{
  "best_metric": 0.46430692076683044,
  "best_model_checkpoint": "/home/datawork-iot-nos/Seatizen/models/multilabel/drone/drone-DinoVdeau-from-probs-large-2024_11_14-batch-size16_freeze_probs/checkpoint-36792",
  "epoch": 94.0,
  "eval_steps": 500,
  "global_step": 41172,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_explained_variance": 0.3296825587749481,
      "eval_kl_divergence": 0.9903448224067688,
      "eval_loss": 0.4933677613735199,
      "eval_mae": 0.129361093044281,
      "eval_rmse": 0.18249033391475677,
      "eval_runtime": 70.7501,
      "eval_samples_per_second": 33.272,
      "eval_steps_per_second": 2.092,
      "learning_rate": 0.001,
      "step": 438
    },
    {
      "epoch": 1.1415525114155252,
      "grad_norm": 0.5407673120498657,
      "learning_rate": 0.001,
      "loss": 0.5313,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_explained_variance": 0.37307724356651306,
      "eval_kl_divergence": 0.6847184896469116,
      "eval_loss": 0.47886165976524353,
      "eval_mae": 0.12624560296535492,
      "eval_rmse": 0.1715712547302246,
      "eval_runtime": 71.8777,
      "eval_samples_per_second": 32.75,
      "eval_steps_per_second": 2.059,
      "learning_rate": 0.001,
      "step": 876
    },
    {
      "epoch": 2.2831050228310503,
      "grad_norm": 0.43918001651763916,
      "learning_rate": 0.001,
      "loss": 0.4831,
      "step": 1000
    },
    {
      "epoch": 3.0,
      "eval_explained_variance": 0.38236939907073975,
      "eval_kl_divergence": 0.5498285293579102,
      "eval_loss": 0.4788369834423065,
      "eval_mae": 0.12710730731487274,
      "eval_rmse": 0.1709037572145462,
      "eval_runtime": 66.8689,
      "eval_samples_per_second": 35.203,
      "eval_steps_per_second": 2.213,
      "learning_rate": 0.001,
      "step": 1314
    },
    {
      "epoch": 3.4246575342465753,
      "grad_norm": 0.28072813153266907,
      "learning_rate": 0.001,
      "loss": 0.4773,
      "step": 1500
    },
    {
      "epoch": 4.0,
      "eval_explained_variance": 0.39789319038391113,
      "eval_kl_divergence": 0.31306192278862,
      "eval_loss": 0.47656363248825073,
      "eval_mae": 0.12784750759601593,
      "eval_rmse": 0.16946662962436676,
      "eval_runtime": 68.7002,
      "eval_samples_per_second": 34.265,
      "eval_steps_per_second": 2.154,
      "learning_rate": 0.001,
      "step": 1752
    },
    {
      "epoch": 4.566210045662101,
      "grad_norm": 0.23626942932605743,
      "learning_rate": 0.001,
      "loss": 0.476,
      "step": 2000
    },
    {
      "epoch": 5.0,
      "eval_explained_variance": 0.3970318138599396,
      "eval_kl_divergence": 0.40128740668296814,
      "eval_loss": 0.4765072166919708,
      "eval_mae": 0.12769053876399994,
      "eval_rmse": 0.1686621755361557,
      "eval_runtime": 69.4052,
      "eval_samples_per_second": 33.917,
      "eval_steps_per_second": 2.132,
      "learning_rate": 0.001,
      "step": 2190
    },
    {
      "epoch": 5.707762557077626,
      "grad_norm": 0.2325536459684372,
      "learning_rate": 0.001,
      "loss": 0.4746,
      "step": 2500
    },
    {
      "epoch": 6.0,
      "eval_explained_variance": 0.392393559217453,
      "eval_kl_divergence": 0.6369896531105042,
      "eval_loss": 0.47649845480918884,
      "eval_mae": 0.12429385632276535,
      "eval_rmse": 0.16892628371715546,
      "eval_runtime": 69.6508,
      "eval_samples_per_second": 33.797,
      "eval_steps_per_second": 2.125,
      "learning_rate": 0.001,
      "step": 2628
    },
    {
      "epoch": 6.8493150684931505,
      "grad_norm": 0.1943524181842804,
      "learning_rate": 0.001,
      "loss": 0.4738,
      "step": 3000
    },
    {
      "epoch": 7.0,
      "eval_explained_variance": 0.3910720646381378,
      "eval_kl_divergence": 0.43142321705818176,
      "eval_loss": 0.47629299759864807,
      "eval_mae": 0.1291760504245758,
      "eval_rmse": 0.16937647759914398,
      "eval_runtime": 68.3166,
      "eval_samples_per_second": 34.457,
      "eval_steps_per_second": 2.166,
      "learning_rate": 0.001,
      "step": 3066
    },
    {
      "epoch": 7.9908675799086755,
      "grad_norm": 0.161671444773674,
      "learning_rate": 0.001,
      "loss": 0.4727,
      "step": 3500
    },
    {
      "epoch": 8.0,
      "eval_explained_variance": 0.407635897397995,
      "eval_kl_divergence": 0.33790889382362366,
      "eval_loss": 0.4755041003227234,
      "eval_mae": 0.1266549676656723,
      "eval_rmse": 0.1681331843137741,
      "eval_runtime": 71.2607,
      "eval_samples_per_second": 33.034,
      "eval_steps_per_second": 2.077,
      "learning_rate": 0.001,
      "step": 3504
    },
    {
      "epoch": 9.0,
      "eval_explained_variance": 0.40718480944633484,
      "eval_kl_divergence": 0.4916018843650818,
      "eval_loss": 0.47342246770858765,
      "eval_mae": 0.1250395029783249,
      "eval_rmse": 0.16618509590625763,
      "eval_runtime": 68.5459,
      "eval_samples_per_second": 34.342,
      "eval_steps_per_second": 2.159,
      "learning_rate": 0.001,
      "step": 3942
    },
    {
      "epoch": 9.132420091324201,
      "grad_norm": 0.19357167184352875,
      "learning_rate": 0.001,
      "loss": 0.4715,
      "step": 4000
    },
    {
      "epoch": 10.0,
      "eval_explained_variance": 0.40620386600494385,
      "eval_kl_divergence": 0.33481186628341675,
      "eval_loss": 0.47546806931495667,
      "eval_mae": 0.1277414709329605,
      "eval_rmse": 0.167744979262352,
      "eval_runtime": 67.1085,
      "eval_samples_per_second": 35.078,
      "eval_steps_per_second": 2.205,
      "learning_rate": 0.001,
      "step": 4380
    },
    {
      "epoch": 10.273972602739725,
      "grad_norm": 0.17573006451129913,
      "learning_rate": 0.001,
      "loss": 0.4714,
      "step": 4500
    },
    {
      "epoch": 11.0,
      "eval_explained_variance": 0.41538166999816895,
      "eval_kl_divergence": 0.35237056016921997,
      "eval_loss": 0.4731104075908661,
      "eval_mae": 0.12549267709255219,
      "eval_rmse": 0.1659194976091385,
      "eval_runtime": 67.2129,
      "eval_samples_per_second": 35.023,
      "eval_steps_per_second": 2.202,
      "learning_rate": 0.001,
      "step": 4818
    },
    {
      "epoch": 11.415525114155251,
      "grad_norm": 0.13790665566921234,
      "learning_rate": 0.001,
      "loss": 0.4713,
      "step": 5000
    },
    {
      "epoch": 12.0,
      "eval_explained_variance": 0.41025033593177795,
      "eval_kl_divergence": 0.23825553059577942,
      "eval_loss": 0.47679492831230164,
      "eval_mae": 0.13060903549194336,
      "eval_rmse": 0.16900604963302612,
