{
  "best_metric": 1.9687457084655762,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2969561989606533,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0014847809948032665,
      "grad_norm": 0.38478878140449524,
      "learning_rate": 1e-05,
      "loss": 2.1167,
      "step": 1
    },
    {
      "epoch": 0.0014847809948032665,
      "eval_loss": 2.15017032623291,
      "eval_runtime": 52.6512,
      "eval_samples_per_second": 86.171,
      "eval_steps_per_second": 5.394,
      "step": 1
    },
    {
      "epoch": 0.002969561989606533,
      "grad_norm": 0.44478335976600647,
      "learning_rate": 2e-05,
      "loss": 2.0261,
      "step": 2
    },
    {
      "epoch": 0.004454342984409799,
      "grad_norm": 0.4396665692329407,
      "learning_rate": 3e-05,
      "loss": 2.1107,
      "step": 3
    },
    {
      "epoch": 0.005939123979213066,
      "grad_norm": 0.3937435448169708,
      "learning_rate": 4e-05,
      "loss": 2.1302,
      "step": 4
    },
    {
      "epoch": 0.007423904974016332,
      "grad_norm": 0.36997291445732117,
      "learning_rate": 5e-05,
      "loss": 2.0641,
      "step": 5
    },
    {
      "epoch": 0.008908685968819599,
      "grad_norm": 0.33418500423431396,
      "learning_rate": 6e-05,
      "loss": 2.0832,
      "step": 6
    },
    {
      "epoch": 0.010393466963622866,
      "grad_norm": 0.3295334279537201,
      "learning_rate": 7e-05,
      "loss": 2.0553,
      "step": 7
    },
    {
      "epoch": 0.011878247958426132,
      "grad_norm": 0.323632150888443,
      "learning_rate": 8e-05,
      "loss": 2.0556,
      "step": 8
    },
    {
      "epoch": 0.013363028953229399,
      "grad_norm": 0.3482978641986847,
      "learning_rate": 9e-05,
      "loss": 2.1089,
      "step": 9
    },
    {
      "epoch": 0.014847809948032665,
      "grad_norm": 0.38115793466567993,
      "learning_rate": 0.0001,
      "loss": 2.1441,
      "step": 10
    },
    {
      "epoch": 0.016332590942835932,
      "grad_norm": 0.37201857566833496,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.1967,
      "step": 11
    },
    {
      "epoch": 0.017817371937639197,
      "grad_norm": 0.3628121018409729,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.1804,
      "step": 12
    },
    {
      "epoch": 0.019302152932442463,
      "grad_norm": 0.301399290561676,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.2342,
      "step": 13
    },
    {
      "epoch": 0.020786933927245732,
      "grad_norm": 0.39368724822998047,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.958,
      "step": 14
    },
    {
      "epoch": 0.022271714922048998,
      "grad_norm": 0.28724896907806396,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.9894,
      "step": 15
    },
    {
      "epoch": 0.023756495916852263,
      "grad_norm": 0.27313336730003357,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.0118,
      "step": 16
    },
    {
      "epoch": 0.025241276911655532,
      "grad_norm": 0.2967425584793091,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.989,
      "step": 17
    },
    {
      "epoch": 0.026726057906458798,
      "grad_norm": 0.29300904273986816,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.0379,
      "step": 18
    },
    {
      "epoch": 0.028210838901262063,
      "grad_norm": 0.29489371180534363,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.0324,
      "step": 19
    },
    {
      "epoch": 0.02969561989606533,
      "grad_norm": 0.28660160303115845,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.0173,
      "step": 20
    },
    {
      "epoch": 0.031180400890868598,
      "grad_norm": 0.28985676169395447,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.1266,
      "step": 21
    },
    {
      "epoch": 0.032665181885671864,
      "grad_norm": 0.29200324416160583,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.078,
      "step": 22
    },
    {
      "epoch": 0.03414996288047513,
      "grad_norm": 0.3007466495037079,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.0785,
      "step": 23
    },
    {
      "epoch": 0.035634743875278395,
      "grad_norm": 0.31664276123046875,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.1601,
      "step": 24
    },
    {
      "epoch": 0.03711952487008166,
      "grad_norm": 0.35239487886428833,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.2085,
      "step": 25
    },
    {
      "epoch": 0.038604305864884926,
      "grad_norm": 0.24031668901443481,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.9822,
      "step": 26
    },
    {
      "epoch": 0.0400890868596882,
      "grad_norm": 0.2429058849811554,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.9006,
      "step": 27
    },
    {
      "epoch": 0.041573867854491464,
      "grad_norm": 0.22939959168434143,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.9815,
