{
  "best_global_step": 740,
  "best_metric": 0.2394678492239468,
  "best_model_checkpoint": "out_qwen_0.6b_sft_hq/checkpoint-740",
  "epoch": 5.828402366863905,
  "eval_steps": 10,
  "global_step": 740,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03944773175542406,
      "grad_norm": 137.58168100959043,
      "learning_rate": 1.9607843137254904e-07,
      "loss": 33.8811,
      "step": 5,
      "true_loss": 4.4586
    },
    {
      "epoch": 0.07889546351084813,
      "grad_norm": 96.05399842775171,
      "learning_rate": 4.4117647058823536e-07,
      "loss": 33.9447,
      "step": 10,
      "true_loss": 4.3732
    },
    {
      "epoch": 0.07889546351084813,
      "eval_accuracy": 0.022172949002217297,
      "eval_loss": 4.352970123291016,
      "eval_runtime": 15.7645,
      "eval_samples_per_second": 57.217,
      "eval_steps_per_second": 7.168,
      "step": 10
    },
    {
      "epoch": 0.11834319526627218,
      "grad_norm": 119.9455370680162,
      "learning_rate": 6.862745098039217e-07,
      "loss": 35.4633,
      "step": 15,
      "true_loss": 4.4363
    },
    {
      "epoch": 0.15779092702169625,
      "grad_norm": 115.69507181998236,
      "learning_rate": 9.313725490196079e-07,
      "loss": 35.4213,
      "step": 20,
      "true_loss": 4.3914
    },
    {
      "epoch": 0.15779092702169625,
      "eval_accuracy": 0.02106430155210643,
      "eval_loss": 4.327692031860352,
      "eval_runtime": 14.936,
      "eval_samples_per_second": 60.391,
      "eval_steps_per_second": 7.566,
      "step": 20
    },
    {
      "epoch": 0.19723865877712032,
      "grad_norm": 109.18491807311037,
      "learning_rate": 1.1764705882352942e-06,
      "loss": 34.7285,
      "step": 25,
      "true_loss": 4.5508
    },
    {
      "epoch": 0.23668639053254437,
      "grad_norm": 114.28553037452185,
      "learning_rate": 1.4215686274509805e-06,
      "loss": 34.658,
      "step": 30,
      "true_loss": 4.4059
    },
    {
      "epoch": 0.23668639053254437,
      "eval_accuracy": 0.017738359201773836,
      "eval_loss": 4.260367393493652,
      "eval_runtime": 14.9567,
      "eval_samples_per_second": 60.308,
      "eval_steps_per_second": 7.555,
      "step": 30
    },
    {
      "epoch": 0.27613412228796846,
      "grad_norm": 109.9826234366543,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 34.1344,
      "step": 35,
      "true_loss": 4.1516
    },
    {
      "epoch": 0.3155818540433925,
      "grad_norm": 123.05318590857651,
      "learning_rate": 1.9117647058823528e-06,
      "loss": 34.1885,
      "step": 40,
      "true_loss": 4.0762
    },
    {
      "epoch": 0.3155818540433925,
      "eval_accuracy": 0.015521064301552107,
      "eval_loss": 4.189310073852539,
      "eval_runtime": 14.9234,
      "eval_samples_per_second": 60.442,
      "eval_steps_per_second": 7.572,
      "step": 40
    },
    {
      "epoch": 0.35502958579881655,
      "grad_norm": 101.89262765170473,
      "learning_rate": 2.1568627450980393e-06,
      "loss": 33.7348,
      "step": 45,
      "true_loss": 4.5547
    },
    {
      "epoch": 0.39447731755424065,
      "grad_norm": 118.10971042116216,
      "learning_rate": 2.401960784313726e-06,
      "loss": 33.5551,
      "step": 50,
      "true_loss": 4.1461
    },
    {
      "epoch": 0.39447731755424065,
      "eval_accuracy": 0.024390243902439025,
      "eval_loss": 4.126463890075684,
      "eval_runtime": 15.0954,
      "eval_samples_per_second": 59.753,
      "eval_steps_per_second": 7.486,
      "step": 50
    },
    {
      "epoch": 0.4339250493096647,
      "grad_norm": 126.8109210779904,
      "learning_rate": 2.647058823529412e-06,
      "loss": 33.3662,
      "step": 55,
      "true_loss": 4.1771
    },
    {
      "epoch": 0.47337278106508873,
      "grad_norm": 107.75818756256407,
      "learning_rate": 2.8921568627450985e-06,
      "loss": 32.4912,
      "step": 60,
      "true_loss": 3.9953
    },
    {
      "epoch": 0.47337278106508873,
      "eval_accuracy": 0.036585365853658534,
      "eval_loss": 4.055848121643066,
      "eval_runtime": 14.9842,
      "eval_samples_per_second": 60.197,
      "eval_steps_per_second": 7.541,
      "step": 60
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 126.96202525436865,
      "learning_rate": 3.1372549019607846e-06,
      "loss": 32.3756,
      "step": 65,
      "true_loss": 4.057
    },
    {
      "epoch": 0.5522682445759369,
      "grad_norm": 121.42814891257407,
      "learning_rate": 3.382352941176471e-06,
      "loss": 32.7852,
      "step": 70,
      "true_loss": 3.8746
    },
    {
      "epoch": 0.5522682445759369,
      "eval_accuracy": 0.04767184035476718,
      "eval_loss": 4.019028663635254,
      "eval_runtime": 15.0235,
      "eval_samples_per_second": 60.039,
      "eval_steps_per_second": 7.522,
      "step": 70
    },
    {
      "epoch": 0.591715976331361,
      "grad_norm": 152.86102648476006,
      "learning_rate": 3.6274509803921573e-06,
      "loss": 31.9488,
      "step": 75,
      "true_loss": 3.9531
    },
    {
      "epoch": 0.631163708086785,
      "grad_norm": 141.9292910226069,
      "learning_rate": 3.872549019607843e-06,
      "loss": 32.1604,
      "step": 80,
      "true_loss": 4.1195
    },
    {
      "epoch": 0.631163708086785,
      "eval_accuracy": 0.05432372505543237,
      "eval_loss": 3.9854836463928223,
      "eval_runtime": 14.9619,
      "eval_samples_per_second": 60.286,
      "eval_steps_per_second": 7.552,
      "step": 80
    },
    {
      "epoch": 0.6706114398422091,
      "grad_norm": 132.12828795355875,
      "learning_rate": 4.11764705882353e-06,
      "loss": 32.0111,
      "step": 85,
      "true_loss": 4.0219
    },
    {
      "epoch": 0.7100591715976331,
      "grad_norm": 153.0719114159677,
      "learning_rate": 4.3627450980392164e-06,
      "loss": 31.752,
      "step": 90,
      "true_loss": 3.9547
    },
    {
      "epoch": 0.7100591715976331,
      "eval_accuracy": 0.058758314855875834,
      "eval_loss": 3.9391109943389893,
      "eval_runtime": 14.9724,
      "eval_samples_per_second": 60.244,
      "eval_steps_per_second": 7.547,