      "eval_runtime": 68.7453,
      "eval_samples_per_second": 34.242,
      "eval_steps_per_second": 2.153,
      "learning_rate": 0.001,
      "step": 5256
    },
    {
      "epoch": 12.557077625570777,
      "grad_norm": 0.1497349739074707,
      "learning_rate": 0.001,
      "loss": 0.4722,
      "step": 5500
    },
    {
      "epoch": 13.0,
      "eval_explained_variance": 0.40282678604125977,
      "eval_kl_divergence": 0.696779727935791,
      "eval_loss": 0.4736888110637665,
      "eval_mae": 0.12227921932935715,
      "eval_rmse": 0.16662709414958954,
      "eval_runtime": 66.9685,
      "eval_samples_per_second": 35.151,
      "eval_steps_per_second": 2.21,
      "learning_rate": 0.001,
      "step": 5694
    },
    {
      "epoch": 13.698630136986301,
      "grad_norm": 0.11125459522008896,
      "learning_rate": 0.001,
      "loss": 0.472,
      "step": 6000
    },
    {
      "epoch": 14.0,
      "eval_explained_variance": 0.4099460542201996,
      "eval_kl_divergence": 0.3982764780521393,
      "eval_loss": 0.4736703634262085,
      "eval_mae": 0.12543360888957977,
      "eval_rmse": 0.16584907472133636,
      "eval_runtime": 70.7133,
      "eval_samples_per_second": 33.289,
      "eval_steps_per_second": 2.093,
      "learning_rate": 0.001,
      "step": 6132
    },
    {
      "epoch": 14.840182648401827,
      "grad_norm": 0.13986219465732574,
      "learning_rate": 0.001,
      "loss": 0.4697,
      "step": 6500
    },
    {
      "epoch": 15.0,
      "eval_explained_variance": 0.40361204743385315,
      "eval_kl_divergence": 0.5619756579399109,
      "eval_loss": 0.4738818407058716,
      "eval_mae": 0.12480553239583969,
      "eval_rmse": 0.16644911468029022,
      "eval_runtime": 67.989,
      "eval_samples_per_second": 34.623,
      "eval_steps_per_second": 2.177,
      "learning_rate": 0.001,
      "step": 6570
    },
    {
      "epoch": 15.981735159817351,
      "grad_norm": 0.10489701479673386,
      "learning_rate": 0.001,
      "loss": 0.4721,
      "step": 7000
    },
    {
      "epoch": 16.0,
      "eval_explained_variance": 0.41589656472206116,
      "eval_kl_divergence": 0.6049214601516724,
      "eval_loss": 0.47202879190444946,
      "eval_mae": 0.12306753545999527,
      "eval_rmse": 0.1648208647966385,
      "eval_runtime": 68.6318,
      "eval_samples_per_second": 34.299,
      "eval_steps_per_second": 2.156,
      "learning_rate": 0.001,
      "step": 7008
    },
    {
      "epoch": 17.0,
      "eval_explained_variance": 0.41705062985420227,
      "eval_kl_divergence": 0.3072437644004822,
      "eval_loss": 0.47406336665153503,
      "eval_mae": 0.12645679712295532,
      "eval_rmse": 0.16636626422405243,
      "eval_runtime": 67.693,
      "eval_samples_per_second": 34.775,
      "eval_steps_per_second": 2.186,
      "learning_rate": 0.001,
      "step": 7446
    },
    {
      "epoch": 17.123287671232877,
      "grad_norm": 0.11510255187749863,
      "learning_rate": 0.001,
      "loss": 0.4709,
      "step": 7500
    },
    {
      "epoch": 18.0,
      "eval_explained_variance": 0.4238598346710205,
      "eval_kl_divergence": 0.3350007236003876,
      "eval_loss": 0.4738321006298065,
      "eval_mae": 0.12525837123394012,
      "eval_rmse": 0.1649676412343979,
      "eval_runtime": 68.1862,
      "eval_samples_per_second": 34.523,
      "eval_steps_per_second": 2.171,
      "learning_rate": 0.001,
      "step": 7884
    },
    {
      "epoch": 18.264840182648403,
      "grad_norm": 0.1282125860452652,
      "learning_rate": 0.001,
      "loss": 0.4711,
      "step": 8000
    },
    {
      "epoch": 19.0,
      "eval_explained_variance": 0.4161747395992279,
      "eval_kl_divergence": 0.2745565176010132,
      "eval_loss": 0.476326048374176,
      "eval_mae": 0.12823572754859924,
      "eval_rmse": 0.16715575754642487,
      "eval_runtime": 68.1658,
      "eval_samples_per_second": 34.533,
      "eval_steps_per_second": 2.171,
      "learning_rate": 0.001,
      "step": 8322
    },
    {
      "epoch": 19.40639269406393,
      "grad_norm": 0.15439559519290924,
      "learning_rate": 0.001,
      "loss": 0.4696,
      "step": 8500
    },
    {
      "epoch": 20.0,
      "eval_explained_variance": 0.40602585673332214,
      "eval_kl_divergence": 0.5658758878707886,
      "eval_loss": 0.4755523204803467,
      "eval_mae": 0.12445954233407974,
      "eval_rmse": 0.1669723242521286,
      "eval_runtime": 66.5326,
      "eval_samples_per_second": 35.381,
      "eval_steps_per_second": 2.224,
      "learning_rate": 0.001,
      "step": 8760
    },
    {
      "epoch": 20.54794520547945,
      "grad_norm": 0.1288071572780609,
      "learning_rate": 0.001,
      "loss": 0.4715,
      "step": 9000
    },
    {
      "epoch": 21.0,
      "eval_explained_variance": 0.40606603026390076,
      "eval_kl_divergence": 0.6153907179832458,
      "eval_loss": 0.47340598702430725,
      "eval_mae": 0.12303853780031204,
      "eval_rmse": 0.16621015965938568,
      "eval_runtime": 66.3328,
      "eval_samples_per_second": 35.488,
      "eval_steps_per_second": 2.231,
      "learning_rate": 0.001,
      "step": 9198
    },
    {
      "epoch": 21.689497716894977,
      "grad_norm": 0.13872142136096954,
      "learning_rate": 0.001,
      "loss": 0.4714,
      "step": 9500
    },
    {
      "epoch": 22.0,
      "eval_explained_variance": 0.402675986289978,
      "eval_kl_divergence": 0.7973779439926147,
      "eval_loss": 0.47443991899490356,
      "eval_mae": 0.12229780107736588,
      "eval_rmse": 0.16771160066127777,
      "eval_runtime": 65.4256,
      "eval_samples_per_second": 35.98,
      "eval_steps_per_second": 2.262,
      "learning_rate": 0.001,
      "step": 9636
    },
    {
      "epoch": 22.831050228310502,
      "grad_norm": 0.12462610751390457,
      "learning_rate": 0.0001,
      "loss": 0.4697,
      "step": 10000
    },
    {
      "epoch": 23.0,
      "eval_explained_variance": 0.43371883034706116,
      "eval_kl_divergence": 0.2306971251964569,
      "eval_loss": 0.47205689549446106,
      "eval_mae": 0.12522436678409576,
      "eval_rmse": 0.16393429040908813,
      "eval_runtime": 67.2945,
      "eval_samples_per_second": 34.981,
      "eval_steps_per_second": 2.199,
      "learning_rate": 0.0001,
      "step": 10074
    },
    {
      "epoch": 23.972602739726028,
      "grad_norm": 0.15716814994812012,
      "learning_rate": 0.0001,
      "loss": 0.4653,
      "step": 10500
    },
    {
      "epoch": 24.0,
      "eval_explained_variance": 0.43139490485191345,
      "eval_kl_divergence": 0.42188531160354614,
      "eval_loss": 0.4706146717071533,
      "eval_mae": 0.12165608257055283,
