3415
    },
    {
      "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": 0.009375,
      "completions/max_length": 495.2,
      "completions/mean_length": 387.44375,
      "completions/min_length": 258.8,
      "epoch": 0.9513212795549374,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6568620204925537,
      "kl": 0.04582350081764162,
      "learning_rate": 6.4654125854204374e-09,
      "loss": 0.0018331578001379966,
      "reward": 3.1052068710327148,
      "reward_std": 0.4515254557132721,
      "rewards/IngredientFormatReward/mean": 0.9904017925262452,
      "rewards/IngredientFormatReward/std": 0.05673844218254089,
      "rewards/IngredientMatchReward/mean": 0.6195003747940063,
      "rewards/IngredientMatchReward/std": 0.27675817608833314,
      "rewards/IngredientQuantityMatchReward/mean": 0.7015547275543212,
      "rewards/IngredientQuantityMatchReward/std": 0.40126625299453733,
      "rewards/TotalKcalExactMatchReward/mean": 0.79375,
      "rewards/TotalKcalExactMatchReward/std": 0.38862606287002566,
      "step": 3420
    },
    {
      "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": 0.00625,
      "completions/max_length": 510.2,
      "completions/mean_length": 387.96875,
      "completions/min_length": 266.2,
      "epoch": 0.952712100139082,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6748784780502319,
      "kl": 0.0493265719152987,
      "learning_rate": 6.101980922157524e-09,
      "loss": 0.001973242685198784,
      "reward": 3.1050642013549803,
      "reward_std": 0.38432427048683165,
      "rewards/IngredientFormatReward/mean": 0.99296875,
      "rewards/IngredientFormatReward/std": 0.06092705130577088,
      "rewards/IngredientMatchReward/mean": 0.6292460322380066,
      "rewards/IngredientMatchReward/std": 0.28708921670913695,
      "rewards/IngredientQuantityMatchReward/mean": 0.6562868833541871,
      "rewards/IngredientQuantityMatchReward/std": 0.4298116385936737,
      "rewards/TotalKcalExactMatchReward/mean": 0.8265625,
      "rewards/TotalKcalExactMatchReward/std": 0.363370805978775,
      "step": 3425
    },
    {
      "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": 0.00625,
      "completions/max_length": 508.2,
      "completions/mean_length": 393.5328125,
      "completions/min_length": 271.0,
      "epoch": 0.9541029207232267,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.663008987903595,
      "kl": 0.048660345608368516,
      "learning_rate": 5.748998743166256e-09,
      "loss": 0.0019468117505311966,
      "reward": 3.1665676116943358,
      "reward_std": 0.3740094184875488,
      "rewards/IngredientFormatReward/mean": 0.9934895753860473,
      "rewards/IngredientFormatReward/std": 0.052749037928879264,
      "rewards/IngredientMatchReward/mean": 0.6485987067222595,
      "rewards/IngredientMatchReward/std": 0.2870753645896912,
      "rewards/IngredientQuantityMatchReward/mean": 0.7197918772697449,
      "rewards/IngredientQuantityMatchReward/std": 0.3927998900413513,
      "rewards/TotalKcalExactMatchReward/mean": 0.8046875,
      "rewards/TotalKcalExactMatchReward/std": 0.39596582651138307,
      "step": 3430
    },
    {
      "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": 0.0046875,
      "completions/max_length": 503.2,
      "completions/mean_length": 389.634375,
      "completions/min_length": 251.6,
      "epoch": 0.9554937413073713,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.65153968334198,
      "kl": 0.04435617581475526,
      "learning_rate": 5.406473516550603e-09,
      "loss": 0.001774563454091549,
      "reward": 3.1701682090759276,
      "reward_std": 0.3287068665027618,
      "rewards/IngredientFormatReward/mean": 0.9945163607597352,
      "rewards/IngredientFormatReward/std": 0.0458319054916501,
      "rewards/IngredientMatchReward/mean": 0.6381113409996033,
      "rewards/IngredientMatchReward/std": 0.29567062854766846,
      "rewards/IngredientQuantityMatchReward/mean": 0.6484779477119446,
      "rewards/IngredientQuantityMatchReward/std": 0.4112301588058472,
      "rewards/TotalKcalExactMatchReward/mean": 0.8890625,
      "rewards/TotalKcalExactMatchReward/std": 0.2916980266571045,
      "step": 3435
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.8,
      "completions/mean_length": 395.403125,
      "completions/min_length": 261.4,
      "epoch": 0.9568845618915159,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6878206133842468,
      "kl": 0.04364196332171559,
      "learning_rate": 5.074412489174895e-09,
      "loss": 0.0017459928989410401,
      "reward": 3.17191801071167,
      "reward_std": 0.3969049096107483,
      "rewards/IngredientFormatReward/mean": 0.9871874809265136,