      "step": 90
    },
    {
      "epoch": 0.7495069033530573,
      "grad_norm": 128.25085846897863,
      "learning_rate": 4.607843137254902e-06,
      "loss": 32.32,
      "step": 95,
      "true_loss": 4.1264
    },
    {
      "epoch": 0.7889546351084813,
      "grad_norm": 132.66281463976432,
      "learning_rate": 4.852941176470589e-06,
      "loss": 31.9783,
      "step": 100,
      "true_loss": 4.1824
    },
    {
      "epoch": 0.7889546351084813,
      "eval_accuracy": 0.082039911308204,
      "eval_loss": 3.891880989074707,
      "eval_runtime": 15.0549,
      "eval_samples_per_second": 59.914,
      "eval_steps_per_second": 7.506,
      "step": 100
    },
    {
      "epoch": 0.8284023668639053,
      "grad_norm": 138.12009424260248,
      "learning_rate": 4.989059080962801e-06,
      "loss": 31.175,
      "step": 105,
      "true_loss": 3.9373
    },
    {
      "epoch": 0.8678500986193294,
      "grad_norm": 139.39677324279805,
      "learning_rate": 4.961706783369804e-06,
      "loss": 31.9854,
      "step": 110,
      "true_loss": 4.1117
    },
    {
      "epoch": 0.8678500986193294,
      "eval_accuracy": 0.09866962305986696,
      "eval_loss": 3.860440254211426,
      "eval_runtime": 14.9426,
      "eval_samples_per_second": 60.364,
      "eval_steps_per_second": 7.562,
      "step": 110
    },
    {
      "epoch": 0.9072978303747534,
      "grad_norm": 160.76229566662045,
      "learning_rate": 4.934354485776805e-06,
      "loss": 31.7863,
      "step": 115,
      "true_loss": 4.0758
    },
    {
      "epoch": 0.9467455621301775,
      "grad_norm": 141.93625692921623,
      "learning_rate": 4.9070021881838075e-06,
      "loss": 31.4898,
      "step": 120,
      "true_loss": 3.8525
    },
    {
      "epoch": 0.9467455621301775,
      "eval_accuracy": 0.12305986696230599,
      "eval_loss": 3.844113826751709,
      "eval_runtime": 14.91,
      "eval_samples_per_second": 60.496,
      "eval_steps_per_second": 7.579,
      "step": 120
    },
    {
      "epoch": 0.9861932938856016,
      "grad_norm": 156.67839380394838,
      "learning_rate": 4.87964989059081e-06,
      "loss": 31.3083,
      "step": 125,
      "true_loss": 3.6941
    },
    {
      "epoch": 1.0236686390532543,
      "grad_norm": 175.06600661459612,
      "learning_rate": 4.852297592997812e-06,
      "loss": 29.4422,
      "step": 130,
      "true_loss": 3.6871
    },
    {
      "epoch": 1.0236686390532543,
      "eval_accuracy": 0.11419068736141907,
      "eval_loss": 3.816354274749756,
      "eval_runtime": 14.9497,
      "eval_samples_per_second": 60.336,
      "eval_steps_per_second": 7.559,
      "step": 130
    },
    {
      "epoch": 1.0631163708086786,
      "grad_norm": 176.4037265950165,
      "learning_rate": 4.824945295404815e-06,
      "loss": 29.6881,
      "step": 135,
      "true_loss": 3.5447
    },
    {
      "epoch": 1.1025641025641026,
      "grad_norm": 175.43736192079575,
      "learning_rate": 4.797592997811816e-06,
      "loss": 30.1351,
      "step": 140,
      "true_loss": 3.9027
    },
    {
      "epoch": 1.1025641025641026,
      "eval_accuracy": 0.12084257206208426,
      "eval_loss": 3.787785053253174,
      "eval_runtime": 14.9544,
      "eval_samples_per_second": 60.317,
      "eval_steps_per_second": 7.556,
      "step": 140
    },
    {
      "epoch": 1.1420118343195267,
      "grad_norm": 172.66609321830506,
      "learning_rate": 4.770240700218819e-06,
      "loss": 31.1361,
      "step": 145,
      "true_loss": 3.9604
    },
    {
      "epoch": 1.1814595660749507,
      "grad_norm": 188.08419409648823,
      "learning_rate": 4.742888402625821e-06,
      "loss": 29.9185,
      "step": 150,
      "true_loss": 3.6713
    },
    {
      "epoch": 1.1814595660749507,
      "eval_accuracy": 0.12195121951219512,
      "eval_loss": 3.7505717277526855,
      "eval_runtime": 15.0675,
      "eval_samples_per_second": 59.864,
      "eval_steps_per_second": 7.5,
      "step": 150
    },
    {
      "epoch": 1.2209072978303748,
      "grad_norm": 184.54494972485557,
      "learning_rate": 4.715536105032823e-06,
      "loss": 29.7549,
      "step": 155,
      "true_loss": 3.7596
    },
    {
      "epoch": 1.2603550295857988,
      "grad_norm": 230.49181032198453,
      "learning_rate": 4.688183807439826e-06,
      "loss": 29.7484,
      "step": 160,
      "true_loss": 3.9594
    },
    {
      "epoch": 1.2603550295857988,
      "eval_accuracy": 0.13747228381374724,
      "eval_loss": 3.7303473949432373,
      "eval_runtime": 15.2085,
      "eval_samples_per_second": 59.309,
      "eval_steps_per_second": 7.43,
      "step": 160
    },
    {
      "epoch": 1.2998027613412229,
      "grad_norm": 186.21055820946583,
      "learning_rate": 4.660831509846827e-06,
      "loss": 29.388,
      "step": 165,
      "true_loss": 3.6656
    },
    {
      "epoch": 1.339250493096647,
      "grad_norm": 203.5666063639384,
      "learning_rate": 4.63347921225383e-06,
      "loss": 29.4438,
      "step": 170,
      "true_loss": 4.0383
    },
    {
      "epoch": 1.339250493096647,
      "eval_accuracy": 0.13747228381374724,
      "eval_loss": 3.683065414428711,
      "eval_runtime": 15.1745,
      "eval_samples_per_second": 59.442,
      "eval_steps_per_second": 7.447,
      "step": 170
    },
    {
      "epoch": 1.378698224852071,
      "grad_norm": 215.86726316889974,
      "learning_rate": 4.606126914660832e-06,
      "loss": 30.1271,
      "step": 175,
      "true_loss": 3.6135
    },
    {
      "epoch": 1.418145956607495,
      "grad_norm": 191.64160050781922,
      "learning_rate": 4.5787746170678345e-06,
      "loss": 29.468,
      "step": 180,
      "true_loss": 3.7248
    },
    {
      "epoch": 1.418145956607495,
      "eval_accuracy": 0.14855875831485588,
      "eval_loss": 3.663315534591675,
      "eval_runtime": 15.0394,
      "eval_samples_per_second": 59.976,
      "eval_steps_per_second": 7.514,
      "step": 180
    },
    {
      "epoch": 1.4575936883629192,
      "grad_norm": 204.6168296859058,
      "learning_rate": 4.551422319474836e-06,
      "loss": 29.6546,
      "step": 185,
      "true_loss": 3.7043
