diff --git "a/BASE_seed1_10k/trainer_state.json" "b/BASE_seed1_10k/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/BASE_seed1_10k/trainer_state.json"
@@ -0,0 +1,7034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.802776275464215,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0018027762754642149,
+      "grad_norm": 5.41559362411499,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.5387,
+      "step": 10
+    },
+    {
+      "epoch": 0.0036055525509284298,
+      "grad_norm": 7.191960334777832,
+      "learning_rate": 6.333333333333333e-07,
+      "loss": 0.5293,
+      "step": 20
+    },
+    {
+      "epoch": 0.005408328826392645,
+      "grad_norm": 4.474244117736816,
+      "learning_rate": 9.666666666666668e-07,
+      "loss": 0.4781,
+      "step": 30
+    },
+    {
+      "epoch": 0.0072111051018568595,
+      "grad_norm": 3.841322183609009,
+      "learning_rate": 1.3e-06,
+      "loss": 0.4167,
+      "step": 40
+    },
+    {
+      "epoch": 0.009013881377321075,
+      "grad_norm": 1.6013221740722656,
+      "learning_rate": 1.6333333333333333e-06,
+      "loss": 0.2892,
+      "step": 50
+    },
+    {
+      "epoch": 0.01081665765278529,
+      "grad_norm": 0.917481005191803,
+      "learning_rate": 1.9666666666666668e-06,
+      "loss": 0.2024,
+      "step": 60
+    },
+    {
+      "epoch": 0.012619433928249504,
+      "grad_norm": 1.1382007598876953,
+      "learning_rate": 2.3e-06,
+      "loss": 0.1743,
+      "step": 70
+    },
+    {
+      "epoch": 0.014422210203713719,
+      "grad_norm": 0.7447443008422852,
+      "learning_rate": 2.6333333333333337e-06,
+      "loss": 0.1568,
+      "step": 80
+    },
+    {
+      "epoch": 0.016224986479177934,
+      "grad_norm": 0.6980589628219604,
+      "learning_rate": 2.966666666666667e-06,
+      "loss": 0.151,
+      "step": 90
+    },
+    {
+      "epoch": 0.01802776275464215,
+      "grad_norm": 0.39905884861946106,
+      "learning_rate": 3.3e-06,
+      "loss": 0.1492,
+      "step": 100
+    },
+    {
+      "epoch": 0.019830539030106363,
+      "grad_norm": 0.4256199598312378,
+      "learning_rate": 3.633333333333334e-06,
+      "loss": 0.1458,
+      "step": 110
+    },
+    {
+      "epoch": 0.02163331530557058,
+      "grad_norm": 0.4188803732395172,
+      "learning_rate": 3.966666666666667e-06,
+      "loss": 0.1271,
+      "step": 120
+    },
+    {
+      "epoch": 0.023436091581034792,
+      "grad_norm": 0.36602726578712463,
+      "learning_rate": 4.2999999999999995e-06,
+      "loss": 0.1223,
+      "step": 130
+    },
+    {
+      "epoch": 0.02523886785649901,
+      "grad_norm": 0.4918166399002075,
+      "learning_rate": 4.633333333333334e-06,
+      "loss": 0.126,
+      "step": 140
+    },
+    {
+      "epoch": 0.02704164413196322,
+      "grad_norm": 0.6279438138008118,
+      "learning_rate": 4.966666666666667e-06,
+      "loss": 0.1206,
+      "step": 150
+    },
+    {
+      "epoch": 0.028844420407427438,
+      "grad_norm": 0.569150984287262,
+      "learning_rate": 5.3e-06,
+      "loss": 0.1107,
+      "step": 160
+    },
+    {
+      "epoch": 0.030647196682891654,
+      "grad_norm": 0.625974714756012,
+      "learning_rate": 5.633333333333333e-06,
+      "loss": 0.1071,
+      "step": 170
+    },
+    {
+      "epoch": 0.03244997295835587,
+      "grad_norm": 0.6144049167633057,
+      "learning_rate": 5.9666666666666666e-06,
+      "loss": 0.1111,
+      "step": 180
+    },
+    {
+      "epoch": 0.03425274923382008,
+      "grad_norm": 0.6697492599487305,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.1167,
+      "step": 190
+    },
+    {
+      "epoch": 0.0360555255092843,
+      "grad_norm": 0.2732592821121216,
+      "learning_rate": 6.633333333333333e-06,
+      "loss": 0.1049,
+      "step": 200
+    },
+    {
+      "epoch": 0.03785830178474851,
+      "grad_norm": 0.6220976114273071,
+      "learning_rate": 6.966666666666667e-06,
+      "loss": 0.1029,
+      "step": 210
+    },
+    {
+      "epoch": 0.039661078060212726,
+      "grad_norm": 0.3747783303260803,
+      "learning_rate": 7.2999999999999996e-06,
+      "loss": 0.0994,
+      "step": 220
+    },
+    {
+      "epoch": 0.041463854335676946,
+      "grad_norm": 0.7402628064155579,
+      "learning_rate": 7.633333333333334e-06,
+      "loss": 0.0955,
+      "step": 230
+    },
+    {
+      "epoch": 0.04326663061114116,
+      "grad_norm": 0.3625622093677521,
+      "learning_rate": 7.966666666666666e-06,
+      "loss": 0.0948,
+      "step": 240
+    },
+    {
+      "epoch": 0.04506940688660537,
+      "grad_norm": 0.33918800950050354,
+      "learning_rate": 8.3e-06,
+      "loss": 0.0887,
+      "step": 250
+    },
+    {
+      "epoch": 0.046872183162069585,
+      "grad_norm": 0.42553257942199707,
+      "learning_rate": 8.633333333333334e-06,
+      "loss": 0.0915,
+      "step": 260
+    },
+    {
+      "epoch": 0.048674959437533805,
+      "grad_norm": 0.7103670835494995,
+      "learning_rate": 8.966666666666668e-06,
+      "loss": 0.0975,
+      "step": 270
+    },
+    {
+      "epoch": 0.05047773571299802,
+      "grad_norm": 0.47956153750419617,
+      "learning_rate": 9.3e-06,
+      "loss": 0.0921,
+      "step": 280
+    },
+    {
+      "epoch": 0.05228051198846223,
+      "grad_norm": 0.52305668592453,
+      "learning_rate": 9.633333333333335e-06,
+      "loss": 0.0956,
+      "step": 290
+    },
+    {
+      "epoch": 0.05408328826392644,
+      "grad_norm": 0.294786274433136,
+      "learning_rate": 9.966666666666667e-06,
+      "loss": 0.0843,
+      "step": 300
+    },
+    {
+      "epoch": 0.05588606453939066,
+      "grad_norm": 0.7143281102180481,
+      "learning_rate": 1.03e-05,
+      "loss": 0.0909,
+      "step": 310
+    },
+    {
+      "epoch": 0.057688840814854876,
+      "grad_norm": 0.37210068106651306,
+      "learning_rate": 1.0633333333333334e-05,
+      "loss": 0.0908,
+      "step": 320
+    },
+    {
+      "epoch": 0.05949161709031909,
+      "grad_norm": 0.3835732936859131,
+      "learning_rate": 1.0966666666666666e-05,
+      "loss": 0.0876,
+      "step": 330
+    },
+    {
+      "epoch": 0.06129439336578331,
+      "grad_norm": 0.4630444347858429,
+      "learning_rate": 1.13e-05,
+      "loss": 0.0941,
+      "step": 340
+    },
+    {
+      "epoch": 0.06309716964124752,
+      "grad_norm": 0.2766304612159729,
+      "learning_rate": 1.1633333333333334e-05,
+      "loss": 0.093,
+      "step": 350
+    },
+    {
+      "epoch": 0.06489994591671173,
+      "grad_norm": 0.5275501012802124,
+      "learning_rate": 1.1966666666666668e-05,
+      "loss": 0.089,
+      "step": 360
+    },
+    {
+      "epoch": 0.06670272219217595,
+      "grad_norm": 0.37696143984794617,
+      "learning_rate": 1.23e-05,
+      "loss": 0.0896,
+      "step": 370
+    },
+    {
+      "epoch": 0.06850549846764016,
+      "grad_norm": 0.32620155811309814,
+      "learning_rate": 1.2633333333333333e-05,
+      "loss": 0.0813,
+      "step": 380
+    },
+    {
+      "epoch": 0.07030827474310439,
+      "grad_norm": 0.4478512704372406,
+      "learning_rate": 1.2966666666666669e-05,
+      "loss": 0.0839,
+      "step": 390
+    },
+    {
+      "epoch": 0.0721110510185686,
+      "grad_norm": 0.37261882424354553,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 0.0837,
+      "step": 400
+    },
+    {
+      "epoch": 0.07391382729403281,
+      "grad_norm": 0.7236140370368958,
+      "learning_rate": 1.3633333333333334e-05,
+      "loss": 0.0822,
+      "step": 410
+    },
+    {
+      "epoch": 0.07571660356949703,
+      "grad_norm": 0.45200467109680176,
+      "learning_rate": 1.3966666666666666e-05,
+      "loss": 0.0783,
+      "step": 420
+    },
+    {
+      "epoch": 0.07751937984496124,
+      "grad_norm": 0.4210900664329529,
+      "learning_rate": 1.43e-05,
+      "loss": 0.0777,
+      "step": 430
+    },
+    {
+      "epoch": 0.07932215612042545,
+      "grad_norm": 0.2809159755706787,
+      "learning_rate": 1.4633333333333334e-05,
+      "loss": 0.0773,
+      "step": 440
+    },
+    {
+      "epoch": 0.08112493239588967,
+      "grad_norm": 0.37136387825012207,
+      "learning_rate": 1.4966666666666668e-05,
+      "loss": 0.0851,
+      "step": 450
+    },
+    {
+      "epoch": 0.08292770867135389,
+      "grad_norm": 0.5091298818588257,
+      "learning_rate": 1.53e-05,
+      "loss": 0.0794,
+      "step": 460
+    },
+    {
+      "epoch": 0.0847304849468181,
+      "grad_norm": 0.3478129506111145,
+      "learning_rate": 1.563333333333333e-05,
+      "loss": 0.0835,
+      "step": 470
+    },
+    {
+      "epoch": 0.08653326122228232,
+      "grad_norm": 0.38393208384513855,
+      "learning_rate": 1.5966666666666667e-05,
+      "loss": 0.0792,
+      "step": 480
+    },
+    {
+      "epoch": 0.08833603749774653,
+      "grad_norm": 0.32405346632003784,
+      "learning_rate": 1.63e-05,
+      "loss": 0.0827,
+      "step": 490
+    },
+    {
+      "epoch": 0.09013881377321074,
+      "grad_norm": 0.5071119666099548,
+      "learning_rate": 1.6633333333333336e-05,
+      "loss": 0.077,
+      "step": 500
+    },
+    {
+      "epoch": 0.09194159004867496,
+      "grad_norm": 0.6446972489356995,
+      "learning_rate": 1.6966666666666668e-05,
+      "loss": 0.0808,
+      "step": 510
+    },
+    {
+      "epoch": 0.09374436632413917,
+      "grad_norm": 0.5842817425727844,
+      "learning_rate": 1.73e-05,
+      "loss": 0.0766,
+      "step": 520
+    },
+    {
+      "epoch": 0.09554714259960338,
+      "grad_norm": 0.4594644606113434,
+      "learning_rate": 1.7633333333333336e-05,
+      "loss": 0.0827,
+      "step": 530
+    },
+    {
+      "epoch": 0.09734991887506761,
+      "grad_norm": 0.3079770803451538,
+      "learning_rate": 1.796666666666667e-05,
+      "loss": 0.0786,
+      "step": 540
+    },
+    {
+      "epoch": 0.09915269515053182,
+      "grad_norm": 0.47132161259651184,
+      "learning_rate": 1.83e-05,
+      "loss": 0.08,
+      "step": 550
+    },
+    {
+      "epoch": 0.10095547142599604,
+      "grad_norm": 0.3440473973751068,
+      "learning_rate": 1.8633333333333333e-05,
+      "loss": 0.0805,
+      "step": 560
+    },
+    {
+      "epoch": 0.10275824770146025,
+      "grad_norm": 0.2881501019001007,
+      "learning_rate": 1.896666666666667e-05,
+      "loss": 0.0756,
+      "step": 570
+    },
+    {
+      "epoch": 0.10456102397692446,
+      "grad_norm": 0.38147321343421936,
+      "learning_rate": 1.93e-05,
+      "loss": 0.0753,
+      "step": 580
+    },
+    {
+      "epoch": 0.10636380025238867,
+      "grad_norm": 0.3498924970626831,
+      "learning_rate": 1.9633333333333334e-05,
+      "loss": 0.0779,
+      "step": 590
+    },
+    {
+      "epoch": 0.10816657652785289,
+      "grad_norm": 0.3204267919063568,
+      "learning_rate": 1.9966666666666666e-05,
+      "loss": 0.073,
+      "step": 600
+    },
+    {
+      "epoch": 0.10996935280331711,
+      "grad_norm": 0.3183678090572357,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 0.0724,
+      "step": 610
+    },
+    {
+      "epoch": 0.11177212907878133,
+      "grad_norm": 0.3010425269603729,
+      "learning_rate": 2.0633333333333335e-05,
+      "loss": 0.0725,
+      "step": 620
+    },
+    {
+      "epoch": 0.11357490535424554,
+      "grad_norm": 0.3266223669052124,
+      "learning_rate": 2.0966666666666667e-05,
+      "loss": 0.0796,
+      "step": 630
+    },
+    {
+      "epoch": 0.11537768162970975,
+      "grad_norm": 0.3957923948764801,
+      "learning_rate": 2.13e-05,
+      "loss": 0.0764,
+      "step": 640
+    },
+    {
+      "epoch": 0.11718045790517397,
+      "grad_norm": 0.36882466077804565,
+      "learning_rate": 2.1633333333333332e-05,
+      "loss": 0.0743,
+      "step": 650
+    },
+    {
+      "epoch": 0.11898323418063818,
+      "grad_norm": 0.40890058875083923,
+      "learning_rate": 2.1966666666666668e-05,
+      "loss": 0.078,
+      "step": 660
+    },
+    {
+      "epoch": 0.12078601045610239,
+      "grad_norm": 0.3890920579433441,
+      "learning_rate": 2.23e-05,
+      "loss": 0.0681,
+      "step": 670
+    },
+    {
+      "epoch": 0.12258878673156662,
+      "grad_norm": 0.3976999521255493,
+      "learning_rate": 2.2633333333333336e-05,
+      "loss": 0.0747,
+      "step": 680
+    },
+    {
+      "epoch": 0.12439156300703083,
+      "grad_norm": 0.3453735113143921,
+      "learning_rate": 2.2966666666666668e-05,
+      "loss": 0.079,
+      "step": 690
+    },
+    {
+      "epoch": 0.12619433928249504,
+      "grad_norm": 0.3744434714317322,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 0.0769,
+      "step": 700
+    },
+    {
+      "epoch": 0.12799711555795926,
+      "grad_norm": 0.4043666422367096,
+      "learning_rate": 2.3633333333333336e-05,
+      "loss": 0.0721,
+      "step": 710
+    },
+    {
+      "epoch": 0.12979989183342347,
+      "grad_norm": 0.452412873506546,
+      "learning_rate": 2.396666666666667e-05,
+      "loss": 0.076,
+      "step": 720
+    },
+    {
+      "epoch": 0.13160266810888768,
+      "grad_norm": 0.3486067056655884,
+      "learning_rate": 2.43e-05,
+      "loss": 0.0742,
+      "step": 730
+    },
+    {
+      "epoch": 0.1334054443843519,
+      "grad_norm": 0.3778965473175049,
+      "learning_rate": 2.4633333333333334e-05,
+      "loss": 0.0727,
+      "step": 740
+    },
+    {
+      "epoch": 0.1352082206598161,
+      "grad_norm": 0.2709180414676666,
+      "learning_rate": 2.496666666666667e-05,
+      "loss": 0.0684,
+      "step": 750
+    },
+    {
+      "epoch": 0.13701099693528032,
+      "grad_norm": 0.32746025919914246,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 0.0735,
+      "step": 760
+    },
+    {
+      "epoch": 0.13881377321074453,
+      "grad_norm": 0.3406435549259186,
+      "learning_rate": 2.5633333333333338e-05,
+      "loss": 0.0747,
+      "step": 770
+    },
+    {
+      "epoch": 0.14061654948620878,
+      "grad_norm": 0.39069485664367676,
+      "learning_rate": 2.5966666666666667e-05,
+      "loss": 0.0638,
+      "step": 780
+    },
+    {
+      "epoch": 0.142419325761673,
+      "grad_norm": 0.3011362552642822,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 0.0709,
+      "step": 790
+    },
+    {
+      "epoch": 0.1442221020371372,
+      "grad_norm": 0.3271448612213135,
+      "learning_rate": 2.663333333333333e-05,
+      "loss": 0.0684,
+      "step": 800
+    },
+    {
+      "epoch": 0.1460248783126014,
+      "grad_norm": 0.3503864109516144,
+      "learning_rate": 2.6966666666666667e-05,
+      "loss": 0.078,
+      "step": 810
+    },
+    {
+      "epoch": 0.14782765458806563,
+      "grad_norm": 0.2866891622543335,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 0.0671,
+      "step": 820
+    },
+    {
+      "epoch": 0.14963043086352984,
+      "grad_norm": 0.36019957065582275,
+      "learning_rate": 2.7633333333333332e-05,
+      "loss": 0.0673,
+      "step": 830
+    },
+    {
+      "epoch": 0.15143320713899405,
+      "grad_norm": 0.34380584955215454,
+      "learning_rate": 2.7966666666666668e-05,
+      "loss": 0.069,
+      "step": 840
+    },
+    {
+      "epoch": 0.15323598341445827,
+      "grad_norm": 0.2428215742111206,
+      "learning_rate": 2.83e-05,
+      "loss": 0.0733,
+      "step": 850
+    },
+    {
+      "epoch": 0.15503875968992248,
+      "grad_norm": 0.29352885484695435,
+      "learning_rate": 2.8633333333333336e-05,
+      "loss": 0.0774,
+      "step": 860
+    },
+    {
+      "epoch": 0.1568415359653867,
+      "grad_norm": 0.4597095251083374,
+      "learning_rate": 2.8966666666666668e-05,
+      "loss": 0.0765,
+      "step": 870
+    },
+    {
+      "epoch": 0.1586443122408509,
+      "grad_norm": 0.31262949109077454,
+      "learning_rate": 2.93e-05,
+      "loss": 0.0664,
+      "step": 880
+    },
+    {
+      "epoch": 0.16044708851631512,
+      "grad_norm": 0.35274234414100647,
+      "learning_rate": 2.9633333333333336e-05,
+      "loss": 0.0662,
+      "step": 890
+    },
+    {
+      "epoch": 0.16224986479177933,
+      "grad_norm": 0.21876050531864166,
+      "learning_rate": 2.9966666666666672e-05,
+      "loss": 0.0641,
+      "step": 900
+    },
+    {
+      "epoch": 0.16405264106724354,
+      "grad_norm": 0.2677417993545532,
+      "learning_rate": 3.03e-05,
+      "loss": 0.0631,
+      "step": 910
+    },
+    {
+      "epoch": 0.16585541734270778,
+      "grad_norm": 0.25009411573410034,
+      "learning_rate": 3.063333333333334e-05,
+      "loss": 0.0728,
+      "step": 920
+    },
+    {
+      "epoch": 0.167658193618172,
+      "grad_norm": 0.2829286456108093,
+      "learning_rate": 3.096666666666666e-05,
+      "loss": 0.0638,
+      "step": 930
+    },
+    {
+      "epoch": 0.1694609698936362,
+      "grad_norm": 0.24688038229942322,
+      "learning_rate": 3.13e-05,
+      "loss": 0.0691,
+      "step": 940
+    },
+    {
+      "epoch": 0.17126374616910042,
+      "grad_norm": 0.34949207305908203,
+      "learning_rate": 3.1633333333333334e-05,
+      "loss": 0.0676,
+      "step": 950
+    },
+    {
+      "epoch": 0.17306652244456464,
+      "grad_norm": 0.2864921987056732,
+      "learning_rate": 3.196666666666667e-05,
+      "loss": 0.0743,
+      "step": 960
+    },
+    {
+      "epoch": 0.17486929872002885,
+      "grad_norm": 0.2108231633901596,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 0.0691,
+      "step": 970
+    },
+    {
+      "epoch": 0.17667207499549306,
+      "grad_norm": 0.30339616537094116,
+      "learning_rate": 3.263333333333333e-05,
+      "loss": 0.0731,
+      "step": 980
+    },
+    {
+      "epoch": 0.17847485127095727,
+      "grad_norm": 0.3775007128715515,
+      "learning_rate": 3.296666666666667e-05,
+      "loss": 0.0694,
+      "step": 990
+    },
+    {
+      "epoch": 0.1802776275464215,
+      "grad_norm": 0.3024364411830902,
+      "learning_rate": 3.33e-05,
+      "loss": 0.0665,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1820804038218857,
+      "grad_norm": 0.2513657212257385,
+      "learning_rate": 3.3633333333333335e-05,
+      "loss": 0.0676,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1838831800973499,
+      "grad_norm": 0.27736595273017883,
+      "learning_rate": 3.396666666666667e-05,
+      "loss": 0.0684,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18568595637281413,
+      "grad_norm": 0.4237976372241974,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 0.064,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18748873264827834,
+      "grad_norm": 0.38221675157546997,
+      "learning_rate": 3.463333333333333e-05,
+      "loss": 0.06,
+      "step": 1040
+    },
+    {
+      "epoch": 0.18929150892374255,
+      "grad_norm": 0.31958162784576416,
+      "learning_rate": 3.496666666666667e-05,
+      "loss": 0.0706,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19109428519920676,
+      "grad_norm": 0.33839792013168335,
+      "learning_rate": 3.53e-05,
+      "loss": 0.0676,
+      "step": 1060
+    },
+    {
+      "epoch": 0.192897061474671,
+      "grad_norm": 0.40528759360313416,
+      "learning_rate": 3.563333333333334e-05,
+      "loss": 0.0694,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19469983775013522,
+      "grad_norm": 0.19379711151123047,
+      "learning_rate": 3.596666666666667e-05,
+      "loss": 0.0613,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19650261402559943,
+      "grad_norm": 0.2642883062362671,
+      "learning_rate": 3.63e-05,
+      "loss": 0.0665,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19830539030106364,
+      "grad_norm": 0.23124559223651886,
+      "learning_rate": 3.6633333333333334e-05,