      "rewards/IngredientFormatReward/std": 0.093497334420681,
      "rewards/IngredientMatchReward/mean": 0.6249485611915588,
      "rewards/IngredientMatchReward/std": 0.2994285225868225,
      "rewards/IngredientQuantityMatchReward/mean": 0.7285319924354553,
      "rewards/IngredientQuantityMatchReward/std": 0.3931278049945831,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.37189761996269227,
      "step": 3440
    },
    {
      "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": 0.0078125,
      "completions/max_length": 503.8,
      "completions/mean_length": 390.659375,
      "completions/min_length": 263.8,
      "epoch": 0.9582753824756607,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6432104706764221,
      "kl": 0.04477330117952079,
      "learning_rate": 4.752822686510727e-09,
      "loss": 0.0017909301444888116,
      "reward": 3.060155248641968,
      "reward_std": 0.3750399827957153,
      "rewards/IngredientFormatReward/mean": 0.9862499952316284,
      "rewards/IngredientFormatReward/std": 0.07456732466816902,
      "rewards/IngredientMatchReward/mean": 0.6478366851806641,
      "rewards/IngredientMatchReward/std": 0.2967373102903366,
      "rewards/IngredientQuantityMatchReward/mean": 0.6291934847831726,
      "rewards/IngredientQuantityMatchReward/std": 0.42095019221305846,
      "rewards/TotalKcalExactMatchReward/mean": 0.796875,
      "rewards/TotalKcalExactMatchReward/std": 0.39375485181808473,
      "step": 3445
    },
    {
      "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": 0.009375,
      "completions/max_length": 508.4,
      "completions/mean_length": 397.0875,
      "completions/min_length": 242.2,
      "epoch": 0.9596662030598053,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6475841403007507,
      "kl": 0.04543045272585004,
      "learning_rate": 4.441710912488239e-09,
      "loss": 0.0018170112743973732,
      "reward": 3.0488508701324464,
      "reward_std": 0.37055699825286864,
      "rewards/IngredientFormatReward/mean": 0.9900520801544189,
      "rewards/IngredientFormatReward/std": 0.08570180833339691,
      "rewards/IngredientMatchReward/mean": 0.6155338644981384,
      "rewards/IngredientMatchReward/std": 0.3055759608745575,
      "rewards/IngredientQuantityMatchReward/mean": 0.6213900089263916,
      "rewards/IngredientQuantityMatchReward/std": 0.4131564795970917,
      "rewards/TotalKcalExactMatchReward/mean": 0.821875,
      "rewards/TotalKcalExactMatchReward/std": 0.3545024037361145,
      "step": 3450
    },
    {
      "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": 0.0125,
      "completions/max_length": 504.2,
      "completions/mean_length": 386.834375,
      "completions/min_length": 243.4,
      "epoch": 0.96105702364395,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6767144799232483,
      "kl": 0.04825890299398452,
      "learning_rate": 4.141083749351959e-09,
      "loss": 0.0019302651286125182,
      "reward": 3.10237512588501,
      "reward_std": 0.41013769507408143,
      "rewards/IngredientFormatReward/mean": 0.9851934432983398,
      "rewards/IngredientFormatReward/std": 0.10111989080905914,
      "rewards/IngredientMatchReward/mean": 0.6478317260742188,
      "rewards/IngredientMatchReward/std": 0.2991852521896362,
      "rewards/IngredientQuantityMatchReward/mean": 0.6787248849868774,
      "rewards/IngredientQuantityMatchReward/std": 0.40713335275650026,
      "rewards/TotalKcalExactMatchReward/mean": 0.790625,
      "rewards/TotalKcalExactMatchReward/std": 0.398161518573761,
      "step": 3455
    },
    {
      "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": 0.021875,
      "completions/max_length": 507.8,
      "completions/mean_length": 402.75,
      "completions/min_length": 274.4,
      "epoch": 0.9624478442280946,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6345236897468567,
      "kl": 0.046873748395591976,
      "learning_rate": 3.8509475575219105e-09,
      "loss": 0.0018753422424197196,
      "reward": 3.0807044982910154,
      "reward_std": 0.426052987575531,
      "rewards/IngredientFormatReward/mean": 0.9768452405929565,
      "rewards/IngredientFormatReward/std": 0.11105858944356442,
      "rewards/IngredientMatchReward/mean": 0.628242838382721,
      "rewards/IngredientMatchReward/std": 0.3001261740922928,
      "rewards/IngredientQuantityMatchReward/mean": 0.6553040742874146,
      "rewards/IngredientQuantityMatchReward/std": 0.4021215379238129,
      "rewards/TotalKcalExactMatchReward/mean": 0.8203125,
      "rewards/TotalKcalExactMatchReward/std": 0.371417498588562,
      "step": 3460
    },
    {
      "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": 0.0078125,