      "step": 28
    },
    {
      "epoch": 0.04305864884929473,
      "grad_norm": 0.23532599210739136,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.991,
      "step": 29
    },
    {
      "epoch": 0.044543429844097995,
      "grad_norm": 0.24697649478912354,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.0062,
      "step": 30
    },
    {
      "epoch": 0.04602821083890126,
      "grad_norm": 0.27173495292663574,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.9554,
      "step": 31
    },
    {
      "epoch": 0.047512991833704527,
      "grad_norm": 0.2755003869533539,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.006,
      "step": 32
    },
    {
      "epoch": 0.04899777282850779,
      "grad_norm": 0.26322251558303833,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.0641,
      "step": 33
    },
    {
      "epoch": 0.050482553823311065,
      "grad_norm": 0.2873618006706238,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.0246,
      "step": 34
    },
    {
      "epoch": 0.05196733481811433,
      "grad_norm": 0.2915358245372772,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.0805,
      "step": 35
    },
    {
      "epoch": 0.053452115812917596,
      "grad_norm": 0.2966706454753876,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.0495,
      "step": 36
    },
    {
      "epoch": 0.05493689680772086,
      "grad_norm": 0.33095332980155945,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.1258,
      "step": 37
    },
    {
      "epoch": 0.05642167780252413,
      "grad_norm": 0.2613125443458557,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.1623,
      "step": 38
    },
    {
      "epoch": 0.05790645879732739,
      "grad_norm": 0.2278221845626831,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.9022,
      "step": 39
    },
    {
      "epoch": 0.05939123979213066,
      "grad_norm": 0.24880492687225342,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.9832,
      "step": 40
    },
    {
      "epoch": 0.06087602078693393,
      "grad_norm": 0.24073165655136108,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.9382,
      "step": 41
    },
    {
      "epoch": 0.062360801781737196,
      "grad_norm": 0.24341227114200592,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.0172,
      "step": 42
    },
    {
      "epoch": 0.06384558277654045,
      "grad_norm": 0.24885092675685883,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.958,
      "step": 43
    },
    {
      "epoch": 0.06533036377134373,
      "grad_norm": 0.2509358525276184,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.9621,
      "step": 44
    },
    {
      "epoch": 0.066815144766147,
      "grad_norm": 0.2590627372264862,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.929,
      "step": 45
    },
    {
      "epoch": 0.06829992576095026,
      "grad_norm": 0.2795308530330658,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.0423,
      "step": 46
    },
    {
      "epoch": 0.06978470675575353,
      "grad_norm": 0.28098785877227783,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.001,
      "step": 47
    },
    {
      "epoch": 0.07126948775055679,
      "grad_norm": 0.30104294419288635,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.9982,
      "step": 48
    },
    {
      "epoch": 0.07275426874536006,
      "grad_norm": 0.3143393397331238,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.0799,
      "step": 49
    },
    {
      "epoch": 0.07423904974016332,
      "grad_norm": 0.33249884843826294,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.1751,
      "step": 50
    },
    {
      "epoch": 0.07423904974016332,
      "eval_loss": 2.0129618644714355,
      "eval_runtime": 19.5295,
      "eval_samples_per_second": 232.315,
      "eval_steps_per_second": 14.542,
      "step": 50
    },
    {
      "epoch": 0.0757238307349666,
      "grad_norm": 0.2234942764043808,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.9796,
      "step": 51
    },
    {
      "epoch": 0.07720861172976985,
      "grad_norm": 0.24600103497505188,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.9358,
      "step": 52
    },
    {
      "epoch": 0.07869339272457312,
      "grad_norm": 0.2583761513233185,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.9341,
      "step": 53
    },
    {
      "epoch": 0.0801781737193764,
      "grad_norm": 0.25638312101364136,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.9443,
      "step": 54
    },
    {
      "epoch": 0.08166295471417966,
      "grad_norm": 0.25296470522880554,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.9503,
      "step": 55
    },
    {
      "epoch": 0.08314773570898293,
      "grad_norm": 0.2555272579193115,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.9624,
      "step": 56
    },
    {
      "epoch": 0.08463251670378619,
