{
  "best_global_step": 13660,
  "best_metric": 0.5841874084919473,
  "best_model_checkpoint": "./saved_models/starencoder/checkpoint-13660",
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 13660,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.018301610541727673,
      "grad_norm": 6.326382637023926,
      "learning_rate": 1.992825768667643e-05,
      "loss": 0.7093,
      "step": 50
    },
    {
      "epoch": 0.036603221083455345,
      "grad_norm": 1.9646369218826294,
      "learning_rate": 1.9855051244509516e-05,
      "loss": 0.6877,
      "step": 100
    },
    {
      "epoch": 0.05490483162518302,
      "grad_norm": 3.757148027420044,
      "learning_rate": 1.9781844802342606e-05,
      "loss": 0.7064,
      "step": 150
    },
    {
      "epoch": 0.07320644216691069,
      "grad_norm": 2.2845559120178223,
      "learning_rate": 1.97086383601757e-05,
      "loss": 0.6941,
      "step": 200
    },
    {
      "epoch": 0.09150805270863836,
      "grad_norm": 3.454932689666748,
      "learning_rate": 1.9635431918008785e-05,
      "loss": 0.697,
      "step": 250
    },
    {
      "epoch": 0.10980966325036604,
      "grad_norm": 2.188014268875122,
      "learning_rate": 1.9562225475841875e-05,
      "loss": 0.6939,
      "step": 300
    },
    {
      "epoch": 0.1281112737920937,
      "grad_norm": 3.7213497161865234,
      "learning_rate": 1.9489019033674965e-05,
      "loss": 0.7021,
      "step": 350
    },
    {
      "epoch": 0.14641288433382138,
      "grad_norm": 4.130824565887451,
      "learning_rate": 1.9415812591508055e-05,
      "loss": 0.6829,
      "step": 400
    },
    {
      "epoch": 0.16471449487554904,
      "grad_norm": 2.1340243816375732,
      "learning_rate": 1.9342606149341144e-05,
      "loss": 0.7163,
      "step": 450
    },
    {
      "epoch": 0.18301610541727673,
      "grad_norm": 2.3844826221466064,
      "learning_rate": 1.926939970717423e-05,
      "loss": 0.6817,
      "step": 500
    },
    {
      "epoch": 0.20131771595900438,
      "grad_norm": 5.518948078155518,
      "learning_rate": 1.919619326500732e-05,
      "loss": 0.6755,
      "step": 550
    },
    {
      "epoch": 0.21961932650073207,
      "grad_norm": 7.737420082092285,
      "learning_rate": 1.9122986822840414e-05,
      "loss": 0.6717,
      "step": 600
    },
    {
      "epoch": 0.23792093704245973,
      "grad_norm": 4.545921802520752,
      "learning_rate": 1.90497803806735e-05,
      "loss": 0.6749,
      "step": 650
    },
    {
      "epoch": 0.2562225475841874,
      "grad_norm": 2.528022289276123,
      "learning_rate": 1.897657393850659e-05,
      "loss": 0.6701,
      "step": 700
    },
    {
      "epoch": 0.2745241581259151,
      "grad_norm": 2.498490571975708,
      "learning_rate": 1.890336749633968e-05,
      "loss": 0.6808,
      "step": 750
    },
    {
      "epoch": 0.29282576866764276,
      "grad_norm": 2.387111186981201,
      "learning_rate": 1.883016105417277e-05,
      "loss": 0.6925,
      "step": 800
    },
    {
      "epoch": 0.31112737920937045,
      "grad_norm": 4.5703253746032715,
      "learning_rate": 1.875695461200586e-05,
      "loss": 0.71,
      "step": 850
    },
    {
      "epoch": 0.3294289897510981,
      "grad_norm": 2.2322046756744385,
      "learning_rate": 1.8683748169838946e-05,
      "loss": 0.6978,
      "step": 900
    },
    {
      "epoch": 0.34773060029282576,
      "grad_norm": 3.186796188354492,
      "learning_rate": 1.8610541727672035e-05,
      "loss": 0.6999,
      "step": 950
    },
    {
      "epoch": 0.36603221083455345,
      "grad_norm": 2.268221855163574,
      "learning_rate": 1.853733528550513e-05,
      "loss": 0.7027,
      "step": 1000
    },
    {
      "epoch": 0.38433382137628114,
      "grad_norm": 3.345116138458252,
      "learning_rate": 1.8464128843338215e-05,
      "loss": 0.687,
      "step": 1050
    },
    {
      "epoch": 0.40263543191800877,
      "grad_norm": 3.4372918605804443,
      "learning_rate": 1.8390922401171305e-05,
      "loss": 0.6892,
      "step": 1100
    },
    {
      "epoch": 0.42093704245973645,
      "grad_norm": 4.888953685760498,
      "learning_rate": 1.8317715959004394e-05,
      "loss": 0.688,
      "step": 1150
    },
    {
      "epoch": 0.43923865300146414,
      "grad_norm": 3.7842814922332764,
      "learning_rate": 1.8244509516837484e-05,
      "loss": 0.6893,
      "step": 1200
    },
    {
      "epoch": 0.4575402635431918,
      "grad_norm": 2.004110336303711,
      "learning_rate": 1.817130307467057e-05,
      "loss": 0.6965,
      "step": 1250
    },
    {
      "epoch": 0.47584187408491946,
      "grad_norm": 3.093764543533325,
      "learning_rate": 1.809809663250366e-05,
      "loss": 0.6946,
      "step": 1300
    },
    {
      "epoch": 0.49414348462664714,
      "grad_norm": 4.060361385345459,
      "learning_rate": 1.802489019033675e-05,
      "loss": 0.6926,
      "step": 1350
    },
    {
      "epoch": 0.5124450951683748,
      "grad_norm": 6.572564601898193,
      "learning_rate": 1.795168374816984e-05,
      "loss": 0.6926,
      "step": 1400
    },
    {
      "epoch": 0.5307467057101025,
      "grad_norm": 5.1151251792907715,
      "learning_rate": 1.787847730600293e-05,
      "loss": 0.6985,
      "step": 1450
    },
    {
      "epoch": 0.5490483162518301,
      "grad_norm": 7.343419075012207,
      "learning_rate": 1.780527086383602e-05,
      "loss": 0.6954,
      "step": 1500
    },
    {
      "epoch": 0.5673499267935578,
      "grad_norm": 4.611226558685303,
      "learning_rate": 1.773206442166911e-05,
      "loss": 0.683,
      "step": 1550
    },
    {
      "epoch": 0.5856515373352855,
      "grad_norm": 1.5302119255065918,
      "learning_rate": 1.76588579795022e-05,
      "loss": 0.6915,
      "step": 1600
    },
    {
      "epoch": 0.6039531478770132,
      "grad_norm": 1.781995415687561,
      "learning_rate": 1.7585651537335285e-05,
      "loss": 0.7022,
      "step": 1650
    },
    {
      "epoch": 0.6222547584187409,
      "grad_norm": 1.6124660968780518,
      "learning_rate": 1.7512445095168375e-05,
      "loss": 0.6913,
      "step": 1700