+      "loss": 0.0645,
+      "step": 1100
+    },
+    {
+      "epoch": 0.20010816657652786,
+      "grad_norm": 0.3091222643852234,
+      "learning_rate": 3.6966666666666666e-05,
+      "loss": 0.0648,
+      "step": 1110
+    },
+    {
+      "epoch": 0.20191094285199207,
+      "grad_norm": 0.3477613031864166,
+      "learning_rate": 3.73e-05,
+      "loss": 0.0628,
+      "step": 1120
+    },
+    {
+      "epoch": 0.20371371912745628,
+      "grad_norm": 0.22648930549621582,
+      "learning_rate": 3.763333333333334e-05,
+      "loss": 0.0612,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2055164954029205,
+      "grad_norm": 0.21195665001869202,
+      "learning_rate": 3.796666666666667e-05,
+      "loss": 0.0668,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2073192716783847,
+      "grad_norm": 0.2780223786830902,
+      "learning_rate": 3.83e-05,
+      "loss": 0.0679,
+      "step": 1150
+    },
+    {
+      "epoch": 0.20912204795384892,
+      "grad_norm": 0.2918558418750763,
+      "learning_rate": 3.8633333333333335e-05,
+      "loss": 0.0673,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21092482422931313,
+      "grad_norm": 0.2419963777065277,
+      "learning_rate": 3.896666666666667e-05,
+      "loss": 0.0708,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21272760050477735,
+      "grad_norm": 0.26872798800468445,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 0.065,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21453037678024156,
+      "grad_norm": 0.25414329767227173,
+      "learning_rate": 3.963333333333333e-05,
+      "loss": 0.0622,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21633315305570577,
+      "grad_norm": 0.25174248218536377,
+      "learning_rate": 3.996666666666667e-05,
+      "loss": 0.0695,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21813592933117001,
+      "grad_norm": 0.22807759046554565,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 0.0701,
+      "step": 1210
+    },
+    {
+      "epoch": 0.21993870560663423,
+      "grad_norm": 0.29292958974838257,
+      "learning_rate": 4.0633333333333336e-05,
+      "loss": 0.0632,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22174148188209844,
+      "grad_norm": 0.2155628800392151,
+      "learning_rate": 4.096666666666667e-05,
+      "loss": 0.0682,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22354425815756265,
+      "grad_norm": 0.20743617415428162,
+      "learning_rate": 4.13e-05,
+      "loss": 0.0641,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22534703443302687,
+      "grad_norm": 0.20022955536842346,
+      "learning_rate": 4.1633333333333333e-05,
+      "loss": 0.0579,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22714981070849108,
+      "grad_norm": 0.3666442334651947,
+      "learning_rate": 4.196666666666667e-05,
+      "loss": 0.0611,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2289525869839553,
+      "grad_norm": 0.29185906052589417,
+      "learning_rate": 4.23e-05,
+      "loss": 0.0641,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2307553632594195,
+      "grad_norm": 0.4205268323421478,
+      "learning_rate": 4.263333333333334e-05,
+      "loss": 0.0666,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23255813953488372,
+      "grad_norm": 0.2673855125904083,
+      "learning_rate": 4.296666666666666e-05,
+      "loss": 0.064,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23436091581034793,
+      "grad_norm": 0.31742236018180847,
+      "learning_rate": 4.33e-05,
+      "loss": 0.0691,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23616369208581214,
+      "grad_norm": 0.2533666789531708,
+      "learning_rate": 4.3633333333333335e-05,
+      "loss": 0.0641,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23796646836127636,
+      "grad_norm": 0.2713998258113861,
+      "learning_rate": 4.396666666666667e-05,
+      "loss": 0.0633,
+      "step": 1320
+    },
+    {
+      "epoch": 0.23976924463674057,
+      "grad_norm": 0.3495473861694336,
+      "learning_rate": 4.43e-05,
+      "loss": 0.061,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24157202091220478,
+      "grad_norm": 0.22186997532844543,
+      "learning_rate": 4.463333333333334e-05,
+      "loss": 0.063,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24337479718766902,
+      "grad_norm": 0.2712762951850891,
+      "learning_rate": 4.496666666666667e-05,
+      "loss": 0.0605,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24517757346313324,
+      "grad_norm": 0.2828688323497772,
+      "learning_rate": 4.53e-05,
+      "loss": 0.0651,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24698034973859745,
+      "grad_norm": 0.32463574409484863,
+      "learning_rate": 4.5633333333333336e-05,
+      "loss": 0.0637,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24878312601406166,
+      "grad_norm": 0.27668318152427673,
+      "learning_rate": 4.596666666666667e-05,
+      "loss": 0.0568,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2505859022895259,
+      "grad_norm": 0.30997949838638306,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 0.0686,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2523886785649901,
+      "grad_norm": 0.2110419124364853,
+      "learning_rate": 4.663333333333333e-05,
+      "loss": 0.0575,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2541914548404543,
+      "grad_norm": 0.2404826134443283,
+      "learning_rate": 4.696666666666667e-05,
+      "loss": 0.0655,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2559942311159185,
+      "grad_norm": 0.23162831366062164,
+      "learning_rate": 4.73e-05,
+      "loss": 0.0624,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2577970073913827,
+      "grad_norm": 0.23692749440670013,
+      "learning_rate": 4.763333333333334e-05,
+      "loss": 0.0613,
+      "step": 1430
+    },
+    {
+      "epoch": 0.25959978366684694,
+      "grad_norm": 0.2759644687175751,
+      "learning_rate": 4.796666666666667e-05,
+      "loss": 0.0654,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26140255994231115,
+      "grad_norm": 0.23253923654556274,
+      "learning_rate": 4.83e-05,
+      "loss": 0.068,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26320533621777537,
+      "grad_norm": 0.26997312903404236,
+      "learning_rate": 4.8633333333333334e-05,
+      "loss": 0.0595,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2650081124932396,
+      "grad_norm": 0.40077781677246094,
+      "learning_rate": 4.8966666666666667e-05,
+      "loss": 0.0622,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2668108887687038,
+      "grad_norm": 0.2376328855752945,
+      "learning_rate": 4.93e-05,
+      "loss": 0.0636,
+      "step": 1480
+    },
+    {
+      "epoch": 0.268613665044168,
+      "grad_norm": 0.2827739715576172,
+      "learning_rate": 4.963333333333334e-05,
+      "loss": 0.0604,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2704164413196322,
+      "grad_norm": 0.4532570540904999,
+      "learning_rate": 4.996666666666667e-05,
+      "loss": 0.0587,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27221921759509643,
+      "grad_norm": 0.43316343426704407,
+      "learning_rate": 5.03e-05,
+      "loss": 0.0641,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27402199387056064,
+      "grad_norm": 0.3506268262863159,
+      "learning_rate": 5.0633333333333335e-05,
+      "loss": 0.0662,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27582477014602486,
+      "grad_norm": 0.24790820479393005,
+      "learning_rate": 5.0966666666666674e-05,
+      "loss": 0.0646,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27762754642148907,
+      "grad_norm": 0.2507699131965637,
+      "learning_rate": 5.130000000000001e-05,
+      "loss": 0.0596,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2794303226969533,
+      "grad_norm": 0.26693329215049744,
+      "learning_rate": 5.163333333333333e-05,
+      "loss": 0.0687,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28123309897241755,
+      "grad_norm": 0.25990283489227295,
+      "learning_rate": 5.196666666666667e-05,
+      "loss": 0.0618,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28303587524788176,
+      "grad_norm": 0.20986329019069672,
+      "learning_rate": 5.2300000000000004e-05,
+      "loss": 0.0631,
+      "step": 1570
+    },
+    {
+      "epoch": 0.284838651523346,
+      "grad_norm": 0.3476528227329254,
+      "learning_rate": 5.2633333333333336e-05,
+      "loss": 0.0577,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2866414277988102,
+      "grad_norm": 0.2593328654766083,
+      "learning_rate": 5.296666666666666e-05,
+      "loss": 0.0605,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2884442040742744,
+      "grad_norm": 0.3481169044971466,
+      "learning_rate": 5.330000000000001e-05,
+      "loss": 0.0606,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2902469803497386,
+      "grad_norm": 0.21679189801216125,
+      "learning_rate": 5.3633333333333334e-05,
+      "loss": 0.0671,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2920497566252028,
+      "grad_norm": 0.2372356802225113,
+      "learning_rate": 5.3966666666666666e-05,
+      "loss": 0.0613,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29385253290066704,
+      "grad_norm": 0.25489747524261475,
+      "learning_rate": 5.4300000000000005e-05,
+      "loss": 0.0601,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29565530917613125,
+      "grad_norm": 0.25877898931503296,
+      "learning_rate": 5.463333333333334e-05,
+      "loss": 0.0633,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29745808545159547,
+      "grad_norm": 0.22300730645656586,
+      "learning_rate": 5.496666666666666e-05,
+      "loss": 0.0642,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2992608617270597,
+      "grad_norm": 0.18823741376399994,
+      "learning_rate": 5.530000000000001e-05,
+      "loss": 0.0584,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3010636380025239,
+      "grad_norm": 0.26364418864250183,
+      "learning_rate": 5.5633333333333335e-05,
+      "loss": 0.0586,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3028664142779881,
+      "grad_norm": 0.2923857271671295,
+      "learning_rate": 5.596666666666667e-05,
+      "loss": 0.0647,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3046691905534523,
+      "grad_norm": 0.2227906584739685,
+      "learning_rate": 5.63e-05,
+      "loss": 0.0625,
+      "step": 1690
+    },
+    {
+      "epoch": 0.30647196682891653,
+      "grad_norm": 0.21054530143737793,
+      "learning_rate": 5.663333333333334e-05,
+      "loss": 0.065,
+      "step": 1700
+    },
+    {
+      "epoch": 0.30827474310438074,
+      "grad_norm": 0.348690390586853,
+      "learning_rate": 5.696666666666667e-05,
+      "loss": 0.0632,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31007751937984496,
+      "grad_norm": 0.20512254536151886,
+      "learning_rate": 5.73e-05,
+      "loss": 0.0643,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31188029565530917,
+      "grad_norm": 0.19749906659126282,
+      "learning_rate": 5.7633333333333336e-05,
+      "loss": 0.0582,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3136830719307734,
+      "grad_norm": 0.2500685155391693,
+      "learning_rate": 5.796666666666667e-05,
+      "loss": 0.0572,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3154858482062376,
+      "grad_norm": 0.20494727790355682,
+      "learning_rate": 5.83e-05,
+      "loss": 0.0618,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3172886244817018,
+      "grad_norm": 0.28706008195877075,
+      "learning_rate": 5.863333333333334e-05,
+      "loss": 0.0595,
+      "step": 1760
+    },
+    {
+      "epoch": 0.319091400757166,
+      "grad_norm": 0.23980054259300232,
+      "learning_rate": 5.896666666666667e-05,
+      "loss": 0.0601,
+      "step": 1770
+    },
+    {
+      "epoch": 0.32089417703263023,
+      "grad_norm": 0.26302099227905273,
+      "learning_rate": 5.93e-05,
+      "loss": 0.0614,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32269695330809445,
+      "grad_norm": 0.34201040863990784,
+      "learning_rate": 5.9633333333333344e-05,
+      "loss": 0.0676,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32449972958355866,
+      "grad_norm": 0.22230128943920135,
+      "learning_rate": 5.996666666666667e-05,
+      "loss": 0.0624,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3263025058590229,
+      "grad_norm": 0.25414663553237915,
+      "learning_rate": 6.03e-05,
+      "loss": 0.0573,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3281052821344871,
+      "grad_norm": 0.20873858034610748,
+      "learning_rate": 6.063333333333333e-05,
+      "loss": 0.058,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3299080584099513,
+      "grad_norm": 0.25061991810798645,
+      "learning_rate": 6.0966666666666674e-05,
+      "loss": 0.0612,
+      "step": 1830
+    },
+    {
+      "epoch": 0.33171083468541557,
+      "grad_norm": 0.24576738476753235,
+      "learning_rate": 6.13e-05,
+      "loss": 0.0635,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3335136109608798,
+      "grad_norm": 0.41412773728370667,
+      "learning_rate": 6.163333333333333e-05,
+      "loss": 0.0635,
+      "step": 1850
+    },
+    {
+      "epoch": 0.335316387236344,
+      "grad_norm": 0.2621055543422699,
+      "learning_rate": 6.196666666666668e-05,
+      "loss": 0.0573,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3371191635118082,
+      "grad_norm": 0.36470827460289,
+      "learning_rate": 6.23e-05,
+      "loss": 0.0597,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3389219397872724,
+      "grad_norm": 0.21115829050540924,
+      "learning_rate": 6.263333333333333e-05,
+      "loss": 0.0562,
+      "step": 1880
+    },
+    {
+      "epoch": 0.34072471606273663,
+      "grad_norm": 0.20064254105091095,
+      "learning_rate": 6.296666666666667e-05,
+      "loss": 0.0596,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34252749233820085,
+      "grad_norm": 0.30089330673217773,
+      "learning_rate": 6.330000000000001e-05,
+      "loss": 0.0574,
+      "step": 1900
+    },
+    {
+      "epoch": 0.34433026861366506,
+      "grad_norm": 0.26324665546417236,
+      "learning_rate": 6.363333333333334e-05,
+      "loss": 0.0628,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34613304488912927,
+      "grad_norm": 0.3654855489730835,
+      "learning_rate": 6.396666666666667e-05,
+      "loss": 0.0603,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3479358211645935,
+      "grad_norm": 0.17230789363384247,
+      "learning_rate": 6.43e-05,
+      "loss": 0.0581,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3497385974400577,
+      "grad_norm": 0.39206016063690186,
+      "learning_rate": 6.463333333333334e-05,
+      "loss": 0.0584,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3515413737155219,
+      "grad_norm": 0.30950891971588135,
+      "learning_rate": 6.496666666666667e-05,
+      "loss": 0.0622,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3533441499909861,
+      "grad_norm": 0.30163729190826416,
+      "learning_rate": 6.53e-05,
+      "loss": 0.0628,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35514692626645034,
+      "grad_norm": 0.3109796345233917,
+      "learning_rate": 6.563333333333333e-05,
+      "loss": 0.0612,
+      "step": 1970
+    },
+    {
+      "epoch": 0.35694970254191455,
+      "grad_norm": 0.3278733491897583,
+      "learning_rate": 6.596666666666667e-05,
+      "loss": 0.0563,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35875247881737876,
+      "grad_norm": 0.33346840739250183,
+      "learning_rate": 6.630000000000001e-05,
+      "loss": 0.06,
+      "step": 1990
+    },
+    {
+      "epoch": 0.360555255092843,
+      "grad_norm": 0.23046278953552246,
+      "learning_rate": 6.663333333333333e-05,
+      "loss": 0.0579,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3623580313683072,
+      "grad_norm": 0.20575708150863647,
+      "learning_rate": 6.696666666666666e-05,
+      "loss": 0.06,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3641608076437714,
+      "grad_norm": 0.2498203068971634,
+      "learning_rate": 6.730000000000001e-05,
+      "loss": 0.0576,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3659635839192356,
+      "grad_norm": 0.3085964322090149,
+      "learning_rate": 6.763333333333334e-05,
+      "loss": 0.0608,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3677663601946998,
+      "grad_norm": 0.30336859822273254,
+      "learning_rate": 6.796666666666666e-05,
+      "loss": 0.0641,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36956913647016404,
+      "grad_norm": 0.21763910353183746,
+      "learning_rate": 6.83e-05,
+      "loss": 0.0565,
+      "step": 2050
+    },
+    {
+      "epoch": 0.37137191274562825,
+      "grad_norm": 0.23763242363929749,
+      "learning_rate": 6.863333333333334e-05,
+      "loss": 0.0569,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37317468902109246,
+      "grad_norm": 0.24654164910316467,
+      "learning_rate": 6.896666666666667e-05,
+      "loss": 0.0588,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3749774652965567,
+      "grad_norm": 0.2766386568546295,
+      "learning_rate": 6.93e-05,
+      "loss": 0.0598,
+      "step": 2080
+    },
+    {
+      "epoch": 0.3767802415720209,
+      "grad_norm": 0.2941744327545166,
+      "learning_rate": 6.963333333333334e-05,
+      "loss": 0.0654,
+      "step": 2090
+    },
+    {
+      "epoch": 0.3785830178474851,
+      "grad_norm": 0.20436793565750122,
+      "learning_rate": 6.996666666666667e-05,
+      "loss": 0.0611,
+      "step": 2100
+    },
+    {
+      "epoch": 0.3803857941229493,
+      "grad_norm": 0.2097810059785843,
+      "learning_rate": 7.03e-05,
+      "loss": 0.0582,
+      "step": 2110
+    },
+    {
+      "epoch": 0.38218857039841353,
+      "grad_norm": 0.19502535462379456,
+      "learning_rate": 7.063333333333333e-05,
+      "loss": 0.057,
+      "step": 2120
+    },
+    {
+      "epoch": 0.3839913466738778,
+      "grad_norm": 0.20433445274829865,
+      "learning_rate": 7.096666666666667e-05,
+      "loss": 0.0599,
+      "step": 2130
+    },
+    {
+      "epoch": 0.385794122949342,
+      "grad_norm": 0.19240501523017883,
+      "learning_rate": 7.13e-05,
+      "loss": 0.0595,
+      "step": 2140
+    },
+    {
+      "epoch": 0.3875968992248062,
+      "grad_norm": 0.25694090127944946,
+      "learning_rate": 7.163333333333334e-05,
+      "loss": 0.0604,
+      "step": 2150
+    },
+    {
+      "epoch": 0.38939967550027044,
+      "grad_norm": 0.29847267270088196,
+      "learning_rate": 7.196666666666668e-05,
+      "loss": 0.057,
+      "step": 2160
+    },
+    {
+      "epoch": 0.39120245177573465,
+      "grad_norm": 0.2709079086780548,
+      "learning_rate": 7.23e-05,
+      "loss": 0.0576,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39300522805119886,
+      "grad_norm": 0.22980639338493347,
+      "learning_rate": 7.263333333333334e-05,
+      "loss": 0.0582,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3948080043266631,
+      "grad_norm": 0.2066880613565445,
+      "learning_rate": 7.296666666666667e-05,
+      "loss": 0.0594,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3966107806021273,
+      "grad_norm": 0.15543051064014435,
+      "learning_rate": 7.33e-05,
+      "loss": 0.0621,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3984135568775915,
+      "grad_norm": 0.30837076902389526,
+      "learning_rate": 7.363333333333334e-05,
