{
  "best_global_step": 7500,
  "best_metric": 0.5665329098701477,
  "best_model_checkpoint": "lfm2.5-230m-code-math/checkpoint-7500",
  "epoch": 1.999466808850973,
  "eval_steps": 100,
  "global_step": 7500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 0.48585736870765683,
      "epoch": 0.006664889362836577,
      "grad_norm": 15.625,
      "learning_rate": 2.6607538802660755e-06,
      "loss": 1.0848355102539062,
      "mean_token_accuracy": 0.7753344154357911,
      "num_tokens": 140976.0,
      "step": 25
    },
    {
      "entropy": 0.5108509111404419,
      "epoch": 0.013329778725673154,
      "grad_norm": 5.03125,
      "learning_rate": 5.432372505543237e-06,
      "loss": 0.8411404418945313,
      "mean_token_accuracy": 0.8020103454589844,
      "num_tokens": 287130.0,
      "step": 50
    },
    {
      "entropy": 0.7612639474868774,
      "epoch": 0.01999466808850973,
      "grad_norm": 4.03125,
      "learning_rate": 8.2039911308204e-06,
      "loss": 0.7567089080810547,
      "mean_token_accuracy": 0.8073056840896606,
      "num_tokens": 432463.0,
      "step": 75
    },
    {
      "entropy": 0.7406812691688538,
      "epoch": 0.026659557451346308,
      "grad_norm": 4.09375,
      "learning_rate": 1.0975609756097562e-05,
      "loss": 0.7424747467041015,
      "mean_token_accuracy": 0.8084338903427124,
      "num_tokens": 580587.0,
      "step": 100
    },
    {
      "epoch": 0.026659557451346308,
      "eval_entropy": 0.7239241289972055,
      "eval_loss": 0.7141907215118408,
      "eval_mean_token_accuracy": 0.8137571396249713,
      "eval_num_tokens": 580587.0,
      "eval_runtime": 9.789,
      "eval_samples_per_second": 645.418,
      "eval_steps_per_second": 20.227,
      "step": 100
    },
    {
      "entropy": 0.733273937702179,
      "epoch": 0.03332444681418288,
      "grad_norm": 3.65625,
      "learning_rate": 1.3747228381374724e-05,
      "loss": 0.7311963653564453,
      "mean_token_accuracy": 0.8083630990982056,
      "num_tokens": 728849.0,
      "step": 125
    },
    {
      "entropy": 0.71931391954422,
      "epoch": 0.03998933617701946,
      "grad_norm": 3.8125,
      "learning_rate": 1.6518847006651886e-05,
      "loss": 0.7112632751464844,
      "mean_token_accuracy": 0.8148289752006531,
      "num_tokens": 873477.0,
      "step": 150
    },
    {
      "entropy": 0.6910139083862304,
      "epoch": 0.046654225539856035,
      "grad_norm": 4.25,
      "learning_rate": 1.9290465631929047e-05,
      "loss": 0.6897886657714843,
      "mean_token_accuracy": 0.818484103679657,
      "num_tokens": 1020564.0,
      "step": 175
    },
    {
      "entropy": 0.6608000230789185,
      "epoch": 0.053319114902692616,
      "grad_norm": 3.984375,
      "learning_rate": 2.206208425720621e-05,
      "loss": 0.6541932678222656,
      "mean_token_accuracy": 0.8244255447387695,
      "num_tokens": 1160309.0,
      "step": 200
    },
    {
      "epoch": 0.053319114902692616,
      "eval_entropy": 0.6792383206011069,
      "eval_loss": 0.6757866740226746,
      "eval_mean_token_accuracy": 0.821125164477512,
      "eval_num_tokens": 1160309.0,
      "eval_runtime": 9.721,
      "eval_samples_per_second": 649.934,
      "eval_steps_per_second": 20.368,
      "step": 200
    },
    {
      "entropy": 0.6763921451568603,
      "epoch": 0.05998400426552919,
      "grad_norm": 3.765625,
      "learning_rate": 2.483370288248337e-05,
      "loss": 0.6713539123535156,
      "mean_token_accuracy": 0.8222450470924377,
      "num_tokens": 1305852.0,
      "step": 225
    },
    {
      "entropy": 0.6528970718383789,
      "epoch": 0.06664889362836576,
      "grad_norm": 4.03125,
      "learning_rate": 2.7605321507760533e-05,
      "loss": 0.6580439758300781,
      "mean_token_accuracy": 0.822701518535614,
      "num_tokens": 1448257.0,
      "step": 250
    },
    {
      "entropy": 0.6856464743614197,
      "epoch": 0.07331378299120235,
      "grad_norm": 4.15625,
      "learning_rate": 3.0376940133037695e-05,
      "loss": 0.6825908660888672,
      "mean_token_accuracy": 0.817496943473816,
      "num_tokens": 1590234.0,
      "step": 275
    },
    {
      "entropy": 0.6945951557159424,
      "epoch": 0.07997867235403892,
      "grad_norm": 3.890625,
      "learning_rate": 3.314855875831486e-05,
      "loss": 0.6921115112304688,
      "mean_token_accuracy": 0.8131667828559875,
      "num_tokens": 1737490.0,
      "step": 300
    },
    {
      "epoch": 0.07997867235403892,
      "eval_entropy": 0.6649126413494649,
      "eval_loss": 0.6639286279678345,
      "eval_mean_token_accuracy": 0.8230556433249001,
      "eval_num_tokens": 1737490.0,
      "eval_runtime": 9.7054,
      "eval_samples_per_second": 650.978,
      "eval_steps_per_second": 20.401,
      "step": 300
    },
    {
      "entropy": 0.6785938477516175,
      "epoch": 0.0866435617168755,
      "grad_norm": 3.953125,
      "learning_rate": 3.5920177383592015e-05,
      "loss": 0.6837982177734375,
      "mean_token_accuracy": 0.8195662879943848,
      "num_tokens": 1880865.0,
      "step": 325
    },
    {
      "entropy": 0.6712707424163818,
      "epoch": 0.09330845107971207,
      "grad_norm": 3.5,
      "learning_rate": 3.869179600886918e-05,
      "loss": 0.6671878814697265,
      "mean_token_accuracy": 0.8222167539596558,
      "num_tokens": 2026401.0,
      "step": 350
    },
    {
      "entropy": 0.6817171144485473,
      "epoch": 0.09997334044254866,
      "grad_norm": 3.921875,
      "learning_rate": 4.146341463414634e-05,
      "loss": 0.6856563568115235,
      "mean_token_accuracy": 0.8147950649261475,
      "num_tokens": 2170209.0,
      "step": 375
    },
    {
      "entropy": 0.6457146716117859,
      "epoch": 0.10663822980538523,
      "grad_norm": 4.0625,
      "learning_rate": 4.42350332594235e-05,
      "loss": 0.6438507080078125,
      "mean_token_accuracy": 0.8271675634384156,
      "num_tokens": 2319994.0,
      "step": 400
    },
    {
      "epoch": 0.10663822980538523,
      "eval_entropy": 0.6214492423365815,
      "eval_loss": 0.6651297807693481,
      "eval_mean_token_accuracy": 0.8227205405933689,
      "eval_num_tokens": 2319994.0,
      "eval_runtime": 9.7418,
      "eval_samples_per_second": 648.543,
      "eval_steps_per_second": 20.325,
      "step": 400
    },
    {
      "entropy": 0.6514025807380677,
      "epoch": 0.1133031191682218,
      "grad_norm": 3.765625,
      "learning_rate": 4.700665188470067e-05,
      "loss": 0.6549468231201172,
      "mean_token_accuracy": 0.8232432770729065,
      "num_tokens": 2464963.0,
      "step": 425
    },
    {
      "entropy": 0.6629467391967774,
      "epoch": 0.11996800853105838,
      "grad_norm": 3.703125,
      "learning_rate": 4.977827050997783e-05,
      "loss": 0.6708641815185546,
      "mean_token_accuracy": 0.8198106718063355,
      "num_tokens": 2612359.0,
      "step": 450
    },
    {
      "entropy": 0.6615508341789246,
      "epoch": 0.12663289789389495,
      "grad_norm": 3.203125,
      "learning_rate": 4.9999691851910445e-05,
      "loss": 0.6611491394042969,
      "mean_token_accuracy": 0.8197968554496765,
      "num_tokens": 2760338.0,
      "step": 475
    },
    {
      "entropy": 0.6625556635856629,
      "epoch": 0.13329778725673153,
      "grad_norm": 3.765625,
      "learning_rate": 4.999865790490649e-05,
      "loss": 0.6705697631835937,
      "mean_token_accuracy": 0.8206671762466431,
      "num_tokens": 2903498.0,
      "step": 500
    },
    {
      "epoch": 0.13329778725673153,
      "eval_entropy": 0.6753983453969763,
      "eval_loss": 0.6686927676200867,
      "eval_mean_token_accuracy": 0.821351081734956,
      "eval_num_tokens": 2903498.0,
      "eval_runtime": 9.7256,
      "eval_samples_per_second": 649.627,
      "eval_steps_per_second": 20.359,
      "step": 500
    },
    {
      "entropy": 0.6689723253250122,
      "epoch": 0.13996267661956813,
      "grad_norm": 3.390625,
      "learning_rate": 4.9996895859055355e-05,
      "loss": 0.6665790557861329,
      "mean_token_accuracy": 0.8203673839569092,
      "num_tokens": 3049335.0,
      "step": 525
    },
    {
      "entropy": 0.6648555302619934,
      "epoch": 0.1466275659824047,
      "grad_norm": 3.5,
      "learning_rate": 4.999440576567755e-05,
      "loss": 0.6728791046142578,
      "mean_token_accuracy": 0.8201075005531311,
      "num_tokens": 3195398.0,
      "step": 550
    },
    {
      "entropy": 0.6789996480941772,
      "epoch": 0.15329245534524127,
      "grad_norm": 3.453125,
      "learning_rate": 4.9991187697298326e-05,
      "loss": 0.6776419830322266,
      "mean_token_accuracy": 0.8184739995002747,
      "num_tokens": 3341120.0,
      "step": 575
    },
    {
      "entropy": 0.6439703822135925,
      "epoch": 0.15995734470807785,
      "grad_norm": 4.125,
      "learning_rate": 4.9987241747645584e-05,
      "loss": 0.6511991882324218,
      "mean_token_accuracy": 0.8268180704116821,
      "num_tokens": 3483811.0,
      "step": 600
    },
    {
      "epoch": 0.15995734470807785,
      "eval_entropy": 0.6606559083618299,
      "eval_loss": 0.6624135375022888,
      "eval_mean_token_accuracy": 0.8232905292751813,
      "eval_num_tokens": 3483811.0,
      "eval_runtime": 9.7389,
      "eval_samples_per_second": 648.737,
      "eval_steps_per_second": 20.331,
      "step": 600
    },
    {
      "entropy": 0.6882552003860474,
      "epoch": 0.16662223407091442,
      "grad_norm": 3.796875,
      "learning_rate": 4.998256803164717e-05,
      "loss": 0.6866973114013671,
      "mean_token_accuracy": 0.8159659147262573,
      "num_tokens": 3633597.0,
      "step": 625
    },
    {
      "entropy": 0.6741298842430115,
      "epoch": 0.173287123433751,
      "grad_norm": 3.46875,
      "learning_rate": 4.997716668542746e-05,
      "loss": 0.6807684326171874,
      "mean_token_accuracy": 0.8177537393569946,
      "num_tokens": 3776237.0,
      "step": 650
    },
    {
      "entropy": 0.6367644453048706,
      "epoch": 0.17995201279658757,
      "grad_norm": 3.734375,
      "learning_rate": 4.997103786630346e-05,
      "loss": 0.6361703491210937,
      "mean_token_accuracy": 0.8284359860420227,
      "num_tokens": 3922071.0,
      "step": 675
    },
    {
      "entropy": 0.6396224927902222,
      "epoch": 0.18661690215942414,
      "grad_norm": 3.3125,
      "learning_rate": 4.9964181752780216e-05,
      "loss": 0.6398828506469727,
      "mean_token_accuracy": 0.8253889107704162,
      "num_tokens": 4062153.0,
      "step": 700
    },
    {
      "epoch": 0.18661690215942414,
      "eval_entropy": 0.6414893803873447,
      "eval_loss": 0.6569135785102844,
      "eval_mean_token_accuracy": 0.8243622755763507,
      "eval_num_tokens": 4062153.0,
      "eval_runtime": 9.715,
      "eval_samples_per_second": 650.334,
      "eval_steps_per_second": 20.381,
      "step": 700
    },
    {
      "entropy": 0.6423980665206909,
      "epoch": 0.19328179152226074,
      "grad_norm": 4.1875,
      "learning_rate": 4.9956598544545566e-05,
      "loss": 0.6402242279052734,
      "mean_token_accuracy": 0.8264222669601441,
      "num_tokens": 4206173.0,
      "step": 725
    },
    {
      "entropy": 0.6451382398605346,
      "epoch": 0.19994668088509732,
      "grad_norm": 3.453125,
      "learning_rate": 4.994828846246437e-05,
      "loss": 0.6507736968994141,
      "mean_token_accuracy": 0.823756353855133,
      "num_tokens": 4350696.0,
      "step": 750
    },
    {
      "entropy": 0.6703963446617126,
      "epoch": 0.2066115702479339,
      "grad_norm": 3.5625,
      "learning_rate": 4.993925174857208e-05,
      "loss": 0.6725439453125,
      "mean_token_accuracy": 0.8190426468849182,
      "num_tokens": 4496659.0,
      "step": 775
    },
    {
      "entropy": 0.6339899516105652,
      "epoch": 0.21327645961077046,
      "grad_norm": 3.296875,
      "learning_rate": 4.9929488666067656e-05,
      "loss": 0.6331915664672851,
      "mean_token_accuracy": 0.8292251420021057,
      "num_tokens": 4639515.0,
      "step": 800
    },
    {
      "epoch": 0.21327645961077046,
      "eval_entropy": 0.6325301860619073,
      "eval_loss": 0.6490293145179749,
      "eval_mean_token_accuracy": 0.8262747673675267,
      "eval_num_tokens": 4639515.0,
      "eval_runtime": 9.7311,
      "eval_samples_per_second": 649.256,
      "eval_steps_per_second": 20.347,
      "step": 800
    },
    {
      "entropy": 0.6715281128883361,
      "epoch": 0.21994134897360704,
      "grad_norm": 4.1875,
      "learning_rate": 4.991899949930594e-05,
      "loss": 0.6789644622802734,
      "mean_token_accuracy": 0.8192856383323669,
      "num_tokens": 4783565.0,
      "step": 825
    },
    {
      "entropy": 0.6412011981010437,
      "epoch": 0.2266062383364436,
      "grad_norm": 3.265625,
      "learning_rate": 4.990778455378933e-05,
      "loss": 0.6406878662109375,
      "mean_token_accuracy": 0.8290301132202148,
      "num_tokens": 4924741.0,
      "step": 850
    },
    {
      "entropy": 0.6668443393707275,
      "epoch": 0.23327112769928018,
      "grad_norm": 3.421875,
      "learning_rate": 4.9895844156158924e-05,
      "loss": 0.6640223693847657,
      "mean_token_accuracy": 0.8229507517814636,
      "num_tokens": 5072594.0,
      "step": 875
    },
    {
      "entropy": 0.6653201484680176,
      "epoch": 0.23993601706211676,
      "grad_norm": 3.90625,
      "learning_rate": 4.9883178654184975e-05,
      "loss": 0.6695989990234374,
      "mean_token_accuracy": 0.8200922751426697,
      "num_tokens": 5217699.0,
      "step": 900
    },
    {
      "epoch": 0.23993601706211676,
      "eval_entropy": 0.6507695416609446,
      "eval_loss": 0.6448351740837097,
      "eval_mean_token_accuracy": 0.826673719317022,
      "eval_num_tokens": 5217699.0,
      "eval_runtime": 9.7495,
      "eval_samples_per_second": 648.03,
      "eval_steps_per_second": 20.309,
      "step": 900
    },
    {
      "entropy": 0.655668329000473,
      "epoch": 0.24660090642495336,
      "grad_norm": 3.765625,
      "learning_rate": 4.9869788416756794e-05,
      "loss": 0.6561538696289062,
      "mean_token_accuracy": 0.8260146856307984,
      "num_tokens": 5360353.0,
      "step": 925
    },
    {
      "entropy": 0.6484189796447754,
      "epoch": 0.2532657957877899,
      "grad_norm": 3.734375,
      "learning_rate": 4.985567383387195e-05,