    },
    {
      "epoch": 1.497041420118343,
      "grad_norm": 219.7996098581931,
      "learning_rate": 4.524070021881838e-06,
      "loss": 29.0906,
      "step": 190,
      "true_loss": 3.6714
    },
    {
      "epoch": 1.497041420118343,
      "eval_accuracy": 0.1385809312638581,
      "eval_loss": 3.6522138118743896,
      "eval_runtime": 14.9508,
      "eval_samples_per_second": 60.331,
      "eval_steps_per_second": 7.558,
      "step": 190
    },
    {
      "epoch": 1.5364891518737673,
      "grad_norm": 177.10887786115575,
      "learning_rate": 4.496717724288841e-06,
      "loss": 29.4486,
      "step": 195,
      "true_loss": 3.9092
    },
    {
      "epoch": 1.5759368836291914,
      "grad_norm": 180.67006503279012,
      "learning_rate": 4.469365426695843e-06,
      "loss": 30.4641,
      "step": 200,
      "true_loss": 3.9328
    },
    {
      "epoch": 1.5759368836291914,
      "eval_accuracy": 0.14079822616407983,
      "eval_loss": 3.65649676322937,
      "eval_runtime": 15.1758,
      "eval_samples_per_second": 59.437,
      "eval_steps_per_second": 7.446,
      "step": 200
    },
    {
      "epoch": 1.6153846153846154,
      "grad_norm": 182.15282062853038,
      "learning_rate": 4.442013129102845e-06,
      "loss": 28.703,
      "step": 205,
      "true_loss": 3.5898
    },
    {
      "epoch": 1.6548323471400395,
      "grad_norm": 187.54953739010512,
      "learning_rate": 4.414660831509847e-06,
      "loss": 29.291,
      "step": 210,
      "true_loss": 3.7754
    },
    {
      "epoch": 1.6548323471400395,
      "eval_accuracy": 0.13414634146341464,
      "eval_loss": 3.650477170944214,
      "eval_runtime": 15.131,
      "eval_samples_per_second": 59.613,
      "eval_steps_per_second": 7.468,
      "step": 210
    },
    {
      "epoch": 1.6942800788954635,
      "grad_norm": 197.72972204694662,
      "learning_rate": 4.3873085339168495e-06,
      "loss": 29.5033,
      "step": 215,
      "true_loss": 3.5539
    },
    {
      "epoch": 1.7337278106508875,
      "grad_norm": 211.1691853811082,
      "learning_rate": 4.359956236323852e-06,
      "loss": 28.2073,
      "step": 220,
      "true_loss": 3.7295
    },
    {
      "epoch": 1.7337278106508875,
      "eval_accuracy": 0.1607538802660754,
      "eval_loss": 3.5888044834136963,
      "eval_runtime": 14.8061,
      "eval_samples_per_second": 60.921,
      "eval_steps_per_second": 7.632,
      "step": 220
    },
    {
      "epoch": 1.7731755424063116,
      "grad_norm": 198.11359915598604,
      "learning_rate": 4.332603938730854e-06,
      "loss": 29.068,
      "step": 225,
      "true_loss": 3.5761
    },
    {
      "epoch": 1.8126232741617359,
      "grad_norm": 208.2967379446751,
      "learning_rate": 4.305251641137856e-06,
      "loss": 29.558,
      "step": 230,
      "true_loss": 3.7469
    },
    {
      "epoch": 1.8126232741617359,
      "eval_accuracy": 0.1662971175166297,
      "eval_loss": 3.567077398300171,
      "eval_runtime": 14.8915,
      "eval_samples_per_second": 60.571,
      "eval_steps_per_second": 7.588,
      "step": 230
    },
    {
      "epoch": 1.8520710059171597,
      "grad_norm": 202.51288282619706,
      "learning_rate": 4.277899343544858e-06,
      "loss": 29.2451,
      "step": 235,
      "true_loss": 3.5877
    },
    {
      "epoch": 1.891518737672584,
      "grad_norm": 204.1287289867651,
      "learning_rate": 4.2505470459518606e-06,
      "loss": 28.581,
      "step": 240,
      "true_loss": 3.4135
    },
    {
      "epoch": 1.891518737672584,
      "eval_accuracy": 0.164079822616408,
      "eval_loss": 3.5610058307647705,
      "eval_runtime": 15.2419,
      "eval_samples_per_second": 59.179,
      "eval_steps_per_second": 7.414,
      "step": 240
    },
    {
      "epoch": 1.9309664694280078,
      "grad_norm": 199.99425110918077,
      "learning_rate": 4.223194748358863e-06,
      "loss": 28.7527,
      "step": 245,
      "true_loss": 3.6467
    },
    {
      "epoch": 1.970414201183432,
      "grad_norm": 209.62816557130563,
      "learning_rate": 4.1958424507658645e-06,
      "loss": 29.8734,
      "step": 250,
      "true_loss": 3.3148
    },
    {
      "epoch": 1.970414201183432,
      "eval_accuracy": 0.1662971175166297,
      "eval_loss": 3.583265542984009,
      "eval_runtime": 14.8799,
      "eval_samples_per_second": 60.619,
      "eval_steps_per_second": 7.594,
      "step": 250
    },
    {
      "epoch": 2.007889546351085,
      "grad_norm": 197.80455811801355,
      "learning_rate": 4.168490153172867e-06,
      "loss": 27.7556,
      "step": 255,
      "true_loss": 4.0312
    },
    {
      "epoch": 2.0473372781065087,
      "grad_norm": 198.8468471778106,
      "learning_rate": 4.141137855579869e-06,
      "loss": 27.6775,
      "step": 260,
      "true_loss": 3.8672
    },
    {
      "epoch": 2.0473372781065087,
      "eval_accuracy": 0.1574279379157428,
      "eval_loss": 3.5533580780029297,
      "eval_runtime": 14.9826,
      "eval_samples_per_second": 60.203,
      "eval_steps_per_second": 7.542,
      "step": 260
    },
    {
      "epoch": 2.086785009861933,
      "grad_norm": 204.45388818689486,
      "learning_rate": 4.113785557986872e-06,
      "loss": 26.7176,
      "step": 265,
      "true_loss": 3.4912
    },
    {
      "epoch": 2.126232741617357,
      "grad_norm": 225.18842668023993,
      "learning_rate": 4.086433260393873e-06,
      "loss": 27.9069,
      "step": 270,
      "true_loss": 3.5805
    },
    {
      "epoch": 2.126232741617357,
      "eval_accuracy": 0.16851441241685144,
      "eval_loss": 3.534597635269165,
      "eval_runtime": 14.8101,
      "eval_samples_per_second": 60.905,
      "eval_steps_per_second": 7.63,
      "step": 270
    },
    {
      "epoch": 2.165680473372781,
      "grad_norm": 232.60579637840817,
      "learning_rate": 4.059080962800876e-06,
      "loss": 26.9553,
      "step": 275,
      "true_loss": 3.1944
    },
    {
      "epoch": 2.2051282051282053,
      "grad_norm": 214.5771122472325,
      "learning_rate": 4.031728665207878e-06,
      "loss": 26.0979,
      "step": 280,
      "true_loss": 3.1361
    },