+      "loss": 0.0587,
+      "step": 2210
+    },
+    {
+      "epoch": 0.4002163331530557,
+      "grad_norm": 0.2566860616207123,
+      "learning_rate": 7.396666666666667e-05,
+      "loss": 0.0551,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4020191094285199,
+      "grad_norm": 0.17758356034755707,
+      "learning_rate": 7.43e-05,
+      "loss": 0.0574,
+      "step": 2230
+    },
+    {
+      "epoch": 0.40382188570398414,
+      "grad_norm": 0.2511194348335266,
+      "learning_rate": 7.463333333333334e-05,
+      "loss": 0.0614,
+      "step": 2240
+    },
+    {
+      "epoch": 0.40562466197944835,
+      "grad_norm": 0.2289959192276001,
+      "learning_rate": 7.496666666666667e-05,
+      "loss": 0.0562,
+      "step": 2250
+    },
+    {
+      "epoch": 0.40742743825491257,
+      "grad_norm": 0.2782787084579468,
+      "learning_rate": 7.53e-05,
+      "loss": 0.0579,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4092302145303768,
+      "grad_norm": 0.14765875041484833,
+      "learning_rate": 7.563333333333333e-05,
+      "loss": 0.0545,
+      "step": 2270
+    },
+    {
+      "epoch": 0.411032990805841,
+      "grad_norm": 0.29629790782928467,
+      "learning_rate": 7.596666666666668e-05,
+      "loss": 0.0592,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4128357670813052,
+      "grad_norm": 0.16385123133659363,
+      "learning_rate": 7.630000000000001e-05,
+      "loss": 0.0591,
+      "step": 2290
+    },
+    {
+      "epoch": 0.4146385433567694,
+      "grad_norm": 0.2961997985839844,
+      "learning_rate": 7.663333333333333e-05,
+      "loss": 0.0567,
+      "step": 2300
+    },
+    {
+      "epoch": 0.41644131963223363,
+      "grad_norm": 0.1871434897184372,
+      "learning_rate": 7.696666666666668e-05,
+      "loss": 0.0569,
+      "step": 2310
+    },
+    {
+      "epoch": 0.41824409590769784,
+      "grad_norm": 0.2781011462211609,
+      "learning_rate": 7.730000000000001e-05,
+      "loss": 0.0547,
+      "step": 2320
+    },
+    {
+      "epoch": 0.42004687218316206,
+      "grad_norm": 0.25767332315444946,
+      "learning_rate": 7.763333333333334e-05,
+      "loss": 0.0637,
+      "step": 2330
+    },
+    {
+      "epoch": 0.42184964845862627,
+      "grad_norm": 0.26878052949905396,
+      "learning_rate": 7.796666666666666e-05,
+      "loss": 0.0558,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4236524247340905,
+      "grad_norm": 0.18930955231189728,
+      "learning_rate": 7.83e-05,
+      "loss": 0.0577,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4254552010095547,
+      "grad_norm": 0.23502376675605774,
+      "learning_rate": 7.863333333333334e-05,
+      "loss": 0.0629,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4272579772850189,
+      "grad_norm": 0.2777710258960724,
+      "learning_rate": 7.896666666666667e-05,
+      "loss": 0.0574,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4290607535604831,
+      "grad_norm": 0.18867869675159454,
+      "learning_rate": 7.93e-05,
+      "loss": 0.0593,
+      "step": 2380
+    },
+    {
+      "epoch": 0.43086352983594733,
+      "grad_norm": 0.24806471168994904,
+      "learning_rate": 7.963333333333334e-05,
+      "loss": 0.0558,
+      "step": 2390
+    },
+    {
+      "epoch": 0.43266630611141155,
+      "grad_norm": 0.21862687170505524,
+      "learning_rate": 7.996666666666667e-05,
+      "loss": 0.0552,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4344690823868758,
+      "grad_norm": 0.2106906920671463,
+      "learning_rate": 8.030000000000001e-05,
+      "loss": 0.0621,
+      "step": 2410
+    },
+    {
+      "epoch": 0.43627185866234003,
+      "grad_norm": 0.20159752666950226,
+      "learning_rate": 8.063333333333333e-05,
+      "loss": 0.0626,
+      "step": 2420
+    },
+    {
+      "epoch": 0.43807463493780424,
+      "grad_norm": 0.2598176896572113,
+      "learning_rate": 8.096666666666667e-05,
+      "loss": 0.0546,
+      "step": 2430
+    },
+    {
+      "epoch": 0.43987741121326845,
+      "grad_norm": 0.19820767641067505,
+      "learning_rate": 8.13e-05,
+      "loss": 0.0575,
+      "step": 2440
+    },
+    {
+      "epoch": 0.44168018748873267,
+      "grad_norm": 0.18782299757003784,
+      "learning_rate": 8.163333333333334e-05,
+      "loss": 0.0551,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4434829637641969,
+      "grad_norm": 0.25559478998184204,
+      "learning_rate": 8.196666666666668e-05,
+      "loss": 0.0573,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4452857400396611,
+      "grad_norm": 0.2523142099380493,
+      "learning_rate": 8.23e-05,
+      "loss": 0.0574,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4470885163151253,
+      "grad_norm": 0.20205341279506683,
+      "learning_rate": 8.263333333333334e-05,
+      "loss": 0.056,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4488912925905895,
+      "grad_norm": 0.2733706533908844,
+      "learning_rate": 8.296666666666667e-05,
+      "loss": 0.0574,
+      "step": 2490
+    },
+    {
+      "epoch": 0.45069406886605373,
+      "grad_norm": 0.2302892506122589,
+      "learning_rate": 8.33e-05,
+      "loss": 0.0593,
+      "step": 2500
+    },
+    {
+      "epoch": 0.45249684514151794,
+      "grad_norm": 0.2586212754249573,
+      "learning_rate": 8.363333333333334e-05,
+      "loss": 0.0551,
+      "step": 2510
+    },
+    {
+      "epoch": 0.45429962141698216,
+      "grad_norm": 0.4032387137413025,
+      "learning_rate": 8.396666666666667e-05,
+      "loss": 0.0593,
+      "step": 2520
+    },
+    {
+      "epoch": 0.45610239769244637,
+      "grad_norm": 0.18610864877700806,
+      "learning_rate": 8.43e-05,
+      "loss": 0.0578,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4579051739679106,
+      "grad_norm": 0.29500970244407654,
+      "learning_rate": 8.463333333333335e-05,
+      "loss": 0.0594,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4597079502433748,
+      "grad_norm": 0.19957338273525238,
+      "learning_rate": 8.496666666666667e-05,
+      "loss": 0.0556,
+      "step": 2550
+    },
+    {
+      "epoch": 0.461510726518839,
+      "grad_norm": 0.27168312668800354,
+      "learning_rate": 8.53e-05,
+      "loss": 0.0612,
+      "step": 2560
+    },
+    {
+      "epoch": 0.4633135027943032,
+      "grad_norm": 0.18914657831192017,
+      "learning_rate": 8.563333333333333e-05,
+      "loss": 0.0564,
+      "step": 2570
+    },
+    {
+      "epoch": 0.46511627906976744,
+      "grad_norm": 0.16713805496692657,
+      "learning_rate": 8.596666666666668e-05,
+      "loss": 0.0574,
+      "step": 2580
+    },
+    {
+      "epoch": 0.46691905534523165,
+      "grad_norm": 0.1571536511182785,
+      "learning_rate": 8.63e-05,
+      "loss": 0.0534,
+      "step": 2590
+    },
+    {
+      "epoch": 0.46872183162069586,
+      "grad_norm": 0.20700669288635254,
+      "learning_rate": 8.663333333333333e-05,
+      "loss": 0.0577,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4705246078961601,
+      "grad_norm": 0.20322678983211517,
+      "learning_rate": 8.696666666666668e-05,
+      "loss": 0.0638,
+      "step": 2610
+    },
+    {
+      "epoch": 0.4723273841716243,
+      "grad_norm": 0.17560529708862305,
+      "learning_rate": 8.730000000000001e-05,
+      "loss": 0.0613,
+      "step": 2620
+    },
+    {
+      "epoch": 0.4741301604470885,
+      "grad_norm": 0.2778451144695282,
+      "learning_rate": 8.763333333333334e-05,
+      "loss": 0.0578,
+      "step": 2630
+    },
+    {
+      "epoch": 0.4759329367225527,
+      "grad_norm": 0.24755309522151947,
+      "learning_rate": 8.796666666666667e-05,
+      "loss": 0.0541,
+      "step": 2640
+    },
+    {
+      "epoch": 0.4777357129980169,
+      "grad_norm": 0.2499132603406906,
+      "learning_rate": 8.83e-05,
+      "loss": 0.0537,
+      "step": 2650
+    },
+    {
+      "epoch": 0.47953848927348114,
+      "grad_norm": 0.21429553627967834,
+      "learning_rate": 8.863333333333334e-05,
+      "loss": 0.0644,
+      "step": 2660
+    },
+    {
+      "epoch": 0.48134126554894535,
+      "grad_norm": 0.25340747833251953,
+      "learning_rate": 8.896666666666667e-05,
+      "loss": 0.0574,
+      "step": 2670
+    },
+    {
+      "epoch": 0.48314404182440956,
+      "grad_norm": 0.197022944688797,
+      "learning_rate": 8.93e-05,
+      "loss": 0.0555,
+      "step": 2680
+    },
+    {
+      "epoch": 0.48494681809987383,
+      "grad_norm": 0.28157272934913635,
+      "learning_rate": 8.963333333333333e-05,
+      "loss": 0.0597,
+      "step": 2690
+    },
+    {
+      "epoch": 0.48674959437533805,
+      "grad_norm": 0.20312272012233734,
+      "learning_rate": 8.996666666666667e-05,
+      "loss": 0.0588,
+      "step": 2700
+    },
+    {
+      "epoch": 0.48855237065080226,
+      "grad_norm": 0.22160930931568146,
+      "learning_rate": 9.030000000000001e-05,
+      "loss": 0.0632,
+      "step": 2710
+    },
+    {
+      "epoch": 0.49035514692626647,
+      "grad_norm": 0.1846475899219513,
+      "learning_rate": 9.063333333333333e-05,
+      "loss": 0.0512,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4921579232017307,
+      "grad_norm": 0.22525529563426971,
+      "learning_rate": 9.096666666666666e-05,
+      "loss": 0.0574,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4939606994771949,
+      "grad_norm": 0.14015792310237885,
+      "learning_rate": 9.130000000000001e-05,
+      "loss": 0.0535,
+      "step": 2740
+    },
+    {
+      "epoch": 0.4957634757526591,
+      "grad_norm": 0.25817930698394775,
+      "learning_rate": 9.163333333333334e-05,
+      "loss": 0.0606,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4975662520281233,
+      "grad_norm": 0.21513380110263824,
+      "learning_rate": 9.196666666666666e-05,
+      "loss": 0.06,
+      "step": 2760
+    },
+    {
+      "epoch": 0.49936902830358754,
+      "grad_norm": 0.22595332562923431,
+      "learning_rate": 9.230000000000001e-05,
+      "loss": 0.0519,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5011718045790517,
+      "grad_norm": 0.27969732880592346,
+      "learning_rate": 9.263333333333334e-05,
+      "loss": 0.0623,
+      "step": 2780
+    },
+    {
+      "epoch": 0.502974580854516,
+      "grad_norm": 0.2222040742635727,
+      "learning_rate": 9.296666666666667e-05,
+      "loss": 0.0565,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5047773571299802,
+      "grad_norm": 0.28888288140296936,
+      "learning_rate": 9.33e-05,
+      "loss": 0.0577,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5065801334054444,
+      "grad_norm": 0.1962571144104004,
+      "learning_rate": 9.363333333333334e-05,
+      "loss": 0.0536,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5083829096809086,
+      "grad_norm": 0.26705968379974365,
+      "learning_rate": 9.396666666666667e-05,
+      "loss": 0.0603,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5101856859563728,
+      "grad_norm": 0.17329153418540955,
+      "learning_rate": 9.43e-05,
+      "loss": 0.0533,
+      "step": 2830
+    },
+    {
+      "epoch": 0.511988462231837,
+      "grad_norm": 0.2211298942565918,
+      "learning_rate": 9.463333333333333e-05,
+      "loss": 0.0608,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5137912385073012,
+      "grad_norm": 0.2022695690393448,
+      "learning_rate": 9.496666666666667e-05,
+      "loss": 0.0579,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5155940147827655,
+      "grad_norm": 0.16921758651733398,
+      "learning_rate": 9.53e-05,
+      "loss": 0.0568,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5173967910582297,
+      "grad_norm": 0.24564726650714874,
+      "learning_rate": 9.563333333333334e-05,
+      "loss": 0.0568,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5191995673336939,
+      "grad_norm": 0.2871924638748169,
+      "learning_rate": 9.596666666666668e-05,
+      "loss": 0.0596,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5210023436091581,
+      "grad_norm": 0.27038854360580444,
+      "learning_rate": 9.63e-05,
+      "loss": 0.0596,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5228051198846223,
+      "grad_norm": 0.2762112319469452,
+      "learning_rate": 9.663333333333334e-05,
+      "loss": 0.0588,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5246078961600865,
+      "grad_norm": 0.1921473741531372,
+      "learning_rate": 9.696666666666667e-05,
+      "loss": 0.0626,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5264106724355507,
+      "grad_norm": 0.27863645553588867,
+      "learning_rate": 9.730000000000001e-05,
+      "loss": 0.0567,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5282134487110149,
+      "grad_norm": 0.23096972703933716,
+      "learning_rate": 9.763333333333334e-05,
+      "loss": 0.0579,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5300162249864792,
+      "grad_norm": 0.23984530568122864,
+      "learning_rate": 9.796666666666667e-05,
+      "loss": 0.0539,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5318190012619434,
+      "grad_norm": 0.26811182498931885,
+      "learning_rate": 9.83e-05,
+      "loss": 0.0532,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5336217775374076,
+      "grad_norm": 0.30521097779273987,
+      "learning_rate": 9.863333333333334e-05,
+      "loss": 0.0604,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5354245538128718,
+      "grad_norm": 0.1678686887025833,
+      "learning_rate": 9.896666666666667e-05,
+      "loss": 0.0566,
+      "step": 2970
+    },
+    {
+      "epoch": 0.537227330088336,
+      "grad_norm": 0.19016125798225403,
+      "learning_rate": 9.93e-05,
+      "loss": 0.0538,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5390301063638002,
+      "grad_norm": 0.30830803513526917,
+      "learning_rate": 9.963333333333333e-05,
+      "loss": 0.0638,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5408328826392644,
+      "grad_norm": 0.17846520245075226,
+      "learning_rate": 9.996666666666668e-05,
+      "loss": 0.0537,
+      "step": 3000
+    },
+    {
+      "epoch": 0.5426356589147286,
+      "grad_norm": 0.15769369900226593,
+      "learning_rate": 9.999999384858465e-05,
+      "loss": 0.0522,
+      "step": 3010
+    },
+    {
+      "epoch": 0.5444384351901929,
+      "grad_norm": 0.2433222085237503,
+      "learning_rate": 9.999997258443473e-05,
+      "loss": 0.0579,
+      "step": 3020
+    },
+    {
+      "epoch": 0.5462412114656571,
+      "grad_norm": 0.24462740123271942,
+      "learning_rate": 9.999993613161331e-05,
+      "loss": 0.0608,
+      "step": 3030
+    },
+    {
+      "epoch": 0.5480439877411213,
+      "grad_norm": 0.2256120890378952,
+      "learning_rate": 9.999988449013146e-05,
+      "loss": 0.0543,
+      "step": 3040
+    },
+    {
+      "epoch": 0.5498467640165855,
+      "grad_norm": 0.23624876141548157,
+      "learning_rate": 9.99998176600049e-05,
+      "loss": 0.055,
+      "step": 3050
+    },
+    {
+      "epoch": 0.5516495402920497,
+      "grad_norm": 0.1689511388540268,
+      "learning_rate": 9.999973564125389e-05,
+      "loss": 0.0563,
+      "step": 3060
+    },
+    {
+      "epoch": 0.5534523165675139,
+      "grad_norm": 0.31788817048072815,
+      "learning_rate": 9.999963843390335e-05,
+      "loss": 0.0554,
+      "step": 3070
+    },
+    {
+      "epoch": 0.5552550928429781,
+      "grad_norm": 0.2632403075695038,
+      "learning_rate": 9.999952603798282e-05,
+      "loss": 0.0517,
+      "step": 3080
+    },
+    {
+      "epoch": 0.5570578691184424,
+      "grad_norm": 0.2732613980770111,
+      "learning_rate": 9.999939845352646e-05,
+      "loss": 0.0599,
+      "step": 3090
+    },
+    {
+      "epoch": 0.5588606453939066,
+      "grad_norm": 0.17078642547130585,
+      "learning_rate": 9.999925568057298e-05,
+      "loss": 0.0597,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5606634216693708,
+      "grad_norm": 0.23614409565925598,
+      "learning_rate": 9.999909771916578e-05,
+      "loss": 0.0559,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5624661979448351,
+      "grad_norm": 0.21851426362991333,
+      "learning_rate": 9.999892456935285e-05,
+      "loss": 0.0564,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5642689742202993,
+      "grad_norm": 0.19016896188259125,
+      "learning_rate": 9.999873623118679e-05,
+      "loss": 0.0577,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5660717504957635,
+      "grad_norm": 0.1992519646883011,
+      "learning_rate": 9.999853270472479e-05,
+      "loss": 0.0599,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5678745267712277,
+      "grad_norm": 0.37866124510765076,
+      "learning_rate": 9.999831399002871e-05,
+      "loss": 0.0569,
+      "step": 3150
+    },
+    {
+      "epoch": 0.569677303046692,
+      "grad_norm": 0.17698729038238525,
+      "learning_rate": 9.999808008716494e-05,
+      "loss": 0.0585,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5714800793221562,
+      "grad_norm": 0.2222241461277008,
+      "learning_rate": 9.999783099620459e-05,
+      "loss": 0.0546,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5732828555976204,
+      "grad_norm": 0.1738797426223755,
+      "learning_rate": 9.999756671722328e-05,
+      "loss": 0.0525,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5750856318730846,
+      "grad_norm": 0.16521303355693817,
+      "learning_rate": 9.99972872503013e-05,
+      "loss": 0.0524,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5768884081485488,
+      "grad_norm": 0.21063007414340973,
+      "learning_rate": 9.999699259552359e-05,
+      "loss": 0.0515,
+      "step": 3200
+    },
+    {
+      "epoch": 0.578691184424013,
+      "grad_norm": 0.20505587756633759,
+      "learning_rate": 9.99966827529796e-05,
+      "loss": 0.0509,
+      "step": 3210
+    },
+    {
+      "epoch": 0.5804939606994772,
+      "grad_norm": 0.24220533668994904,
+      "learning_rate": 9.999635772276348e-05,
+      "loss": 0.0577,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5822967369749414,
+      "grad_norm": 0.20930424332618713,
+      "learning_rate": 9.999601750497396e-05,
+      "loss": 0.0526,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5840995132504057,
+      "grad_norm": 0.17883487045764923,
+      "learning_rate": 9.99956620997144e-05,
+      "loss": 0.0529,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5859022895258699,
+      "grad_norm": 0.27819836139678955,
+      "learning_rate": 9.999529150709275e-05,
+      "loss": 0.0573,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5877050658013341,
+      "grad_norm": 0.31870806217193604,
+      "learning_rate": 9.999490572722158e-05,
+      "loss": 0.0576,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5895078420767983,
+      "grad_norm": 0.2694283723831177,
+      "learning_rate": 9.99945047602181e-05,
+      "loss": 0.0544,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5913106183522625,
+      "grad_norm": 0.2328466773033142,
+      "learning_rate": 9.99940886062041e-05,
+      "loss": 0.0572,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5931133946277267,
+      "grad_norm": 0.2757991850376129,
+      "learning_rate": 9.999365726530599e-05,
+      "loss": 0.0567,
+      "step": 3290
+    },
+    {
+      "epoch": 0.5949161709031909,
+      "grad_norm": 0.1877506971359253,
+      "learning_rate": 9.999321073765481e-05,
+      "loss": 0.0521,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5967189471786551,
+      "grad_norm": 0.17714393138885498,
+      "learning_rate": 9.99927490233862e-05,
+      "loss": 0.052,
+      "step": 3310
+    },
+    {