      "eval_rmse": 0.1630544662475586,
      "eval_runtime": 69.0079,
      "eval_samples_per_second": 34.112,
      "eval_steps_per_second": 2.145,
      "learning_rate": 0.0001,
      "step": 10512
    },
    {
      "epoch": 25.0,
      "eval_explained_variance": 0.4370768964290619,
      "eval_kl_divergence": 0.5242214798927307,
      "eval_loss": 0.46876564621925354,
      "eval_mae": 0.11949952691793442,
      "eval_rmse": 0.16123180091381073,
      "eval_runtime": 67.8342,
      "eval_samples_per_second": 34.702,
      "eval_steps_per_second": 2.182,
      "learning_rate": 0.0001,
      "step": 10950
    },
    {
      "epoch": 25.114155251141554,
      "grad_norm": 0.12071619927883148,
      "learning_rate": 0.0001,
      "loss": 0.4665,
      "step": 11000
    },
    {
      "epoch": 26.0,
      "eval_explained_variance": 0.43381467461586,
      "eval_kl_divergence": 0.6159161925315857,
      "eval_loss": 0.46925392746925354,
      "eval_mae": 0.11895165592432022,
      "eval_rmse": 0.16197769343852997,
      "eval_runtime": 66.3554,
      "eval_samples_per_second": 35.476,
      "eval_steps_per_second": 2.23,
      "learning_rate": 0.0001,
      "step": 11388
    },
    {
      "epoch": 26.255707762557076,
      "grad_norm": 0.13487227261066437,
      "learning_rate": 0.0001,
      "loss": 0.4638,
      "step": 11500
    },
    {
      "epoch": 27.0,
      "eval_explained_variance": 0.4415856897830963,
      "eval_kl_divergence": 0.4046337604522705,
      "eval_loss": 0.46849024295806885,
      "eval_mae": 0.12057051062583923,
      "eval_rmse": 0.16070334613323212,
      "eval_runtime": 67.413,
      "eval_samples_per_second": 34.919,
      "eval_steps_per_second": 2.195,
      "learning_rate": 0.0001,
      "step": 11826
    },
    {
      "epoch": 27.397260273972602,
      "grad_norm": 0.122219979763031,
      "learning_rate": 0.0001,
      "loss": 0.4647,
      "step": 12000
    },
    {
      "epoch": 28.0,
      "eval_explained_variance": 0.4443088471889496,
      "eval_kl_divergence": 0.28601282835006714,
      "eval_loss": 0.46939656138420105,
      "eval_mae": 0.12197214365005493,
      "eval_rmse": 0.16160771250724792,
      "eval_runtime": 68.24,
      "eval_samples_per_second": 34.496,
      "eval_steps_per_second": 2.169,
      "learning_rate": 0.0001,
      "step": 12264
    },
    {
      "epoch": 28.538812785388128,
      "grad_norm": 0.1166076809167862,
      "learning_rate": 0.0001,
      "loss": 0.4644,
      "step": 12500
    },
    {
      "epoch": 29.0,
      "eval_explained_variance": 0.44008609652519226,
      "eval_kl_divergence": 0.42698290944099426,
      "eval_loss": 0.46892231702804565,
      "eval_mae": 0.11971781402826309,
      "eval_rmse": 0.16136477887630463,
      "eval_runtime": 67.0859,
      "eval_samples_per_second": 35.089,
      "eval_steps_per_second": 2.206,
      "learning_rate": 0.0001,
      "step": 12702
    },
    {
      "epoch": 29.680365296803654,
      "grad_norm": 0.1203789934515953,
      "learning_rate": 0.0001,
      "loss": 0.4638,
      "step": 13000
    },
    {
      "epoch": 30.0,
      "eval_explained_variance": 0.44359269738197327,
      "eval_kl_divergence": 0.2625378370285034,
      "eval_loss": 0.46987923979759216,
      "eval_mae": 0.12246056646108627,
      "eval_rmse": 0.16186781227588654,
      "eval_runtime": 67.1358,
      "eval_samples_per_second": 35.063,
      "eval_steps_per_second": 2.204,
      "learning_rate": 0.0001,
      "step": 13140
    },
    {
      "epoch": 30.82191780821918,
      "grad_norm": 0.11070472747087479,
      "learning_rate": 0.0001,
      "loss": 0.4636,
      "step": 13500
    },
    {
      "epoch": 31.0,
      "eval_explained_variance": 0.44310399889945984,
      "eval_kl_divergence": 0.38758543133735657,
      "eval_loss": 0.46842578053474426,
      "eval_mae": 0.11974402517080307,
      "eval_rmse": 0.16068722307682037,
      "eval_runtime": 68.317,
      "eval_samples_per_second": 34.457,
      "eval_steps_per_second": 2.166,
      "learning_rate": 0.0001,
      "step": 13578
    },
    {
      "epoch": 31.963470319634702,
      "grad_norm": 0.18743829429149628,
      "learning_rate": 0.0001,
      "loss": 0.463,
      "step": 14000
    },
    {
      "epoch": 32.0,
      "eval_explained_variance": 0.4466596841812134,
      "eval_kl_divergence": 0.4060189425945282,
      "eval_loss": 0.46784707903862,
      "eval_mae": 0.11951278895139694,
      "eval_rmse": 0.16003534197807312,
      "eval_runtime": 66.0349,
      "eval_samples_per_second": 35.648,
      "eval_steps_per_second": 2.241,
      "learning_rate": 0.0001,
      "step": 14016
    },
    {
      "epoch": 33.0,
      "eval_explained_variance": 0.44936656951904297,
      "eval_kl_divergence": 0.3688030540943146,
      "eval_loss": 0.46755874156951904,
      "eval_mae": 0.11934500932693481,
      "eval_rmse": 0.15963692963123322,
      "eval_runtime": 72.5096,
      "eval_samples_per_second": 32.465,
      "eval_steps_per_second": 2.041,
      "learning_rate": 0.0001,
      "step": 14454
    },
    {
      "epoch": 33.10502283105023,
      "grad_norm": 0.15971647202968597,
      "learning_rate": 0.0001,
      "loss": 0.4628,
      "step": 14500
    },
    {
      "epoch": 34.0,
      "eval_explained_variance": 0.4491260051727295,
      "eval_kl_divergence": 0.3900001645088196,
      "eval_loss": 0.46766504645347595,
      "eval_mae": 0.11943826079368591,
      "eval_rmse": 0.15995797514915466,
      "eval_runtime": 67.335,
      "eval_samples_per_second": 34.96,
      "eval_steps_per_second": 2.198,
      "learning_rate": 0.0001,
      "step": 14892
    },
    {
      "epoch": 34.24657534246575,
      "grad_norm": 0.15495429933071136,
      "learning_rate": 0.0001,
      "loss": 0.4616,
      "step": 15000
    },
    {
      "epoch": 35.0,
      "eval_explained_variance": 0.44995394349098206,
      "eval_kl_divergence": 0.42819926142692566,
      "eval_loss": 0.4670174717903137,
      "eval_mae": 0.11885016411542892,
      "eval_rmse": 0.1593056619167328,
      "eval_runtime": 69.6729,
      "eval_samples_per_second": 33.786,
      "eval_steps_per_second": 2.124,
      "learning_rate": 0.0001,
      "step": 15330
    },
    {
      "epoch": 35.38812785388128,
      "grad_norm": 0.1067986786365509,
      "learning_rate": 0.0001,
      "loss": 0.4634,
      "step": 15500
    },
    {
      "epoch": 36.0,
      "eval_explained_variance": 0.45064660906791687,
      "eval_kl_divergence": 0.44459816813468933,
      "eval_loss": 0.46679624915122986,
      "eval_mae": 0.11801303178071976,
      "eval_rmse": 0.15909221768379211,
      "eval_runtime": 70.3703,