      "loss": 0.6564521026611329,
      "mean_token_accuracy": 0.8237798285484313,
      "num_tokens": 5506229.0,
      "step": 950
    },
    {
      "entropy": 0.6464210343360901,
      "epoch": 0.2599306851506265,
      "grad_norm": 3.921875,
      "learning_rate": 4.9840835316625e-05,
      "loss": 0.6414363098144531,
      "mean_token_accuracy": 0.8268412518501281,
      "num_tokens": 5646061.0,
      "step": 975
    },
    {
      "entropy": 0.6468575143814087,
      "epoch": 0.26659557451346305,
      "grad_norm": 3.453125,
      "learning_rate": 4.9825273297195426e-05,
      "loss": 0.650556411743164,
      "mean_token_accuracy": 0.8242916202545166,
      "num_tokens": 5788093.0,
      "step": 1000
    },
    {
      "epoch": 0.26659557451346305,
      "eval_entropy": 0.6249580481136688,
      "eval_loss": 0.6412637829780579,
      "eval_mean_token_accuracy": 0.8275833551329795,
      "eval_num_tokens": 5788093.0,
      "eval_runtime": 9.7134,
      "eval_samples_per_second": 650.443,
      "eval_steps_per_second": 20.384,
      "step": 1000
    },
    {
      "entropy": 0.6641205096244812,
      "epoch": 0.27326046387629965,
      "grad_norm": 3.625,
      "learning_rate": 4.98089882288351e-05,
      "loss": 0.6628132629394531,
      "mean_token_accuracy": 0.8196079039573669,
      "num_tokens": 5926515.0,
      "step": 1025
    },
    {
      "entropy": 0.6678084516525269,
      "epoch": 0.27992535323913625,
      "grad_norm": 3.484375,
      "learning_rate": 4.979198058585507e-05,
      "loss": 0.670301284790039,
      "mean_token_accuracy": 0.8206749320030212,
      "num_tokens": 6070258.0,
      "step": 1050
    },
    {
      "entropy": 0.6584195876121521,
      "epoch": 0.2865902426019728,
      "grad_norm": 3.234375,
      "learning_rate": 4.977425086361172e-05,
      "loss": 0.6617906951904297,
      "mean_token_accuracy": 0.8234014248847962,
      "num_tokens": 6216448.0,
      "step": 1075
    },
    {
      "entropy": 0.6359080982208252,
      "epoch": 0.2932551319648094,
      "grad_norm": 3.0,
      "learning_rate": 4.975579957849239e-05,
      "loss": 0.6384500885009765,
      "mean_token_accuracy": 0.8259626984596252,
      "num_tokens": 6362151.0,
      "step": 1100
    },
    {
      "epoch": 0.2932551319648094,
      "eval_entropy": 0.6356519805361526,
      "eval_loss": 0.6366018652915955,
      "eval_mean_token_accuracy": 0.8290430208047231,
      "eval_num_tokens": 6362151.0,
      "eval_runtime": 9.7203,
      "eval_samples_per_second": 649.977,
      "eval_steps_per_second": 20.37,
      "step": 1100
    },
    {
      "entropy": 0.6123459696769714,
      "epoch": 0.29992002132764595,
      "grad_norm": 3.40625,
      "learning_rate": 4.9736627267900314e-05,
      "loss": 0.6200491333007813,
      "mean_token_accuracy": 0.8320556664466858,
      "num_tokens": 6504435.0,
      "step": 1125
    },
    {
      "entropy": 0.6508487749099732,
      "epoch": 0.30658491069048255,
      "grad_norm": 3.984375,
      "learning_rate": 4.9716734490238964e-05,
      "loss": 0.646004409790039,
      "mean_token_accuracy": 0.8272648477554321,
      "num_tokens": 6652727.0,
      "step": 1150
    },
    {
      "entropy": 0.6557350659370422,
      "epoch": 0.3132498000533191,
      "grad_norm": 3.21875,
      "learning_rate": 4.969612182489577e-05,
      "loss": 0.6570596313476562,
      "mean_token_accuracy": 0.8228574538230896,
      "num_tokens": 6801453.0,
      "step": 1175
    },
    {
      "entropy": 0.6752374565601349,
      "epoch": 0.3199146894161557,
      "grad_norm": 4.3125,
      "learning_rate": 4.967478987222529e-05,
      "loss": 0.6750936889648438,
      "mean_token_accuracy": 0.8177144861221314,
      "num_tokens": 6943864.0,
      "step": 1200
    },
    {
      "epoch": 0.3199146894161557,
      "eval_entropy": 0.638334623641438,
      "eval_loss": 0.6348886489868164,
      "eval_mean_token_accuracy": 0.8285927025958745,
      "eval_num_tokens": 6943864.0,
      "eval_runtime": 9.7282,
      "eval_samples_per_second": 649.451,
      "eval_steps_per_second": 20.353,
      "step": 1200
    },
    {
      "entropy": 0.6177306866645813,
      "epoch": 0.32657957877899224,
      "grad_norm": 3.453125,
      "learning_rate": 4.965273925353167e-05,
      "loss": 0.6229856872558593,
      "mean_token_accuracy": 0.8295817518234253,
      "num_tokens": 7089484.0,
      "step": 1225
    },
    {
      "entropy": 0.6389037203788758,
      "epoch": 0.33324446814182884,
      "grad_norm": 3.390625,
      "learning_rate": 4.96299706110506e-05,
      "loss": 0.6413640594482422,
      "mean_token_accuracy": 0.8270923471450806,
      "num_tokens": 7235186.0,
      "step": 1250
    },
    {
      "entropy": 0.6154287505149841,
      "epoch": 0.33990935750466544,
      "grad_norm": 3.171875,
      "learning_rate": 4.960648460793057e-05,
      "loss": 0.6111659622192382,
      "mean_token_accuracy": 0.8333053183555603,
      "num_tokens": 7377949.0,
      "step": 1275
    },
    {
      "entropy": 0.6321292018890381,
      "epoch": 0.346574246867502,
      "grad_norm": 3.640625,
      "learning_rate": 4.958228192821357e-05,
      "loss": 0.637452392578125,
      "mean_token_accuracy": 0.8257576179504394,
      "num_tokens": 7520757.0,
      "step": 1300
    },
    {
      "epoch": 0.346574246867502,
      "eval_entropy": 0.619249304436674,
      "eval_loss": 0.6298498511314392,
      "eval_mean_token_accuracy": 0.830113782726153,
      "eval_num_tokens": 7520757.0,
      "eval_runtime": 9.748,
      "eval_samples_per_second": 648.134,
      "eval_steps_per_second": 20.312,
      "step": 1300
    },
    {
      "entropy": 0.6197018504142762,
      "epoch": 0.3532391362303386,
      "grad_norm": 3.375,
      "learning_rate": 4.955736327681517e-05,
      "loss": 0.6249597549438477,
      "mean_token_accuracy": 0.8315670347213745,
      "num_tokens": 7665785.0,
      "step": 1325
    },
    {
      "entropy": 0.625680284500122,
      "epoch": 0.35990402559317514,
      "grad_norm": 3.453125,
      "learning_rate": 4.9531729379503996e-05,
      "loss": 0.6232044982910157,
      "mean_token_accuracy": 0.8310526013374329,
      "num_tokens": 7810058.0,
      "step": 1350
    },
    {
      "entropy": 0.6432394456863403,
      "epoch": 0.36656891495601174,
      "grad_norm": 3.25,
      "learning_rate": 4.9505380982880566e-05,
      "loss": 0.6420378875732422,
      "mean_token_accuracy": 0.8282063484191895,
      "num_tokens": 7957328.0,
      "step": 1375
    },
    {
      "entropy": 0.6322685599327087,
      "epoch": 0.3732338043188483,
      "grad_norm": 3.359375,
      "learning_rate": 4.947831885435557e-05,
      "loss": 0.6345468521118164,
      "mean_token_accuracy": 0.8284816265106201,
      "num_tokens": 8100329.0,
      "step": 1400
    },
    {
      "epoch": 0.3732338043188483,
      "eval_entropy": 0.6233037412166595,
      "eval_loss": 0.6273157596588135,
      "eval_mean_token_accuracy": 0.8308055271403958,
      "eval_num_tokens": 8100329.0,
      "eval_runtime": 9.7245,
      "eval_samples_per_second": 649.702,
      "eval_steps_per_second": 20.361,
      "step": 1400
    },
    {
      "entropy": 0.6449450099468231,
      "epoch": 0.3798986936816849,
      "grad_norm": 3.4375,
      "learning_rate": 4.94505437821275e-05,
      "loss": 0.6465570068359375,
      "mean_token_accuracy": 0.82611088514328,
      "num_tokens": 8248137.0,
      "step": 1425
    },
    {
      "entropy": 0.6307063841819763,
      "epoch": 0.3865635830445215,
      "grad_norm": 4.09375,
      "learning_rate": 4.94220565751597e-05,
      "loss": 0.631772346496582,
      "mean_token_accuracy": 0.830638415813446,
      "num_tokens": 8391888.0,
      "step": 1450
    },
    {
      "entropy": 0.6287683236598969,
      "epoch": 0.39322847240735803,
      "grad_norm": 3.40625,
      "learning_rate": 4.939285806315682e-05,
      "loss": 0.6340837097167968,
      "mean_token_accuracy": 0.8278608989715576,
      "num_tokens": 8536717.0,
      "step": 1475
    },
    {
      "entropy": 0.6098583114147186,
      "epoch": 0.39989336177019463,
      "grad_norm": 3.515625,
      "learning_rate": 4.936294909654064e-05,
      "loss": 0.6120164108276367,
      "mean_token_accuracy": 0.8323467993736267,
      "num_tokens": 8680630.0,
      "step": 1500
    },
    {
      "epoch": 0.39989336177019463,
      "eval_entropy": 0.6439532580399754,
      "eval_loss": 0.6256674528121948,
      "eval_mean_token_accuracy": 0.8305191478945992,
      "eval_num_tokens": 8680630.0,
      "eval_runtime": 9.7366,
      "eval_samples_per_second": 648.89,
      "eval_steps_per_second": 20.336,
      "step": 1500
    },
    {
      "entropy": 0.6400356459617614,
      "epoch": 0.4065582511330312,
      "grad_norm": 3.328125,
      "learning_rate": 4.933233054642528e-05,
      "loss": 0.6397489547729492,
      "mean_token_accuracy": 0.8285423231124878,
      "num_tokens": 8825048.0,
      "step": 1525
    },
    {
      "entropy": 0.6009359300136566,
      "epoch": 0.4132231404958678,
      "grad_norm": 3.1875,
      "learning_rate": 4.930100330459186e-05,
      "loss": 0.6036084747314453,
      "mean_token_accuracy": 0.8358356285095215,
      "num_tokens": 8969117.0,
      "step": 1550
    },
    {
      "entropy": 0.6396456003189087,
      "epoch": 0.4198880298587043,
      "grad_norm": 3.34375,
      "learning_rate": 4.92689682834625e-05,
      "loss": 0.6477825927734375,
      "mean_token_accuracy": 0.8254559183120728,
      "num_tokens": 9112823.0,
      "step": 1575
    },
    {
      "entropy": 0.6181868577003479,
      "epoch": 0.4265529192215409,
      "grad_norm": 3.359375,
      "learning_rate": 4.9236226416073784e-05,
      "loss": 0.616800422668457,
      "mean_token_accuracy": 0.8312000560760499,
      "num_tokens": 9258813.0,
      "step": 1600
    },
    {
      "epoch": 0.4265529192215409,
      "eval_entropy": 0.641190735830201,
      "eval_loss": 0.6236028075218201,
      "eval_mean_token_accuracy": 0.8311248879240016,
      "eval_num_tokens": 9258813.0,
      "eval_runtime": 9.748,
      "eval_samples_per_second": 648.133,
      "eval_steps_per_second": 20.312,
      "step": 1600
    },
    {
      "entropy": 0.6314822697639465,
      "epoch": 0.4332178085843775,
      "grad_norm": 3.171875,
      "learning_rate": 4.920277865604952e-05,
      "loss": 0.6371616744995117,
      "mean_token_accuracy": 0.8289465856552124,
      "num_tokens": 9403553.0,
      "step": 1625
    },
    {
      "entropy": 0.6129474115371704,
      "epoch": 0.4398826979472141,
      "grad_norm": 3.03125,
      "learning_rate": 4.916862597757305e-05,
      "loss": 0.619774513244629,
      "mean_token_accuracy": 0.8317501187324524,
      "num_tokens": 9544995.0,
      "step": 1650
    },
    {
      "entropy": 0.627348735332489,
      "epoch": 0.4465475873100507,
      "grad_norm": 3.4375,
      "learning_rate": 4.913376937535878e-05,
      "loss": 0.6189299011230469,
      "mean_token_accuracy": 0.8326731014251709,
      "num_tokens": 9685611.0,
      "step": 1675
    },
    {
      "entropy": 0.6220496940612793,
      "epoch": 0.4532124766728872,
      "grad_norm": 3.328125,
      "learning_rate": 4.909820986462331e-05,
      "loss": 0.625998649597168,
      "mean_token_accuracy": 0.8305903482437134,
      "num_tokens": 9831278.0,
      "step": 1700
    },
    {
      "epoch": 0.4532124766728872,
      "eval_entropy": 0.6165648003419241,
      "eval_loss": 0.6189141273498535,
      "eval_mean_token_accuracy": 0.8325730523075721,
      "eval_num_tokens": 9831278.0,
      "eval_runtime": 9.7319,
      "eval_samples_per_second": 649.206,
      "eval_steps_per_second": 20.345,
      "step": 1700
    },
    {
      "entropy": 0.611033536195755,
      "epoch": 0.4598773660357238,
      "grad_norm": 3.203125,
      "learning_rate": 4.906194848105578e-05,
      "loss": 0.6132270812988281,
      "mean_token_accuracy": 0.8337287878990174,
      "num_tokens": 9976646.0,
      "step": 1725
    },
    {
      "entropy": 0.612893455028534,
      "epoch": 0.46654225539856037,
      "grad_norm": 3.25,
      "learning_rate": 4.902498628078776e-05,
      "loss": 0.6200115966796875,
      "mean_token_accuracy": 0.8298955512046814,
      "num_tokens": 10119925.0,
      "step": 1750
    },
    {
      "entropy": 0.6513232159614563,
      "epoch": 0.47320714476139697,
      "grad_norm": 3.5625,
      "learning_rate": 4.898732434036244e-05,
      "loss": 0.6499297332763672,
      "mean_token_accuracy": 0.8228607320785523,
      "num_tokens": 10261366.0,
      "step": 1775
    },
    {
      "entropy": 0.6138845670223236,
      "epoch": 0.4798720341242335,
      "grad_norm": 3.609375,
      "learning_rate": 4.8948963756703335e-05,
      "loss": 0.6122502899169922,
      "mean_token_accuracy": 0.8334995889663697,
      "num_tokens": 10407143.0,
      "step": 1800
    },
    {
      "epoch": 0.4798720341242335,
      "eval_entropy": 0.624958168978643,
      "eval_loss": 0.6164339184761047,
      "eval_mean_token_accuracy": 0.832638972034358,
      "eval_num_tokens": 10407143.0,
      "eval_runtime": 9.7654,
      "eval_samples_per_second": 646.975,
      "eval_steps_per_second": 20.276,
      "step": 1800
    },
    {
      "entropy": 0.6197100532054901,
      "epoch": 0.4865369234870701,
      "grad_norm": 3.71875,
      "learning_rate": 4.89099056470823e-05,
      "loss": 0.6245113372802734,
      "mean_token_accuracy": 0.831277985572815,
      "num_tokens": 10553806.0,
      "step": 1825
    },
    {
      "entropy": 0.629082190990448,
      "epoch": 0.4932018128499067,
      "grad_norm": 3.8125,
      "learning_rate": 4.887015114908697e-05,
      "loss": 0.6327330398559571,
      "mean_token_accuracy": 0.8294685196876526,
      "num_tokens": 10696634.0,
      "step": 1850
    },
    {
      "entropy": 0.6304358422756196,
      "epoch": 0.49986670221274326,
      "grad_norm": 3.0625,
      "learning_rate": 4.88297014205877e-05,
      "loss": 0.6345623779296875,
      "mean_token_accuracy": 0.8299706888198852,
      "num_tokens": 10842933.0,
      "step": 1875
    },
    {
      "entropy": 0.6218569898605346,
      "epoch": 0.5065315915755798,
      "grad_norm": 3.3125,
      "learning_rate": 4.8788557639703735e-05,
      "loss": 0.6220085144042968,
      "mean_token_accuracy": 0.831146297454834,