      "completions/max_length": 505.0,
      "completions/mean_length": 388.6921875,
      "completions/min_length": 260.0,
      "epoch": 0.9638386648122392,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.7613881826400757,
      "kl": 0.04958433983847499,
      "learning_rate": 3.571308475458723e-09,
      "loss": 0.0019834831357002257,
      "reward": 3.1493412971496584,
      "reward_std": 0.4270430862903595,
      "rewards/IngredientFormatReward/mean": 0.9888020753860474,
      "rewards/IngredientFormatReward/std": 0.0908542349934578,
      "rewards/IngredientMatchReward/mean": 0.6713510751724243,
      "rewards/IngredientMatchReward/std": 0.3045841991901398,
      "rewards/IngredientQuantityMatchReward/mean": 0.6938755869865417,
      "rewards/IngredientQuantityMatchReward/std": 0.4049495041370392,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.39825695753097534,
      "step": 3465
    },
    {
      "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": 0.00625,
      "completions/max_length": 503.2,
      "completions/mean_length": 387.459375,
      "completions/min_length": 250.8,
      "epoch": 0.9652294853963839,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.6586174964904785,
      "kl": 0.04571211233269423,
      "learning_rate": 3.3021724195340107e-09,
      "loss": 0.001828974112868309,
      "reward": 3.149783134460449,
      "reward_std": 0.4005468308925629,
      "rewards/IngredientFormatReward/mean": 0.9925000071525574,
      "rewards/IngredientFormatReward/std": 0.052525454014539716,
      "rewards/IngredientMatchReward/mean": 0.6844698548316955,
      "rewards/IngredientMatchReward/std": 0.2855565369129181,
      "rewards/IngredientQuantityMatchReward/mean": 0.6696882605552673,
      "rewards/IngredientQuantityMatchReward/std": 0.4112094700336456,
      "rewards/TotalKcalExactMatchReward/mean": 0.803125,
      "rewards/TotalKcalExactMatchReward/std": 0.3755046874284744,
      "step": 3470
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.8,
      "completions/mean_length": 388.734375,
      "completions/min_length": 249.2,
      "epoch": 0.9666203059805285,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6234393119812012,
      "kl": 0.049782629730179905,
      "learning_rate": 3.0435450839049194e-09,
      "loss": 0.001991119608283043,
      "reward": 3.1384727478027346,
      "reward_std": 0.34530335664749146,
      "rewards/IngredientFormatReward/mean": 0.9880580425262451,
      "rewards/IngredientFormatReward/std": 0.07789933532476426,
      "rewards/IngredientMatchReward/mean": 0.6632006525993347,
      "rewards/IngredientMatchReward/std": 0.27832908630371095,
      "rewards/IngredientQuantityMatchReward/mean": 0.637213945388794,
      "rewards/IngredientQuantityMatchReward/std": 0.41195704936981203,
      "rewards/TotalKcalExactMatchReward/mean": 0.85,
      "rewards/TotalKcalExactMatchReward/std": 0.32829630076885224,
      "step": 3475
    },
    {
      "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": 0.003125,
      "completions/max_length": 505.0,
      "completions/mean_length": 385.3453125,
      "completions/min_length": 241.8,
      "epoch": 0.9680111265646731,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6432631015777588,
      "kl": 0.747702275426127,
      "learning_rate": 2.7954319403940553e-09,
      "loss": 0.029897397756576537,
      "reward": 3.144414186477661,
      "reward_std": 0.40637815594673155,
      "rewards/IngredientFormatReward/mean": 0.9936718821525574,
      "rewards/IngredientFormatReward/std": 0.05400313213467598,
      "rewards/IngredientMatchReward/mean": 0.6715754985809326,
      "rewards/IngredientMatchReward/std": 0.28783435225486753,
      "rewards/IngredientQuantityMatchReward/mean": 0.708854329586029,
      "rewards/IngredientQuantityMatchReward/std": 0.3918882191181183,
      "rewards/TotalKcalExactMatchReward/mean": 0.7703125,
      "rewards/TotalKcalExactMatchReward/std": 0.40544432401657104,
      "step": 3480
    },
    {
      "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": 0.0125,
      "completions/max_length": 503.6,
      "completions/mean_length": 381.9390625,
      "completions/min_length": 245.4,
      "epoch": 0.9694019471488178,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6335321664810181,
      "kl": 0.04744428307749331,
      "learning_rate": 2.5578382383732442e-09,
      "loss": 0.0018980761989951134,
      "reward": 3.157121992111206,
      "reward_std": 0.3908619523048401,
      "rewards/IngredientFormatReward/mean": 0.9844103455543518,
      "rewards/IngredientFormatReward/std": 0.08667321354150773,
      "rewards/IngredientMatchReward/mean": 0.6567723035812378,