      "grad_norm": 0.262175053358078,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.0029,
      "step": 57
    },
    {
      "epoch": 0.08611729769858946,
      "grad_norm": 0.2732691466808319,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.9835,
      "step": 58
    },
    {
      "epoch": 0.08760207869339272,
      "grad_norm": 0.27853691577911377,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.0423,
      "step": 59
    },
    {
      "epoch": 0.08908685968819599,
      "grad_norm": 0.28769248723983765,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.0913,
      "step": 60
    },
    {
      "epoch": 0.09057164068299926,
      "grad_norm": 0.30930954217910767,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.1202,
      "step": 61
    },
    {
      "epoch": 0.09205642167780252,
      "grad_norm": 0.3266344666481018,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.1216,
      "step": 62
    },
    {
      "epoch": 0.0935412026726058,
      "grad_norm": 0.24870085716247559,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.1835,
      "step": 63
    },
    {
      "epoch": 0.09502598366740905,
      "grad_norm": 0.21612682938575745,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.868,
      "step": 64
    },
    {
      "epoch": 0.09651076466221233,
      "grad_norm": 0.23229877650737762,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.9209,
      "step": 65
    },
    {
      "epoch": 0.09799554565701558,
      "grad_norm": 0.2308695763349533,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.9352,
      "step": 66
    },
    {
      "epoch": 0.09948032665181886,
      "grad_norm": 0.23685020208358765,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.0197,
      "step": 67
    },
    {
      "epoch": 0.10096510764662213,
      "grad_norm": 0.23842664062976837,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.9883,
      "step": 68
    },
    {
      "epoch": 0.10244988864142539,
      "grad_norm": 0.25624990463256836,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.9837,
      "step": 69
    },
    {
      "epoch": 0.10393466963622866,
      "grad_norm": 0.2614176869392395,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.9812,
      "step": 70
    },
    {
      "epoch": 0.10541945063103192,
      "grad_norm": 0.27144908905029297,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.9914,
      "step": 71
    },
    {
      "epoch": 0.10690423162583519,
      "grad_norm": 0.28132250905036926,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.0116,
      "step": 72
    },
    {
      "epoch": 0.10838901262063845,
      "grad_norm": 0.2873154878616333,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.0177,
      "step": 73
    },
    {
      "epoch": 0.10987379361544172,
      "grad_norm": 0.30333152413368225,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.0645,
      "step": 74
    },
    {
      "epoch": 0.111358574610245,
      "grad_norm": 0.32396766543388367,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.214,
      "step": 75
    },
    {
      "epoch": 0.11284335560504825,
      "grad_norm": 0.23978875577449799,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.9924,
      "step": 76
    },
    {
      "epoch": 0.11432813659985153,
      "grad_norm": 0.2556561231613159,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.8942,
      "step": 77
    },
    {
      "epoch": 0.11581291759465479,
      "grad_norm": 0.24838104844093323,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.9889,
      "step": 78
    },
    {
      "epoch": 0.11729769858945806,
      "grad_norm": 0.24040506780147552,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.0221,
      "step": 79
    },
    {
      "epoch": 0.11878247958426132,
      "grad_norm": 0.24800269305706024,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.9852,
      "step": 80
    },
    {
      "epoch": 0.12026726057906459,
      "grad_norm": 0.24833384156227112,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.9952,
      "step": 81
    },
    {
      "epoch": 0.12175204157386786,
      "grad_norm": 0.2735847234725952,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.9449,
      "step": 82
    },
    {
      "epoch": 0.12323682256867112,
      "grad_norm": 0.2768011689186096,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.0202,
      "step": 83
    },
    {
      "epoch": 0.12472160356347439,
      "grad_norm": 0.28170162439346313,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.0467,
      "step": 84
    },
    {
      "epoch": 0.12620638455827765,
      "grad_norm": 0.2887212038040161,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.0257,
      "step": 85
    },
    {
      "epoch": 0.1276911655530809,
      "grad_norm": 0.3012441396713257,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.0491,
      "step": 86
