{
  "best_metric": 0.9168,
  "best_model_checkpoint": "beit-base-patch16-224-pt22k-ft22k-finetuning/checkpoint-1053",
  "epoch": 2.9936034115138592,
  "eval_steps": 500,
  "global_step": 1053,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.028429282160625444,
      "grad_norm": 10.06214427947998,
      "learning_rate": 4.716981132075472e-06,
      "loss": 2.362,
      "step": 10
    },
    {
      "epoch": 0.05685856432125089,
      "grad_norm": 8.617206573486328,
      "learning_rate": 9.433962264150944e-06,
      "loss": 2.1938,
      "step": 20
    },
    {
      "epoch": 0.08528784648187633,
      "grad_norm": 8.857650756835938,
      "learning_rate": 1.4150943396226415e-05,
      "loss": 1.9819,
      "step": 30
    },
    {
      "epoch": 0.11371712864250177,
      "grad_norm": 10.247784614562988,
      "learning_rate": 1.8867924528301888e-05,
      "loss": 1.6578,
      "step": 40
    },
    {
      "epoch": 0.14214641080312723,
      "grad_norm": 13.104848861694336,
      "learning_rate": 2.358490566037736e-05,
      "loss": 1.3054,
      "step": 50
    },
    {
      "epoch": 0.17057569296375266,
      "grad_norm": 10.21298885345459,
      "learning_rate": 2.830188679245283e-05,
      "loss": 1.0335,
      "step": 60
    },
    {
      "epoch": 0.19900497512437812,
      "grad_norm": 10.701081275939941,
      "learning_rate": 3.30188679245283e-05,
      "loss": 0.8563,
      "step": 70
    },
    {
      "epoch": 0.22743425728500355,
      "grad_norm": 11.215727806091309,
      "learning_rate": 3.7735849056603776e-05,
      "loss": 0.7059,
      "step": 80
    },
    {
      "epoch": 0.255863539445629,
      "grad_norm": 8.667176246643066,
      "learning_rate": 4.245283018867925e-05,
      "loss": 0.638,
      "step": 90
    },
    {
      "epoch": 0.28429282160625446,
      "grad_norm": 12.863126754760742,
      "learning_rate": 4.716981132075472e-05,
      "loss": 0.6131,
      "step": 100
    },
    {
      "epoch": 0.31272210376687987,
      "grad_norm": 8.988736152648926,
      "learning_rate": 4.978880675818374e-05,
      "loss": 0.5919,
      "step": 110
    },
    {
      "epoch": 0.3411513859275053,
      "grad_norm": 8.396913528442383,
      "learning_rate": 4.9260823653643085e-05,
      "loss": 0.5408,
      "step": 120
    },
    {
      "epoch": 0.3695806680881308,
      "grad_norm": 6.962805271148682,
      "learning_rate": 4.8732840549102435e-05,
      "loss": 0.5122,
      "step": 130
    },
    {
      "epoch": 0.39800995024875624,
      "grad_norm": 6.434642314910889,
      "learning_rate": 4.820485744456177e-05,
      "loss": 0.5033,
      "step": 140
    },
    {
      "epoch": 0.42643923240938164,
      "grad_norm": 7.300764083862305,
      "learning_rate": 4.767687434002112e-05,
      "loss": 0.4988,
      "step": 150
    },
    {
      "epoch": 0.4548685145700071,
      "grad_norm": 7.479175567626953,
      "learning_rate": 4.7148891235480466e-05,
      "loss": 0.4234,
      "step": 160
    },
    {
      "epoch": 0.48329779673063256,
      "grad_norm": 6.13791036605835,
      "learning_rate": 4.662090813093981e-05,
      "loss": 0.4344,
      "step": 170
    },
    {
      "epoch": 0.511727078891258,
      "grad_norm": 9.220943450927734,
      "learning_rate": 4.609292502639916e-05,
      "loss": 0.4222,
      "step": 180
    },
    {
      "epoch": 0.5401563610518835,
      "grad_norm": 7.470913410186768,
      "learning_rate": 4.55649419218585e-05,
      "loss": 0.4733,
      "step": 190
    },
    {
      "epoch": 0.5685856432125089,
      "grad_norm": 6.389730930328369,
      "learning_rate": 4.503695881731785e-05,
      "loss": 0.4312,
      "step": 200
    },
    {
      "epoch": 0.5970149253731343,
      "grad_norm": 9.146849632263184,
      "learning_rate": 4.45089757127772e-05,
      "loss": 0.4623,
      "step": 210
    },
    {
      "epoch": 0.6254442075337597,