      "rewards/IngredientMatchReward/std": 0.3101677715778351,
      "rewards/IngredientQuantityMatchReward/mean": 0.6675018310546875,
      "rewards/IngredientQuantityMatchReward/std": 0.43211646676063536,
      "rewards/TotalKcalExactMatchReward/mean": 0.8484375,
      "rewards/TotalKcalExactMatchReward/std": 0.35224978923797606,
      "step": 3485
    },
    {
      "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": 0.0046875,
      "completions/max_length": 508.8,
      "completions/mean_length": 389.928125,
      "completions/min_length": 236.0,
      "epoch": 0.9707927677329624,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6075333952903748,
      "kl": 0.04749975092709065,
      "learning_rate": 2.3307690046527883e-09,
      "loss": 0.00190051831305027,
      "reward": 3.052085304260254,
      "reward_std": 0.3895645797252655,
      "rewards/IngredientFormatReward/mean": 0.9751562595367431,
      "rewards/IngredientFormatReward/std": 0.12122566476464272,
      "rewards/IngredientMatchReward/mean": 0.6185956478118897,
      "rewards/IngredientMatchReward/std": 0.26451897621154785,
      "rewards/IngredientQuantityMatchReward/mean": 0.6395834565162659,
      "rewards/IngredientQuantityMatchReward/std": 0.4328662157058716,
      "rewards/TotalKcalExactMatchReward/mean": 0.81875,
      "rewards/TotalKcalExactMatchReward/std": 0.37133659422397614,
      "step": 3490
    },
    {
      "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": 0.0078125,
      "completions/max_length": 507.6,
      "completions/mean_length": 388.0703125,
      "completions/min_length": 274.4,
      "epoch": 0.972183588317107,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.6728726625442505,
      "kl": 0.047019163495860994,
      "learning_rate": 2.1142290433750486e-09,
      "loss": 0.0018808633089065553,
      "reward": 3.1073230266571046,
      "reward_std": 0.367638099193573,
      "rewards/IngredientFormatReward/mean": 0.9906510472297668,
      "rewards/IngredientFormatReward/std": 0.07325609177350997,
      "rewards/IngredientMatchReward/mean": 0.6417634010314941,
      "rewards/IngredientMatchReward/std": 0.28443962931632993,
      "rewards/IngredientQuantityMatchReward/mean": 0.6686586856842041,
      "rewards/IngredientQuantityMatchReward/std": 0.41565218567848206,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3815740138292313,
      "step": 3495
    },
    {
      "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": 0.0125,
      "completions/max_length": 502.0,
      "completions/mean_length": 383.71875,
      "completions/min_length": 259.0,
      "epoch": 0.9735744089012517,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7703426480293274,
      "kl": 0.04528216728940606,
      "learning_rate": 1.908222935912751e-09,
      "loss": 0.001811334304511547,
      "reward": 3.0197416305541993,
      "reward_std": 0.4097736120223999,
      "rewards/IngredientFormatReward/mean": 0.9861718654632569,
      "rewards/IngredientFormatReward/std": 0.06792713664472103,
      "rewards/IngredientMatchReward/mean": 0.6406938433647156,
      "rewards/IngredientMatchReward/std": 0.2943361341953278,
      "rewards/IngredientQuantityMatchReward/mean": 0.57568838596344,
      "rewards/IngredientQuantityMatchReward/std": 0.4376201629638672,
      "rewards/TotalKcalExactMatchReward/mean": 0.8171875,
      "rewards/TotalKcalExactMatchReward/std": 0.3830027997493744,
      "step": 3500
    },
    {
      "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": 0.0109375,
      "completions/max_length": 501.6,
      "completions/mean_length": 388.3625,
      "completions/min_length": 245.2,
      "epoch": 0.9749652294853964,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6825488805770874,
      "kl": 0.052455898956395686,
      "learning_rate": 1.7127550407721181e-09,
      "loss": 0.0020987868309020998,
      "reward": 3.1052551746368406,
      "reward_std": 0.421875661611557,
      "rewards/IngredientFormatReward/mean": 0.9838802099227906,
      "rewards/IngredientFormatReward/std": 0.10591072142124176,
      "rewards/IngredientMatchReward/mean": 0.6220045924186707,
      "rewards/IngredientMatchReward/std": 0.2856017410755157,
      "rewards/IngredientQuantityMatchReward/mean": 0.6649953603744507,
      "rewards/IngredientQuantityMatchReward/std": 0.4061083495616913,
      "rewards/TotalKcalExactMatchReward/mean": 0.834375,
      "rewards/TotalKcalExactMatchReward/std": 0.3393280774354935,
      "step": 3505
    },
    {
      "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": 0.003125,
      "completions/max_length": 502.8,
      "completions/mean_length": 390.6109375,