      "eval_samples_per_second": 33.452,
      "eval_steps_per_second": 2.103,
      "learning_rate": 0.0001,
      "step": 15768
    },
    {
      "epoch": 36.529680365296805,
      "grad_norm": 0.17229855060577393,
      "learning_rate": 0.0001,
      "loss": 0.462,
      "step": 16000
    },
    {
      "epoch": 37.0,
      "eval_explained_variance": 0.4528321325778961,
      "eval_kl_divergence": 0.39424213767051697,
      "eval_loss": 0.46689239144325256,
      "eval_mae": 0.11852964758872986,
      "eval_rmse": 0.15902292728424072,
      "eval_runtime": 68.7832,
      "eval_samples_per_second": 34.223,
      "eval_steps_per_second": 2.152,
      "learning_rate": 0.0001,
      "step": 16206
    },
    {
      "epoch": 37.67123287671233,
      "grad_norm": 0.16489605605602264,
      "learning_rate": 0.0001,
      "loss": 0.4631,
      "step": 16500
    },
    {
      "epoch": 38.0,
      "eval_explained_variance": 0.4512017071247101,
      "eval_kl_divergence": 0.47831401228904724,
      "eval_loss": 0.4664570987224579,
      "eval_mae": 0.11769810318946838,
      "eval_rmse": 0.15879899263381958,
      "eval_runtime": 69.0328,
      "eval_samples_per_second": 34.1,
      "eval_steps_per_second": 2.144,
      "learning_rate": 0.0001,
      "step": 16644
    },
    {
      "epoch": 38.81278538812786,
      "grad_norm": 0.24003809690475464,
      "learning_rate": 0.0001,
      "loss": 0.4603,
      "step": 17000
    },
    {
      "epoch": 39.0,
      "eval_explained_variance": 0.4500379264354706,
      "eval_kl_divergence": 0.386750727891922,
      "eval_loss": 0.4674011468887329,
      "eval_mae": 0.11898898333311081,
      "eval_rmse": 0.15971434116363525,
      "eval_runtime": 69.047,
      "eval_samples_per_second": 34.093,
      "eval_steps_per_second": 2.143,
      "learning_rate": 0.0001,
      "step": 17082
    },
    {
      "epoch": 39.954337899543376,
      "grad_norm": 0.1894679218530655,
      "learning_rate": 0.0001,
      "loss": 0.4614,
      "step": 17500
    },
    {
      "epoch": 40.0,
      "eval_explained_variance": 0.4498337507247925,
      "eval_kl_divergence": 0.3627259135246277,
      "eval_loss": 0.4677062928676605,
      "eval_mae": 0.11947864294052124,
      "eval_rmse": 0.15986855328083038,
      "eval_runtime": 68.3668,
      "eval_samples_per_second": 34.432,
      "eval_steps_per_second": 2.165,
      "learning_rate": 0.0001,
      "step": 17520
    },
    {
      "epoch": 41.0,
      "eval_explained_variance": 0.4539731740951538,
      "eval_kl_divergence": 0.2654862701892853,
      "eval_loss": 0.46822381019592285,
      "eval_mae": 0.12105310708284378,
      "eval_rmse": 0.16022303700447083,
      "eval_runtime": 68.0126,
      "eval_samples_per_second": 34.611,
      "eval_steps_per_second": 2.176,
      "learning_rate": 0.0001,
      "step": 17958
    },
    {
      "epoch": 41.0958904109589,
      "grad_norm": 0.13239748775959015,
      "learning_rate": 0.0001,
      "loss": 0.4612,
      "step": 18000
    },
    {
      "epoch": 42.0,
      "eval_explained_variance": 0.45140552520751953,
      "eval_kl_divergence": 0.5071607828140259,
      "eval_loss": 0.4664672613143921,
      "eval_mae": 0.11723508685827255,
      "eval_rmse": 0.15891815721988678,
      "eval_runtime": 66.3853,
      "eval_samples_per_second": 35.46,
      "eval_steps_per_second": 2.229,
      "learning_rate": 0.0001,
      "step": 18396
    },
    {
      "epoch": 42.23744292237443,
      "grad_norm": 0.17189915478229523,
      "learning_rate": 0.0001,
      "loss": 0.462,
      "step": 18500
    },
    {
      "epoch": 43.0,
      "eval_explained_variance": 0.4554661214351654,
      "eval_kl_divergence": 0.430560827255249,
      "eval_loss": 0.46638762950897217,
      "eval_mae": 0.11766376346349716,
      "eval_rmse": 0.15850575268268585,
      "eval_runtime": 65.8594,
      "eval_samples_per_second": 35.743,
      "eval_steps_per_second": 2.247,
      "learning_rate": 0.0001,
      "step": 18834
    },
    {
      "epoch": 43.37899543378995,
      "grad_norm": 0.1566254049539566,
      "learning_rate": 0.0001,
      "loss": 0.4603,
      "step": 19000
    },
    {
      "epoch": 44.0,
      "eval_explained_variance": 0.45294663310050964,
      "eval_kl_divergence": 0.4114503562450409,
      "eval_loss": 0.46708741784095764,
      "eval_mae": 0.11923228949308395,
      "eval_rmse": 0.15942519903182983,
      "eval_runtime": 70.2132,
      "eval_samples_per_second": 33.526,
      "eval_steps_per_second": 2.108,
      "learning_rate": 0.0001,
      "step": 19272
    },
    {
      "epoch": 44.52054794520548,
      "grad_norm": 0.1487286388874054,
      "learning_rate": 0.0001,
      "loss": 0.4599,
      "step": 19500
    },
    {
      "epoch": 45.0,
      "eval_explained_variance": 0.45276129245758057,
      "eval_kl_divergence": 0.441719114780426,
      "eval_loss": 0.46663177013397217,
      "eval_mae": 0.11713916063308716,
      "eval_rmse": 0.15901675820350647,
      "eval_runtime": 67.4478,
      "eval_samples_per_second": 34.901,
      "eval_steps_per_second": 2.194,
      "learning_rate": 0.0001,
      "step": 19710
    },
    {
      "epoch": 45.662100456621005,
      "grad_norm": 0.17072182893753052,
      "learning_rate": 0.0001,
      "loss": 0.4612,
      "step": 20000
    },
    {
      "epoch": 46.0,
      "eval_explained_variance": 0.4574218690395355,
      "eval_kl_divergence": 0.3686365485191345,
      "eval_loss": 0.4663327634334564,
      "eval_mae": 0.11786039918661118,
      "eval_rmse": 0.15854045748710632,
      "eval_runtime": 66.0464,
      "eval_samples_per_second": 35.642,
      "eval_steps_per_second": 2.241,
      "learning_rate": 0.0001,
      "step": 20148
    },
    {
      "epoch": 46.80365296803653,
      "grad_norm": 0.17141404747962952,
      "learning_rate": 0.0001,
      "loss": 0.4596,
      "step": 20500
    },
    {
      "epoch": 47.0,
      "eval_explained_variance": 0.4567171633243561,
      "eval_kl_divergence": 0.5089961886405945,
      "eval_loss": 0.46577510237693787,
      "eval_mae": 0.11723583936691284,
      "eval_rmse": 0.15815743803977966,
      "eval_runtime": 68.7513,
      "eval_samples_per_second": 34.239,
      "eval_steps_per_second": 2.153,
      "learning_rate": 0.0001,
      "step": 20586
    },
    {
      "epoch": 47.945205479452056,
      "grad_norm": 0.13870340585708618,
      "learning_rate": 0.0001,
      "loss": 0.4603,
      "step": 21000
    },
    {
      "epoch": 48.0,
      "eval_explained_variance": 0.4547780752182007,
      "eval_kl_divergence": 0.5279051065444946,
      "eval_loss": 0.46634000539779663,