      "grad_norm": 6.792445659637451,
      "learning_rate": 4.398099260823654e-05,
      "loss": 0.469,
      "step": 220
    },
    {
      "epoch": 0.6538734896943852,
      "grad_norm": 7.154394149780273,
      "learning_rate": 4.3453009503695884e-05,
      "loss": 0.422,
      "step": 230
    },
    {
      "epoch": 0.6823027718550106,
      "grad_norm": 8.260627746582031,
      "learning_rate": 4.292502639915523e-05,
      "loss": 0.4369,
      "step": 240
    },
    {
      "epoch": 0.7107320540156361,
      "grad_norm": 4.797882080078125,
      "learning_rate": 4.239704329461457e-05,
      "loss": 0.3836,
      "step": 250
    },
    {
      "epoch": 0.7391613361762616,
      "grad_norm": 7.595874309539795,
      "learning_rate": 4.186906019007392e-05,
      "loss": 0.4172,
      "step": 260
    },
    {
      "epoch": 0.767590618336887,
      "grad_norm": 5.995832443237305,
      "learning_rate": 4.1341077085533265e-05,
      "loss": 0.3898,
      "step": 270
    },
    {
      "epoch": 0.7960199004975125,
      "grad_norm": 6.300724029541016,
      "learning_rate": 4.081309398099261e-05,
      "loss": 0.351,
      "step": 280
    },
    {
      "epoch": 0.8244491826581379,
      "grad_norm": 6.866366863250732,
      "learning_rate": 4.028511087645195e-05,
      "loss": 0.3872,
      "step": 290
    },
    {
      "epoch": 0.8528784648187633,
      "grad_norm": 7.383920669555664,
      "learning_rate": 3.97571277719113e-05,
      "loss": 0.3951,
      "step": 300
    },
    {
      "epoch": 0.8813077469793887,
      "grad_norm": 6.516490459442139,
      "learning_rate": 3.9229144667370646e-05,
      "loss": 0.4339,
      "step": 310
    },
    {
      "epoch": 0.9097370291400142,
      "grad_norm": 7.601709365844727,
      "learning_rate": 3.870116156282999e-05,
      "loss": 0.3799,
      "step": 320
    },
    {
      "epoch": 0.9381663113006397,
      "grad_norm": 6.864746570587158,
      "learning_rate": 3.817317845828934e-05,
      "loss": 0.3533,
      "step": 330
    },
    {
      "epoch": 0.9665955934612651,
      "grad_norm": 7.082674980163574,
      "learning_rate": 3.764519535374868e-05,
      "loss": 0.3558,
      "step": 340
    },
    {
      "epoch": 0.9950248756218906,
      "grad_norm": 7.578101634979248,
      "learning_rate": 3.711721224920803e-05,
      "loss": 0.3814,
      "step": 350
    },
    {
      "epoch": 0.997867803837953,
      "eval_accuracy": 0.8838,
      "eval_loss": 0.35002440214157104,
      "eval_runtime": 77.9571,
      "eval_samples_per_second": 64.138,
      "eval_steps_per_second": 2.014,
      "step": 351
    },
    {
      "epoch": 1.023454157782516,
      "grad_norm": 6.228023052215576,
      "learning_rate": 3.658922914466738e-05,
      "loss": 0.3687,
      "step": 360
    },
    {
      "epoch": 1.0518834399431414,
      "grad_norm": 5.145815849304199,
      "learning_rate": 3.6061246040126714e-05,
      "loss": 0.3351,
      "step": 370
    },
    {
      "epoch": 1.080312722103767,
      "grad_norm": 5.885372638702393,
      "learning_rate": 3.5533262935586064e-05,
      "loss": 0.3126,
      "step": 380
    },
    {
      "epoch": 1.1087420042643923,
      "grad_norm": 8.223782539367676,
      "learning_rate": 3.500527983104541e-05,
      "loss": 0.2893,
      "step": 390
    },
    {
      "epoch": 1.1371712864250179,
      "grad_norm": 5.879673004150391,
      "learning_rate": 3.447729672650475e-05,
      "loss": 0.3841,
      "step": 400
    },
    {
      "epoch": 1.1656005685856432,
      "grad_norm": 7.912990570068359,
      "learning_rate": 3.3949313621964095e-05,
      "loss": 0.3618,
      "step": 410
    },
    {
      "epoch": 1.1940298507462686,
      "grad_norm": 7.397537708282471,
      "learning_rate": 3.3421330517423445e-05,
      "loss": 0.3303,
      "step": 420
    },
    {
      "epoch": 1.2224591329068941,
      "grad_norm": 6.526395797729492,