      "num_tokens": 10986898.0,
      "step": 1900
    },
    {
      "epoch": 0.5065315915755798,
      "eval_entropy": 0.6090830788768903,
      "eval_loss": 0.615010678768158,
      "eval_mean_token_accuracy": 0.8334405316848947,
      "eval_num_tokens": 10986898.0,
      "eval_runtime": 9.7409,
      "eval_samples_per_second": 648.604,
      "eval_steps_per_second": 20.327,
      "step": 1900
    },
    {
      "entropy": 0.6116416215896606,
      "epoch": 0.5131964809384164,
      "grad_norm": 3.734375,
      "learning_rate": 4.8746721004768995e-05,
      "loss": 0.6076314926147461,
      "mean_token_accuracy": 0.8343944025039672,
      "num_tokens": 11130894.0,
      "step": 1925
    },
    {
      "entropy": 0.6040491485595703,
      "epoch": 0.519861370301253,
      "grad_norm": 3.640625,
      "learning_rate": 4.870419273429713e-05,
      "loss": 0.6067220306396485,
      "mean_token_accuracy": 0.8341487884521485,
      "num_tokens": 11275078.0,
      "step": 1950
    },
    {
      "entropy": 0.6298557305335999,
      "epoch": 0.5265262596640896,
      "grad_norm": 3.265625,
      "learning_rate": 4.8660974066946e-05,
      "loss": 0.6338196563720703,
      "mean_token_accuracy": 0.8293531250953674,
      "num_tokens": 11421763.0,
      "step": 1975
    },
    {
      "entropy": 0.6021210646629334,
      "epoch": 0.5331911490269261,
      "grad_norm": 3.078125,
      "learning_rate": 4.861706626148167e-05,
      "loss": 0.5998428726196289,
      "mean_token_accuracy": 0.8349133777618408,
      "num_tokens": 11567920.0,
      "step": 2000
    },
    {
      "epoch": 0.5331911490269261,
      "eval_entropy": 0.6047542828201043,
      "eval_loss": 0.6115385293960571,
      "eval_mean_token_accuracy": 0.8337489107642511,
      "eval_num_tokens": 11567920.0,
      "eval_runtime": 9.754,
      "eval_samples_per_second": 647.733,
      "eval_steps_per_second": 20.299,
      "step": 2000
    },
    {
      "entropy": 0.6188659286499023,
      "epoch": 0.5398560383897627,
      "grad_norm": 3.234375,
      "learning_rate": 4.8572470596741705e-05,
      "loss": 0.6212138366699219,
      "mean_token_accuracy": 0.8316786718368531,
      "num_tokens": 11712518.0,
      "step": 2025
    },
    {
      "entropy": 0.6001254725456238,
      "epoch": 0.5465209277525993,
      "grad_norm": 3.109375,
      "learning_rate": 4.85271883715979e-05,
      "loss": 0.6023494720458984,
      "mean_token_accuracy": 0.8325961565971375,
      "num_tokens": 11854417.0,
      "step": 2050
    },
    {
      "entropy": 0.6078185975551605,
      "epoch": 0.5531858171154359,
      "grad_norm": 3.578125,
      "learning_rate": 4.8481220904918495e-05,
      "loss": 0.6093574905395508,
      "mean_token_accuracy": 0.8349392509460449,
      "num_tokens": 11995124.0,
      "step": 2075
    },
    {
      "entropy": 0.6496124482154846,
      "epoch": 0.5598507064782725,
      "grad_norm": 3.296875,
      "learning_rate": 4.843456953552975e-05,
      "loss": 0.65424072265625,
      "mean_token_accuracy": 0.8237144875526429,
      "num_tokens": 12141204.0,
      "step": 2100
    },
    {
      "epoch": 0.5598507064782725,
      "eval_entropy": 0.6072330154252775,
      "eval_loss": 0.609747588634491,
      "eval_mean_token_accuracy": 0.8345236287574576,
      "eval_num_tokens": 12141204.0,
      "eval_runtime": 9.7734,
      "eval_samples_per_second": 646.45,
      "eval_steps_per_second": 20.259,
      "step": 2100
    },
    {
      "entropy": 0.6275837254524231,
      "epoch": 0.566515595841109,
      "grad_norm": 3.296875,
      "learning_rate": 4.8387235622176944e-05,
      "loss": 0.6322821426391602,
      "mean_token_accuracy": 0.827671000957489,
      "num_tokens": 12285028.0,
      "step": 2125
    },
    {
      "entropy": 0.6162643671035767,
      "epoch": 0.5731804852039456,
      "grad_norm": 3.5,
      "learning_rate": 4.833922054348478e-05,
      "loss": 0.6133514404296875,
      "mean_token_accuracy": 0.8318719363212586,
      "num_tokens": 12428096.0,
      "step": 2150
    },
    {
      "entropy": 0.6257944464683532,
      "epoch": 0.5798453745667822,
      "grad_norm": 3.734375,
      "learning_rate": 4.8290525697917286e-05,
      "loss": 0.6283451461791992,
      "mean_token_accuracy": 0.8297429800033569,
      "num_tokens": 12569954.0,
      "step": 2175
    },
    {
      "entropy": 0.6130402791500091,
      "epoch": 0.5865102639296188,
      "grad_norm": 3.0,
      "learning_rate": 4.8241152503737016e-05,
      "loss": 0.6102632904052734,
      "mean_token_accuracy": 0.835267539024353,
      "num_tokens": 12714381.0,
      "step": 2200
    },
    {
      "epoch": 0.5865102639296188,
      "eval_entropy": 0.606271753407488,
      "eval_loss": 0.6074169874191284,
      "eval_mean_token_accuracy": 0.8349123759703203,
      "eval_num_tokens": 12714381.0,
      "eval_runtime": 9.7388,
      "eval_samples_per_second": 648.744,
      "eval_steps_per_second": 20.331,
      "step": 2200
    },
    {
      "entropy": 0.6239155578613281,
      "epoch": 0.5931751532924553,
      "grad_norm": 2.8125,
      "learning_rate": 4.819110239896383e-05,
      "loss": 0.6234345626831055,
      "mean_token_accuracy": 0.830019359588623,
      "num_tokens": 12859892.0,
      "step": 2225
    },
    {
      "entropy": 0.6134936881065368,
      "epoch": 0.5998400426552919,
      "grad_norm": 2.828125,
      "learning_rate": 4.81403768413329e-05,
      "loss": 0.6236824417114257,
      "mean_token_accuracy": 0.832092502117157,
      "num_tokens": 13004200.0,
      "step": 2250
    },
    {
      "entropy": 0.6160032296180725,
      "epoch": 0.6065049320181285,
      "grad_norm": 3.265625,
      "learning_rate": 4.8088977308252356e-05,
      "loss": 0.6103633880615235,
      "mean_token_accuracy": 0.8334192895889282,
      "num_tokens": 13152148.0,
      "step": 2275
    },
    {
      "entropy": 0.5942094624042511,
      "epoch": 0.6131698213809651,
      "grad_norm": 3.546875,
      "learning_rate": 4.803690529676019e-05,
      "loss": 0.5954260635375976,
      "mean_token_accuracy": 0.837100396156311,
      "num_tokens": 13297514.0,
      "step": 2300
    },
    {
      "epoch": 0.6131698213809651,
      "eval_entropy": 0.6259117246878267,
      "eval_loss": 0.6047196984291077,
      "eval_mean_token_accuracy": 0.834894511434767,
      "eval_num_tokens": 13297514.0,
      "eval_runtime": 9.7463,
      "eval_samples_per_second": 648.245,
      "eval_steps_per_second": 20.315,
      "step": 2300
    },
    {
      "entropy": 0.6091474604606628,
      "epoch": 0.6198347107438017,
      "grad_norm": 3.375,
      "learning_rate": 4.7984162323480686e-05,
      "loss": 0.6083653259277344,
      "mean_token_accuracy": 0.8344989585876464,
      "num_tokens": 13445685.0,
      "step": 2325
    },
    {
      "entropy": 0.6305983853340149,
      "epoch": 0.6264996001066382,
      "grad_norm": 3.546875,
      "learning_rate": 4.793074992458023e-05,
      "loss": 0.6344556427001953,
      "mean_token_accuracy": 0.827897560596466,
      "num_tokens": 13587286.0,
      "step": 2350
    },
    {
      "entropy": 0.6205494022369384,
      "epoch": 0.6331644894694748,
      "grad_norm": 3.359375,
      "learning_rate": 4.787666965572257e-05,
      "loss": 0.6239939498901367,
      "mean_token_accuracy": 0.8304968309402466,
      "num_tokens": 13734511.0,
      "step": 2375
    },
    {
      "entropy": 0.5878440809249877,
      "epoch": 0.6398293788323114,
      "grad_norm": 3.625,
      "learning_rate": 4.7821923092023536e-05,
      "loss": 0.5924504089355469,
      "mean_token_accuracy": 0.8374314546585083,
      "num_tokens": 13876465.0,
      "step": 2400
    },
    {
      "epoch": 0.6398293788323114,
      "eval_entropy": 0.6051560039773132,
      "eval_loss": 0.6047720909118652,
      "eval_mean_token_accuracy": 0.8355716957588388,
      "eval_num_tokens": 13876465.0,
      "eval_runtime": 9.7349,
      "eval_samples_per_second": 649.008,
      "eval_steps_per_second": 20.339,
      "step": 2400
    },
    {
      "entropy": 0.6027912616729736,
      "epoch": 0.646494268195148,
      "grad_norm": 3.09375,
      "learning_rate": 4.7766511828005126e-05,
      "loss": 0.5983250045776367,
      "mean_token_accuracy": 0.8355687427520752,
      "num_tokens": 14019216.0,
      "step": 2425
    },
    {
      "entropy": 0.6003220212459565,
      "epoch": 0.6531591575579845,
      "grad_norm": 3.109375,
      "learning_rate": 4.7710437477549055e-05,
      "loss": 0.603050880432129,
      "mean_token_accuracy": 0.8355051445960998,
      "num_tokens": 14166649.0,
      "step": 2450
    },
    {
      "entropy": 0.6266344451904297,
      "epoch": 0.6598240469208211,
      "grad_norm": 2.953125,
      "learning_rate": 4.7653701673849814e-05,
      "loss": 0.6223006820678711,
      "mean_token_accuracy": 0.8295617151260376,
      "num_tokens": 14311767.0,
      "step": 2475
    },
    {
      "entropy": 0.598486008644104,
      "epoch": 0.6664889362836577,
      "grad_norm": 3.375,
      "learning_rate": 4.759630606936704e-05,
      "loss": 0.6029941940307617,
      "mean_token_accuracy": 0.8345848822593689,
      "num_tokens": 14452316.0,
      "step": 2500
    },
    {
      "epoch": 0.6664889362836577,
      "eval_entropy": 0.6187347394950462,
      "eval_loss": 0.6028246879577637,
      "eval_mean_token_accuracy": 0.8359824545455702,
      "eval_num_tokens": 14452316.0,
      "eval_runtime": 9.9389,
      "eval_samples_per_second": 635.686,
      "eval_steps_per_second": 19.922,
      "step": 2500
    },
    {
      "entropy": 0.621828726530075,
      "epoch": 0.6731538256464943,
      "grad_norm": 2.9375,
      "learning_rate": 4.753825233577741e-05,
      "loss": 0.6235171890258789,
      "mean_token_accuracy": 0.8317538905143738,
      "num_tokens": 14598677.0,
      "step": 2525
    },
    {
      "entropy": 0.5931375324726105,
      "epoch": 0.6798187150093309,
      "grad_norm": 3.25,
      "learning_rate": 4.747954216392596e-05,
      "loss": 0.5964409255981445,
      "mean_token_accuracy": 0.8366534495353699,
      "num_tokens": 14745223.0,
      "step": 2550
    },
    {
      "entropy": 0.5765382301807404,
      "epoch": 0.6864836043721674,
      "grad_norm": 3.21875,
      "learning_rate": 4.7420177263776835e-05,
      "loss": 0.582974739074707,
      "mean_token_accuracy": 0.8398167967796326,
      "num_tokens": 14883405.0,
      "step": 2575
    },
    {
      "entropy": 0.5904065752029419,
      "epoch": 0.693148493735004,
      "grad_norm": 3.46875,
      "learning_rate": 4.736015936436344e-05,
      "loss": 0.5854922866821289,
      "mean_token_accuracy": 0.8389309740066528,
      "num_tokens": 15023854.0,
      "step": 2600
    },
    {
      "epoch": 0.693148493735004,
      "eval_entropy": 0.5971191399025194,
      "eval_loss": 0.6014995574951172,
      "eval_mean_token_accuracy": 0.8363629308613864,
      "eval_num_tokens": 15023854.0,
      "eval_runtime": 9.7377,
      "eval_samples_per_second": 648.818,
      "eval_steps_per_second": 20.333,
      "step": 2600
    },
    {
      "entropy": 0.5968545866012573,
      "epoch": 0.6998133830978406,
      "grad_norm": 2.984375,
      "learning_rate": 4.729949021373818e-05,
      "loss": 0.5970866775512695,
      "mean_token_accuracy": 0.837434937953949,
      "num_tokens": 15169699.0,
      "step": 2625
    },
    {
      "entropy": 0.6048703622817994,
      "epoch": 0.7064782724606772,
      "grad_norm": 3.25,
      "learning_rate": 4.723817157892145e-05,
      "loss": 0.60951171875,
      "mean_token_accuracy": 0.8339747905731201,
      "num_tokens": 15315904.0,
      "step": 2650
    },
    {
      "entropy": 0.6201922380924225,
      "epoch": 0.7131431618235138,
      "grad_norm": 2.953125,
      "learning_rate": 4.7176205245850226e-05,
      "loss": 0.6170035171508789,
      "mean_token_accuracy": 0.8327391886711121,
      "num_tokens": 15456500.0,
      "step": 2675
    },
    {
      "entropy": 0.5827772569656372,
      "epoch": 0.7198080511863503,
      "grad_norm": 2.96875,
      "learning_rate": 4.711359301932601e-05,
      "loss": 0.5839195632934571,
      "mean_token_accuracy": 0.8380605459213257,
      "num_tokens": 15604039.0,
      "step": 2700
    },
    {
      "epoch": 0.7198080511863503,
      "eval_entropy": 0.5931673052937093,
      "eval_loss": 0.6001288890838623,
      "eval_mean_token_accuracy": 0.8362962377793861,
      "eval_num_tokens": 15604039.0,
      "eval_runtime": 9.7318,
      "eval_samples_per_second": 649.21,
      "eval_steps_per_second": 20.346,
      "step": 2700
    },
    {
      "entropy": 0.5997409081459045,
      "epoch": 0.7264729405491869,
      "grad_norm": 3.328125,
      "learning_rate": 4.705033672296231e-05,
      "loss": 0.6017365264892578,
      "mean_token_accuracy": 0.8349942350387574,
      "num_tokens": 15750080.0,
      "step": 2725
    },
    {
      "entropy": 0.5990082120895386,
      "epoch": 0.7331378299120235,
      "grad_norm": 2.875,
      "learning_rate": 4.698643819913149e-05,
      "loss": 0.5988827133178711,
      "mean_token_accuracy": 0.8350508284568786,
      "num_tokens": 15894877.0,
      "step": 2750
    },
    {
      "entropy": 0.5866609215736389,
      "epoch": 0.7398027192748601,
      "grad_norm": 3.203125,
      "learning_rate": 4.692189930891111e-05,
      "loss": 0.5887953567504883,
      "mean_token_accuracy": 0.8390545415878295,
      "num_tokens": 16043347.0,
      "step": 2775
    },
    {
      "entropy": 0.6289522290229798,
      "epoch": 0.7464676086376966,
      "grad_norm": 3.578125,
      "learning_rate": 4.6856721932029755e-05,
      "loss": 0.6291543960571289,
      "mean_token_accuracy": 0.8286815953254699,
      "num_tokens": 16186193.0,
      "step": 2800
    },
    {
      "epoch": 0.7464676086376966,
      "eval_entropy": 0.5972515185071965,
      "eval_loss": 0.5981862545013428,
      "eval_mean_token_accuracy": 0.8366569593699291,
      "eval_num_tokens": 16186193.0,
      "eval_runtime": 9.738,
      "eval_samples_per_second": 648.798,
      "eval_steps_per_second": 20.333,
      "step": 2800
    },
    {
      "entropy": 0.5985104608535766,
      "epoch": 0.7531324980005332,
      "grad_norm": 2.78125,
      "learning_rate": 4.679090796681225e-05,
      "loss": 0.5956599426269531,