    },
    {
      "epoch": 0.1291759465478842,
      "grad_norm": 0.3268975019454956,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.1463,
      "step": 87
    },
    {
      "epoch": 0.13066072754268745,
      "grad_norm": 0.27081364393234253,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.1791,
      "step": 88
    },
    {
      "epoch": 0.1321455085374907,
      "grad_norm": 0.22752290964126587,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.8788,
      "step": 89
    },
    {
      "epoch": 0.133630289532294,
      "grad_norm": 0.23880364000797272,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.8947,
      "step": 90
    },
    {
      "epoch": 0.13511507052709726,
      "grad_norm": 0.24774649739265442,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.9353,
      "step": 91
    },
    {
      "epoch": 0.13659985152190052,
      "grad_norm": 0.25214967131614685,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.9933,
      "step": 92
    },
    {
      "epoch": 0.13808463251670378,
      "grad_norm": 0.25763872265815735,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.9347,
      "step": 93
    },
    {
      "epoch": 0.13956941351150706,
      "grad_norm": 0.2547333538532257,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.8945,
      "step": 94
    },
    {
      "epoch": 0.14105419450631032,
      "grad_norm": 0.2755119800567627,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.0194,
      "step": 95
    },
    {
      "epoch": 0.14253897550111358,
      "grad_norm": 0.27266886830329895,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.9371,
      "step": 96
    },
    {
      "epoch": 0.14402375649591687,
      "grad_norm": 0.2779705822467804,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.0305,
      "step": 97
    },
    {
      "epoch": 0.14550853749072012,
      "grad_norm": 0.3047863841056824,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.0574,
      "step": 98
    },
    {
      "epoch": 0.14699331848552338,
      "grad_norm": 0.3120288848876953,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.1114,
      "step": 99
    },
    {
      "epoch": 0.14847809948032664,
      "grad_norm": 0.3263275623321533,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.184,
      "step": 100
    },
    {
      "epoch": 0.14847809948032664,
      "eval_loss": 1.9841808080673218,
      "eval_runtime": 19.6074,
      "eval_samples_per_second": 231.392,
      "eval_steps_per_second": 14.484,
      "step": 100
    },
    {
      "epoch": 0.14996288047512993,
      "grad_norm": 0.2335229367017746,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.9675,
      "step": 101
    },
    {
      "epoch": 0.1514476614699332,
      "grad_norm": 0.22848185896873474,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.8204,
      "step": 102
    },
    {
      "epoch": 0.15293244246473645,
      "grad_norm": 0.2401008903980255,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.9805,
      "step": 103
    },
    {
      "epoch": 0.1544172234595397,
      "grad_norm": 0.2521536350250244,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.999,
      "step": 104
    },
    {
      "epoch": 0.155902004454343,
      "grad_norm": 0.25453588366508484,
      "learning_rate": 5e-05,
      "loss": 1.9693,
      "step": 105
    },
    {
      "epoch": 0.15738678544914625,
      "grad_norm": 0.2561187446117401,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.9207,
      "step": 106
    },
    {
      "epoch": 0.1588715664439495,
      "grad_norm": 0.2744416296482086,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.9337,
      "step": 107
    },
    {
      "epoch": 0.1603563474387528,
      "grad_norm": 0.280840665102005,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.9139,
      "step": 108
    },
    {
      "epoch": 0.16184112843355605,
      "grad_norm": 0.28846701979637146,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.0067,
      "step": 109
    },
    {
      "epoch": 0.1633259094283593,
      "grad_norm": 0.3039008378982544,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.015,
      "step": 110
    },
    {
      "epoch": 0.16481069042316257,
      "grad_norm": 0.3108331263065338,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.0476,
      "step": 111
    },
    {
      "epoch": 0.16629547141796586,
      "grad_norm": 0.3125537931919098,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.0825,
      "step": 112
    },
    {
      "epoch": 0.16778025241276912,
      "grad_norm": 0.2841561436653137,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.1979,
      "step": 113
    },
    {
      "epoch": 0.16926503340757237,
      "grad_norm": 0.23780225217342377,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.845,
      "step": 114
    },
    {
      "epoch": 0.17074981440237566,