    },
    {
      "epoch": 0.6405563689604685,
      "grad_norm": 4.99559211730957,
      "learning_rate": 1.7439238653001465e-05,
      "loss": 0.6996,
      "step": 1750
    },
    {
      "epoch": 0.6588579795021962,
      "grad_norm": 9.72033405303955,
      "learning_rate": 1.7366032210834554e-05,
      "loss": 0.7121,
      "step": 1800
    },
    {
      "epoch": 0.6771595900439239,
      "grad_norm": 2.2888541221618652,
      "learning_rate": 1.7292825768667644e-05,
      "loss": 0.6852,
      "step": 1850
    },
    {
      "epoch": 0.6954612005856515,
      "grad_norm": 7.631497383117676,
      "learning_rate": 1.7219619326500734e-05,
      "loss": 0.6971,
      "step": 1900
    },
    {
      "epoch": 0.7137628111273792,
      "grad_norm": 1.9625825881958008,
      "learning_rate": 1.7146412884333824e-05,
      "loss": 0.6881,
      "step": 1950
    },
    {
      "epoch": 0.7320644216691069,
      "grad_norm": 10.556265830993652,
      "learning_rate": 1.7073206442166913e-05,
      "loss": 0.6978,
      "step": 2000
    },
    {
      "epoch": 0.7503660322108345,
      "grad_norm": 5.823647499084473,
      "learning_rate": 1.7e-05,
      "loss": 0.6974,
      "step": 2050
    },
    {
      "epoch": 0.7686676427525623,
      "grad_norm": 5.508345603942871,
      "learning_rate": 1.692679355783309e-05,
      "loss": 0.7045,
      "step": 2100
    },
    {
      "epoch": 0.7869692532942899,
      "grad_norm": 7.553188800811768,
      "learning_rate": 1.685358711566618e-05,
      "loss": 0.7011,
      "step": 2150
    },
    {
      "epoch": 0.8052708638360175,
      "grad_norm": 3.6769418716430664,
      "learning_rate": 1.678038067349927e-05,
      "loss": 0.6934,
      "step": 2200
    },
    {
      "epoch": 0.8235724743777453,
      "grad_norm": 2.2349374294281006,
      "learning_rate": 1.670717423133236e-05,
      "loss": 0.6951,
      "step": 2250
    },
    {
      "epoch": 0.8418740849194729,
      "grad_norm": 1.6531909704208374,
      "learning_rate": 1.663396778916545e-05,
      "loss": 0.7001,
      "step": 2300
    },
    {
      "epoch": 0.8601756954612005,
      "grad_norm": 4.906118869781494,
      "learning_rate": 1.656076134699854e-05,
      "loss": 0.6966,
      "step": 2350
    },
    {
      "epoch": 0.8784773060029283,
      "grad_norm": 7.1544013023376465,
      "learning_rate": 1.6487554904831625e-05,
      "loss": 0.6882,
      "step": 2400
    },
    {
      "epoch": 0.8967789165446559,
      "grad_norm": 1.9310983419418335,
      "learning_rate": 1.6414348462664715e-05,
      "loss": 0.7183,
      "step": 2450
    },
    {
      "epoch": 0.9150805270863837,
      "grad_norm": 5.646407127380371,
      "learning_rate": 1.6341142020497804e-05,
      "loss": 0.6884,
      "step": 2500
    },
    {
      "epoch": 0.9333821376281113,
      "grad_norm": 1.564105749130249,
      "learning_rate": 1.6267935578330894e-05,
      "loss": 0.6979,
      "step": 2550
    },
    {
      "epoch": 0.9516837481698389,
      "grad_norm": 2.552931308746338,
      "learning_rate": 1.6194729136163984e-05,
      "loss": 0.6921,
      "step": 2600
    },
    {
      "epoch": 0.9699853587115667,
      "grad_norm": 4.336668968200684,
      "learning_rate": 1.6121522693997074e-05,
      "loss": 0.6956,
      "step": 2650
    },
    {
      "epoch": 0.9882869692532943,
      "grad_norm": 1.3156540393829346,
      "learning_rate": 1.6048316251830163e-05,
      "loss": 0.6962,
      "step": 2700
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5655197657393851,
      "eval_loss": 0.6866211295127869,
      "eval_runtime": 30.1911,
      "eval_samples_per_second": 90.49,
      "eval_steps_per_second": 5.664,
      "step": 2732
    },
    {
      "epoch": 1.006588579795022,
      "grad_norm": 4.257384777069092,
      "learning_rate": 1.5975109809663253e-05,
      "loss": 0.6824,
      "step": 2750
    },
    {
      "epoch": 1.0248901903367496,
      "grad_norm": 4.433679580688477,
      "learning_rate": 1.590190336749634e-05,
      "loss": 0.6886,
      "step": 2800
    },
    {
      "epoch": 1.0431918008784773,
      "grad_norm": 2.416440010070801,
      "learning_rate": 1.582869692532943e-05,
      "loss": 0.7093,
      "step": 2850
    },
    {
      "epoch": 1.061493411420205,
      "grad_norm": 3.4436511993408203,
      "learning_rate": 1.575549048316252e-05,
      "loss": 0.6854,
      "step": 2900
    },
    {
      "epoch": 1.0797950219619326,
      "grad_norm": 3.391012668609619,
      "learning_rate": 1.568228404099561e-05,
      "loss": 0.6916,
      "step": 2950
    },
    {
      "epoch": 1.0980966325036603,
      "grad_norm": 2.20050311088562,
      "learning_rate": 1.56090775988287e-05,
      "loss": 0.6928,
      "step": 3000
    },
    {
      "epoch": 1.116398243045388,
      "grad_norm": 3.6996657848358154,
      "learning_rate": 1.5535871156661788e-05,
      "loss": 0.6978,
      "step": 3050
    },
    {
      "epoch": 1.1346998535871156,
      "grad_norm": 5.966168403625488,
      "learning_rate": 1.5462664714494878e-05,
      "loss": 0.6835,
      "step": 3100
    },
    {
      "epoch": 1.1530014641288433,
      "grad_norm": 7.096467971801758,
      "learning_rate": 1.5389458272327968e-05,
      "loss": 0.7008,
      "step": 3150
    },
    {
      "epoch": 1.171303074670571,
      "grad_norm": 2.978032350540161,
      "learning_rate": 1.5316251830161054e-05,
      "loss": 0.6923,
      "step": 3200
    },
    {
      "epoch": 1.1896046852122986,
      "grad_norm": 2.5770692825317383,
      "learning_rate": 1.5243045387994144e-05,
      "loss": 0.6988,
      "step": 3250
    },
    {
      "epoch": 1.2079062957540263,
      "grad_norm": 1.8154376745224,
      "learning_rate": 1.5169838945827234e-05,
      "loss": 0.693,
      "step": 3300
    },
    {
      "epoch": 1.226207906295754,
      "grad_norm": 5.659831523895264,
      "learning_rate": 1.5096632503660322e-05,
      "loss": 0.6951,
      "step": 3350
    },
    {
      "epoch": 1.2445095168374818,
      "grad_norm": 3.904559373855591,
      "learning_rate": 1.5023426061493413e-05,
      "loss": 0.6901,
      "step": 3400