      "learning_rate": 3.289334741288279e-05,
      "loss": 0.3611,
      "step": 430
    },
    {
      "epoch": 1.2508884150675195,
      "grad_norm": 6.138469696044922,
      "learning_rate": 3.236536430834213e-05,
      "loss": 0.3382,
      "step": 440
    },
    {
      "epoch": 1.279317697228145,
      "grad_norm": 9.07867431640625,
      "learning_rate": 3.183738120380148e-05,
      "loss": 0.3566,
      "step": 450
    },
    {
      "epoch": 1.3077469793887704,
      "grad_norm": 6.109983444213867,
      "learning_rate": 3.130939809926082e-05,
      "loss": 0.3187,
      "step": 460
    },
    {
      "epoch": 1.336176261549396,
      "grad_norm": 6.44178581237793,
      "learning_rate": 3.078141499472017e-05,
      "loss": 0.3115,
      "step": 470
    },
    {
      "epoch": 1.3646055437100213,
      "grad_norm": 7.65872049331665,
      "learning_rate": 3.0253431890179517e-05,
      "loss": 0.3082,
      "step": 480
    },
    {
      "epoch": 1.3930348258706466,
      "grad_norm": 8.232040405273438,
      "learning_rate": 2.972544878563886e-05,
      "loss": 0.3293,
      "step": 490
    },
    {
      "epoch": 1.4214641080312722,
      "grad_norm": 5.998810291290283,
      "learning_rate": 2.9197465681098207e-05,
      "loss": 0.3737,
      "step": 500
    },
    {
      "epoch": 1.4498933901918978,
      "grad_norm": 7.02003288269043,
      "learning_rate": 2.8669482576557548e-05,
      "loss": 0.3748,
      "step": 510
    },
    {
      "epoch": 1.4783226723525231,
      "grad_norm": 5.83160400390625,
      "learning_rate": 2.8141499472016898e-05,
      "loss": 0.3263,
      "step": 520
    },
    {
      "epoch": 1.5067519545131485,
      "grad_norm": 6.051623344421387,
      "learning_rate": 2.7613516367476245e-05,
      "loss": 0.304,
      "step": 530
    },
    {
      "epoch": 1.535181236673774,
      "grad_norm": 5.6155877113342285,
      "learning_rate": 2.7085533262935585e-05,
      "loss": 0.2972,
      "step": 540
    },
    {
      "epoch": 1.5636105188343994,
      "grad_norm": 7.925964832305908,
      "learning_rate": 2.6557550158394935e-05,
      "loss": 0.3265,
      "step": 550
    },
    {
      "epoch": 1.5920398009950247,
      "grad_norm": 6.579248905181885,
      "learning_rate": 2.6029567053854276e-05,
      "loss": 0.3129,
      "step": 560
    },
    {
      "epoch": 1.6204690831556503,
      "grad_norm": 6.650197982788086,
      "learning_rate": 2.5501583949313622e-05,
      "loss": 0.3005,
      "step": 570
    },
    {
      "epoch": 1.6488983653162759,
      "grad_norm": 5.006053924560547,
      "learning_rate": 2.497360084477297e-05,
      "loss": 0.2965,
      "step": 580
    },
    {
      "epoch": 1.6773276474769012,
      "grad_norm": 5.139395236968994,
      "learning_rate": 2.4445617740232313e-05,
      "loss": 0.3101,
      "step": 590
    },
    {
      "epoch": 1.7057569296375266,
      "grad_norm": 5.386786937713623,
      "learning_rate": 2.391763463569166e-05,
      "loss": 0.3151,
      "step": 600
    },
    {
      "epoch": 1.7341862117981521,
      "grad_norm": 5.373327732086182,
      "learning_rate": 2.3389651531151003e-05,
      "loss": 0.279,
      "step": 610
    },
    {
      "epoch": 1.7626154939587777,
      "grad_norm": 7.425722599029541,
      "learning_rate": 2.286166842661035e-05,
      "loss": 0.2881,
      "step": 620
    },
    {
      "epoch": 1.7910447761194028,
      "grad_norm": 5.681941509246826,
      "learning_rate": 2.2333685322069694e-05,
      "loss": 0.2912,
      "step": 630
    },
    {
      "epoch": 1.8194740582800284,
      "grad_norm": 6.3065266609191895,
      "learning_rate": 2.180570221752904e-05,
      "loss": 0.2947,
      "step": 640
    },
    {
      "epoch": 1.847903340440654,
      "grad_norm": 7.426571369171143,
      "learning_rate": 2.1277719112988384e-05,
      "loss": 0.2655,
      "step": 650
    },
    {
      "epoch": 1.8763326226012793,