      "mean_token_accuracy": 0.8357338571548462,
      "num_tokens": 16331212.0,
      "step": 2825
    },
    {
      "entropy": 0.59714763879776,
      "epoch": 0.7597973873633698,
      "grad_norm": 3.046875,
      "learning_rate": 4.672445933012438e-05,
      "loss": 0.6016803359985352,
      "mean_token_accuracy": 0.8357126069068909,
      "num_tokens": 16469346.0,
      "step": 2850
    },
    {
      "entropy": 0.6107793045043945,
      "epoch": 0.7664622767262064,
      "grad_norm": 3.515625,
      "learning_rate": 4.6657377957317086e-05,
      "loss": 0.6100077056884765,
      "mean_token_accuracy": 0.8330074310302734,
      "num_tokens": 16614031.0,
      "step": 2875
    },
    {
      "entropy": 0.5867998445034027,
      "epoch": 0.773127166089043,
      "grad_norm": 2.796875,
      "learning_rate": 4.6589665802170054e-05,
      "loss": 0.5865866470336915,
      "mean_token_accuracy": 0.8399538493156433,
      "num_tokens": 16759677.0,
      "step": 2900
    },
    {
      "epoch": 0.773127166089043,
      "eval_entropy": 0.5893042432238357,
      "eval_loss": 0.5955391526222229,
      "eval_mean_token_accuracy": 0.8377770549721189,
      "eval_num_tokens": 16759677.0,
      "eval_runtime": 9.7434,
      "eval_samples_per_second": 648.439,
      "eval_steps_per_second": 20.321,
      "step": 2900
    },
    {
      "entropy": 0.5851580178737641,
      "epoch": 0.7797920554518795,
      "grad_norm": 3.265625,
      "learning_rate": 4.652132483683484e-05,
      "loss": 0.5916387939453125,
      "mean_token_accuracy": 0.8391216373443604,
      "num_tokens": 16906978.0,
      "step": 2925
    },
    {
      "entropy": 0.601825475692749,
      "epoch": 0.7864569448147161,
      "grad_norm": 2.890625,
      "learning_rate": 4.645235705177743e-05,
      "loss": 0.6006249237060547,
      "mean_token_accuracy": 0.8339450812339783,
      "num_tokens": 17051201.0,
      "step": 2950
    },
    {
      "entropy": 0.606091765165329,
      "epoch": 0.7931218341775527,
      "grad_norm": 3.109375,
      "learning_rate": 4.638276445572025e-05,
      "loss": 0.6129763412475586,
      "mean_token_accuracy": 0.8339709901809692,
      "num_tokens": 17196698.0,
      "step": 2975
    },
    {
      "entropy": 0.5800209677219391,
      "epoch": 0.7997867235403893,
      "grad_norm": 2.828125,
      "learning_rate": 4.631254907558366e-05,
      "loss": 0.5749567031860352,
      "mean_token_accuracy": 0.842078640460968,
      "num_tokens": 17343101.0,
      "step": 3000
    },
    {
      "epoch": 0.7997867235403893,
      "eval_entropy": 0.6018539712284551,
      "eval_loss": 0.5932691097259521,
      "eval_mean_token_accuracy": 0.8380928972754815,
      "eval_num_tokens": 17343101.0,
      "eval_runtime": 9.7353,
      "eval_samples_per_second": 648.98,
      "eval_steps_per_second": 20.338,
      "step": 3000
    },
    {
      "entropy": 0.5907849824428558,
      "epoch": 0.8064516129032258,
      "grad_norm": 2.96875,
      "learning_rate": 4.6241712956426955e-05,
      "loss": 0.5954814147949219,
      "mean_token_accuracy": 0.8374426198005677,
      "num_tokens": 17485438.0,
      "step": 3025
    },
    {
      "entropy": 0.6114890503883362,
      "epoch": 0.8131165022660624,
      "grad_norm": 3.078125,
      "learning_rate": 4.6170258161388744e-05,
      "loss": 0.6149655151367187,
      "mean_token_accuracy": 0.8324610614776611,
      "num_tokens": 17631751.0,
      "step": 3050
    },
    {
      "entropy": 0.5884647488594055,
      "epoch": 0.819781391628899,
      "grad_norm": 3.390625,
      "learning_rate": 4.609818677162693e-05,
      "loss": 0.583940658569336,
      "mean_token_accuracy": 0.8390086889266968,
      "num_tokens": 17780307.0,
      "step": 3075
    },
    {
      "entropy": 0.5901286149024964,
      "epoch": 0.8264462809917356,
      "grad_norm": 3.578125,
      "learning_rate": 4.602550088625801e-05,
      "loss": 0.5915086364746094,
      "mean_token_accuracy": 0.8386432647705078,
      "num_tokens": 17925149.0,
      "step": 3100
    },
    {
      "epoch": 0.8264462809917356,
      "eval_entropy": 0.572156646185451,
      "eval_loss": 0.5926044583320618,
      "eval_mean_token_accuracy": 0.8384992110006737,
      "eval_num_tokens": 17925149.0,
      "eval_runtime": 9.7242,
      "eval_samples_per_second": 649.72,
      "eval_steps_per_second": 20.362,
      "step": 3100
    },
    {
      "entropy": 0.5913607490062713,
      "epoch": 0.8331111703545722,
      "grad_norm": 3.265625,
      "learning_rate": 4.595220262229601e-05,
      "loss": 0.5934914016723633,
      "mean_token_accuracy": 0.8361663150787354,
      "num_tokens": 18067408.0,
      "step": 3125
    },
    {
      "entropy": 0.5997444796562195,
      "epoch": 0.8397760597174087,
      "grad_norm": 2.875,
      "learning_rate": 4.5878294114590826e-05,
      "loss": 0.6025965118408203,
      "mean_token_accuracy": 0.8340276598930358,
      "num_tokens": 18212385.0,
      "step": 3150
    },
    {
      "entropy": 0.5851596581935883,
      "epoch": 0.8464409490802453,
      "grad_norm": 2.5625,
      "learning_rate": 4.580377751576596e-05,
      "loss": 0.584029655456543,
      "mean_token_accuracy": 0.8390402150154114,
      "num_tokens": 18358833.0,
      "step": 3175
    },
    {
      "entropy": 0.5820559048652649,
      "epoch": 0.8531058384430819,
      "grad_norm": 2.953125,
      "learning_rate": 4.5728654996155905e-05,
      "loss": 0.5894980621337891,
      "mean_token_accuracy": 0.8393650794029236,
      "num_tokens": 18506775.0,
      "step": 3200
    },
    {
      "epoch": 0.8531058384430819,
      "eval_entropy": 0.6181450245356319,
      "eval_loss": 0.5922074317932129,
      "eval_mean_token_accuracy": 0.8380140466521485,
      "eval_num_tokens": 18506775.0,
      "eval_runtime": 9.7812,
      "eval_samples_per_second": 645.935,
      "eval_steps_per_second": 20.243,
      "step": 3200
    },
    {
      "entropy": 0.6041733551025391,
      "epoch": 0.8597707278059185,
      "grad_norm": 3.234375,
      "learning_rate": 4.565292874374294e-05,
      "loss": 0.6035808563232422,
      "mean_token_accuracy": 0.8352038097381592,
      "num_tokens": 18648458.0,
      "step": 3225
    },
    {
      "entropy": 0.5962919354438782,
      "epoch": 0.866435617168755,
      "grad_norm": 3.046875,
      "learning_rate": 4.557660096409334e-05,
      "loss": 0.5908296203613281,
      "mean_token_accuracy": 0.8380798316001892,
      "num_tokens": 18793299.0,
      "step": 3250
    },
    {
      "entropy": 0.5811218416690827,
      "epoch": 0.8731005065315915,
      "grad_norm": 3.234375,
      "learning_rate": 4.549967388029318e-05,
      "loss": 0.5825255584716796,
      "mean_token_accuracy": 0.8401248335838318,
      "num_tokens": 18937026.0,
      "step": 3275
    },
    {
      "entropy": 0.5571352910995483,
      "epoch": 0.8797653958944281,
      "grad_norm": 2.734375,
      "learning_rate": 4.54221497328836e-05,
      "loss": 0.5632460021972656,
      "mean_token_accuracy": 0.8463289976119995,
      "num_tokens": 19082870.0,
      "step": 3300
    },
    {
      "epoch": 0.8797653958944281,
      "eval_entropy": 0.60099324689369,
      "eval_loss": 0.5893176794052124,
      "eval_mean_token_accuracy": 0.8387525632525935,
      "eval_num_tokens": 19082870.0,
      "eval_runtime": 9.7403,
      "eval_samples_per_second": 648.647,
      "eval_steps_per_second": 20.328,
      "step": 3300
    },
    {
      "entropy": 0.6014616990089416,
      "epoch": 0.8864302852572647,
      "grad_norm": 3.421875,
      "learning_rate": 4.5344030779795486e-05,
      "loss": 0.6033432006835937,
      "mean_token_accuracy": 0.834429316520691,
      "num_tokens": 19230669.0,
      "step": 3325
    },
    {
      "entropy": 0.6081087231636048,
      "epoch": 0.8930951746201014,
      "grad_norm": 3.015625,
      "learning_rate": 4.526531929628379e-05,
      "loss": 0.6076117324829101,
      "mean_token_accuracy": 0.8338310408592224,
      "num_tokens": 19375850.0,
      "step": 3350
    },
    {
      "entropy": 0.606181526184082,
      "epoch": 0.8997600639829378,
      "grad_norm": 2.765625,
      "learning_rate": 4.518601757486118e-05,
      "loss": 0.6070707702636718,
      "mean_token_accuracy": 0.8326127576828003,
      "num_tokens": 19519180.0,
      "step": 3375
    },
    {
      "entropy": 0.5862989127635956,
      "epoch": 0.9064249533457744,
      "grad_norm": 2.734375,
      "learning_rate": 4.5106127925231346e-05,
      "loss": 0.5848813629150391,
      "mean_token_accuracy": 0.8414177536964417,
      "num_tokens": 19663093.0,
      "step": 3400
    },
    {
      "epoch": 0.9064249533457744,
      "eval_entropy": 0.5915440427835541,
      "eval_loss": 0.5878822207450867,
      "eval_mean_token_accuracy": 0.839049695718168,
      "eval_num_tokens": 19663093.0,
      "eval_runtime": 9.7566,
      "eval_samples_per_second": 647.564,
      "eval_steps_per_second": 20.294,
      "step": 3400
    },
    {
      "entropy": 0.5722174990177155,
      "epoch": 0.913089842708611,
      "grad_norm": 2.984375,
      "learning_rate": 4.502565267422166e-05,
      "loss": 0.5784879302978516,
      "mean_token_accuracy": 0.8408189225196838,
      "num_tokens": 19809648.0,
      "step": 3425
    },
    {
      "entropy": 0.5941082048416138,
      "epoch": 0.9197547320714476,
      "grad_norm": 3.15625,
      "learning_rate": 4.494459416571545e-05,
      "loss": 0.5904175186157227,
      "mean_token_accuracy": 0.8363987565040588,
      "num_tokens": 19953752.0,
      "step": 3450
    },
    {
      "entropy": 0.5655630517005921,
      "epoch": 0.9264196214342842,
      "grad_norm": 3.15625,
      "learning_rate": 4.486295476058372e-05,
      "loss": 0.5638382720947266,
      "mean_token_accuracy": 0.8425051403045655,
      "num_tokens": 20097535.0,
      "step": 3475
    },
    {
      "entropy": 0.5954309177398681,
      "epoch": 0.9330845107971207,
      "grad_norm": 3.21875,
      "learning_rate": 4.478073683661641e-05,
      "loss": 0.5978251266479492,
      "mean_token_accuracy": 0.8378843522071838,
      "num_tokens": 20243811.0,
      "step": 3500
    },
    {
      "epoch": 0.9330845107971207,
      "eval_entropy": 0.5841524459496893,
      "eval_loss": 0.5865455865859985,
      "eval_mean_token_accuracy": 0.8397409310846617,
      "eval_num_tokens": 20243811.0,
      "eval_runtime": 9.7567,
      "eval_samples_per_second": 647.557,
      "eval_steps_per_second": 20.294,
      "step": 3500
    },
    {
      "entropy": 0.5696915936470032,
      "epoch": 0.9397494001599573,
      "grad_norm": 2.734375,
      "learning_rate": 4.469794278845307e-05,
      "loss": 0.5706735229492188,
      "mean_token_accuracy": 0.8425429725646972,
      "num_tokens": 20387397.0,
      "step": 3525
    },
    {
      "entropy": 0.5941010534763336,
      "epoch": 0.9464142895227939,
      "grad_norm": 3.0,
      "learning_rate": 4.4614575027513205e-05,
      "loss": 0.5982907485961914,
      "mean_token_accuracy": 0.8367367315292359,
      "num_tokens": 20535831.0,
      "step": 3550
    },
    {
      "entropy": 0.6068766963481903,
      "epoch": 0.9530791788856305,
      "grad_norm": 2.796875,
      "learning_rate": 4.4530635981926006e-05,
      "loss": 0.6052367401123047,
      "mean_token_accuracy": 0.8348676657676697,
      "num_tokens": 20678706.0,
      "step": 3575
    },
    {
      "entropy": 0.5945624005794525,
      "epoch": 0.959744068248467,
      "grad_norm": 2.9375,
      "learning_rate": 4.444612809645959e-05,
      "loss": 0.5949686431884765,
      "mean_token_accuracy": 0.8374036860466003,
      "num_tokens": 20822990.0,
      "step": 3600
    },
    {
      "epoch": 0.959744068248467,
      "eval_entropy": 0.5832300323127496,
      "eval_loss": 0.5857488512992859,
      "eval_mean_token_accuracy": 0.8397423621982035,
      "eval_num_tokens": 20822990.0,
      "eval_runtime": 9.7301,
      "eval_samples_per_second": 649.324,
      "eval_steps_per_second": 20.349,
      "step": 3600
    },
    {
      "entropy": 0.5896636867523193,
      "epoch": 0.9664089576113036,
      "grad_norm": 3.0625,
      "learning_rate": 4.436105383244986e-05,
      "loss": 0.5859294891357422,
      "mean_token_accuracy": 0.8385764217376709,
      "num_tokens": 20966208.0,
      "step": 3625
    },
    {
      "entropy": 0.5848815441131592,
      "epoch": 0.9730738469741402,
      "grad_norm": 3.234375,
      "learning_rate": 4.427541566772878e-05,
      "loss": 0.5898666000366211,
      "mean_token_accuracy": 0.8376466727256775,
      "num_tokens": 21109678.0,
      "step": 3650
    },
    {
      "entropy": 0.5694864594936371,
      "epoch": 0.9797387363369768,
      "grad_norm": 3.265625,
      "learning_rate": 4.41892160965522e-05,
      "loss": 0.5709371185302734,
      "mean_token_accuracy": 0.8425320792198181,
      "num_tokens": 21257932.0,
      "step": 3675
    },
    {
      "entropy": 0.5713275444507598,
      "epoch": 0.9864036256998134,
      "grad_norm": 3.328125,
      "learning_rate": 4.410245762952725e-05,
      "loss": 0.573502311706543,
      "mean_token_accuracy": 0.8421503901481628,
      "num_tokens": 21400583.0,
      "step": 3700
    },
    {
      "epoch": 0.9864036256998134,
      "eval_entropy": 0.5846781367906416,
      "eval_loss": 0.5837874412536621,
      "eval_mean_token_accuracy": 0.840186777741018,
      "eval_num_tokens": 21400583.0,
      "eval_runtime": 9.7378,
      "eval_samples_per_second": 648.815,
      "eval_steps_per_second": 20.333,
      "step": 3700
    },
    {
      "entropy": 0.5944207239151001,
      "epoch": 0.9930685150626499,
      "grad_norm": 3.1875,
      "learning_rate": 4.4015142793539164e-05,
      "loss": 0.5886489486694336,
      "mean_token_accuracy": 0.8381674599647522,
      "num_tokens": 21545777.0,
      "step": 3725
    },
    {
      "entropy": 0.59309396982193,
      "epoch": 0.9997334044254865,
      "grad_norm": 2.75,
      "learning_rate": 4.3927274131677734e-05,
      "loss": 0.6018276596069336,
      "mean_token_accuracy": 0.8352336382865906,
      "num_tokens": 21692055.0,
      "step": 3750
    },
    {
      "entropy": 0.5392892348766327,
      "epoch": 1.0063982937883231,
      "grad_norm": 2.625,
      "learning_rate": 4.3838854203163196e-05,
      "loss": 0.5211112213134765,
      "mean_token_accuracy": 0.8531012725830078,
