diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4621 @@
+{
+  "best_metric": 1.9382596015930176,
+  "best_model_checkpoint": "./codegen-350M-mono-QLoRa-flytech/checkpoint-2000",
+  "epoch": 0.4605907075824745,
+  "eval_steps": 5,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "eval_loss": 2.743941068649292,
+      "eval_runtime": 749.578,
+      "eval_samples_per_second": 19.862,
+      "eval_steps_per_second": 2.483,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.3329066038131714,
+      "learning_rate": 0.000199,
+      "loss": 2.9969,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 2.6796510219573975,
+      "eval_runtime": 751.2092,
+      "eval_samples_per_second": 19.819,
+      "eval_steps_per_second": 2.477,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 2.5984177589416504,
+      "eval_runtime": 764.9736,
+      "eval_samples_per_second": 19.462,
+      "eval_steps_per_second": 2.433,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.283966600894928,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 2.5613,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 2.510923147201538,
+      "eval_runtime": 775.4648,
+      "eval_samples_per_second": 19.199,
+      "eval_steps_per_second": 2.4,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.4384284019470215,
+      "eval_runtime": 807.2199,
+      "eval_samples_per_second": 18.444,
+      "eval_steps_per_second": 2.305,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.3212601840496063,
+      "learning_rate": 0.00019700000000000002,
+      "loss": 2.4609,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.398897886276245,
+      "eval_runtime": 806.9842,
+      "eval_samples_per_second": 18.449,
+      "eval_steps_per_second": 2.306,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.3749544620513916,
+      "eval_runtime": 807.3296,
+      "eval_samples_per_second": 18.441,
+      "eval_steps_per_second": 2.305,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5253788232803345,
+      "learning_rate": 0.000196,
+      "loss": 2.4028,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.3531863689422607,
+      "eval_runtime": 807.2812,
+      "eval_samples_per_second": 18.442,
+      "eval_steps_per_second": 2.305,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.3311145305633545,
+      "eval_runtime": 807.5901,
+      "eval_samples_per_second": 18.435,
+      "eval_steps_per_second": 2.304,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4847645163536072,
+      "learning_rate": 0.000195,
+      "loss": 2.4339,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.31105637550354,
+      "eval_runtime": 807.447,
+      "eval_samples_per_second": 18.438,
+      "eval_steps_per_second": 2.305,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.294403314590454,
+      "eval_runtime": 804.5301,
+      "eval_samples_per_second": 18.505,
+      "eval_steps_per_second": 2.313,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5184943675994873,
+      "learning_rate": 0.000194,
+      "loss": 2.3291,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.28222393989563,
+      "eval_runtime": 744.3103,
+      "eval_samples_per_second": 20.002,
+      "eval_steps_per_second": 2.5,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 2.271087169647217,
+      "eval_runtime": 744.2352,
+      "eval_samples_per_second": 20.004,
+      "eval_steps_per_second": 2.501,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.42745935916900635,
+      "learning_rate": 0.000193,
+      "loss": 2.3128,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.2600862979888916,
+      "eval_runtime": 747.8797,
+      "eval_samples_per_second": 19.907,
+      "eval_steps_per_second": 2.488,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.251253128051758,
+      "eval_runtime": 755.1321,
+      "eval_samples_per_second": 19.716,
+      "eval_steps_per_second": 2.464,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5841810703277588,
+      "learning_rate": 0.000192,
+      "loss": 2.6201,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.2421493530273438,
+      "eval_runtime": 750.5937,
+      "eval_samples_per_second": 19.835,
+      "eval_steps_per_second": 2.479,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.2358789443969727,
+      "eval_runtime": 750.3335,
+      "eval_samples_per_second": 19.842,
+      "eval_steps_per_second": 2.48,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.34617361426353455,
+      "learning_rate": 0.000191,
+      "loss": 2.2322,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.2303481101989746,
+      "eval_runtime": 748.12,
+      "eval_samples_per_second": 19.901,
+      "eval_steps_per_second": 2.488,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.2245254516601562,
+      "eval_runtime": 763.3784,
+      "eval_samples_per_second": 19.503,
+      "eval_steps_per_second": 2.438,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5058037638664246,
+      "learning_rate": 0.00019,
+      "loss": 2.3576,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.217428207397461,
+      "eval_runtime": 800.3219,
+      "eval_samples_per_second": 18.603,
+      "eval_steps_per_second": 2.325,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.2102160453796387,
+      "eval_runtime": 806.6841,
+      "eval_samples_per_second": 18.456,
+      "eval_steps_per_second": 2.307,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5872398018836975,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 2.3798,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.203975200653076,
+      "eval_runtime": 778.7436,
+      "eval_samples_per_second": 19.118,
+      "eval_steps_per_second": 2.39,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.198529005050659,
+      "eval_runtime": 754.0178,
+      "eval_samples_per_second": 19.745,
+      "eval_steps_per_second": 2.468,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4760008752346039,
+      "learning_rate": 0.000188,
+      "loss": 2.2499,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.1929078102111816,
+      "eval_runtime": 770.8151,
+      "eval_samples_per_second": 19.315,
+      "eval_steps_per_second": 2.414,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.1861350536346436,
+      "eval_runtime": 749.444,
+      "eval_samples_per_second": 19.865,
+      "eval_steps_per_second": 2.483,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6719084978103638,
+      "learning_rate": 0.00018700000000000002,
+      "loss": 2.2002,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.1807847023010254,
+      "eval_runtime": 790.3227,
+      "eval_samples_per_second": 18.838,
+      "eval_steps_per_second": 2.355,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.177309274673462,
+      "eval_runtime": 767.7802,
+      "eval_samples_per_second": 19.391,
+      "eval_steps_per_second": 2.424,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6400887966156006,
+      "learning_rate": 0.00018600000000000002,
+      "loss": 2.3217,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.1720025539398193,
+      "eval_runtime": 794.6729,
+      "eval_samples_per_second": 18.735,
+      "eval_steps_per_second": 2.342,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.1658318042755127,
+      "eval_runtime": 774.1526,
+      "eval_samples_per_second": 19.231,
+      "eval_steps_per_second": 2.404,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48335766792297363,
+      "learning_rate": 0.00018500000000000002,
+      "loss": 2.1297,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 2.1612164974212646,
+      "eval_runtime": 778.6453,
+      "eval_samples_per_second": 19.12,
+      "eval_steps_per_second": 2.39,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.1569318771362305,
+      "eval_runtime": 761.275,
+      "eval_samples_per_second": 19.557,
+      "eval_steps_per_second": 2.445,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8584076166152954,
+      "learning_rate": 0.00018400000000000003,
+      "loss": 2.1718,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.1521499156951904,
+      "eval_runtime": 746.5904,
+      "eval_samples_per_second": 19.941,
+      "eval_steps_per_second": 2.493,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.1496639251708984,
+      "eval_runtime": 751.6198,
+      "eval_samples_per_second": 19.808,
+      "eval_steps_per_second": 2.476,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7087550163269043,
+      "learning_rate": 0.000183,
+      "loss": 2.3469,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.146427631378174,
+      "eval_runtime": 768.8707,
+      "eval_samples_per_second": 19.363,
+      "eval_steps_per_second": 2.42,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.1417484283447266,
+      "eval_runtime": 775.3535,
+      "eval_samples_per_second": 19.202,
+      "eval_steps_per_second": 2.4,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6968621611595154,
+      "learning_rate": 0.000182,
+      "loss": 2.2498,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.138108015060425,
+      "eval_runtime": 773.4912,
+      "eval_samples_per_second": 19.248,
+      "eval_steps_per_second": 2.406,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.1360790729522705,
+      "eval_runtime": 773.9293,
+      "eval_samples_per_second": 19.237,
+      "eval_steps_per_second": 2.405,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7121121287345886,
+      "learning_rate": 0.000181,
+      "loss": 2.3611,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.1335365772247314,
+      "eval_runtime": 763.0423,
+      "eval_samples_per_second": 19.511,
+      "eval_steps_per_second": 2.439,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 2.1294620037078857,
+      "eval_runtime": 755.8471,
+      "eval_samples_per_second": 19.697,
+      "eval_steps_per_second": 2.462,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6406265497207642,
+      "learning_rate": 0.00018,
+      "loss": 2.1718,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.125330924987793,
+      "eval_runtime": 764.112,
+      "eval_samples_per_second": 19.484,
+      "eval_steps_per_second": 2.436,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.121659278869629,
+      "eval_runtime": 784.0069,
+      "eval_samples_per_second": 18.99,
+      "eval_steps_per_second": 2.374,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.94074547290802,
+      "learning_rate": 0.00017900000000000001,
+      "loss": 2.2531,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.1184146404266357,
+      "eval_runtime": 795.2056,
+      "eval_samples_per_second": 18.722,
+      "eval_steps_per_second": 2.34,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.11686635017395,
+      "eval_runtime": 782.4427,
+      "eval_samples_per_second": 19.028,
+      "eval_steps_per_second": 2.378,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4468848407268524,
+      "learning_rate": 0.00017800000000000002,
+      "loss": 2.276,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.1175310611724854,
+      "eval_runtime": 755.2557,
+      "eval_samples_per_second": 19.713,
+      "eval_steps_per_second": 2.464,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.1144332885742188,
+      "eval_runtime": 777.8858,
+      "eval_samples_per_second": 19.139,
+      "eval_steps_per_second": 2.392,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.697229266166687,
+      "learning_rate": 0.00017700000000000002,
+      "loss": 2.2138,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.111482620239258,
+      "eval_runtime": 751.8011,
+      "eval_samples_per_second": 19.803,
+      "eval_steps_per_second": 2.475,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 2.1105000972747803,
+      "eval_runtime": 751.7683,
+      "eval_samples_per_second": 19.804,
+      "eval_steps_per_second": 2.475,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9181985855102539,
+      "learning_rate": 0.00017600000000000002,
+      "loss": 2.3692,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.109123468399048,
+      "eval_runtime": 749.086,
+      "eval_samples_per_second": 19.875,
+      "eval_steps_per_second": 2.484,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.10528564453125,
+      "eval_runtime": 750.8258,
+      "eval_samples_per_second": 19.829,
+      "eval_steps_per_second": 2.479,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48959431052207947,
+      "learning_rate": 0.000175,
+      "loss": 2.2648,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.103628396987915,
+      "eval_runtime": 785.1888,
+      "eval_samples_per_second": 18.961,
+      "eval_steps_per_second": 2.37,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.1015634536743164,
+      "eval_runtime": 771.9757,
+      "eval_samples_per_second": 19.286,
+      "eval_steps_per_second": 2.411,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6428173780441284,