    {
      "epoch": 2.2051282051282053,
      "eval_accuracy": 0.15299334811529933,
      "eval_loss": 3.5224153995513916,
      "eval_runtime": 15.107,
      "eval_samples_per_second": 59.707,
      "eval_steps_per_second": 7.48,
      "step": 280
    },
    {
      "epoch": 2.244575936883629,
      "grad_norm": 229.00786163080798,
      "learning_rate": 4.00437636761488e-06,
      "loss": 27.3225,
      "step": 285,
      "true_loss": 3.3937
    },
    {
      "epoch": 2.2840236686390534,
      "grad_norm": 220.9809902657408,
      "learning_rate": 3.977024070021882e-06,
      "loss": 27.3824,
      "step": 290,
      "true_loss": 3.3252
    },
    {
      "epoch": 2.2840236686390534,
      "eval_accuracy": 0.16186252771618626,
      "eval_loss": 3.5029642581939697,
      "eval_runtime": 14.9319,
      "eval_samples_per_second": 60.408,
      "eval_steps_per_second": 7.568,
      "step": 290
    },
    {
      "epoch": 2.323471400394477,
      "grad_norm": 224.23746222333077,
      "learning_rate": 3.949671772428884e-06,
      "loss": 27.8245,
      "step": 295,
      "true_loss": 3.3297
    },
    {
      "epoch": 2.3629191321499015,
      "grad_norm": 196.24143904821017,
      "learning_rate": 3.922319474835887e-06,
      "loss": 27.9647,
      "step": 300,
      "true_loss": 3.6294
    },
    {
      "epoch": 2.3629191321499015,
      "eval_accuracy": 0.18181818181818182,
      "eval_loss": 3.5059654712677,
      "eval_runtime": 14.9636,
      "eval_samples_per_second": 60.279,
      "eval_steps_per_second": 7.552,
      "step": 300
    },
    {
      "epoch": 2.4023668639053253,
      "grad_norm": 210.16619384673945,
      "learning_rate": 3.894967177242889e-06,
      "loss": 27.5051,
      "step": 305,
      "true_loss": 3.3774
    },
    {
      "epoch": 2.4418145956607495,
      "grad_norm": 213.67456162572208,
      "learning_rate": 3.867614879649891e-06,
      "loss": 27.3462,
      "step": 310,
      "true_loss": 3.3658
    },
    {
      "epoch": 2.4418145956607495,
      "eval_accuracy": 0.17849223946784923,
      "eval_loss": 3.5055649280548096,
      "eval_runtime": 15.1871,
      "eval_samples_per_second": 59.393,
      "eval_steps_per_second": 7.441,
      "step": 310
    },
    {
      "epoch": 2.4812623274161734,
      "grad_norm": 225.9401296492373,
      "learning_rate": 3.840262582056893e-06,
      "loss": 27.1061,
      "step": 315,
      "true_loss": 3.478
    },
    {
      "epoch": 2.5207100591715976,
      "grad_norm": 205.8362032729107,
      "learning_rate": 3.8129102844638954e-06,
      "loss": 26.3703,
      "step": 320,
      "true_loss": 3.1416
    },
    {
      "epoch": 2.5207100591715976,
      "eval_accuracy": 0.17073170731707318,
      "eval_loss": 3.4906609058380127,
      "eval_runtime": 15.0707,
      "eval_samples_per_second": 59.851,
      "eval_steps_per_second": 7.498,
      "step": 320
    },
    {
      "epoch": 2.560157790927022,
      "grad_norm": 203.21971371657304,
      "learning_rate": 3.7855579868708973e-06,
      "loss": 27.3517,
      "step": 325,
      "true_loss": 3.2373
    },
    {
      "epoch": 2.5996055226824457,
      "grad_norm": 219.93911587157675,
      "learning_rate": 3.7582056892778997e-06,
      "loss": 27.8681,
      "step": 330,
      "true_loss": 3.125
    },
    {
      "epoch": 2.5996055226824457,
      "eval_accuracy": 0.17184035476718404,
      "eval_loss": 3.4929237365722656,
      "eval_runtime": 15.1428,
      "eval_samples_per_second": 59.566,
      "eval_steps_per_second": 7.462,
      "step": 330
    },
    {
      "epoch": 2.63905325443787,
      "grad_norm": 232.49538276234148,
      "learning_rate": 3.7308533916849017e-06,
      "loss": 27.0143,
      "step": 335,
      "true_loss": 3.5833
    },
    {
      "epoch": 2.678500986193294,
      "grad_norm": 218.79623472715534,
      "learning_rate": 3.703501094091904e-06,
      "loss": 27.1542,
      "step": 340,
      "true_loss": 3.425
    },
    {
      "epoch": 2.678500986193294,
      "eval_accuracy": 0.1674057649667406,
      "eval_loss": 3.4812633991241455,
      "eval_runtime": 14.9427,
      "eval_samples_per_second": 60.364,
      "eval_steps_per_second": 7.562,
      "step": 340
    },
    {
      "epoch": 2.717948717948718,
      "grad_norm": 210.9796490946478,
      "learning_rate": 3.676148796498906e-06,
      "loss": 27.0408,
      "step": 345,
      "true_loss": 3.2768
    },
    {
      "epoch": 2.757396449704142,
      "grad_norm": 217.69737467031442,
      "learning_rate": 3.6487964989059084e-06,
      "loss": 27.45,
      "step": 350,
      "true_loss": 3.4411
    },
    {
      "epoch": 2.757396449704142,
      "eval_accuracy": 0.17738359201773837,
      "eval_loss": 3.468329906463623,
      "eval_runtime": 14.9595,
      "eval_samples_per_second": 60.296,
      "eval_steps_per_second": 7.554,
      "step": 350
    },
    {
      "epoch": 2.796844181459566,
      "grad_norm": 223.8506652627376,
      "learning_rate": 3.621444201312911e-06,
      "loss": 26.4817,
      "step": 355,
      "true_loss": 3.4845
    },
    {
      "epoch": 2.83629191321499,
      "grad_norm": 222.52387593334151,
      "learning_rate": 3.5940919037199128e-06,
      "loss": 28.3421,
      "step": 360,
      "true_loss": 3.386
    },
    {
      "epoch": 2.83629191321499,
      "eval_accuracy": 0.19290465631929046,
      "eval_loss": 3.4540300369262695,
      "eval_runtime": 15.1328,
      "eval_samples_per_second": 59.605,
      "eval_steps_per_second": 7.467,
      "step": 360
    },
    {
      "epoch": 2.8757396449704142,
      "grad_norm": 233.9031521180998,
      "learning_rate": 3.5667396061269147e-06,
      "loss": 27.4952,
      "step": 365,
      "true_loss": 3.5988
    },
    {
      "epoch": 2.9151873767258385,
      "grad_norm": 220.40770079671375,
      "learning_rate": 3.539387308533917e-06,
      "loss": 27.7365,
      "step": 370,
      "true_loss": 3.8111
    },
    {
      "epoch": 2.9151873767258385,
      "eval_accuracy": 0.2039911308203991,
      "eval_loss": 3.445399045944214,
      "eval_runtime": 15.0467,
      "eval_samples_per_second": 59.947,