      "grad_norm": 0.2500491738319397,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.9024,
      "step": 115
    },
    {
      "epoch": 0.17223459539717892,
      "grad_norm": 0.24424073100090027,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.8818,
      "step": 116
    },
    {
      "epoch": 0.17371937639198218,
      "grad_norm": 0.25010746717453003,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.923,
      "step": 117
    },
    {
      "epoch": 0.17520415738678544,
      "grad_norm": 0.2599712014198303,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.9959,
      "step": 118
    },
    {
      "epoch": 0.17668893838158872,
      "grad_norm": 0.26703882217407227,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.9729,
      "step": 119
    },
    {
      "epoch": 0.17817371937639198,
      "grad_norm": 0.27770882844924927,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.9894,
      "step": 120
    },
    {
      "epoch": 0.17965850037119524,
      "grad_norm": 0.2750227153301239,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.9627,
      "step": 121
    },
    {
      "epoch": 0.18114328136599853,
      "grad_norm": 0.2809279263019562,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.9626,
      "step": 122
    },
    {
      "epoch": 0.18262806236080179,
      "grad_norm": 0.30249762535095215,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.0459,
      "step": 123
    },
    {
      "epoch": 0.18411284335560504,
      "grad_norm": 0.30125948786735535,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.0274,
      "step": 124
    },
    {
      "epoch": 0.1855976243504083,
      "grad_norm": 0.34516477584838867,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.1014,
      "step": 125
    },
    {
      "epoch": 0.1870824053452116,
      "grad_norm": 0.22669430077075958,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.9298,
      "step": 126
    },
    {
      "epoch": 0.18856718634001485,
      "grad_norm": 0.22487293183803558,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.8454,
      "step": 127
    },
    {
      "epoch": 0.1900519673348181,
      "grad_norm": 0.24330025911331177,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.9126,
      "step": 128
    },
    {
      "epoch": 0.1915367483296214,
      "grad_norm": 0.2637227475643158,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.9629,
      "step": 129
    },
    {
      "epoch": 0.19302152932442465,
      "grad_norm": 0.2630452513694763,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.9063,
      "step": 130
    },
    {
      "epoch": 0.1945063103192279,
      "grad_norm": 0.28585508465766907,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.9545,
      "step": 131
    },
    {
      "epoch": 0.19599109131403117,
      "grad_norm": 0.2664121985435486,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.8999,
      "step": 132
    },
    {
      "epoch": 0.19747587230883445,
      "grad_norm": 0.2800696790218353,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.9773,
      "step": 133
    },
    {
      "epoch": 0.1989606533036377,
      "grad_norm": 0.2868615388870239,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.9594,
      "step": 134
    },
    {
      "epoch": 0.20044543429844097,
      "grad_norm": 0.3034520745277405,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.0202,
      "step": 135
    },
    {
      "epoch": 0.20193021529324426,
      "grad_norm": 0.30918583273887634,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.0719,
      "step": 136
    },
    {
      "epoch": 0.20341499628804752,
      "grad_norm": 0.33706042170524597,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.1243,
      "step": 137
    },
    {
      "epoch": 0.20489977728285078,
      "grad_norm": 0.2731862962245941,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.1643,
      "step": 138
    },
    {
      "epoch": 0.20638455827765403,
      "grad_norm": 0.23040184378623962,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.8253,
      "step": 139
    },
    {
      "epoch": 0.20786933927245732,
      "grad_norm": 0.2567889392375946,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.8995,
      "step": 140
    },
    {
      "epoch": 0.20935412026726058,
      "grad_norm": 0.25470319390296936,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.8981,
      "step": 141
    },
    {
      "epoch": 0.21083890126206384,
      "grad_norm": 0.2646588981151581,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.9524,
      "step": 142
    },
    {
      "epoch": 0.21232368225686712,
      "grad_norm": 0.2872580885887146,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.931,
      "step": 143
    },
    {
      "epoch": 0.21380846325167038,
      "grad_norm": 0.2654319107532501,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.9036,