      "eval_mae": 0.11751776933670044,
      "eval_rmse": 0.15890929102897644,
      "eval_runtime": 66.5777,
      "eval_samples_per_second": 35.357,
      "eval_steps_per_second": 2.223,
      "learning_rate": 0.0001,
      "step": 21024
    },
    {
      "epoch": 49.0,
      "eval_explained_variance": 0.4532240927219391,
      "eval_kl_divergence": 0.44965481758117676,
      "eval_loss": 0.46656596660614014,
      "eval_mae": 0.11831226199865341,
      "eval_rmse": 0.15911053121089935,
      "eval_runtime": 67.2369,
      "eval_samples_per_second": 35.011,
      "eval_steps_per_second": 2.201,
      "learning_rate": 0.0001,
      "step": 21462
    },
    {
      "epoch": 49.08675799086758,
      "grad_norm": 0.16372357308864594,
      "learning_rate": 0.0001,
      "loss": 0.4599,
      "step": 21500
    },
    {
      "epoch": 50.0,
      "eval_explained_variance": 0.45799562335014343,
      "eval_kl_divergence": 0.27116483449935913,
      "eval_loss": 0.46755433082580566,
      "eval_mae": 0.1204712763428688,
      "eval_rmse": 0.15946535766124725,
      "eval_runtime": 67.5446,
      "eval_samples_per_second": 34.851,
      "eval_steps_per_second": 2.191,
      "learning_rate": 0.0001,
      "step": 21900
    },
    {
      "epoch": 50.22831050228311,
      "grad_norm": 0.15486378967761993,
      "learning_rate": 0.0001,
      "loss": 0.4594,
      "step": 22000
    },
    {
      "epoch": 51.0,
      "eval_explained_variance": 0.4551831781864166,
      "eval_kl_divergence": 0.4008127450942993,
      "eval_loss": 0.46639156341552734,
      "eval_mae": 0.11721646040678024,
      "eval_rmse": 0.15861637890338898,
      "eval_runtime": 64.0368,
      "eval_samples_per_second": 36.76,
      "eval_steps_per_second": 2.311,
      "learning_rate": 0.0001,
      "step": 22338
    },
    {
      "epoch": 51.36986301369863,
      "grad_norm": 0.2131696492433548,
      "learning_rate": 0.0001,
      "loss": 0.4593,
      "step": 22500
    },
    {
      "epoch": 52.0,
      "eval_explained_variance": 0.4556889832019806,
      "eval_kl_divergence": 0.4921821057796478,
      "eval_loss": 0.46591076254844666,
      "eval_mae": 0.11625734716653824,
      "eval_rmse": 0.1582706868648529,
      "eval_runtime": 68.9844,
      "eval_samples_per_second": 34.124,
      "eval_steps_per_second": 2.145,
      "learning_rate": 0.0001,
      "step": 22776
    },
    {
      "epoch": 52.51141552511415,
      "grad_norm": 0.3151676654815674,
      "learning_rate": 0.0001,
      "loss": 0.4614,
      "step": 23000
    },
    {
      "epoch": 53.0,
      "eval_explained_variance": 0.4601232707500458,
      "eval_kl_divergence": 0.4273536205291748,
      "eval_loss": 0.4656851887702942,
      "eval_mae": 0.1177864745259285,
      "eval_rmse": 0.15793031454086304,
      "eval_runtime": 68.0941,
      "eval_samples_per_second": 34.57,
      "eval_steps_per_second": 2.173,
      "learning_rate": 0.0001,
      "step": 23214
    },
    {
      "epoch": 53.65296803652968,
      "grad_norm": 0.1899169534444809,
      "learning_rate": 0.0001,
      "loss": 0.4592,
      "step": 23500
    },
    {
      "epoch": 54.0,
      "eval_explained_variance": 0.456957072019577,
      "eval_kl_divergence": 0.4574394226074219,
      "eval_loss": 0.46629655361175537,
      "eval_mae": 0.11577478051185608,
      "eval_rmse": 0.15850554406642914,
      "eval_runtime": 65.2277,
      "eval_samples_per_second": 36.089,
      "eval_steps_per_second": 2.269,
      "learning_rate": 0.0001,
      "step": 23652
    },
    {
      "epoch": 54.794520547945204,
      "grad_norm": 0.27214938402175903,
      "learning_rate": 0.0001,
      "loss": 0.4601,
      "step": 24000
    },
    {
      "epoch": 55.0,
      "eval_explained_variance": 0.45804765820503235,
      "eval_kl_divergence": 0.34863969683647156,
      "eval_loss": 0.46644341945648193,
      "eval_mae": 0.11888903379440308,
      "eval_rmse": 0.1585645228624344,
      "eval_runtime": 62.989,
      "eval_samples_per_second": 37.372,
      "eval_steps_per_second": 2.35,
      "learning_rate": 0.0001,
      "step": 24090
    },
    {
      "epoch": 55.93607305936073,
      "grad_norm": 0.32561782002449036,
      "learning_rate": 0.0001,
      "loss": 0.4589,
      "step": 24500
    },
    {
      "epoch": 56.0,
      "eval_explained_variance": 0.46123751997947693,
      "eval_kl_divergence": 0.3015596568584442,
      "eval_loss": 0.4661739766597748,
      "eval_mae": 0.1184321939945221,
      "eval_rmse": 0.15838663280010223,
      "eval_runtime": 62.9298,
      "eval_samples_per_second": 37.407,
      "eval_steps_per_second": 2.352,
      "learning_rate": 0.0001,
      "step": 24528
    },
    {
      "epoch": 57.0,
      "eval_explained_variance": 0.45534956455230713,
      "eval_kl_divergence": 0.4162982106208801,
      "eval_loss": 0.46634721755981445,
      "eval_mae": 0.11808174103498459,
      "eval_rmse": 0.15866883099079132,
      "eval_runtime": 63.3411,
      "eval_samples_per_second": 37.164,
      "eval_steps_per_second": 2.337,
      "learning_rate": 0.0001,
      "step": 24966
    },
    {
      "epoch": 57.077625570776256,
      "grad_norm": 0.21059946715831757,
      "learning_rate": 0.0001,
      "loss": 0.4588,
      "step": 25000
    },
    {
      "epoch": 58.0,
      "eval_explained_variance": 0.4557063579559326,
      "eval_kl_divergence": 0.3399294316768646,
      "eval_loss": 0.4673805236816406,
      "eval_mae": 0.11887022852897644,
      "eval_rmse": 0.15934236347675323,
      "eval_runtime": 65.9311,
      "eval_samples_per_second": 35.704,
      "eval_steps_per_second": 2.245,
      "learning_rate": 0.0001,
      "step": 25404
    },
    {
      "epoch": 58.21917808219178,
      "grad_norm": 0.1977730244398117,
      "learning_rate": 0.0001,
      "loss": 0.4595,
      "step": 25500
    },
    {
      "epoch": 59.0,
      "eval_explained_variance": 0.4646029472351074,
      "eval_kl_divergence": 0.3685876429080963,
      "eval_loss": 0.4650005102157593,
      "eval_mae": 0.11696922034025192,
      "eval_rmse": 0.15720517933368683,
      "eval_runtime": 63.6041,
      "eval_samples_per_second": 37.01,
      "eval_steps_per_second": 2.327,
      "learning_rate": 0.0001,
      "step": 25842
    },
    {
      "epoch": 59.36073059360731,
      "grad_norm": 0.22561948001384735,
      "learning_rate": 0.0001,
      "loss": 0.4594,
      "step": 26000
    },
    {
      "epoch": 60.0,
      "eval_explained_variance": 0.4566896855831146,
      "eval_kl_divergence": 0.45347684621810913,
      "eval_loss": 0.46599113941192627,
      "eval_mae": 0.11718535423278809,
      "eval_rmse": 0.15840259194374084,