      "grad_norm": 5.058150768280029,
      "learning_rate": 2.074973600844773e-05,
      "loss": 0.2874,
      "step": 660
    },
    {
      "epoch": 1.9047619047619047,
      "grad_norm": 3.7765002250671387,
      "learning_rate": 2.0221752903907075e-05,
      "loss": 0.2908,
      "step": 670
    },
    {
      "epoch": 1.9331911869225302,
      "grad_norm": 7.601891040802002,
      "learning_rate": 1.9693769799366422e-05,
      "loss": 0.2994,
      "step": 680
    },
    {
      "epoch": 1.9616204690831558,
      "grad_norm": 5.20227575302124,
      "learning_rate": 1.9165786694825765e-05,
      "loss": 0.253,
      "step": 690
    },
    {
      "epoch": 1.9900497512437811,
      "grad_norm": 4.2544426918029785,
      "learning_rate": 1.863780359028511e-05,
      "loss": 0.3251,
      "step": 700
    },
    {
      "epoch": 1.9985785358919688,
      "eval_accuracy": 0.9042,
      "eval_loss": 0.27974340319633484,
      "eval_runtime": 76.4828,
      "eval_samples_per_second": 65.374,
      "eval_steps_per_second": 2.053,
      "step": 703
    },
    {
      "epoch": 2.0184790334044065,
      "grad_norm": 5.491783618927002,
      "learning_rate": 1.810982048574446e-05,
      "loss": 0.2544,
      "step": 710
    },
    {
      "epoch": 2.046908315565032,
      "grad_norm": 5.285998821258545,
      "learning_rate": 1.7581837381203803e-05,
      "loss": 0.2764,
      "step": 720
    },
    {
      "epoch": 2.0753375977256576,
      "grad_norm": 5.578101634979248,
      "learning_rate": 1.7053854276663146e-05,
      "loss": 0.3035,
      "step": 730
    },
    {
      "epoch": 2.1037668798862827,
      "grad_norm": 5.175171375274658,
      "learning_rate": 1.6525871172122493e-05,
      "loss": 0.2497,
      "step": 740
    },
    {
      "epoch": 2.1321961620469083,
      "grad_norm": 6.48363733291626,
      "learning_rate": 1.5997888067581837e-05,
      "loss": 0.2896,
      "step": 750
    },
    {
      "epoch": 2.160625444207534,
      "grad_norm": 6.3542351722717285,
      "learning_rate": 1.5469904963041184e-05,
      "loss": 0.267,
      "step": 760
    },
    {
      "epoch": 2.189054726368159,
      "grad_norm": 7.133031368255615,
      "learning_rate": 1.4941921858500529e-05,
      "loss": 0.2719,
      "step": 770
    },
    {
      "epoch": 2.2174840085287846,
      "grad_norm": 6.279868125915527,
      "learning_rate": 1.4413938753959874e-05,
      "loss": 0.2712,
      "step": 780
    },
    {
      "epoch": 2.24591329068941,
      "grad_norm": 7.219544410705566,
      "learning_rate": 1.388595564941922e-05,
      "loss": 0.2266,
      "step": 790
    },
    {
      "epoch": 2.2743425728500357,
      "grad_norm": 4.286072731018066,
      "learning_rate": 1.3357972544878563e-05,
      "loss": 0.2469,
      "step": 800
    },
    {
      "epoch": 2.302771855010661,
      "grad_norm": 5.2917609214782715,
      "learning_rate": 1.2829989440337912e-05,
      "loss": 0.2824,
      "step": 810
    },
    {
      "epoch": 2.3312011371712864,
      "grad_norm": 6.390634536743164,
      "learning_rate": 1.2302006335797255e-05,
      "loss": 0.3072,
      "step": 820
    },
    {
      "epoch": 2.359630419331912,
      "grad_norm": 6.700084686279297,
      "learning_rate": 1.17740232312566e-05,
      "loss": 0.2336,
      "step": 830
    },
    {
      "epoch": 2.388059701492537,
      "grad_norm": 7.671480655670166,
      "learning_rate": 1.1246040126715946e-05,
      "loss": 0.2622,
      "step": 840
    },
    {
      "epoch": 2.4164889836531627,
      "grad_norm": 6.457718849182129,
      "learning_rate": 1.0718057022175291e-05,
      "loss": 0.2437,
      "step": 850
    },
    {
      "epoch": 2.4449182658137882,
      "grad_norm": 5.916180610656738,
      "learning_rate": 1.0190073917634636e-05,
      "loss": 0.2584,
      "step": 860
    },
    {
      "epoch": 2.473347547974414,
      "grad_norm": 5.509181976318359,