      "completions/min_length": 269.4,
      "epoch": 0.9763560500695411,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6667228937149048,
      "kl": 0.04870023645926267,
      "learning_rate": 1.5278294935006098e-09,
      "loss": 0.0019479893147945405,
      "reward": 3.113843631744385,
      "reward_std": 0.35846092104911803,
      "rewards/IngredientFormatReward/mean": 0.9938801884651184,
      "rewards/IngredientFormatReward/std": 0.05015102569013834,
      "rewards/IngredientMatchReward/mean": 0.6636923670768737,
      "rewards/IngredientMatchReward/std": 0.27049159705638887,
      "rewards/IngredientQuantityMatchReward/mean": 0.6203335046768188,
      "rewards/IngredientQuantityMatchReward/std": 0.41901023387908937,
      "rewards/TotalKcalExactMatchReward/mean": 0.8359375,
      "rewards/TotalKcalExactMatchReward/std": 0.3319972887635231,
      "step": 3510
    },
    {
      "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": 0.0109375,
      "completions/max_length": 512.4,
      "completions/mean_length": 389.965625,
      "completions/min_length": 265.8,
      "epoch": 0.9777468706536857,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.7143624424934387,
      "kl": 0.046095013013109566,
      "learning_rate": 1.3534502065993826e-09,
      "loss": 0.001844063401222229,
      "reward": 3.1194585800170898,
      "reward_std": 0.45072959661483764,
      "rewards/IngredientFormatReward/mean": 0.987109375,
      "rewards/IngredientFormatReward/std": 0.09570224285125732,
      "rewards/IngredientMatchReward/mean": 0.6516096472740174,
      "rewards/IngredientMatchReward/std": 0.29113835096359253,
      "rewards/IngredientQuantityMatchReward/mean": 0.6744894981384277,
      "rewards/IngredientQuantityMatchReward/std": 0.40634567141532896,
      "rewards/TotalKcalExactMatchReward/mean": 0.80625,
      "rewards/TotalKcalExactMatchReward/std": 0.3906187415122986,
      "step": 3515
    },
    {
      "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": 0.0046875,
      "completions/max_length": 493.0,
      "completions/mean_length": 382.9546875,
      "completions/min_length": 244.8,
      "epoch": 0.9791376912378303,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.683193027973175,
      "kl": 0.050947370985522863,
      "learning_rate": 1.1896208694406884e-09,
      "loss": 0.002037731558084488,
      "reward": 3.0674956798553468,
      "reward_std": 0.45058631896972656,
      "rewards/IngredientFormatReward/mean": 0.9909598112106324,
      "rewards/IngredientFormatReward/std": 0.08356152959167958,
      "rewards/IngredientMatchReward/mean": 0.6089496612548828,
      "rewards/IngredientMatchReward/std": 0.29325642585754397,
      "rewards/IngredientQuantityMatchReward/mean": 0.6363362312316895,
      "rewards/IngredientQuantityMatchReward/std": 0.4325535476207733,
      "rewards/TotalKcalExactMatchReward/mean": 0.83125,
      "rewards/TotalKcalExactMatchReward/std": 0.3584339439868927,
      "step": 3520
    },
    {
      "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": 0.0078125,
      "completions/max_length": 504.4,
      "completions/mean_length": 396.371875,
      "completions/min_length": 274.4,
      "epoch": 0.980528511821975,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6610113382339478,
      "kl": 0.04606618359684944,
      "learning_rate": 1.0363449481896603e-09,
      "loss": 0.0018428176641464233,
      "reward": 3.0702935695648192,
      "reward_std": 0.3585042953491211,
      "rewards/IngredientFormatReward/mean": 0.9895312428474426,
      "rewards/IngredientFormatReward/std": 0.0739785360172391,
      "rewards/IngredientMatchReward/mean": 0.647214138507843,
      "rewards/IngredientMatchReward/std": 0.2919449031352997,
      "rewards/IngredientQuantityMatchReward/mean": 0.650735592842102,
      "rewards/IngredientQuantityMatchReward/std": 0.38301952481269835,
      "rewards/TotalKcalExactMatchReward/mean": 0.7828125,
      "rewards/TotalKcalExactMatchReward/std": 0.4121102452278137,
      "step": 3525
    },
    {
      "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": 0.0046875,
      "completions/max_length": 494.4,
      "completions/mean_length": 381.7484375,
      "completions/min_length": 251.4,
      "epoch": 0.9819193324061196,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6419509053230286,
      "kl": 0.04725923398509622,
      "learning_rate": 8.9362568573087e-10,
      "loss": 0.0018903814256191255,
      "reward": 3.1441524028778076,
      "reward_std": 0.4015710175037384,
      "rewards/IngredientFormatReward/mean": 0.993958342075348,
      "rewards/IngredientFormatReward/std": 0.04699918553233147,
      "rewards/IngredientMatchReward/mean": 0.6546136975288391,