      "num_tokens": 21832168.0,
      "step": 3775
    },
    {
      "entropy": 0.5008841550350189,
      "epoch": 1.0130631831511596,
      "grad_norm": 3.34375,
      "learning_rate": 4.374988558327171e-05,
      "loss": 0.4891099166870117,
      "mean_token_accuracy": 0.8606327605247498,
      "num_tokens": 21977587.0,
      "step": 3800
    },
    {
      "epoch": 1.0130631831511596,
      "eval_entropy": 0.5184622164627518,
      "eval_loss": 0.5895727276802063,
      "eval_mean_token_accuracy": 0.8396453758080801,
      "eval_num_tokens": 21977587.0,
      "eval_runtime": 9.7284,
      "eval_samples_per_second": 649.438,
      "eval_steps_per_second": 20.353,
      "step": 3800
    },
    {
      "entropy": 0.50399031996727,
      "epoch": 1.0197280725139963,
      "grad_norm": 3.375,
      "learning_rate": 4.366037086326039e-05,
      "loss": 0.4957071304321289,
      "mean_token_accuracy": 0.857741436958313,
      "num_tokens": 22124405.0,
      "step": 3825
    },
    {
      "entropy": 0.5011958622932434,
      "epoch": 1.0263929618768328,
      "grad_norm": 2.75,
      "learning_rate": 4.357031265029177e-05,
      "loss": 0.4955809783935547,
      "mean_token_accuracy": 0.8605676531791687,
      "num_tokens": 22270623.0,
      "step": 3850
    },
    {
      "entropy": 0.5175231349468231,
      "epoch": 1.0330578512396693,
      "grad_norm": 2.71875,
      "learning_rate": 4.3479713567357886e-05,
      "loss": 0.509754409790039,
      "mean_token_accuracy": 0.8568044590950012,
      "num_tokens": 22414418.0,
      "step": 3875
    },
    {
      "entropy": 0.48299696803092956,
      "epoch": 1.039722740602506,
      "grad_norm": 3.171875,
      "learning_rate": 4.338857625320394e-05,
      "loss": 0.4777442169189453,
      "mean_token_accuracy": 0.8638524317741394,
      "num_tokens": 22554058.0,
      "step": 3900
    },
    {
      "epoch": 1.039722740602506,
      "eval_entropy": 0.5268884427619703,
      "eval_loss": 0.5868653059005737,
      "eval_mean_token_accuracy": 0.8400596973269877,
      "eval_num_tokens": 22554058.0,
      "eval_runtime": 9.7363,
      "eval_samples_per_second": 648.912,
      "eval_steps_per_second": 20.336,
      "step": 3900
    },
    {
      "entropy": 0.5102592813968658,
      "epoch": 1.0463876299653425,
      "grad_norm": 3.53125,
      "learning_rate": 4.3296903362251336e-05,
      "loss": 0.5050753784179688,
      "mean_token_accuracy": 0.8573951268196106,
      "num_tokens": 22697566.0,
      "step": 3925
    },
    {
      "entropy": 0.47698324084281923,
      "epoch": 1.0530525193281792,
      "grad_norm": 3.609375,
      "learning_rate": 4.320469756452048e-05,
      "loss": 0.46883365631103513,
      "mean_token_accuracy": 0.8662173438072205,
      "num_tokens": 22843336.0,
      "step": 3950
    },
    {
      "entropy": 0.49658790946006776,
      "epoch": 1.0597174086910157,
      "grad_norm": 3.171875,
      "learning_rate": 4.311196154555294e-05,
      "loss": 0.49169235229492186,
      "mean_token_accuracy": 0.8591770315170288,
      "num_tokens": 22988698.0,
      "step": 3975
    },
    {
      "entropy": 0.5155219399929046,
      "epoch": 1.0663822980538522,
      "grad_norm": 2.8125,
      "learning_rate": 4.301869800633325e-05,
      "loss": 0.51111083984375,
      "mean_token_accuracy": 0.8561162400245667,
      "num_tokens": 23132070.0,
      "step": 4000
    },
    {
      "epoch": 1.0663822980538522,
      "eval_entropy": 0.5093354819097904,
      "eval_loss": 0.587713360786438,
      "eval_mean_token_accuracy": 0.8403223054577605,
      "eval_num_tokens": 23132070.0,
      "eval_runtime": 9.7501,
      "eval_samples_per_second": 647.992,
      "eval_steps_per_second": 20.307,
      "step": 4000
    },
    {
      "entropy": 0.511484671831131,
      "epoch": 1.073047187416689,
      "grad_norm": 3.0625,
      "learning_rate": 4.292490966321025e-05,
      "loss": 0.5033857345581054,
      "mean_token_accuracy": 0.8568600273132324,
      "num_tokens": 23280385.0,
      "step": 4025
    },
    {
      "entropy": 0.5142236185073853,
      "epoch": 1.0797120767795254,
      "grad_norm": 2.90625,
      "learning_rate": 4.283059924781797e-05,
      "loss": 0.5116531372070312,
      "mean_token_accuracy": 0.8551531672477722,
      "num_tokens": 23423929.0,
      "step": 4050
    },
    {
      "entropy": 0.5283914136886597,
      "epoch": 1.0863769661423621,
      "grad_norm": 3.0,
      "learning_rate": 4.273576950699605e-05,
      "loss": 0.5269992065429687,
      "mean_token_accuracy": 0.8525123023986816,
      "num_tokens": 23569889.0,
      "step": 4075
    },
    {
      "entropy": 0.5085922884941101,
      "epoch": 1.0930418555051986,
      "grad_norm": 3.796875,
      "learning_rate": 4.2640423202709765e-05,
      "loss": 0.5062469482421875,
      "mean_token_accuracy": 0.8572459363937378,
      "num_tokens": 23714230.0,
      "step": 4100
    },
    {
      "epoch": 1.0930418555051986,
      "eval_entropy": 0.5201174911826548,
      "eval_loss": 0.5870948433876038,
      "eval_mean_token_accuracy": 0.8402606760612642,
      "eval_num_tokens": 23714230.0,
      "eval_runtime": 9.7413,
      "eval_samples_per_second": 648.581,
      "eval_steps_per_second": 20.326,
      "step": 4100
    },
    {
      "entropy": 0.5255253279209137,
      "epoch": 1.099706744868035,
      "grad_norm": 3.046875,
      "learning_rate": 4.2544563111969575e-05,
      "loss": 0.5161056900024414,
      "mean_token_accuracy": 0.8543310570716858,
      "num_tokens": 23858838.0,
      "step": 4125
    },
    {
      "entropy": 0.4847089183330536,
      "epoch": 1.1063716342308718,
      "grad_norm": 2.78125,
      "learning_rate": 4.244819202675021e-05,
      "loss": 0.48134525299072267,
      "mean_token_accuracy": 0.8621941256523132,
      "num_tokens": 24002318.0,
      "step": 4150
    },
    {
      "entropy": 0.5117491590976715,
      "epoch": 1.1130365235937083,
      "grad_norm": 3.0,
      "learning_rate": 4.235131275390941e-05,
      "loss": 0.5072209548950195,
      "mean_token_accuracy": 0.8567190027236938,
      "num_tokens": 24142993.0,
      "step": 4175
    },
    {
      "entropy": 0.5290677130222321,
      "epoch": 1.119701412956545,
      "grad_norm": 2.953125,
      "learning_rate": 4.225392811510613e-05,
      "loss": 0.5246530151367188,
      "mean_token_accuracy": 0.8522943139076233,
      "num_tokens": 24285920.0,
      "step": 4200
    },
    {
      "epoch": 1.119701412956545,
      "eval_entropy": 0.516965294877688,
      "eval_loss": 0.5866913795471191,
      "eval_mean_token_accuracy": 0.8407578353930001,
      "eval_num_tokens": 24285920.0,
      "eval_runtime": 9.7343,
      "eval_samples_per_second": 649.044,
      "eval_steps_per_second": 20.34,
      "step": 4200
    },
    {
      "entropy": 0.4997833299636841,
      "epoch": 1.1263663023193815,
      "grad_norm": 2.796875,
      "learning_rate": 4.215604094671835e-05,
      "loss": 0.4927844619750977,
      "mean_token_accuracy": 0.8590528678894043,
      "num_tokens": 24428318.0,
      "step": 4225
    },
    {
      "entropy": 0.4754242181777954,
      "epoch": 1.133031191682218,
      "grad_norm": 3.4375,
      "learning_rate": 4.20576540997605e-05,
      "loss": 0.4746293640136719,
      "mean_token_accuracy": 0.8637139797210693,
      "num_tokens": 24574268.0,
      "step": 4250
    },
    {
      "entropy": 0.4979741084575653,
      "epoch": 1.1396960810450547,
      "grad_norm": 3.09375,
      "learning_rate": 4.195877043980042e-05,
      "loss": 0.49032203674316405,
      "mean_token_accuracy": 0.860429105758667,
      "num_tokens": 24720940.0,
      "step": 4275
    },
    {
      "entropy": 0.5015843975543975,
      "epoch": 1.1463609704078912,
      "grad_norm": 2.734375,
      "learning_rate": 4.185939284687588e-05,
      "loss": 0.49562347412109375,
      "mean_token_accuracy": 0.8594220519065857,
      "num_tokens": 24866114.0,
      "step": 4300
    },
    {
      "epoch": 1.1463609704078912,
      "eval_entropy": 0.5156128767764929,
      "eval_loss": 0.5860560536384583,
      "eval_mean_token_accuracy": 0.8406131875635398,
      "eval_num_tokens": 24866114.0,
      "eval_runtime": 9.7262,
      "eval_samples_per_second": 649.588,
      "eval_steps_per_second": 20.357,
      "step": 4300
    },
    {
      "entropy": 0.507763022184372,
      "epoch": 1.153025859770728,
      "grad_norm": 3.109375,
      "learning_rate": 4.175952421541068e-05,
      "loss": 0.5014072799682617,
      "mean_token_accuracy": 0.8574363899230957,
      "num_tokens": 25010005.0,
      "step": 4325
    },
    {
      "entropy": 0.5023082315921783,
      "epoch": 1.1596907491335644,
      "grad_norm": 3.21875,
      "learning_rate": 4.165916745413041e-05,
      "loss": 0.49743896484375,
      "mean_token_accuracy": 0.8583322453498841,
      "num_tokens": 25155499.0,
      "step": 4350
    },
    {
      "entropy": 0.5104747486114501,
      "epoch": 1.166355638496401,
      "grad_norm": 2.859375,
      "learning_rate": 4.155832548597766e-05,
      "loss": 0.509513168334961,
      "mean_token_accuracy": 0.8577624201774597,
      "num_tokens": 25302266.0,
      "step": 4375
    },
    {
      "entropy": 0.4983500790596008,
      "epoch": 1.1730205278592376,
      "grad_norm": 2.96875,
      "learning_rate": 4.145700124802693e-05,
      "loss": 0.4983134078979492,
      "mean_token_accuracy": 0.8582618188858032,
      "num_tokens": 25443830.0,
      "step": 4400
    },
    {
      "epoch": 1.1730205278592376,
      "eval_entropy": 0.520526113835248,
      "eval_loss": 0.5842269062995911,
      "eval_mean_token_accuracy": 0.8410851266046967,
      "eval_num_tokens": 25443830.0,
      "eval_runtime": 9.7556,
      "eval_samples_per_second": 647.625,
      "eval_steps_per_second": 20.296,
      "step": 4400
    },
    {
      "entropy": 0.49977203011512755,
      "epoch": 1.179685417222074,
      "grad_norm": 3.109375,
      "learning_rate": 4.1355197691399085e-05,
      "loss": 0.4933819961547852,
      "mean_token_accuracy": 0.8594501066207886,
      "num_tokens": 25591260.0,
      "step": 4425
    },
    {
      "entropy": 0.5259402668476105,
      "epoch": 1.1863503065849108,
      "grad_norm": 3.34375,
      "learning_rate": 4.1252917781175396e-05,
      "loss": 0.5194372177124024,
      "mean_token_accuracy": 0.856159439086914,
      "num_tokens": 25737069.0,
      "step": 4450
    },
    {
      "entropy": 0.5153890872001647,
      "epoch": 1.1930151959477473,
      "grad_norm": 3.015625,
      "learning_rate": 4.115016449631116e-05,
      "loss": 0.505796241760254,
      "mean_token_accuracy": 0.8548446869850159,
      "num_tokens": 25878998.0,
      "step": 4475
    },
    {
      "entropy": 0.4993034374713898,
      "epoch": 1.1996800853105838,
      "grad_norm": 3.25,
      "learning_rate": 4.104694082954898e-05,
      "loss": 0.4987041091918945,
      "mean_token_accuracy": 0.8577118849754334,
      "num_tokens": 26023729.0,
      "step": 4500
    },
    {
      "epoch": 1.1996800853105838,
      "eval_entropy": 0.507043658482908,
      "eval_loss": 0.5852404236793518,
      "eval_mean_token_accuracy": 0.8408991965380582,
      "eval_num_tokens": 26023729.0,
      "eval_runtime": 9.7663,
      "eval_samples_per_second": 646.921,
      "eval_steps_per_second": 20.274,
      "step": 4500
    },
    {
      "entropy": 0.5043972539901733,
      "epoch": 1.2063449746734205,
      "grad_norm": 2.859375,
      "learning_rate": 4.094324978733157e-05,
      "loss": 0.49994281768798826,
      "mean_token_accuracy": 0.8591324949264526,
      "num_tokens": 26169098.0,
      "step": 4525
    },
    {
      "entropy": 0.5257331693172455,
      "epoch": 1.213009864036257,
      "grad_norm": 2.859375,
      "learning_rate": 4.083909438971417e-05,
      "loss": 0.5249554824829101,
      "mean_token_accuracy": 0.8541541981697083,
      "num_tokens": 26313901.0,
      "step": 4550
    },
    {
      "entropy": 0.5089606499671936,
      "epoch": 1.2196747533990935,
      "grad_norm": 3.421875,
      "learning_rate": 4.073447767027666e-05,
      "loss": 0.5029035568237304,
      "mean_token_accuracy": 0.8567742228507995,
      "num_tokens": 26456396.0,
      "step": 4575
    },
    {
      "entropy": 0.5000081765651703,
      "epoch": 1.2263396427619302,
      "grad_norm": 3.0625,
      "learning_rate": 4.0629402676035094e-05,
      "loss": 0.4968996047973633,
      "mean_token_accuracy": 0.8593900799751282,
      "num_tokens": 26603709.0,
      "step": 4600
    },
    {
      "epoch": 1.2263396427619302,
      "eval_entropy": 0.5124151544137434,
      "eval_loss": 0.5833254456520081,
      "eval_mean_token_accuracy": 0.841457466284434,
      "eval_num_tokens": 26603709.0,
      "eval_runtime": 9.7384,
      "eval_samples_per_second": 648.774,
      "eval_steps_per_second": 20.332,
      "step": 4600
    },
    {
      "entropy": 0.5004336619377137,
      "epoch": 1.2330045321247667,
      "grad_norm": 3.078125,
      "learning_rate": 4.0523872467353075e-05,
      "loss": 0.493514404296875,
      "mean_token_accuracy": 0.8571614718437195,
      "num_tokens": 26748859.0,
      "step": 4625
    },
    {
      "entropy": 0.5111635398864746,
      "epoch": 1.2396694214876034,
      "grad_norm": 3.3125,
      "learning_rate": 4.041789011785253e-05,
      "loss": 0.5096990585327148,
      "mean_token_accuracy": 0.8550857090950013,
      "num_tokens": 26889011.0,
      "step": 4650
    },
    {
      "entropy": 0.49170732975006104,
      "epoch": 1.2463343108504399,
      "grad_norm": 3.28125,
      "learning_rate": 4.031145871432424e-05,
      "loss": 0.4863494110107422,
      "mean_token_accuracy": 0.860098683834076,
      "num_tokens": 27032375.0,
      "step": 4675
    },
    {
      "entropy": 0.5117895793914795,
      "epoch": 1.2529992002132766,
      "grad_norm": 2.953125,
      "learning_rate": 4.020458135663789e-05,
      "loss": 0.5120787048339843,
      "mean_token_accuracy": 0.8560007190704346,
      "num_tokens": 27177271.0,
      "step": 4700
    },
    {
      "epoch": 1.2529992002132766,
      "eval_entropy": 0.5179021790473148,
      "eval_loss": 0.5813733339309692,
      "eval_mean_token_accuracy": 0.8414854235721357,
      "eval_num_tokens": 27177271.0,
      "eval_runtime": 9.7357,
      "eval_samples_per_second": 648.95,