      "eval_runtime": 62.901,
      "eval_samples_per_second": 37.424,
      "eval_steps_per_second": 2.353,
      "learning_rate": 0.0001,
      "step": 26280
    },
    {
      "epoch": 60.50228310502283,
      "grad_norm": 0.19251562654972076,
      "learning_rate": 0.0001,
      "loss": 0.4599,
      "step": 26500
    },
    {
      "epoch": 61.0,
      "eval_explained_variance": 0.4583926200866699,
      "eval_kl_divergence": 0.375108540058136,
      "eval_loss": 0.4662201702594757,
      "eval_mae": 0.11791958659887314,
      "eval_rmse": 0.1585090458393097,
      "eval_runtime": 61.9039,
      "eval_samples_per_second": 38.027,
      "eval_steps_per_second": 2.391,
      "learning_rate": 0.0001,
      "step": 26718
    },
    {
      "epoch": 61.64383561643836,
      "grad_norm": 0.2111298143863678,
      "learning_rate": 0.0001,
      "loss": 0.4584,
      "step": 27000
    },
    {
      "epoch": 62.0,
      "eval_explained_variance": 0.4587836265563965,
      "eval_kl_divergence": 0.35335177183151245,
      "eval_loss": 0.46614503860473633,
      "eval_mae": 0.11732319742441177,
      "eval_rmse": 0.15833334624767303,
      "eval_runtime": 63.6491,
      "eval_samples_per_second": 36.984,
      "eval_steps_per_second": 2.325,
      "learning_rate": 0.0001,
      "step": 27156
    },
    {
      "epoch": 62.78538812785388,
      "grad_norm": 0.21951091289520264,
      "learning_rate": 0.0001,
      "loss": 0.4575,
      "step": 27500
    },
    {
      "epoch": 63.0,
      "eval_explained_variance": 0.4576869010925293,
      "eval_kl_divergence": 0.40475067496299744,
      "eval_loss": 0.4660026431083679,
      "eval_mae": 0.11627298593521118,
      "eval_rmse": 0.1582755148410797,
      "eval_runtime": 62.5657,
      "eval_samples_per_second": 37.624,
      "eval_steps_per_second": 2.366,
      "learning_rate": 0.0001,
      "step": 27594
    },
    {
      "epoch": 63.926940639269404,
      "grad_norm": 0.21228961646556854,
      "learning_rate": 0.0001,
      "loss": 0.4598,
      "step": 28000
    },
    {
      "epoch": 64.0,
      "eval_explained_variance": 0.46285080909729004,
      "eval_kl_divergence": 0.2471023052930832,
      "eval_loss": 0.46711620688438416,
      "eval_mae": 0.11878780275583267,
      "eval_rmse": 0.15883709490299225,
      "eval_runtime": 62.9069,
      "eval_samples_per_second": 37.42,
      "eval_steps_per_second": 2.353,
      "learning_rate": 0.0001,
      "step": 28032
    },
    {
      "epoch": 65.0,
      "eval_explained_variance": 0.46037113666534424,
      "eval_kl_divergence": 0.4526395797729492,
      "eval_loss": 0.46536803245544434,
      "eval_mae": 0.11657585948705673,
      "eval_rmse": 0.15768831968307495,
      "eval_runtime": 65.6932,
      "eval_samples_per_second": 35.833,
      "eval_steps_per_second": 2.253,
      "learning_rate": 0.0001,
      "step": 28470
    },
    {
      "epoch": 65.06849315068493,
      "grad_norm": 0.23406238853931427,
      "learning_rate": 1e-05,
      "loss": 0.4582,
      "step": 28500
    },
    {
      "epoch": 66.0,
      "eval_explained_variance": 0.45918479561805725,
      "eval_kl_divergence": 0.5259271264076233,
      "eval_loss": 0.46573594212532043,
      "eval_mae": 0.11610028892755508,
      "eval_rmse": 0.15816400945186615,
      "eval_runtime": 63.8294,
      "eval_samples_per_second": 36.88,
      "eval_steps_per_second": 2.319,
      "learning_rate": 1e-05,
      "step": 28908
    },
    {
      "epoch": 66.21004566210046,
      "grad_norm": 0.20290181040763855,
      "learning_rate": 1e-05,
      "loss": 0.4592,
      "step": 29000
    },
    {
      "epoch": 67.0,
      "eval_explained_variance": 0.4623439311981201,
      "eval_kl_divergence": 0.4251798987388611,
      "eval_loss": 0.4653942584991455,
      "eval_mae": 0.11725304275751114,
      "eval_rmse": 0.1573910117149353,
      "eval_runtime": 65.929,
      "eval_samples_per_second": 35.705,
      "eval_steps_per_second": 2.245,
      "learning_rate": 1e-05,
      "step": 29346
    },
    {
      "epoch": 67.35159817351598,
      "grad_norm": 0.210421621799469,
      "learning_rate": 1e-05,
      "loss": 0.4573,
      "step": 29500
    },
    {
      "epoch": 68.0,
      "eval_explained_variance": 0.46140334010124207,
      "eval_kl_divergence": 0.49885210394859314,
      "eval_loss": 0.46487176418304443,
      "eval_mae": 0.11542203277349472,
      "eval_rmse": 0.15722759068012238,
      "eval_runtime": 63.8681,
      "eval_samples_per_second": 36.857,
      "eval_steps_per_second": 2.317,
      "learning_rate": 1e-05,
      "step": 29784
    },
    {
      "epoch": 68.4931506849315,
      "grad_norm": 0.3686019480228424,
      "learning_rate": 1e-05,
      "loss": 0.4576,
      "step": 30000
    },
    {
      "epoch": 69.0,
      "eval_explained_variance": 0.46442678570747375,
      "eval_kl_divergence": 0.40228068828582764,
      "eval_loss": 0.465110719203949,
      "eval_mae": 0.11607488244771957,
      "eval_rmse": 0.15701350569725037,
      "eval_runtime": 68.8094,
      "eval_samples_per_second": 34.21,
      "eval_steps_per_second": 2.151,
      "learning_rate": 1e-05,
      "step": 30222
    },
    {
      "epoch": 69.63470319634703,
      "grad_norm": 0.3072531223297119,
      "learning_rate": 1e-05,
      "loss": 0.4556,
      "step": 30500
    },
    {
      "epoch": 70.0,
      "eval_explained_variance": 0.4622003734111786,
      "eval_kl_divergence": 0.41175633668899536,
      "eval_loss": 0.466043084859848,
      "eval_mae": 0.1165715679526329,
      "eval_rmse": 0.15760619938373566,
      "eval_runtime": 67.6756,
      "eval_samples_per_second": 34.784,
      "eval_steps_per_second": 2.187,
      "learning_rate": 1e-05,
      "step": 30660
    },
    {
      "epoch": 70.77625570776256,
      "grad_norm": 0.2429497092962265,
      "learning_rate": 1e-05,
      "loss": 0.4591,
      "step": 31000
    },
    {
      "epoch": 71.0,
      "eval_explained_variance": 0.4643818140029907,
      "eval_kl_divergence": 0.3074641227722168,
      "eval_loss": 0.46608996391296387,
      "eval_mae": 0.11772569268941879,
      "eval_rmse": 0.15784674882888794,
      "eval_runtime": 72.04,
      "eval_samples_per_second": 32.676,
      "eval_steps_per_second": 2.054,
      "learning_rate": 1e-05,
      "step": 31098
    },
    {
      "epoch": 71.91780821917808,
      "grad_norm": 0.1753670871257782,
      "learning_rate": 1e-05,
      "loss": 0.4563,
      "step": 31500
    },
    {
      "epoch": 72.0,
      "eval_explained_variance": 0.46214058995246887,
      "eval_kl_divergence": 0.38358867168426514,
      "eval_loss": 0.46582332253456116,
      "eval_mae": 0.11707664281129837,