    },
    {
      "epoch": 1.2628111273792093,
      "grad_norm": 3.0381088256835938,
      "learning_rate": 1.4950219619326503e-05,
      "loss": 0.6871,
      "step": 3450
    },
    {
      "epoch": 1.281112737920937,
      "grad_norm": 3.667318344116211,
      "learning_rate": 1.4877013177159591e-05,
      "loss": 0.6912,
      "step": 3500
    },
    {
      "epoch": 1.2994143484626648,
      "grad_norm": 6.238102912902832,
      "learning_rate": 1.480380673499268e-05,
      "loss": 0.6983,
      "step": 3550
    },
    {
      "epoch": 1.3177159590043923,
      "grad_norm": 4.712520599365234,
      "learning_rate": 1.4730600292825769e-05,
      "loss": 0.6898,
      "step": 3600
    },
    {
      "epoch": 1.33601756954612,
      "grad_norm": 4.511780261993408,
      "learning_rate": 1.4657393850658859e-05,
      "loss": 0.6972,
      "step": 3650
    },
    {
      "epoch": 1.3543191800878478,
      "grad_norm": 2.4388058185577393,
      "learning_rate": 1.4584187408491948e-05,
      "loss": 0.6962,
      "step": 3700
    },
    {
      "epoch": 1.3726207906295755,
      "grad_norm": 2.305985927581787,
      "learning_rate": 1.4510980966325036e-05,
      "loss": 0.696,
      "step": 3750
    },
    {
      "epoch": 1.390922401171303,
      "grad_norm": 5.038622856140137,
      "learning_rate": 1.4437774524158128e-05,
      "loss": 0.6945,
      "step": 3800
    },
    {
      "epoch": 1.4092240117130308,
      "grad_norm": 1.7405680418014526,
      "learning_rate": 1.4364568081991218e-05,
      "loss": 0.6975,
      "step": 3850
    },
    {
      "epoch": 1.4275256222547585,
      "grad_norm": 2.3643574714660645,
      "learning_rate": 1.4291361639824306e-05,
      "loss": 0.6961,
      "step": 3900
    },
    {
      "epoch": 1.445827232796486,
      "grad_norm": 5.455554008483887,
      "learning_rate": 1.4218155197657395e-05,
      "loss": 0.6974,
      "step": 3950
    },
    {
      "epoch": 1.4641288433382138,
      "grad_norm": 1.3421140909194946,
      "learning_rate": 1.4144948755490484e-05,
      "loss": 0.6929,
      "step": 4000
    },
    {
      "epoch": 1.4824304538799415,
      "grad_norm": 2.995839834213257,
      "learning_rate": 1.4071742313323573e-05,
      "loss": 0.6974,
      "step": 4050
    },
    {
      "epoch": 1.500732064421669,
      "grad_norm": 4.329990386962891,
      "learning_rate": 1.3998535871156661e-05,
      "loss": 0.6968,
      "step": 4100
    },
    {
      "epoch": 1.5190336749633968,
      "grad_norm": 2.4263505935668945,
      "learning_rate": 1.3925329428989751e-05,
      "loss": 0.694,
      "step": 4150
    },
    {
      "epoch": 1.5373352855051245,
      "grad_norm": 5.850309371948242,
      "learning_rate": 1.3852122986822843e-05,
      "loss": 0.6975,
      "step": 4200
    },
    {
      "epoch": 1.555636896046852,
      "grad_norm": 1.353864073753357,
      "learning_rate": 1.377891654465593e-05,
      "loss": 0.7005,
      "step": 4250
    },
    {
      "epoch": 1.5739385065885798,
      "grad_norm": 2.510244369506836,
      "learning_rate": 1.370571010248902e-05,
      "loss": 0.6906,
      "step": 4300
    },
    {
      "epoch": 1.5922401171303076,
      "grad_norm": 3.187347888946533,
      "learning_rate": 1.363250366032211e-05,
      "loss": 0.6902,
      "step": 4350
    },
    {
      "epoch": 1.610541727672035,
      "grad_norm": 2.019270420074463,
      "learning_rate": 1.3559297218155198e-05,
      "loss": 0.6959,
      "step": 4400
    },
    {
      "epoch": 1.6288433382137628,
      "grad_norm": 3.3284361362457275,
      "learning_rate": 1.3486090775988288e-05,
      "loss": 0.6901,
      "step": 4450
    },
    {
      "epoch": 1.6471449487554906,
      "grad_norm": 2.3892571926116943,
      "learning_rate": 1.3412884333821376e-05,
      "loss": 0.6831,
      "step": 4500
    },
    {
      "epoch": 1.665446559297218,
      "grad_norm": 2.138828992843628,
      "learning_rate": 1.3339677891654466e-05,
      "loss": 0.6961,
      "step": 4550
    },
    {
      "epoch": 1.6837481698389458,
      "grad_norm": 2.1902518272399902,
      "learning_rate": 1.3266471449487557e-05,
      "loss": 0.7011,
      "step": 4600
    },
    {
      "epoch": 1.7020497803806736,
      "grad_norm": 1.9349195957183838,
      "learning_rate": 1.3193265007320645e-05,
      "loss": 0.6964,
      "step": 4650
    },
    {
      "epoch": 1.720351390922401,
      "grad_norm": 2.414310932159424,
      "learning_rate": 1.3120058565153735e-05,
      "loss": 0.6993,
      "step": 4700
    },
    {
      "epoch": 1.7386530014641288,
      "grad_norm": 2.0503060817718506,
      "learning_rate": 1.3046852122986823e-05,
      "loss": 0.6945,
      "step": 4750
    },
    {
      "epoch": 1.7569546120058566,
      "grad_norm": 3.2496321201324463,
      "learning_rate": 1.2973645680819913e-05,
      "loss": 0.6984,
      "step": 4800
    },
    {
      "epoch": 1.775256222547584,
      "grad_norm": 4.175572395324707,
      "learning_rate": 1.2900439238653003e-05,
      "loss": 0.6945,
      "step": 4850
    },
    {
      "epoch": 1.7935578330893118,
      "grad_norm": 1.1053935289382935,
      "learning_rate": 1.282723279648609e-05,
      "loss": 0.6947,
      "step": 4900
    },
    {
      "epoch": 1.8118594436310396,
      "grad_norm": 4.932358741760254,
      "learning_rate": 1.2754026354319182e-05,
      "loss": 0.6893,
      "step": 4950
    },
    {
      "epoch": 1.830161054172767,
      "grad_norm": 1.429621934890747,
      "learning_rate": 1.2680819912152272e-05,
      "loss": 0.6711,
      "step": 5000
    },
    {
      "epoch": 1.8484626647144948,
      "grad_norm": 3.933276891708374,
      "learning_rate": 1.260761346998536e-05,
      "loss": 0.6942,
      "step": 5050
    },
    {
      "epoch": 1.8667642752562226,
      "grad_norm": 1.696254849433899,
      "learning_rate": 1.253440702781845e-05,
      "loss": 0.6922,
      "step": 5100
    },
    {
      "epoch": 1.88506588579795,
      "grad_norm": 2.4555442333221436,
      "learning_rate": 1.2461200585651538e-05,
      "loss": 0.6923,
      "step": 5150
    },
    {
      "epoch": 1.903367496339678,
