{
  "best_metric": 0.20436730980873108,
  "best_model_checkpoint": "./ryan03302024/checkpoint-2700",
  "epoch": 3.0,
  "eval_steps": 100,
  "global_step": 8319,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 0.9509561657905579,
      "learning_rate": 9.969948311095084e-05,
      "loss": 0.5096,
      "step": 25
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5529997944831848,
      "learning_rate": 9.939896622190168e-05,
      "loss": 0.3519,
      "step": 50
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6374471187591553,
      "learning_rate": 9.909844933285252e-05,
      "loss": 0.3431,
      "step": 75
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.235526204109192,
      "learning_rate": 9.879793244380335e-05,
      "loss": 0.3617,
      "step": 100
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.3133379817008972,
      "eval_na_accuracy": 0.876288652420044,
      "eval_ordinal_accuracy": 0.41741588711738586,
      "eval_ordinal_mae": 0.8465587496757507,
      "eval_runtime": 338.0185,
      "eval_samples_per_second": 13.239,
      "eval_steps_per_second": 1.657,
      "step": 100
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0370614528656006,
      "learning_rate": 9.849741555475418e-05,
      "loss": 0.3721,
      "step": 125
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.619168996810913,
      "learning_rate": 9.819689866570502e-05,
      "loss": 0.3498,
      "step": 150
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6790041923522949,
      "learning_rate": 9.789638177665586e-05,
      "loss": 0.3633,
      "step": 175
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47009700536727905,
      "learning_rate": 9.759586488760669e-05,
      "loss": 0.2891,
      "step": 200
    },
    {
      "epoch": 0.07,
      "eval_loss": 0.3022407293319702,
      "eval_na_accuracy": 0.699312686920166,
      "eval_ordinal_accuracy": 0.5044952630996704,
      "eval_ordinal_mae": 0.7738743424415588,
      "eval_runtime": 195.8973,
      "eval_samples_per_second": 22.844,
      "eval_steps_per_second": 2.859,
      "step": 200
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6545381546020508,
      "learning_rate": 9.73073686741195e-05,
      "loss": 0.3277,
      "step": 225
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2690508961677551,
      "learning_rate": 9.700685178507033e-05,
      "loss": 0.3505,
      "step": 250
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.571977138519287,
      "learning_rate": 9.670633489602116e-05,
      "loss": 0.2819,
      "step": 275
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7339119911193848,
      "learning_rate": 9.6405818006972e-05,
      "loss": 0.3163,
      "step": 300
    },
    {
      "epoch": 0.11,
      "eval_loss": 0.26153865456581116,
      "eval_na_accuracy": 0.7903780341148376,
      "eval_ordinal_accuracy": 0.5597226023674011,
      "eval_ordinal_mae": 0.6166509985923767,
      "eval_runtime": 200.5643,
      "eval_samples_per_second": 22.312,
      "eval_steps_per_second": 2.792,
      "step": 300
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.86951744556427,
      "learning_rate": 9.610530111792283e-05,
      "loss": 0.2707,
      "step": 325
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3769865036010742,
      "learning_rate": 9.580478422887367e-05,
      "loss": 0.2917,
      "step": 350
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39000704884529114,
      "learning_rate": 9.55042673398245e-05,
      "loss": 0.2765,
      "step": 375
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7430490255355835,
      "learning_rate": 9.520375045077534e-05,
      "loss": 0.2781,
      "step": 400
    },
    {
      "epoch": 0.14,
      "eval_loss": 0.2597866952419281,
      "eval_na_accuracy": 0.8298969268798828,
      "eval_ordinal_accuracy": 0.5936295986175537,
      "eval_ordinal_mae": 0.5432500839233398,
      "eval_runtime": 199.7937,
      "eval_samples_per_second": 22.398,
      "eval_steps_per_second": 2.803,
      "step": 400
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8847980499267578,
      "learning_rate": 9.490323356172616e-05,
      "loss": 0.2643,
      "step": 425
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0712331533432007,
      "learning_rate": 9.461473734823898e-05,
      "loss": 0.2747,
      "step": 450
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9433643221855164,
      "learning_rate": 9.431422045918981e-05,
      "loss": 0.295,
      "step": 475
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2171131372451782,
      "learning_rate": 9.401370357014065e-05,
      "loss": 0.2731,
      "step": 500
    },
    {
      "epoch": 0.18,
      "eval_loss": 0.26127612590789795,
      "eval_na_accuracy": 0.8453608155250549,
      "eval_ordinal_accuracy": 0.5566401481628418,
      "eval_ordinal_mae": 0.5651282668113708,
      "eval_runtime": 199.6257,
      "eval_samples_per_second": 22.417,
      "eval_steps_per_second": 2.805,
      "step": 500
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1272056102752686,
      "learning_rate": 9.371318668109149e-05,
      "loss": 0.2878,
      "step": 525
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4902811348438263,
      "learning_rate": 9.341266979204232e-05,
      "loss": 0.2426,
      "step": 550
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5631658434867859,
      "learning_rate": 9.311215290299315e-05,
      "loss": 0.2487,
      "step": 575
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6327888369560242,
      "learning_rate": 9.281163601394399e-05,
      "loss": 0.2926,
      "step": 600
    },
    {
      "epoch": 0.22,
      "eval_loss": 0.2734401226043701,
      "eval_na_accuracy": 0.9089347124099731,
      "eval_ordinal_accuracy": 0.5494477152824402,
      "eval_ordinal_mae": 0.5305272340774536,
      "eval_runtime": 196.452,
      "eval_samples_per_second": 22.779,
      "eval_steps_per_second": 2.851,
      "step": 600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6612924337387085,
      "learning_rate": 9.251111912489481e-05,
      "loss": 0.2871,
      "step": 625
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.513719081878662,
      "learning_rate": 9.221060223584565e-05,
      "loss": 0.2538,
      "step": 650
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1069471836090088,
      "learning_rate": 9.191008534679649e-05,
      "loss": 0.21,
      "step": 675
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.458164691925049,
      "learning_rate": 9.160956845774733e-05,
      "loss": 0.2686,
      "step": 700
    },
    {
      "epoch": 0.25,
      "eval_loss": 0.23620270192623138,
      "eval_na_accuracy": 0.7886598110198975,
      "eval_ordinal_accuracy": 0.6249678730964661,
      "eval_ordinal_mae": 0.485275000333786,
      "eval_runtime": 190.0791,
      "eval_samples_per_second": 23.543,
      "eval_steps_per_second": 2.946,
      "step": 700
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5598509907722473,
      "learning_rate": 9.130905156869817e-05,
      "loss": 0.2525,
      "step": 725
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2617449760437012,
      "learning_rate": 9.1008534679649e-05,
      "loss": 0.2453,
      "step": 750
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5153591632843018,
      "learning_rate": 9.070801779059983e-05,
      "loss": 0.225,
      "step": 775
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.562745451927185,
      "learning_rate": 9.040750090155067e-05,
      "loss": 0.2715,
      "step": 800
    },
    {
      "epoch": 0.29,
      "eval_loss": 0.24541285634040833,
      "eval_na_accuracy": 0.7714776396751404,
      "eval_ordinal_accuracy": 0.6254816055297852,
      "eval_ordinal_mae": 0.49135151505470276,
      "eval_runtime": 193.3675,
      "eval_samples_per_second": 23.142,
      "eval_steps_per_second": 2.896,
      "step": 800
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5192123651504517,
      "learning_rate": 9.010698401250151e-05,
      "loss": 0.2424,
      "step": 825
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8706183433532715,
      "learning_rate": 8.980646712345235e-05,
      "loss": 0.2341,
      "step": 850
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0854986906051636,
      "learning_rate": 8.950595023440318e-05,
      "loss": 0.2694,
      "step": 875
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.208933115005493,
      "learning_rate": 8.920543334535402e-05,
      "loss": 0.2459,
      "step": 900
    },
    {
      "epoch": 0.32,
      "eval_loss": 0.2451752871274948,
      "eval_na_accuracy": 0.7800687551498413,
      "eval_ordinal_accuracy": 0.6072437763214111,
      "eval_ordinal_mae": 0.47625866532325745,
      "eval_runtime": 192.5483,
      "eval_samples_per_second": 23.241,
      "eval_steps_per_second": 2.908,
      "step": 900
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8433920741081238,
      "learning_rate": 8.890491645630485e-05,
      "loss": 0.2375,
      "step": 925
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.9415366649627686,
      "learning_rate": 8.860439956725569e-05,
      "loss": 0.2918,
      "step": 950
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6653732061386108,
      "learning_rate": 8.830388267820651e-05,
      "loss": 0.2615,
      "step": 975
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.47592782974243164,
      "learning_rate": 8.800336578915735e-05,
      "loss": 0.2033,
      "step": 1000
    },
    {
      "epoch": 0.36,
      "eval_loss": 0.23654678463935852,
      "eval_na_accuracy": 0.7663230299949646,
      "eval_ordinal_accuracy": 0.6105831265449524,
      "eval_ordinal_mae": 0.4967080354690552,
      "eval_runtime": 192.911,
      "eval_samples_per_second": 23.197,
      "eval_steps_per_second": 2.903,
      "step": 1000
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8534409403800964,
      "learning_rate": 8.770284890010819e-05,
      "loss": 0.2082,
      "step": 1025
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6388253569602966,
      "learning_rate": 8.740233201105903e-05,
      "loss": 0.2238,
      "step": 1050
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5685051679611206,