+      "epoch": 0.5985217234541194,
+      "grad_norm": 0.2916216254234314,
+      "learning_rate": 9.999227212264043e-05,
+      "loss": 0.0577,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6003244997295836,
+      "grad_norm": 0.16772031784057617,
+      "learning_rate": 9.999178003556236e-05,
+      "loss": 0.0554,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6021272760050478,
+      "grad_norm": 0.2048356533050537,
+      "learning_rate": 9.999127276230146e-05,
+      "loss": 0.0521,
+      "step": 3340
+    },
+    {
+      "epoch": 0.603930052280512,
+      "grad_norm": 0.1914864331483841,
+      "learning_rate": 9.999075030301184e-05,
+      "loss": 0.0575,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6057328285559762,
+      "grad_norm": 0.22202041745185852,
+      "learning_rate": 9.999021265785221e-05,
+      "loss": 0.0548,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6075356048314404,
+      "grad_norm": 0.14372526109218597,
+      "learning_rate": 9.998965982698589e-05,
+      "loss": 0.0547,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6093383811069046,
+      "grad_norm": 0.22338172793388367,
+      "learning_rate": 9.998909181058082e-05,
+      "loss": 0.0596,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6111411573823688,
+      "grad_norm": 0.23771338164806366,
+      "learning_rate": 9.998850860880953e-05,
+      "loss": 0.0542,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6129439336578331,
+      "grad_norm": 0.14361989498138428,
+      "learning_rate": 9.998791022184922e-05,
+      "loss": 0.0493,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6147467099332973,
+      "grad_norm": 0.24462664127349854,
+      "learning_rate": 9.99872966498816e-05,
+      "loss": 0.0544,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6165494862087615,
+      "grad_norm": 0.26684728264808655,
+      "learning_rate": 9.998666789309313e-05,
+      "loss": 0.0569,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6183522624842257,
+      "grad_norm": 0.22234764695167542,
+      "learning_rate": 9.998602395167475e-05,
+      "loss": 0.0514,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6201550387596899,
+      "grad_norm": 0.2795373797416687,
+      "learning_rate": 9.998536482582213e-05,
+      "loss": 0.0583,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6219578150351541,
+      "grad_norm": 0.22455738484859467,
+      "learning_rate": 9.998469051573544e-05,
+      "loss": 0.0482,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6237605913106183,
+      "grad_norm": 0.204681396484375,
+      "learning_rate": 9.998400102161954e-05,
+      "loss": 0.0541,
+      "step": 3460
+    },
+    {
+      "epoch": 0.6255633675860826,
+      "grad_norm": 0.1677926778793335,
+      "learning_rate": 9.998329634368388e-05,
+      "loss": 0.0556,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6273661438615468,
+      "grad_norm": 0.16572725772857666,
+      "learning_rate": 9.998257648214253e-05,
+      "loss": 0.0511,
+      "step": 3480
+    },
+    {
+      "epoch": 0.629168920137011,
+      "grad_norm": 0.18733033537864685,
+      "learning_rate": 9.998184143721417e-05,
+      "loss": 0.0543,
+      "step": 3490
+    },
+    {
+      "epoch": 0.6309716964124752,
+      "grad_norm": 0.23857447504997253,
+      "learning_rate": 9.998109120912206e-05,
+      "loss": 0.0558,
+      "step": 3500
+    },
+    {
+      "epoch": 0.6327744726879394,
+      "grad_norm": 0.20467570424079895,
+      "learning_rate": 9.998032579809411e-05,
+      "loss": 0.0525,
+      "step": 3510
+    },
+    {
+      "epoch": 0.6345772489634036,
+      "grad_norm": 0.1384439617395401,
+      "learning_rate": 9.997954520436286e-05,
+      "loss": 0.0541,
+      "step": 3520
+    },
+    {
+      "epoch": 0.6363800252388678,
+      "grad_norm": 0.19725823402404785,
+      "learning_rate": 9.997874942816538e-05,
+      "loss": 0.0569,
+      "step": 3530
+    },
+    {
+      "epoch": 0.638182801514332,
+      "grad_norm": 0.16356350481510162,
+      "learning_rate": 9.997793846974345e-05,
+      "loss": 0.0501,
+      "step": 3540
+    },
+    {
+      "epoch": 0.6399855777897963,
+      "grad_norm": 0.20610161125659943,
+      "learning_rate": 9.997711232934341e-05,
+      "loss": 0.0551,
+      "step": 3550
+    },
+    {
+      "epoch": 0.6417883540652605,
+      "grad_norm": 0.16926410794258118,
+      "learning_rate": 9.99762710072162e-05,
+      "loss": 0.0594,
+      "step": 3560
+    },
+    {
+      "epoch": 0.6435911303407247,
+      "grad_norm": 0.17915554344654083,
+      "learning_rate": 9.997541450361743e-05,
+      "loss": 0.0545,
+      "step": 3570
+    },
+    {
+      "epoch": 0.6453939066161889,
+      "grad_norm": 0.2203824371099472,
+      "learning_rate": 9.997454281880723e-05,
+      "loss": 0.058,
+      "step": 3580
+    },
+    {
+      "epoch": 0.6471966828916531,
+      "grad_norm": 0.2686256766319275,
+      "learning_rate": 9.997365595305044e-05,
+      "loss": 0.06,
+      "step": 3590
+    },
+    {
+      "epoch": 0.6489994591671173,
+      "grad_norm": 0.2311336100101471,
+      "learning_rate": 9.997275390661644e-05,
+      "loss": 0.0536,
+      "step": 3600
+    },
+    {
+      "epoch": 0.6508022354425815,
+      "grad_norm": 0.2140345573425293,
+      "learning_rate": 9.997183667977926e-05,
+      "loss": 0.0581,
+      "step": 3610
+    },
+    {
+      "epoch": 0.6526050117180457,
+      "grad_norm": 0.2858904302120209,
+      "learning_rate": 9.997090427281752e-05,
+      "loss": 0.0492,
+      "step": 3620
+    },
+    {
+      "epoch": 0.65440778799351,
+      "grad_norm": 0.2953670024871826,
+      "learning_rate": 9.996995668601448e-05,
+      "loss": 0.055,
+      "step": 3630
+    },
+    {
+      "epoch": 0.6562105642689742,
+      "grad_norm": 0.2625373303890228,
+      "learning_rate": 9.996899391965798e-05,
+      "loss": 0.0521,
+      "step": 3640
+    },
+    {
+      "epoch": 0.6580133405444384,
+      "grad_norm": 0.16617342829704285,
+      "learning_rate": 9.996801597404048e-05,
+      "loss": 0.0524,
+      "step": 3650
+    },
+    {
+      "epoch": 0.6598161168199026,
+      "grad_norm": 0.19370779395103455,
+      "learning_rate": 9.996702284945905e-05,
+      "loss": 0.0541,
+      "step": 3660
+    },
+    {
+      "epoch": 0.6616188930953668,
+      "grad_norm": 0.22694896161556244,
+      "learning_rate": 9.996601454621539e-05,
+      "loss": 0.0543,
+      "step": 3670
+    },
+    {
+      "epoch": 0.6634216693708311,
+      "grad_norm": 0.16225895285606384,
+      "learning_rate": 9.996499106461577e-05,
+      "loss": 0.0513,
+      "step": 3680
+    },
+    {
+      "epoch": 0.6652244456462953,
+      "grad_norm": 0.13752566277980804,
+      "learning_rate": 9.996395240497112e-05,
+      "loss": 0.0552,
+      "step": 3690
+    },
+    {
+      "epoch": 0.6670272219217596,
+      "grad_norm": 0.17287544906139374,
+      "learning_rate": 9.996289856759696e-05,
+      "loss": 0.055,
+      "step": 3700
+    },
+    {
+      "epoch": 0.6688299981972238,
+      "grad_norm": 0.1903308480978012,
+      "learning_rate": 9.996182955281342e-05,
+      "loss": 0.0541,
+      "step": 3710
+    },
+    {
+      "epoch": 0.670632774472688,
+      "grad_norm": 0.17485389113426208,
+      "learning_rate": 9.996074536094519e-05,
+      "loss": 0.0511,
+      "step": 3720
+    },
+    {
+      "epoch": 0.6724355507481522,
+      "grad_norm": 0.18762964010238647,
+      "learning_rate": 9.995964599232168e-05,
+      "loss": 0.0574,
+      "step": 3730
+    },
+    {
+      "epoch": 0.6742383270236164,
+      "grad_norm": 0.17530937492847443,
+      "learning_rate": 9.995853144727683e-05,
+      "loss": 0.0603,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6760411032990806,
+      "grad_norm": 0.1973475217819214,
+      "learning_rate": 9.99574017261492e-05,
+      "loss": 0.0544,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6778438795745448,
+      "grad_norm": 0.2093428671360016,
+      "learning_rate": 9.995625682928198e-05,
+      "loss": 0.0559,
+      "step": 3760
+    },
+    {
+      "epoch": 0.679646655850009,
+      "grad_norm": 0.14786028861999512,
+      "learning_rate": 9.995509675702295e-05,
+      "loss": 0.0529,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6814494321254733,
+      "grad_norm": 0.2190335988998413,
+      "learning_rate": 9.995392150972451e-05,
+      "loss": 0.0643,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6832522084009375,
+      "grad_norm": 0.22454246878623962,
+      "learning_rate": 9.995273108774366e-05,
+      "loss": 0.0522,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6850549846764017,
+      "grad_norm": 0.18701009452342987,
+      "learning_rate": 9.995152549144205e-05,
+      "loss": 0.0547,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6868577609518659,
+      "grad_norm": 0.21698643267154694,
+      "learning_rate": 9.995030472118587e-05,
+      "loss": 0.0543,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6886605372273301,
+      "grad_norm": 0.2331894040107727,
+      "learning_rate": 9.9949068777346e-05,
+      "loss": 0.0578,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6904633135027943,
+      "grad_norm": 0.23350775241851807,
+      "learning_rate": 9.994781766029786e-05,
+      "loss": 0.0556,
+      "step": 3830
+    },
+    {
+      "epoch": 0.6922660897782585,
+      "grad_norm": 0.17531098425388336,
+      "learning_rate": 9.994655137042151e-05,
+      "loss": 0.0514,
+      "step": 3840
+    },
+    {
+      "epoch": 0.6940688660537228,
+      "grad_norm": 0.17184023559093475,
+      "learning_rate": 9.99452699081016e-05,
+      "loss": 0.0495,
+      "step": 3850
+    },
+    {
+      "epoch": 0.695871642329187,
+      "grad_norm": 0.16536134481430054,
+      "learning_rate": 9.994397327372743e-05,
+      "loss": 0.0539,
+      "step": 3860
+    },
+    {
+      "epoch": 0.6976744186046512,
+      "grad_norm": 0.1660694181919098,
+      "learning_rate": 9.994266146769286e-05,
+      "loss": 0.053,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6994771948801154,
+      "grad_norm": 0.20537647604942322,
+      "learning_rate": 9.994133449039642e-05,
+      "loss": 0.0578,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7012799711555796,
+      "grad_norm": 0.18250614404678345,
+      "learning_rate": 9.993999234224118e-05,
+      "loss": 0.0511,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7030827474310438,
+      "grad_norm": 0.21514642238616943,
+      "learning_rate": 9.993863502363485e-05,
+      "loss": 0.0553,
+      "step": 3900
+    },
+    {
+      "epoch": 0.704885523706508,
+      "grad_norm": 0.2605758309364319,
+      "learning_rate": 9.993726253498976e-05,
+      "loss": 0.0529,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7066882999819722,
+      "grad_norm": 0.20881788432598114,
+      "learning_rate": 9.993587487672282e-05,
+      "loss": 0.0555,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7084910762574365,
+      "grad_norm": 0.18526031076908112,
+      "learning_rate": 9.993447204925558e-05,
+      "loss": 0.0529,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7102938525329007,
+      "grad_norm": 0.17271843552589417,
+      "learning_rate": 9.993305405301416e-05,
+      "loss": 0.0539,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7120966288083649,
+      "grad_norm": 0.17218437790870667,
+      "learning_rate": 9.993162088842935e-05,
+      "loss": 0.053,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7138994050838291,
+      "grad_norm": 0.20867522060871124,
+      "learning_rate": 9.993017255593646e-05,
+      "loss": 0.0544,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7157021813592933,
+      "grad_norm": 0.18345223367214203,
+      "learning_rate": 9.992870905597548e-05,
+      "loss": 0.0522,
+      "step": 3970
+    },
+    {
+      "epoch": 0.7175049576347575,
+      "grad_norm": 0.14037351310253143,
+      "learning_rate": 9.9927230388991e-05,
+      "loss": 0.0507,
+      "step": 3980
+    },
+    {
+      "epoch": 0.7193077339102217,
+      "grad_norm": 0.1470947265625,
+      "learning_rate": 9.992573655543215e-05,
+      "loss": 0.0525,
+      "step": 3990
+    },
+    {
+      "epoch": 0.721110510185686,
+      "grad_norm": 0.15597905218601227,
+      "learning_rate": 9.992422755575277e-05,
+      "loss": 0.0486,
+      "step": 4000
+    },
+    {
+      "epoch": 0.7229132864611502,
+      "grad_norm": 0.15142855048179626,
+      "learning_rate": 9.992270339041123e-05,
+      "loss": 0.0477,
+      "step": 4010
+    },
+    {
+      "epoch": 0.7247160627366144,
+      "grad_norm": 0.19573090970516205,
+      "learning_rate": 9.992116405987053e-05,
+      "loss": 0.0564,
+      "step": 4020
+    },
+    {
+      "epoch": 0.7265188390120786,
+      "grad_norm": 0.13741803169250488,
+      "learning_rate": 9.991960956459828e-05,
+      "loss": 0.0533,
+      "step": 4030
+    },
+    {
+      "epoch": 0.7283216152875428,
+      "grad_norm": 0.19118353724479675,
+      "learning_rate": 9.991803990506669e-05,
+      "loss": 0.0509,
+      "step": 4040
+    },
+    {
+      "epoch": 0.730124391563007,
+      "grad_norm": 0.18628346920013428,
+      "learning_rate": 9.991645508175258e-05,
+      "loss": 0.0568,
+      "step": 4050
+    },
+    {
+      "epoch": 0.7319271678384712,
+      "grad_norm": 0.2145071029663086,
+      "learning_rate": 9.99148550951374e-05,
+      "loss": 0.0523,
+      "step": 4060
+    },
+    {
+      "epoch": 0.7337299441139354,
+      "grad_norm": 0.17149919271469116,
+      "learning_rate": 9.991323994570716e-05,
+      "loss": 0.0483,
+      "step": 4070
+    },
+    {
+      "epoch": 0.7355327203893997,
+      "grad_norm": 0.22665870189666748,
+      "learning_rate": 9.99116096339525e-05,
+      "loss": 0.0567,
+      "step": 4080
+    },
+    {
+      "epoch": 0.7373354966648639,
+      "grad_norm": 0.17213742434978485,
+      "learning_rate": 9.990996416036869e-05,
+      "loss": 0.0519,
+      "step": 4090
+    },
+    {
+      "epoch": 0.7391382729403281,
+      "grad_norm": 0.21811003983020782,
+      "learning_rate": 9.990830352545555e-05,
+      "loss": 0.0507,
+      "step": 4100
+    },
+    {
+      "epoch": 0.7409410492157923,
+      "grad_norm": 0.19745135307312012,
+      "learning_rate": 9.990662772971756e-05,
+      "loss": 0.0522,
+      "step": 4110
+    },
+    {
+      "epoch": 0.7427438254912565,
+      "grad_norm": 0.15801742672920227,
+      "learning_rate": 9.990493677366376e-05,
+      "loss": 0.0531,
+      "step": 4120
+    },
+    {
+      "epoch": 0.7445466017667207,
+      "grad_norm": 0.19218087196350098,
+      "learning_rate": 9.990323065780786e-05,
+      "loss": 0.057,
+      "step": 4130
+    },
+    {
+      "epoch": 0.7463493780421849,
+      "grad_norm": 0.20360451936721802,
+      "learning_rate": 9.990150938266808e-05,
+      "loss": 0.0525,
+      "step": 4140
+    },
+    {
+      "epoch": 0.7481521543176491,
+      "grad_norm": 0.19944339990615845,
+      "learning_rate": 9.989977294876733e-05,
+      "loss": 0.0533,
+      "step": 4150
+    },
+    {
+      "epoch": 0.7499549305931134,
+      "grad_norm": 0.16528679430484772,
+      "learning_rate": 9.989802135663308e-05,
+      "loss": 0.0532,
+      "step": 4160
+    },
+    {
+      "epoch": 0.7517577068685776,
+      "grad_norm": 0.15513837337493896,
+      "learning_rate": 9.989625460679743e-05,
+      "loss": 0.0493,
+      "step": 4170
+    },
+    {
+      "epoch": 0.7535604831440418,
+      "grad_norm": 0.2322871834039688,
+      "learning_rate": 9.989447269979706e-05,
+      "loss": 0.0529,
+      "step": 4180
+    },
+    {
+      "epoch": 0.755363259419506,
+      "grad_norm": 0.1429019719362259,
+      "learning_rate": 9.989267563617328e-05,
+      "loss": 0.0524,
+      "step": 4190
+    },
+    {
+      "epoch": 0.7571660356949702,
+      "grad_norm": 0.15561752021312714,
+      "learning_rate": 9.989086341647198e-05,
+      "loss": 0.048,
+      "step": 4200
+    },
+    {
+      "epoch": 0.7589688119704344,
+      "grad_norm": 0.15190979838371277,
+      "learning_rate": 9.988903604124366e-05,
+      "loss": 0.0502,
+      "step": 4210
+    },
+    {
+      "epoch": 0.7607715882458986,
+      "grad_norm": 0.19186142086982727,
+      "learning_rate": 9.988719351104343e-05,
+      "loss": 0.0484,
+      "step": 4220
+    },
+    {
+      "epoch": 0.7625743645213628,
+      "grad_norm": 0.2058783620595932,
+      "learning_rate": 9.9885335826431e-05,
+      "loss": 0.0463,
+      "step": 4230
+    },
+    {
+      "epoch": 0.7643771407968271,
+      "grad_norm": 0.19337031245231628,
+      "learning_rate": 9.988346298797071e-05,
+      "loss": 0.0493,
+      "step": 4240
+    },
+    {
+      "epoch": 0.7661799170722914,
+      "grad_norm": 0.24919281899929047,
+      "learning_rate": 9.988157499623146e-05,
+      "loss": 0.0566,
+      "step": 4250
+    },
+    {
+      "epoch": 0.7679826933477556,
+      "grad_norm": 0.23650670051574707,
+      "learning_rate": 9.987967185178677e-05,
+      "loss": 0.0529,
+      "step": 4260
+    },
+    {
+      "epoch": 0.7697854696232198,
+      "grad_norm": 0.15392479300498962,
+      "learning_rate": 9.987775355521476e-05,
+      "loss": 0.0468,
+      "step": 4270
+    },
+    {
+      "epoch": 0.771588245898684,
+      "grad_norm": 0.2416786551475525,
+      "learning_rate": 9.987582010709817e-05,
+      "loss": 0.0479,
+      "step": 4280
+    },
+    {
+      "epoch": 0.7733910221741482,
+      "grad_norm": 0.15816278755664825,
+      "learning_rate": 9.987387150802431e-05,
+      "loss": 0.0517,
+      "step": 4290
+    },
+    {
+      "epoch": 0.7751937984496124,
+      "grad_norm": 0.16642552614212036,
+      "learning_rate": 9.987190775858517e-05,
+      "loss": 0.0498,
+      "step": 4300
+    },
+    {
+      "epoch": 0.7769965747250767,
+      "grad_norm": 0.12232939898967743,
+      "learning_rate": 9.98699288593772e-05,
+      "loss": 0.0528,
+      "step": 4310
+    },
+    {
+      "epoch": 0.7787993510005409,
+      "grad_norm": 0.1731266975402832,
+      "learning_rate": 9.986793481100161e-05,
+      "loss": 0.048,
+      "step": 4320
+    },
+    {
+      "epoch": 0.7806021272760051,
+      "grad_norm": 0.21021099388599396,
+      "learning_rate": 9.986592561406412e-05,
+      "loss": 0.0483,
+      "step": 4330
+    },
+    {
+      "epoch": 0.7824049035514693,
+      "grad_norm": 0.1725785732269287,
+      "learning_rate": 9.986390126917503e-05,
+      "loss": 0.0534,
+      "step": 4340
+    },
+    {
+      "epoch": 0.7842076798269335,
+      "grad_norm": 0.2088138461112976,
+      "learning_rate": 9.986186177694933e-05,
+      "loss": 0.0547,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7860104561023977,
+      "grad_norm": 0.21956445276737213,
+      "learning_rate": 9.985980713800656e-05,
+      "loss": 0.0478,
+      "step": 4360
+    },
+    {
+      "epoch": 0.7878132323778619,
+      "grad_norm": 0.16561521589756012,
+      "learning_rate": 9.985773735297084e-05,
+      "loss": 0.0521,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7896160086533262,
+      "grad_norm": 0.16367164254188538,
+      "learning_rate": 9.985565242247092e-05,
+      "loss": 0.0538,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7914187849287904,
+      "grad_norm": 0.19190697371959686,
+      "learning_rate": 9.985355234714016e-05,
+      "loss": 0.0496,
+      "step": 4390
+    },
+    {
+      "epoch": 0.7932215612042546,
+      "grad_norm": 0.14845453202724457,
+      "learning_rate": 9.985143712761652e-05,
+      "loss": 0.0483,
+      "step": 4400
+    },
+    {
+      "epoch": 0.7950243374797188,