      "grad_norm": 3.334120035171509,
      "learning_rate": 1.2387994143484628e-05,
      "loss": 0.6872,
      "step": 5200
    },
    {
      "epoch": 1.9216691068814056,
      "grad_norm": 1.637233018875122,
      "learning_rate": 1.2314787701317716e-05,
      "loss": 0.6832,
      "step": 5250
    },
    {
      "epoch": 1.939970717423133,
      "grad_norm": 1.511348009109497,
      "learning_rate": 1.2241581259150805e-05,
      "loss": 0.6892,
      "step": 5300
    },
    {
      "epoch": 1.958272327964861,
      "grad_norm": 3.0342915058135986,
      "learning_rate": 1.2168374816983897e-05,
      "loss": 0.6877,
      "step": 5350
    },
    {
      "epoch": 1.9765739385065886,
      "grad_norm": 4.36331033706665,
      "learning_rate": 1.2095168374816985e-05,
      "loss": 0.7022,
      "step": 5400
    },
    {
      "epoch": 1.994875549048316,
      "grad_norm": 1.6790521144866943,
      "learning_rate": 1.2021961932650075e-05,
      "loss": 0.689,
      "step": 5450
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.43448023426061494,
      "eval_loss": 0.6969389319419861,
      "eval_runtime": 9.7517,
      "eval_samples_per_second": 280.155,
      "eval_steps_per_second": 17.535,
      "step": 5464
    },
    {
      "epoch": 2.013177159590044,
      "grad_norm": 2.8888535499572754,
      "learning_rate": 1.1948755490483164e-05,
      "loss": 0.6928,
      "step": 5500
    },
    {
      "epoch": 2.0314787701317716,
      "grad_norm": 1.8324801921844482,
      "learning_rate": 1.1875549048316253e-05,
      "loss": 0.6899,
      "step": 5550
    },
    {
      "epoch": 2.049780380673499,
      "grad_norm": 1.6425199508666992,
      "learning_rate": 1.1802342606149342e-05,
      "loss": 0.6923,
      "step": 5600
    },
    {
      "epoch": 2.068081991215227,
      "grad_norm": 2.4402670860290527,
      "learning_rate": 1.172913616398243e-05,
      "loss": 0.6882,
      "step": 5650
    },
    {
      "epoch": 2.0863836017569546,
      "grad_norm": 1.4472464323043823,
      "learning_rate": 1.165592972181552e-05,
      "loss": 0.6899,
      "step": 5700
    },
    {
      "epoch": 2.104685212298682,
      "grad_norm": 2.157292127609253,
      "learning_rate": 1.1582723279648612e-05,
      "loss": 0.7026,
      "step": 5750
    },
    {
      "epoch": 2.12298682284041,
      "grad_norm": 3.735875129699707,
      "learning_rate": 1.15095168374817e-05,
      "loss": 0.6818,
      "step": 5800
    },
    {
      "epoch": 2.1412884333821376,
      "grad_norm": 3.0418055057525635,
      "learning_rate": 1.143631039531479e-05,
      "loss": 0.6909,
      "step": 5850
    },
    {
      "epoch": 2.159590043923865,
      "grad_norm": 1.1220214366912842,
      "learning_rate": 1.1363103953147877e-05,
      "loss": 0.6977,
      "step": 5900
    },
    {
      "epoch": 2.177891654465593,
      "grad_norm": 1.926895022392273,
      "learning_rate": 1.1289897510980967e-05,
      "loss": 0.6937,
      "step": 5950
    },
    {
      "epoch": 2.1961932650073206,
      "grad_norm": 1.1366084814071655,
      "learning_rate": 1.1216691068814057e-05,
      "loss": 0.6981,
      "step": 6000
    },
    {
      "epoch": 2.214494875549048,
      "grad_norm": 2.684913396835327,
      "learning_rate": 1.1143484626647145e-05,
      "loss": 0.6923,
      "step": 6050
    },
    {
      "epoch": 2.232796486090776,
      "grad_norm": 1.1763278245925903,
      "learning_rate": 1.1070278184480235e-05,
      "loss": 0.6884,
      "step": 6100
    },
    {
      "epoch": 2.2510980966325036,
      "grad_norm": 0.9922837615013123,
      "learning_rate": 1.0997071742313326e-05,
      "loss": 0.6972,
      "step": 6150
    },
    {
      "epoch": 2.269399707174231,
      "grad_norm": 1.7432096004486084,
      "learning_rate": 1.0923865300146414e-05,
      "loss": 0.6956,
      "step": 6200
    },
    {
      "epoch": 2.287701317715959,
      "grad_norm": 5.431227207183838,
      "learning_rate": 1.0850658857979504e-05,
      "loss": 0.6949,
      "step": 6250
    },
    {
      "epoch": 2.3060029282576866,
      "grad_norm": 2.5441296100616455,
      "learning_rate": 1.0777452415812592e-05,
      "loss": 0.6922,
      "step": 6300
    },
    {
      "epoch": 2.3243045387994146,
      "grad_norm": 2.507460832595825,
      "learning_rate": 1.0704245973645682e-05,
      "loss": 0.6776,
      "step": 6350
    },
    {
      "epoch": 2.342606149341142,
      "grad_norm": 4.670897006988525,
      "learning_rate": 1.063103953147877e-05,
      "loss": 0.6888,
      "step": 6400
    },
    {
      "epoch": 2.3609077598828696,
      "grad_norm": 5.1314167976379395,
      "learning_rate": 1.055783308931186e-05,
      "loss": 0.6966,
      "step": 6450
    },
    {
      "epoch": 2.379209370424597,
      "grad_norm": 2.4066569805145264,
      "learning_rate": 1.048462664714495e-05,
      "loss": 0.6983,
      "step": 6500
    },
    {
      "epoch": 2.397510980966325,
      "grad_norm": 3.4087448120117188,
      "learning_rate": 1.041142020497804e-05,
      "loss": 0.6973,
      "step": 6550
    },
    {
      "epoch": 2.4158125915080526,
      "grad_norm": 1.7125242948532104,
      "learning_rate": 1.0338213762811129e-05,
      "loss": 0.6948,
      "step": 6600
    },
    {
      "epoch": 2.4341142020497806,
      "grad_norm": 3.0995583534240723,
      "learning_rate": 1.0265007320644219e-05,
      "loss": 0.6973,
      "step": 6650
    },
    {
      "epoch": 2.452415812591508,
      "grad_norm": 4.187213897705078,
      "learning_rate": 1.0191800878477307e-05,
      "loss": 0.6839,
      "step": 6700
    },
    {
      "epoch": 2.4707174231332356,
      "grad_norm": 2.849050521850586,
      "learning_rate": 1.0118594436310397e-05,
      "loss": 0.6927,
      "step": 6750
    },
    {
      "epoch": 2.4890190336749636,
      "grad_norm": 3.120654821395874,
      "learning_rate": 1.0045387994143485e-05,
      "loss": 0.6887,
      "step": 6800
    },
    {
      "epoch": 2.507320644216691,
      "grad_norm": 3.0376622676849365,
      "learning_rate": 9.972181551976574e-06,
      "loss": 0.6786,
      "step": 6850
    },
    {