      "learning_rate": 8.710181512200987e-05,
      "loss": 0.2775,
      "step": 1075
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9940025806427002,
      "learning_rate": 8.680129823296069e-05,
      "loss": 0.2234,
      "step": 1100
    },
    {
      "epoch": 0.4,
      "eval_loss": 0.22986453771591187,
      "eval_na_accuracy": 0.8676975965499878,
      "eval_ordinal_accuracy": 0.6180323362350464,
      "eval_ordinal_mae": 0.49474066495895386,
      "eval_runtime": 191.9691,
      "eval_samples_per_second": 23.311,
      "eval_steps_per_second": 2.917,
      "step": 1100
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6706956624984741,
      "learning_rate": 8.650078134391153e-05,
      "loss": 0.2027,
      "step": 1125
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4717120826244354,
      "learning_rate": 8.620026445486237e-05,
      "loss": 0.2072,
      "step": 1150
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5387229919433594,
      "learning_rate": 8.58997475658132e-05,
      "loss": 0.2328,
      "step": 1175
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0794016122817993,
      "learning_rate": 8.559923067676405e-05,
      "loss": 0.2035,
      "step": 1200
    },
    {
      "epoch": 0.43,
      "eval_loss": 0.2313634753227234,
      "eval_na_accuracy": 0.7835051417350769,
      "eval_ordinal_accuracy": 0.6308759450912476,
      "eval_ordinal_mae": 0.4743907153606415,
      "eval_runtime": 191.4648,
      "eval_samples_per_second": 23.372,
      "eval_steps_per_second": 2.925,
      "step": 1200
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2199459075927734,
      "learning_rate": 8.529871378771488e-05,
      "loss": 0.2769,
      "step": 1225
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5869888067245483,
      "learning_rate": 8.499819689866571e-05,
      "loss": 0.2194,
      "step": 1250
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1209378242492676,
      "learning_rate": 8.469768000961655e-05,
      "loss": 0.2536,
      "step": 1275
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4376941919326782,
      "learning_rate": 8.439716312056739e-05,
      "loss": 0.2277,
      "step": 1300
    },
    {
      "epoch": 0.47,
      "eval_loss": 0.23891955614089966,
      "eval_na_accuracy": 0.730240523815155,
      "eval_ordinal_accuracy": 0.643462598323822,
      "eval_ordinal_mae": 0.46490678191185,
      "eval_runtime": 193.3761,
      "eval_samples_per_second": 23.141,
      "eval_steps_per_second": 2.896,
      "step": 1300
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8051934242248535,
      "learning_rate": 8.409664623151821e-05,
      "loss": 0.2007,
      "step": 1325
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5033048391342163,
      "learning_rate": 8.379612934246905e-05,
      "loss": 0.1993,
      "step": 1350
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.577111542224884,
      "learning_rate": 8.349561245341989e-05,
      "loss": 0.2337,
      "step": 1375
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8847935795783997,
      "learning_rate": 8.319509556437071e-05,
      "loss": 0.2535,
      "step": 1400
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.22593823075294495,
      "eval_na_accuracy": 0.8247422575950623,
      "eval_ordinal_accuracy": 0.6021063327789307,
      "eval_ordinal_mae": 0.45086583495140076,
      "eval_runtime": 192.5485,
      "eval_samples_per_second": 23.241,
      "eval_steps_per_second": 2.908,
      "step": 1400
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.888088345527649,
      "learning_rate": 8.289457867532155e-05,
      "loss": 0.2162,
      "step": 1425
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9532464146614075,
      "learning_rate": 8.259406178627239e-05,
      "loss": 0.2632,
      "step": 1450
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.7865819931030273,
      "learning_rate": 8.229354489722323e-05,
      "loss": 0.2005,
      "step": 1475
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0594526529312134,
      "learning_rate": 8.199302800817407e-05,
      "loss": 0.2209,
      "step": 1500
    },
    {
      "epoch": 0.54,
      "eval_loss": 0.23685050010681152,
      "eval_na_accuracy": 0.7577319741249084,
      "eval_ordinal_accuracy": 0.6362702250480652,
      "eval_ordinal_mae": 0.45073509216308594,
      "eval_runtime": 192.1913,
      "eval_samples_per_second": 23.284,
      "eval_steps_per_second": 2.914,
      "step": 1500
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9557788372039795,
      "learning_rate": 8.16925111191249e-05,
      "loss": 0.2009,
      "step": 1525
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7707318067550659,
      "learning_rate": 8.139199423007573e-05,
      "loss": 0.2172,
      "step": 1550
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.569989562034607,
      "learning_rate": 8.109147734102657e-05,
      "loss": 0.2229,
      "step": 1575
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5089187622070312,
      "learning_rate": 8.079096045197741e-05,
      "loss": 0.2007,
      "step": 1600
    },
    {
      "epoch": 0.58,
      "eval_loss": 0.21611952781677246,
      "eval_na_accuracy": 0.831615149974823,
      "eval_ordinal_accuracy": 0.6539943218231201,
      "eval_ordinal_mae": 0.427180677652359,
      "eval_runtime": 190.3474,
      "eval_samples_per_second": 23.51,
      "eval_steps_per_second": 2.942,
      "step": 1600
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9716092348098755,
      "learning_rate": 8.049044356292825e-05,
      "loss": 0.2375,
      "step": 1625
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.04927659034729,
      "learning_rate": 8.018992667387908e-05,
      "loss": 0.2548,
      "step": 1650
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1613515615463257,
      "learning_rate": 7.988940978482991e-05,
      "loss": 0.3238,
      "step": 1675
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4977881908416748,
      "learning_rate": 7.958889289578075e-05,
      "loss": 0.2013,
      "step": 1700
    },
    {
      "epoch": 0.61,
      "eval_loss": 0.24333100020885468,
      "eval_na_accuracy": 0.7731958627700806,
      "eval_ordinal_accuracy": 0.6128949522972107,
      "eval_ordinal_mae": 0.43257907032966614,
      "eval_runtime": 193.154,
      "eval_samples_per_second": 23.168,
      "eval_steps_per_second": 2.899,
      "step": 1700
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6059219837188721,
      "learning_rate": 7.928837600673157e-05,
      "loss": 0.273,
      "step": 1725
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3245080709457397,
      "learning_rate": 7.898785911768241e-05,
      "loss": 0.2148,
      "step": 1750
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5151876211166382,
      "learning_rate": 7.868734222863325e-05,
      "loss": 0.1901,
      "step": 1775
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37926343083381653,
      "learning_rate": 7.838682533958409e-05,
      "loss": 0.1999,
      "step": 1800
    },
    {
      "epoch": 0.65,
      "eval_loss": 0.22266778349876404,
      "eval_na_accuracy": 0.8247422575950623,
      "eval_ordinal_accuracy": 0.6552786827087402,
      "eval_ordinal_mae": 0.4460422396659851,
      "eval_runtime": 193.6404,
      "eval_samples_per_second": 23.11,
      "eval_steps_per_second": 2.892,
      "step": 1800
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.746518611907959,
      "learning_rate": 7.808630845053493e-05,
      "loss": 0.2169,
      "step": 1825
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3521455228328705,
      "learning_rate": 7.778579156148575e-05,
      "loss": 0.228,
      "step": 1850
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3322985172271729,
      "learning_rate": 7.748527467243659e-05,
      "loss": 0.2047,
      "step": 1875
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6390167474746704,
      "learning_rate": 7.718475778338743e-05,
      "loss": 0.2157,
      "step": 1900
    },
    {
      "epoch": 0.69,
      "eval_loss": 0.2134324014186859,
      "eval_na_accuracy": 0.8161512017250061,
      "eval_ordinal_accuracy": 0.6362702250480652,
      "eval_ordinal_mae": 0.4728001058101654,
      "eval_runtime": 191.9038,
      "eval_samples_per_second": 23.319,
      "eval_steps_per_second": 2.918,
      "step": 1900
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.581801176071167,
      "learning_rate": 7.688424089433827e-05,
      "loss": 0.2335,
      "step": 1925
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3418828248977661,
      "learning_rate": 7.65837240052891e-05,
      "loss": 0.2446,
      "step": 1950
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.605707585811615,
      "learning_rate": 7.628320711623994e-05,
      "loss": 0.2144,
      "step": 1975
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6443113088607788,
      "learning_rate": 7.598269022719077e-05,
      "loss": 0.2154,
      "step": 2000
    },
    {
      "epoch": 0.72,
      "eval_loss": 0.2238776683807373,
      "eval_na_accuracy": 0.8573883175849915,
      "eval_ordinal_accuracy": 0.5787310600280762,
      "eval_ordinal_mae": 0.4734483063220978,
      "eval_runtime": 192.57,
      "eval_samples_per_second": 23.238,
      "eval_steps_per_second": 2.908,
      "step": 2000
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4812798500061035,
      "learning_rate": 7.568217333814161e-05,
      "loss": 0.2172,
      "step": 2025
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5400125980377197,
      "learning_rate": 7.538165644909245e-05,
      "loss": 0.1907,
      "step": 2050
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6313005685806274,
      "learning_rate": 7.508113956004327e-05,
      "loss": 0.2078,
      "step": 2075
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3446621894836426,
      "learning_rate": 7.478062267099411e-05,
      "loss": 0.2169,
      "step": 2100
    },
    {
      "epoch": 0.76,
      "eval_loss": 0.23939213156700134,
      "eval_na_accuracy": 0.8848797082901001,
      "eval_ordinal_accuracy": 0.6254816055297852,