+      "grad_norm": 0.13262321054935455,
+      "learning_rate": 9.984930676454252e-05,
+      "loss": 0.0553,
+      "step": 4410
+    },
+    {
+      "epoch": 0.796827113755183,
+      "grad_norm": 0.20816828310489655,
+      "learning_rate": 9.984716125856532e-05,
+      "loss": 0.0507,
+      "step": 4420
+    },
+    {
+      "epoch": 0.7986298900306472,
+      "grad_norm": 0.15905918180942535,
+      "learning_rate": 9.984500061033667e-05,
+      "loss": 0.0472,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8004326663061114,
+      "grad_norm": 0.16600371897220612,
+      "learning_rate": 9.984282482051293e-05,
+      "loss": 0.0499,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8022354425815756,
+      "grad_norm": 0.2226187288761139,
+      "learning_rate": 9.9840633889755e-05,
+      "loss": 0.0513,
+      "step": 4450
+    },
+    {
+      "epoch": 0.8040382188570399,
+      "grad_norm": 0.13732576370239258,
+      "learning_rate": 9.983842781872848e-05,
+      "loss": 0.0514,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8058409951325041,
+      "grad_norm": 0.16772523522377014,
+      "learning_rate": 9.98362066081035e-05,
+      "loss": 0.0553,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8076437714079683,
+      "grad_norm": 0.15782472491264343,
+      "learning_rate": 9.983397025855479e-05,
+      "loss": 0.0514,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8094465476834325,
+      "grad_norm": 0.20454518496990204,
+      "learning_rate": 9.983171877076171e-05,
+      "loss": 0.0503,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8112493239588967,
+      "grad_norm": 0.23142272233963013,
+      "learning_rate": 9.98294521454082e-05,
+      "loss": 0.0515,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8130521002343609,
+      "grad_norm": 0.17832320928573608,
+      "learning_rate": 9.98271703831828e-05,
+      "loss": 0.0495,
+      "step": 4510
+    },
+    {
+      "epoch": 0.8148548765098251,
+      "grad_norm": 0.16449207067489624,
+      "learning_rate": 9.982487348477865e-05,
+      "loss": 0.0577,
+      "step": 4520
+    },
+    {
+      "epoch": 0.8166576527852893,
+      "grad_norm": 0.2397778481245041,
+      "learning_rate": 9.982256145089347e-05,
+      "loss": 0.0437,
+      "step": 4530
+    },
+    {
+      "epoch": 0.8184604290607536,
+      "grad_norm": 0.13208262622356415,
+      "learning_rate": 9.982023428222962e-05,
+      "loss": 0.0477,
+      "step": 4540
+    },
+    {
+      "epoch": 0.8202632053362178,
+      "grad_norm": 0.168897807598114,
+      "learning_rate": 9.981789197949403e-05,
+      "loss": 0.0475,
+      "step": 4550
+    },
+    {
+      "epoch": 0.822065981611682,
+      "grad_norm": 0.15609736740589142,
+      "learning_rate": 9.98155345433982e-05,
+      "loss": 0.0481,
+      "step": 4560
+    },
+    {
+      "epoch": 0.8238687578871462,
+      "grad_norm": 0.11921030282974243,
+      "learning_rate": 9.981316197465831e-05,
+      "loss": 0.0582,
+      "step": 4570
+    },
+    {
+      "epoch": 0.8256715341626104,
+      "grad_norm": 0.23547197878360748,
+      "learning_rate": 9.981077427399504e-05,
+      "loss": 0.0474,
+      "step": 4580
+    },
+    {
+      "epoch": 0.8274743104380746,
+      "grad_norm": 0.15833474695682526,
+      "learning_rate": 9.980837144213371e-05,
+      "loss": 0.0477,
+      "step": 4590
+    },
+    {
+      "epoch": 0.8292770867135388,
+      "grad_norm": 0.11992108076810837,
+      "learning_rate": 9.980595347980426e-05,
+      "loss": 0.0494,
+      "step": 4600
+    },
+    {
+      "epoch": 0.831079862989003,
+      "grad_norm": 0.1470881551504135,
+      "learning_rate": 9.980352038774119e-05,
+      "loss": 0.0523,
+      "step": 4610
+    },
+    {
+      "epoch": 0.8328826392644673,
+      "grad_norm": 0.17348624765872955,
+      "learning_rate": 9.98010721666836e-05,
+      "loss": 0.0509,
+      "step": 4620
+    },
+    {
+      "epoch": 0.8346854155399315,
+      "grad_norm": 0.19449074566364288,
+      "learning_rate": 9.979860881737523e-05,
+      "loss": 0.046,
+      "step": 4630
+    },
+    {
+      "epoch": 0.8364881918153957,
+      "grad_norm": 0.18979020416736603,
+      "learning_rate": 9.979613034056434e-05,
+      "loss": 0.0487,
+      "step": 4640
+    },
+    {
+      "epoch": 0.8382909680908599,
+      "grad_norm": 0.214563250541687,
+      "learning_rate": 9.979363673700386e-05,
+      "loss": 0.0484,
+      "step": 4650
+    },
+    {
+      "epoch": 0.8400937443663241,
+      "grad_norm": 0.14514483511447906,
+      "learning_rate": 9.979112800745124e-05,
+      "loss": 0.0505,
+      "step": 4660
+    },
+    {
+      "epoch": 0.8418965206417883,
+      "grad_norm": 0.1271432340145111,
+      "learning_rate": 9.978860415266861e-05,
+      "loss": 0.0514,
+      "step": 4670
+    },
+    {
+      "epoch": 0.8436992969172525,
+      "grad_norm": 0.25574085116386414,
+      "learning_rate": 9.978606517342262e-05,
+      "loss": 0.052,
+      "step": 4680
+    },
+    {
+      "epoch": 0.8455020731927168,
+      "grad_norm": 0.16129499673843384,
+      "learning_rate": 9.978351107048456e-05,
+      "loss": 0.0516,
+      "step": 4690
+    },
+    {
+      "epoch": 0.847304849468181,
+      "grad_norm": 0.208964541554451,
+      "learning_rate": 9.978094184463029e-05,
+      "loss": 0.0485,
+      "step": 4700
+    },
+    {
+      "epoch": 0.8491076257436452,
+      "grad_norm": 0.18033969402313232,
+      "learning_rate": 9.977835749664029e-05,
+      "loss": 0.0486,
+      "step": 4710
+    },
+    {
+      "epoch": 0.8509104020191094,
+      "grad_norm": 0.17518360912799835,
+      "learning_rate": 9.97757580272996e-05,
+      "loss": 0.0498,
+      "step": 4720
+    },
+    {
+      "epoch": 0.8527131782945736,
+      "grad_norm": 0.1941109299659729,
+      "learning_rate": 9.977314343739786e-05,
+      "loss": 0.0508,
+      "step": 4730
+    },
+    {
+      "epoch": 0.8545159545700378,
+      "grad_norm": 0.1365840882062912,
+      "learning_rate": 9.977051372772934e-05,
+      "loss": 0.0469,
+      "step": 4740
+    },
+    {
+      "epoch": 0.856318730845502,
+      "grad_norm": 0.20867887139320374,
+      "learning_rate": 9.976786889909286e-05,
+      "loss": 0.0479,
+      "step": 4750
+    },
+    {
+      "epoch": 0.8581215071209662,
+      "grad_norm": 0.16011196374893188,
+      "learning_rate": 9.976520895229185e-05,
+      "loss": 0.0477,
+      "step": 4760
+    },
+    {
+      "epoch": 0.8599242833964305,
+      "grad_norm": 0.14465907216072083,
+      "learning_rate": 9.976253388813433e-05,
+      "loss": 0.0513,
+      "step": 4770
+    },
+    {
+      "epoch": 0.8617270596718947,
+      "grad_norm": 0.18770858645439148,
+      "learning_rate": 9.975984370743293e-05,
+      "loss": 0.0491,
+      "step": 4780
+    },
+    {
+      "epoch": 0.8635298359473589,
+      "grad_norm": 0.13379095494747162,
+      "learning_rate": 9.975713841100485e-05,
+      "loss": 0.0508,
+      "step": 4790
+    },
+    {
+      "epoch": 0.8653326122228231,
+      "grad_norm": 0.17976029217243195,
+      "learning_rate": 9.975441799967187e-05,
+      "loss": 0.05,
+      "step": 4800
+    },
+    {
+      "epoch": 0.8671353884982873,
+      "grad_norm": 0.1620328575372696,
+      "learning_rate": 9.975168247426039e-05,
+      "loss": 0.0445,
+      "step": 4810
+    },
+    {
+      "epoch": 0.8689381647737516,
+      "grad_norm": 0.25882643461227417,
+      "learning_rate": 9.974893183560139e-05,
+      "loss": 0.0545,
+      "step": 4820
+    },
+    {
+      "epoch": 0.8707409410492158,
+      "grad_norm": 0.1808311641216278,
+      "learning_rate": 9.974616608453045e-05,
+      "loss": 0.0558,
+      "step": 4830
+    },
+    {
+      "epoch": 0.8725437173246801,
+      "grad_norm": 0.17577724158763885,
+      "learning_rate": 9.974338522188772e-05,
+      "loss": 0.0484,
+      "step": 4840
+    },
+    {
+      "epoch": 0.8743464936001443,
+      "grad_norm": 0.2507362365722656,
+      "learning_rate": 9.974058924851797e-05,
+      "loss": 0.0499,
+      "step": 4850
+    },
+    {
+      "epoch": 0.8761492698756085,
+      "grad_norm": 0.21123144030570984,
+      "learning_rate": 9.973777816527051e-05,
+      "loss": 0.0526,
+      "step": 4860
+    },
+    {
+      "epoch": 0.8779520461510727,
+      "grad_norm": 0.18370167911052704,
+      "learning_rate": 9.973495197299931e-05,
+      "loss": 0.0465,
+      "step": 4870
+    },
+    {
+      "epoch": 0.8797548224265369,
+      "grad_norm": 0.19995036721229553,
+      "learning_rate": 9.973211067256287e-05,
+      "loss": 0.0539,
+      "step": 4880
+    },
+    {
+      "epoch": 0.8815575987020011,
+      "grad_norm": 0.16384492814540863,
+      "learning_rate": 9.97292542648243e-05,
+      "loss": 0.0487,
+      "step": 4890
+    },
+    {
+      "epoch": 0.8833603749774653,
+      "grad_norm": 0.17125661671161652,
+      "learning_rate": 9.972638275065131e-05,
+      "loss": 0.0489,
+      "step": 4900
+    },
+    {
+      "epoch": 0.8851631512529295,
+      "grad_norm": 0.14433567225933075,
+      "learning_rate": 9.972349613091621e-05,
+      "loss": 0.0468,
+      "step": 4910
+    },
+    {
+      "epoch": 0.8869659275283938,
+      "grad_norm": 0.18537750840187073,
+      "learning_rate": 9.972059440649584e-05,
+      "loss": 0.0507,
+      "step": 4920
+    },
+    {
+      "epoch": 0.888768703803858,
+      "grad_norm": 0.1987074315547943,
+      "learning_rate": 9.971767757827168e-05,
+      "loss": 0.0474,
+      "step": 4930
+    },
+    {
+      "epoch": 0.8905714800793222,
+      "grad_norm": 0.15364907681941986,
+      "learning_rate": 9.971474564712982e-05,
+      "loss": 0.0454,
+      "step": 4940
+    },
+    {
+      "epoch": 0.8923742563547864,
+      "grad_norm": 0.13974200189113617,
+      "learning_rate": 9.971179861396084e-05,
+      "loss": 0.0472,
+      "step": 4950
+    },
+    {
+      "epoch": 0.8941770326302506,
+      "grad_norm": 0.1537025272846222,
+      "learning_rate": 9.970883647966003e-05,
+      "loss": 0.0495,
+      "step": 4960
+    },
+    {
+      "epoch": 0.8959798089057148,
+      "grad_norm": 0.12050797045230865,
+      "learning_rate": 9.970585924512717e-05,
+      "loss": 0.0474,
+      "step": 4970
+    },
+    {
+      "epoch": 0.897782585181179,
+      "grad_norm": 0.17876748740673065,
+      "learning_rate": 9.970286691126669e-05,
+      "loss": 0.0482,
+      "step": 4980
+    },
+    {
+      "epoch": 0.8995853614566433,
+      "grad_norm": 0.1352316439151764,
+      "learning_rate": 9.969985947898756e-05,
+      "loss": 0.0495,
+      "step": 4990
+    },
+    {
+      "epoch": 0.9013881377321075,
+      "grad_norm": 0.17774096131324768,
+      "learning_rate": 9.969683694920337e-05,
+      "loss": 0.0494,
+      "step": 5000
+    },
+    {
+      "epoch": 0.9031909140075717,
+      "grad_norm": 0.11138828843832016,
+      "learning_rate": 9.969379932283228e-05,
+      "loss": 0.0468,
+      "step": 5010
+    },
+    {
+      "epoch": 0.9049936902830359,
+      "grad_norm": 0.14681695401668549,
+      "learning_rate": 9.969074660079704e-05,
+      "loss": 0.0425,
+      "step": 5020
+    },
+    {
+      "epoch": 0.9067964665585001,
+      "grad_norm": 0.15625491738319397,
+      "learning_rate": 9.968767878402501e-05,
+      "loss": 0.05,
+      "step": 5030
+    },
+    {
+      "epoch": 0.9085992428339643,
+      "grad_norm": 0.13043609261512756,
+      "learning_rate": 9.968459587344808e-05,
+      "loss": 0.0478,
+      "step": 5040
+    },
+    {
+      "epoch": 0.9104020191094285,
+      "grad_norm": 0.2065897434949875,
+      "learning_rate": 9.968149787000278e-05,
+      "loss": 0.0468,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9122047953848927,
+      "grad_norm": 0.17964307963848114,
+      "learning_rate": 9.967838477463018e-05,
+      "loss": 0.05,
+      "step": 5060
+    },
+    {
+      "epoch": 0.914007571660357,
+      "grad_norm": 0.1859792321920395,
+      "learning_rate": 9.967525658827597e-05,
+      "loss": 0.0506,
+      "step": 5070
+    },
+    {
+      "epoch": 0.9158103479358212,
+      "grad_norm": 0.15018241107463837,
+      "learning_rate": 9.967211331189042e-05,
+      "loss": 0.0502,
+      "step": 5080
+    },
+    {
+      "epoch": 0.9176131242112854,
+      "grad_norm": 0.17236633598804474,
+      "learning_rate": 9.966895494642834e-05,
+      "loss": 0.0472,
+      "step": 5090
+    },
+    {
+      "epoch": 0.9194159004867496,
+      "grad_norm": 0.1674015372991562,
+      "learning_rate": 9.96657814928492e-05,
+      "loss": 0.0498,
+      "step": 5100
+    },
+    {
+      "epoch": 0.9212186767622138,
+      "grad_norm": 0.1864183396100998,
+      "learning_rate": 9.966259295211697e-05,
+      "loss": 0.048,
+      "step": 5110
+    },
+    {
+      "epoch": 0.923021453037678,
+      "grad_norm": 0.18949656188488007,
+      "learning_rate": 9.965938932520028e-05,
+      "loss": 0.0522,
+      "step": 5120
+    },
+    {
+      "epoch": 0.9248242293131422,
+      "grad_norm": 0.1557510942220688,
+      "learning_rate": 9.965617061307229e-05,
+      "loss": 0.0503,
+      "step": 5130
+    },
+    {
+      "epoch": 0.9266270055886064,
+      "grad_norm": 0.19600127637386322,
+      "learning_rate": 9.965293681671077e-05,
+      "loss": 0.0507,
+      "step": 5140
+    },
+    {
+      "epoch": 0.9284297818640707,
+      "grad_norm": 0.22450041770935059,
+      "learning_rate": 9.964968793709804e-05,
+      "loss": 0.0451,
+      "step": 5150
+    },
+    {
+      "epoch": 0.9302325581395349,
+      "grad_norm": 0.18441647291183472,
+      "learning_rate": 9.964642397522106e-05,
+      "loss": 0.0552,
+      "step": 5160
+    },
+    {
+      "epoch": 0.9320353344149991,
+      "grad_norm": 0.18776282668113708,
+      "learning_rate": 9.96431449320713e-05,
+      "loss": 0.045,
+      "step": 5170
+    },
+    {
+      "epoch": 0.9338381106904633,
+      "grad_norm": 0.14953307807445526,
+      "learning_rate": 9.963985080864486e-05,
+      "loss": 0.0467,
+      "step": 5180
+    },
+    {
+      "epoch": 0.9356408869659275,
+      "grad_norm": 0.19671177864074707,
+      "learning_rate": 9.96365416059424e-05,
+      "loss": 0.0498,
+      "step": 5190
+    },
+    {
+      "epoch": 0.9374436632413917,
+      "grad_norm": 0.1434078812599182,
+      "learning_rate": 9.963321732496919e-05,
+      "loss": 0.0456,
+      "step": 5200
+    },
+    {
+      "epoch": 0.9392464395168559,
+      "grad_norm": 0.26493221521377563,
+      "learning_rate": 9.962987796673506e-05,
+      "loss": 0.0474,
+      "step": 5210
+    },
+    {
+      "epoch": 0.9410492157923201,
+      "grad_norm": 0.1555505245923996,
+      "learning_rate": 9.962652353225438e-05,
+      "loss": 0.0492,
+      "step": 5220
+    },
+    {
+      "epoch": 0.9428519920677844,
+      "grad_norm": 0.17275793850421906,
+      "learning_rate": 9.962315402254619e-05,
+      "loss": 0.0475,
+      "step": 5230
+    },
+    {
+      "epoch": 0.9446547683432486,
+      "grad_norm": 0.16553552448749542,
+      "learning_rate": 9.9619769438634e-05,
+      "loss": 0.0491,
+      "step": 5240
+    },
+    {
+      "epoch": 0.9464575446187128,
+      "grad_norm": 0.2248888462781906,
+      "learning_rate": 9.9616369781546e-05,
+      "loss": 0.0513,
+      "step": 5250
+    },
+    {
+      "epoch": 0.948260320894177,
+      "grad_norm": 0.18081867694854736,
+      "learning_rate": 9.961295505231491e-05,
+      "loss": 0.0497,
+      "step": 5260
+    },
+    {
+      "epoch": 0.9500630971696412,
+      "grad_norm": 0.24891573190689087,
+      "learning_rate": 9.960952525197804e-05,
+      "loss": 0.0513,
+      "step": 5270
+    },
+    {
+      "epoch": 0.9518658734451054,
+      "grad_norm": 0.19671489298343658,
+      "learning_rate": 9.960608038157724e-05,
+      "loss": 0.0477,
+      "step": 5280
+    },
+    {
+      "epoch": 0.9536686497205696,
+      "grad_norm": 0.23478513956069946,
+      "learning_rate": 9.960262044215901e-05,
+      "loss": 0.0521,
+      "step": 5290
+    },
+    {
+      "epoch": 0.9554714259960339,
+      "grad_norm": 0.2092088758945465,
+      "learning_rate": 9.959914543477435e-05,
+      "loss": 0.0446,
+      "step": 5300
+    },
+    {
+      "epoch": 0.9572742022714981,
+      "grad_norm": 0.14573600888252258,
+      "learning_rate": 9.959565536047892e-05,
+      "loss": 0.0448,
+      "step": 5310
+    },
+    {
+      "epoch": 0.9590769785469623,
+      "grad_norm": 0.17377443611621857,
+      "learning_rate": 9.959215022033288e-05,
+      "loss": 0.053,
+      "step": 5320
+    },
+    {
+      "epoch": 0.9608797548224265,
+      "grad_norm": 0.19106821715831757,
+      "learning_rate": 9.9588630015401e-05,
+      "loss": 0.0504,
+      "step": 5330
+    },
+    {
+      "epoch": 0.9626825310978907,
+      "grad_norm": 0.18199145793914795,
+      "learning_rate": 9.958509474675264e-05,
+      "loss": 0.0449,
+      "step": 5340
+    },
+    {
+      "epoch": 0.9644853073733549,
+      "grad_norm": 0.2071419656276703,
+      "learning_rate": 9.958154441546171e-05,
+      "loss": 0.0465,
+      "step": 5350
+    },
+    {
+      "epoch": 0.9662880836488191,
+      "grad_norm": 0.1954154074192047,
+      "learning_rate": 9.957797902260673e-05,
+      "loss": 0.0496,
+      "step": 5360
+    },
+    {
+      "epoch": 0.9680908599242833,
+      "grad_norm": 0.15930332243442535,
+      "learning_rate": 9.957439856927073e-05,
+      "loss": 0.0479,
+      "step": 5370
+    },
+    {
+      "epoch": 0.9698936361997477,
+      "grad_norm": 0.1356603056192398,
+      "learning_rate": 9.957080305654139e-05,
+      "loss": 0.0459,
+      "step": 5380
+    },
+    {
+      "epoch": 0.9716964124752119,
+      "grad_norm": 0.16535735130310059,
+      "learning_rate": 9.956719248551092e-05,
+      "loss": 0.0485,
+      "step": 5390
+    },
+    {
+      "epoch": 0.9734991887506761,
+      "grad_norm": 0.13410362601280212,
+      "learning_rate": 9.956356685727612e-05,
+      "loss": 0.0499,
+      "step": 5400
+    },
+    {
+      "epoch": 0.9753019650261403,
+      "grad_norm": 0.15987256169319153,
+      "learning_rate": 9.955992617293836e-05,
+      "loss": 0.0532,
+      "step": 5410
+    },
+    {
+      "epoch": 0.9771047413016045,
+      "grad_norm": 0.15461158752441406,
+      "learning_rate": 9.955627043360358e-05,
+      "loss": 0.0528,
+      "step": 5420
+    },
+    {
+      "epoch": 0.9789075175770687,
+      "grad_norm": 0.1482284665107727,
+      "learning_rate": 9.955259964038231e-05,
+      "loss": 0.0514,
+      "step": 5430
+    },
+    {
+      "epoch": 0.9807102938525329,
+      "grad_norm": 0.14993350207805634,
+      "learning_rate": 9.954891379438962e-05,
+      "loss": 0.0496,
+      "step": 5440
+    },
+    {
+      "epoch": 0.9825130701279972,
+      "grad_norm": 0.17867375910282135,
+      "learning_rate": 9.954521289674519e-05,
+      "loss": 0.0509,
+      "step": 5450
+    },
+    {
+      "epoch": 0.9843158464034614,
+      "grad_norm": 0.1541956067085266,
+      "learning_rate": 9.954149694857325e-05,
+      "loss": 0.0535,
+      "step": 5460
+    },
+    {
+      "epoch": 0.9861186226789256,
+      "grad_norm": 0.18034759163856506,
+      "learning_rate": 9.953776595100258e-05,
+      "loss": 0.0475,
+      "step": 5470
+    },
+    {
+      "epoch": 0.9879213989543898,
+      "grad_norm": 0.14206287264823914,
+      "learning_rate": 9.95340199051666e-05,
+      "loss": 0.0472,
+      "step": 5480
+    },
+    {
+      "epoch": 0.989724175229854,
+      "grad_norm": 0.1905004382133484,
+      "learning_rate": 9.953025881220325e-05,
+      "loss": 0.057,
+      "step": 5490
+    },
+    {
+      "epoch": 0.9915269515053182,