      "rewards/IngredientMatchReward/std": 0.29563533067703246,
      "rewards/IngredientQuantityMatchReward/mean": 0.7018303513526917,
      "rewards/IngredientQuantityMatchReward/std": 0.3950989067554474,
      "rewards/TotalKcalExactMatchReward/mean": 0.79375,
      "rewards/TotalKcalExactMatchReward/std": 0.39758574962615967,
      "step": 3530
    },
    {
      "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": 0.009375,
      "completions/max_length": 511.4,
      "completions/mean_length": 390.678125,
      "completions/min_length": 256.0,
      "epoch": 0.9833101529902643,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 0.5843694806098938,
      "kl": 0.05070257929619402,
      "learning_rate": 7.614661016001056e-10,
      "loss": 0.002028765343129635,
      "reward": 3.0022071838378905,
      "reward_std": 0.4141098916530609,
      "rewards/IngredientFormatReward/mean": 0.9861049175262451,
      "rewards/IngredientFormatReward/std": 0.11176733374595642,
      "rewards/IngredientMatchReward/mean": 0.6246558666229248,
      "rewards/IngredientMatchReward/std": 0.3036438703536987,
      "rewards/IngredientQuantityMatchReward/mean": 0.6586337327957154,
      "rewards/IngredientQuantityMatchReward/std": 0.41347108483314515,
      "rewards/TotalKcalExactMatchReward/mean": 0.7328125,
      "rewards/TotalKcalExactMatchReward/std": 0.4390223383903503,
      "step": 3535
    },
    {
      "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": 0.0171875,
      "completions/max_length": 508.6,
      "completions/mean_length": 393.465625,
      "completions/min_length": 254.4,
      "epoch": 0.9847009735744089,
      "frac_reward_zero_std": 0.1375,
      "grad_norm": 0.808208703994751,
      "kl": 0.05155458953231573,
      "learning_rate": 6.398689919201449e-10,
      "loss": 0.0020628392696380613,
      "reward": 3.1207000732421877,
      "reward_std": 0.3916473984718323,
      "rewards/IngredientFormatReward/mean": 0.9782217264175415,
      "rewards/IngredientFormatReward/std": 0.1250761903822422,
      "rewards/IngredientMatchReward/mean": 0.6547699689865112,
      "rewards/IngredientMatchReward/std": 0.2909015595912933,
      "rewards/IngredientQuantityMatchReward/mean": 0.7017708659172058,
      "rewards/IngredientQuantityMatchReward/std": 0.3939378708600998,
      "rewards/TotalKcalExactMatchReward/mean": 0.7859375,
      "rewards/TotalKcalExactMatchReward/std": 0.40576318502426145,
      "step": 3540
    },
    {
      "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": 0.0109375,
      "completions/max_length": 511.6,
      "completions/mean_length": 389.2640625,
      "completions/min_length": 263.8,
      "epoch": 0.9860917941585535,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6409344673156738,
      "kl": 0.04587732988875359,
      "learning_rate": 5.288369293415807e-10,
      "loss": 0.0018349820747971535,
      "reward": 3.138898420333862,
      "reward_std": 0.4375850737094879,
      "rewards/IngredientFormatReward/mean": 0.987485122680664,
      "rewards/IngredientFormatReward/std": 0.09236232042312623,
      "rewards/IngredientMatchReward/mean": 0.6686526775360108,
      "rewards/IngredientMatchReward/std": 0.2893941581249237,
      "rewards/IngredientQuantityMatchReward/mean": 0.6874481797218323,
      "rewards/IngredientQuantityMatchReward/std": 0.40284740924835205,
      "rewards/TotalKcalExactMatchReward/mean": 0.7953125,
      "rewards/TotalKcalExactMatchReward/std": 0.39978498220443726,
      "step": 3545
    },
    {
      "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": 0.015625,
      "completions/max_length": 507.4,
      "completions/mean_length": 391.2296875,
      "completions/min_length": 273.2,
      "epoch": 0.9874826147426982,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.6869775652885437,
      "kl": 0.048427975224331024,
      "learning_rate": 4.2837226298875206e-10,
      "loss": 0.0019370142370462417,
      "reward": 3.165076398849487,
      "reward_std": 0.41075287461280824,
      "rewards/IngredientFormatReward/mean": 0.9800595283508301,
      "rewards/IngredientFormatReward/std": 0.12871189117431642,
      "rewards/IngredientMatchReward/mean": 0.657729423046112,
      "rewards/IngredientMatchReward/std": 0.2861269056797028,
      "rewards/IngredientQuantityMatchReward/mean": 0.7179124355316162,
      "rewards/IngredientQuantityMatchReward/std": 0.3858673632144928,
      "rewards/TotalKcalExactMatchReward/mean": 0.809375,
      "rewards/TotalKcalExactMatchReward/std": 0.3873170852661133,
      "step": 3550
    },
    {
      "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": 0.0015625,
      "completions/max_length": 495.4,