      "eval_steps_per_second": 7.51,
      "step": 370
    },
    {
      "epoch": 2.9546351084812623,
      "grad_norm": 214.3532111476442,
      "learning_rate": 3.5120350109409195e-06,
      "loss": 26.3055,
      "step": 375,
      "true_loss": 3.1235
    },
    {
      "epoch": 2.994082840236686,
      "grad_norm": 217.53508025403696,
      "learning_rate": 3.4846827133479215e-06,
      "loss": 28.0186,
      "step": 380,
      "true_loss": 3.4562
    },
    {
      "epoch": 2.994082840236686,
      "eval_accuracy": 0.1917960088691796,
      "eval_loss": 3.44553542137146,
      "eval_runtime": 14.9341,
      "eval_samples_per_second": 60.399,
      "eval_steps_per_second": 7.567,
      "step": 380
    },
    {
      "epoch": 3.0315581854043394,
      "grad_norm": 213.65664891050722,
      "learning_rate": 3.457330415754924e-06,
      "loss": 24.4655,
      "step": 385,
      "true_loss": 3.3045
    },
    {
      "epoch": 3.0710059171597632,
      "grad_norm": 238.57756114961336,
      "learning_rate": 3.429978118161926e-06,
      "loss": 24.9568,
      "step": 390,
      "true_loss": 3.0463
    },
    {
      "epoch": 3.0710059171597632,
      "eval_accuracy": 0.18957871396895787,
      "eval_loss": 3.4360036849975586,
      "eval_runtime": 15.1499,
      "eval_samples_per_second": 59.538,
      "eval_steps_per_second": 7.459,
      "step": 390
    },
    {
      "epoch": 3.1104536489151875,
      "grad_norm": 242.90427569755158,
      "learning_rate": 3.402625820568928e-06,
      "loss": 25.6816,
      "step": 395,
      "true_loss": 3.1437
    },
    {
      "epoch": 3.1499013806706113,
      "grad_norm": 250.93466465674842,
      "learning_rate": 3.37527352297593e-06,
      "loss": 24.8387,
      "step": 400,
      "true_loss": 3.1525
    },
    {
      "epoch": 3.1499013806706113,
      "eval_accuracy": 0.18514412416851442,
      "eval_loss": 3.435126781463623,
      "eval_runtime": 15.0403,
      "eval_samples_per_second": 59.972,
      "eval_steps_per_second": 7.513,
      "step": 400
    },
    {
      "epoch": 3.1893491124260356,
      "grad_norm": 256.79950170149914,
      "learning_rate": 3.3479212253829326e-06,
      "loss": 25.7926,
      "step": 405,
      "true_loss": 3.0679
    },
    {
      "epoch": 3.2287968441814594,
      "grad_norm": 240.77762669362124,
      "learning_rate": 3.3205689277899345e-06,
      "loss": 26.0472,
      "step": 410,
      "true_loss": 3.2098
    },
    {
      "epoch": 3.2287968441814594,
      "eval_accuracy": 0.19290465631929046,
      "eval_loss": 3.428030014038086,
      "eval_runtime": 15.1004,
      "eval_samples_per_second": 59.733,
      "eval_steps_per_second": 7.483,
      "step": 410
    },
    {
      "epoch": 3.2682445759368837,
      "grad_norm": 237.38142361903007,
      "learning_rate": 3.293216630196937e-06,
      "loss": 25.423,
      "step": 415,
      "true_loss": 3.0787
    },
    {
      "epoch": 3.3076923076923075,
      "grad_norm": 242.6891012315701,
      "learning_rate": 3.2658643326039393e-06,
      "loss": 25.1691,
      "step": 420,
      "true_loss": 3.208
    },
    {
      "epoch": 3.3076923076923075,
      "eval_accuracy": 0.18403547671840353,
      "eval_loss": 3.431478261947632,
      "eval_runtime": 14.9427,
      "eval_samples_per_second": 60.364,
      "eval_steps_per_second": 7.562,
      "step": 420
    },
    {
      "epoch": 3.3471400394477318,
      "grad_norm": 228.6708954920748,
      "learning_rate": 3.2385120350109413e-06,
      "loss": 26.0412,
      "step": 425,
      "true_loss": 3.3517
    },
    {
      "epoch": 3.386587771203156,
      "grad_norm": 244.34477651789024,
      "learning_rate": 3.2111597374179432e-06,
      "loss": 24.4891,
      "step": 430,
      "true_loss": 3.1938
    },
    {
      "epoch": 3.386587771203156,
      "eval_accuracy": 0.2017738359201774,
      "eval_loss": 3.4007480144500732,
      "eval_runtime": 14.9507,
      "eval_samples_per_second": 60.332,
      "eval_steps_per_second": 7.558,
      "step": 430
    },
    {
      "epoch": 3.42603550295858,
      "grad_norm": 247.97686568101082,
      "learning_rate": 3.1838074398249456e-06,
      "loss": 25.6189,
      "step": 435,
      "true_loss": 3.2995
    },
    {
      "epoch": 3.465483234714004,
      "grad_norm": 241.1156299032968,
      "learning_rate": 3.156455142231948e-06,
      "loss": 24.7909,
      "step": 440,
      "true_loss": 2.9935
    },
    {
      "epoch": 3.465483234714004,
      "eval_accuracy": 0.18070953436807094,
      "eval_loss": 3.395143985748291,
      "eval_runtime": 15.0313,
      "eval_samples_per_second": 60.008,
      "eval_steps_per_second": 7.518,
      "step": 440
    },
    {
      "epoch": 3.504930966469428,
      "grad_norm": 257.4476404600661,
      "learning_rate": 3.12910284463895e-06,
      "loss": 25.0061,
      "step": 445,
      "true_loss": 3.4912
    },
    {
      "epoch": 3.544378698224852,
      "grad_norm": 260.2937525280967,
      "learning_rate": 3.101750547045952e-06,
      "loss": 25.6066,
      "step": 450,
      "true_loss": 3.3814
    },
    {
      "epoch": 3.544378698224852,
      "eval_accuracy": 0.18957871396895787,
      "eval_loss": 3.395690679550171,
      "eval_runtime": 14.9346,
      "eval_samples_per_second": 60.397,
      "eval_steps_per_second": 7.566,
      "step": 450
    },
    {
      "epoch": 3.583826429980276,
      "grad_norm": 240.92415923092258,
      "learning_rate": 3.0743982494529543e-06,
      "loss": 26.1029,
      "step": 455,
      "true_loss": 3.2856
    },
    {
      "epoch": 3.6232741617357003,
      "grad_norm": 250.4817822446221,
      "learning_rate": 3.0470459518599567e-06,
      "loss": 24.8986,
      "step": 460,
      "true_loss": 2.8531
    },
    {
      "epoch": 3.6232741617357003,
      "eval_accuracy": 0.18181818181818182,
      "eval_loss": 3.4077982902526855,
      "eval_runtime": 15.0046,
      "eval_samples_per_second": 60.115,
      "eval_steps_per_second": 7.531,
      "step": 460
    },
    {
      "epoch": 3.662721893491124,
      "grad_norm": 273.72732435650465,
      "learning_rate": 3.0196936542669587e-06,
      "loss": 26.2365,