+      "learning_rate": 0.000174,
+      "loss": 2.1139,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.097482919692993,
+      "eval_runtime": 763.409,
+      "eval_samples_per_second": 19.502,
+      "eval_steps_per_second": 2.438,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.095787286758423,
+      "eval_runtime": 760.3894,
+      "eval_samples_per_second": 19.579,
+      "eval_steps_per_second": 2.447,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5767454504966736,
+      "learning_rate": 0.000173,
+      "loss": 2.1961,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.0945658683776855,
+      "eval_runtime": 758.8729,
+      "eval_samples_per_second": 19.619,
+      "eval_steps_per_second": 2.452,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.0915117263793945,
+      "eval_runtime": 748.4004,
+      "eval_samples_per_second": 19.893,
+      "eval_steps_per_second": 2.487,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6312183737754822,
+      "learning_rate": 0.000172,
+      "loss": 2.2072,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 2.0887210369110107,
+      "eval_runtime": 762.6911,
+      "eval_samples_per_second": 19.52,
+      "eval_steps_per_second": 2.44,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.086923360824585,
+      "eval_runtime": 806.6105,
+      "eval_samples_per_second": 18.457,
+      "eval_steps_per_second": 2.307,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6048781275749207,
+      "learning_rate": 0.000171,
+      "loss": 2.0356,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.085240125656128,
+      "eval_runtime": 806.3511,
+      "eval_samples_per_second": 18.463,
+      "eval_steps_per_second": 2.308,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.084153175354004,
+      "eval_runtime": 806.3163,
+      "eval_samples_per_second": 18.464,
+      "eval_steps_per_second": 2.308,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7954710125923157,
+      "learning_rate": 0.00017,
+      "loss": 2.0189,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.082386016845703,
+      "eval_runtime": 806.423,
+      "eval_samples_per_second": 18.462,
+      "eval_steps_per_second": 2.308,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.0789432525634766,
+      "eval_runtime": 806.4464,
+      "eval_samples_per_second": 18.461,
+      "eval_steps_per_second": 2.308,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7538740038871765,
+      "learning_rate": 0.00016900000000000002,
+      "loss": 1.8978,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.076179027557373,
+      "eval_runtime": 806.5403,
+      "eval_samples_per_second": 18.459,
+      "eval_steps_per_second": 2.307,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.075080156326294,
+      "eval_runtime": 784.5471,
+      "eval_samples_per_second": 18.977,
+      "eval_steps_per_second": 2.372,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2843266427516937,
+      "learning_rate": 0.000168,
+      "loss": 2.1163,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.074650764465332,
+      "eval_runtime": 752.9254,
+      "eval_samples_per_second": 19.774,
+      "eval_steps_per_second": 2.472,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 2.0736653804779053,
+      "eval_runtime": 757.9611,
+      "eval_samples_per_second": 19.642,
+      "eval_steps_per_second": 2.455,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7260650396347046,
+      "learning_rate": 0.000167,
+      "loss": 2.2219,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.073575496673584,
+      "eval_runtime": 806.2254,
+      "eval_samples_per_second": 18.466,
+      "eval_steps_per_second": 2.308,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.0729920864105225,
+      "eval_runtime": 779.9527,
+      "eval_samples_per_second": 19.088,
+      "eval_steps_per_second": 2.386,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7835765480995178,
+      "learning_rate": 0.000166,
+      "loss": 2.2111,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.070727825164795,
+      "eval_runtime": 758.1738,
+      "eval_samples_per_second": 19.637,
+      "eval_steps_per_second": 2.455,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.069014549255371,
+      "eval_runtime": 747.7568,
+      "eval_samples_per_second": 19.91,
+      "eval_steps_per_second": 2.489,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5928293466567993,
+      "learning_rate": 0.000165,
+      "loss": 2.1936,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.0676755905151367,
+      "eval_runtime": 798.6576,
+      "eval_samples_per_second": 18.641,
+      "eval_steps_per_second": 2.33,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.0652225017547607,
+      "eval_runtime": 806.0571,
+      "eval_samples_per_second": 18.47,
+      "eval_steps_per_second": 2.309,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7964838147163391,
+      "learning_rate": 0.000164,
+      "loss": 2.1177,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.0634725093841553,
+      "eval_runtime": 805.9924,
+      "eval_samples_per_second": 18.472,
+      "eval_steps_per_second": 2.309,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 2.063730478286743,
+      "eval_runtime": 805.9942,
+      "eval_samples_per_second": 18.472,
+      "eval_steps_per_second": 2.309,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5622439980506897,
+      "learning_rate": 0.000163,
+      "loss": 2.1331,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.06245756149292,
+      "eval_runtime": 806.0779,
+      "eval_samples_per_second": 18.47,
+      "eval_steps_per_second": 2.309,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.0626060962677,
+      "eval_runtime": 806.2506,
+      "eval_samples_per_second": 18.466,
+      "eval_steps_per_second": 2.308,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7700399160385132,
+      "learning_rate": 0.000162,
+      "loss": 2.3266,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.0606532096862793,
+      "eval_runtime": 806.0725,
+      "eval_samples_per_second": 18.47,
+      "eval_steps_per_second": 2.309,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.060309410095215,
+      "eval_runtime": 806.3975,
+      "eval_samples_per_second": 18.462,
+      "eval_steps_per_second": 2.308,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7712746858596802,
+      "learning_rate": 0.000161,
+      "loss": 2.0627,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.0599002838134766,
+      "eval_runtime": 806.2192,
+      "eval_samples_per_second": 18.466,
+      "eval_steps_per_second": 2.308,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.0586650371551514,
+      "eval_runtime": 806.299,
+      "eval_samples_per_second": 18.465,
+      "eval_steps_per_second": 2.308,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5348338484764099,
+      "learning_rate": 0.00016,
+      "loss": 2.3373,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.056227445602417,
+      "eval_runtime": 806.1063,
+      "eval_samples_per_second": 18.469,
+      "eval_steps_per_second": 2.309,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.054941177368164,
+      "eval_runtime": 806.3058,
+      "eval_samples_per_second": 18.464,
+      "eval_steps_per_second": 2.308,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.577366828918457,
+      "learning_rate": 0.00015900000000000002,
+      "loss": 2.0422,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 2.0555078983306885,
+      "eval_runtime": 806.2774,
+      "eval_samples_per_second": 18.465,
+      "eval_steps_per_second": 2.308,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.054901361465454,
+      "eval_runtime": 806.2591,
+      "eval_samples_per_second": 18.466,
+      "eval_steps_per_second": 2.308,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7189317345619202,
+      "learning_rate": 0.00015800000000000002,
+      "loss": 2.1189,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.0532333850860596,
+      "eval_runtime": 806.381,
+      "eval_samples_per_second": 18.463,
+      "eval_steps_per_second": 2.308,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.0509543418884277,
+      "eval_runtime": 806.2323,
+      "eval_samples_per_second": 18.466,
+      "eval_steps_per_second": 2.308,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.548724353313446,
+      "learning_rate": 0.00015700000000000002,
+      "loss": 2.0847,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.0487470626831055,
+      "eval_runtime": 806.372,
+      "eval_samples_per_second": 18.463,
+      "eval_steps_per_second": 2.308,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.0478439331054688,
+      "eval_runtime": 806.31,
+      "eval_samples_per_second": 18.464,
+      "eval_steps_per_second": 2.308,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5101466774940491,
+      "learning_rate": 0.00015600000000000002,
+      "loss": 2.0727,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.0454509258270264,
+      "eval_runtime": 806.399,
+      "eval_samples_per_second": 18.462,
+      "eval_steps_per_second": 2.308,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.043710708618164,
+      "eval_runtime": 806.2814,
+      "eval_samples_per_second": 18.465,
+      "eval_steps_per_second": 2.308,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7228645086288452,
+      "learning_rate": 0.000155,
+      "loss": 2.0467,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.042335033416748,
+      "eval_runtime": 806.3613,
+      "eval_samples_per_second": 18.463,
+      "eval_steps_per_second": 2.308,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.0407748222351074,
+      "eval_runtime": 806.5248,
+      "eval_samples_per_second": 18.459,
+      "eval_steps_per_second": 2.307,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7004246115684509,
+      "learning_rate": 0.000154,
+      "loss": 2.2736,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.0401110649108887,
+      "eval_runtime": 807.0877,
+      "eval_samples_per_second": 18.447,
+      "eval_steps_per_second": 2.306,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.039562702178955,
+      "eval_runtime": 808.9767,
+      "eval_samples_per_second": 18.403,
+      "eval_steps_per_second": 2.3,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5844886898994446,
+      "learning_rate": 0.000153,
+      "loss": 2.0475,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.039823055267334,
+      "eval_runtime": 768.1785,
+      "eval_samples_per_second": 19.381,
+      "eval_steps_per_second": 2.423,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.0399796962738037,
+      "eval_runtime": 745.2193,
+      "eval_samples_per_second": 19.978,
+      "eval_steps_per_second": 2.497,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9306110143661499,
+      "learning_rate": 0.000152,
+      "loss": 2.1269,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.0386431217193604,
+      "eval_runtime": 744.6033,
+      "eval_samples_per_second": 19.995,
+      "eval_steps_per_second": 2.499,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.038109302520752,
+      "eval_runtime": 744.4904,
+      "eval_samples_per_second": 19.998,
+      "eval_steps_per_second": 2.5,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7265554666519165,
+      "learning_rate": 0.000151,
+      "loss": 2.1965,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.039163589477539,
+      "eval_runtime": 745.1648,
+      "eval_samples_per_second": 19.979,
+      "eval_steps_per_second": 2.497,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 2.039876699447632,
+      "eval_runtime": 745.0115,
+      "eval_samples_per_second": 19.984,
+      "eval_steps_per_second": 2.498,
+      "step": 495
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7146623134613037,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.9578,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.0375914573669434,
+      "eval_runtime": 744.9842,
+      "eval_samples_per_second": 19.984,
+      "eval_steps_per_second": 2.498,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.0346121788024902,
+      "eval_runtime": 746.8332,
+      "eval_samples_per_second": 19.935,
+      "eval_steps_per_second": 2.492,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.0184926986694336,
+      "learning_rate": 0.00014900000000000002,
+      "loss": 2.1056,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.032033920288086,
+      "eval_runtime": 755.7533,
+      "eval_samples_per_second": 19.7,
+      "eval_steps_per_second": 2.462,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.0300207138061523,
+      "eval_runtime": 752.8834,
+      "eval_samples_per_second": 19.775,
+      "eval_steps_per_second": 2.472,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6594335436820984,
+      "learning_rate": 0.000148,