+      "grad_norm": 0.18126565217971802,
+      "learning_rate": 9.952648267325504e-05,
+      "loss": 0.0466,
+      "step": 5500
+    },
+    {
+      "epoch": 0.9933297277807824,
+      "grad_norm": 0.17522726953029633,
+      "learning_rate": 9.952269148946905e-05,
+      "loss": 0.0523,
+      "step": 5510
+    },
+    {
+      "epoch": 0.9951325040562466,
+      "grad_norm": 0.16590476036071777,
+      "learning_rate": 9.951888526199697e-05,
+      "loss": 0.0515,
+      "step": 5520
+    },
+    {
+      "epoch": 0.9969352803317109,
+      "grad_norm": 0.15113309025764465,
+      "learning_rate": 9.951506399199501e-05,
+      "loss": 0.0519,
+      "step": 5530
+    },
+    {
+      "epoch": 0.9987380566071751,
+      "grad_norm": 0.14152631163597107,
+      "learning_rate": 9.951122768062399e-05,
+      "loss": 0.0413,
+      "step": 5540
+    },
+    {
+      "epoch": 1.0005408328826393,
+      "grad_norm": 0.149987131357193,
+      "learning_rate": 9.950737632904927e-05,
+      "loss": 0.0499,
+      "step": 5550
+    },
+    {
+      "epoch": 1.0023436091581035,
+      "grad_norm": 0.1638782024383545,
+      "learning_rate": 9.950350993844077e-05,
+      "loss": 0.0478,
+      "step": 5560
+    },
+    {
+      "epoch": 1.0041463854335677,
+      "grad_norm": 0.12248101830482483,
+      "learning_rate": 9.949962850997303e-05,
+      "loss": 0.0437,
+      "step": 5570
+    },
+    {
+      "epoch": 1.005949161709032,
+      "grad_norm": 0.14365188777446747,
+      "learning_rate": 9.949573204482512e-05,
+      "loss": 0.0471,
+      "step": 5580
+    },
+    {
+      "epoch": 1.0077519379844961,
+      "grad_norm": 0.1796455979347229,
+      "learning_rate": 9.949182054418064e-05,
+      "loss": 0.0484,
+      "step": 5590
+    },
+    {
+      "epoch": 1.0095547142599604,
+      "grad_norm": 0.20651739835739136,
+      "learning_rate": 9.948789400922787e-05,
+      "loss": 0.0474,
+      "step": 5600
+    },
+    {
+      "epoch": 1.0113574905354246,
+      "grad_norm": 0.14700564742088318,
+      "learning_rate": 9.948395244115953e-05,
+      "loss": 0.0453,
+      "step": 5610
+    },
+    {
+      "epoch": 1.0131602668108888,
+      "grad_norm": 0.17185929417610168,
+      "learning_rate": 9.9479995841173e-05,
+      "loss": 0.0453,
+      "step": 5620
+    },
+    {
+      "epoch": 1.014963043086353,
+      "grad_norm": 0.1785033792257309,
+      "learning_rate": 9.947602421047017e-05,
+      "loss": 0.0453,
+      "step": 5630
+    },
+    {
+      "epoch": 1.0167658193618172,
+      "grad_norm": 0.13014046847820282,
+      "learning_rate": 9.947203755025753e-05,
+      "loss": 0.0495,
+      "step": 5640
+    },
+    {
+      "epoch": 1.0185685956372814,
+      "grad_norm": 0.16697810590267181,
+      "learning_rate": 9.946803586174611e-05,
+      "loss": 0.0463,
+      "step": 5650
+    },
+    {
+      "epoch": 1.0203713719127456,
+      "grad_norm": 0.16410310566425323,
+      "learning_rate": 9.946401914615151e-05,
+      "loss": 0.0475,
+      "step": 5660
+    },
+    {
+      "epoch": 1.0221741481882098,
+      "grad_norm": 0.2110091745853424,
+      "learning_rate": 9.945998740469394e-05,
+      "loss": 0.0502,
+      "step": 5670
+    },
+    {
+      "epoch": 1.023976924463674,
+      "grad_norm": 0.17573440074920654,
+      "learning_rate": 9.945594063859809e-05,
+      "loss": 0.044,
+      "step": 5680
+    },
+    {
+      "epoch": 1.0257797007391383,
+      "grad_norm": 0.17109978199005127,
+      "learning_rate": 9.94518788490933e-05,
+      "loss": 0.0477,
+      "step": 5690
+    },
+    {
+      "epoch": 1.0275824770146025,
+      "grad_norm": 0.14034892618656158,
+      "learning_rate": 9.944780203741341e-05,
+      "loss": 0.0481,
+      "step": 5700
+    },
+    {
+      "epoch": 1.0293852532900667,
+      "grad_norm": 0.11823529750108719,
+      "learning_rate": 9.944371020479686e-05,
+      "loss": 0.0418,
+      "step": 5710
+    },
+    {
+      "epoch": 1.031188029565531,
+      "grad_norm": 0.23299425840377808,
+      "learning_rate": 9.943960335248662e-05,
+      "loss": 0.0496,
+      "step": 5720
+    },
+    {
+      "epoch": 1.0329908058409951,
+      "grad_norm": 0.13195425271987915,
+      "learning_rate": 9.943548148173027e-05,
+      "loss": 0.049,
+      "step": 5730
+    },
+    {
+      "epoch": 1.0347935821164593,
+      "grad_norm": 0.20911426842212677,
+      "learning_rate": 9.943134459377992e-05,
+      "loss": 0.0392,
+      "step": 5740
+    },
+    {
+      "epoch": 1.0365963583919235,
+      "grad_norm": 0.1547815203666687,
+      "learning_rate": 9.942719268989222e-05,
+      "loss": 0.0445,
+      "step": 5750
+    },
+    {
+      "epoch": 1.0383991346673878,
+      "grad_norm": 0.16118600964546204,
+      "learning_rate": 9.942302577132844e-05,
+      "loss": 0.0501,
+      "step": 5760
+    },
+    {
+      "epoch": 1.040201910942852,
+      "grad_norm": 0.20199312269687653,
+      "learning_rate": 9.941884383935438e-05,
+      "loss": 0.0505,
+      "step": 5770
+    },
+    {
+      "epoch": 1.0420046872183162,
+      "grad_norm": 0.20491407811641693,
+      "learning_rate": 9.941464689524039e-05,
+      "loss": 0.053,
+      "step": 5780
+    },
+    {
+      "epoch": 1.0438074634937804,
+      "grad_norm": 0.12579475343227386,
+      "learning_rate": 9.941043494026139e-05,
+      "loss": 0.0468,
+      "step": 5790
+    },
+    {
+      "epoch": 1.0456102397692446,
+      "grad_norm": 0.1562582403421402,
+      "learning_rate": 9.940620797569685e-05,
+      "loss": 0.0456,
+      "step": 5800
+    },
+    {
+      "epoch": 1.0474130160447088,
+      "grad_norm": 0.16877920925617218,
+      "learning_rate": 9.940196600283082e-05,
+      "loss": 0.0446,
+      "step": 5810
+    },
+    {
+      "epoch": 1.049215792320173,
+      "grad_norm": 0.16113650798797607,
+      "learning_rate": 9.939770902295192e-05,
+      "loss": 0.0496,
+      "step": 5820
+    },
+    {
+      "epoch": 1.0510185685956372,
+      "grad_norm": 0.1584942638874054,
+      "learning_rate": 9.939343703735329e-05,
+      "loss": 0.0461,
+      "step": 5830
+    },
+    {
+      "epoch": 1.0528213448711015,
+      "grad_norm": 0.13135896623134613,
+      "learning_rate": 9.938915004733264e-05,
+      "loss": 0.047,
+      "step": 5840
+    },
+    {
+      "epoch": 1.0546241211465657,
+      "grad_norm": 0.15665724873542786,
+      "learning_rate": 9.938484805419224e-05,
+      "loss": 0.0436,
+      "step": 5850
+    },
+    {
+      "epoch": 1.0564268974220299,
+      "grad_norm": 0.12544065713882446,
+      "learning_rate": 9.938053105923894e-05,
+      "loss": 0.0459,
+      "step": 5860
+    },
+    {
+      "epoch": 1.058229673697494,
+      "grad_norm": 0.18433156609535217,
+      "learning_rate": 9.937619906378413e-05,
+      "loss": 0.0469,
+      "step": 5870
+    },
+    {
+      "epoch": 1.0600324499729583,
+      "grad_norm": 0.1148960292339325,
+      "learning_rate": 9.937185206914374e-05,
+      "loss": 0.045,
+      "step": 5880
+    },
+    {
+      "epoch": 1.0618352262484225,
+      "grad_norm": 0.14018428325653076,
+      "learning_rate": 9.936749007663829e-05,
+      "loss": 0.0452,
+      "step": 5890
+    },
+    {
+      "epoch": 1.0636380025238867,
+      "grad_norm": 0.11849283427000046,
+      "learning_rate": 9.93631130875928e-05,
+      "loss": 0.0471,
+      "step": 5900
+    },
+    {
+      "epoch": 1.065440778799351,
+      "grad_norm": 0.22733385860919952,
+      "learning_rate": 9.935872110333692e-05,
+      "loss": 0.0418,
+      "step": 5910
+    },
+    {
+      "epoch": 1.0672435550748152,
+      "grad_norm": 0.1406257450580597,
+      "learning_rate": 9.935431412520484e-05,
+      "loss": 0.0428,
+      "step": 5920
+    },
+    {
+      "epoch": 1.0690463313502794,
+      "grad_norm": 0.1938510537147522,
+      "learning_rate": 9.934989215453523e-05,
+      "loss": 0.0502,
+      "step": 5930
+    },
+    {
+      "epoch": 1.0708491076257436,
+      "grad_norm": 0.15984466671943665,
+      "learning_rate": 9.934545519267139e-05,
+      "loss": 0.0409,
+      "step": 5940
+    },
+    {
+      "epoch": 1.0726518839012078,
+      "grad_norm": 0.1253848671913147,
+      "learning_rate": 9.934100324096117e-05,
+      "loss": 0.0517,
+      "step": 5950
+    },
+    {
+      "epoch": 1.074454660176672,
+      "grad_norm": 0.24231602251529694,
+      "learning_rate": 9.933653630075692e-05,
+      "loss": 0.0446,
+      "step": 5960
+    },
+    {
+      "epoch": 1.0762574364521362,
+      "grad_norm": 0.14123372733592987,
+      "learning_rate": 9.93320543734156e-05,
+      "loss": 0.0454,
+      "step": 5970
+    },
+    {
+      "epoch": 1.0780602127276004,
+      "grad_norm": 0.18040312826633453,
+      "learning_rate": 9.932755746029871e-05,
+      "loss": 0.0506,
+      "step": 5980
+    },
+    {
+      "epoch": 1.0798629890030647,
+      "grad_norm": 0.16003407537937164,
+      "learning_rate": 9.932304556277228e-05,
+      "loss": 0.0462,
+      "step": 5990
+    },
+    {
+      "epoch": 1.0816657652785289,
+      "grad_norm": 0.14502501487731934,
+      "learning_rate": 9.93185186822069e-05,
+      "loss": 0.0534,
+      "step": 6000
+    },
+    {
+      "epoch": 1.083468541553993,
+      "grad_norm": 0.14464087784290314,
+      "learning_rate": 9.931397681997773e-05,
+      "loss": 0.0426,
+      "step": 6010
+    },
+    {
+      "epoch": 1.0852713178294573,
+      "grad_norm": 0.18443164229393005,
+      "learning_rate": 9.930941997746446e-05,
+      "loss": 0.0471,
+      "step": 6020
+    },
+    {
+      "epoch": 1.0870740941049215,
+      "grad_norm": 0.11963239312171936,
+      "learning_rate": 9.930484815605134e-05,
+      "loss": 0.0457,
+      "step": 6030
+    },
+    {
+      "epoch": 1.0888768703803857,
+      "grad_norm": 0.14174722135066986,
+      "learning_rate": 9.930026135712717e-05,
+      "loss": 0.0453,
+      "step": 6040
+    },
+    {
+      "epoch": 1.09067964665585,
+      "grad_norm": 0.17676784098148346,
+      "learning_rate": 9.92956595820853e-05,
+      "loss": 0.0456,
+      "step": 6050
+    },
+    {
+      "epoch": 1.0924824229313141,
+      "grad_norm": 0.15777680277824402,
+      "learning_rate": 9.929104283232362e-05,
+      "loss": 0.0476,
+      "step": 6060
+    },
+    {
+      "epoch": 1.0942851992067784,
+      "grad_norm": 0.17012213170528412,
+      "learning_rate": 9.92864111092446e-05,
+      "loss": 0.0441,
+      "step": 6070
+    },
+    {
+      "epoch": 1.0960879754822426,
+      "grad_norm": 0.21466368436813354,
+      "learning_rate": 9.92817644142552e-05,
+      "loss": 0.0451,
+      "step": 6080
+    },
+    {
+      "epoch": 1.0978907517577068,
+      "grad_norm": 0.16265292465686798,
+      "learning_rate": 9.927710274876698e-05,
+      "loss": 0.0434,
+      "step": 6090
+    },
+    {
+      "epoch": 1.099693528033171,
+      "grad_norm": 0.2022707313299179,
+      "learning_rate": 9.927242611419603e-05,
+      "loss": 0.045,
+      "step": 6100
+    },
+    {
+      "epoch": 1.1014963043086352,
+      "grad_norm": 0.2022763192653656,
+      "learning_rate": 9.926773451196301e-05,
+      "loss": 0.0482,
+      "step": 6110
+    },
+    {
+      "epoch": 1.1032990805840994,
+      "grad_norm": 0.17981594800949097,
+      "learning_rate": 9.926302794349306e-05,
+      "loss": 0.047,
+      "step": 6120
+    },
+    {
+      "epoch": 1.1051018568595636,
+      "grad_norm": 0.18052178621292114,
+      "learning_rate": 9.925830641021594e-05,
+      "loss": 0.0444,
+      "step": 6130
+    },
+    {
+      "epoch": 1.1069046331350278,
+      "grad_norm": 0.19416724145412445,
+      "learning_rate": 9.925356991356593e-05,
+      "loss": 0.0459,
+      "step": 6140
+    },
+    {
+      "epoch": 1.108707409410492,
+      "grad_norm": 0.19651128351688385,
+      "learning_rate": 9.924881845498184e-05,
+      "loss": 0.0415,
+      "step": 6150
+    },
+    {
+      "epoch": 1.1105101856859563,
+      "grad_norm": 0.17051537334918976,
+      "learning_rate": 9.924405203590705e-05,
+      "loss": 0.0445,
+      "step": 6160
+    },
+    {
+      "epoch": 1.1123129619614205,
+      "grad_norm": 0.15216341614723206,
+      "learning_rate": 9.923927065778946e-05,
+      "loss": 0.042,
+      "step": 6170
+    },
+    {
+      "epoch": 1.1141157382368847,
+      "grad_norm": 0.1388617604970932,
+      "learning_rate": 9.923447432208154e-05,
+      "loss": 0.0455,
+      "step": 6180
+    },
+    {
+      "epoch": 1.115918514512349,
+      "grad_norm": 0.14735934138298035,
+      "learning_rate": 9.922966303024027e-05,
+      "loss": 0.0443,
+      "step": 6190
+    },
+    {
+      "epoch": 1.1177212907878133,
+      "grad_norm": 0.15511681139469147,
+      "learning_rate": 9.922483678372721e-05,
+      "loss": 0.0414,
+      "step": 6200
+    },
+    {
+      "epoch": 1.1195240670632773,
+      "grad_norm": 0.1833156943321228,
+      "learning_rate": 9.921999558400845e-05,
+      "loss": 0.0489,
+      "step": 6210
+    },
+    {
+      "epoch": 1.1213268433387418,
+      "grad_norm": 0.15968109667301178,
+      "learning_rate": 9.92151394325546e-05,
+      "loss": 0.047,
+      "step": 6220
+    },
+    {
+      "epoch": 1.1231296196142058,
+      "grad_norm": 0.12436918169260025,
+      "learning_rate": 9.921026833084084e-05,
+      "loss": 0.0478,
+      "step": 6230
+    },
+    {
+      "epoch": 1.1249323958896702,
+      "grad_norm": 0.1411399096250534,
+      "learning_rate": 9.920538228034689e-05,
+      "loss": 0.0454,
+      "step": 6240
+    },
+    {
+      "epoch": 1.1267351721651342,
+      "grad_norm": 0.1757981926202774,
+      "learning_rate": 9.920048128255699e-05,
+      "loss": 0.0449,
+      "step": 6250
+    },
+    {
+      "epoch": 1.1285379484405986,
+      "grad_norm": 0.14229696989059448,
+      "learning_rate": 9.919556533895995e-05,
+      "loss": 0.0433,
+      "step": 6260
+    },
+    {
+      "epoch": 1.1303407247160626,
+      "grad_norm": 0.199496790766716,
+      "learning_rate": 9.919063445104907e-05,
+      "loss": 0.0456,
+      "step": 6270
+    },
+    {
+      "epoch": 1.132143500991527,
+      "grad_norm": 0.15765699744224548,
+      "learning_rate": 9.918568862032227e-05,
+      "loss": 0.0509,
+      "step": 6280
+    },
+    {
+      "epoch": 1.133946277266991,
+      "grad_norm": 0.20775863528251648,
+      "learning_rate": 9.918072784828194e-05,
+      "loss": 0.0419,
+      "step": 6290
+    },
+    {
+      "epoch": 1.1357490535424555,
+      "grad_norm": 0.13819508254528046,
+      "learning_rate": 9.917575213643501e-05,
+      "loss": 0.0465,
+      "step": 6300
+    },
+    {
+      "epoch": 1.1375518298179197,
+      "grad_norm": 0.17002363502979279,
+      "learning_rate": 9.917076148629302e-05,
+      "loss": 0.0449,
+      "step": 6310
+    },
+    {
+      "epoch": 1.139354606093384,
+      "grad_norm": 0.16458258032798767,
+      "learning_rate": 9.916575589937196e-05,
+      "loss": 0.0446,
+      "step": 6320
+    },
+    {
+      "epoch": 1.1411573823688481,
+      "grad_norm": 0.2087150514125824,
+      "learning_rate": 9.916073537719239e-05,
+      "loss": 0.0476,
+      "step": 6330
+    },
+    {
+      "epoch": 1.1429601586443123,
+      "grad_norm": 0.20809558033943176,
+      "learning_rate": 9.915569992127944e-05,
+      "loss": 0.049,
+      "step": 6340
+    },
+    {
+      "epoch": 1.1447629349197765,
+      "grad_norm": 0.17088796198368073,
+      "learning_rate": 9.915064953316273e-05,
+      "loss": 0.0401,
+      "step": 6350
+    },
+    {
+      "epoch": 1.1465657111952408,
+      "grad_norm": 0.18331198394298553,
+      "learning_rate": 9.914558421437645e-05,
+      "loss": 0.0422,
+      "step": 6360
+    },
+    {
+      "epoch": 1.148368487470705,
+      "grad_norm": 0.1356252282857895,
+      "learning_rate": 9.914050396645929e-05,
+      "loss": 0.0431,
+      "step": 6370
+    },
+    {
+      "epoch": 1.1501712637461692,
+      "grad_norm": 0.15957140922546387,
+      "learning_rate": 9.913540879095452e-05,
+      "loss": 0.0454,
+      "step": 6380
+    },
+    {
+      "epoch": 1.1519740400216334,
+      "grad_norm": 0.16391707956790924,
+      "learning_rate": 9.913029868940987e-05,
+      "loss": 0.0462,
+      "step": 6390
+    },
+    {
+      "epoch": 1.1537768162970976,
+      "grad_norm": 0.1389417201280594,
+      "learning_rate": 9.912517366337772e-05,
+      "loss": 0.0442,
+      "step": 6400
+    },
+    {
+      "epoch": 1.1555795925725618,
+      "grad_norm": 0.15863144397735596,
+      "learning_rate": 9.912003371441487e-05,
+      "loss": 0.0477,
+      "step": 6410
+    },
+    {
+      "epoch": 1.157382368848026,
+      "grad_norm": 0.1406799852848053,
+      "learning_rate": 9.911487884408271e-05,
+      "loss": 0.0483,
+      "step": 6420
+    },
+    {
+      "epoch": 1.1591851451234902,
+      "grad_norm": 0.1771913468837738,
+      "learning_rate": 9.910970905394719e-05,
+      "loss": 0.0467,
+      "step": 6430
+    },
+    {
+      "epoch": 1.1609879213989545,
+      "grad_norm": 0.19251134991645813,
+      "learning_rate": 9.91045243455787e-05,
+      "loss": 0.0451,
+      "step": 6440
+    },
+    {
+      "epoch": 1.1627906976744187,
+      "grad_norm": 0.1681923270225525,
+      "learning_rate": 9.909932472055225e-05,
+      "loss": 0.0473,
+      "step": 6450
+    },
+    {
+      "epoch": 1.1645934739498829,
+      "grad_norm": 0.1934024542570114,
+      "learning_rate": 9.909411018044734e-05,
+      "loss": 0.0452,
+      "step": 6460
+    },
+    {
+      "epoch": 1.166396250225347,
+      "grad_norm": 0.19895634055137634,
+      "learning_rate": 9.908888072684802e-05,
+      "loss": 0.0463,
+      "step": 6470
+    },
+    {
+      "epoch": 1.1681990265008113,
+      "grad_norm": 0.20210351049900055,
+      "learning_rate": 9.908363636134285e-05,
+      "loss": 0.0479,
+      "step": 6480
+    },
+    {
+      "epoch": 1.1700018027762755,
+      "grad_norm": 0.15985043346881866,
+      "learning_rate": 9.907837708552493e-05,
+      "loss": 0.0452,
+      "step": 6490
+    },
+    {
+      "epoch": 1.1718045790517397,
+      "grad_norm": 0.10954516381025314,
+      "learning_rate": 9.90731029009919e-05,
+      "loss": 0.0429,
+      "step": 6500
+    },
+    {
+      "epoch": 1.173607355327204,
+      "grad_norm": 0.2081098109483719,
+      "learning_rate": 9.906781380934589e-05,
+      "loss": 0.0466,
+      "step": 6510
+    },
+    {
+      "epoch": 1.1754101316026682,
+      "grad_norm": 0.16629405319690704,
+      "learning_rate": 9.906250981219362e-05,
+      "loss": 0.04,
+      "step": 6520
+    },
+    {
+      "epoch": 1.1772129078781324,
+      "grad_norm": 0.19642026722431183,
+      "learning_rate": 9.905719091114628e-05,
+      "loss": 0.0498,
+      "step": 6530
+    },
+    {
+      "epoch": 1.1790156841535966,
+      "grad_norm": 0.23067142069339752,
+      "learning_rate": 9.905185710781964e-05,
+      "loss": 0.0527,
+      "step": 6540
+    },
+    {
+      "epoch": 1.1808184604290608,
+      "grad_norm": 0.16386303305625916,
+      "learning_rate": 9.904650840383392e-05,
+      "loss": 0.0477,
+      "step": 6550
+    },
+    {
+      "epoch": 1.182621236704525,
+      "grad_norm": 0.1786251962184906,
+      "learning_rate": 9.904114480081397e-05,
+      "loss": 0.0454,
+      "step": 6560
+    },
+    {
+      "epoch": 1.1844240129799892,
+      "grad_norm": 0.1505107283592224,
+      "learning_rate": 9.903576630038906e-05,
+      "loss": 0.0435,
+      "step": 6570
+    },
+    {
+      "epoch": 1.1862267892554534,
+      "grad_norm": 0.11944922804832458,
+      "learning_rate": 9.903037290419309e-05,
+      "loss": 0.0453,
+      "step": 6580
+    },
+    {
+      "epoch": 1.1880295655309177,