      "eval_steps_per_second": 20.337,
      "step": 4700
    },
    {
      "entropy": 0.491740335226059,
      "epoch": 1.259664089576113,
      "grad_norm": 3.171875,
      "learning_rate": 4.009726115765183e-05,
      "loss": 0.48845561981201174,
      "mean_token_accuracy": 0.8611953997612,
      "num_tokens": 27324590.0,
      "step": 4725
    },
    {
      "entropy": 0.5393763053417205,
      "epoch": 1.2663289789389496,
      "grad_norm": 2.734375,
      "learning_rate": 3.998950124312239e-05,
      "loss": 0.5358795928955078,
      "mean_token_accuracy": 0.850636191368103,
      "num_tokens": 27469048.0,
      "step": 4750
    },
    {
      "entropy": 0.501517174243927,
      "epoch": 1.272993868301786,
      "grad_norm": 3.125,
      "learning_rate": 3.988130475161286e-05,
      "loss": 0.4910397720336914,
      "mean_token_accuracy": 0.8588261079788208,
      "num_tokens": 27614413.0,
      "step": 4775
    },
    {
      "entropy": 0.5018379390239716,
      "epoch": 1.2796587576646228,
      "grad_norm": 3.0625,
      "learning_rate": 3.9772674834402014e-05,
      "loss": 0.5001216125488281,
      "mean_token_accuracy": 0.8595121622085571,
      "num_tokens": 27762519.0,
      "step": 4800
    },
    {
      "epoch": 1.2796587576646228,
      "eval_entropy": 0.5208656595210837,
      "eval_loss": 0.5812972187995911,
      "eval_mean_token_accuracy": 0.841524803277218,
      "eval_num_tokens": 27762519.0,
      "eval_runtime": 9.7704,
      "eval_samples_per_second": 646.646,
      "eval_steps_per_second": 20.265,
      "step": 4800
    },
    {
      "entropy": 0.5212963402271271,
      "epoch": 1.2863236470274593,
      "grad_norm": 2.59375,
      "learning_rate": 3.966361465539242e-05,
      "loss": 0.5138653182983398,
      "mean_token_accuracy": 0.8536733984947205,
      "num_tokens": 27903469.0,
      "step": 4825
    },
    {
      "entropy": 0.5033409285545349,
      "epoch": 1.292988536390296,
      "grad_norm": 2.84375,
      "learning_rate": 3.955412739101822e-05,
      "loss": 0.5018418884277344,
      "mean_token_accuracy": 0.8581663250923157,
      "num_tokens": 28046419.0,
      "step": 4850
    },
    {
      "entropy": 0.5039857482910156,
      "epoch": 1.2996534257531325,
      "grad_norm": 3.109375,
      "learning_rate": 3.944421623015264e-05,
      "loss": 0.49739479064941405,
      "mean_token_accuracy": 0.8591382002830505,
      "num_tokens": 28191329.0,
      "step": 4875
    },
    {
      "entropy": 0.5154730951786042,
      "epoch": 1.306318315115969,
      "grad_norm": 2.71875,
      "learning_rate": 3.933388437401512e-05,
      "loss": 0.511655387878418,
      "mean_token_accuracy": 0.8580222272872925,
      "num_tokens": 28332894.0,
      "step": 4900
    },
    {
      "epoch": 1.306318315115969,
      "eval_entropy": 0.5046350797920516,
      "eval_loss": 0.5813717246055603,
      "eval_mean_token_accuracy": 0.8419145174098738,
      "eval_num_tokens": 28332894.0,
      "eval_runtime": 9.7529,
      "eval_samples_per_second": 647.807,
      "eval_steps_per_second": 20.302,
      "step": 4900
    },
    {
      "entropy": 0.4988634371757507,
      "epoch": 1.3129832044788057,
      "grad_norm": 3.640625,
      "learning_rate": 3.9223135036078064e-05,
      "loss": 0.5002391052246093,
      "mean_token_accuracy": 0.8578930592536926,
      "num_tokens": 28473007.0,
      "step": 4925
    },
    {
      "entropy": 0.5041767644882202,
      "epoch": 1.3196480938416422,
      "grad_norm": 3.703125,
      "learning_rate": 3.911197144197324e-05,
      "loss": 0.49866897583007813,
      "mean_token_accuracy": 0.8579932999610901,
      "num_tokens": 28617949.0,
      "step": 4950
    },
    {
      "entropy": 0.49564511895179747,
      "epoch": 1.3263129832044789,
      "grad_norm": 2.75,
      "learning_rate": 3.9000396829397846e-05,
      "loss": 0.48963977813720705,
      "mean_token_accuracy": 0.8593028116226197,
      "num_tokens": 28763742.0,
      "step": 4975
    },
    {
      "entropy": 0.4947984671592712,
      "epoch": 1.3329778725673154,
      "grad_norm": 3.109375,
      "learning_rate": 3.888841444802022e-05,
      "loss": 0.49226444244384765,
      "mean_token_accuracy": 0.8594289588928222,
      "num_tokens": 28905102.0,
      "step": 5000
    },
    {
      "epoch": 1.3329778725673154,
      "eval_entropy": 0.5156144879081033,
      "eval_loss": 0.5803172588348389,
      "eval_mean_token_accuracy": 0.8419654944328346,
      "eval_num_tokens": 28905102.0,
      "eval_runtime": 9.7414,
      "eval_samples_per_second": 648.573,
      "eval_steps_per_second": 20.326,
      "step": 5000
    },
    {
      "entropy": 0.492349898815155,
      "epoch": 1.3396427619301519,
      "grad_norm": 3.140625,
      "learning_rate": 3.8776027559385156e-05,
      "loss": 0.48780059814453125,
      "mean_token_accuracy": 0.8593190836906434,
      "num_tokens": 29049322.0,
      "step": 5025
    },
    {
      "entropy": 0.4934911644458771,
      "epoch": 1.3463076512929886,
      "grad_norm": 2.78125,
      "learning_rate": 3.866323943681894e-05,
      "loss": 0.49066509246826173,
      "mean_token_accuracy": 0.8612163662910461,
      "num_tokens": 29193147.0,
      "step": 5050
    },
    {
      "entropy": 0.5232586395740509,
      "epoch": 1.352972540655825,
      "grad_norm": 3.03125,
      "learning_rate": 3.855005336533402e-05,
      "loss": 0.5178916931152344,
      "mean_token_accuracy": 0.8533386588096619,
      "num_tokens": 29337358.0,
      "step": 5075
    },
    {
      "entropy": 0.4796565890312195,
      "epoch": 1.3596374300186618,
      "grad_norm": 2.6875,
      "learning_rate": 3.8436472641533275e-05,
      "loss": 0.47610820770263673,
      "mean_token_accuracy": 0.8629685068130493,
      "num_tokens": 29487222.0,
      "step": 5100
    },
    {
      "epoch": 1.3596374300186618,
      "eval_entropy": 0.5127187901073031,
      "eval_loss": 0.5798896551132202,
      "eval_mean_token_accuracy": 0.8420344657368131,
      "eval_num_tokens": 29487222.0,
      "eval_runtime": 9.7545,
      "eval_samples_per_second": 647.703,
      "eval_steps_per_second": 20.298,
      "step": 5100
    },
    {
      "entropy": 0.5090840709209442,
      "epoch": 1.3663023193814983,
      "grad_norm": 2.859375,
      "learning_rate": 3.832250057351407e-05,
      "loss": 0.5035771560668946,
      "mean_token_accuracy": 0.8572527241706848,
      "num_tokens": 29630625.0,
      "step": 5125
    },
    {
      "entropy": 0.48668288707733154,
      "epoch": 1.3729672087443348,
      "grad_norm": 3.359375,
      "learning_rate": 3.8208140480771856e-05,
      "loss": 0.48361099243164063,
      "mean_token_accuracy": 0.8621697497367858,
      "num_tokens": 29773326.0,
      "step": 5150
    },
    {
      "entropy": 0.49902196645736696,
      "epoch": 1.3796320981071715,
      "grad_norm": 3.15625,
      "learning_rate": 3.809339569410353e-05,
      "loss": 0.49409725189208986,
      "mean_token_accuracy": 0.8606331825256348,
      "num_tokens": 29916238.0,
      "step": 5175
    },
    {
      "entropy": 0.488629195690155,
      "epoch": 1.386296987470008,
      "grad_norm": 2.796875,
      "learning_rate": 3.797826955551036e-05,
      "loss": 0.4872911834716797,
      "mean_token_accuracy": 0.8595227003097534,
      "num_tokens": 30059543.0,
      "step": 5200
    },
    {
      "epoch": 1.386296987470008,
      "eval_entropy": 0.5199090155086132,
      "eval_loss": 0.5779443383216858,
      "eval_mean_token_accuracy": 0.8421979239492705,
      "eval_num_tokens": 30059543.0,
      "eval_runtime": 9.7654,
      "eval_samples_per_second": 646.98,
      "eval_steps_per_second": 20.276,
      "step": 5200
    },
    {
      "entropy": 0.48608951807022094,
      "epoch": 1.3929618768328447,
      "grad_norm": 3.296875,
      "learning_rate": 3.78627654181007e-05,
      "loss": 0.4822881317138672,
      "mean_token_accuracy": 0.8628294610977173,
      "num_tokens": 30203206.0,
      "step": 5225
    },
    {
      "entropy": 0.48682778716087344,
      "epoch": 1.3996267661956812,
      "grad_norm": 3.203125,
      "learning_rate": 3.774688664599233e-05,
      "loss": 0.4814950942993164,
      "mean_token_accuracy": 0.8632514381408691,
      "num_tokens": 30349617.0,
      "step": 5250
    },
    {
      "entropy": 0.5095673787593842,
      "epoch": 1.4062916555585177,
      "grad_norm": 2.828125,
      "learning_rate": 3.763063661421445e-05,
      "loss": 0.5060235214233398,
      "mean_token_accuracy": 0.8573762965202332,
      "num_tokens": 30495020.0,
      "step": 5275
    },
    {
      "entropy": 0.4841842985153198,
      "epoch": 1.4129565449213544,
      "grad_norm": 2.984375,
      "learning_rate": 3.75140187086094e-05,
      "loss": 0.48105899810791014,
      "mean_token_accuracy": 0.861905689239502,
      "num_tokens": 30642347.0,
      "step": 5300
    },
    {
      "epoch": 1.4129565449213544,
      "eval_entropy": 0.5103263499760868,
      "eval_loss": 0.5780280232429504,
      "eval_mean_token_accuracy": 0.8424878018070953,
      "eval_num_tokens": 30642347.0,
      "eval_runtime": 9.7319,
      "eval_samples_per_second": 649.204,
      "eval_steps_per_second": 20.345,
      "step": 5300
    },
    {
      "entropy": 0.4989655911922455,
      "epoch": 1.4196214342841909,
      "grad_norm": 3.203125,
      "learning_rate": 3.7397036325734e-05,
      "loss": 0.49534820556640624,
      "mean_token_accuracy": 0.8585507464408875,
      "num_tokens": 30785003.0,
      "step": 5325
    },
    {
      "entropy": 0.519689199924469,
      "epoch": 1.4262863236470276,
      "grad_norm": 3.703125,
      "learning_rate": 3.7279692872760705e-05,
      "loss": 0.5148283386230469,
      "mean_token_accuracy": 0.8537467694282532,
      "num_tokens": 30928037.0,
      "step": 5350
    },
    {
      "entropy": 0.496384756565094,
      "epoch": 1.432951213009864,
      "grad_norm": 2.875,
      "learning_rate": 3.7161991767378304e-05,
      "loss": 0.49060142517089844,
      "mean_token_accuracy": 0.8604137516021728,
      "num_tokens": 31074977.0,
      "step": 5375
    },
    {
      "entropy": 0.48764374256134035,
      "epoch": 1.4396161023727005,
      "grad_norm": 3.203125,
      "learning_rate": 3.704393643769238e-05,
      "loss": 0.4834042739868164,
      "mean_token_accuracy": 0.8625716900825501,
      "num_tokens": 31222400.0,
      "step": 5400
    },
    {
      "epoch": 1.4396161023727005,
      "eval_entropy": 0.5097903979547096,
      "eval_loss": 0.5779716968536377,
      "eval_mean_token_accuracy": 0.8427059776855238,
      "eval_num_tokens": 31222400.0,
      "eval_runtime": 9.792,
      "eval_samples_per_second": 645.223,
      "eval_steps_per_second": 20.221,
      "step": 5400
    },
    {
      "entropy": 0.5231076347827911,
      "epoch": 1.4462809917355373,
      "grad_norm": 3.765625,
      "learning_rate": 3.6925530322125504e-05,
      "loss": 0.5226813125610351,
      "mean_token_accuracy": 0.854303572177887,
      "num_tokens": 31367998.0,
      "step": 5425
    },
    {
      "entropy": 0.5331024742126464,
      "epoch": 1.4529458810983737,
      "grad_norm": 2.875,
      "learning_rate": 3.680677686931707e-05,
      "loss": 0.5296571731567383,
      "mean_token_accuracy": 0.851455614566803,
      "num_tokens": 31516181.0,
      "step": 5450
    },
    {
      "entropy": 0.5272133600711822,
      "epoch": 1.4596107704612105,
      "grad_norm": 2.859375,
      "learning_rate": 3.668767953802283e-05,
      "loss": 0.5236498641967774,
      "mean_token_accuracy": 0.851326093673706,
      "num_tokens": 31657722.0,
      "step": 5475
    },
    {
      "entropy": 0.501412365436554,
      "epoch": 1.466275659824047,
      "grad_norm": 2.71875,
      "learning_rate": 3.65682417970142e-05,
      "loss": 0.49632007598876954,
      "mean_token_accuracy": 0.8590896248817443,
      "num_tokens": 31803079.0,
      "step": 5500
    },
    {
      "epoch": 1.466275659824047,
      "eval_entropy": 0.5149196815310102,
      "eval_loss": 0.5756713151931763,
      "eval_mean_token_accuracy": 0.8430842567573894,
      "eval_num_tokens": 31803079.0,
      "eval_runtime": 9.7663,
      "eval_samples_per_second": 646.92,
      "eval_steps_per_second": 20.274,
      "step": 5500
    },
    {
      "entropy": 0.49319090723991393,
      "epoch": 1.4729405491868834,
      "grad_norm": 3.078125,
      "learning_rate": 3.6448467124977175e-05,
      "loss": 0.48577377319335935,
      "mean_token_accuracy": 0.8619768643379211,
      "num_tokens": 31947212.0,
      "step": 5525
    },
    {
      "entropy": 0.5195786714553833,
      "epoch": 1.4796054385497202,
      "grad_norm": 3.359375,
      "learning_rate": 3.6328359010411076e-05,
      "loss": 0.5172930526733398,
      "mean_token_accuracy": 0.8530046033859253,
      "num_tokens": 32093638.0,
      "step": 5550
    },
    {
      "entropy": 0.5013323855400086,
      "epoch": 1.4862703279125566,
      "grad_norm": 3.296875,
      "learning_rate": 3.6207920951526866e-05,
      "loss": 0.49422168731689453,
      "mean_token_accuracy": 0.859441409111023,
      "num_tokens": 32238776.0,
      "step": 5575
    },
    {
      "entropy": 0.4985967803001404,
      "epoch": 1.4929352172753934,
      "grad_norm": 3.34375,
      "learning_rate": 3.608715645614534e-05,
      "loss": 0.49081180572509764,
      "mean_token_accuracy": 0.8593882775306702,
      "num_tokens": 32384009.0,
      "step": 5600
    },
    {
      "epoch": 1.4929352172753934,
      "eval_entropy": 0.5067303588595053,
      "eval_loss": 0.5764662623405457,
      "eval_mean_token_accuracy": 0.8430337418209423,
      "eval_num_tokens": 32384009.0,
      "eval_runtime": 9.772,
      "eval_samples_per_second": 646.54,
      "eval_steps_per_second": 20.262,
      "step": 5600
    },
    {
      "entropy": 0.47397088646888735,
      "epoch": 1.4996001066382298,
      "grad_norm": 3.171875,
      "learning_rate": 3.596606904159492e-05,
      "loss": 0.46940338134765625,
      "mean_token_accuracy": 0.8659658598899841,
      "num_tokens": 32533004.0,
      "step": 5625
    },
    {
      "entropy": 0.4999105751514435,
      "epoch": 1.5062649960010663,
      "grad_norm": 3.390625,
      "learning_rate": 3.584466223460921e-05,
      "loss": 0.49122276306152346,
      "mean_token_accuracy": 0.8597587943077087,