      "completions/mean_length": 382.4734375,
      "completions/min_length": 269.8,
      "epoch": 0.9888734353268428,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6728901863098145,
      "kl": 0.04554827213287353,
      "learning_rate": 3.384771184095081e-10,
      "loss": 0.0018219701945781709,
      "reward": 3.1922850608825684,
      "reward_std": 0.38457239866256715,
      "rewards/IngredientFormatReward/mean": 0.996875,
      "rewards/IngredientFormatReward/std": 0.03535533845424652,
      "rewards/IngredientMatchReward/mean": 0.7049667954444885,
      "rewards/IngredientMatchReward/std": 0.2878182172775269,
      "rewards/IngredientQuantityMatchReward/mean": 0.652943241596222,
      "rewards/IngredientQuantityMatchReward/std": 0.4103390693664551,
      "rewards/TotalKcalExactMatchReward/mean": 0.8375,
      "rewards/TotalKcalExactMatchReward/std": 0.36109898686409,
      "step": 3555
    },
    {
      "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": 0.0109375,
      "completions/max_length": 506.8,
      "completions/mean_length": 390.6703125,
      "completions/min_length": 256.2,
      "epoch": 0.9902642559109874,
      "frac_reward_zero_std": 0.075,
      "grad_norm": 7.2769775390625,
      "kl": 0.0716410128865391,
      "learning_rate": 2.5915339753085354e-10,
      "loss": 0.0028650620952248573,
      "reward": 3.1179696559906005,
      "reward_std": 0.3656330227851868,
      "rewards/IngredientFormatReward/mean": 0.9906473159790039,
      "rewards/IngredientFormatReward/std": 0.0729976112022996,
      "rewards/IngredientMatchReward/mean": 0.6654321789741516,
      "rewards/IngredientMatchReward/std": 0.30215807557106017,
      "rewards/IngredientQuantityMatchReward/mean": 0.6509526252746582,
      "rewards/IngredientQuantityMatchReward/std": 0.42949780225753786,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.38284428119659425,
      "step": 3560
    },
    {
      "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": 0.0140625,
      "completions/max_length": 508.8,
      "completions/mean_length": 384.6453125,
      "completions/min_length": 228.2,
      "epoch": 0.9916550764951322,
      "frac_reward_zero_std": 0.1,
      "grad_norm": 0.6099084615707397,
      "kl": 0.04533320358023048,
      "learning_rate": 1.9040277861814835e-10,
      "loss": 0.0018132233992218972,
      "reward": 3.1440773010253906,
      "reward_std": 0.3812756478786469,
      "rewards/IngredientFormatReward/mean": 0.9850111722946167,
      "rewards/IngredientFormatReward/std": 0.10009037107229232,
      "rewards/IngredientMatchReward/mean": 0.646268618106842,
      "rewards/IngredientMatchReward/std": 0.28203052580356597,
      "rewards/IngredientQuantityMatchReward/mean": 0.6581100583076477,
      "rewards/IngredientQuantityMatchReward/std": 0.39739391803741453,
      "rewards/TotalKcalExactMatchReward/mean": 0.8546875,
      "rewards/TotalKcalExactMatchReward/std": 0.34881080985069274,
      "step": 3565
    },
    {
      "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": 0.0125,
      "completions/max_length": 513.0,
      "completions/mean_length": 392.1140625,
      "completions/min_length": 230.4,
      "epoch": 0.9930458970792768,
      "frac_reward_zero_std": 0.1125,
      "grad_norm": 0.6250317096710205,
      "kl": 0.04590991784352809,
      "learning_rate": 1.3222671623991377e-10,
      "loss": 0.0018364254385232926,
      "reward": 3.1440375804901124,
      "reward_std": 0.4134489595890045,
      "rewards/IngredientFormatReward/mean": 0.9813243865966796,
      "rewards/IngredientFormatReward/std": 0.12313078790903091,
      "rewards/IngredientMatchReward/mean": 0.6408395528793335,
      "rewards/IngredientMatchReward/std": 0.29529817700386046,
      "rewards/IngredientQuantityMatchReward/mean": 0.6984362483024598,
      "rewards/IngredientQuantityMatchReward/std": 0.41224990487098695,
      "rewards/TotalKcalExactMatchReward/mean": 0.8234375,
      "rewards/TotalKcalExactMatchReward/std": 0.38107048273086547,
      "step": 3570
    },
    {
      "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": 0.00625,
      "completions/max_length": 503.4,
      "completions/mean_length": 392.796875,
      "completions/min_length": 257.0,
      "epoch": 0.9944367176634215,
      "frac_reward_zero_std": 0.0875,
      "grad_norm": 0.6811825633049011,
      "kl": 0.04669976490549743,
      "learning_rate": 8.462644123696794e-11,
      "loss": 0.001867879182100296,
      "reward": 3.0588041305541993,
      "reward_std": 0.4007099688053131,
      "rewards/IngredientFormatReward/mean": 0.992005217075348,
      "rewards/IngredientFormatReward/std": 0.06722570732235908,