      "eval_ordinal_mae": 0.4392252564430237,
      "eval_runtime": 194.2506,
      "eval_samples_per_second": 23.037,
      "eval_steps_per_second": 2.883,
      "step": 2100
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5709229111671448,
      "learning_rate": 7.448010578194495e-05,
      "loss": 0.2155,
      "step": 2125
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.9002838134765625,
      "learning_rate": 7.417958889289577e-05,
      "loss": 0.2534,
      "step": 2150
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3746689558029175,
      "learning_rate": 7.387907200384661e-05,
      "loss": 0.2489,
      "step": 2175
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1106956005096436,
      "learning_rate": 7.357855511479745e-05,
      "loss": 0.2719,
      "step": 2200
    },
    {
      "epoch": 0.79,
      "eval_loss": 0.2283402532339096,
      "eval_na_accuracy": 0.8780068755149841,
      "eval_ordinal_accuracy": 0.6229129433631897,
      "eval_ordinal_mae": 0.4323791265487671,
      "eval_runtime": 193.3183,
      "eval_samples_per_second": 23.148,
      "eval_steps_per_second": 2.897,
      "step": 2200
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5689876079559326,
      "learning_rate": 7.327803822574829e-05,
      "loss": 0.2054,
      "step": 2225
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.14363694190979,
      "learning_rate": 7.297752133669913e-05,
      "loss": 0.213,
      "step": 2250
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4198272228240967,
      "learning_rate": 7.267700444764997e-05,
      "loss": 0.1729,
      "step": 2275
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5297597646713257,
      "learning_rate": 7.237648755860079e-05,
      "loss": 0.2244,
      "step": 2300
    },
    {
      "epoch": 0.83,
      "eval_loss": 0.2139568328857422,
      "eval_na_accuracy": 0.8728522062301636,
      "eval_ordinal_accuracy": 0.6313896775245667,
      "eval_ordinal_mae": 0.4482755959033966,
      "eval_runtime": 193.8188,
      "eval_samples_per_second": 23.089,
      "eval_steps_per_second": 2.889,
      "step": 2300
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4800972044467926,
      "learning_rate": 7.207597066955163e-05,
      "loss": 0.231,
      "step": 2325
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.125603199005127,
      "learning_rate": 7.177545378050247e-05,
      "loss": 0.2371,
      "step": 2350
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.553396224975586,
      "learning_rate": 7.147493689145331e-05,
      "loss": 0.217,
      "step": 2375
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5980515480041504,
      "learning_rate": 7.117442000240415e-05,
      "loss": 0.2072,
      "step": 2400
    },
    {
      "epoch": 0.87,
      "eval_loss": 0.2198052555322647,
      "eval_na_accuracy": 0.8213058710098267,
      "eval_ordinal_accuracy": 0.6439763903617859,
      "eval_ordinal_mae": 0.4330386817455292,
      "eval_runtime": 193.8423,
      "eval_samples_per_second": 23.086,
      "eval_steps_per_second": 2.889,
      "step": 2400
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0013697147369385,
      "learning_rate": 7.087390311335498e-05,
      "loss": 0.2141,
      "step": 2425
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.5208072662353516,
      "learning_rate": 7.057338622430581e-05,
      "loss": 0.2109,
      "step": 2450
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3612011969089508,
      "learning_rate": 7.027286933525665e-05,
      "loss": 0.1701,
      "step": 2475
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6919033527374268,
      "learning_rate": 6.997235244620747e-05,
      "loss": 0.1754,
      "step": 2500
    },
    {
      "epoch": 0.9,
      "eval_loss": 0.20990537106990814,
      "eval_na_accuracy": 0.8419243693351746,
      "eval_ordinal_accuracy": 0.6712047457695007,
      "eval_ordinal_mae": 0.41983720660209656,
      "eval_runtime": 193.4285,
      "eval_samples_per_second": 23.135,
      "eval_steps_per_second": 2.895,
      "step": 2500
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.494234323501587,
      "learning_rate": 6.967183555715831e-05,
      "loss": 0.2547,
      "step": 2525
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2194478511810303,
      "learning_rate": 6.937131866810915e-05,
      "loss": 0.2105,
      "step": 2550
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8242117166519165,
      "learning_rate": 6.907080177905999e-05,
      "loss": 0.185,
      "step": 2575
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.185267210006714,
      "learning_rate": 6.877028489001081e-05,
      "loss": 0.1773,
      "step": 2600
    },
    {
      "epoch": 0.94,
      "eval_loss": 0.20526131987571716,
      "eval_na_accuracy": 0.8642611503601074,
      "eval_ordinal_accuracy": 0.6586180329322815,
      "eval_ordinal_mae": 0.4104856550693512,
      "eval_runtime": 191.4771,
      "eval_samples_per_second": 23.371,
      "eval_steps_per_second": 2.925,
      "step": 2600
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5727041959762573,
      "learning_rate": 6.846976800096165e-05,
      "loss": 0.1994,
      "step": 2625
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3567266464233398,
      "learning_rate": 6.816925111191249e-05,
      "loss": 0.2087,
      "step": 2650
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43862801790237427,
      "learning_rate": 6.786873422286333e-05,
      "loss": 0.2394,
      "step": 2675
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7140023708343506,
      "learning_rate": 6.756821733381417e-05,
      "loss": 0.2378,
      "step": 2700
    },
    {
      "epoch": 0.97,
      "eval_loss": 0.20436730980873108,
      "eval_na_accuracy": 0.8333333134651184,
      "eval_ordinal_accuracy": 0.6647829413414001,
      "eval_ordinal_mae": 0.4324062764644623,
      "eval_runtime": 193.6819,
      "eval_samples_per_second": 23.105,
      "eval_steps_per_second": 2.891,
      "step": 2700
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44048529863357544,
      "learning_rate": 6.7267700444765e-05,
      "loss": 0.172,
      "step": 2725
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.4433631896972656,
      "learning_rate": 6.696718355571583e-05,
      "loss": 0.1666,
      "step": 2750
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6148251891136169,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.2,
      "step": 2775
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.5114859342575073,
      "learning_rate": 6.636614977761751e-05,
      "loss": 0.1295,
      "step": 2800
    },
    {
      "epoch": 1.01,
      "eval_loss": 0.20437294244766235,
      "eval_na_accuracy": 0.8247422575950623,
      "eval_ordinal_accuracy": 0.6843051910400391,
      "eval_ordinal_mae": 0.4015503525733948,
      "eval_runtime": 191.7797,
      "eval_samples_per_second": 23.334,
      "eval_steps_per_second": 2.92,
      "step": 2800
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.5778698921203613,
      "learning_rate": 6.606563288856835e-05,
      "loss": 0.1506,
      "step": 2825
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.3983612358570099,
      "learning_rate": 6.576511599951917e-05,
      "loss": 0.1358,
      "step": 2850
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8497461080551147,
      "learning_rate": 6.546459911047001e-05,
      "loss": 0.1288,
      "step": 2875
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2482887506484985,
      "learning_rate": 6.516408222142085e-05,
      "loss": 0.1126,
      "step": 2900
    },
    {
      "epoch": 1.05,
      "eval_loss": 0.2301819771528244,
      "eval_na_accuracy": 0.7577319741249084,
      "eval_ordinal_accuracy": 0.6804521083831787,
      "eval_ordinal_mae": 0.402468740940094,
      "eval_runtime": 195.9356,
      "eval_samples_per_second": 22.839,
      "eval_steps_per_second": 2.858,
      "step": 2900
    },
    {
      "epoch": 1.05,
      "grad_norm": 12.926443099975586,
      "learning_rate": 6.486356533237167e-05,
      "loss": 0.1244,
      "step": 2925
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.5321182012557983,
      "learning_rate": 6.456304844332251e-05,
      "loss": 0.1368,
      "step": 2950
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.4846901297569275,
      "learning_rate": 6.426253155427335e-05,
      "loss": 0.1837,
      "step": 2975
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6585441827774048,
      "learning_rate": 6.396201466522419e-05,
      "loss": 0.1262,
      "step": 3000
    },
    {
      "epoch": 1.08,
      "eval_loss": 0.22050338983535767,
      "eval_na_accuracy": 0.8092783689498901,
      "eval_ordinal_accuracy": 0.6516824960708618,
      "eval_ordinal_mae": 0.4016842246055603,
      "eval_runtime": 191.162,
      "eval_samples_per_second": 23.409,
      "eval_steps_per_second": 2.929,
      "step": 3000
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.30136457085609436,
      "learning_rate": 6.366149777617503e-05,
      "loss": 0.1417,
      "step": 3025
    },
    {
      "epoch": 1.1,
      "grad_norm": 2.017369508743286,
      "learning_rate": 6.336098088712587e-05,
      "loss": 0.1155,
      "step": 3050
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6489114165306091,
      "learning_rate": 6.306046399807669e-05,
      "loss": 0.1074,
      "step": 3075
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.5059970617294312,
      "learning_rate": 6.275994710902753e-05,
      "loss": 0.1104,
      "step": 3100
    },
    {
      "epoch": 1.12,
      "eval_loss": 0.21169866621494293,
      "eval_na_accuracy": 0.8453608155250549,
      "eval_ordinal_accuracy": 0.6778833866119385,
      "eval_ordinal_mae": 0.39305010437965393,
      "eval_runtime": 189.5391,
      "eval_samples_per_second": 23.61,
      "eval_steps_per_second": 2.955,
      "step": 3100
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.127930998802185,
      "learning_rate": 6.245943021997837e-05,
      "loss": 0.1353,
      "step": 3125
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.9614742994308472,
      "learning_rate": 6.215891333092921e-05,
      "loss": 0.1069,
      "step": 3150