      "eval_rmse": 0.1579793244600296,
      "eval_runtime": 66.3861,
      "eval_samples_per_second": 35.459,
      "eval_steps_per_second": 2.229,
      "learning_rate": 1e-05,
      "step": 31536
    },
    {
      "epoch": 73.0,
      "eval_explained_variance": 0.464040070772171,
      "eval_kl_divergence": 0.45435720682144165,
      "eval_loss": 0.4648771584033966,
      "eval_mae": 0.11544410139322281,
      "eval_rmse": 0.15694700181484222,
      "eval_runtime": 64.6485,
      "eval_samples_per_second": 36.412,
      "eval_steps_per_second": 2.289,
      "learning_rate": 1e-05,
      "step": 31974
    },
    {
      "epoch": 73.05936073059361,
      "grad_norm": 0.2539682686328888,
      "learning_rate": 1e-05,
      "loss": 0.4577,
      "step": 32000
    },
    {
      "epoch": 74.0,
      "eval_explained_variance": 0.46597161889076233,
      "eval_kl_divergence": 0.4537515938282013,
      "eval_loss": 0.46466848254203796,
      "eval_mae": 0.11628029495477676,
      "eval_rmse": 0.1567227691411972,
      "eval_runtime": 64.3001,
      "eval_samples_per_second": 36.61,
      "eval_steps_per_second": 2.302,
      "learning_rate": 1e-05,
      "step": 32412
    },
    {
      "epoch": 74.20091324200914,
      "grad_norm": 0.19458268582820892,
      "learning_rate": 1e-05,
      "loss": 0.4576,
      "step": 32500
    },
    {
      "epoch": 75.0,
      "eval_explained_variance": 0.46580255031585693,
      "eval_kl_divergence": 0.33480265736579895,
      "eval_loss": 0.46563470363616943,
      "eval_mae": 0.11664699763059616,
      "eval_rmse": 0.1572960466146469,
      "eval_runtime": 65.0705,
      "eval_samples_per_second": 36.176,
      "eval_steps_per_second": 2.274,
      "learning_rate": 1e-05,
      "step": 32850
    },
    {
      "epoch": 75.34246575342466,
      "grad_norm": 0.23031042516231537,
      "learning_rate": 1e-05,
      "loss": 0.457,
      "step": 33000
    },
    {
      "epoch": 76.0,
      "eval_explained_variance": 0.4644790589809418,
      "eval_kl_divergence": 0.49758121371269226,
      "eval_loss": 0.4647076725959778,
      "eval_mae": 0.11578535288572311,
      "eval_rmse": 0.15706199407577515,
      "eval_runtime": 64.4752,
      "eval_samples_per_second": 36.51,
      "eval_steps_per_second": 2.295,
      "learning_rate": 1e-05,
      "step": 33288
    },
    {
      "epoch": 76.48401826484019,
      "grad_norm": 0.21923455595970154,
      "learning_rate": 1e-05,
      "loss": 0.4574,
      "step": 33500
    },
    {
      "epoch": 77.0,
      "eval_explained_variance": 0.46530094742774963,
      "eval_kl_divergence": 0.3934381902217865,
      "eval_loss": 0.4650570750236511,
      "eval_mae": 0.11632921546697617,
      "eval_rmse": 0.15704087913036346,
      "eval_runtime": 65.0363,
      "eval_samples_per_second": 36.195,
      "eval_steps_per_second": 2.276,
      "learning_rate": 1e-05,
      "step": 33726
    },
    {
      "epoch": 77.62557077625571,
      "grad_norm": 0.27394065260887146,
      "learning_rate": 1e-05,
      "loss": 0.457,
      "step": 34000
    },
    {
      "epoch": 78.0,
      "eval_explained_variance": 0.4654146730899811,
      "eval_kl_divergence": 0.39355090260505676,
      "eval_loss": 0.46495845913887024,
      "eval_mae": 0.11605050414800644,
      "eval_rmse": 0.15708379447460175,
      "eval_runtime": 64.2371,
      "eval_samples_per_second": 36.645,
      "eval_steps_per_second": 2.304,
      "learning_rate": 1e-05,
      "step": 34164
    },
    {
      "epoch": 78.76712328767124,
      "grad_norm": 0.20854905247688293,
      "learning_rate": 1e-05,
      "loss": 0.4566,
      "step": 34500
    },
    {
      "epoch": 79.0,
      "eval_explained_variance": 0.4653468132019043,
      "eval_kl_divergence": 0.37590739130973816,
      "eval_loss": 0.46530288457870483,
      "eval_mae": 0.11588699370622635,
      "eval_rmse": 0.15725918114185333,
      "eval_runtime": 62.4443,
      "eval_samples_per_second": 37.698,
      "eval_steps_per_second": 2.37,
      "learning_rate": 1e-05,
      "step": 34602
    },
    {
      "epoch": 79.90867579908675,
      "grad_norm": 0.21241584420204163,
      "learning_rate": 1e-05,
      "loss": 0.458,
      "step": 35000
    },
    {
      "epoch": 80.0,
      "eval_explained_variance": 0.46603894233703613,
      "eval_kl_divergence": 0.41887620091438293,
      "eval_loss": 0.4647064805030823,
      "eval_mae": 0.11623198539018631,
      "eval_rmse": 0.15669378638267517,
      "eval_runtime": 62.0038,
      "eval_samples_per_second": 37.965,
      "eval_steps_per_second": 2.387,
      "learning_rate": 1e-05,
      "step": 35040
    },
    {
      "epoch": 81.0,
      "eval_explained_variance": 0.4646754562854767,
      "eval_kl_divergence": 0.47510233521461487,
      "eval_loss": 0.46485888957977295,
      "eval_mae": 0.11576662212610245,
      "eval_rmse": 0.15710778534412384,
      "eval_runtime": 61.3344,
      "eval_samples_per_second": 38.38,
      "eval_steps_per_second": 2.413,
      "learning_rate": 1.0000000000000002e-06,
      "step": 35478
    },
    {
      "epoch": 81.05022831050228,
      "grad_norm": 0.23253624141216278,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.456,
      "step": 35500
    },
    {
      "epoch": 82.0,
      "eval_explained_variance": 0.4651080071926117,
      "eval_kl_divergence": 0.43354716897010803,
      "eval_loss": 0.4654049277305603,
      "eval_mae": 0.11612025648355484,
      "eval_rmse": 0.1572476029396057,
      "eval_runtime": 62.2235,
      "eval_samples_per_second": 37.831,
      "eval_steps_per_second": 2.379,
      "learning_rate": 1.0000000000000002e-06,
      "step": 35916
    },
    {
      "epoch": 82.1917808219178,
      "grad_norm": 0.23126010596752167,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4564,
      "step": 36000
    },
    {
      "epoch": 83.0,
      "eval_explained_variance": 0.46666058897972107,
      "eval_kl_divergence": 0.3906138837337494,
      "eval_loss": 0.46466442942619324,
      "eval_mae": 0.11605874449014664,
      "eval_rmse": 0.15662842988967896,
      "eval_runtime": 61.3486,
      "eval_samples_per_second": 38.371,
      "eval_steps_per_second": 2.412,
      "learning_rate": 1.0000000000000002e-06,
      "step": 36354
    },
    {
      "epoch": 83.33333333333333,
      "grad_norm": 0.2322608381509781,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4575,
      "step": 36500
    },
    {
      "epoch": 84.0,
      "eval_explained_variance": 0.46770432591438293,
      "eval_kl_divergence": 0.3855327367782593,
      "eval_loss": 0.46430692076683044,
      "eval_mae": 0.11573296785354614,