+      "loss": 2.3502,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.028749465942383,
+      "eval_runtime": 782.3884,
+      "eval_samples_per_second": 19.029,
+      "eval_steps_per_second": 2.379,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.0278258323669434,
+      "eval_runtime": 785.0029,
+      "eval_samples_per_second": 18.966,
+      "eval_steps_per_second": 2.371,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8392504453659058,
+      "learning_rate": 0.000147,
+      "loss": 2.203,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.02721905708313,
+      "eval_runtime": 744.3301,
+      "eval_samples_per_second": 20.002,
+      "eval_steps_per_second": 2.5,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.0271682739257812,
+      "eval_runtime": 744.0199,
+      "eval_samples_per_second": 20.01,
+      "eval_steps_per_second": 2.501,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7347747683525085,
+      "learning_rate": 0.000146,
+      "loss": 2.2609,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 2.0258853435516357,
+      "eval_runtime": 744.263,
+      "eval_samples_per_second": 20.004,
+      "eval_steps_per_second": 2.5,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.026393413543701,
+      "eval_runtime": 744.209,
+      "eval_samples_per_second": 20.005,
+      "eval_steps_per_second": 2.501,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4300149083137512,
+      "learning_rate": 0.000145,
+      "loss": 1.9083,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0268478393554688,
+      "eval_runtime": 744.0279,
+      "eval_samples_per_second": 20.01,
+      "eval_steps_per_second": 2.501,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0266873836517334,
+      "eval_runtime": 744.0752,
+      "eval_samples_per_second": 20.009,
+      "eval_steps_per_second": 2.501,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.46820569038391113,
+      "learning_rate": 0.000144,
+      "loss": 2.1976,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0266172885894775,
+      "eval_runtime": 744.0609,
+      "eval_samples_per_second": 20.009,
+      "eval_steps_per_second": 2.501,
+      "step": 560
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0261571407318115,
+      "eval_runtime": 744.0034,
+      "eval_samples_per_second": 20.011,
+      "eval_steps_per_second": 2.501,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7849207520484924,
+      "learning_rate": 0.000143,
+      "loss": 1.8726,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0230391025543213,
+      "eval_runtime": 744.0221,
+      "eval_samples_per_second": 20.01,
+      "eval_steps_per_second": 2.501,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0199837684631348,
+      "eval_runtime": 744.0989,
+      "eval_samples_per_second": 20.008,
+      "eval_steps_per_second": 2.501,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6005426645278931,
+      "learning_rate": 0.000142,
+      "loss": 2.0767,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0186009407043457,
+      "eval_runtime": 744.1002,
+      "eval_samples_per_second": 20.008,
+      "eval_steps_per_second": 2.501,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 2.0188121795654297,
+      "eval_runtime": 768.6396,
+      "eval_samples_per_second": 19.369,
+      "eval_steps_per_second": 2.421,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4590970575809479,
+      "learning_rate": 0.000141,
+      "loss": 2.0767,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.01908278465271,
+      "eval_runtime": 767.2243,
+      "eval_samples_per_second": 19.405,
+      "eval_steps_per_second": 2.426,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.0185952186584473,
+      "eval_runtime": 744.1236,
+      "eval_samples_per_second": 20.007,
+      "eval_steps_per_second": 2.501,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8970805406570435,
+      "learning_rate": 0.00014,
+      "loss": 2.1128,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.0173678398132324,
+      "eval_runtime": 744.3957,
+      "eval_samples_per_second": 20.0,
+      "eval_steps_per_second": 2.5,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.015655279159546,
+      "eval_runtime": 744.2471,
+      "eval_samples_per_second": 20.004,
+      "eval_steps_per_second": 2.501,
+      "step": 605
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6531006693840027,
+      "learning_rate": 0.000139,
+      "loss": 2.0945,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.0149075984954834,
+      "eval_runtime": 744.0697,
+      "eval_samples_per_second": 20.009,
+      "eval_steps_per_second": 2.501,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.0139400959014893,
+      "eval_runtime": 744.1056,
+      "eval_samples_per_second": 20.008,
+      "eval_steps_per_second": 2.501,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6212542653083801,
+      "learning_rate": 0.000138,
+      "loss": 2.0035,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.0124478340148926,
+      "eval_runtime": 747.8028,
+      "eval_samples_per_second": 19.909,
+      "eval_steps_per_second": 2.489,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 2.0124478340148926,
+      "eval_runtime": 763.4128,
+      "eval_samples_per_second": 19.502,
+      "eval_steps_per_second": 2.438,
+      "step": 625
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6082640290260315,
+      "learning_rate": 0.00013700000000000002,
+      "loss": 2.1103,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.0123772621154785,
+      "eval_runtime": 745.1902,
+      "eval_samples_per_second": 19.979,
+      "eval_steps_per_second": 2.497,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.011105537414551,
+      "eval_runtime": 799.942,
+      "eval_samples_per_second": 18.611,
+      "eval_steps_per_second": 2.326,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7771211862564087,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 2.1028,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.009859323501587,
+      "eval_runtime": 796.318,
+      "eval_samples_per_second": 18.696,
+      "eval_steps_per_second": 2.337,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.009398937225342,
+      "eval_runtime": 745.0813,
+      "eval_samples_per_second": 19.982,
+      "eval_steps_per_second": 2.498,
+      "step": 645
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8797178268432617,
+      "learning_rate": 0.00013500000000000003,
+      "loss": 2.2711,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.0092194080352783,
+      "eval_runtime": 743.6761,
+      "eval_samples_per_second": 20.019,
+      "eval_steps_per_second": 2.502,
+      "step": 650
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.00980806350708,
+      "eval_runtime": 742.9948,
+      "eval_samples_per_second": 20.038,
+      "eval_steps_per_second": 2.505,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.1657720804214478,
+      "learning_rate": 0.000134,
+      "loss": 2.243,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.0087344646453857,
+      "eval_runtime": 742.7774,
+      "eval_samples_per_second": 20.044,
+      "eval_steps_per_second": 2.505,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.0070958137512207,
+      "eval_runtime": 748.9587,
+      "eval_samples_per_second": 19.878,
+      "eval_steps_per_second": 2.485,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6517012715339661,
+      "learning_rate": 0.000133,
+      "loss": 1.9059,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 2.0066118240356445,
+      "eval_runtime": 746.8276,
+      "eval_samples_per_second": 19.935,
+      "eval_steps_per_second": 2.492,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.006216287612915,
+      "eval_runtime": 746.1771,
+      "eval_samples_per_second": 19.952,
+      "eval_steps_per_second": 2.494,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6520020365715027,
+      "learning_rate": 0.000132,
+      "loss": 1.9953,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.0057804584503174,
+      "eval_runtime": 800.2909,
+      "eval_samples_per_second": 18.603,
+      "eval_steps_per_second": 2.325,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.0061757564544678,
+      "eval_runtime": 743.2104,
+      "eval_samples_per_second": 20.032,
+      "eval_steps_per_second": 2.504,
+      "step": 685
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7605012059211731,
+      "learning_rate": 0.000131,
+      "loss": 2.0684,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.0061566829681396,
+      "eval_runtime": 743.1127,
+      "eval_samples_per_second": 20.035,
+      "eval_steps_per_second": 2.504,
+      "step": 690
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.005591869354248,
+      "eval_runtime": 743.2532,
+      "eval_samples_per_second": 20.031,
+      "eval_steps_per_second": 2.504,
+      "step": 695
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8598791360855103,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.9624,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.00502347946167,
+      "eval_runtime": 743.0596,
+      "eval_samples_per_second": 20.036,
+      "eval_steps_per_second": 2.505,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.0032622814178467,
+      "eval_runtime": 743.2083,
+      "eval_samples_per_second": 20.032,
+      "eval_steps_per_second": 2.504,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5449765920639038,
+      "learning_rate": 0.00012900000000000002,
+      "loss": 2.0671,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.0018348693847656,
+      "eval_runtime": 742.4305,
+      "eval_samples_per_second": 20.053,
+      "eval_steps_per_second": 2.507,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 2.001499652862549,
+      "eval_runtime": 742.3408,
+      "eval_samples_per_second": 20.055,
+      "eval_steps_per_second": 2.507,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7031337022781372,
+      "learning_rate": 0.00012800000000000002,
+      "loss": 2.1012,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.0012526512145996,
+      "eval_runtime": 742.3685,
+      "eval_samples_per_second": 20.055,
+      "eval_steps_per_second": 2.507,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.0008952617645264,
+      "eval_runtime": 742.5083,
+      "eval_samples_per_second": 20.051,
+      "eval_steps_per_second": 2.506,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9199779033660889,
+      "learning_rate": 0.000127,
+      "loss": 2.1093,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.0010104179382324,
+      "eval_runtime": 742.3134,
+      "eval_samples_per_second": 20.056,
+      "eval_steps_per_second": 2.507,
+      "step": 730
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 2.000081777572632,
+      "eval_runtime": 742.5728,
+      "eval_samples_per_second": 20.049,
+      "eval_steps_per_second": 2.506,
+      "step": 735
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.5495649576187134,
+      "learning_rate": 0.000126,
+      "loss": 2.0745,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 1.9993902444839478,
+      "eval_runtime": 742.2628,
+      "eval_samples_per_second": 20.058,
+      "eval_steps_per_second": 2.507,
+      "step": 740
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 1.9984277486801147,
+      "eval_runtime": 742.413,
+      "eval_samples_per_second": 20.054,
+      "eval_steps_per_second": 2.507,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6500754952430725,
+      "learning_rate": 0.000125,
+      "loss": 2.1672,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 1.9985421895980835,
+      "eval_runtime": 742.5432,
+      "eval_samples_per_second": 20.05,
+      "eval_steps_per_second": 2.506,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 1.9995174407958984,
+      "eval_runtime": 742.7051,
+      "eval_samples_per_second": 20.046,
+      "eval_steps_per_second": 2.506,
+      "step": 755
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6539570689201355,
+      "learning_rate": 0.000124,
+      "loss": 2.0329,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.999580979347229,
+      "eval_runtime": 742.8251,
+      "eval_samples_per_second": 20.042,
+      "eval_steps_per_second": 2.505,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.9998208284378052,
+      "eval_runtime": 742.7139,
+      "eval_samples_per_second": 20.045,
+      "eval_steps_per_second": 2.506,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7342298626899719,
+      "learning_rate": 0.000123,
+      "loss": 2.1127,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.998628854751587,
+      "eval_runtime": 742.6862,
+      "eval_samples_per_second": 20.046,
+      "eval_steps_per_second": 2.506,
+      "step": 770
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.9968189001083374,