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.5779272317886353,
      "learning_rate": 6.185839644188005e-05,
      "loss": 0.1161,
      "step": 3175
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.6223546266555786,
      "learning_rate": 6.155787955283087e-05,
      "loss": 0.1657,
      "step": 3200
    },
    {
      "epoch": 1.15,
      "eval_loss": 0.21738822758197784,
      "eval_na_accuracy": 0.8591065406799316,
      "eval_ordinal_accuracy": 0.6665810346603394,
      "eval_ordinal_mae": 0.38904985785484314,
      "eval_runtime": 193.8286,
      "eval_samples_per_second": 23.087,
      "eval_steps_per_second": 2.889,
      "step": 3200
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1987367868423462,
      "learning_rate": 6.125736266378171e-05,
      "loss": 0.1139,
      "step": 3225
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.510863184928894,
      "learning_rate": 6.095684577473254e-05,
      "loss": 0.1286,
      "step": 3250
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7094324827194214,
      "learning_rate": 6.065632888568338e-05,
      "loss": 0.1231,
      "step": 3275
    },
    {
      "epoch": 1.19,
      "grad_norm": 2.9405245780944824,
      "learning_rate": 6.035581199663422e-05,
      "loss": 0.1186,
      "step": 3300
    },
    {
      "epoch": 1.19,
      "eval_loss": 0.22988717257976532,
      "eval_na_accuracy": 0.8058419227600098,
      "eval_ordinal_accuracy": 0.6622142195701599,
      "eval_ordinal_mae": 0.40129175782203674,
      "eval_runtime": 191.4373,
      "eval_samples_per_second": 23.376,
      "eval_steps_per_second": 2.925,
      "step": 3300
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.9970686435699463,
      "learning_rate": 6.005529510758505e-05,
      "loss": 0.1492,
      "step": 3325
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.4853156805038452,
      "learning_rate": 5.975477821853589e-05,
      "loss": 0.0923,
      "step": 3350
    },
    {
      "epoch": 1.22,
      "grad_norm": 3.004967451095581,
      "learning_rate": 5.9454261329486714e-05,
      "loss": 0.1395,
      "step": 3375
    },
    {
      "epoch": 1.23,
      "grad_norm": 2.5411417484283447,
      "learning_rate": 5.915374444043755e-05,
      "loss": 0.1304,
      "step": 3400
    },
    {
      "epoch": 1.23,
      "eval_loss": 0.21757462620735168,
      "eval_na_accuracy": 0.8109965920448303,
      "eval_ordinal_accuracy": 0.6902132034301758,
      "eval_ordinal_mae": 0.3801174759864807,
      "eval_runtime": 191.9938,
      "eval_samples_per_second": 23.308,
      "eval_steps_per_second": 2.917,
      "step": 3400
    },
    {
      "epoch": 1.24,
      "grad_norm": 2.031360626220703,
      "learning_rate": 5.885322755138839e-05,
      "loss": 0.161,
      "step": 3425
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9166708588600159,
      "learning_rate": 5.855271066233923e-05,
      "loss": 0.1062,
      "step": 3450
    },
    {
      "epoch": 1.25,
      "grad_norm": 2.4028913974761963,
      "learning_rate": 5.825219377329007e-05,
      "loss": 0.1085,
      "step": 3475
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.99100261926651,
      "learning_rate": 5.79516768842409e-05,
      "loss": 0.1081,
      "step": 3500
    },
    {
      "epoch": 1.26,
      "eval_loss": 0.23295216262340546,
      "eval_na_accuracy": 0.831615149974823,
      "eval_ordinal_accuracy": 0.664269208908081,
      "eval_ordinal_mae": 0.3867188096046448,
      "eval_runtime": 194.3732,
      "eval_samples_per_second": 23.023,
      "eval_steps_per_second": 2.881,
      "step": 3500
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.6745060682296753,
      "learning_rate": 5.765115999519173e-05,
      "loss": 0.1079,
      "step": 3525
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.5718048810958862,
      "learning_rate": 5.735064310614256e-05,
      "loss": 0.1353,
      "step": 3550
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.4868764579296112,
      "learning_rate": 5.70501262170934e-05,
      "loss": 0.1872,
      "step": 3575
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2596338987350464,
      "learning_rate": 5.674960932804424e-05,
      "loss": 0.1281,
      "step": 3600
    },
    {
      "epoch": 1.3,
      "eval_loss": 0.23198525607585907,
      "eval_na_accuracy": 0.7680412530899048,
      "eval_ordinal_accuracy": 0.6902132034301758,
      "eval_ordinal_mae": 0.3953803777694702,
      "eval_runtime": 193.0669,
      "eval_samples_per_second": 23.178,
      "eval_steps_per_second": 2.901,
      "step": 3600
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.5509974360466003,
      "learning_rate": 5.644909243899508e-05,
      "loss": 0.1085,
      "step": 3625
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8018646240234375,
      "learning_rate": 5.614857554994592e-05,
      "loss": 0.1134,
      "step": 3650
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.46525055170059204,
      "learning_rate": 5.584805866089674e-05,
      "loss": 0.1789,
      "step": 3675
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.4697776734828949,
      "learning_rate": 5.554754177184758e-05,
      "loss": 0.1192,
      "step": 3700
    },
    {
      "epoch": 1.33,
      "eval_loss": 0.231234610080719,
      "eval_na_accuracy": 0.7989690899848938,
      "eval_ordinal_accuracy": 0.6768559217453003,
      "eval_ordinal_mae": 0.4108929932117462,
      "eval_runtime": 192.9099,
      "eval_samples_per_second": 23.197,
      "eval_steps_per_second": 2.903,
      "step": 3700
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.9536067247390747,
      "learning_rate": 5.524702488279841e-05,
      "loss": 0.1531,
      "step": 3725
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.6636119484901428,
      "learning_rate": 5.494650799374925e-05,
      "loss": 0.1152,
      "step": 3750
    },
    {
      "epoch": 1.36,
      "grad_norm": 3.2403175830841064,
      "learning_rate": 5.464599110470009e-05,
      "loss": 0.133,
      "step": 3775
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7575627565383911,
      "learning_rate": 5.434547421565093e-05,
      "loss": 0.1029,
      "step": 3800
    },
    {
      "epoch": 1.37,
      "eval_loss": 0.21953611075878143,
      "eval_na_accuracy": 0.8024054765701294,
      "eval_ordinal_accuracy": 0.681993305683136,
      "eval_ordinal_mae": 0.38697296380996704,
      "eval_runtime": 192.4936,
      "eval_samples_per_second": 23.248,
      "eval_steps_per_second": 2.909,
      "step": 3800
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.6479007601737976,
      "learning_rate": 5.404495732660175e-05,
      "loss": 0.1174,
      "step": 3825
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.8630564212799072,
      "learning_rate": 5.374444043755259e-05,
      "loss": 0.1398,
      "step": 3850
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.6482840180397034,
      "learning_rate": 5.344392354850343e-05,
      "loss": 0.1144,
      "step": 3875
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.39722609519958496,
      "learning_rate": 5.314340665945426e-05,
      "loss": 0.1159,
      "step": 3900
    },
    {
      "epoch": 1.41,
      "eval_loss": 0.22003380954265594,
      "eval_na_accuracy": 0.7903780341148376,
      "eval_ordinal_accuracy": 0.6812227368354797,
      "eval_ordinal_mae": 0.3859783113002777,
      "eval_runtime": 193.9688,
      "eval_samples_per_second": 23.071,
      "eval_steps_per_second": 2.887,
      "step": 3900
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7128223776817322,
      "learning_rate": 5.28428897704051e-05,
      "loss": 0.1437,
      "step": 3925
    },
    {
      "epoch": 1.42,
      "grad_norm": 2.3041298389434814,
      "learning_rate": 5.254237288135594e-05,
      "loss": 0.1413,
      "step": 3950
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9336991310119629,
      "learning_rate": 5.2241855992306764e-05,
      "loss": 0.1112,
      "step": 3975
    },
    {
      "epoch": 1.44,
      "grad_norm": 3.5301501750946045,
      "learning_rate": 5.19413391032576e-05,
      "loss": 0.1159,
      "step": 4000
    },
    {
      "epoch": 1.44,
      "eval_loss": 0.21585588157176971,
      "eval_na_accuracy": 0.7989690899848938,
      "eval_ordinal_accuracy": 0.6981762051582336,
      "eval_ordinal_mae": 0.3712264895439148,
      "eval_runtime": 191.7494,
      "eval_samples_per_second": 23.338,
      "eval_steps_per_second": 2.92,
      "step": 4000
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8215810060501099,
      "learning_rate": 5.164082221420844e-05,
      "loss": 0.0948,
      "step": 4025
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8609411716461182,
      "learning_rate": 5.134030532515928e-05,
      "loss": 0.0936,
      "step": 4050
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.7595614790916443,
      "learning_rate": 5.103978843611011e-05,
      "loss": 0.1181,
      "step": 4075
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7951629757881165,
      "learning_rate": 5.073927154706095e-05,
      "loss": 0.107,
      "step": 4100
    },
    {
      "epoch": 1.48,
      "eval_loss": 0.22618594765663147,
      "eval_na_accuracy": 0.8213058710098267,
      "eval_ordinal_accuracy": 0.6904700994491577,
      "eval_ordinal_mae": 0.37572237849235535,
      "eval_runtime": 193.8218,
      "eval_samples_per_second": 23.088,
      "eval_steps_per_second": 2.889,
      "step": 4100
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.809657096862793,
      "learning_rate": 5.0438754658011775e-05,
      "loss": 0.1275,
      "step": 4125
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.6124809384346008,
      "learning_rate": 5.0138237768962613e-05,
      "loss": 0.0975,
      "step": 4150
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.5735268592834473,
      "learning_rate": 4.983772087991345e-05,
      "loss": 0.1267,
      "step": 4175
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2053624987602234,
      "learning_rate": 4.953720399086429e-05,
      "loss": 0.1262,
      "step": 4200
    },
    {
      "epoch": 1.51,
      "eval_loss": 0.22913989424705505,
      "eval_na_accuracy": 0.8247422575950623,