      "epoch": 2.5256222547584186,
      "grad_norm": 2.4937405586242676,
      "learning_rate": 9.898975109809664e-06,
      "loss": 0.7005,
      "step": 6900
    },
    {
      "epoch": 2.5439238653001466,
      "grad_norm": 1.2871942520141602,
      "learning_rate": 9.825768667642754e-06,
      "loss": 0.6933,
      "step": 6950
    },
    {
      "epoch": 2.562225475841874,
      "grad_norm": 1.9713603258132935,
      "learning_rate": 9.752562225475842e-06,
      "loss": 0.6986,
      "step": 7000
    },
    {
      "epoch": 2.5805270863836016,
      "grad_norm": 3.823519706726074,
      "learning_rate": 9.679355783308932e-06,
      "loss": 0.6926,
      "step": 7050
    },
    {
      "epoch": 2.5988286969253296,
      "grad_norm": 1.7224080562591553,
      "learning_rate": 9.606149341142022e-06,
      "loss": 0.6894,
      "step": 7100
    },
    {
      "epoch": 2.617130307467057,
      "grad_norm": 1.2425187826156616,
      "learning_rate": 9.532942898975111e-06,
      "loss": 0.6898,
      "step": 7150
    },
    {
      "epoch": 2.6354319180087846,
      "grad_norm": 3.008572816848755,
      "learning_rate": 9.4597364568082e-06,
      "loss": 0.6902,
      "step": 7200
    },
    {
      "epoch": 2.6537335285505126,
      "grad_norm": 2.5645570755004883,
      "learning_rate": 9.386530014641289e-06,
      "loss": 0.6828,
      "step": 7250
    },
    {
      "epoch": 2.67203513909224,
      "grad_norm": 1.7012056112289429,
      "learning_rate": 9.313323572474379e-06,
      "loss": 0.6969,
      "step": 7300
    },
    {
      "epoch": 2.6903367496339676,
      "grad_norm": 6.18578577041626,
      "learning_rate": 9.240117130307467e-06,
      "loss": 0.68,
      "step": 7350
    },
    {
      "epoch": 2.7086383601756956,
      "grad_norm": 2.9371464252471924,
      "learning_rate": 9.166910688140557e-06,
      "loss": 0.7052,
      "step": 7400
    },
    {
      "epoch": 2.726939970717423,
      "grad_norm": 1.306884527206421,
      "learning_rate": 9.093704245973646e-06,
      "loss": 0.6945,
      "step": 7450
    },
    {
      "epoch": 2.745241581259151,
      "grad_norm": 2.4502739906311035,
      "learning_rate": 9.020497803806736e-06,
      "loss": 0.6998,
      "step": 7500
    },
    {
      "epoch": 2.7635431918008786,
      "grad_norm": 3.148775100708008,
      "learning_rate": 8.947291361639824e-06,
      "loss": 0.6931,
      "step": 7550
    },
    {
      "epoch": 2.781844802342606,
      "grad_norm": 2.4222412109375,
      "learning_rate": 8.874084919472914e-06,
      "loss": 0.6939,
      "step": 7600
    },
    {
      "epoch": 2.8001464128843336,
      "grad_norm": 2.543954610824585,
      "learning_rate": 8.800878477306004e-06,
      "loss": 0.6934,
      "step": 7650
    },
    {
      "epoch": 2.8184480234260616,
      "grad_norm": 1.2464910745620728,
      "learning_rate": 8.727672035139094e-06,
      "loss": 0.6919,
      "step": 7700
    },
    {
      "epoch": 2.836749633967789,
      "grad_norm": 3.3286757469177246,
      "learning_rate": 8.654465592972182e-06,
      "loss": 0.6955,
      "step": 7750
    },
    {
      "epoch": 2.855051244509517,
      "grad_norm": 1.269946575164795,
      "learning_rate": 8.581259150805271e-06,
      "loss": 0.6904,
      "step": 7800
    },
    {
      "epoch": 2.8733528550512446,
      "grad_norm": 2.8248860836029053,
      "learning_rate": 8.508052708638361e-06,
      "loss": 0.6876,
      "step": 7850
    },
    {
      "epoch": 2.891654465592972,
      "grad_norm": 4.0137553215026855,
      "learning_rate": 8.434846266471451e-06,
      "loss": 0.6901,
      "step": 7900
    },
    {
      "epoch": 2.9099560761346996,
      "grad_norm": 4.750382900238037,
      "learning_rate": 8.361639824304539e-06,
      "loss": 0.6986,
      "step": 7950
    },
    {
      "epoch": 2.9282576866764276,
      "grad_norm": 3.539860963821411,
      "learning_rate": 8.288433382137629e-06,
      "loss": 0.6838,
      "step": 8000
    },
    {
      "epoch": 2.946559297218155,
      "grad_norm": 2.6545543670654297,
      "learning_rate": 8.215226939970719e-06,
      "loss": 0.6942,
      "step": 8050
    },
    {
      "epoch": 2.964860907759883,
      "grad_norm": 2.351973056793213,
      "learning_rate": 8.142020497803808e-06,
      "loss": 0.6991,
      "step": 8100
    },
    {
      "epoch": 2.9831625183016106,
      "grad_norm": 1.9938989877700806,
      "learning_rate": 8.068814055636896e-06,
      "loss": 0.6928,
      "step": 8150
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.5655197657393851,
      "eval_loss": 0.6848952770233154,
      "eval_runtime": 9.8156,
      "eval_samples_per_second": 278.332,
      "eval_steps_per_second": 17.421,
      "step": 8196
    },
    {
      "epoch": 3.001464128843338,
      "grad_norm": 3.9596476554870605,
      "learning_rate": 7.995607613469986e-06,
      "loss": 0.6903,
      "step": 8200
    },
    {
      "epoch": 3.019765739385066,
      "grad_norm": 4.190953254699707,
      "learning_rate": 7.922401171303076e-06,
      "loss": 0.6778,
      "step": 8250
    },
    {
      "epoch": 3.0380673499267936,
      "grad_norm": 3.9022440910339355,
      "learning_rate": 7.849194729136164e-06,
      "loss": 0.7014,
      "step": 8300
    },
    {
      "epoch": 3.056368960468521,
      "grad_norm": 3.035680055618286,
      "learning_rate": 7.775988286969254e-06,
      "loss": 0.6879,
      "step": 8350
    },
    {
      "epoch": 3.074670571010249,
      "grad_norm": 6.70428466796875,
      "learning_rate": 7.702781844802343e-06,
      "loss": 0.6871,
      "step": 8400
    },
    {
      "epoch": 3.0929721815519766,
      "grad_norm": 2.6576058864593506,
      "learning_rate": 7.629575402635433e-06,
      "loss": 0.6956,
      "step": 8450
    },
    {
      "epoch": 3.111273792093704,
      "grad_norm": 3.4227051734924316,
      "learning_rate": 7.556368960468522e-06,
      "loss": 0.6949,
      "step": 8500
    },
    {
      "epoch": 3.129575402635432,
      "grad_norm": 6.5497145652771,
      "learning_rate": 7.483162518301611e-06,
      "loss": 0.6882,
      "step": 8550
    },
    {
      "epoch": 3.1478770131771596,