+      "grad_norm": 0.13519299030303955,
+      "learning_rate": 9.902496461386439e-05,
+      "loss": 0.0522,
+      "step": 6590
+    },
+    {
+      "epoch": 1.1898323418063819,
+      "grad_norm": 0.17141962051391602,
+      "learning_rate": 9.901954143104588e-05,
+      "loss": 0.0429,
+      "step": 6600
+    },
+    {
+      "epoch": 1.191635118081846,
+      "grad_norm": 0.15148812532424927,
+      "learning_rate": 9.901410335738496e-05,
+      "loss": 0.0475,
+      "step": 6610
+    },
+    {
+      "epoch": 1.1934378943573103,
+      "grad_norm": 0.1400941014289856,
+      "learning_rate": 9.900865039453358e-05,
+      "loss": 0.0487,
+      "step": 6620
+    },
+    {
+      "epoch": 1.1952406706327745,
+      "grad_norm": 0.22592294216156006,
+      "learning_rate": 9.900318254414821e-05,
+      "loss": 0.0445,
+      "step": 6630
+    },
+    {
+      "epoch": 1.1970434469082387,
+      "grad_norm": 0.1541605442762375,
+      "learning_rate": 9.899769980788985e-05,
+      "loss": 0.041,
+      "step": 6640
+    },
+    {
+      "epoch": 1.198846223183703,
+      "grad_norm": 0.15684810280799866,
+      "learning_rate": 9.899220218742398e-05,
+      "loss": 0.0481,
+      "step": 6650
+    },
+    {
+      "epoch": 1.2006489994591671,
+      "grad_norm": 0.13748125731945038,
+      "learning_rate": 9.898668968442066e-05,
+      "loss": 0.0518,
+      "step": 6660
+    },
+    {
+      "epoch": 1.2024517757346314,
+      "grad_norm": 0.17705634236335754,
+      "learning_rate": 9.898116230055443e-05,
+      "loss": 0.0485,
+      "step": 6670
+    },
+    {
+      "epoch": 1.2042545520100956,
+      "grad_norm": 0.13691218197345734,
+      "learning_rate": 9.897562003750437e-05,
+      "loss": 0.0488,
+      "step": 6680
+    },
+    {
+      "epoch": 1.2060573282855598,
+      "grad_norm": 0.14242900907993317,
+      "learning_rate": 9.897006289695407e-05,
+      "loss": 0.0433,
+      "step": 6690
+    },
+    {
+      "epoch": 1.207860104561024,
+      "grad_norm": 0.23364323377609253,
+      "learning_rate": 9.896449088059164e-05,
+      "loss": 0.0468,
+      "step": 6700
+    },
+    {
+      "epoch": 1.2096628808364882,
+      "grad_norm": 0.15275025367736816,
+      "learning_rate": 9.89589039901097e-05,
+      "loss": 0.0476,
+      "step": 6710
+    },
+    {
+      "epoch": 1.2114656571119524,
+      "grad_norm": 0.14401565492153168,
+      "learning_rate": 9.895330222720542e-05,
+      "loss": 0.0446,
+      "step": 6720
+    },
+    {
+      "epoch": 1.2132684333874166,
+      "grad_norm": 0.13415250182151794,
+      "learning_rate": 9.894768559358047e-05,
+      "loss": 0.0482,
+      "step": 6730
+    },
+    {
+      "epoch": 1.2150712096628808,
+      "grad_norm": 0.12522150576114655,
+      "learning_rate": 9.894205409094101e-05,
+      "loss": 0.0435,
+      "step": 6740
+    },
+    {
+      "epoch": 1.216873985938345,
+      "grad_norm": 0.13278326392173767,
+      "learning_rate": 9.893640772099777e-05,
+      "loss": 0.0506,
+      "step": 6750
+    },
+    {
+      "epoch": 1.2186767622138093,
+      "grad_norm": 0.2089739739894867,
+      "learning_rate": 9.893074648546595e-05,
+      "loss": 0.0458,
+      "step": 6760
+    },
+    {
+      "epoch": 1.2204795384892735,
+      "grad_norm": 0.14211158454418182,
+      "learning_rate": 9.892507038606528e-05,
+      "loss": 0.0479,
+      "step": 6770
+    },
+    {
+      "epoch": 1.2222823147647377,
+      "grad_norm": 0.20254890620708466,
+      "learning_rate": 9.891937942452003e-05,
+      "loss": 0.0434,
+      "step": 6780
+    },
+    {
+      "epoch": 1.224085091040202,
+      "grad_norm": 0.14579035341739655,
+      "learning_rate": 9.891367360255895e-05,
+      "loss": 0.0479,
+      "step": 6790
+    },
+    {
+      "epoch": 1.2258878673156661,
+      "grad_norm": 0.19807976484298706,
+      "learning_rate": 9.890795292191532e-05,
+      "loss": 0.0502,
+      "step": 6800
+    },
+    {
+      "epoch": 1.2276906435911303,
+      "grad_norm": 0.1949124038219452,
+      "learning_rate": 9.890221738432694e-05,
+      "loss": 0.0431,
+      "step": 6810
+    },
+    {
+      "epoch": 1.2294934198665946,
+      "grad_norm": 0.14246895909309387,
+      "learning_rate": 9.88964669915361e-05,
+      "loss": 0.0416,
+      "step": 6820
+    },
+    {
+      "epoch": 1.2312961961420588,
+      "grad_norm": 0.20334585011005402,
+      "learning_rate": 9.889070174528963e-05,
+      "loss": 0.0479,
+      "step": 6830
+    },
+    {
+      "epoch": 1.233098972417523,
+      "grad_norm": 0.20166416466236115,
+      "learning_rate": 9.888492164733883e-05,
+      "loss": 0.0452,
+      "step": 6840
+    },
+    {
+      "epoch": 1.2349017486929872,
+      "grad_norm": 0.17774198949337006,
+      "learning_rate": 9.88791266994396e-05,
+      "loss": 0.0432,
+      "step": 6850
+    },
+    {
+      "epoch": 1.2367045249684514,
+      "grad_norm": 0.15698379278182983,
+      "learning_rate": 9.887331690335223e-05,
+      "loss": 0.0472,
+      "step": 6860
+    },
+    {
+      "epoch": 1.2385073012439156,
+      "grad_norm": 0.15600785613059998,
+      "learning_rate": 9.886749226084163e-05,
+      "loss": 0.0352,
+      "step": 6870
+    },
+    {
+      "epoch": 1.2403100775193798,
+      "grad_norm": 0.18297724425792694,
+      "learning_rate": 9.886165277367714e-05,
+      "loss": 0.0469,
+      "step": 6880
+    },
+    {
+      "epoch": 1.242112853794844,
+      "grad_norm": 0.13390932977199554,
+      "learning_rate": 9.885579844363265e-05,
+      "loss": 0.0419,
+      "step": 6890
+    },
+    {
+      "epoch": 1.2439156300703083,
+      "grad_norm": 0.15496668219566345,
+      "learning_rate": 9.884992927248656e-05,
+      "loss": 0.0481,
+      "step": 6900
+    },
+    {
+      "epoch": 1.2457184063457725,
+      "grad_norm": 0.1544051617383957,
+      "learning_rate": 9.884404526202178e-05,
+      "loss": 0.0448,
+      "step": 6910
+    },
+    {
+      "epoch": 1.2475211826212367,
+      "grad_norm": 0.12542957067489624,
+      "learning_rate": 9.883814641402568e-05,
+      "loss": 0.0448,
+      "step": 6920
+    },
+    {
+      "epoch": 1.249323958896701,
+      "grad_norm": 0.17185518145561218,
+      "learning_rate": 9.88322327302902e-05,
+      "loss": 0.0509,
+      "step": 6930
+    },
+    {
+      "epoch": 1.251126735172165,
+      "grad_norm": 0.13775959610939026,
+      "learning_rate": 9.882630421261176e-05,
+      "loss": 0.0433,
+      "step": 6940
+    },
+    {
+      "epoch": 1.2529295114476293,
+      "grad_norm": 0.10725994408130646,
+      "learning_rate": 9.88203608627913e-05,
+      "loss": 0.0421,
+      "step": 6950
+    },
+    {
+      "epoch": 1.2547322877230935,
+      "grad_norm": 0.13983775675296783,
+      "learning_rate": 9.881440268263422e-05,
+      "loss": 0.0444,
+      "step": 6960
+    },
+    {
+      "epoch": 1.2565350639985577,
+      "grad_norm": 0.15295752882957458,
+      "learning_rate": 9.880842967395048e-05,
+      "loss": 0.0423,
+      "step": 6970
+    },
+    {
+      "epoch": 1.258337840274022,
+      "grad_norm": 0.14794494211673737,
+      "learning_rate": 9.880244183855452e-05,
+      "loss": 0.041,
+      "step": 6980
+    },
+    {
+      "epoch": 1.2601406165494862,
+      "grad_norm": 0.10952970385551453,
+      "learning_rate": 9.879643917826527e-05,
+      "loss": 0.0478,
+      "step": 6990
+    },
+    {
+      "epoch": 1.2619433928249504,
+      "grad_norm": 0.1916574239730835,
+      "learning_rate": 9.87904216949062e-05,
+      "loss": 0.0486,
+      "step": 7000
+    },
+    {
+      "epoch": 1.2637461691004146,
+      "grad_norm": 0.15800578892230988,
+      "learning_rate": 9.878438939030526e-05,
+      "loss": 0.0422,
+      "step": 7010
+    },
+    {
+      "epoch": 1.2655489453758788,
+      "grad_norm": 0.16109836101531982,
+      "learning_rate": 9.877834226629489e-05,
+      "loss": 0.045,
+      "step": 7020
+    },
+    {
+      "epoch": 1.267351721651343,
+      "grad_norm": 0.15866237878799438,
+      "learning_rate": 9.877228032471206e-05,
+      "loss": 0.0399,
+      "step": 7030
+    },
+    {
+      "epoch": 1.2691544979268072,
+      "grad_norm": 0.16993622481822968,
+      "learning_rate": 9.876620356739823e-05,
+      "loss": 0.0409,
+      "step": 7040
+    },
+    {
+      "epoch": 1.2709572742022714,
+      "grad_norm": 0.14860498905181885,
+      "learning_rate": 9.876011199619935e-05,
+      "loss": 0.0463,
+      "step": 7050
+    },
+    {
+      "epoch": 1.2727600504777357,
+      "grad_norm": 0.16061100363731384,
+      "learning_rate": 9.875400561296589e-05,
+      "loss": 0.0418,
+      "step": 7060
+    },
+    {
+      "epoch": 1.2745628267531999,
+      "grad_norm": 0.11456069350242615,
+      "learning_rate": 9.874788441955278e-05,
+      "loss": 0.0464,
+      "step": 7070
+    },
+    {
+      "epoch": 1.276365603028664,
+      "grad_norm": 0.1331043392419815,
+      "learning_rate": 9.874174841781951e-05,
+      "loss": 0.0437,
+      "step": 7080
+    },
+    {
+      "epoch": 1.2781683793041283,
+      "grad_norm": 0.1368376463651657,
+      "learning_rate": 9.873559760963003e-05,
+      "loss": 0.0425,
+      "step": 7090
+    },
+    {
+      "epoch": 1.2799711555795925,
+      "grad_norm": 0.13278715312480927,
+      "learning_rate": 9.872943199685278e-05,
+      "loss": 0.043,
+      "step": 7100
+    },
+    {
+      "epoch": 1.2817739318550567,
+      "grad_norm": 0.13766273856163025,
+      "learning_rate": 9.872325158136071e-05,
+      "loss": 0.0436,
+      "step": 7110
+    },
+    {
+      "epoch": 1.283576708130521,
+      "grad_norm": 0.23797397315502167,
+      "learning_rate": 9.871705636503128e-05,
+      "loss": 0.0385,
+      "step": 7120
+    },
+    {
+      "epoch": 1.2853794844059852,
+      "grad_norm": 0.12689067423343658,
+      "learning_rate": 9.871084634974641e-05,
+      "loss": 0.043,
+      "step": 7130
+    },
+    {
+      "epoch": 1.2871822606814494,
+      "grad_norm": 0.21763867139816284,
+      "learning_rate": 9.870462153739257e-05,
+      "loss": 0.0434,
+      "step": 7140
+    },
+    {
+      "epoch": 1.2889850369569136,
+      "grad_norm": 0.10984236001968384,
+      "learning_rate": 9.869838192986067e-05,
+      "loss": 0.0415,
+      "step": 7150
+    },
+    {
+      "epoch": 1.2907878132323778,
+      "grad_norm": 0.1933307945728302,
+      "learning_rate": 9.869212752904616e-05,
+      "loss": 0.0444,
+      "step": 7160
+    },
+    {
+      "epoch": 1.292590589507842,
+      "grad_norm": 0.16871808469295502,
+      "learning_rate": 9.868585833684894e-05,
+      "loss": 0.0408,
+      "step": 7170
+    },
+    {
+      "epoch": 1.2943933657833062,
+      "grad_norm": 0.11162350326776505,
+      "learning_rate": 9.867957435517342e-05,
+      "loss": 0.0425,
+      "step": 7180
+    },
+    {
+      "epoch": 1.2961961420587704,
+      "grad_norm": 0.17216373980045319,
+      "learning_rate": 9.867327558592854e-05,
+      "loss": 0.0426,
+      "step": 7190
+    },
+    {
+      "epoch": 1.2979989183342346,
+      "grad_norm": 0.22052639722824097,
+      "learning_rate": 9.866696203102766e-05,
+      "loss": 0.0401,
+      "step": 7200
+    },
+    {
+      "epoch": 1.2998016946096989,
+      "grad_norm": 0.16249202191829681,
+      "learning_rate": 9.86606336923887e-05,
+      "loss": 0.042,
+      "step": 7210
+    },
+    {
+      "epoch": 1.301604470885163,
+      "grad_norm": 0.1349180042743683,
+      "learning_rate": 9.865429057193403e-05,
+      "loss": 0.0427,
+      "step": 7220
+    },
+    {
+      "epoch": 1.3034072471606275,
+      "grad_norm": 0.15774761140346527,
+      "learning_rate": 9.864793267159053e-05,
+      "loss": 0.0443,
+      "step": 7230
+    },
+    {
+      "epoch": 1.3052100234360915,
+      "grad_norm": 0.18511465191841125,
+      "learning_rate": 9.864155999328957e-05,
+      "loss": 0.0454,
+      "step": 7240
+    },
+    {
+      "epoch": 1.307012799711556,
+      "grad_norm": 0.13638466596603394,
+      "learning_rate": 9.8635172538967e-05,
+      "loss": 0.0471,
+      "step": 7250
+    },
+    {
+      "epoch": 1.30881557598702,
+      "grad_norm": 0.14766816794872284,
+      "learning_rate": 9.862877031056312e-05,
+      "loss": 0.0417,
+      "step": 7260
+    },
+    {
+      "epoch": 1.3106183522624844,
+      "grad_norm": 0.13629595935344696,
+      "learning_rate": 9.862235331002279e-05,
+      "loss": 0.042,
+      "step": 7270
+    },
+    {
+      "epoch": 1.3124211285379483,
+      "grad_norm": 0.1394883245229721,
+      "learning_rate": 9.861592153929533e-05,
+      "loss": 0.044,
+      "step": 7280
+    },
+    {
+      "epoch": 1.3142239048134128,
+      "grad_norm": 0.16391190886497498,
+      "learning_rate": 9.860947500033455e-05,
+      "loss": 0.0448,
+      "step": 7290
+    },
+    {
+      "epoch": 1.3160266810888768,
+      "grad_norm": 0.11683732271194458,
+      "learning_rate": 9.86030136950987e-05,
+      "loss": 0.0455,
+      "step": 7300
+    },
+    {
+      "epoch": 1.3178294573643412,
+      "grad_norm": 0.17655129730701447,
+      "learning_rate": 9.85965376255506e-05,
+      "loss": 0.0451,
+      "step": 7310
+    },
+    {
+      "epoch": 1.3196322336398052,
+      "grad_norm": 0.16985571384429932,
+      "learning_rate": 9.859004679365747e-05,
+      "loss": 0.0415,
+      "step": 7320
+    },
+    {
+      "epoch": 1.3214350099152696,
+      "grad_norm": 0.20975105464458466,
+      "learning_rate": 9.858354120139108e-05,
+      "loss": 0.0448,
+      "step": 7330
+    },
+    {
+      "epoch": 1.3232377861907336,
+      "grad_norm": 0.13291749358177185,
+      "learning_rate": 9.857702085072764e-05,
+      "loss": 0.0452,
+      "step": 7340
+    },
+    {
+      "epoch": 1.325040562466198,
+      "grad_norm": 0.20697391033172607,
+      "learning_rate": 9.857048574364787e-05,
+      "loss": 0.0442,
+      "step": 7350
+    },
+    {
+      "epoch": 1.326843338741662,
+      "grad_norm": 0.15438371896743774,
+      "learning_rate": 9.856393588213698e-05,
+      "loss": 0.0454,
+      "step": 7360
+    },
+    {
+      "epoch": 1.3286461150171265,
+      "grad_norm": 0.17755204439163208,
+      "learning_rate": 9.855737126818458e-05,
+      "loss": 0.0413,
+      "step": 7370
+    },
+    {
+      "epoch": 1.3304488912925905,
+      "grad_norm": 0.14504413306713104,
+      "learning_rate": 9.855079190378491e-05,
+      "loss": 0.0383,
+      "step": 7380
+    },
+    {
+      "epoch": 1.332251667568055,
+      "grad_norm": 0.13375242054462433,
+      "learning_rate": 9.854419779093655e-05,
+      "loss": 0.0437,
+      "step": 7390
+    },
+    {
+      "epoch": 1.334054443843519,
+      "grad_norm": 0.11517435312271118,
+      "learning_rate": 9.853758893164264e-05,
+      "loss": 0.045,
+      "step": 7400
+    },
+    {
+      "epoch": 1.3358572201189833,
+      "grad_norm": 0.15008369088172913,
+      "learning_rate": 9.853096532791078e-05,
+      "loss": 0.0479,
+      "step": 7410
+    },
+    {
+      "epoch": 1.3376599963944473,
+      "grad_norm": 0.22088518738746643,
+      "learning_rate": 9.852432698175304e-05,
+      "loss": 0.0442,
+      "step": 7420
+    },
+    {
+      "epoch": 1.3394627726699118,
+      "grad_norm": 0.15875470638275146,
+      "learning_rate": 9.851767389518597e-05,
+      "loss": 0.047,
+      "step": 7430
+    },
+    {
+      "epoch": 1.3412655489453758,
+      "grad_norm": 0.16996069252490997,
+      "learning_rate": 9.85110060702306e-05,
+      "loss": 0.0431,
+      "step": 7440
+    },
+    {
+      "epoch": 1.3430683252208402,
+      "grad_norm": 0.17854364216327667,
+      "learning_rate": 9.850432350891245e-05,
+      "loss": 0.0453,
+      "step": 7450
+    },
+    {
+      "epoch": 1.3448711014963042,
+      "grad_norm": 0.17406010627746582,
+      "learning_rate": 9.84976262132615e-05,
+      "loss": 0.0448,
+      "step": 7460
+    },
+    {
+      "epoch": 1.3466738777717686,
+      "grad_norm": 0.16046424210071564,
+      "learning_rate": 9.849091418531222e-05,
+      "loss": 0.042,
+      "step": 7470
+    },
+    {
+      "epoch": 1.3484766540472326,
+      "grad_norm": 0.17490343749523163,
+      "learning_rate": 9.848418742710353e-05,
+      "loss": 0.0432,
+      "step": 7480
+    },
+    {
+      "epoch": 1.350279430322697,
+      "grad_norm": 0.19233691692352295,
+      "learning_rate": 9.847744594067885e-05,
+      "loss": 0.0411,
+      "step": 7490
+    },
+    {
+      "epoch": 1.3520822065981613,
+      "grad_norm": 0.1614624410867691,
+      "learning_rate": 9.847068972808607e-05,
+      "loss": 0.0428,
+      "step": 7500
+    },
+    {
+      "epoch": 1.3538849828736255,
+      "grad_norm": 0.1238013207912445,
+      "learning_rate": 9.846391879137756e-05,
+      "loss": 0.042,
+      "step": 7510
+    },
+    {
+      "epoch": 1.3556877591490897,
+      "grad_norm": 0.15535993874073029,
+      "learning_rate": 9.845713313261012e-05,
+      "loss": 0.0459,
+      "step": 7520
+    },
+    {
+      "epoch": 1.357490535424554,
+      "grad_norm": 0.1513490229845047,
+      "learning_rate": 9.845033275384505e-05,
+      "loss": 0.0447,
+      "step": 7530
+    },
+    {
+      "epoch": 1.359293311700018,
+      "grad_norm": 0.1245562881231308,
+      "learning_rate": 9.844351765714818e-05,
+      "loss": 0.0487,
+      "step": 7540
+    },
+    {
+      "epoch": 1.3610960879754823,
+      "grad_norm": 0.2179003357887268,
+      "learning_rate": 9.843668784458971e-05,
+      "loss": 0.0454,
+      "step": 7550
+    },
+    {
+      "epoch": 1.3628988642509465,
+      "grad_norm": 0.17515775561332703,
+      "learning_rate": 9.842984331824437e-05,
+      "loss": 0.0427,
+      "step": 7560
+    },
+    {
+      "epoch": 1.3647016405264107,
+      "grad_norm": 0.1258850246667862,
+      "learning_rate": 9.842298408019133e-05,
+      "loss": 0.0424,
+      "step": 7570
+    },
+    {
+      "epoch": 1.366504416801875,
+      "grad_norm": 0.155188649892807,
+      "learning_rate": 9.841611013251429e-05,
+      "loss": 0.046,
+      "step": 7580
+    },
+    {
+      "epoch": 1.3683071930773392,
+      "grad_norm": 0.15004678070545197,
+      "learning_rate": 9.840922147730133e-05,
+      "loss": 0.043,
+      "step": 7590
+    },
+    {
+      "epoch": 1.3701099693528034,
+      "grad_norm": 0.13265766203403473,
+      "learning_rate": 9.840231811664506e-05,
+      "loss": 0.0504,
+      "step": 7600
+    },
+    {
+      "epoch": 1.3719127456282676,
+      "grad_norm": 0.15950389206409454,
+      "learning_rate": 9.839540005264252e-05,
+      "loss": 0.0455,
+      "step": 7610
+    },
+    {
+      "epoch": 1.3737155219037318,
+      "grad_norm": 0.12739747762680054,
+      "learning_rate": 9.838846728739527e-05,
+      "loss": 0.0406,
+      "step": 7620
+    },
+    {
+      "epoch": 1.375518298179196,
+      "grad_norm": 0.15198072791099548,
+      "learning_rate": 9.838151982300927e-05,
+      "loss": 0.0393,
+      "step": 7630
+    },
+    {
+      "epoch": 1.3773210744546602,
+      "grad_norm": 0.15584009885787964,
+      "learning_rate": 9.8374557661595e-05,
+      "loss": 0.0365,
+      "step": 7640
+    },
+    {
+      "epoch": 1.3791238507301244,
+      "grad_norm": 0.20010541379451752,
+      "learning_rate": 9.836758080526735e-05,
+      "loss": 0.0461,
+      "step": 7650
+    },
+    {
+      "epoch": 1.3809266270055887,
+      "grad_norm": 0.14808468520641327,
+      "learning_rate": 9.836058925614575e-05,
+      "loss": 0.0409,
+      "step": 7660
+    },
+    {
+      "epoch": 1.3827294032810529,
+      "grad_norm": 0.19547396898269653,
+      "learning_rate": 9.8353583016354e-05,
+      "loss": 0.0441,
+      "step": 7670
+    },
+    {
+      "epoch": 1.384532179556517,