      "num_tokens": 32678523.0,
      "step": 5650
    },
    {
      "entropy": 0.5061901009082794,
      "epoch": 1.5129298853639028,
      "grad_norm": 2.625,
      "learning_rate": 3.57229395712243e-05,
      "loss": 0.5069366836547852,
      "mean_token_accuracy": 0.8579014086723328,
      "num_tokens": 32826187.0,
      "step": 5675
    },
    {
      "entropy": 0.49365248799324035,
      "epoch": 1.5195947747267395,
      "grad_norm": 3.046875,
      "learning_rate": 3.560090459667572e-05,
      "loss": 0.4923128128051758,
      "mean_token_accuracy": 0.8574353384971619,
      "num_tokens": 32971071.0,
      "step": 5700
    },
    {
      "epoch": 1.5195947747267395,
      "eval_entropy": 0.5142042793408789,
      "eval_loss": 0.5750279426574707,
      "eval_mean_token_accuracy": 0.8430195231028278,
      "eval_num_tokens": 32971071.0,
      "eval_runtime": 9.7601,
      "eval_samples_per_second": 647.329,
      "eval_steps_per_second": 20.287,
      "step": 5700
    },
    {
      "entropy": 0.49530532121658327,
      "epoch": 1.5262596640895763,
      "grad_norm": 3.203125,
      "learning_rate": 3.547856086529527e-05,
      "loss": 0.48944438934326173,
      "mean_token_accuracy": 0.8606444954872131,
      "num_tokens": 33114470.0,
      "step": 5725
    },
    {
      "entropy": 0.49027254700660705,
      "epoch": 1.5329245534524127,
      "grad_norm": 2.84375,
      "learning_rate": 3.5355911940407435e-05,
      "loss": 0.4825494003295898,
      "mean_token_accuracy": 0.8620397138595581,
      "num_tokens": 33257823.0,
      "step": 5750
    },
    {
      "entropy": 0.4736780667304993,
      "epoch": 1.5395894428152492,
      "grad_norm": 2.890625,
      "learning_rate": 3.5232961394225606e-05,
      "loss": 0.47526275634765625,
      "mean_token_accuracy": 0.8632498574256897,
      "num_tokens": 33402525.0,
      "step": 5775
    },
    {
      "entropy": 0.4866432321071625,
      "epoch": 1.5462543321780857,
      "grad_norm": 3.359375,
      "learning_rate": 3.510971280774807e-05,
      "loss": 0.48605846405029296,
      "mean_token_accuracy": 0.8616202688217163,
      "num_tokens": 33544042.0,
      "step": 5800
    },
    {
      "epoch": 1.5462543321780857,
      "eval_entropy": 0.5177262515732737,
      "eval_loss": 0.5741413235664368,
      "eval_mean_token_accuracy": 0.8431975621767719,
      "eval_num_tokens": 33544042.0,
      "eval_runtime": 9.7502,
      "eval_samples_per_second": 647.986,
      "eval_steps_per_second": 20.307,
      "step": 5800
    },
    {
      "entropy": 0.5127836644649506,
      "epoch": 1.5529192215409224,
      "grad_norm": 3.109375,
      "learning_rate": 3.4986169770653685e-05,
      "loss": 0.5063155746459961,
      "mean_token_accuracy": 0.8584924054145813,
      "num_tokens": 33688957.0,
      "step": 5825
    },
    {
      "entropy": 0.4810756719112396,
      "epoch": 1.5595841109037591,
      "grad_norm": 3.0625,
      "learning_rate": 3.486233588119732e-05,
      "loss": 0.47811145782470704,
      "mean_token_accuracy": 0.8630669164657593,
      "num_tokens": 33840021.0,
      "step": 5850
    },
    {
      "entropy": 0.512798752784729,
      "epoch": 1.5662490002665956,
      "grad_norm": 3.1875,
      "learning_rate": 3.47382147461051e-05,
      "loss": 0.5082141876220703,
      "mean_token_accuracy": 0.8553845024108887,
      "num_tokens": 33981313.0,
      "step": 5875
    },
    {
      "entropy": 0.4986651134490967,
      "epoch": 1.5729138896294321,
      "grad_norm": 3.484375,
      "learning_rate": 3.461380998046932e-05,
      "loss": 0.4948250198364258,
      "mean_token_accuracy": 0.860196053981781,
      "num_tokens": 34127089.0,
      "step": 5900
    },
    {
      "epoch": 1.5729138896294321,
      "eval_entropy": 0.514950349625915,
      "eval_loss": 0.5741170644760132,
      "eval_mean_token_accuracy": 0.843467779231794,
      "eval_num_tokens": 34127089.0,
      "eval_runtime": 9.7876,
      "eval_samples_per_second": 645.509,
      "eval_steps_per_second": 20.23,
      "step": 5900
    },
    {
      "entropy": 0.5126995348930359,
      "epoch": 1.5795787789922686,
      "grad_norm": 3.25,
      "learning_rate": 3.4489125207643124e-05,
      "loss": 0.5107212448120118,
      "mean_token_accuracy": 0.856031756401062,
      "num_tokens": 34274965.0,
      "step": 5925
    },
    {
      "entropy": 0.48536608219146726,
      "epoch": 1.5862436683551053,
      "grad_norm": 2.8125,
      "learning_rate": 3.436416405913506e-05,
      "loss": 0.4784140777587891,
      "mean_token_accuracy": 0.8637364315986633,
      "num_tokens": 34420906.0,
      "step": 5950
    },
    {
      "entropy": 0.5012638473510742,
      "epoch": 1.592908557717942,
      "grad_norm": 3.28125,
      "learning_rate": 3.4238930174503245e-05,
      "loss": 0.5017874145507812,
      "mean_token_accuracy": 0.8579901337623597,
      "num_tokens": 34563858.0,
      "step": 5975
    },
    {
      "entropy": 0.4964439618587494,
      "epoch": 1.5995734470807785,
      "grad_norm": 3.59375,
      "learning_rate": 3.411342720124935e-05,
      "loss": 0.4876750946044922,
      "mean_token_accuracy": 0.8590740132331848,
      "num_tokens": 34709615.0,
      "step": 6000
    },
    {
      "epoch": 1.5995734470807785,
      "eval_entropy": 0.5166444302809359,
      "eval_loss": 0.5727928280830383,
      "eval_mean_token_accuracy": 0.8436289360426893,
      "eval_num_tokens": 34709615.0,
      "eval_runtime": 9.7585,
      "eval_samples_per_second": 647.439,
      "eval_steps_per_second": 20.29,
      "step": 6000
    },
    {
      "entropy": 0.5095919418334961,
      "epoch": 1.606238336443615,
      "grad_norm": 2.90625,
      "learning_rate": 3.398765879471243e-05,
      "loss": 0.5062215805053711,
      "mean_token_accuracy": 0.8556327033042908,
      "num_tokens": 34848011.0,
      "step": 6025
    },
    {
      "entropy": 0.48980743050575254,
      "epoch": 1.6129032258064515,
      "grad_norm": 3.21875,
      "learning_rate": 3.3861628617962405e-05,
      "loss": 0.4866968536376953,
      "mean_token_accuracy": 0.8607393360137939,
      "num_tokens": 34991660.0,
      "step": 6050
    },
    {
      "entropy": 0.4836041057109833,
      "epoch": 1.6195681151692882,
      "grad_norm": 3.046875,
      "learning_rate": 3.373534034169338e-05,
      "loss": 0.4809332275390625,
      "mean_token_accuracy": 0.8616054844856262,
      "num_tokens": 35133930.0,
      "step": 6075
    },
    {
      "entropy": 0.49632117271423337,
      "epoch": 1.6262330045321247,
      "grad_norm": 3.890625,
      "learning_rate": 3.360879764411675e-05,
      "loss": 0.4947856903076172,
      "mean_token_accuracy": 0.8588719320297241,
      "num_tokens": 35277569.0,
      "step": 6100
    },
    {
      "epoch": 1.6262330045321247,
      "eval_entropy": 0.5053547709578216,
      "eval_loss": 0.5732483863830566,
      "eval_mean_token_accuracy": 0.843414017648408,
      "eval_num_tokens": 35277569.0,
      "eval_runtime": 9.7626,
      "eval_samples_per_second": 647.161,
      "eval_steps_per_second": 20.281,
      "step": 6100
    },
    {
      "entropy": 0.504781242609024,
      "epoch": 1.6328978938949614,
      "grad_norm": 3.15625,
      "learning_rate": 3.3482004210854066e-05,
      "loss": 0.5003813552856445,
      "mean_token_accuracy": 0.8580192995071411,
      "num_tokens": 35421807.0,
      "step": 6125
    },
    {
      "entropy": 0.5044054019451142,
      "epoch": 1.639562783257798,
      "grad_norm": 2.859375,
      "learning_rate": 3.335496373482969e-05,
      "loss": 0.4991204452514648,
      "mean_token_accuracy": 0.8587190723419189,
      "num_tokens": 35565851.0,
      "step": 6150
    },
    {
      "entropy": 0.48343592166900634,
      "epoch": 1.6462276726206344,
      "grad_norm": 2.578125,
      "learning_rate": 3.322767991616321e-05,
      "loss": 0.4804368591308594,
      "mean_token_accuracy": 0.8624867463111877,
      "num_tokens": 35712823.0,
      "step": 6175
    },
    {
      "entropy": 0.4863465356826782,
      "epoch": 1.6528925619834711,
      "grad_norm": 2.734375,
      "learning_rate": 3.3100156462061716e-05,
      "loss": 0.48566131591796874,
      "mean_token_accuracy": 0.8613511276245117,
      "num_tokens": 35856646.0,
      "step": 6200
    },
    {
      "epoch": 1.6528925619834711,
      "eval_entropy": 0.5089494792198894,
      "eval_loss": 0.5731666684150696,
      "eval_mean_token_accuracy": 0.8438188578143264,
      "eval_num_tokens": 35856646.0,
      "eval_runtime": 9.7435,
      "eval_samples_per_second": 648.435,
      "eval_steps_per_second": 20.321,
      "step": 6200
    },
    {
      "entropy": 0.5108045828342438,
      "epoch": 1.6595574513463076,
      "grad_norm": 2.859375,
      "learning_rate": 3.297239708671179e-05,
      "loss": 0.5045115661621093,
      "mean_token_accuracy": 0.8573472452163696,
      "num_tokens": 36000083.0,
      "step": 6225
    },
    {
      "entropy": 0.5037453770637512,
      "epoch": 1.6662223407091443,
      "grad_norm": 2.921875,
      "learning_rate": 3.284440551117135e-05,
      "loss": 0.49875808715820313,
      "mean_token_accuracy": 0.8576627206802369,
      "num_tokens": 36145480.0,
      "step": 6250
    },
    {
      "entropy": 0.4987742877006531,
      "epoch": 1.6728872300719808,
      "grad_norm": 3.59375,
      "learning_rate": 3.2716185463261275e-05,
      "loss": 0.4970406723022461,
      "mean_token_accuracy": 0.8577119374275207,
      "num_tokens": 36289656.0,
      "step": 6275
    },
    {
      "entropy": 0.5184943521022797,
      "epoch": 1.6795521194348173,
      "grad_norm": 2.796875,
      "learning_rate": 3.258774067745679e-05,
      "loss": 0.5158401489257812,
      "mean_token_accuracy": 0.8556339120864869,
      "num_tokens": 36434160.0,
      "step": 6300
    },
    {
      "epoch": 1.6795521194348173,
      "eval_entropy": 0.516247521295692,
      "eval_loss": 0.5708701610565186,
      "eval_mean_token_accuracy": 0.8439788658811589,
      "eval_num_tokens": 36434160.0,
      "eval_runtime": 9.7299,
      "eval_samples_per_second": 649.338,
      "eval_steps_per_second": 20.35,
      "step": 6300
    },
    {
      "entropy": 0.5124625873565674,
      "epoch": 1.6862170087976538,
      "grad_norm": 2.421875,
      "learning_rate": 3.245907489477877e-05,
      "loss": 0.5039989089965821,
      "mean_token_accuracy": 0.8565243029594422,
      "num_tokens": 36577669.0,
      "step": 6325
    },
    {
      "entropy": 0.4982371270656586,
      "epoch": 1.6928818981604905,
      "grad_norm": 3.078125,
      "learning_rate": 3.2330191862684686e-05,
      "loss": 0.5009930419921875,
      "mean_token_accuracy": 0.8570191264152527,
      "num_tokens": 36719642.0,
      "step": 6350
    },
    {
      "entropy": 0.5068892049789429,
      "epoch": 1.6995467875233272,
      "grad_norm": 3.609375,
      "learning_rate": 3.2201095334959564e-05,
      "loss": 0.49831764221191405,
      "mean_token_accuracy": 0.8586463546752929,
      "num_tokens": 36862722.0,
      "step": 6375
    },
    {
      "entropy": 0.5055415034294128,
      "epoch": 1.7062116768861637,
      "grad_norm": 3.140625,
      "learning_rate": 3.207178907160656e-05,
      "loss": 0.5018918991088868,
      "mean_token_accuracy": 0.8585444331169129,
      "num_tokens": 37009725.0,
      "step": 6400
    },
    {
      "epoch": 1.7062116768861637,
      "eval_entropy": 0.5109388096766039,
      "eval_loss": 0.5707947015762329,
      "eval_mean_token_accuracy": 0.8440865744846036,
      "eval_num_tokens": 37009725.0,
      "eval_runtime": 9.7406,
      "eval_samples_per_second": 648.628,
      "eval_steps_per_second": 20.327,
      "step": 6400
    },
    {
      "entropy": 0.5061308288574219,
      "epoch": 1.7128765662490002,
      "grad_norm": 2.859375,
      "learning_rate": 3.194227683873751e-05,
      "loss": 0.5021739196777344,
      "mean_token_accuracy": 0.8566106986999512,
      "num_tokens": 37156764.0,
      "step": 6425
    },
    {
      "entropy": 0.4989737367630005,
      "epoch": 1.7195414556118367,
      "grad_norm": 3.34375,
      "learning_rate": 3.181256240846322e-05,
      "loss": 0.4957284164428711,
      "mean_token_accuracy": 0.8585223269462585,
      "num_tokens": 37300679.0,
      "step": 6450
    },
    {
      "entropy": 0.5016226756572724,
      "epoch": 1.7262063449746734,
      "grad_norm": 3.3125,
      "learning_rate": 3.168264955878357e-05,
      "loss": 0.4927267074584961,
      "mean_token_accuracy": 0.8600739169120789,
      "num_tokens": 37443806.0,
      "step": 6475
    },
    {
      "entropy": 0.49852060914039614,
      "epoch": 1.7328712343375101,
      "grad_norm": 3.71875,
      "learning_rate": 3.1552542073477555e-05,
      "loss": 0.49685848236083985,
      "mean_token_accuracy": 0.8604237747192383,
      "num_tokens": 37589360.0,
      "step": 6500
    },
    {
      "epoch": 1.7328712343375101,
      "eval_entropy": 0.49901027947363225,
      "eval_loss": 0.5710014700889587,
      "eval_mean_token_accuracy": 0.844109443101016,
      "eval_num_tokens": 37589360.0,
      "eval_runtime": 9.7867,
      "eval_samples_per_second": 645.572,
      "eval_steps_per_second": 20.232,
      "step": 6500
    },
    {
      "entropy": 0.4926502549648285,
      "epoch": 1.7395361237003466,
      "grad_norm": 3.15625,
      "learning_rate": 3.1422243741993004e-05,
      "loss": 0.49145046234130857,
      "mean_token_accuracy": 0.861022162437439,
      "num_tokens": 37734141.0,
      "step": 6525
    },
    {
      "entropy": 0.4888096308708191,
      "epoch": 1.746201013063183,
      "grad_norm": 2.921875,
      "learning_rate": 3.129175835933624e-05,
      "loss": 0.48122230529785154,
      "mean_token_accuracy": 0.8632526302337646,
      "num_tokens": 37877679.0,
      "step": 6550
    },
    {
      "entropy": 0.4736886584758759,
      "epoch": 1.7528659024260196,
      "grad_norm": 2.78125,
      "learning_rate": 3.116108972596155e-05,
      "loss": 0.47542301177978513,
      "mean_token_accuracy": 0.8642934894561768,
      "num_tokens": 38018387.0,
      "step": 6575
    },
    {
      "entropy": 0.5113972222805023,
      "epoch": 1.7595307917888563,
      "grad_norm": 3.3125,
      "learning_rate": 3.103024164766051e-05,
      "loss": 0.5091039657592773,
      "mean_token_accuracy": 0.855690279006958,