      "step": 465,
      "true_loss": 3.1754
    },
    {
      "epoch": 3.7021696252465484,
      "grad_norm": 266.52186007353083,
      "learning_rate": 2.9923413566739606e-06,
      "loss": 24.3942,
      "step": 470,
      "true_loss": 2.8515
    },
    {
      "epoch": 3.7021696252465484,
      "eval_accuracy": 0.19623059866962306,
      "eval_loss": 3.4032986164093018,
      "eval_runtime": 14.9785,
      "eval_samples_per_second": 60.22,
      "eval_steps_per_second": 7.544,
      "step": 470
    },
    {
      "epoch": 3.7416173570019726,
      "grad_norm": 277.7285610260442,
      "learning_rate": 2.9649890590809634e-06,
      "loss": 24.7561,
      "step": 475,
      "true_loss": 3.0628
    },
    {
      "epoch": 3.7810650887573964,
      "grad_norm": 258.65135809575645,
      "learning_rate": 2.9376367614879654e-06,
      "loss": 27.2582,
      "step": 480,
      "true_loss": 3.2503
    },
    {
      "epoch": 3.7810650887573964,
      "eval_accuracy": 0.19290465631929046,
      "eval_loss": 3.4006612300872803,
      "eval_runtime": 15.0539,
      "eval_samples_per_second": 59.918,
      "eval_steps_per_second": 7.506,
      "step": 480
    },
    {
      "epoch": 3.8205128205128203,
      "grad_norm": 276.16011426554854,
      "learning_rate": 2.9102844638949674e-06,
      "loss": 25.0956,
      "step": 485,
      "true_loss": 2.8415
    },
    {
      "epoch": 3.8599605522682445,
      "grad_norm": 265.77475990424085,
      "learning_rate": 2.8829321663019693e-06,
      "loss": 24.7172,
      "step": 490,
      "true_loss": 3.2856
    },
    {
      "epoch": 3.8599605522682445,
      "eval_accuracy": 0.19401330376940132,
      "eval_loss": 3.405134916305542,
      "eval_runtime": 14.9878,
      "eval_samples_per_second": 60.182,
      "eval_steps_per_second": 7.539,
      "step": 490
    },
    {
      "epoch": 3.899408284023669,
      "grad_norm": 273.4345528471514,
      "learning_rate": 2.855579868708972e-06,
      "loss": 25.455,
      "step": 495,
      "true_loss": 3.431
    },
    {
      "epoch": 3.9388560157790926,
      "grad_norm": 273.95487836913196,
      "learning_rate": 2.828227571115974e-06,
      "loss": 25.3041,
      "step": 500,
      "true_loss": 3.1038
    },
    {
      "epoch": 3.9388560157790926,
      "eval_accuracy": 0.2017738359201774,
      "eval_loss": 3.3893778324127197,
      "eval_runtime": 15.0204,
      "eval_samples_per_second": 60.052,
      "eval_steps_per_second": 7.523,
      "step": 500
    },
    {
      "epoch": 3.978303747534517,
      "grad_norm": 265.6659951241656,
      "learning_rate": 2.800875273522976e-06,
      "loss": 24.9082,
      "step": 505,
      "true_loss": 3.1013
    },
    {
      "epoch": 4.01577909270217,
      "grad_norm": 258.2230652518369,
      "learning_rate": 2.773522975929978e-06,
      "loss": 23.6265,
      "step": 510,
      "true_loss": 2.474
    },
    {
      "epoch": 4.01577909270217,
      "eval_accuracy": 0.2006651884700665,
      "eval_loss": 3.3711023330688477,
      "eval_runtime": 15.1319,
      "eval_samples_per_second": 59.609,
      "eval_steps_per_second": 7.468,
      "step": 510
    },
    {
      "epoch": 4.055226824457594,
      "grad_norm": 265.77115170095055,
      "learning_rate": 2.746170678336981e-06,
      "loss": 23.0143,
      "step": 515,
      "true_loss": 2.4705
    },
    {
      "epoch": 4.094674556213017,
      "grad_norm": 305.1591165434374,
      "learning_rate": 2.718818380743983e-06,
      "loss": 23.0705,
      "step": 520,
      "true_loss": 2.9254
    },
    {
      "epoch": 4.094674556213017,
      "eval_accuracy": 0.2017738359201774,
      "eval_loss": 3.3792917728424072,
      "eval_runtime": 15.0429,
      "eval_samples_per_second": 59.962,
      "eval_steps_per_second": 7.512,
      "step": 520
    },
    {
      "epoch": 4.134122287968442,
      "grad_norm": 301.013844993568,
      "learning_rate": 2.6914660831509848e-06,
      "loss": 23.3767,
      "step": 525,
      "true_loss": 2.2937
    },
    {
      "epoch": 4.173570019723866,
      "grad_norm": 289.0773672920475,
      "learning_rate": 2.6641137855579867e-06,
      "loss": 24.2713,
      "step": 530,
      "true_loss": 3.0639
    },
    {
      "epoch": 4.173570019723866,
      "eval_accuracy": 0.21175166297117518,
      "eval_loss": 3.387169122695923,
      "eval_runtime": 14.9666,
      "eval_samples_per_second": 60.267,
      "eval_steps_per_second": 7.55,
      "step": 530
    },
    {
      "epoch": 4.21301775147929,
      "grad_norm": 292.88346898040646,
      "learning_rate": 2.6367614879649895e-06,
      "loss": 23.0921,
      "step": 535,
      "true_loss": 2.8408
    },
    {
      "epoch": 4.252465483234714,
      "grad_norm": 308.73457670985425,
      "learning_rate": 2.6094091903719915e-06,
      "loss": 22.7046,
      "step": 540,
      "true_loss": 2.7119
    },
    {
      "epoch": 4.252465483234714,
      "eval_accuracy": 0.20288248337028825,
      "eval_loss": 3.378486156463623,
      "eval_runtime": 15.0097,
      "eval_samples_per_second": 60.094,
      "eval_steps_per_second": 7.528,
      "step": 540
    },
    {
      "epoch": 4.291913214990138,
      "grad_norm": 304.39435113456375,
      "learning_rate": 2.5820568927789935e-06,
      "loss": 23.06,
      "step": 545,
      "true_loss": 2.8129
    },
    {
      "epoch": 4.331360946745562,
      "grad_norm": 297.18920459782885,
      "learning_rate": 2.5547045951859954e-06,
      "loss": 22.8824,
      "step": 550,
      "true_loss": 2.8139
    },
    {
      "epoch": 4.331360946745562,
      "eval_accuracy": 0.21286031042128603,
      "eval_loss": 3.3715484142303467,
      "eval_runtime": 15.0858,
      "eval_samples_per_second": 59.791,
      "eval_steps_per_second": 7.491,
      "step": 550
    },
    {
      "epoch": 4.370808678500986,
      "grad_norm": 296.890352701281,
      "learning_rate": 2.5273522975929982e-06,
      "loss": 23.1762,
      "step": 555,
      "true_loss": 2.6411
    },
    {
      "epoch": 4.410256410256411,
      "grad_norm": 301.7124359937404,
      "learning_rate": 2.5e-06,
      "loss": 23.1173,
      "step": 560,