      "eval_rmse": 0.15642325580120087,
      "eval_runtime": 66.0549,
      "eval_samples_per_second": 35.637,
      "eval_steps_per_second": 2.241,
      "learning_rate": 1.0000000000000002e-06,
      "step": 36792
    },
    {
      "epoch": 84.47488584474885,
      "grad_norm": 0.23277735710144043,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4557,
      "step": 37000
    },
    {
      "epoch": 85.0,
      "eval_explained_variance": 0.46685031056404114,
      "eval_kl_divergence": 0.3372012972831726,
      "eval_loss": 0.46528080105781555,
      "eval_mae": 0.11732091754674911,
      "eval_rmse": 0.1570904403924942,
      "eval_runtime": 64.5003,
      "eval_samples_per_second": 36.496,
      "eval_steps_per_second": 2.295,
      "learning_rate": 1.0000000000000002e-06,
      "step": 37230
    },
    {
      "epoch": 85.61643835616438,
      "grad_norm": 0.40314728021621704,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4587,
      "step": 37500
    },
    {
      "epoch": 86.0,
      "eval_explained_variance": 0.46858659386634827,
      "eval_kl_divergence": 0.29690617322921753,
      "eval_loss": 0.46546733379364014,
      "eval_mae": 0.11843166500329971,
      "eval_rmse": 0.15721069276332855,
      "eval_runtime": 63.2115,
      "eval_samples_per_second": 37.24,
      "eval_steps_per_second": 2.341,
      "learning_rate": 1.0000000000000002e-06,
      "step": 37668
    },
    {
      "epoch": 86.7579908675799,
      "grad_norm": 0.2404189556837082,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4564,
      "step": 38000
    },
    {
      "epoch": 87.0,
      "eval_explained_variance": 0.4670089781284332,
      "eval_kl_divergence": 0.3571859300136566,
      "eval_loss": 0.4651782214641571,
      "eval_mae": 0.11727560311555862,
      "eval_rmse": 0.15713386237621307,
      "eval_runtime": 63.9279,
      "eval_samples_per_second": 36.823,
      "eval_steps_per_second": 2.315,
      "learning_rate": 1.0000000000000002e-06,
      "step": 38106
    },
    {
      "epoch": 87.89954337899543,
      "grad_norm": 0.27116990089416504,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4565,
      "step": 38500
    },
    {
      "epoch": 88.0,
      "eval_explained_variance": 0.4626586139202118,
      "eval_kl_divergence": 0.5179023146629333,
      "eval_loss": 0.4655611217021942,
      "eval_mae": 0.11513540893793106,
      "eval_rmse": 0.15777353942394257,
      "eval_runtime": 64.6969,
      "eval_samples_per_second": 36.385,
      "eval_steps_per_second": 2.288,
      "learning_rate": 1.0000000000000002e-06,
      "step": 38544
    },
    {
      "epoch": 89.0,
      "eval_explained_variance": 0.4670461118221283,
      "eval_kl_divergence": 0.29475125670433044,
      "eval_loss": 0.4654468297958374,
      "eval_mae": 0.11768680065870285,
      "eval_rmse": 0.15736806392669678,
      "eval_runtime": 64.3025,
      "eval_samples_per_second": 36.608,
      "eval_steps_per_second": 2.302,
      "learning_rate": 1.0000000000000002e-06,
      "step": 38982
    },
    {
      "epoch": 89.04109589041096,
      "grad_norm": 0.35173681378364563,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.4569,
      "step": 39000
    },
    {
      "epoch": 90.0,
      "eval_explained_variance": 0.46744585037231445,
      "eval_kl_divergence": 0.34272825717926025,
      "eval_loss": 0.4649873971939087,
      "eval_mae": 0.11670031398534775,
      "eval_rmse": 0.15687011182308197,
      "eval_runtime": 63.1964,
      "eval_samples_per_second": 37.249,
      "eval_steps_per_second": 2.342,
      "learning_rate": 1.0000000000000002e-06,
      "step": 39420
    },
    {
      "epoch": 90.18264840182648,
      "grad_norm": 0.27881479263305664,
      "learning_rate": 1.0000000000000002e-07,
      "loss": 0.4561,
      "step": 39500
    },
    {
      "epoch": 91.0,
      "eval_explained_variance": 0.4690658748149872,
      "eval_kl_divergence": 0.27898427844047546,
      "eval_loss": 0.4655340015888214,
      "eval_mae": 0.11734825372695923,
      "eval_rmse": 0.15720801055431366,
      "eval_runtime": 62.9781,
      "eval_samples_per_second": 37.378,
      "eval_steps_per_second": 2.35,
      "learning_rate": 1.0000000000000002e-07,
      "step": 39858
    },
    {
      "epoch": 91.32420091324201,
      "grad_norm": 0.18466131389141083,
      "learning_rate": 1.0000000000000002e-07,
      "loss": 0.4575,
      "step": 40000
    },
    {
      "epoch": 92.0,
      "eval_explained_variance": 0.4671539068222046,
      "eval_kl_divergence": 0.41531136631965637,
      "eval_loss": 0.4645555317401886,
      "eval_mae": 0.11530810594558716,
      "eval_rmse": 0.15662522614002228,
      "eval_runtime": 65.5509,
      "eval_samples_per_second": 35.911,
      "eval_steps_per_second": 2.258,
      "learning_rate": 1.0000000000000002e-07,
      "step": 40296
    },
    {
      "epoch": 92.46575342465754,
      "grad_norm": 0.18533748388290405,
      "learning_rate": 1.0000000000000002e-07,
      "loss": 0.4569,
      "step": 40500
    },
    {
      "epoch": 93.0,
      "eval_explained_variance": 0.464511513710022,
      "eval_kl_divergence": 0.46641936898231506,
      "eval_loss": 0.4648568034172058,
      "eval_mae": 0.11531826853752136,
      "eval_rmse": 0.1570696085691452,
      "eval_runtime": 64.6542,
      "eval_samples_per_second": 36.409,
      "eval_steps_per_second": 2.289,
      "learning_rate": 1.0000000000000002e-07,
      "step": 40734
    },
    {
      "epoch": 93.60730593607306,
      "grad_norm": 0.27212274074554443,
      "learning_rate": 1.0000000000000002e-07,
      "loss": 0.456,
      "step": 41000
    },
    {
      "epoch": 94.0,
      "eval_explained_variance": 0.4661947786808014,
      "eval_kl_divergence": 0.3858942687511444,
      "eval_loss": 0.4652610421180725,
      "eval_mae": 0.11592459678649902,
      "eval_rmse": 0.15684206783771515,
      "eval_runtime": 63.3717,
      "eval_samples_per_second": 37.146,
      "eval_steps_per_second": 2.335,
      "learning_rate": 1.0000000000000002e-07,
      "step": 41172
    },
    {
      "epoch": 94.0,
      "learning_rate": 1.0000000000000002e-07,
      "step": 41172,
      "total_flos": 9.743166860733663e+19,
      "train_loss": 0.46343856458701926,
      "train_runtime": 26810.1674,
      "train_samples_per_second": 39.181,
      "train_steps_per_second": 2.451
    }
  ],
  "logging_steps": 500,
  "max_steps": 65700,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 150,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 10,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.743166860733663e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}