+      "eval_runtime": 742.6582,
+      "eval_samples_per_second": 20.047,
+      "eval_steps_per_second": 2.506,
+      "step": 775
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.5857991576194763,
+      "learning_rate": 0.000122,
+      "loss": 1.83,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.9953583478927612,
+      "eval_runtime": 742.6099,
+      "eval_samples_per_second": 20.048,
+      "eval_steps_per_second": 2.506,
+      "step": 780
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.9942070245742798,
+      "eval_runtime": 742.6392,
+      "eval_samples_per_second": 20.047,
+      "eval_steps_per_second": 2.506,
+      "step": 785
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7684934735298157,
+      "learning_rate": 0.000121,
+      "loss": 2.1953,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.9930860996246338,
+      "eval_runtime": 742.5366,
+      "eval_samples_per_second": 20.05,
+      "eval_steps_per_second": 2.506,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.9929825067520142,
+      "eval_runtime": 742.4635,
+      "eval_samples_per_second": 20.052,
+      "eval_steps_per_second": 2.507,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7895638346672058,
+      "learning_rate": 0.00012,
+      "loss": 2.0523,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 1.993416428565979,
+      "eval_runtime": 742.3749,
+      "eval_samples_per_second": 20.055,
+      "eval_steps_per_second": 2.507,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.9942388534545898,
+      "eval_runtime": 742.6099,
+      "eval_samples_per_second": 20.048,
+      "eval_steps_per_second": 2.506,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.7561328411102295,
+      "learning_rate": 0.000119,
+      "loss": 1.9476,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.9943628311157227,
+      "eval_runtime": 742.6406,
+      "eval_samples_per_second": 20.047,
+      "eval_steps_per_second": 2.506,
+      "step": 810
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.993873953819275,
+      "eval_runtime": 742.6764,
+      "eval_samples_per_second": 20.046,
+      "eval_steps_per_second": 2.506,
+      "step": 815
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6752389073371887,
+      "learning_rate": 0.000118,
+      "loss": 2.2179,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.993316411972046,
+      "eval_runtime": 742.8432,
+      "eval_samples_per_second": 20.042,
+      "eval_steps_per_second": 2.505,
+      "step": 820
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.9925029277801514,
+      "eval_runtime": 742.8122,
+      "eval_samples_per_second": 20.043,
+      "eval_steps_per_second": 2.505,
+      "step": 825
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8891268372535706,
+      "learning_rate": 0.000117,
+      "loss": 2.0997,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.9914040565490723,
+      "eval_runtime": 742.5596,
+      "eval_samples_per_second": 20.05,
+      "eval_steps_per_second": 2.506,
+      "step": 830
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.989838719367981,
+      "eval_runtime": 742.7205,
+      "eval_samples_per_second": 20.045,
+      "eval_steps_per_second": 2.506,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8138695359230042,
+      "learning_rate": 0.000116,
+      "loss": 2.0146,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.9878920316696167,
+      "eval_runtime": 746.0661,
+      "eval_samples_per_second": 19.955,
+      "eval_steps_per_second": 2.494,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.986674427986145,
+      "eval_runtime": 753.8888,
+      "eval_samples_per_second": 19.748,
+      "eval_steps_per_second": 2.469,
+      "step": 845
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8682060241699219,
+      "learning_rate": 0.00011499999999999999,
+      "loss": 2.1124,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9865528345108032,
+      "eval_runtime": 754.5751,
+      "eval_samples_per_second": 19.73,
+      "eval_steps_per_second": 2.466,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9874132871627808,
+      "eval_runtime": 758.8328,
+      "eval_samples_per_second": 19.62,
+      "eval_steps_per_second": 2.452,
+      "step": 855
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8115680813789368,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 2.3409,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9879497289657593,
+      "eval_runtime": 752.4654,
+      "eval_samples_per_second": 19.786,
+      "eval_steps_per_second": 2.473,
+      "step": 860
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9877504110336304,
+      "eval_runtime": 777.0301,
+      "eval_samples_per_second": 19.16,
+      "eval_steps_per_second": 2.395,
+      "step": 865
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8121163845062256,
+      "learning_rate": 0.000113,
+      "loss": 2.1718,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9862942695617676,
+      "eval_runtime": 789.4357,
+      "eval_samples_per_second": 18.859,
+      "eval_steps_per_second": 2.357,
+      "step": 870
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9849509000778198,
+      "eval_runtime": 766.4313,
+      "eval_samples_per_second": 19.425,
+      "eval_steps_per_second": 2.428,
+      "step": 875
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6633480787277222,
+      "learning_rate": 0.00011200000000000001,
+      "loss": 2.0154,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9856353998184204,
+      "eval_runtime": 797.5238,
+      "eval_samples_per_second": 18.668,
+      "eval_steps_per_second": 2.333,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9872477054595947,
+      "eval_runtime": 796.5052,
+      "eval_samples_per_second": 18.692,
+      "eval_steps_per_second": 2.336,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6087526679039001,
+      "learning_rate": 0.00011100000000000001,
+      "loss": 2.03,
+      "step": 890
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 1.9866758584976196,
+      "eval_runtime": 776.7422,
+      "eval_samples_per_second": 19.167,
+      "eval_steps_per_second": 2.396,
+      "step": 890
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.985320806503296,
+      "eval_runtime": 801.3372,
+      "eval_samples_per_second": 18.579,
+      "eval_steps_per_second": 2.322,
+      "step": 895
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8433945178985596,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 2.0573,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.9836574792861938,
+      "eval_runtime": 767.1796,
+      "eval_samples_per_second": 19.406,
+      "eval_steps_per_second": 2.426,
+      "step": 900
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.9830906391143799,
+      "eval_runtime": 755.8391,
+      "eval_samples_per_second": 19.697,
+      "eval_steps_per_second": 2.462,
+      "step": 905
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9100953340530396,
+      "learning_rate": 0.000109,
+      "loss": 1.9915,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.9824570417404175,
+      "eval_runtime": 766.9525,
+      "eval_samples_per_second": 19.412,
+      "eval_steps_per_second": 2.426,
+      "step": 910
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.983730435371399,
+      "eval_runtime": 761.0826,
+      "eval_samples_per_second": 19.562,
+      "eval_steps_per_second": 2.445,
+      "step": 915
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5524155497550964,
+      "learning_rate": 0.00010800000000000001,
+      "loss": 2.4334,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.9855434894561768,
+      "eval_runtime": 806.2529,
+      "eval_samples_per_second": 18.466,
+      "eval_steps_per_second": 2.308,
+      "step": 920
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.9850364923477173,
+      "eval_runtime": 759.49,
+      "eval_samples_per_second": 19.603,
+      "eval_steps_per_second": 2.45,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8319659233093262,
+      "learning_rate": 0.00010700000000000001,
+      "loss": 2.0895,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.9837418794631958,
+      "eval_runtime": 742.155,
+      "eval_samples_per_second": 20.06,
+      "eval_steps_per_second": 2.508,
+      "step": 930
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9824495315551758,
+      "eval_runtime": 742.1038,
+      "eval_samples_per_second": 20.062,
+      "eval_steps_per_second": 2.508,
+      "step": 935
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.5936623811721802,
+      "learning_rate": 0.00010600000000000002,
+      "loss": 2.018,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9815988540649414,
+      "eval_runtime": 742.0378,
+      "eval_samples_per_second": 20.064,
+      "eval_steps_per_second": 2.508,
+      "step": 940
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9809609651565552,
+      "eval_runtime": 742.3078,
+      "eval_samples_per_second": 20.056,
+      "eval_steps_per_second": 2.507,
+      "step": 945
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6863654255867004,
+      "learning_rate": 0.000105,
+      "loss": 2.1028,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9804768562316895,
+      "eval_runtime": 742.0981,
+      "eval_samples_per_second": 20.062,
+      "eval_steps_per_second": 2.508,
+      "step": 950
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.979957103729248,
+      "eval_runtime": 742.167,
+      "eval_samples_per_second": 20.06,
+      "eval_steps_per_second": 2.508,
+      "step": 955
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8558996915817261,
+      "learning_rate": 0.00010400000000000001,
+      "loss": 2.0259,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9790544509887695,
+      "eval_runtime": 742.0942,
+      "eval_samples_per_second": 20.062,
+      "eval_steps_per_second": 2.508,
+      "step": 960
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9782884120941162,
+      "eval_runtime": 742.2666,
+      "eval_samples_per_second": 20.057,
+      "eval_steps_per_second": 2.507,
+      "step": 965
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0001534223556519,
+      "learning_rate": 0.00010300000000000001,
+      "loss": 1.9648,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9775859117507935,
+      "eval_runtime": 742.263,
+      "eval_samples_per_second": 20.058,
+      "eval_steps_per_second": 2.507,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 1.9767318964004517,
+      "eval_runtime": 742.3891,
+      "eval_samples_per_second": 20.054,
+      "eval_steps_per_second": 2.507,
+      "step": 975
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8744415640830994,
+      "learning_rate": 0.00010200000000000001,
+      "loss": 2.0649,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.976646065711975,
+      "eval_runtime": 742.1965,
+      "eval_samples_per_second": 20.059,
+      "eval_steps_per_second": 2.507,
+      "step": 980
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.9765182733535767,
+      "eval_runtime": 742.453,
+      "eval_samples_per_second": 20.052,
+      "eval_steps_per_second": 2.507,
+      "step": 985
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7335282564163208,
+      "learning_rate": 0.000101,
+      "loss": 2.0636,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.9757829904556274,
+      "eval_runtime": 742.3078,
+      "eval_samples_per_second": 20.056,
+      "eval_steps_per_second": 2.507,
+      "step": 990
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.9756759405136108,
+      "eval_runtime": 742.3788,
+      "eval_samples_per_second": 20.054,
+      "eval_steps_per_second": 2.507,
+      "step": 995
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.9827843904495239,
+      "learning_rate": 0.0001,
+      "loss": 1.9848,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.9751605987548828,
+      "eval_runtime": 742.3183,
+      "eval_samples_per_second": 20.056,
+      "eval_steps_per_second": 2.507,
+      "step": 1000
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.974611759185791,
+      "eval_runtime": 741.7661,
+      "eval_samples_per_second": 20.071,
+      "eval_steps_per_second": 2.509,
+      "step": 1005
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8337591290473938,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.8473,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.974379301071167,
+      "eval_runtime": 741.9424,
+      "eval_samples_per_second": 20.066,
+      "eval_steps_per_second": 2.508,
+      "step": 1010
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.973697304725647,
+      "eval_runtime": 741.9395,
+      "eval_samples_per_second": 20.066,
+      "eval_steps_per_second": 2.508,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5600916147232056,
+      "learning_rate": 9.8e-05,