      "grad_norm": 1.9692039489746094,
      "learning_rate": 7.4099560761347e-06,
      "loss": 0.6857,
      "step": 8600
    },
    {
      "epoch": 3.166178623718887,
      "grad_norm": 1.5063157081604004,
      "learning_rate": 7.3367496339677906e-06,
      "loss": 0.703,
      "step": 8650
    },
    {
      "epoch": 3.184480234260615,
      "grad_norm": 2.114001750946045,
      "learning_rate": 7.2635431918008795e-06,
      "loss": 0.6931,
      "step": 8700
    },
    {
      "epoch": 3.2027818448023426,
      "grad_norm": 4.059252738952637,
      "learning_rate": 7.190336749633968e-06,
      "loss": 0.7012,
      "step": 8750
    },
    {
      "epoch": 3.22108345534407,
      "grad_norm": 1.4372018575668335,
      "learning_rate": 7.117130307467057e-06,
      "loss": 0.6911,
      "step": 8800
    },
    {
      "epoch": 3.239385065885798,
      "grad_norm": 1.4479070901870728,
      "learning_rate": 7.043923865300147e-06,
      "loss": 0.6925,
      "step": 8850
    },
    {
      "epoch": 3.2576866764275256,
      "grad_norm": 1.4054046869277954,
      "learning_rate": 6.970717423133237e-06,
      "loss": 0.6948,
      "step": 8900
    },
    {
      "epoch": 3.275988286969253,
      "grad_norm": 2.7354447841644287,
      "learning_rate": 6.897510980966326e-06,
      "loss": 0.6931,
      "step": 8950
    },
    {
      "epoch": 3.294289897510981,
      "grad_norm": 6.485820293426514,
      "learning_rate": 6.824304538799415e-06,
      "loss": 0.6924,
      "step": 9000
    },
    {
      "epoch": 3.3125915080527086,
      "grad_norm": 2.2250216007232666,
      "learning_rate": 6.751098096632504e-06,
      "loss": 0.6955,
      "step": 9050
    },
    {
      "epoch": 3.330893118594436,
      "grad_norm": 3.7703425884246826,
      "learning_rate": 6.677891654465593e-06,
      "loss": 0.6978,
      "step": 9100
    },
    {
      "epoch": 3.349194729136164,
      "grad_norm": 1.3543486595153809,
      "learning_rate": 6.604685212298682e-06,
      "loss": 0.6851,
      "step": 9150
    },
    {
      "epoch": 3.3674963396778916,
      "grad_norm": 1.9245567321777344,
      "learning_rate": 6.531478770131772e-06,
      "loss": 0.6971,
      "step": 9200
    },
    {
      "epoch": 3.385797950219619,
      "grad_norm": 1.5131146907806396,
      "learning_rate": 6.458272327964862e-06,
      "loss": 0.6898,
      "step": 9250
    },
    {
      "epoch": 3.404099560761347,
      "grad_norm": 1.524381399154663,
      "learning_rate": 6.385065885797951e-06,
      "loss": 0.6881,
      "step": 9300
    },
    {
      "epoch": 3.4224011713030746,
      "grad_norm": 2.4183127880096436,
      "learning_rate": 6.3118594436310396e-06,
      "loss": 0.6957,
      "step": 9350
    },
    {
      "epoch": 3.440702781844802,
      "grad_norm": 6.259820461273193,
      "learning_rate": 6.2386530014641285e-06,
      "loss": 0.6945,
      "step": 9400
    },
    {
      "epoch": 3.45900439238653,
      "grad_norm": 4.742863655090332,
      "learning_rate": 6.165446559297219e-06,
      "loss": 0.6918,
      "step": 9450
    },
    {
      "epoch": 3.4773060029282576,
      "grad_norm": 3.144963026046753,
      "learning_rate": 6.092240117130308e-06,
      "loss": 0.6862,
      "step": 9500
    },
    {
      "epoch": 3.4956076134699856,
      "grad_norm": 2.9106500148773193,
      "learning_rate": 6.019033674963397e-06,
      "loss": 0.6889,
      "step": 9550
    },
    {
      "epoch": 3.513909224011713,
      "grad_norm": 6.509891510009766,
      "learning_rate": 5.945827232796486e-06,
      "loss": 0.6945,
      "step": 9600
    },
    {
      "epoch": 3.5322108345534406,
      "grad_norm": 2.27295184135437,
      "learning_rate": 5.8726207906295764e-06,
      "loss": 0.6943,
      "step": 9650
    },
    {
      "epoch": 3.550512445095168,
      "grad_norm": 7.141515731811523,
      "learning_rate": 5.799414348462665e-06,
      "loss": 0.6876,
      "step": 9700
    },
    {
      "epoch": 3.568814055636896,
      "grad_norm": 2.371335744857788,
      "learning_rate": 5.726207906295754e-06,
      "loss": 0.6988,
      "step": 9750
    },
    {
      "epoch": 3.5871156661786237,
      "grad_norm": 1.0689224004745483,
      "learning_rate": 5.653001464128843e-06,
      "loss": 0.6956,
      "step": 9800
    },
    {
      "epoch": 3.6054172767203516,
      "grad_norm": 3.6092991828918457,
      "learning_rate": 5.579795021961934e-06,
      "loss": 0.6926,
      "step": 9850
    },
    {
      "epoch": 3.623718887262079,
      "grad_norm": 1.2732123136520386,
      "learning_rate": 5.506588579795023e-06,
      "loss": 0.6901,
      "step": 9900
    },
    {
      "epoch": 3.6420204978038067,
      "grad_norm": 1.149488925933838,
      "learning_rate": 5.433382137628112e-06,
      "loss": 0.6947,
      "step": 9950
    },
    {
      "epoch": 3.660322108345534,
      "grad_norm": 1.580837607383728,
      "learning_rate": 5.3601756954612005e-06,
      "loss": 0.6931,
      "step": 10000
    },
    {
      "epoch": 3.678623718887262,
      "grad_norm": 4.544098854064941,
      "learning_rate": 5.28696925329429e-06,
      "loss": 0.6885,
      "step": 10050
    },
    {
      "epoch": 3.6969253294289897,
      "grad_norm": 2.011023998260498,
      "learning_rate": 5.21376281112738e-06,
      "loss": 0.6962,
      "step": 10100
    },
    {
      "epoch": 3.7152269399707176,
      "grad_norm": 5.078649997711182,
      "learning_rate": 5.140556368960469e-06,
      "loss": 0.6936,
      "step": 10150
    },
    {
      "epoch": 3.733528550512445,
      "grad_norm": 2.257343053817749,
      "learning_rate": 5.067349926793558e-06,
      "loss": 0.6945,
      "step": 10200
    },
    {
      "epoch": 3.7518301610541727,
      "grad_norm": 1.252966046333313,
      "learning_rate": 4.994143484626648e-06,
      "loss": 0.6892,
      "step": 10250
    },
    {
      "epoch": 3.7701317715959,
      "grad_norm": 2.1973466873168945,
      "learning_rate": 4.9209370424597365e-06,
      "loss": 0.6868,
      "step": 10300
    },
    {
      "epoch": 3.788433382137628,
      "grad_norm": 1.9186662435531616,