      "rewards/IngredientMatchReward/mean": 0.6472081303596496,
      "rewards/IngredientMatchReward/std": 0.29389954805374147,
      "rewards/IngredientQuantityMatchReward/mean": 0.6477158069610596,
      "rewards/IngredientQuantityMatchReward/std": 0.4318304181098938,
      "rewards/TotalKcalExactMatchReward/mean": 0.771875,
      "rewards/TotalKcalExactMatchReward/std": 0.41408060789108275,
      "step": 3575
    },
    {
      "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": 0.0046875,
      "completions/max_length": 510.0,
      "completions/mean_length": 387.1984375,
      "completions/min_length": 256.8,
      "epoch": 0.9958275382475661,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6889552474021912,
      "kl": 0.04746832814998925,
      "learning_rate": 4.7602960696391246e-11,
      "loss": 0.001899052783846855,
      "reward": 3.145450496673584,
      "reward_std": 0.3926717579364777,
      "rewards/IngredientFormatReward/mean": 0.9950000047683716,
      "rewards/IngredientFormatReward/std": 0.05335577577352524,
      "rewards/IngredientMatchReward/mean": 0.6480660915374756,
      "rewards/IngredientMatchReward/std": 0.2908414363861084,
      "rewards/IngredientQuantityMatchReward/mean": 0.6492594242095947,
      "rewards/IngredientQuantityMatchReward/std": 0.4212400853633881,
      "rewards/TotalKcalExactMatchReward/mean": 0.853125,
      "rewards/TotalKcalExactMatchReward/std": 0.3392543256282806,
      "step": 3580
    },
    {
      "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": 0.009375,
      "completions/max_length": 509.2,
      "completions/mean_length": 387.4125,
      "completions/min_length": 249.2,
      "epoch": 0.9972183588317107,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 0.7055279612541199,
      "kl": 0.04739529467187822,
      "learning_rate": 2.1157057930321077e-11,
      "loss": 0.0018960034474730491,
      "reward": 3.0336164474487304,
      "reward_std": 0.4345869064331055,
      "rewards/IngredientFormatReward/mean": 0.9876562476158142,
      "rewards/IngredientFormatReward/std": 0.0809405880048871,
      "rewards/IngredientMatchReward/mean": 0.6347569465637207,
      "rewards/IngredientMatchReward/std": 0.2958140969276428,
      "rewards/IngredientQuantityMatchReward/mean": 0.6315157651901245,
      "rewards/IngredientQuantityMatchReward/std": 0.42673113346099856,
      "rewards/TotalKcalExactMatchReward/mean": 0.7796875,
      "rewards/TotalKcalExactMatchReward/std": 0.40893966555595396,
      "step": 3585
    },
    {
      "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": 0.0109375,
      "completions/max_length": 508.8,
      "completions/mean_length": 389.1203125,
      "completions/min_length": 255.8,
      "epoch": 0.9986091794158554,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6596475839614868,
      "kl": 0.05180376642383635,
      "learning_rate": 5.289292459187411e-12,
      "loss": 0.002072012796998024,
      "reward": 3.072105598449707,
      "reward_std": 0.37645214796066284,
      "rewards/IngredientFormatReward/mean": 0.9853515744209289,
      "rewards/IngredientFormatReward/std": 0.0902503028512001,
      "rewards/IngredientMatchReward/mean": 0.6567869186401367,
      "rewards/IngredientMatchReward/std": 0.28849694728851316,
      "rewards/IngredientQuantityMatchReward/mean": 0.6143420994281769,
      "rewards/IngredientQuantityMatchReward/std": 0.4065409183502197,
      "rewards/TotalKcalExactMatchReward/mean": 0.815625,
      "rewards/TotalKcalExactMatchReward/std": 0.37442034780979155,
      "step": 3590
    },
    {
      "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": 0.00625,
      "completions/max_length": 508.0,
      "completions/mean_length": 393.1953125,
      "completions/min_length": 273.2,
      "epoch": 1.0,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.6491743326187134,
      "kl": 0.046436529234051706,
      "learning_rate": 0.0,
      "loss": 0.0018576562404632568,
      "reward": 3.1264251708984374,
      "reward_std": 0.3975203990936279,
      "rewards/IngredientFormatReward/mean": 0.988683032989502,
      "rewards/IngredientFormatReward/std": 0.07184891402721405,
      "rewards/IngredientMatchReward/mean": 0.6666737198829651,
      "rewards/IngredientMatchReward/std": 0.2867902278900146,
      "rewards/IngredientQuantityMatchReward/mean": 0.6601309657096863,
      "rewards/IngredientQuantityMatchReward/std": 0.4129492461681366,
      "rewards/TotalKcalExactMatchReward/mean": 0.8109375,
      "rewards/TotalKcalExactMatchReward/std": 0.38998509049415586,
      "step": 3595
    }
  ],
  "logging_steps": 5,
  "max_steps": 3595,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}