+      "loss": 1.9928,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 1.9728336334228516,
+      "eval_runtime": 742.058,
+      "eval_samples_per_second": 20.063,
+      "eval_steps_per_second": 2.508,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9720138311386108,
+      "eval_runtime": 741.909,
+      "eval_samples_per_second": 20.067,
+      "eval_steps_per_second": 2.508,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7983824014663696,
+      "learning_rate": 9.7e-05,
+      "loss": 1.9903,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9715585708618164,
+      "eval_runtime": 741.9989,
+      "eval_samples_per_second": 20.065,
+      "eval_steps_per_second": 2.508,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.971234917640686,
+      "eval_runtime": 742.1566,
+      "eval_samples_per_second": 20.06,
+      "eval_steps_per_second": 2.508,
+      "step": 1035
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0410112142562866,
+      "learning_rate": 9.6e-05,
+      "loss": 2.2528,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9711899757385254,
+      "eval_runtime": 742.4115,
+      "eval_samples_per_second": 20.054,
+      "eval_steps_per_second": 2.507,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9705572128295898,
+      "eval_runtime": 742.1151,
+      "eval_samples_per_second": 20.062,
+      "eval_steps_per_second": 2.508,
+      "step": 1045
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.9996401071548462,
+      "learning_rate": 9.5e-05,
+      "loss": 1.8971,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9699151515960693,
+      "eval_runtime": 742.02,
+      "eval_samples_per_second": 20.064,
+      "eval_steps_per_second": 2.508,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9693446159362793,
+      "eval_runtime": 742.2312,
+      "eval_samples_per_second": 20.058,
+      "eval_steps_per_second": 2.507,
+      "step": 1055
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6173760294914246,
+      "learning_rate": 9.4e-05,
+      "loss": 1.9011,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9691505432128906,
+      "eval_runtime": 742.0593,
+      "eval_samples_per_second": 20.063,
+      "eval_steps_per_second": 2.508,
+      "step": 1060
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9691288471221924,
+      "eval_runtime": 741.9923,
+      "eval_samples_per_second": 20.065,
+      "eval_steps_per_second": 2.508,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7394955158233643,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 2.0518,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9698320627212524,
+      "eval_runtime": 742.0516,
+      "eval_samples_per_second": 20.063,
+      "eval_steps_per_second": 2.508,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9706062078475952,
+      "eval_runtime": 746.4611,
+      "eval_samples_per_second": 19.945,
+      "eval_steps_per_second": 2.493,
+      "step": 1075
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6462586522102356,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 2.04,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9709925651550293,
+      "eval_runtime": 793.086,
+      "eval_samples_per_second": 18.772,
+      "eval_steps_per_second": 2.347,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9699233770370483,
+      "eval_runtime": 772.4813,
+      "eval_samples_per_second": 19.273,
+      "eval_steps_per_second": 2.409,
+      "step": 1085
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8134130835533142,
+      "learning_rate": 9.1e-05,
+      "loss": 1.9992,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9690241813659668,
+      "eval_runtime": 742.7771,
+      "eval_samples_per_second": 20.044,
+      "eval_steps_per_second": 2.505,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9685667753219604,
+      "eval_runtime": 742.8374,
+      "eval_samples_per_second": 20.042,
+      "eval_steps_per_second": 2.505,
+      "step": 1095
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.065739631652832,
+      "learning_rate": 9e-05,
+      "loss": 2.2447,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.9679232835769653,
+      "eval_runtime": 743.058,
+      "eval_samples_per_second": 20.036,
+      "eval_steps_per_second": 2.505,
+      "step": 1100
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.967824101448059,
+      "eval_runtime": 742.6615,
+      "eval_samples_per_second": 20.047,
+      "eval_steps_per_second": 2.506,
+      "step": 1105
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8471609950065613,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 2.0276,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.9677886962890625,
+      "eval_runtime": 742.686,
+      "eval_samples_per_second": 20.046,
+      "eval_steps_per_second": 2.506,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.9673609733581543,
+      "eval_runtime": 742.7476,
+      "eval_samples_per_second": 20.044,
+      "eval_steps_per_second": 2.506,
+      "step": 1115
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.0944584608078003,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.9626,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.9674118757247925,
+      "eval_runtime": 742.8656,
+      "eval_samples_per_second": 20.041,
+      "eval_steps_per_second": 2.505,
+      "step": 1120
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.966983675956726,
+      "eval_runtime": 742.5557,
+      "eval_samples_per_second": 20.05,
+      "eval_steps_per_second": 2.506,
+      "step": 1125
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.0942668914794922,
+      "learning_rate": 8.7e-05,
+      "loss": 2.1137,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.966120719909668,
+      "eval_runtime": 743.2494,
+      "eval_samples_per_second": 20.031,
+      "eval_steps_per_second": 2.504,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.9657062292099,
+      "eval_runtime": 742.7716,
+      "eval_samples_per_second": 20.044,
+      "eval_steps_per_second": 2.505,
+      "step": 1135
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.9553364515304565,
+      "learning_rate": 8.6e-05,
+      "loss": 1.9172,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.9652293920516968,
+      "eval_runtime": 742.8386,
+      "eval_samples_per_second": 20.042,
+      "eval_steps_per_second": 2.505,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.965266466140747,
+      "eval_runtime": 742.7224,
+      "eval_samples_per_second": 20.045,
+      "eval_steps_per_second": 2.506,
+      "step": 1145
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7700403332710266,
+      "learning_rate": 8.5e-05,
+      "loss": 1.969,
+      "step": 1150
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.9645556211471558,
+      "eval_runtime": 742.7935,
+      "eval_samples_per_second": 20.043,
+      "eval_steps_per_second": 2.505,
+      "step": 1150
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.9638077020645142,
+      "eval_runtime": 742.9212,
+      "eval_samples_per_second": 20.04,
+      "eval_steps_per_second": 2.505,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6622421741485596,
+      "learning_rate": 8.4e-05,
+      "loss": 2.0767,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.9633898735046387,
+      "eval_runtime": 743.0434,
+      "eval_samples_per_second": 20.037,
+      "eval_steps_per_second": 2.505,
+      "step": 1160
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.9631668329238892,
+      "eval_runtime": 743.0892,
+      "eval_samples_per_second": 20.035,
+      "eval_steps_per_second": 2.504,
+      "step": 1165
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.902953565120697,
+      "learning_rate": 8.3e-05,
+      "loss": 1.9907,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.9637709856033325,
+      "eval_runtime": 742.9531,
+      "eval_samples_per_second": 20.039,
+      "eval_steps_per_second": 2.505,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.964076042175293,
+      "eval_runtime": 752.1865,
+      "eval_samples_per_second": 19.793,
+      "eval_steps_per_second": 2.474,
+      "step": 1175
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6186763644218445,
+      "learning_rate": 8.2e-05,
+      "loss": 2.0733,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.9636982679367065,
+      "eval_runtime": 806.7383,
+      "eval_samples_per_second": 18.455,
+      "eval_steps_per_second": 2.307,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.9626468420028687,
+      "eval_runtime": 794.7659,
+      "eval_samples_per_second": 18.733,
+      "eval_steps_per_second": 2.342,
+      "step": 1185
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.3641421794891357,
+      "learning_rate": 8.1e-05,
+      "loss": 2.2387,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 1.962171196937561,
+      "eval_runtime": 742.946,
+      "eval_samples_per_second": 20.039,
+      "eval_steps_per_second": 2.505,
+      "step": 1190
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9623751640319824,
+      "eval_runtime": 748.6093,
+      "eval_samples_per_second": 19.888,
+      "eval_steps_per_second": 2.486,
+      "step": 1195
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8681169748306274,
+      "learning_rate": 8e-05,
+      "loss": 1.7697,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9623966217041016,
+      "eval_runtime": 747.3889,
+      "eval_samples_per_second": 19.92,
+      "eval_steps_per_second": 2.49,
+      "step": 1200
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9618523120880127,
+      "eval_runtime": 747.8306,
+      "eval_samples_per_second": 19.908,
+      "eval_steps_per_second": 2.489,
+      "step": 1205
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8061439990997314,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 1.8372,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9610604047775269,
+      "eval_runtime": 754.4243,
+      "eval_samples_per_second": 19.734,
+      "eval_steps_per_second": 2.467,
+      "step": 1210
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9600740671157837,
+      "eval_runtime": 773.0575,
+      "eval_samples_per_second": 19.259,
+      "eval_steps_per_second": 2.407,
+      "step": 1215
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6680261492729187,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.8579,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9600106477737427,
+      "eval_runtime": 758.3783,
+      "eval_samples_per_second": 19.631,
+      "eval_steps_per_second": 2.454,
+      "step": 1220
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.960121512413025,
+      "eval_runtime": 742.6905,
+      "eval_samples_per_second": 20.046,
+      "eval_steps_per_second": 2.506,
+      "step": 1225
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.7453675866127014,
+      "learning_rate": 7.7e-05,
+      "loss": 2.0669,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9599344730377197,
+      "eval_runtime": 742.6985,
+      "eval_samples_per_second": 20.046,
+      "eval_steps_per_second": 2.506,
+      "step": 1230
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9595812559127808,
+      "eval_runtime": 743.459,
+      "eval_samples_per_second": 20.025,
+      "eval_steps_per_second": 2.503,
+      "step": 1235
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8137094378471375,
+      "learning_rate": 7.6e-05,
+      "loss": 1.8052,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.959162950515747,
+      "eval_runtime": 762.4324,
+      "eval_samples_per_second": 19.527,
+      "eval_steps_per_second": 2.441,
+      "step": 1240
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.958164930343628,
+      "eval_runtime": 760.8758,
+      "eval_samples_per_second": 19.567,
+      "eval_steps_per_second": 2.446,
+      "step": 1245
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9641257524490356,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 2.012,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.9580010175704956,
+      "eval_runtime": 758.6208,
+      "eval_samples_per_second": 19.625,
+      "eval_steps_per_second": 2.453,
+      "step": 1250
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.9579390287399292,
+      "eval_runtime": 756.8034,
+      "eval_samples_per_second": 19.672,
+      "eval_steps_per_second": 2.459,
+      "step": 1255
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9015198349952698,
+      "learning_rate": 7.4e-05,
+      "loss": 1.7978,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.957844614982605,
+      "eval_runtime": 762.8272,
+      "eval_samples_per_second": 19.517,
+      "eval_steps_per_second": 2.44,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.9587702751159668,
+      "eval_runtime": 757.6095,
+      "eval_samples_per_second": 19.651,
+      "eval_steps_per_second": 2.456,
+      "step": 1265