      "learning_rate": 4.847730600292826e-06,
      "loss": 0.6854,
      "step": 10350
    },
    {
      "epoch": 3.8067349926793557,
      "grad_norm": 3.1210763454437256,
      "learning_rate": 4.774524158125915e-06,
      "loss": 0.6977,
      "step": 10400
    },
    {
      "epoch": 3.8250366032210836,
      "grad_norm": 2.7756359577178955,
      "learning_rate": 4.701317715959005e-06,
      "loss": 0.696,
      "step": 10450
    },
    {
      "epoch": 3.843338213762811,
      "grad_norm": 1.0838842391967773,
      "learning_rate": 4.628111273792094e-06,
      "loss": 0.6921,
      "step": 10500
    },
    {
      "epoch": 3.8616398243045387,
      "grad_norm": 3.4335124492645264,
      "learning_rate": 4.554904831625183e-06,
      "loss": 0.6956,
      "step": 10550
    },
    {
      "epoch": 3.8799414348462666,
      "grad_norm": 1.28254234790802,
      "learning_rate": 4.4816983894582726e-06,
      "loss": 0.691,
      "step": 10600
    },
    {
      "epoch": 3.898243045387994,
      "grad_norm": 2.6828558444976807,
      "learning_rate": 4.4084919472913615e-06,
      "loss": 0.6926,
      "step": 10650
    },
    {
      "epoch": 3.9165446559297217,
      "grad_norm": 1.131003499031067,
      "learning_rate": 4.335285505124451e-06,
      "loss": 0.6894,
      "step": 10700
    },
    {
      "epoch": 3.9348462664714496,
      "grad_norm": 3.3838775157928467,
      "learning_rate": 4.26207906295754e-06,
      "loss": 0.6936,
      "step": 10750
    },
    {
      "epoch": 3.953147877013177,
      "grad_norm": 1.3978344202041626,
      "learning_rate": 4.18887262079063e-06,
      "loss": 0.6873,
      "step": 10800
    },
    {
      "epoch": 3.9714494875549047,
      "grad_norm": 4.850183010101318,
      "learning_rate": 4.115666178623719e-06,
      "loss": 0.6855,
      "step": 10850
    },
    {
      "epoch": 3.9897510980966326,
      "grad_norm": 3.063927412033081,
      "learning_rate": 4.0424597364568086e-06,
      "loss": 0.6942,
      "step": 10900
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.5655197657393851,
      "eval_loss": 0.6846572756767273,
      "eval_runtime": 9.7626,
      "eval_samples_per_second": 279.843,
      "eval_steps_per_second": 17.516,
      "step": 10928
    },
    {
      "epoch": 4.008052708638361,
      "grad_norm": 2.9117279052734375,
      "learning_rate": 3.9692532942898975e-06,
      "loss": 0.6924,
      "step": 10950
    },
    {
      "epoch": 4.026354319180088,
      "grad_norm": 1.5743687152862549,
      "learning_rate": 3.896046852122987e-06,
      "loss": 0.6929,
      "step": 11000
    },
    {
      "epoch": 4.044655929721816,
      "grad_norm": 1.4719305038452148,
      "learning_rate": 3.822840409956076e-06,
      "loss": 0.6887,
      "step": 11050
    },
    {
      "epoch": 4.062957540263543,
      "grad_norm": 1.4639744758605957,
      "learning_rate": 3.749633967789166e-06,
      "loss": 0.6786,
      "step": 11100
    },
    {
      "epoch": 4.081259150805271,
      "grad_norm": 6.5527729988098145,
      "learning_rate": 3.676427525622255e-06,
      "loss": 0.6868,
      "step": 11150
    },
    {
      "epoch": 4.099560761346998,
      "grad_norm": 1.1454309225082397,
      "learning_rate": 3.6032210834553446e-06,
      "loss": 0.6991,
      "step": 11200
    },
    {
      "epoch": 4.117862371888727,
      "grad_norm": 2.6722428798675537,
      "learning_rate": 3.5300146412884335e-06,
      "loss": 0.6871,
      "step": 11250
    },
    {
      "epoch": 4.136163982430454,
      "grad_norm": 1.562593936920166,
      "learning_rate": 3.456808199121523e-06,
      "loss": 0.6911,
      "step": 11300
    },
    {
      "epoch": 4.154465592972182,
      "grad_norm": 1.5078984498977661,
      "learning_rate": 3.383601756954612e-06,
      "loss": 0.699,
      "step": 11350
    },
    {
      "epoch": 4.172767203513909,
      "grad_norm": 4.862717151641846,
      "learning_rate": 3.3103953147877015e-06,
      "loss": 0.6848,
      "step": 11400
    },
    {
      "epoch": 4.191068814055637,
      "grad_norm": 1.2064334154129028,
      "learning_rate": 3.2371888726207904e-06,
      "loss": 0.6784,
      "step": 11450
    },
    {
      "epoch": 4.209370424597364,
      "grad_norm": 1.4410967826843262,
      "learning_rate": 3.16398243045388e-06,
      "loss": 0.6867,
      "step": 11500
    },
    {
      "epoch": 4.227672035139093,
      "grad_norm": 1.5525456666946411,
      "learning_rate": 3.090775988286969e-06,
      "loss": 0.6878,
      "step": 11550
    },
    {
      "epoch": 4.24597364568082,
      "grad_norm": 2.721654176712036,
      "learning_rate": 3.017569546120059e-06,
      "loss": 0.6911,
      "step": 11600
    },
    {
      "epoch": 4.264275256222548,
      "grad_norm": 2.7285444736480713,
      "learning_rate": 2.944363103953148e-06,
      "loss": 0.6843,
      "step": 11650
    },
    {
      "epoch": 4.282576866764275,
      "grad_norm": 1.2089800834655762,
      "learning_rate": 2.8711566617862375e-06,
      "loss": 0.6989,
      "step": 11700
    },
    {
      "epoch": 4.300878477306003,
      "grad_norm": 3.368563175201416,
      "learning_rate": 2.797950219619327e-06,
      "loss": 0.683,
      "step": 11750
    },
    {
      "epoch": 4.31918008784773,
      "grad_norm": 1.4972318410873413,
      "learning_rate": 2.724743777452416e-06,
      "loss": 0.6897,
      "step": 11800
    },
    {
      "epoch": 4.337481698389459,
      "grad_norm": 3.92740797996521,
      "learning_rate": 2.6515373352855055e-06,
      "loss": 0.6876,
      "step": 11850
    },
    {
      "epoch": 4.355783308931186,
      "grad_norm": 2.367027521133423,
      "learning_rate": 2.5783308931185944e-06,
      "loss": 0.6884,
      "step": 11900
    },
    {
      "epoch": 4.374084919472914,
      "grad_norm": 3.318098306655884,
      "learning_rate": 2.505124450951684e-06,
      "loss": 0.6867,
      "step": 11950
    },
    {
      "epoch": 4.392386530014641,
      "grad_norm": 2.5479702949523926,
      "learning_rate": 2.431918008784773e-06,
      "loss": 0.6787,
      "step": 12000
    },
    {
      "epoch": 4.410688140556369,
      "grad_norm": 2.361260175704956,