      "true_loss": 2.8016
    },
    {
      "epoch": 4.410256410256411,
      "eval_accuracy": 0.2006651884700665,
      "eval_loss": 3.3617937564849854,
      "eval_runtime": 15.4105,
      "eval_samples_per_second": 58.531,
      "eval_steps_per_second": 7.333,
      "step": 560
    },
    {
      "epoch": 4.449704142011834,
      "grad_norm": 301.5417177978464,
      "learning_rate": 2.4726477024070026e-06,
      "loss": 23.2135,
      "step": 565,
      "true_loss": 2.6408
    },
    {
      "epoch": 4.489151873767258,
      "grad_norm": 307.2723305395378,
      "learning_rate": 2.4452954048140046e-06,
      "loss": 22.2622,
      "step": 570,
      "true_loss": 2.9552
    },
    {
      "epoch": 4.489151873767258,
      "eval_accuracy": 0.1951219512195122,
      "eval_loss": 3.3558921813964844,
      "eval_runtime": 14.9957,
      "eval_samples_per_second": 60.15,
      "eval_steps_per_second": 7.535,
      "step": 570
    },
    {
      "epoch": 4.5285996055226825,
      "grad_norm": 316.2854049798904,
      "learning_rate": 2.417943107221007e-06,
      "loss": 23.6898,
      "step": 575,
      "true_loss": 3.0164
    },
    {
      "epoch": 4.568047337278107,
      "grad_norm": 285.3077007071203,
      "learning_rate": 2.390590809628009e-06,
      "loss": 21.7338,
      "step": 580,
      "true_loss": 2.3795
    },
    {
      "epoch": 4.568047337278107,
      "eval_accuracy": 0.21507760532150777,
      "eval_loss": 3.3453264236450195,
      "eval_runtime": 14.9619,
      "eval_samples_per_second": 60.286,
      "eval_steps_per_second": 7.553,
      "step": 580
    },
    {
      "epoch": 4.607495069033531,
      "grad_norm": 317.59610581151685,
      "learning_rate": 2.3632385120350113e-06,
      "loss": 23.3327,
      "step": 585,
      "true_loss": 3.1207
    },
    {
      "epoch": 4.646942800788954,
      "grad_norm": 321.934617091813,
      "learning_rate": 2.3358862144420133e-06,
      "loss": 23.3021,
      "step": 590,
      "true_loss": 2.9562
    },
    {
      "epoch": 4.646942800788954,
      "eval_accuracy": 0.21840354767184036,
      "eval_loss": 3.3475186824798584,
      "eval_runtime": 14.8088,
      "eval_samples_per_second": 60.91,
      "eval_steps_per_second": 7.631,
      "step": 590
    },
    {
      "epoch": 4.686390532544379,
      "grad_norm": 318.62665027252604,
      "learning_rate": 2.3085339168490156e-06,
      "loss": 23.6124,
      "step": 595,
      "true_loss": 3.19
    },
    {
      "epoch": 4.725838264299803,
      "grad_norm": 323.13589935545605,
      "learning_rate": 2.2811816192560176e-06,
      "loss": 23.3896,
      "step": 600,
      "true_loss": 2.9138
    },
    {
      "epoch": 4.725838264299803,
      "eval_accuracy": 0.2073170731707317,
      "eval_loss": 3.3516340255737305,
      "eval_runtime": 15.0949,
      "eval_samples_per_second": 59.755,
      "eval_steps_per_second": 7.486,
      "step": 600
    },
    {
      "epoch": 4.765285996055227,
      "grad_norm": 306.75577469369307,
      "learning_rate": 2.25382932166302e-06,
      "loss": 23.1344,
      "step": 605,
      "true_loss": 2.9071
    },
    {
      "epoch": 4.804733727810651,
      "grad_norm": 305.64568552389716,
      "learning_rate": 2.226477024070022e-06,
      "loss": 22.918,
      "step": 610,
      "true_loss": 3.205
    },
    {
      "epoch": 4.804733727810651,
      "eval_accuracy": 0.21286031042128603,
      "eval_loss": 3.344453811645508,
      "eval_runtime": 14.8514,
      "eval_samples_per_second": 60.735,
      "eval_steps_per_second": 7.609,
      "step": 610
    },
    {
      "epoch": 4.844181459566075,
      "grad_norm": 345.3630194507602,
      "learning_rate": 2.1991247264770243e-06,
      "loss": 22.4193,
      "step": 615,
      "true_loss": 3.0203
    },
    {
      "epoch": 4.883629191321499,
      "grad_norm": 310.8043549889887,
      "learning_rate": 2.1717724288840263e-06,
      "loss": 22.9716,
      "step": 620,
      "true_loss": 2.7156
    },
    {
      "epoch": 4.883629191321499,
      "eval_accuracy": 0.2261640798226164,
      "eval_loss": 3.3267650604248047,
      "eval_runtime": 15.0181,
      "eval_samples_per_second": 60.061,
      "eval_steps_per_second": 7.524,
      "step": 620
    },
    {
      "epoch": 4.923076923076923,
      "grad_norm": 328.17455994439047,
      "learning_rate": 2.1444201312910287e-06,
      "loss": 23.5778,
      "step": 625,
      "true_loss": 3.0593
    },
    {
      "epoch": 4.962524654832347,
      "grad_norm": 314.85030625337566,
      "learning_rate": 2.1170678336980307e-06,
      "loss": 23.4656,
      "step": 630,
      "true_loss": 2.7659
    },
    {
      "epoch": 4.962524654832347,
      "eval_accuracy": 0.22283813747228381,
      "eval_loss": 3.313983201980591,
      "eval_runtime": 15.004,
      "eval_samples_per_second": 60.117,
      "eval_steps_per_second": 7.531,
      "step": 630
    },
    {
      "epoch": 5.0,
      "grad_norm": 264.024205323258,
      "learning_rate": 2.089715536105033e-06,
      "loss": 22.7471,
      "step": 635,
      "true_loss": 2.9564
    },
    {
      "epoch": 5.039447731755424,
      "grad_norm": 306.42431977803875,
      "learning_rate": 2.062363238512035e-06,
      "loss": 21.5082,
      "step": 640,
      "true_loss": 2.7247
    },
    {
      "epoch": 5.039447731755424,
      "eval_accuracy": 0.22838137472283815,
      "eval_loss": 3.3180108070373535,
      "eval_runtime": 14.9697,
      "eval_samples_per_second": 60.255,
      "eval_steps_per_second": 7.549,
      "step": 640
    },
    {
      "epoch": 5.0788954635108485,
      "grad_norm": 340.14969305624527,
      "learning_rate": 2.0350109409190374e-06,
      "loss": 21.563,
      "step": 645,
      "true_loss": 2.5149
    },
    {
      "epoch": 5.118343195266272,
      "grad_norm": 323.01121928019046,
      "learning_rate": 2.0076586433260394e-06,
      "loss": 20.9225,
      "step": 650,
      "true_loss": 2.632
    },
    {
      "epoch": 5.118343195266272,
      "eval_accuracy": 0.21618625277161863,
      "eval_loss": 3.321817398071289,
      "eval_runtime": 15.0606,
      "eval_samples_per_second": 59.891,