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6779741644859314,
+      "learning_rate": 7.3e-05,
+      "loss": 2.2184,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.9595640897750854,
+      "eval_runtime": 752.2491,
+      "eval_samples_per_second": 19.791,
+      "eval_steps_per_second": 2.474,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.9594327211380005,
+      "eval_runtime": 753.8257,
+      "eval_samples_per_second": 19.75,
+      "eval_steps_per_second": 2.469,
+      "step": 1275
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.2027363777160645,
+      "learning_rate": 7.2e-05,
+      "loss": 2.1702,
+      "step": 1280
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.9583239555358887,
+      "eval_runtime": 762.8807,
+      "eval_samples_per_second": 19.516,
+      "eval_steps_per_second": 2.439,
+      "step": 1280
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.9568368196487427,
+      "eval_runtime": 760.0025,
+      "eval_samples_per_second": 19.589,
+      "eval_steps_per_second": 2.449,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.690773069858551,
+      "learning_rate": 7.1e-05,
+      "loss": 2.1691,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.9561883211135864,
+      "eval_runtime": 763.7974,
+      "eval_samples_per_second": 19.492,
+      "eval_steps_per_second": 2.437,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.9560072422027588,
+      "eval_runtime": 750.1629,
+      "eval_samples_per_second": 19.846,
+      "eval_steps_per_second": 2.481,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5568638443946838,
+      "learning_rate": 7e-05,
+      "loss": 2.0352,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.9558581113815308,
+      "eval_runtime": 771.8238,
+      "eval_samples_per_second": 19.289,
+      "eval_steps_per_second": 2.411,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.9561656713485718,
+      "eval_runtime": 761.3654,
+      "eval_samples_per_second": 19.554,
+      "eval_steps_per_second": 2.444,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.9243978261947632,
+      "learning_rate": 6.9e-05,
+      "loss": 2.1049,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.9566450119018555,
+      "eval_runtime": 774.2138,
+      "eval_samples_per_second": 19.23,
+      "eval_steps_per_second": 2.404,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.956655502319336,
+      "eval_runtime": 757.6655,
+      "eval_samples_per_second": 19.65,
+      "eval_steps_per_second": 2.456,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.799391508102417,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.9632,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 1.9563167095184326,
+      "eval_runtime": 775.3068,
+      "eval_samples_per_second": 19.203,
+      "eval_steps_per_second": 2.4,
+      "step": 1320
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.955706000328064,
+      "eval_runtime": 775.6046,
+      "eval_samples_per_second": 19.195,
+      "eval_steps_per_second": 2.399,
+      "step": 1325
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5652315020561218,
+      "learning_rate": 6.7e-05,
+      "loss": 2.1231,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9552385807037354,
+      "eval_runtime": 774.2845,
+      "eval_samples_per_second": 19.228,
+      "eval_steps_per_second": 2.404,
+      "step": 1330
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9548553228378296,
+      "eval_runtime": 804.564,
+      "eval_samples_per_second": 18.504,
+      "eval_steps_per_second": 2.313,
+      "step": 1335
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.838487446308136,
+      "learning_rate": 6.6e-05,
+      "loss": 2.1477,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9545986652374268,
+      "eval_runtime": 804.7596,
+      "eval_samples_per_second": 18.5,
+      "eval_steps_per_second": 2.312,
+      "step": 1340
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.954168438911438,
+      "eval_runtime": 804.8099,
+      "eval_samples_per_second": 18.499,
+      "eval_steps_per_second": 2.312,
+      "step": 1345
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6252363920211792,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.9284,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9540753364562988,
+      "eval_runtime": 756.5934,
+      "eval_samples_per_second": 19.678,
+      "eval_steps_per_second": 2.46,
+      "step": 1350
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9538133144378662,
+      "eval_runtime": 753.786,
+      "eval_samples_per_second": 19.751,
+      "eval_steps_per_second": 2.469,
+      "step": 1355
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7139986753463745,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.9112,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9534461498260498,
+      "eval_runtime": 752.6982,
+      "eval_samples_per_second": 19.78,
+      "eval_steps_per_second": 2.472,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9531402587890625,
+      "eval_runtime": 753.832,
+      "eval_samples_per_second": 19.75,
+      "eval_steps_per_second": 2.469,
+      "step": 1365
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7954660058021545,
+      "learning_rate": 6.3e-05,
+      "loss": 1.879,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9530184268951416,
+      "eval_runtime": 749.4503,
+      "eval_samples_per_second": 19.865,
+      "eval_steps_per_second": 2.483,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9531172513961792,
+      "eval_runtime": 751.563,
+      "eval_samples_per_second": 19.809,
+      "eval_steps_per_second": 2.476,
+      "step": 1375
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6721971035003662,
+      "learning_rate": 6.2e-05,
+      "loss": 1.793,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9533251523971558,
+      "eval_runtime": 760.6235,
+      "eval_samples_per_second": 19.573,
+      "eval_steps_per_second": 2.447,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9533405303955078,
+      "eval_runtime": 762.9017,
+      "eval_samples_per_second": 19.515,
+      "eval_steps_per_second": 2.439,
+      "step": 1385
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9299044609069824,
+      "learning_rate": 6.1e-05,
+      "loss": 2.0359,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9533050060272217,
+      "eval_runtime": 770.8252,
+      "eval_samples_per_second": 19.314,
+      "eval_steps_per_second": 2.414,
+      "step": 1390
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9530737400054932,
+      "eval_runtime": 774.7678,
+      "eval_samples_per_second": 19.216,
+      "eval_steps_per_second": 2.402,
+      "step": 1395
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8456938862800598,
+      "learning_rate": 6e-05,
+      "loss": 2.1071,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9526972770690918,
+      "eval_runtime": 785.1544,
+      "eval_samples_per_second": 18.962,
+      "eval_steps_per_second": 2.37,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.9520927667617798,
+      "eval_runtime": 774.7517,
+      "eval_samples_per_second": 19.216,
+      "eval_steps_per_second": 2.402,
+      "step": 1405
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8644620180130005,
+      "learning_rate": 5.9e-05,
+      "loss": 1.922,
+      "step": 1410
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.951416015625,
+      "eval_runtime": 770.2102,
+      "eval_samples_per_second": 19.33,
+      "eval_steps_per_second": 2.416,
+      "step": 1410
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.951120138168335,
+      "eval_runtime": 776.0343,
+      "eval_samples_per_second": 19.185,
+      "eval_steps_per_second": 2.398,
+      "step": 1415
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7249743342399597,
+      "learning_rate": 5.8e-05,
+      "loss": 1.8578,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.9509921073913574,
+      "eval_runtime": 786.6,
+      "eval_samples_per_second": 18.927,
+      "eval_steps_per_second": 2.366,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.9510488510131836,
+      "eval_runtime": 776.8377,
+      "eval_samples_per_second": 19.165,
+      "eval_steps_per_second": 2.396,
+      "step": 1425
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6929222941398621,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 2.2087,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.9513132572174072,
+      "eval_runtime": 759.5413,
+      "eval_samples_per_second": 19.601,
+      "eval_steps_per_second": 2.45,
+      "step": 1430
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.9508575201034546,
+      "eval_runtime": 762.6392,
+      "eval_samples_per_second": 19.522,
+      "eval_steps_per_second": 2.44,
+      "step": 1435
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.7323941588401794,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 2.1452,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.950052261352539,
+      "eval_runtime": 744.7027,
+      "eval_samples_per_second": 19.992,
+      "eval_steps_per_second": 2.499,
+      "step": 1440
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.9493366479873657,
+      "eval_runtime": 740.8795,
+      "eval_samples_per_second": 20.095,
+      "eval_steps_per_second": 2.512,
+      "step": 1445
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.9700380563735962,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.945,
+      "step": 1450
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.9489691257476807,
+      "eval_runtime": 740.6194,
+      "eval_samples_per_second": 20.102,
+      "eval_steps_per_second": 2.513,
+      "step": 1450
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.9490820169448853,
+      "eval_runtime": 740.5722,
+      "eval_samples_per_second": 20.103,
+      "eval_steps_per_second": 2.513,
+      "step": 1455
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.704833984375,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 2.158,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.949270486831665,
+      "eval_runtime": 743.446,
+      "eval_samples_per_second": 20.026,
+      "eval_steps_per_second": 2.503,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.9491441249847412,
+      "eval_runtime": 768.2882,
+      "eval_samples_per_second": 19.378,
+      "eval_steps_per_second": 2.422,
+      "step": 1465
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7452517747879028,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 2.1391,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.949285864830017,
+      "eval_runtime": 784.4371,
+      "eval_samples_per_second": 18.979,
+      "eval_steps_per_second": 2.372,
+      "step": 1470
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.9491676092147827,
+      "eval_runtime": 779.509,
+      "eval_samples_per_second": 19.099,
+      "eval_steps_per_second": 2.387,
+      "step": 1475
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.1882747411727905,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 2.0945,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.9490594863891602,
+      "eval_runtime": 773.5132,
+      "eval_samples_per_second": 19.247,
+      "eval_steps_per_second": 2.406,
+      "step": 1480
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.9491394758224487,
+      "eval_runtime": 769.1156,
+      "eval_samples_per_second": 19.357,
+      "eval_steps_per_second": 2.42,
+      "step": 1485
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.636154055595398,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 1.812,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.9490303993225098,
+      "eval_runtime": 800.4965,
+      "eval_samples_per_second": 18.598,
+      "eval_steps_per_second": 2.325,
+      "step": 1490
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.9486448764801025,
+      "eval_runtime": 800.2384,
+      "eval_samples_per_second": 18.604,
+      "eval_steps_per_second": 2.326,
+      "step": 1495
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7470951676368713,
+      "learning_rate": 5e-05,
+      "loss": 2.0111,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9481943845748901,
+      "eval_runtime": 799.7192,
+      "eval_samples_per_second": 18.617,
+      "eval_steps_per_second": 2.327,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9475739002227783,
+      "eval_runtime": 799.7239,
+      "eval_samples_per_second": 18.616,
+      "eval_steps_per_second": 2.327,
+      "step": 1505
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1213610172271729,
+      "learning_rate": 4.9e-05,
+      "loss": 2.2712,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9474812746047974,
+      "eval_runtime": 799.7984,
+      "eval_samples_per_second": 18.615,
+      "eval_steps_per_second": 2.327,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.947679877281189,
+      "eval_runtime": 799.9014,