      "step": 144
    },
    {
      "epoch": 0.21529324424647364,
      "grad_norm": 0.2975412607192993,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.9443,
      "step": 145
    },
    {
      "epoch": 0.2167780252412769,
      "grad_norm": 0.2751605212688446,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.8959,
      "step": 146
    },
    {
      "epoch": 0.2182628062360802,
      "grad_norm": 0.2967882752418518,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.0217,
      "step": 147
    },
    {
      "epoch": 0.21974758723088345,
      "grad_norm": 0.30078670382499695,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.0321,
      "step": 148
    },
    {
      "epoch": 0.2212323682256867,
      "grad_norm": 0.31555402278900146,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.0162,
      "step": 149
    },
    {
      "epoch": 0.22271714922049,
      "grad_norm": 0.34547528624534607,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.1877,
      "step": 150
    },
    {
      "epoch": 0.22271714922049,
      "eval_loss": 1.972423791885376,
      "eval_runtime": 19.565,
      "eval_samples_per_second": 231.893,
      "eval_steps_per_second": 14.516,
      "step": 150
    },
    {
      "epoch": 0.22420193021529325,
      "grad_norm": 0.23464946448802948,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.9895,
      "step": 151
    },
    {
      "epoch": 0.2256867112100965,
      "grad_norm": 0.2432500422000885,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.8907,
      "step": 152
    },
    {
      "epoch": 0.22717149220489977,
      "grad_norm": 0.246421217918396,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.8851,
      "step": 153
    },
    {
      "epoch": 0.22865627319970305,
      "grad_norm": 0.24729110300540924,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.844,
      "step": 154
    },
    {
      "epoch": 0.2301410541945063,
      "grad_norm": 0.2709508538246155,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.9877,
      "step": 155
    },
    {
      "epoch": 0.23162583518930957,
      "grad_norm": 0.2616763710975647,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.9132,
      "step": 156
    },
    {
      "epoch": 0.23311061618411286,
      "grad_norm": 0.2788386046886444,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.913,
      "step": 157
    },
    {
      "epoch": 0.23459539717891612,
      "grad_norm": 0.27590465545654297,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.9731,
      "step": 158
    },
    {
      "epoch": 0.23608017817371937,
      "grad_norm": 0.2902533710002899,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.9987,
      "step": 159
    },
    {
      "epoch": 0.23756495916852263,
      "grad_norm": 0.3047283887863159,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.9959,
      "step": 160
    },
    {
      "epoch": 0.23904974016332592,
      "grad_norm": 0.31083446741104126,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.0085,
      "step": 161
    },
    {
      "epoch": 0.24053452115812918,
      "grad_norm": 0.3371656537055969,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.0768,
      "step": 162
    },
    {
      "epoch": 0.24201930215293244,
      "grad_norm": 0.264365017414093,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.1202,
      "step": 163
    },
    {
      "epoch": 0.24350408314773572,
      "grad_norm": 0.23497478663921356,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.8068,
      "step": 164
    },
    {
      "epoch": 0.24498886414253898,
      "grad_norm": 0.2511955499649048,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.9219,
      "step": 165
    },
    {
      "epoch": 0.24647364513734224,
      "grad_norm": 0.26190921664237976,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.895,
      "step": 166
    },
    {
      "epoch": 0.2479584261321455,
      "grad_norm": 0.2678329348564148,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.9421,
      "step": 167
    },
    {
      "epoch": 0.24944320712694878,
      "grad_norm": 0.2638859450817108,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.9652,
      "step": 168
    },
    {
      "epoch": 0.25092798812175204,
      "grad_norm": 0.2646275460720062,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.8834,
      "step": 169
    },
    {
      "epoch": 0.2524127691165553,
      "grad_norm": 0.27891165018081665,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.9645,
      "step": 170
    },
    {
      "epoch": 0.25389755011135856,
      "grad_norm": 0.308635950088501,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.9686,
      "step": 171
    },
    {
      "epoch": 0.2553823311061618,
      "grad_norm": 0.3122882843017578,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.9763,
      "step": 172
    },
    {
      "epoch": 0.25686711210096513,