      "eval_steps_per_second": 7.503,
      "step": 650
    },
    {
      "epoch": 5.157790927021696,
      "grad_norm": 348.8138229734559,
      "learning_rate": 1.9803063457330417e-06,
      "loss": 20.005,
      "step": 655,
      "true_loss": 2.4218
    },
    {
      "epoch": 5.19723865877712,
      "grad_norm": 336.229278513962,
      "learning_rate": 1.9529540481400437e-06,
      "loss": 19.3711,
      "step": 660,
      "true_loss": 2.5052
    },
    {
      "epoch": 5.19723865877712,
      "eval_accuracy": 0.20953436807095344,
      "eval_loss": 3.3247318267822266,
      "eval_runtime": 14.9674,
      "eval_samples_per_second": 60.264,
      "eval_steps_per_second": 7.55,
      "step": 660
    },
    {
      "epoch": 5.236686390532545,
      "grad_norm": 359.810565898114,
      "learning_rate": 1.925601750547046e-06,
      "loss": 20.7458,
      "step": 665,
      "true_loss": 2.2577
    },
    {
      "epoch": 5.276134122287968,
      "grad_norm": 326.2113183093954,
      "learning_rate": 1.8982494529540483e-06,
      "loss": 21.3133,
      "step": 670,
      "true_loss": 2.5225
    },
    {
      "epoch": 5.276134122287968,
      "eval_accuracy": 0.21729490022172948,
      "eval_loss": 3.3202853202819824,
      "eval_runtime": 14.8455,
      "eval_samples_per_second": 60.759,
      "eval_steps_per_second": 7.612,
      "step": 670
    },
    {
      "epoch": 5.315581854043392,
      "grad_norm": 347.21516127995676,
      "learning_rate": 1.8708971553610504e-06,
      "loss": 20.257,
      "step": 675,
      "true_loss": 2.7666
    },
    {
      "epoch": 5.355029585798817,
      "grad_norm": 344.59999127493575,
      "learning_rate": 1.8435448577680526e-06,
      "loss": 20.2119,
      "step": 680,
      "true_loss": 2.548
    },
    {
      "epoch": 5.355029585798817,
      "eval_accuracy": 0.21618625277161863,
      "eval_loss": 3.327068328857422,
      "eval_runtime": 15.0698,
      "eval_samples_per_second": 59.855,
      "eval_steps_per_second": 7.498,
      "step": 680
    },
    {
      "epoch": 5.394477317554241,
      "grad_norm": 356.6528216246677,
      "learning_rate": 1.816192560175055e-06,
      "loss": 20.4713,
      "step": 685,
      "true_loss": 2.7505
    },
    {
      "epoch": 5.433925049309664,
      "grad_norm": 373.80585141045236,
      "learning_rate": 1.788840262582057e-06,
      "loss": 21.3783,
      "step": 690,
      "true_loss": 2.5356
    },
    {
      "epoch": 5.433925049309664,
      "eval_accuracy": 0.2073170731707317,
      "eval_loss": 3.330681085586548,
      "eval_runtime": 15.1298,
      "eval_samples_per_second": 59.617,
      "eval_steps_per_second": 7.469,
      "step": 690
    },
    {
      "epoch": 5.4733727810650885,
      "grad_norm": 385.2402312951401,
      "learning_rate": 1.7614879649890594e-06,
      "loss": 20.441,
      "step": 695,
      "true_loss": 2.6452
    },
    {
      "epoch": 5.512820512820513,
      "grad_norm": 369.2755541375683,
      "learning_rate": 1.7341356673960613e-06,
      "loss": 22.1389,
      "step": 700,
      "true_loss": 2.5363
    },
    {
      "epoch": 5.512820512820513,
      "eval_accuracy": 0.22172949002217296,
      "eval_loss": 3.3146209716796875,
      "eval_runtime": 15.1379,
      "eval_samples_per_second": 59.586,
      "eval_steps_per_second": 7.465,
      "step": 700
    },
    {
      "epoch": 5.552268244575937,
      "grad_norm": 360.7604924002524,
      "learning_rate": 1.7067833698030637e-06,
      "loss": 19.7835,
      "step": 705,
      "true_loss": 2.455
    },
    {
      "epoch": 5.591715976331361,
      "grad_norm": 345.01754212392365,
      "learning_rate": 1.6794310722100657e-06,
      "loss": 22.7283,
      "step": 710,
      "true_loss": 3.471
    },
    {
      "epoch": 5.591715976331361,
      "eval_accuracy": 0.22838137472283815,
      "eval_loss": 3.3121492862701416,
      "eval_runtime": 15.0858,
      "eval_samples_per_second": 59.791,
      "eval_steps_per_second": 7.49,
      "step": 710
    },
    {
      "epoch": 5.631163708086785,
      "grad_norm": 381.2083321719042,
      "learning_rate": 1.652078774617068e-06,
      "loss": 22.344,
      "step": 715,
      "true_loss": 2.7354
    },
    {
      "epoch": 5.670611439842209,
      "grad_norm": 364.2642170009335,
      "learning_rate": 1.62472647702407e-06,
      "loss": 20.889,
      "step": 720,
      "true_loss": 2.6664
    },
    {
      "epoch": 5.670611439842209,
      "eval_accuracy": 0.23059866962305986,
      "eval_loss": 3.314051389694214,
      "eval_runtime": 14.9034,
      "eval_samples_per_second": 60.523,
      "eval_steps_per_second": 7.582,
      "step": 720
    },
    {
      "epoch": 5.710059171597633,
      "grad_norm": 395.4957897175653,
      "learning_rate": 1.5973741794310724e-06,
      "loss": 20.6087,
      "step": 725,
      "true_loss": 2.8175
    },
    {
      "epoch": 5.7495069033530575,
      "grad_norm": 425.38224141217324,
      "learning_rate": 1.5700218818380744e-06,
      "loss": 20.4895,
      "step": 730,
      "true_loss": 2.452
    },
    {
      "epoch": 5.7495069033530575,
      "eval_accuracy": 0.23503325942350334,
      "eval_loss": 3.308513641357422,
      "eval_runtime": 15.0023,
      "eval_samples_per_second": 60.124,
      "eval_steps_per_second": 7.532,
      "step": 730
    },
    {
      "epoch": 5.788954635108482,
      "grad_norm": 370.3276392958097,
      "learning_rate": 1.5426695842450768e-06,
      "loss": 20.9818,
      "step": 735,
      "true_loss": 2.7038
    },
    {
      "epoch": 5.828402366863905,
      "grad_norm": 367.66652541868655,
      "learning_rate": 1.5153172866520787e-06,
      "loss": 21.7155,
      "step": 740,
      "true_loss": 2.613
    },
    {
      "epoch": 5.828402366863905,
      "eval_accuracy": 0.2394678492239468,
      "eval_loss": 3.302647829055786,
      "eval_runtime": 14.9417,
      "eval_samples_per_second": 60.368,
      "eval_steps_per_second": 7.563,
      "step": 740
    }
  ],
  "logging_steps": 5,
  "max_steps": 1016,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "save_steps": 10,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}