+      "eval_samples_per_second": 18.612,
+      "eval_steps_per_second": 2.327,
+      "step": 1515
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1590468883514404,
+      "learning_rate": 4.8e-05,
+      "loss": 2.0296,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.947396159172058,
+      "eval_runtime": 799.8052,
+      "eval_samples_per_second": 18.615,
+      "eval_steps_per_second": 2.327,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9468810558319092,
+      "eval_runtime": 799.9551,
+      "eval_samples_per_second": 18.611,
+      "eval_steps_per_second": 2.326,
+      "step": 1525
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7615826725959778,
+      "learning_rate": 4.7e-05,
+      "loss": 1.8507,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9464668035507202,
+      "eval_runtime": 799.7366,
+      "eval_samples_per_second": 18.616,
+      "eval_steps_per_second": 2.327,
+      "step": 1530
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9460117816925049,
+      "eval_runtime": 799.5384,
+      "eval_samples_per_second": 18.621,
+      "eval_steps_per_second": 2.328,
+      "step": 1535
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.255834698677063,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.9103,
+      "step": 1540
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9458439350128174,
+      "eval_runtime": 799.7058,
+      "eval_samples_per_second": 18.617,
+      "eval_steps_per_second": 2.327,
+      "step": 1540
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.9456100463867188,
+      "eval_runtime": 799.6359,
+      "eval_samples_per_second": 18.618,
+      "eval_steps_per_second": 2.327,
+      "step": 1545
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8211413621902466,
+      "learning_rate": 4.5e-05,
+      "loss": 1.9487,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.94538152217865,
+      "eval_runtime": 799.8615,
+      "eval_samples_per_second": 18.613,
+      "eval_steps_per_second": 2.327,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.9452861547470093,
+      "eval_runtime": 799.6185,
+      "eval_samples_per_second": 18.619,
+      "eval_steps_per_second": 2.327,
+      "step": 1555
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0679750442504883,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 2.0504,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.9451079368591309,
+      "eval_runtime": 799.6427,
+      "eval_samples_per_second": 18.618,
+      "eval_steps_per_second": 2.327,
+      "step": 1560
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.945074439048767,
+      "eval_runtime": 799.6799,
+      "eval_samples_per_second": 18.617,
+      "eval_steps_per_second": 2.327,
+      "step": 1565
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6025732159614563,
+      "learning_rate": 4.3e-05,
+      "loss": 2.1497,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.9450632333755493,
+      "eval_runtime": 799.6958,
+      "eval_samples_per_second": 18.617,
+      "eval_steps_per_second": 2.327,
+      "step": 1570
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.9450970888137817,
+      "eval_runtime": 799.7035,
+      "eval_samples_per_second": 18.617,
+      "eval_steps_per_second": 2.327,
+      "step": 1575
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9095989465713501,
+      "learning_rate": 4.2e-05,
+      "loss": 1.9836,
+      "step": 1580
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.9450123310089111,
+      "eval_runtime": 799.6334,
+      "eval_samples_per_second": 18.619,
+      "eval_steps_per_second": 2.327,
+      "step": 1580
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9449422359466553,
+      "eval_runtime": 799.7912,
+      "eval_samples_per_second": 18.615,
+      "eval_steps_per_second": 2.327,
+      "step": 1585
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.714409351348877,
+      "learning_rate": 4.1e-05,
+      "loss": 1.8096,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.944868803024292,
+      "eval_runtime": 799.8495,
+      "eval_samples_per_second": 18.614,
+      "eval_steps_per_second": 2.327,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9445998668670654,
+      "eval_runtime": 799.4538,
+      "eval_samples_per_second": 18.623,
+      "eval_steps_per_second": 2.328,
+      "step": 1595
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5383341908454895,
+      "learning_rate": 4e-05,
+      "loss": 2.0759,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9442588090896606,
+      "eval_runtime": 777.9784,
+      "eval_samples_per_second": 19.137,
+      "eval_steps_per_second": 2.392,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9438750743865967,
+      "eval_runtime": 738.7615,
+      "eval_samples_per_second": 20.153,
+      "eval_steps_per_second": 2.519,
+      "step": 1605
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8301254510879517,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 2.0214,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9435011148452759,
+      "eval_runtime": 739.0062,
+      "eval_samples_per_second": 20.146,
+      "eval_steps_per_second": 2.518,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9431562423706055,
+      "eval_runtime": 738.2469,
+      "eval_samples_per_second": 20.167,
+      "eval_steps_per_second": 2.521,
+      "step": 1615
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9721934795379639,
+      "learning_rate": 3.8e-05,
+      "loss": 1.9382,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9430102109909058,
+      "eval_runtime": 738.2353,
+      "eval_samples_per_second": 20.167,
+      "eval_steps_per_second": 2.521,
+      "step": 1620
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 1.9429086446762085,
+      "eval_runtime": 738.1793,
+      "eval_samples_per_second": 20.169,
+      "eval_steps_per_second": 2.521,
+      "step": 1625
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9744455218315125,
+      "learning_rate": 3.7e-05,
+      "loss": 2.0086,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.942738652229309,
+      "eval_runtime": 738.1789,
+      "eval_samples_per_second": 20.169,
+      "eval_steps_per_second": 2.521,
+      "step": 1630
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9425232410430908,
+      "eval_runtime": 738.3004,
+      "eval_samples_per_second": 20.165,
+      "eval_steps_per_second": 2.521,
+      "step": 1635
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9008609652519226,
+      "learning_rate": 3.6e-05,
+      "loss": 2.1409,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9424079656600952,
+      "eval_runtime": 738.2426,
+      "eval_samples_per_second": 20.167,
+      "eval_steps_per_second": 2.521,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9424738883972168,
+      "eval_runtime": 738.2907,
+      "eval_samples_per_second": 20.165,
+      "eval_steps_per_second": 2.521,
+      "step": 1645
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7677657604217529,
+      "learning_rate": 3.5e-05,
+      "loss": 2.067,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9425007104873657,
+      "eval_runtime": 742.0127,
+      "eval_samples_per_second": 20.064,
+      "eval_steps_per_second": 2.508,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9425512552261353,
+      "eval_runtime": 750.5258,
+      "eval_samples_per_second": 19.837,
+      "eval_steps_per_second": 2.48,
+      "step": 1655
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6732711791992188,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.7463,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9426753520965576,
+      "eval_runtime": 748.9114,
+      "eval_samples_per_second": 19.88,
+      "eval_steps_per_second": 2.485,
+      "step": 1660
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.942758560180664,
+      "eval_runtime": 746.9293,
+      "eval_samples_per_second": 19.932,
+      "eval_steps_per_second": 2.492,
+      "step": 1665
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9339891076087952,
+      "learning_rate": 3.3e-05,
+      "loss": 2.001,
+      "step": 1670
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9427272081375122,
+      "eval_runtime": 797.2689,
+      "eval_samples_per_second": 18.674,
+      "eval_steps_per_second": 2.334,
+      "step": 1670
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9428359270095825,
+      "eval_runtime": 800.5836,
+      "eval_samples_per_second": 18.596,
+      "eval_steps_per_second": 2.325,
+      "step": 1675
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8810248374938965,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.8671,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9427249431610107,
+      "eval_runtime": 799.8884,
+      "eval_samples_per_second": 18.613,
+      "eval_steps_per_second": 2.327,
+      "step": 1680
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9424289464950562,
+      "eval_runtime": 799.7098,
+      "eval_samples_per_second": 18.617,
+      "eval_steps_per_second": 2.327,
+      "step": 1685
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9490606784820557,
+      "learning_rate": 3.1e-05,
+      "loss": 1.9094,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9420220851898193,
+      "eval_runtime": 752.2407,
+      "eval_samples_per_second": 19.792,
+      "eval_steps_per_second": 2.474,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9419386386871338,
+      "eval_runtime": 738.4421,
+      "eval_samples_per_second": 20.161,
+      "eval_steps_per_second": 2.52,
+      "step": 1695
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.7382946610450745,
+      "learning_rate": 3e-05,
+      "loss": 2.048,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9419728517532349,
+      "eval_runtime": 738.4148,
+      "eval_samples_per_second": 20.162,
+      "eval_steps_per_second": 2.52,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9419692754745483,
+      "eval_runtime": 744.3255,
+      "eval_samples_per_second": 20.002,
+      "eval_steps_per_second": 2.5,
+      "step": 1705
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8462529182434082,
+      "learning_rate": 2.9e-05,
+      "loss": 1.9912,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9418373107910156,
+      "eval_runtime": 745.3964,
+      "eval_samples_per_second": 19.973,
+      "eval_steps_per_second": 2.497,
+      "step": 1710
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 1.9416935443878174,
+      "eval_runtime": 745.4722,
+      "eval_samples_per_second": 19.971,
+      "eval_steps_per_second": 2.496,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0404905080795288,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.1166,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9415353536605835,
+      "eval_runtime": 744.4599,
+      "eval_samples_per_second": 19.998,
+      "eval_steps_per_second": 2.5,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.941475510597229,
+      "eval_runtime": 747.4114,
+      "eval_samples_per_second": 19.919,
+      "eval_steps_per_second": 2.49,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9904809594154358,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.8904,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9413694143295288,
+      "eval_runtime": 745.7476,
+      "eval_samples_per_second": 19.964,
+      "eval_steps_per_second": 2.495,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9412492513656616,
+      "eval_runtime": 746.5069,
+      "eval_samples_per_second": 19.944,
+      "eval_steps_per_second": 2.493,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8735015988349915,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 1.926,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9410345554351807,
+      "eval_runtime": 745.8571,
+      "eval_samples_per_second": 19.961,
+      "eval_steps_per_second": 2.495,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.940790057182312,
+      "eval_runtime": 745.3703,
+      "eval_samples_per_second": 19.974,
+      "eval_steps_per_second": 2.497,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.8582627177238464,
+      "learning_rate": 2.5e-05,
+      "loss": 2.0692,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9406797885894775,
+      "eval_runtime": 759.2984,
+      "eval_samples_per_second": 19.608,
+      "eval_steps_per_second": 2.451,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9406994581222534,
+      "eval_runtime": 799.6272,
+      "eval_samples_per_second": 18.619,
+      "eval_steps_per_second": 2.327,
+      "step": 1755
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.8529316782951355,
+      "learning_rate": 2.4e-05,
+      "loss": 2.0355,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9407473802566528,
+      "eval_runtime": 799.5818,
+      "eval_samples_per_second": 18.62,
+      "eval_steps_per_second": 2.327,