      "grad_norm": 0.3170386850833893,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.0198,
      "step": 173
    },
    {
      "epoch": 0.2583518930957684,
      "grad_norm": 0.3260398507118225,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.0544,
      "step": 174
    },
    {
      "epoch": 0.25983667409057165,
      "grad_norm": 0.35422399640083313,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.1192,
      "step": 175
    },
    {
      "epoch": 0.2613214550853749,
      "grad_norm": 0.2319270670413971,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.9963,
      "step": 176
    },
    {
      "epoch": 0.26280623608017817,
      "grad_norm": 0.234304279088974,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.8388,
      "step": 177
    },
    {
      "epoch": 0.2642910170749814,
      "grad_norm": 0.2515241801738739,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.9062,
      "step": 178
    },
    {
      "epoch": 0.2657757980697847,
      "grad_norm": 0.2559499740600586,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.9525,
      "step": 179
    },
    {
      "epoch": 0.267260579064588,
      "grad_norm": 0.2721082866191864,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.8927,
      "step": 180
    },
    {
      "epoch": 0.26874536005939126,
      "grad_norm": 0.2704063951969147,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.9709,
      "step": 181
    },
    {
      "epoch": 0.2702301410541945,
      "grad_norm": 0.2726072371006012,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.9695,
      "step": 182
    },
    {
      "epoch": 0.2717149220489978,
      "grad_norm": 0.2886669635772705,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.9648,
      "step": 183
    },
    {
      "epoch": 0.27319970304380103,
      "grad_norm": 0.30987393856048584,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.9806,
      "step": 184
    },
    {
      "epoch": 0.2746844840386043,
      "grad_norm": 0.304328978061676,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.0566,
      "step": 185
    },
    {
      "epoch": 0.27616926503340755,
      "grad_norm": 0.3262093961238861,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.1196,
      "step": 186
    },
    {
      "epoch": 0.27765404602821087,
      "grad_norm": 0.33939123153686523,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.1053,
      "step": 187
    },
    {
      "epoch": 0.2791388270230141,
      "grad_norm": 0.2607070505619049,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.1044,
      "step": 188
    },
    {
      "epoch": 0.2806236080178174,
      "grad_norm": 0.23139512538909912,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.864,
      "step": 189
    },
    {
      "epoch": 0.28210838901262064,
      "grad_norm": 0.24001334607601166,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.8862,
      "step": 190
    },
    {
      "epoch": 0.2835931700074239,
      "grad_norm": 0.25511983036994934,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.9277,
      "step": 191
    },
    {
      "epoch": 0.28507795100222716,
      "grad_norm": 0.26207661628723145,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.9528,
      "step": 192
    },
    {
      "epoch": 0.2865627319970304,
      "grad_norm": 0.26416125893592834,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.9352,
      "step": 193
    },
    {
      "epoch": 0.28804751299183373,
      "grad_norm": 0.26347023248672485,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.9231,
      "step": 194
    },
    {
      "epoch": 0.289532293986637,
      "grad_norm": 0.27364036440849304,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.9293,
      "step": 195
    },
    {
      "epoch": 0.29101707498144025,
      "grad_norm": 0.28394055366516113,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.9283,
      "step": 196
    },
    {
      "epoch": 0.2925018559762435,
      "grad_norm": 0.3116365373134613,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.9831,
      "step": 197
    },
    {
      "epoch": 0.29398663697104677,
      "grad_norm": 0.31009212136268616,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.9754,
      "step": 198
    },
    {
      "epoch": 0.29547141796585,
      "grad_norm": 0.31543102860450745,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.0441,
      "step": 199
    },
    {
      "epoch": 0.2969561989606533,
      "grad_norm": 0.35349082946777344,
      "learning_rate": 0.0,
      "loss": 2.1441,
      "step": 200
    },
    {
      "epoch": 0.2969561989606533,
      "eval_loss": 1.9687457084655762,
      "eval_runtime": 19.5217,
      "eval_samples_per_second": 232.408,
      "eval_steps_per_second": 14.548,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.3560930160594125e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}