      "eval_ordinal_accuracy": 0.683534562587738,
      "eval_ordinal_mae": 0.3841015100479126,
      "eval_runtime": 192.0713,
      "eval_samples_per_second": 23.299,
      "eval_steps_per_second": 2.916,
      "step": 4200
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.163071632385254,
      "learning_rate": 4.923668710181513e-05,
      "loss": 0.112,
      "step": 4225
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.48347094655036926,
      "learning_rate": 4.893617021276596e-05,
      "loss": 0.1312,
      "step": 4250
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.0154635906219482,
      "learning_rate": 4.863565332371679e-05,
      "loss": 0.1057,
      "step": 4275
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.7999942302703857,
      "learning_rate": 4.833513643466763e-05,
      "loss": 0.1437,
      "step": 4300
    },
    {
      "epoch": 1.55,
      "eval_loss": 0.23114901781082153,
      "eval_na_accuracy": 0.800687313079834,
      "eval_ordinal_accuracy": 0.6922681927680969,
      "eval_ordinal_mae": 0.3750612139701843,
      "eval_runtime": 193.0261,
      "eval_samples_per_second": 23.183,
      "eval_steps_per_second": 2.901,
      "step": 4300
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7880312204360962,
      "learning_rate": 4.803461954561846e-05,
      "loss": 0.1099,
      "step": 4325
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.5259920954704285,
      "learning_rate": 4.77341026565693e-05,
      "loss": 0.1355,
      "step": 4350
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.49388599395751953,
      "learning_rate": 4.743358576752014e-05,
      "loss": 0.1331,
      "step": 4375
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.7215922474861145,
      "learning_rate": 4.714508955403294e-05,
      "loss": 0.0916,
      "step": 4400
    },
    {
      "epoch": 1.59,
      "eval_loss": 0.23432399332523346,
      "eval_na_accuracy": 0.8659793734550476,
      "eval_ordinal_accuracy": 0.6791677474975586,
      "eval_ordinal_mae": 0.37433725595474243,
      "eval_runtime": 191.7082,
      "eval_samples_per_second": 23.343,
      "eval_steps_per_second": 2.921,
      "step": 4400
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.4619504511356354,
      "learning_rate": 4.684457266498378e-05,
      "loss": 0.11,
      "step": 4425
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.470409393310547,
      "learning_rate": 4.654405577593461e-05,
      "loss": 0.138,
      "step": 4450
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.46536511182785034,
      "learning_rate": 4.6243538886885443e-05,
      "loss": 0.1233,
      "step": 4475
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.4539566934108734,
      "learning_rate": 4.594302199783628e-05,
      "loss": 0.1266,
      "step": 4500
    },
    {
      "epoch": 1.62,
      "eval_loss": 0.22511447966098785,
      "eval_na_accuracy": 0.8505154848098755,
      "eval_ordinal_accuracy": 0.6861032843589783,
      "eval_ordinal_mae": 0.37239742279052734,
      "eval_runtime": 185.8881,
      "eval_samples_per_second": 24.074,
      "eval_steps_per_second": 3.013,
      "step": 4500
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.497732013463974,
      "learning_rate": 4.5642505108787114e-05,
      "loss": 0.109,
      "step": 4525
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.833667278289795,
      "learning_rate": 4.534198821973795e-05,
      "loss": 0.1342,
      "step": 4550
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.6105310320854187,
      "learning_rate": 4.504147133068879e-05,
      "loss": 0.125,
      "step": 4575
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.5035513639450073,
      "learning_rate": 4.474095444163962e-05,
      "loss": 0.1185,
      "step": 4600
    },
    {
      "epoch": 1.66,
      "eval_loss": 0.22424614429473877,
      "eval_na_accuracy": 0.8264604806900024,
      "eval_ordinal_accuracy": 0.6902132034301758,
      "eval_ordinal_mae": 0.36660563945770264,
      "eval_runtime": 188.1827,
      "eval_samples_per_second": 23.78,
      "eval_steps_per_second": 2.976,
      "step": 4600
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.6165774464607239,
      "learning_rate": 4.444043755259046e-05,
      "loss": 0.1273,
      "step": 4625
    },
    {
      "epoch": 1.68,
      "grad_norm": 2.2787024974823,
      "learning_rate": 4.413992066354129e-05,
      "loss": 0.1537,
      "step": 4650
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2412303686141968,
      "learning_rate": 4.3839403774492125e-05,
      "loss": 0.1197,
      "step": 4675
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.671398162841797,
      "learning_rate": 4.353888688544296e-05,
      "loss": 0.1037,
      "step": 4700
    },
    {
      "epoch": 1.69,
      "eval_loss": 0.22189220786094666,
      "eval_na_accuracy": 0.8522336483001709,
      "eval_ordinal_accuracy": 0.6845620274543762,
      "eval_ordinal_mae": 0.3699798583984375,
      "eval_runtime": 185.1462,
      "eval_samples_per_second": 24.17,
      "eval_steps_per_second": 3.025,
      "step": 4700
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7951823472976685,
      "learning_rate": 4.32383699963938e-05,
      "loss": 0.1335,
      "step": 4725
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.5750038027763367,
      "learning_rate": 4.2937853107344634e-05,
      "loss": 0.1225,
      "step": 4750
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.599660336971283,
      "learning_rate": 4.263733621829547e-05,
      "loss": 0.1153,
      "step": 4775
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.5971084237098694,
      "learning_rate": 4.233681932924631e-05,
      "loss": 0.1264,
      "step": 4800
    },
    {
      "epoch": 1.73,
      "eval_loss": 0.22111621499061584,
      "eval_na_accuracy": 0.8350515365600586,
      "eval_ordinal_accuracy": 0.6891857385635376,
      "eval_ordinal_mae": 0.36766940355300903,
      "eval_runtime": 187.791,
      "eval_samples_per_second": 23.83,
      "eval_steps_per_second": 2.982,
      "step": 4800
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.6335272789001465,
      "learning_rate": 4.203630244019714e-05,
      "loss": 0.1309,
      "step": 4825
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4906469583511353,
      "learning_rate": 4.1735785551147974e-05,
      "loss": 0.1481,
      "step": 4850
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.5834776163101196,
      "learning_rate": 4.143526866209881e-05,
      "loss": 0.1082,
      "step": 4875
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.3094954788684845,
      "learning_rate": 4.1134751773049644e-05,
      "loss": 0.1404,
      "step": 4900
    },
    {
      "epoch": 1.77,
      "eval_loss": 0.22064529359340668,
      "eval_na_accuracy": 0.7938144207000732,
      "eval_ordinal_accuracy": 0.6945800185203552,
      "eval_ordinal_mae": 0.371782511472702,
      "eval_runtime": 187.3419,
      "eval_samples_per_second": 23.887,
      "eval_steps_per_second": 2.989,
      "step": 4900
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.43263766169548035,
      "learning_rate": 4.083423488400048e-05,
      "loss": 0.0901,
      "step": 4925
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0660347938537598,
      "learning_rate": 4.053371799495132e-05,
      "loss": 0.1197,
      "step": 4950
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7330833077430725,
      "learning_rate": 4.023320110590215e-05,
      "loss": 0.1052,
      "step": 4975
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.152076005935669,
      "learning_rate": 3.993268421685299e-05,
      "loss": 0.1238,
      "step": 5000
    },
    {
      "epoch": 1.8,
      "eval_loss": 0.20976723730564117,
      "eval_na_accuracy": 0.8264604806900024,
      "eval_ordinal_accuracy": 0.6948369145393372,
      "eval_ordinal_mae": 0.37225744128227234,
      "eval_runtime": 186.1293,
      "eval_samples_per_second": 24.042,
      "eval_steps_per_second": 3.009,
      "step": 5000
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.28938254714012146,
      "learning_rate": 3.9632167327803824e-05,
      "loss": 0.1089,
      "step": 5025
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.5542977452278137,
      "learning_rate": 3.9331650438754655e-05,
      "loss": 0.1034,
      "step": 5050
    },
    {
      "epoch": 1.83,
      "grad_norm": 3.0670604705810547,
      "learning_rate": 3.9031133549705494e-05,
      "loss": 0.1207,
      "step": 5075
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1476399898529053,
      "learning_rate": 3.873061666065633e-05,
      "loss": 0.0868,
      "step": 5100
    },
    {
      "epoch": 1.84,
      "eval_loss": 0.2088630199432373,
      "eval_na_accuracy": 0.8144329786300659,
      "eval_ordinal_accuracy": 0.7025430202484131,
      "eval_ordinal_mae": 0.3573513329029083,
      "eval_runtime": 186.6283,
      "eval_samples_per_second": 23.978,
      "eval_steps_per_second": 3.001,
      "step": 5100
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.155088186264038,
      "learning_rate": 3.8430099771607164e-05,
      "loss": 0.1285,
      "step": 5125
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.079723358154297,
      "learning_rate": 3.8129582882558e-05,
      "loss": 0.1503,
      "step": 5150
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3054969310760498,
      "learning_rate": 3.782906599350884e-05,
      "loss": 0.1117,
      "step": 5175
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.676068902015686,
      "learning_rate": 3.752854910445967e-05,
      "loss": 0.0828,
      "step": 5200
    },
    {
      "epoch": 1.88,
      "eval_loss": 0.2203822135925293,
      "eval_na_accuracy": 0.7817869186401367,
      "eval_ordinal_accuracy": 0.7030567526817322,
      "eval_ordinal_mae": 0.3679908215999603,
      "eval_runtime": 186.9501,
      "eval_samples_per_second": 23.937,
      "eval_steps_per_second": 2.995,
      "step": 5200
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.5627509355545044,
      "learning_rate": 3.7228032215410505e-05,
      "loss": 0.1013,
      "step": 5225