+      "step": 1760
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9407432079315186,
+      "eval_runtime": 799.2768,
+      "eval_samples_per_second": 18.627,
+      "eval_steps_per_second": 2.328,
+      "step": 1765
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9113526940345764,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 1.9542,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9405388832092285,
+      "eval_runtime": 799.4163,
+      "eval_samples_per_second": 18.624,
+      "eval_steps_per_second": 2.328,
+      "step": 1770
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9403679370880127,
+      "eval_runtime": 799.618,
+      "eval_samples_per_second": 18.619,
+      "eval_steps_per_second": 2.327,
+      "step": 1775
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6997771859169006,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.9599,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9402695894241333,
+      "eval_runtime": 799.4944,
+      "eval_samples_per_second": 18.622,
+      "eval_steps_per_second": 2.328,
+      "step": 1780
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9402552843093872,
+      "eval_runtime": 799.5449,
+      "eval_samples_per_second": 18.621,
+      "eval_steps_per_second": 2.328,
+      "step": 1785
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6518684029579163,
+      "learning_rate": 2.1e-05,
+      "loss": 1.8072,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9402391910552979,
+      "eval_runtime": 799.5838,
+      "eval_samples_per_second": 18.62,
+      "eval_steps_per_second": 2.327,
+      "step": 1790
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9402625560760498,
+      "eval_runtime": 764.0189,
+      "eval_samples_per_second": 19.486,
+      "eval_steps_per_second": 2.436,
+      "step": 1795
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0698702335357666,
+      "learning_rate": 2e-05,
+      "loss": 2.0761,
+      "step": 1800
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 1.9402350187301636,
+      "eval_runtime": 766.71,
+      "eval_samples_per_second": 19.418,
+      "eval_steps_per_second": 2.427,
+      "step": 1800
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.94016695022583,
+      "eval_runtime": 767.2932,
+      "eval_samples_per_second": 19.403,
+      "eval_steps_per_second": 2.425,
+      "step": 1805
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5021382570266724,
+      "learning_rate": 1.9e-05,
+      "loss": 2.1248,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9400701522827148,
+      "eval_runtime": 751.0521,
+      "eval_samples_per_second": 19.823,
+      "eval_steps_per_second": 2.478,
+      "step": 1810
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9399473667144775,
+      "eval_runtime": 781.0617,
+      "eval_samples_per_second": 19.061,
+      "eval_steps_per_second": 2.383,
+      "step": 1815
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5781351327896118,
+      "learning_rate": 1.8e-05,
+      "loss": 1.8231,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9398096799850464,
+      "eval_runtime": 786.8736,
+      "eval_samples_per_second": 18.92,
+      "eval_steps_per_second": 2.365,
+      "step": 1820
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.93966805934906,
+      "eval_runtime": 768.2598,
+      "eval_samples_per_second": 19.379,
+      "eval_steps_per_second": 2.422,
+      "step": 1825
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6509262919425964,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 2.0489,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9395979642868042,
+      "eval_runtime": 751.1152,
+      "eval_samples_per_second": 19.821,
+      "eval_steps_per_second": 2.478,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9395313262939453,
+      "eval_runtime": 759.4106,
+      "eval_samples_per_second": 19.605,
+      "eval_steps_per_second": 2.451,
+      "step": 1835
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.2815340757369995,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.0146,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.939477562904358,
+      "eval_runtime": 749.6182,
+      "eval_samples_per_second": 19.861,
+      "eval_steps_per_second": 2.483,
+      "step": 1840
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9394245147705078,
+      "eval_runtime": 752.8584,
+      "eval_samples_per_second": 19.775,
+      "eval_steps_per_second": 2.472,
+      "step": 1845
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.43942534923553467,
+      "learning_rate": 1.5e-05,
+      "loss": 1.942,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.9393796920776367,
+      "eval_runtime": 759.4255,
+      "eval_samples_per_second": 19.604,
+      "eval_steps_per_second": 2.451,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.9393460750579834,
+      "eval_runtime": 804.7704,
+      "eval_samples_per_second": 18.5,
+      "eval_steps_per_second": 2.312,
+      "step": 1855
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7411811947822571,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 2.1642,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.9392962455749512,
+      "eval_runtime": 804.5633,
+      "eval_samples_per_second": 18.504,
+      "eval_steps_per_second": 2.313,
+      "step": 1860
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.939192533493042,
+      "eval_runtime": 795.592,
+      "eval_samples_per_second": 18.713,
+      "eval_steps_per_second": 2.339,
+      "step": 1865
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.844436764717102,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.8236,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.939056634902954,
+      "eval_runtime": 752.4134,
+      "eval_samples_per_second": 19.787,
+      "eval_steps_per_second": 2.473,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.9389359951019287,
+      "eval_runtime": 748.7084,
+      "eval_samples_per_second": 19.885,
+      "eval_steps_per_second": 2.486,
+      "step": 1875
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8722310662269592,
+      "learning_rate": 1.2e-05,
+      "loss": 2.134,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.9388750791549683,
+      "eval_runtime": 751.6317,
+      "eval_samples_per_second": 19.808,
+      "eval_steps_per_second": 2.476,
+      "step": 1880
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.9388281106948853,
+      "eval_runtime": 754.3186,
+      "eval_samples_per_second": 19.737,
+      "eval_steps_per_second": 2.467,
+      "step": 1885
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8985447287559509,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.9291,
+      "step": 1890
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.938772439956665,
+      "eval_runtime": 769.6876,
+      "eval_samples_per_second": 19.343,
+      "eval_steps_per_second": 2.418,
+      "step": 1890
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.9387296438217163,
+      "eval_runtime": 804.3454,
+      "eval_samples_per_second": 18.509,
+      "eval_steps_per_second": 2.314,
+      "step": 1895
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.002416968345642,
+      "learning_rate": 1e-05,
+      "loss": 1.9172,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.938694953918457,
+      "eval_runtime": 804.4582,
+      "eval_samples_per_second": 18.507,
+      "eval_steps_per_second": 2.313,
+      "step": 1900
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.9386274814605713,
+      "eval_runtime": 804.296,
+      "eval_samples_per_second": 18.511,
+      "eval_steps_per_second": 2.314,
+      "step": 1905
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7875113487243652,
+      "learning_rate": 9e-06,
+      "loss": 2.0179,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.93858003616333,
+      "eval_runtime": 767.3044,
+      "eval_samples_per_second": 19.403,
+      "eval_steps_per_second": 2.425,
+      "step": 1910
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.9385550022125244,
+      "eval_runtime": 753.0534,
+      "eval_samples_per_second": 19.77,
+      "eval_steps_per_second": 2.471,
+      "step": 1915
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8589398264884949,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.1198,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.9385381937026978,
+      "eval_runtime": 752.7415,
+      "eval_samples_per_second": 19.778,
+      "eval_steps_per_second": 2.472,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.9385038614273071,
+      "eval_runtime": 751.344,
+      "eval_samples_per_second": 19.815,
+      "eval_steps_per_second": 2.477,
+      "step": 1925
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7445068359375,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.7865,
+      "step": 1930
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 1.938494086265564,
+      "eval_runtime": 754.4106,
+      "eval_samples_per_second": 19.735,
+      "eval_steps_per_second": 2.467,
+      "step": 1930
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9384897947311401,
+      "eval_runtime": 751.5617,
+      "eval_samples_per_second": 19.809,
+      "eval_steps_per_second": 2.476,
+      "step": 1935
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9787230491638184,
+      "learning_rate": 6e-06,
+      "loss": 1.8503,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9384899139404297,
+      "eval_runtime": 762.4984,
+      "eval_samples_per_second": 19.525,
+      "eval_steps_per_second": 2.441,
+      "step": 1940
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9384738206863403,
+      "eval_runtime": 759.6919,
+      "eval_samples_per_second": 19.597,
+      "eval_steps_per_second": 2.45,
+      "step": 1945
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5382779240608215,
+      "learning_rate": 5e-06,
+      "loss": 1.9748,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9384503364562988,
+      "eval_runtime": 765.0189,
+      "eval_samples_per_second": 19.461,
+      "eval_steps_per_second": 2.433,
+      "step": 1950
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9384208917617798,
+      "eval_runtime": 759.892,
+      "eval_samples_per_second": 19.592,
+      "eval_steps_per_second": 2.449,
+      "step": 1955
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.9979357123374939,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.8969,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.938381314277649,
+      "eval_runtime": 779.852,
+      "eval_samples_per_second": 19.091,
+      "eval_steps_per_second": 2.386,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9383498430252075,
+      "eval_runtime": 790.6207,
+      "eval_samples_per_second": 18.831,
+      "eval_steps_per_second": 2.354,
+      "step": 1965
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6065915822982788,
+      "learning_rate": 3e-06,
+      "loss": 1.9022,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9383246898651123,
+      "eval_runtime": 768.8896,
+      "eval_samples_per_second": 19.363,
+      "eval_steps_per_second": 2.42,
+      "step": 1970
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.9383031129837036,
+      "eval_runtime": 777.816,
+      "eval_samples_per_second": 19.141,
+      "eval_steps_per_second": 2.393,
+      "step": 1975
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9139683246612549,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.9378,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.938288927078247,
+      "eval_runtime": 800.6894,
+      "eval_samples_per_second": 18.594,
+      "eval_steps_per_second": 2.324,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.9382776021957397,
+      "eval_runtime": 796.1175,
+      "eval_samples_per_second": 18.701,
+      "eval_steps_per_second": 2.338,
+      "step": 1985
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0108898878097534,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.0676,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.938266396522522,
+      "eval_runtime": 764.489,
+      "eval_samples_per_second": 19.474,
+      "eval_steps_per_second": 2.434,
+      "step": 1990
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.9382603168487549,
+      "eval_runtime": 774.5865,
+      "eval_samples_per_second": 19.221,
+      "eval_steps_per_second": 2.403,
+      "step": 1995
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8565602898597717,
+      "learning_rate": 0.0,
+      "loss": 1.9374,
+      "step": 2000
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 1.9382596015930176,
+      "eval_runtime": 749.4454,
+      "eval_samples_per_second": 19.865,
+      "eval_steps_per_second": 2.483,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 2000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 1244221462020096.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}