      "num_tokens": 38163941.0,
      "step": 6600
    },
    {
      "epoch": 1.7595307917888563,
      "eval_entropy": 0.5185635198246349,
      "eval_loss": 0.5696987509727478,
      "eval_mean_token_accuracy": 0.8440857758425703,
      "eval_num_tokens": 38163941.0,
      "eval_runtime": 9.7585,
      "eval_samples_per_second": 647.433,
      "eval_steps_per_second": 20.29,
      "step": 6600
    },
    {
      "entropy": 0.5031931173801422,
      "epoch": 1.766195681151693,
      "grad_norm": 3.0,
      "learning_rate": 3.08992179354511e-05,
      "loss": 0.5011128234863281,
      "mean_token_accuracy": 0.8584766578674317,
      "num_tokens": 38308591.0,
      "step": 6625
    },
    {
      "entropy": 0.5072784328460693,
      "epoch": 1.7728605705145295,
      "grad_norm": 3.140625,
      "learning_rate": 3.076802240546673e-05,
      "loss": 0.49884384155273437,
      "mean_token_accuracy": 0.8589383006095886,
      "num_tokens": 38454830.0,
      "step": 6650
    },
    {
      "entropy": 0.4829687595367432,
      "epoch": 1.779525459877366,
      "grad_norm": 3.046875,
      "learning_rate": 3.0636658878845115e-05,
      "loss": 0.485014762878418,
      "mean_token_accuracy": 0.8610359907150269,
      "num_tokens": 38600055.0,
      "step": 6675
    },
    {
      "entropy": 0.5141796481609344,
      "epoch": 1.7861903492402025,
      "grad_norm": 3.203125,
      "learning_rate": 3.0505131181616924e-05,
      "loss": 0.5080356597900391,
      "mean_token_accuracy": 0.8556164121627807,
      "num_tokens": 38741783.0,
      "step": 6700
    },
    {
      "epoch": 1.7861903492402025,
      "eval_entropy": 0.5062719550397661,
      "eval_loss": 0.5697488784790039,
      "eval_mean_token_accuracy": 0.8441806059293072,
      "eval_num_tokens": 38741783.0,
      "eval_runtime": 9.7587,
      "eval_samples_per_second": 647.425,
      "eval_steps_per_second": 20.29,
      "step": 6700
    },
    {
      "entropy": 0.5165781116485596,
      "epoch": 1.7928552386030392,
      "grad_norm": 2.921875,
      "learning_rate": 3.0373443144594417e-05,
      "loss": 0.5115277099609375,
      "mean_token_accuracy": 0.8551289105415344,
      "num_tokens": 38886287.0,
      "step": 6725
    },
    {
      "entropy": 0.5084601831436157,
      "epoch": 1.799520127965876,
      "grad_norm": 3.125,
      "learning_rate": 3.0241598603259802e-05,
      "loss": 0.5081824874877929,
      "mean_token_accuracy": 0.8570996880531311,
      "num_tokens": 39026302.0,
      "step": 6750
    },
    {
      "entropy": 0.4965075993537903,
      "epoch": 1.8061850173287124,
      "grad_norm": 3.453125,
      "learning_rate": 3.0109601397653576e-05,
      "loss": 0.48743534088134766,
      "mean_token_accuracy": 0.8598896431922912,
      "num_tokens": 39168386.0,
      "step": 6775
    },
    {
      "entropy": 0.5045452630519867,
      "epoch": 1.8128499066915489,
      "grad_norm": 2.71875,
      "learning_rate": 2.9977455372262648e-05,
      "loss": 0.5021046066284179,
      "mean_token_accuracy": 0.8576905584335327,
      "num_tokens": 39312926.0,
      "step": 6800
    },
    {
      "epoch": 1.8128499066915489,
      "eval_entropy": 0.5086117632160283,
      "eval_loss": 0.5695776343345642,
      "eval_mean_token_accuracy": 0.8444965972442819,
      "eval_num_tokens": 39312926.0,
      "eval_runtime": 9.7468,
      "eval_samples_per_second": 648.215,
      "eval_steps_per_second": 20.314,
      "step": 6800
    },
    {
      "entropy": 0.4822874164581299,
      "epoch": 1.8195147960543854,
      "grad_norm": 2.75,
      "learning_rate": 2.9845164375908392e-05,
      "loss": 0.4734258270263672,
      "mean_token_accuracy": 0.8644535994529724,
      "num_tokens": 39458863.0,
      "step": 6825
    },
    {
      "entropy": 0.47904945015907285,
      "epoch": 1.826179685417222,
      "grad_norm": 3.09375,
      "learning_rate": 2.9712732261634524e-05,
      "loss": 0.4761051559448242,
      "mean_token_accuracy": 0.864869990348816,
      "num_tokens": 39604291.0,
      "step": 6850
    },
    {
      "entropy": 0.5039735972881317,
      "epoch": 1.8328445747800588,
      "grad_norm": 2.421875,
      "learning_rate": 2.9580162886594927e-05,
      "loss": 0.5048819732666016,
      "mean_token_accuracy": 0.8575394487380982,
      "num_tokens": 39746988.0,
      "step": 6875
    },
    {
      "entropy": 0.5102309358119964,
      "epoch": 1.8395094641428953,
      "grad_norm": 3.390625,
      "learning_rate": 2.944746011194121e-05,
      "loss": 0.501180534362793,
      "mean_token_accuracy": 0.8581396126747132,
      "num_tokens": 39896227.0,
      "step": 6900
    },
    {
      "epoch": 1.8395094641428953,
      "eval_entropy": 0.49915121029121706,
      "eval_loss": 0.5695159435272217,
      "eval_mean_token_accuracy": 0.8448578269794734,
      "eval_num_tokens": 39896227.0,
      "eval_runtime": 9.7856,
      "eval_samples_per_second": 645.64,
      "eval_steps_per_second": 20.234,
      "step": 6900
    },
    {
      "entropy": 0.5096822345256805,
      "epoch": 1.8461743535057318,
      "grad_norm": 3.078125,
      "learning_rate": 2.931462780271037e-05,
      "loss": 0.5166770935058593,
      "mean_token_accuracy": 0.8526557636260986,
      "num_tokens": 40044922.0,
      "step": 6925
    },
    {
      "entropy": 0.49531367778778074,
      "epoch": 1.8528392428685683,
      "grad_norm": 3.21875,
      "learning_rate": 2.9181669827712154e-05,
      "loss": 0.4857900619506836,
      "mean_token_accuracy": 0.861467981338501,
      "num_tokens": 40186746.0,
      "step": 6950
    },
    {
      "entropy": 0.4967197239398956,
      "epoch": 1.859504132231405,
      "grad_norm": 3.59375,
      "learning_rate": 2.9048590059416374e-05,
      "loss": 0.49272445678710936,
      "mean_token_accuracy": 0.8590538740158081,
      "num_tokens": 40334495.0,
      "step": 6975
    },
    {
      "entropy": 0.5200507783889771,
      "epoch": 1.8661690215942417,
      "grad_norm": 3.34375,
      "learning_rate": 2.8915392373840155e-05,
      "loss": 0.520328369140625,
      "mean_token_accuracy": 0.8518788957595825,
      "num_tokens": 40479579.0,
      "step": 7000
    },
    {
      "epoch": 1.8661690215942417,
      "eval_entropy": 0.510908834560953,
      "eval_loss": 0.567832887172699,
      "eval_mean_token_accuracy": 0.84469555905371,
      "eval_num_tokens": 40479579.0,
      "eval_runtime": 9.7404,
      "eval_samples_per_second": 648.637,
      "eval_steps_per_second": 20.328,
      "step": 7000
    },
    {
      "entropy": 0.492070107460022,
      "epoch": 1.8728339109570782,
      "grad_norm": 3.5,
      "learning_rate": 2.8782080650435006e-05,
      "loss": 0.48320369720458983,
      "mean_token_accuracy": 0.8616286325454712,
      "num_tokens": 40624190.0,
      "step": 7025
    },
    {
      "entropy": 0.4801434397697449,
      "epoch": 1.8794988003199147,
      "grad_norm": 3.3125,
      "learning_rate": 2.8648658771973845e-05,
      "loss": 0.47655860900878905,
      "mean_token_accuracy": 0.8648339509963989,
      "num_tokens": 40766872.0,
      "step": 7050
    },
    {
      "entropy": 0.4984069561958313,
      "epoch": 1.8861636896827512,
      "grad_norm": 3.234375,
      "learning_rate": 2.8515130624437947e-05,
      "loss": 0.5001305770874024,
      "mean_token_accuracy": 0.8588688015937805,
      "num_tokens": 40911531.0,
      "step": 7075
    },
    {
      "entropy": 0.5127998423576355,
      "epoch": 1.8928285790455879,
      "grad_norm": 3.15625,
      "learning_rate": 2.8381500096903675e-05,
      "loss": 0.5017485427856445,
      "mean_token_accuracy": 0.8586782503128052,
      "num_tokens": 41057748.0,
      "step": 7100
    },
    {
      "epoch": 1.8928285790455879,
      "eval_entropy": 0.5053765608204736,
      "eval_loss": 0.5682053565979004,
      "eval_mean_token_accuracy": 0.8449366836234776,
      "eval_num_tokens": 41057748.0,
      "eval_runtime": 9.7654,
      "eval_samples_per_second": 646.981,
      "eval_steps_per_second": 20.276,
      "step": 7100
    },
    {
      "entropy": 0.48358229756355287,
      "epoch": 1.8994934684084244,
      "grad_norm": 2.84375,
      "learning_rate": 2.8247771081429314e-05,
      "loss": 0.48294887542724607,
      "mean_token_accuracy": 0.8613047981262207,
      "num_tokens": 41200923.0,
      "step": 7125
    },
    {
      "entropy": 0.4903048026561737,
      "epoch": 1.906158357771261,
      "grad_norm": 3.59375,
      "learning_rate": 2.8113947472941626e-05,
      "loss": 0.48239028930664063,
      "mean_token_accuracy": 0.8625064349174499,
      "num_tokens": 41350705.0,
      "step": 7150
    },
    {
      "entropy": 0.4935940110683441,
      "epoch": 1.9128232471340976,
      "grad_norm": 3.125,
      "learning_rate": 2.7980033169122454e-05,
      "loss": 0.49292110443115233,
      "mean_token_accuracy": 0.8594210958480835,
      "num_tokens": 41497018.0,
      "step": 7175
    },
    {
      "entropy": 0.46626673221588133,
      "epoch": 1.919488136496934,
      "grad_norm": 3.203125,
      "learning_rate": 2.78460320702952e-05,
      "loss": 0.4608114624023438,
      "mean_token_accuracy": 0.8685131549835206,
      "num_tokens": 41641476.0,
      "step": 7200
    },
    {
      "epoch": 1.919488136496934,
      "eval_entropy": 0.502743075140799,
      "eval_loss": 0.568652868270874,
      "eval_mean_token_accuracy": 0.8447539123019787,
      "eval_num_tokens": 41641476.0,
      "eval_runtime": 9.7447,
      "eval_samples_per_second": 648.352,
      "eval_steps_per_second": 20.319,
      "step": 7200
    },
    {
      "entropy": 0.4976526975631714,
      "epoch": 1.9261530258597708,
      "grad_norm": 3.0625,
      "learning_rate": 2.7711948079311208e-05,
      "loss": 0.49244281768798825,
      "mean_token_accuracy": 0.8588139820098877,
      "num_tokens": 41787681.0,
      "step": 7225
    },
    {
      "entropy": 0.5004709732532501,
      "epoch": 1.9328179152226073,
      "grad_norm": 2.890625,
      "learning_rate": 2.7577785101436093e-05,
      "loss": 0.4953230667114258,
      "mean_token_accuracy": 0.8588651013374329,
      "num_tokens": 41932769.0,
      "step": 7250
    },
    {
      "entropy": 0.5028802716732025,
      "epoch": 1.939482804585444,
      "grad_norm": 2.953125,
      "learning_rate": 2.7443547044236005e-05,
      "loss": 0.5052907180786133,
      "mean_token_accuracy": 0.8557114601135254,
      "num_tokens": 42081263.0,
      "step": 7275
    },
    {
      "entropy": 0.4994118547439575,
      "epoch": 1.9461476939482805,
      "grad_norm": 3.28125,
      "learning_rate": 2.7309237817463834e-05,
      "loss": 0.4907569122314453,
      "mean_token_accuracy": 0.8586895990371705,
      "num_tokens": 42226329.0,
      "step": 7300
    },
    {
      "epoch": 1.9461476939482805,
      "eval_entropy": 0.5078689188066156,
      "eval_loss": 0.5674465894699097,
      "eval_mean_token_accuracy": 0.8448128823680107,
      "eval_num_tokens": 42226329.0,
      "eval_runtime": 9.7816,
      "eval_samples_per_second": 645.906,
      "eval_steps_per_second": 20.242,
      "step": 7300
    },
    {
      "entropy": 0.5100186550617218,
      "epoch": 1.952812583311117,
      "grad_norm": 2.90625,
      "learning_rate": 2.717486133294529e-05,
      "loss": 0.5126686477661133,
      "mean_token_accuracy": 0.8554815769195556,
      "num_tokens": 42372348.0,
      "step": 7325
    },
    {
      "entropy": 0.5032690846920014,
      "epoch": 1.9594774726739534,
      "grad_norm": 3.28125,
      "learning_rate": 2.704042150446504e-05,
      "loss": 0.4902206802368164,
      "mean_token_accuracy": 0.8585510444641113,
      "num_tokens": 42520268.0,
      "step": 7350
    },
    {
      "entropy": 0.47538632750511167,
      "epoch": 1.9661423620367902,
      "grad_norm": 3.0625,
      "learning_rate": 2.690592224765265e-05,
      "loss": 0.4766672515869141,
      "mean_token_accuracy": 0.8637108612060547,
      "num_tokens": 42663953.0,
      "step": 7375
    },
    {
      "entropy": 0.48658151149749757,
      "epoch": 1.9728072513996269,
      "grad_norm": 2.953125,
      "learning_rate": 2.6771367479868555e-05,
      "loss": 0.481767463684082,
      "mean_token_accuracy": 0.8610238194465637,
      "num_tokens": 42805722.0,
      "step": 7400
    },
    {
      "epoch": 1.9728072513996269,
      "eval_entropy": 0.5022936142755278,
      "eval_loss": 0.5677194595336914,
      "eval_mean_token_accuracy": 0.8449866741594642,
      "eval_num_tokens": 42805722.0,
      "eval_runtime": 9.7558,
      "eval_samples_per_second": 647.617,
      "eval_steps_per_second": 20.296,
      "step": 7400
    },
    {
      "entropy": 0.47726617693901063,
      "epoch": 1.9794721407624634,
      "grad_norm": 2.796875,
      "learning_rate": 2.663676112009001e-05,
      "loss": 0.4719820785522461,
      "mean_token_accuracy": 0.8651622080802918,
      "num_tokens": 42951632.0,
      "step": 7425
    },
    {
      "entropy": 0.4992399764060974,
      "epoch": 1.9861370301252999,
      "grad_norm": 3.25,
      "learning_rate": 2.6502107088796884e-05,
      "loss": 0.49747184753417967,
      "mean_token_accuracy": 0.8609014844894409,
      "num_tokens": 43093577.0,
      "step": 7450
    },
    {
      "entropy": 0.4883934187889099,
      "epoch": 1.9928019194881363,
      "grad_norm": 3.484375,
      "learning_rate": 2.636740930785754e-05,
      "loss": 0.48338459014892576,
      "mean_token_accuracy": 0.86230060338974,
      "num_tokens": 43239387.0,
      "step": 7475
    },
    {
      "entropy": 0.48324010491371155,
      "epoch": 1.999466808850973,
      "grad_norm": 3.171875,
      "learning_rate": 2.623267170041452e-05,
      "loss": 0.47823593139648435,
      "mean_token_accuracy": 0.8628026294708252,
      "num_tokens": 43383771.0,
      "step": 7500
    },
    {
      "epoch": 1.999466808850973,
      "eval_entropy": 0.5054499470826351,
      "eval_loss": 0.5665329098701477,
      "eval_mean_token_accuracy": 0.8452398024424158,
      "eval_num_tokens": 43383771.0,
      "eval_runtime": 9.7522,
      "eval_samples_per_second": 647.852,
      "eval_steps_per_second": 20.303,
      "step": 7500
    }
  ],
  "logging_steps": 25,
  "max_steps": 15004,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 9.880736163988992e+16,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}