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.74530029296875,
      "learning_rate": 3.6927515326361343e-05,
      "loss": 0.1163,
      "step": 5250
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.4432946741580963,
      "learning_rate": 3.6626998437312175e-05,
      "loss": 0.1076,
      "step": 5275
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8598417639732361,
      "learning_rate": 3.6326481548263014e-05,
      "loss": 0.0986,
      "step": 5300
    },
    {
      "epoch": 1.91,
      "eval_loss": 0.21255970001220703,
      "eval_na_accuracy": 0.8127147555351257,
      "eval_ordinal_accuracy": 0.6981762051582336,
      "eval_ordinal_mae": 0.35426145792007446,
      "eval_runtime": 188.2634,
      "eval_samples_per_second": 23.77,
      "eval_steps_per_second": 2.975,
      "step": 5300
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.5436437726020813,
      "learning_rate": 3.602596465921385e-05,
      "loss": 0.1368,
      "step": 5325
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.3918437063694,
      "learning_rate": 3.5725447770164684e-05,
      "loss": 0.1062,
      "step": 5350
    },
    {
      "epoch": 1.94,
      "grad_norm": 3.4165215492248535,
      "learning_rate": 3.542493088111552e-05,
      "loss": 0.1394,
      "step": 5375
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.5507489442825317,
      "learning_rate": 3.5124413992066354e-05,
      "loss": 0.0869,
      "step": 5400
    },
    {
      "epoch": 1.95,
      "eval_loss": 0.2247212827205658,
      "eval_na_accuracy": 0.80756014585495,
      "eval_ordinal_accuracy": 0.7107629179954529,
      "eval_ordinal_mae": 0.35320159792900085,
      "eval_runtime": 186.4594,
      "eval_samples_per_second": 24.0,
      "eval_steps_per_second": 3.003,
      "step": 5400
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.36401164531707764,
      "learning_rate": 3.4823897103017186e-05,
      "loss": 0.1216,
      "step": 5425
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.5114885568618774,
      "learning_rate": 3.4523380213968025e-05,
      "loss": 0.1034,
      "step": 5450
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.5193465352058411,
      "learning_rate": 3.422286332491886e-05,
      "loss": 0.095,
      "step": 5475
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.3319372236728668,
      "learning_rate": 3.3922346435869695e-05,
      "loss": 0.1006,
      "step": 5500
    },
    {
      "epoch": 1.98,
      "eval_loss": 0.22681304812431335,
      "eval_na_accuracy": 0.8161512017250061,
      "eval_ordinal_accuracy": 0.702799916267395,
      "eval_ordinal_mae": 0.3637341260910034,
      "eval_runtime": 187.993,
      "eval_samples_per_second": 23.804,
      "eval_steps_per_second": 2.979,
      "step": 5500
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.6313589215278625,
      "learning_rate": 3.3621829546820533e-05,
      "loss": 0.1265,
      "step": 5525
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3126126825809479,
      "learning_rate": 3.332131265777137e-05,
      "loss": 0.1002,
      "step": 5550
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.7703510522842407,
      "learning_rate": 3.3020795768722204e-05,
      "loss": 0.065,
      "step": 5575
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.5371888875961304,
      "learning_rate": 3.2720278879673036e-05,
      "loss": 0.0639,
      "step": 5600
    },
    {
      "epoch": 2.02,
      "eval_loss": 0.22521378099918365,
      "eval_na_accuracy": 0.8109965920448303,
      "eval_ordinal_accuracy": 0.7069098353385925,
      "eval_ordinal_mae": 0.3478536903858185,
      "eval_runtime": 186.1883,
      "eval_samples_per_second": 24.035,
      "eval_steps_per_second": 3.008,
      "step": 5600
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.3779759109020233,
      "learning_rate": 3.2419761990623874e-05,
      "loss": 0.0491,
      "step": 5625
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.3881787657737732,
      "learning_rate": 3.2119245101574706e-05,
      "loss": 0.0475,
      "step": 5650
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.2288794368505478,
      "learning_rate": 3.1818728212525544e-05,
      "loss": 0.062,
      "step": 5675
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.4537404477596283,
      "learning_rate": 3.151821132347638e-05,
      "loss": 0.0569,
      "step": 5700
    },
    {
      "epoch": 2.06,
      "eval_loss": 0.23154324293136597,
      "eval_na_accuracy": 0.80756014585495,
      "eval_ordinal_accuracy": 0.7166709303855896,
      "eval_ordinal_mae": 0.3399028480052948,
      "eval_runtime": 188.0319,
      "eval_samples_per_second": 23.799,
      "eval_steps_per_second": 2.978,
      "step": 5700
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.4112221300601959,
      "learning_rate": 3.1217694434427215e-05,
      "loss": 0.0513,
      "step": 5725
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.4015844166278839,
      "learning_rate": 3.091717754537805e-05,
      "loss": 0.0608,
      "step": 5750
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.43563222885131836,
      "learning_rate": 3.0616660656328885e-05,
      "loss": 0.0463,
      "step": 5775
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.328296422958374,
      "learning_rate": 3.031614376727972e-05,
      "loss": 0.0626,
      "step": 5800
    },
    {
      "epoch": 2.09,
      "eval_loss": 0.2304299920797348,
      "eval_na_accuracy": 0.8127147555351257,
      "eval_ordinal_accuracy": 0.702799916267395,
      "eval_ordinal_mae": 0.34806403517723083,
      "eval_runtime": 186.3734,
      "eval_samples_per_second": 24.011,
      "eval_steps_per_second": 3.005,
      "step": 5800
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.5192309021949768,
      "learning_rate": 3.001562687823056e-05,
      "loss": 0.0506,
      "step": 5825
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.7943512201309204,
      "learning_rate": 2.9715109989181394e-05,
      "loss": 0.0612,
      "step": 5850
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.5543213486671448,
      "learning_rate": 2.9414593100132226e-05,
      "loss": 0.0571,
      "step": 5875
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.32637158036231995,
      "learning_rate": 2.9114076211083064e-05,
      "loss": 0.0502,
      "step": 5900
    },
    {
      "epoch": 2.13,
      "eval_loss": 0.23814311623573303,
      "eval_na_accuracy": 0.8092783689498901,
      "eval_ordinal_accuracy": 0.6953506469726562,
      "eval_ordinal_mae": 0.3623509407043457,
      "eval_runtime": 185.1929,
      "eval_samples_per_second": 24.164,
      "eval_steps_per_second": 3.024,
      "step": 5900
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.4265578091144562,
      "learning_rate": 2.88135593220339e-05,
      "loss": 0.0529,
      "step": 5925
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.5854742527008057,
      "learning_rate": 2.8513042432984738e-05,
      "loss": 0.0458,
      "step": 5950
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.6325811147689819,
      "learning_rate": 2.821252554393557e-05,
      "loss": 0.0422,
      "step": 5975
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.3043919503688812,
      "learning_rate": 2.7912008654886408e-05,
      "loss": 0.0541,
      "step": 6000
    },
    {
      "epoch": 2.16,
      "eval_loss": 0.2298140674829483,
      "eval_na_accuracy": 0.8109965920448303,
      "eval_ordinal_accuracy": 0.7159003615379333,
      "eval_ordinal_mae": 0.34048062562942505,
      "eval_runtime": 188.0938,
      "eval_samples_per_second": 23.791,
      "eval_steps_per_second": 2.977,
      "step": 6000
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.5921896696090698,
      "learning_rate": 2.7611491765837243e-05,
      "loss": 0.0441,
      "step": 6025
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.5739880204200745,
      "learning_rate": 2.7310974876788075e-05,
      "loss": 0.0399,
      "step": 6050
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.6747081279754639,
      "learning_rate": 2.7010457987738914e-05,
      "loss": 0.0467,
      "step": 6075
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.5710461139678955,
      "learning_rate": 2.670994109868975e-05,
      "loss": 0.0671,
      "step": 6100
    },
    {
      "epoch": 2.2,
      "eval_loss": 0.24321676790714264,
      "eval_na_accuracy": 0.7989690899848938,
      "eval_ordinal_accuracy": 0.7030567526817322,
      "eval_ordinal_mae": 0.3529086709022522,
      "eval_runtime": 188.2916,
      "eval_samples_per_second": 23.766,
      "eval_steps_per_second": 2.974,
      "step": 6100
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.3675624430179596,
      "learning_rate": 2.640942420964058e-05,
      "loss": 0.0466,
      "step": 6125
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.31548529863357544,
      "learning_rate": 2.610890732059142e-05,
      "loss": 0.0764,
      "step": 6150
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.43977534770965576,
      "learning_rate": 2.5808390431542258e-05,
      "loss": 0.0475,
      "step": 6175
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.6533438563346863,
      "learning_rate": 2.550787354249309e-05,
      "loss": 0.0672,
      "step": 6200
    },
    {
      "epoch": 2.24,
      "eval_loss": 0.2430579513311386,
      "eval_na_accuracy": 0.7714776396751404,
      "eval_ordinal_accuracy": 0.7194965481758118,
      "eval_ordinal_mae": 0.3360508680343628,
      "eval_runtime": 186.7478,
      "eval_samples_per_second": 23.963,
      "eval_steps_per_second": 2.999,
      "step": 6200
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.2441914677619934,
      "learning_rate": 2.5207356653443925e-05,
      "loss": 0.0503,
      "step": 6225
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.3184777796268463,
      "learning_rate": 2.490683976439476e-05,
      "loss": 0.0641,
      "step": 6250
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.5295473337173462,
      "learning_rate": 2.4606322875345595e-05,
      "loss": 0.0453,
      "step": 6275
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.6061444282531738,
      "learning_rate": 2.430580598629643e-05,