      "learning_rate": 2.3587115666178625e-06,
      "loss": 0.6904,
      "step": 12050
    },
    {
      "epoch": 4.428989751098096,
      "grad_norm": 2.41310715675354,
      "learning_rate": 2.285505124450952e-06,
      "loss": 0.6814,
      "step": 12100
    },
    {
      "epoch": 4.447291361639825,
      "grad_norm": 2.390275716781616,
      "learning_rate": 2.212298682284041e-06,
      "loss": 0.667,
      "step": 12150
    },
    {
      "epoch": 4.465592972181552,
      "grad_norm": 1.9900853633880615,
      "learning_rate": 2.1390922401171305e-06,
      "loss": 0.6873,
      "step": 12200
    },
    {
      "epoch": 4.48389458272328,
      "grad_norm": 5.233725547790527,
      "learning_rate": 2.06588579795022e-06,
      "loss": 0.6844,
      "step": 12250
    },
    {
      "epoch": 4.502196193265007,
      "grad_norm": 2.4165773391723633,
      "learning_rate": 1.992679355783309e-06,
      "loss": 0.6803,
      "step": 12300
    },
    {
      "epoch": 4.520497803806735,
      "grad_norm": 1.9948047399520874,
      "learning_rate": 1.9194729136163985e-06,
      "loss": 0.6885,
      "step": 12350
    },
    {
      "epoch": 4.538799414348462,
      "grad_norm": 1.5085363388061523,
      "learning_rate": 1.8462664714494876e-06,
      "loss": 0.6883,
      "step": 12400
    },
    {
      "epoch": 4.557101024890191,
      "grad_norm": 5.91818380355835,
      "learning_rate": 1.773060029282577e-06,
      "loss": 0.6696,
      "step": 12450
    },
    {
      "epoch": 4.575402635431918,
      "grad_norm": 3.4706273078918457,
      "learning_rate": 1.6998535871156663e-06,
      "loss": 0.6798,
      "step": 12500
    },
    {
      "epoch": 4.593704245973646,
      "grad_norm": 3.0562493801116943,
      "learning_rate": 1.6266471449487556e-06,
      "loss": 0.6763,
      "step": 12550
    },
    {
      "epoch": 4.612005856515373,
      "grad_norm": 5.213473796844482,
      "learning_rate": 1.553440702781845e-06,
      "loss": 0.6793,
      "step": 12600
    },
    {
      "epoch": 4.630307467057101,
      "grad_norm": 2.782263994216919,
      "learning_rate": 1.4802342606149343e-06,
      "loss": 0.6748,
      "step": 12650
    },
    {
      "epoch": 4.648609077598829,
      "grad_norm": 12.179971694946289,
      "learning_rate": 1.4070278184480234e-06,
      "loss": 0.6653,
      "step": 12700
    },
    {
      "epoch": 4.666910688140557,
      "grad_norm": 5.823268413543701,
      "learning_rate": 1.3338213762811127e-06,
      "loss": 0.6916,
      "step": 12750
    },
    {
      "epoch": 4.685212298682284,
      "grad_norm": 4.365823745727539,
      "learning_rate": 1.260614934114202e-06,
      "loss": 0.6892,
      "step": 12800
    },
    {
      "epoch": 4.703513909224012,
      "grad_norm": 6.108221530914307,
      "learning_rate": 1.1874084919472914e-06,
      "loss": 0.6896,
      "step": 12850
    },
    {
      "epoch": 4.721815519765739,
      "grad_norm": 6.130873203277588,
      "learning_rate": 1.1142020497803808e-06,
      "loss": 0.6818,
      "step": 12900
    },
    {
      "epoch": 4.740117130307467,
      "grad_norm": 4.07578706741333,
      "learning_rate": 1.04099560761347e-06,
      "loss": 0.6795,
      "step": 12950
    },
    {
      "epoch": 4.758418740849194,
      "grad_norm": 3.392340898513794,
      "learning_rate": 9.677891654465594e-07,
      "loss": 0.6655,
      "step": 13000
    },
    {
      "epoch": 4.776720351390923,
      "grad_norm": 6.404242992401123,
      "learning_rate": 8.945827232796487e-07,
      "loss": 0.6769,
      "step": 13050
    },
    {
      "epoch": 4.79502196193265,
      "grad_norm": 3.1065897941589355,
      "learning_rate": 8.213762811127379e-07,
      "loss": 0.683,
      "step": 13100
    },
    {
      "epoch": 4.813323572474378,
      "grad_norm": 6.093231201171875,
      "learning_rate": 7.481698389458272e-07,
      "loss": 0.6798,
      "step": 13150
    },
    {
      "epoch": 4.831625183016105,
      "grad_norm": 7.247190952301025,
      "learning_rate": 6.749633967789166e-07,
      "loss": 0.6737,
      "step": 13200
    },
    {
      "epoch": 4.849926793557833,
      "grad_norm": 5.564077377319336,
      "learning_rate": 6.017569546120059e-07,
      "loss": 0.6805,
      "step": 13250
    },
    {
      "epoch": 4.868228404099561,
      "grad_norm": 3.6535987854003906,
      "learning_rate": 5.285505124450952e-07,
      "loss": 0.6612,
      "step": 13300
    },
    {
      "epoch": 4.886530014641289,
      "grad_norm": 3.1486198902130127,
      "learning_rate": 4.553440702781845e-07,
      "loss": 0.6833,
      "step": 13350
    },
    {
      "epoch": 4.904831625183016,
      "grad_norm": 2.8693318367004395,
      "learning_rate": 3.821376281112738e-07,
      "loss": 0.6775,
      "step": 13400
    },
    {
      "epoch": 4.923133235724744,
      "grad_norm": 3.7927544116973877,
      "learning_rate": 3.0893118594436313e-07,
      "loss": 0.673,
      "step": 13450
    },
    {
      "epoch": 4.941434846266471,
      "grad_norm": 3.532392740249634,
      "learning_rate": 2.3572474377745242e-07,
      "loss": 0.6713,
      "step": 13500
    },
    {
      "epoch": 4.959736456808199,
      "grad_norm": 3.34582257270813,
      "learning_rate": 1.6251830161054173e-07,
      "loss": 0.6542,
      "step": 13550
    },
    {
      "epoch": 4.978038067349927,
      "grad_norm": 4.675662517547607,
      "learning_rate": 8.931185944363105e-08,
      "loss": 0.6832,
      "step": 13600
    },
    {
      "epoch": 4.996339677891655,
      "grad_norm": 3.3549957275390625,
      "learning_rate": 1.610541727672035e-08,
      "loss": 0.6633,
      "step": 13650
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.5841874084919473,
      "eval_loss": 0.6651186347007751,
      "eval_runtime": 9.7608,
      "eval_samples_per_second": 279.896,
      "eval_steps_per_second": 17.519,
      "step": 13660
    }
  ],
  "logging_steps": 50,
  "max_steps": 13660,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "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": 7187536254796800.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}