      "learning_rate": 9.662090813093982e-06,
      "loss": 0.2525,
      "step": 870
    },
    {
      "epoch": 2.501776830135039,
      "grad_norm": 5.262068748474121,
      "learning_rate": 9.134107708553327e-06,
      "loss": 0.2278,
      "step": 880
    },
    {
      "epoch": 2.5302061122956645,
      "grad_norm": 5.783515930175781,
      "learning_rate": 8.606124604012672e-06,
      "loss": 0.2466,
      "step": 890
    },
    {
      "epoch": 2.55863539445629,
      "grad_norm": 5.991846084594727,
      "learning_rate": 8.078141499472017e-06,
      "loss": 0.2186,
      "step": 900
    },
    {
      "epoch": 2.587064676616915,
      "grad_norm": 6.3764777183532715,
      "learning_rate": 7.5501583949313625e-06,
      "loss": 0.2636,
      "step": 910
    },
    {
      "epoch": 2.6154939587775408,
      "grad_norm": 7.736949920654297,
      "learning_rate": 7.022175290390708e-06,
      "loss": 0.251,
      "step": 920
    },
    {
      "epoch": 2.6439232409381663,
      "grad_norm": 4.5385332107543945,
      "learning_rate": 6.494192185850054e-06,
      "loss": 0.2642,
      "step": 930
    },
    {
      "epoch": 2.672352523098792,
      "grad_norm": 4.693294525146484,
      "learning_rate": 5.966209081309398e-06,
      "loss": 0.2302,
      "step": 940
    },
    {
      "epoch": 2.7007818052594175,
      "grad_norm": 6.726281642913818,
      "learning_rate": 5.438225976768744e-06,
      "loss": 0.2356,
      "step": 950
    },
    {
      "epoch": 2.7292110874200426,
      "grad_norm": 6.241092205047607,
      "learning_rate": 4.910242872228089e-06,
      "loss": 0.2105,
      "step": 960
    },
    {
      "epoch": 2.757640369580668,
      "grad_norm": 4.900135040283203,
      "learning_rate": 4.382259767687434e-06,
      "loss": 0.2375,
      "step": 970
    },
    {
      "epoch": 2.7860696517412933,
      "grad_norm": 6.798654556274414,
      "learning_rate": 3.854276663146779e-06,
      "loss": 0.2401,
      "step": 980
    },
    {
      "epoch": 2.814498933901919,
      "grad_norm": 4.920904159545898,
      "learning_rate": 3.326293558606125e-06,
      "loss": 0.2181,
      "step": 990
    },
    {
      "epoch": 2.8429282160625444,
      "grad_norm": 6.352721691131592,
      "learning_rate": 2.79831045406547e-06,
      "loss": 0.2228,
      "step": 1000
    },
    {
      "epoch": 2.87135749822317,
      "grad_norm": 6.661321640014648,
      "learning_rate": 2.2703273495248154e-06,
      "loss": 0.2012,
      "step": 1010
    },
    {
      "epoch": 2.8997867803837956,
      "grad_norm": 7.170318603515625,
      "learning_rate": 1.7423442449841606e-06,
      "loss": 0.2292,
      "step": 1020
    },
    {
      "epoch": 2.9282160625444207,
      "grad_norm": 4.110240459442139,
      "learning_rate": 1.2143611404435059e-06,
      "loss": 0.1975,
      "step": 1030
    },
    {
      "epoch": 2.9566453447050463,
      "grad_norm": 3.190889358520508,
      "learning_rate": 6.863780359028511e-07,
      "loss": 0.2381,
      "step": 1040
    },
    {
      "epoch": 2.9850746268656714,
      "grad_norm": 7.6400532722473145,
      "learning_rate": 1.5839493136219642e-07,
      "loss": 0.2368,
      "step": 1050
    },
    {
      "epoch": 2.9936034115138592,
      "eval_accuracy": 0.9168,
      "eval_loss": 0.24364404380321503,
      "eval_runtime": 76.8968,
      "eval_samples_per_second": 65.022,
      "eval_steps_per_second": 2.042,
      "step": 1053
    },
    {
      "epoch": 2.9936034115138592,
      "step": 1053,
      "total_flos": 1.043724675699553e+19,
      "train_loss": 0.41841287151021495,
      "train_runtime": 6863.6795,
      "train_samples_per_second": 19.669,
      "train_steps_per_second": 0.153
    }
  ],
  "logging_steps": 10,
  "max_steps": 1053,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.043724675699553e+19,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}