      "loss": 0.0446,
      "step": 6300
    },
    {
      "epoch": 2.27,
      "eval_loss": 0.24466152489185333,
      "eval_na_accuracy": 0.7938144207000732,
      "eval_ordinal_accuracy": 0.7141022086143494,
      "eval_ordinal_mae": 0.34008586406707764,
      "eval_runtime": 184.0098,
      "eval_samples_per_second": 24.319,
      "eval_steps_per_second": 3.043,
      "step": 6300
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.7771974802017212,
      "learning_rate": 2.4005289097247265e-05,
      "loss": 0.0506,
      "step": 6325
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.48779726028442383,
      "learning_rate": 2.37047722081981e-05,
      "loss": 0.0625,
      "step": 6350
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.5689972639083862,
      "learning_rate": 2.340425531914894e-05,
      "loss": 0.0504,
      "step": 6375
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.35768938064575195,
      "learning_rate": 2.310373843009977e-05,
      "loss": 0.0424,
      "step": 6400
    },
    {
      "epoch": 2.31,
      "eval_loss": 0.24263423681259155,
      "eval_na_accuracy": 0.8161512017250061,
      "eval_ordinal_accuracy": 0.7017723917961121,
      "eval_ordinal_mae": 0.34848281741142273,
      "eval_runtime": 184.8006,
      "eval_samples_per_second": 24.215,
      "eval_steps_per_second": 3.03,
      "step": 6400
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.28792133927345276,
      "learning_rate": 2.280322154105061e-05,
      "loss": 0.0444,
      "step": 6425
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.5131263136863708,
      "learning_rate": 2.2502704652001444e-05,
      "loss": 0.0494,
      "step": 6450
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.4703335464000702,
      "learning_rate": 2.220218776295228e-05,
      "loss": 0.0568,
      "step": 6475
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.48926159739494324,
      "learning_rate": 2.1901670873903115e-05,
      "loss": 0.0386,
      "step": 6500
    },
    {
      "epoch": 2.34,
      "eval_loss": 0.24884438514709473,
      "eval_na_accuracy": 0.8127147555351257,
      "eval_ordinal_accuracy": 0.7123041152954102,
      "eval_ordinal_mae": 0.33866390585899353,
      "eval_runtime": 182.7756,
      "eval_samples_per_second": 24.484,
      "eval_steps_per_second": 3.064,
      "step": 6500
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.7936327457427979,
      "learning_rate": 2.160115398485395e-05,
      "loss": 0.0553,
      "step": 6525
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.41186073422431946,
      "learning_rate": 2.1300637095804785e-05,
      "loss": 0.0441,
      "step": 6550
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.7716944813728333,
      "learning_rate": 2.1000120206755623e-05,
      "loss": 0.0562,
      "step": 6575
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.40576568245887756,
      "learning_rate": 2.0699603317706455e-05,
      "loss": 0.0736,
      "step": 6600
    },
    {
      "epoch": 2.38,
      "eval_loss": 0.24542821943759918,
      "eval_na_accuracy": 0.831615149974823,
      "eval_ordinal_accuracy": 0.7053686380386353,
      "eval_ordinal_mae": 0.3381500244140625,
      "eval_runtime": 188.0054,
      "eval_samples_per_second": 23.803,
      "eval_steps_per_second": 2.979,
      "step": 6600
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.147202968597412,
      "learning_rate": 2.039908642865729e-05,
      "loss": 0.0552,
      "step": 6625
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0888266563415527,
      "learning_rate": 2.009856953960813e-05,
      "loss": 0.0598,
      "step": 6650
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.6044988036155701,
      "learning_rate": 1.9798052650558964e-05,
      "loss": 0.039,
      "step": 6675
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.5873610973358154,
      "learning_rate": 1.9497535761509796e-05,
      "loss": 0.0421,
      "step": 6700
    },
    {
      "epoch": 2.42,
      "eval_loss": 0.25130367279052734,
      "eval_na_accuracy": 0.831615149974823,
      "eval_ordinal_accuracy": 0.712047278881073,
      "eval_ordinal_mae": 0.3393591046333313,
      "eval_runtime": 185.2317,
      "eval_samples_per_second": 24.159,
      "eval_steps_per_second": 3.023,
      "step": 6700
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.5890676975250244,
      "learning_rate": 1.9197018872460634e-05,
      "loss": 0.0717,
      "step": 6725
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.3779636025428772,
      "learning_rate": 1.889650198341147e-05,
      "loss": 0.0559,
      "step": 6750
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.38733378052711487,
      "learning_rate": 1.8595985094362305e-05,
      "loss": 0.0458,
      "step": 6775
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.6435078978538513,
      "learning_rate": 1.829546820531314e-05,
      "loss": 0.0607,
      "step": 6800
    },
    {
      "epoch": 2.45,
      "eval_loss": 0.2546432614326477,
      "eval_na_accuracy": 0.8264604806900024,
      "eval_ordinal_accuracy": 0.7092216610908508,
      "eval_ordinal_mae": 0.33696386218070984,
      "eval_runtime": 185.2223,
      "eval_samples_per_second": 24.16,
      "eval_steps_per_second": 3.023,
      "step": 6800
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.34649285674095154,
      "learning_rate": 1.7994951316263975e-05,
      "loss": 0.0459,
      "step": 6825
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.7860547304153442,
      "learning_rate": 1.769443442721481e-05,
      "loss": 0.0483,
      "step": 6850
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.4681091904640198,
      "learning_rate": 1.7393917538165645e-05,
      "loss": 0.0481,
      "step": 6875
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.788452386856079,
      "learning_rate": 1.709340064911648e-05,
      "loss": 0.0517,
      "step": 6900
    },
    {
      "epoch": 2.49,
      "eval_loss": 0.25944724678993225,
      "eval_na_accuracy": 0.8298969268798828,
      "eval_ordinal_accuracy": 0.7081941962242126,
      "eval_ordinal_mae": 0.3375920057296753,
      "eval_runtime": 187.8946,
      "eval_samples_per_second": 23.817,
      "eval_steps_per_second": 2.98,
      "step": 6900
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.35276201367378235,
      "learning_rate": 1.6792883760067316e-05,
      "loss": 0.0602,
      "step": 6925
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.607291042804718,
      "learning_rate": 1.6492366871018154e-05,
      "loss": 0.0426,
      "step": 6950
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.29660192131996155,
      "learning_rate": 1.6191849981968986e-05,
      "loss": 0.0499,
      "step": 6975
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.6227554082870483,
      "learning_rate": 1.589133309291982e-05,
      "loss": 0.062,
      "step": 7000
    },
    {
      "epoch": 2.52,
      "eval_loss": 0.2532944679260254,
      "eval_na_accuracy": 0.8109965920448303,
      "eval_ordinal_accuracy": 0.710506021976471,
      "eval_ordinal_mae": 0.33686304092407227,
      "eval_runtime": 184.4841,
      "eval_samples_per_second": 24.257,
      "eval_steps_per_second": 3.035,
      "step": 7000
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.37592121958732605,
      "learning_rate": 1.559081620387066e-05,
      "loss": 0.0529,
      "step": 7025
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.7044575810432434,
      "learning_rate": 1.5290299314821495e-05,
      "loss": 0.0555,
      "step": 7050
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.47716042399406433,
      "learning_rate": 1.4989782425772328e-05,
      "loss": 0.0424,
      "step": 7075
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.6846262812614441,
      "learning_rate": 1.4689265536723165e-05,
      "loss": 0.0664,
      "step": 7100
    },
    {
      "epoch": 2.56,
      "eval_loss": 0.2534230649471283,
      "eval_na_accuracy": 0.8024054765701294,
      "eval_ordinal_accuracy": 0.7184690237045288,
      "eval_ordinal_mae": 0.33289140462875366,
      "eval_runtime": 185.2999,
      "eval_samples_per_second": 24.15,
      "eval_steps_per_second": 3.022,
      "step": 7100
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8216082453727722,
      "learning_rate": 1.4388748647674e-05,
      "loss": 0.0481,
      "step": 7125
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8088984489440918,
      "learning_rate": 1.4088231758624834e-05,
      "loss": 0.0575,
      "step": 7150
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.7712150812149048,
      "learning_rate": 1.3787714869575672e-05,
      "loss": 0.0464,
      "step": 7175
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.2501738369464874,
      "learning_rate": 1.3487197980526506e-05,
      "loss": 0.0389,
      "step": 7200
    },
    {
      "epoch": 2.6,
      "eval_loss": 0.246970534324646,
      "eval_na_accuracy": 0.8092783689498901,
      "eval_ordinal_accuracy": 0.7259182929992676,
      "eval_ordinal_mae": 0.328827440738678,
      "eval_runtime": 185.0735,
      "eval_samples_per_second": 24.18,
      "eval_steps_per_second": 3.026,
      "step": 7200
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.6835588216781616,
      "learning_rate": 1.318668109147734e-05,
      "loss": 0.0475,
      "step": 7225
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.5996441841125488,
      "learning_rate": 1.2886164202428178e-05,
      "loss": 0.0443,
      "step": 7250
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0018677711486816,
      "learning_rate": 1.2585647313379013e-05,
      "loss": 0.0425,
      "step": 7275
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.300843209028244,
      "learning_rate": 1.2285130424329848e-05,
      "loss": 0.0671,
      "step": 7300
    },
    {
      "epoch": 2.63,
      "eval_loss": 0.25160375237464905,
      "eval_na_accuracy": 0.8041236996650696,
      "eval_ordinal_accuracy": 0.7159003615379333,
      "eval_ordinal_mae": 0.3293640613555908,
      "eval_runtime": 186.9929,
      "eval_samples_per_second": 23.931,
      "eval_steps_per_second": 2.995,
      "step": 7300
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.677975058555603,
      "learning_rate": 1.1984613535280683e-05,
      "loss": 0.0438,
      "step": 7325
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.45172086358070374,
      "learning_rate": 1.1684096646231518e-05,
      "loss": 0.0503,
      "step": 7350
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8617995381355286,
      "learning_rate": 1.1383579757182353e-05,
      "loss": 0.0426,
      "step": 7375
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.47193852066993713,
      "learning_rate": 1.108306286813319e-05,
      "loss": 0.0416,
      "step": 7400
    },
    {
      "epoch": 2.67,
      "eval_loss": 0.25071558356285095,
      "eval_na_accuracy": 0.8058419227600098,
      "eval_ordinal_accuracy": 0.7133316397666931,
      "eval_ordinal_mae": 0.33067333698272705,
      "eval_runtime": 187.7549,
      "eval_samples_per_second": 23.834,
      "eval_steps_per_second": 2.983,
      "step": 7400
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0205371379852295,
      "learning_rate": 1.0782545979084024e-05,
      "loss": 0.0495,
      "step": 7425
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8885800838470459,
      "learning_rate": 1.048202909003486e-05,
      "loss": 0.0407,
      "step": 7450
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.6127232313156128,
      "learning_rate": 1.0181512200985696e-05,
      "loss": 0.0611,
      "step": 7475
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.7351526021957397,
      "learning_rate": 9.880995311936533e-06,
      "loss": 0.0541,
      "step": 7500
    },
    {
      "epoch": 2.7,
      "eval_loss": 0.25292566418647766,
      "eval_na_accuracy": 0.8058419227600098,
      "eval_ordinal_accuracy": 0.71101975440979,
      "eval_ordinal_mae": 0.335502564907074,
      "eval_runtime": 185.7438,
      "eval_samples_per_second": 24.092,
      "eval_steps_per_second": 3.015,
      "step": 7500
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.715017557144165,
      "learning_rate": 9.580478422887366e-06,
      "loss": 0.0581,
      "step": 7525
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.2852751910686493,
      "learning_rate": 9.279961533838203e-06,
      "loss": 0.0507,
      "step": 7550
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.5301045775413513,
      "learning_rate": 8.979444644789038e-06,
      "loss": 0.0427,
      "step": 7575
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.3962666094303131,
      "learning_rate": 8.678927755739873e-06,
      "loss": 0.0374,
      "step": 7600
    },
    {
      "epoch": 2.74,
      "eval_loss": 0.2529982626438141,
      "eval_na_accuracy": 0.8109965920448303,
      "eval_ordinal_accuracy": 0.7148728370666504,
      "eval_ordinal_mae": 0.33152100443840027,
      "eval_runtime": 184.9088,
      "eval_samples_per_second": 24.201,
      "eval_steps_per_second": 3.029,
      "step": 7600
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.5724599957466125,
      "learning_rate": 8.378410866690708e-06,
      "loss": 0.0383,
      "step": 7625
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.6337906718254089,
      "learning_rate": 8.077893977641545e-06,
      "loss": 0.0339,
      "step": 7650
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.645459771156311,
      "learning_rate": 7.777377088592379e-06,
      "loss": 0.0485,
      "step": 7675
    },
    {
      "epoch": 2.78,
      "grad_norm": 5.039384841918945,
      "learning_rate": 7.476860199543215e-06,
      "loss": 0.04,
      "step": 7700
    },
    {
      "epoch": 2.78,
      "eval_loss": 0.2519625723361969,
      "eval_na_accuracy": 0.80756014585495,
      "eval_ordinal_accuracy": 0.7166709303855896,
      "eval_ordinal_mae": 0.32904064655303955,
      "eval_runtime": 188.802,
      "eval_samples_per_second": 23.702,
      "eval_steps_per_second": 2.966,
      "step": 7700
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.30601269006729126,
      "learning_rate": 7.176343310494051e-06,
      "loss": 0.053,
      "step": 7725
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.6143732070922852,
      "learning_rate": 6.875826421444885e-06,
      "loss": 0.0433,
      "step": 7750
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.5583890080451965,
      "learning_rate": 6.575309532395721e-06,
      "loss": 0.0534,
      "step": 7775
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.5204710960388184,
      "learning_rate": 6.274792643346557e-06,
      "loss": 0.0507,
      "step": 7800
    },
    {
      "epoch": 2.81,
      "eval_loss": 0.2555212676525116,
      "eval_na_accuracy": 0.8127147555351257,
      "eval_ordinal_accuracy": 0.710506021976471,
      "eval_ordinal_mae": 0.3297020196914673,
      "eval_runtime": 185.0982,
      "eval_samples_per_second": 24.176,
      "eval_steps_per_second": 3.025,
      "step": 7800
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.11941999197006226,
      "learning_rate": 5.974275754297392e-06,
      "loss": 0.0392,
      "step": 7825
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.20734691619873047,
      "learning_rate": 5.673758865248227e-06,
      "loss": 0.06,
      "step": 7850
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.48085981607437134,
      "learning_rate": 5.373241976199063e-06,
      "loss": 0.0502,
      "step": 7875
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.5097972750663757,
      "learning_rate": 5.072725087149898e-06,
      "loss": 0.0379,
      "step": 7900
    },
    {
      "epoch": 2.85,
      "eval_loss": 0.2531285285949707,
      "eval_na_accuracy": 0.8127147555351257,
      "eval_ordinal_accuracy": 0.7161571979522705,
      "eval_ordinal_mae": 0.3273853659629822,
      "eval_runtime": 183.7697,
      "eval_samples_per_second": 24.351,
      "eval_steps_per_second": 3.047,
      "step": 7900
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.1783066987991333,
      "learning_rate": 4.7722081981007335e-06,
      "loss": 0.0455,
      "step": 7925
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.6303412318229675,
      "learning_rate": 4.4716913090515695e-06,
      "loss": 0.0639,
      "step": 7950
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.7201049327850342,
      "learning_rate": 4.171174420002405e-06,
      "loss": 0.0373,
      "step": 7975
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.31524857878685,
      "learning_rate": 3.87065753095324e-06,
      "loss": 0.0736,
      "step": 8000
    },
    {
      "epoch": 2.88,
      "eval_loss": 0.25256654620170593,
      "eval_na_accuracy": 0.8195876479148865,
      "eval_ordinal_accuracy": 0.7164140939712524,
      "eval_ordinal_mae": 0.3279343843460083,
      "eval_runtime": 187.2067,
      "eval_samples_per_second": 23.904,
      "eval_steps_per_second": 2.991,
      "step": 8000
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.1482267528772354,
      "learning_rate": 3.5701406419040754e-06,
      "loss": 0.0363,
      "step": 8025
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.6438891887664795,
      "learning_rate": 3.2696237528549105e-06,
      "loss": 0.043,
      "step": 8050
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.5657501816749573,
      "learning_rate": 2.969106863805746e-06,
      "loss": 0.0351,
      "step": 8075
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.35640329122543335,
      "learning_rate": 2.6685899747565817e-06,
      "loss": 0.0589,
      "step": 8100
    },
    {
      "epoch": 2.92,
      "eval_loss": 0.2521790862083435,
      "eval_na_accuracy": 0.8161512017250061,
      "eval_ordinal_accuracy": 0.7143591046333313,
      "eval_ordinal_mae": 0.3267035186290741,
      "eval_runtime": 185.6368,
      "eval_samples_per_second": 24.106,
      "eval_steps_per_second": 3.017,
      "step": 8100
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.6037762761116028,
      "learning_rate": 2.368073085707417e-06,
      "loss": 0.052,
      "step": 8125
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.5858399868011475,
      "learning_rate": 2.067556196658252e-06,
      "loss": 0.043,
      "step": 8150
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9954311847686768,
      "learning_rate": 1.7670393076090878e-06,
      "loss": 0.0708,
      "step": 8175
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.448538601398468,
      "learning_rate": 1.4665224185599232e-06,
      "loss": 0.0449,
      "step": 8200
    },
    {
      "epoch": 2.96,
      "eval_loss": 0.252143532037735,
      "eval_na_accuracy": 0.8161512017250061,
      "eval_ordinal_accuracy": 0.7148728370666504,
      "eval_ordinal_mae": 0.3271646499633789,
      "eval_runtime": 188.3236,
      "eval_samples_per_second": 23.762,
      "eval_steps_per_second": 2.974,
      "step": 8200
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.3916318416595459,
      "learning_rate": 1.1660055295107585e-06,
      "loss": 0.0519,
      "step": 8225
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.7591487765312195,
      "learning_rate": 8.654886404615941e-07,
      "loss": 0.0447,
      "step": 8250
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.33720019459724426,
      "learning_rate": 5.649717514124295e-07,
      "loss": 0.0486,
      "step": 8275
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.5342977046966553,
      "learning_rate": 2.644548623632648e-07,
      "loss": 0.0498,
      "step": 8300
    },
    {
      "epoch": 2.99,
      "eval_loss": 0.2520281970500946,
      "eval_na_accuracy": 0.8144329786300659,
      "eval_ordinal_accuracy": 0.7166709303855896,
      "eval_ordinal_mae": 0.3264659643173218,
      "eval_runtime": 186.96,
      "eval_samples_per_second": 23.936,
      "eval_steps_per_second": 2.995,
      "step": 8300
    },
    {
      "epoch": 3.0,
      "step": 8319,
      "total_flos": 1.0314863567841853e+19,
      "train_loss": 0.13915026724546314,
      "train_runtime": 27077.9759,
      "train_samples_per_second": 4.916,
      "train_steps_per_second": 0.307
    }
  ],
  "logging_steps": 25,
  "max_steps": 8319,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 1.0314863567841853e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}