+      "grad_norm": 0.1321982890367508,
+      "learning_rate": 9.834656208802044e-05,
+      "loss": 0.0471,
+      "step": 7680
+    },
+    {
+      "epoch": 1.3863349558319813,
+      "grad_norm": 0.15596811473369598,
+      "learning_rate": 9.833952647327784e-05,
+      "loss": 0.049,
+      "step": 7690
+    },
+    {
+      "epoch": 1.3881377321074455,
+      "grad_norm": 0.1391909122467041,
+      "learning_rate": 9.833247617426342e-05,
+      "loss": 0.0451,
+      "step": 7700
+    },
+    {
+      "epoch": 1.3899405083829097,
+      "grad_norm": 0.1682356894016266,
+      "learning_rate": 9.832541119311889e-05,
+      "loss": 0.0447,
+      "step": 7710
+    },
+    {
+      "epoch": 1.391743284658374,
+      "grad_norm": 0.15864479541778564,
+      "learning_rate": 9.83183315319904e-05,
+      "loss": 0.0457,
+      "step": 7720
+    },
+    {
+      "epoch": 1.3935460609338381,
+      "grad_norm": 0.1691732257604599,
+      "learning_rate": 9.831123719302855e-05,
+      "loss": 0.0416,
+      "step": 7730
+    },
+    {
+      "epoch": 1.3953488372093024,
+      "grad_norm": 0.1667153537273407,
+      "learning_rate": 9.830412817838842e-05,
+      "loss": 0.043,
+      "step": 7740
+    },
+    {
+      "epoch": 1.3971516134847666,
+      "grad_norm": 0.14120270311832428,
+      "learning_rate": 9.829700449022956e-05,
+      "loss": 0.0436,
+      "step": 7750
+    },
+    {
+      "epoch": 1.3989543897602308,
+      "grad_norm": 0.23471084237098694,
+      "learning_rate": 9.828986613071593e-05,
+      "loss": 0.0416,
+      "step": 7760
+    },
+    {
+      "epoch": 1.400757166035695,
+      "grad_norm": 0.187713623046875,
+      "learning_rate": 9.828271310201601e-05,
+      "loss": 0.046,
+      "step": 7770
+    },
+    {
+      "epoch": 1.4025599423111592,
+      "grad_norm": 0.14881864190101624,
+      "learning_rate": 9.827554540630268e-05,
+      "loss": 0.0395,
+      "step": 7780
+    },
+    {
+      "epoch": 1.4043627185866234,
+      "grad_norm": 0.1740046590566635,
+      "learning_rate": 9.826836304575329e-05,
+      "loss": 0.0431,
+      "step": 7790
+    },
+    {
+      "epoch": 1.4061654948620876,
+      "grad_norm": 0.11747528612613678,
+      "learning_rate": 9.826116602254966e-05,
+      "loss": 0.0442,
+      "step": 7800
+    },
+    {
+      "epoch": 1.4079682711375519,
+      "grad_norm": 0.1879766881465912,
+      "learning_rate": 9.825395433887805e-05,
+      "loss": 0.0442,
+      "step": 7810
+    },
+    {
+      "epoch": 1.409771047413016,
+      "grad_norm": 0.13589905202388763,
+      "learning_rate": 9.824672799692917e-05,
+      "loss": 0.0441,
+      "step": 7820
+    },
+    {
+      "epoch": 1.4115738236884803,
+      "grad_norm": 0.12330379337072372,
+      "learning_rate": 9.823948699889823e-05,
+      "loss": 0.0449,
+      "step": 7830
+    },
+    {
+      "epoch": 1.4133765999639445,
+      "grad_norm": 0.16740384697914124,
+      "learning_rate": 9.823223134698483e-05,
+      "loss": 0.0416,
+      "step": 7840
+    },
+    {
+      "epoch": 1.4151793762394087,
+      "grad_norm": 0.1375618875026703,
+      "learning_rate": 9.822496104339303e-05,
+      "loss": 0.0428,
+      "step": 7850
+    },
+    {
+      "epoch": 1.416982152514873,
+      "grad_norm": 0.1519220471382141,
+      "learning_rate": 9.821767609033138e-05,
+      "loss": 0.0421,
+      "step": 7860
+    },
+    {
+      "epoch": 1.4187849287903371,
+      "grad_norm": 0.15725424885749817,
+      "learning_rate": 9.821037649001284e-05,
+      "loss": 0.0427,
+      "step": 7870
+    },
+    {
+      "epoch": 1.4205877050658013,
+      "grad_norm": 0.10759121179580688,
+      "learning_rate": 9.820306224465486e-05,
+      "loss": 0.0432,
+      "step": 7880
+    },
+    {
+      "epoch": 1.4223904813412656,
+      "grad_norm": 0.16982968151569366,
+      "learning_rate": 9.819573335647928e-05,
+      "loss": 0.044,
+      "step": 7890
+    },
+    {
+      "epoch": 1.4241932576167298,
+      "grad_norm": 0.09863346070051193,
+      "learning_rate": 9.818838982771246e-05,
+      "loss": 0.0457,
+      "step": 7900
+    },
+    {
+      "epoch": 1.425996033892194,
+      "grad_norm": 0.10342240333557129,
+      "learning_rate": 9.818103166058514e-05,
+      "loss": 0.0499,
+      "step": 7910
+    },
+    {
+      "epoch": 1.4277988101676582,
+      "grad_norm": 0.10769131779670715,
+      "learning_rate": 9.817365885733254e-05,
+      "loss": 0.0411,
+      "step": 7920
+    },
+    {
+      "epoch": 1.4296015864431224,
+      "grad_norm": 0.192804753780365,
+      "learning_rate": 9.816627142019434e-05,
+      "loss": 0.0413,
+      "step": 7930
+    },
+    {
+      "epoch": 1.4314043627185866,
+      "grad_norm": 0.13700097799301147,
+      "learning_rate": 9.815886935141463e-05,
+      "loss": 0.046,
+      "step": 7940
+    },
+    {
+      "epoch": 1.4332071389940508,
+      "grad_norm": 0.11457335948944092,
+      "learning_rate": 9.8151452653242e-05,
+      "loss": 0.0453,
+      "step": 7950
+    },
+    {
+      "epoch": 1.435009915269515,
+      "grad_norm": 0.13416250050067902,
+      "learning_rate": 9.814402132792939e-05,
+      "loss": 0.0408,
+      "step": 7960
+    },
+    {
+      "epoch": 1.4368126915449793,
+      "grad_norm": 0.20427556335926056,
+      "learning_rate": 9.813657537773428e-05,
+      "loss": 0.0444,
+      "step": 7970
+    },
+    {
+      "epoch": 1.4386154678204435,
+      "grad_norm": 0.21200846135616302,
+      "learning_rate": 9.812911480491854e-05,
+      "loss": 0.0454,
+      "step": 7980
+    },
+    {
+      "epoch": 1.4404182440959077,
+      "grad_norm": 0.13624627888202667,
+      "learning_rate": 9.81216396117485e-05,
+      "loss": 0.0393,
+      "step": 7990
+    },
+    {
+      "epoch": 1.442221020371372,
+      "grad_norm": 0.20008771121501923,
+      "learning_rate": 9.811414980049491e-05,
+      "loss": 0.0443,
+      "step": 8000
+    },
+    {
+      "epoch": 1.444023796646836,
+      "grad_norm": 0.13469800353050232,
+      "learning_rate": 9.810664537343301e-05,
+      "loss": 0.0445,
+      "step": 8010
+    },
+    {
+      "epoch": 1.4458265729223003,
+      "grad_norm": 0.12516558170318604,
+      "learning_rate": 9.809912633284243e-05,
+      "loss": 0.0441,
+      "step": 8020
+    },
+    {
+      "epoch": 1.4476293491977645,
+      "grad_norm": 0.12844984233379364,
+      "learning_rate": 9.809159268100725e-05,
+      "loss": 0.046,
+      "step": 8030
+    },
+    {
+      "epoch": 1.4494321254732287,
+      "grad_norm": 0.12819664180278778,
+      "learning_rate": 9.808404442021599e-05,
+      "loss": 0.0418,
+      "step": 8040
+    },
+    {
+      "epoch": 1.451234901748693,
+      "grad_norm": 0.12778475880622864,
+      "learning_rate": 9.807648155276163e-05,
+      "loss": 0.0407,
+      "step": 8050
+    },
+    {
+      "epoch": 1.4530376780241572,
+      "grad_norm": 0.16056707501411438,
+      "learning_rate": 9.806890408094156e-05,
+      "loss": 0.0444,
+      "step": 8060
+    },
+    {
+      "epoch": 1.4548404542996214,
+      "grad_norm": 0.09849455207586288,
+      "learning_rate": 9.806131200705761e-05,
+      "loss": 0.0389,
+      "step": 8070
+    },
+    {
+      "epoch": 1.4566432305750856,
+      "grad_norm": 0.09808798879384995,
+      "learning_rate": 9.805370533341605e-05,
+      "loss": 0.0444,
+      "step": 8080
+    },
+    {
+      "epoch": 1.4584460068505498,
+      "grad_norm": 0.13618935644626617,
+      "learning_rate": 9.804608406232762e-05,
+      "loss": 0.0418,
+      "step": 8090
+    },
+    {
+      "epoch": 1.460248783126014,
+      "grad_norm": 0.14985764026641846,
+      "learning_rate": 9.803844819610741e-05,
+      "loss": 0.0401,
+      "step": 8100
+    },
+    {
+      "epoch": 1.4620515594014782,
+      "grad_norm": 0.23928481340408325,
+      "learning_rate": 9.803079773707504e-05,
+      "loss": 0.0411,
+      "step": 8110
+    },
+    {
+      "epoch": 1.4638543356769425,
+      "grad_norm": 0.10376619547605515,
+      "learning_rate": 9.802313268755447e-05,
+      "loss": 0.0393,
+      "step": 8120
+    },
+    {
+      "epoch": 1.4656571119524067,
+      "grad_norm": 0.1542768031358719,
+      "learning_rate": 9.801545304987419e-05,
+      "loss": 0.0423,
+      "step": 8130
+    },
+    {
+      "epoch": 1.4674598882278709,
+      "grad_norm": 0.14499475061893463,
+      "learning_rate": 9.800775882636704e-05,
+      "loss": 0.0437,
+      "step": 8140
+    },
+    {
+      "epoch": 1.469262664503335,
+      "grad_norm": 0.2057196944952011,
+      "learning_rate": 9.800005001937034e-05,
+      "loss": 0.0423,
+      "step": 8150
+    },
+    {
+      "epoch": 1.4710654407787993,
+      "grad_norm": 0.14442934095859528,
+      "learning_rate": 9.79923266312258e-05,
+      "loss": 0.0425,
+      "step": 8160
+    },
+    {
+      "epoch": 1.4728682170542635,
+      "grad_norm": 0.1957738995552063,
+      "learning_rate": 9.79845886642796e-05,
+      "loss": 0.0365,
+      "step": 8170
+    },
+    {
+      "epoch": 1.4746709933297277,
+      "grad_norm": 0.1717921495437622,
+      "learning_rate": 9.797683612088233e-05,
+      "loss": 0.0427,
+      "step": 8180
+    },
+    {
+      "epoch": 1.476473769605192,
+      "grad_norm": 0.15808957815170288,
+      "learning_rate": 9.796906900338898e-05,
+      "loss": 0.0393,
+      "step": 8190
+    },
+    {
+      "epoch": 1.4782765458806562,
+      "grad_norm": 0.19438597559928894,
+      "learning_rate": 9.796128731415903e-05,
+      "loss": 0.0435,
+      "step": 8200
+    },
+    {
+      "epoch": 1.4800793221561204,
+      "grad_norm": 0.19207032024860382,
+      "learning_rate": 9.795349105555634e-05,
+      "loss": 0.0401,
+      "step": 8210
+    },
+    {
+      "epoch": 1.4818820984315846,
+      "grad_norm": 0.1565093696117401,
+      "learning_rate": 9.794568022994922e-05,
+      "loss": 0.0408,
+      "step": 8220
+    },
+    {
+      "epoch": 1.4836848747070488,
+      "grad_norm": 0.15701228380203247,
+      "learning_rate": 9.793785483971034e-05,
+      "loss": 0.0442,
+      "step": 8230
+    },
+    {
+      "epoch": 1.485487650982513,
+      "grad_norm": 0.1439363807439804,
+      "learning_rate": 9.793001488721691e-05,
+      "loss": 0.0508,
+      "step": 8240
+    },
+    {
+      "epoch": 1.4872904272579772,
+      "grad_norm": 0.11535283923149109,
+      "learning_rate": 9.792216037485047e-05,
+      "loss": 0.0414,
+      "step": 8250
+    },
+    {
+      "epoch": 1.4890932035334414,
+      "grad_norm": 0.12835390865802765,
+      "learning_rate": 9.791429130499704e-05,
+      "loss": 0.047,
+      "step": 8260
+    },
+    {
+      "epoch": 1.4908959798089056,
+      "grad_norm": 0.1851019561290741,
+      "learning_rate": 9.790640768004698e-05,
+      "loss": 0.0443,
+      "step": 8270
+    },
+    {
+      "epoch": 1.4926987560843699,
+      "grad_norm": 0.14740526676177979,
+      "learning_rate": 9.789850950239518e-05,
+      "loss": 0.0412,
+      "step": 8280
+    },
+    {
+      "epoch": 1.494501532359834,
+      "grad_norm": 0.1528676599264145,
+      "learning_rate": 9.789059677444089e-05,
+      "loss": 0.0397,
+      "step": 8290
+    },
+    {
+      "epoch": 1.4963043086352983,
+      "grad_norm": 0.18221311271190643,
+      "learning_rate": 9.788266949858776e-05,
+      "loss": 0.0432,
+      "step": 8300
+    },
+    {
+      "epoch": 1.4981070849107625,
+      "grad_norm": 0.12342915683984756,
+      "learning_rate": 9.787472767724392e-05,
+      "loss": 0.0407,
+      "step": 8310
+    },
+    {
+      "epoch": 1.4999098611862267,
+      "grad_norm": 0.12390574812889099,
+      "learning_rate": 9.786677131282185e-05,
+      "loss": 0.0432,
+      "step": 8320
+    },
+    {
+      "epoch": 1.501712637461691,
+      "grad_norm": 0.13368622958660126,
+      "learning_rate": 9.785880040773853e-05,
+      "loss": 0.0414,
+      "step": 8330
+    },
+    {
+      "epoch": 1.5035154137371554,
+      "grad_norm": 0.17940254509449005,
+      "learning_rate": 9.785081496441527e-05,
+      "loss": 0.0423,
+      "step": 8340
+    },
+    {
+      "epoch": 1.5053181900126194,
+      "grad_norm": 0.1522243469953537,
+      "learning_rate": 9.784281498527785e-05,
+      "loss": 0.044,
+      "step": 8350
+    },
+    {
+      "epoch": 1.5071209662880838,
+      "grad_norm": 0.17423155903816223,
+      "learning_rate": 9.783480047275646e-05,
+      "loss": 0.0486,
+      "step": 8360
+    },
+    {
+      "epoch": 1.5089237425635478,
+      "grad_norm": 0.20396697521209717,
+      "learning_rate": 9.78267714292857e-05,
+      "loss": 0.0418,
+      "step": 8370
+    },
+    {
+      "epoch": 1.5107265188390122,
+      "grad_norm": 0.1652018427848816,
+      "learning_rate": 9.781872785730454e-05,
+      "loss": 0.0402,
+      "step": 8380
+    },
+    {
+      "epoch": 1.5125292951144762,
+      "grad_norm": 0.13339237868785858,
+      "learning_rate": 9.781066975925646e-05,
+      "loss": 0.0439,
+      "step": 8390
+    },
+    {
+      "epoch": 1.5143320713899406,
+      "grad_norm": 0.14576540887355804,
+      "learning_rate": 9.780259713758928e-05,
+      "loss": 0.0398,
+      "step": 8400
+    },
+    {
+      "epoch": 1.5161348476654046,
+      "grad_norm": 0.11785120517015457,
+      "learning_rate": 9.779450999475524e-05,
+      "loss": 0.0409,
+      "step": 8410
+    },
+    {
+      "epoch": 1.517937623940869,
+      "grad_norm": 0.13513927161693573,
+      "learning_rate": 9.7786408333211e-05,
+      "loss": 0.0454,
+      "step": 8420
+    },
+    {
+      "epoch": 1.519740400216333,
+      "grad_norm": 0.12213265895843506,
+      "learning_rate": 9.777829215541764e-05,
+      "loss": 0.0424,
+      "step": 8430
+    },
+    {
+      "epoch": 1.5215431764917975,
+      "grad_norm": 0.15755802392959595,
+      "learning_rate": 9.777016146384064e-05,
+      "loss": 0.0433,
+      "step": 8440
+    },
+    {
+      "epoch": 1.5233459527672615,
+      "grad_norm": 0.21495431661605835,
+      "learning_rate": 9.776201626094988e-05,
+      "loss": 0.0406,
+      "step": 8450
+    },
+    {
+      "epoch": 1.525148729042726,
+      "grad_norm": 0.19009213149547577,
+      "learning_rate": 9.775385654921965e-05,
+      "loss": 0.0404,
+      "step": 8460
+    },
+    {
+      "epoch": 1.52695150531819,
+      "grad_norm": 0.12882180511951447,
+      "learning_rate": 9.774568233112868e-05,
+      "loss": 0.0384,
+      "step": 8470
+    },
+    {
+      "epoch": 1.5287542815936543,
+      "grad_norm": 0.11674314737319946,
+      "learning_rate": 9.773749360916007e-05,
+      "loss": 0.04,
+      "step": 8480
+    },
+    {
+      "epoch": 1.5305570578691183,
+      "grad_norm": 0.13835161924362183,
+      "learning_rate": 9.772929038580134e-05,
+      "loss": 0.0409,
+      "step": 8490
+    },
+    {
+      "epoch": 1.5323598341445828,
+      "grad_norm": 0.15066763758659363,
+      "learning_rate": 9.772107266354439e-05,
+      "loss": 0.0451,
+      "step": 8500
+    },
+    {
+      "epoch": 1.5341626104200468,
+      "grad_norm": 0.1178978905081749,
+      "learning_rate": 9.77128404448856e-05,
+      "loss": 0.0434,
+      "step": 8510
+    },
+    {
+      "epoch": 1.5359653866955112,
+      "grad_norm": 0.10479331016540527,
+      "learning_rate": 9.770459373232565e-05,
+      "loss": 0.0425,
+      "step": 8520
+    },
+    {
+      "epoch": 1.5377681629709752,
+      "grad_norm": 0.1205153539776802,
+      "learning_rate": 9.769633252836969e-05,
+      "loss": 0.0437,
+      "step": 8530
+    },
+    {
+      "epoch": 1.5395709392464396,
+      "grad_norm": 0.20385809242725372,
+      "learning_rate": 9.768805683552724e-05,
+      "loss": 0.0457,
+      "step": 8540
+    },
+    {
+      "epoch": 1.5413737155219036,
+      "grad_norm": 0.15422265231609344,
+      "learning_rate": 9.767976665631228e-05,
+      "loss": 0.0435,
+      "step": 8550
+    },
+    {
+      "epoch": 1.543176491797368,
+      "grad_norm": 0.1509687304496765,
+      "learning_rate": 9.767146199324311e-05,
+      "loss": 0.0442,
+      "step": 8560
+    },
+    {
+      "epoch": 1.544979268072832,
+      "grad_norm": 0.1412591189146042,
+      "learning_rate": 9.766314284884249e-05,
+      "loss": 0.0407,
+      "step": 8570
+    },
+    {
+      "epoch": 1.5467820443482965,
+      "grad_norm": 0.1641855239868164,
+      "learning_rate": 9.765480922563752e-05,
+      "loss": 0.0409,
+      "step": 8580
+    },
+    {
+      "epoch": 1.5485848206237605,
+      "grad_norm": 0.17688889801502228,
+      "learning_rate": 9.764646112615978e-05,
+      "loss": 0.0451,
+      "step": 8590
+    },
+    {
+      "epoch": 1.550387596899225,
+      "grad_norm": 0.10193276405334473,
+      "learning_rate": 9.763809855294517e-05,
+      "loss": 0.0424,
+      "step": 8600
+    },
+    {
+      "epoch": 1.5521903731746889,
+      "grad_norm": 0.1310090571641922,
+      "learning_rate": 9.762972150853404e-05,
+      "loss": 0.0442,
+      "step": 8610
+    },
+    {
+      "epoch": 1.5539931494501533,
+      "grad_norm": 0.15439288318157196,
+      "learning_rate": 9.762132999547111e-05,
+      "loss": 0.046,
+      "step": 8620
+    },
+    {
+      "epoch": 1.5557959257256173,
+      "grad_norm": 0.14456555247306824,
+      "learning_rate": 9.761292401630549e-05,
+      "loss": 0.039,
+      "step": 8630
+    },
+    {
+      "epoch": 1.5575987020010817,
+      "grad_norm": 0.14294250309467316,
+      "learning_rate": 9.76045035735907e-05,
+      "loss": 0.0408,
+      "step": 8640
+    },
+    {
+      "epoch": 1.5594014782765457,
+      "grad_norm": 0.17914031445980072,
+      "learning_rate": 9.759606866988464e-05,
+      "loss": 0.0387,
+      "step": 8650
+    },
+    {
+      "epoch": 1.5612042545520102,
+      "grad_norm": 0.14985167980194092,
+      "learning_rate": 9.758761930774963e-05,
+      "loss": 0.0477,
+      "step": 8660
+    },
+    {
+      "epoch": 1.5630070308274742,
+      "grad_norm": 0.15756645798683167,
+      "learning_rate": 9.757915548975235e-05,
+      "loss": 0.044,
+      "step": 8670
+    },
+    {
+      "epoch": 1.5648098071029386,
+      "grad_norm": 0.14678172767162323,
+      "learning_rate": 9.757067721846389e-05,
+      "loss": 0.0432,
+      "step": 8680
+    },
+    {
+      "epoch": 1.5666125833784026,
+      "grad_norm": 0.14178459346294403,
+      "learning_rate": 9.756218449645971e-05,
+      "loss": 0.0516,
+      "step": 8690
+    },
+    {
+      "epoch": 1.568415359653867,
+      "grad_norm": 0.1318591982126236,
+      "learning_rate": 9.75536773263197e-05,
+      "loss": 0.0439,
+      "step": 8700
+    },
+    {
+      "epoch": 1.570218135929331,
+      "grad_norm": 0.13863953948020935,
+      "learning_rate": 9.75451557106281e-05,
+      "loss": 0.045,
+      "step": 8710
+    },
+    {
+      "epoch": 1.5720209122047955,
+      "grad_norm": 0.12420926988124847,
+      "learning_rate": 9.753661965197354e-05,
+      "loss": 0.0447,
+      "step": 8720
+    },
+    {
+      "epoch": 1.5738236884802594,
+      "grad_norm": 0.12888164818286896,
+      "learning_rate": 9.752806915294908e-05,
+      "loss": 0.0453,
+      "step": 8730
+    },
+    {
+      "epoch": 1.5756264647557239,
+      "grad_norm": 0.15761391818523407,
+      "learning_rate": 9.75195042161521e-05,
+      "loss": 0.0452,
+      "step": 8740
+    },
+    {
+      "epoch": 1.5774292410311879,
+      "grad_norm": 0.15469135344028473,
+      "learning_rate": 9.751092484418442e-05,
+      "loss": 0.0446,
+      "step": 8750
+    },
+    {
+      "epoch": 1.5792320173066523,
+      "grad_norm": 0.1579580157995224,
+      "learning_rate": 9.750233103965224e-05,
+      "loss": 0.0391,
+      "step": 8760
+    },
+    {
+      "epoch": 1.5810347935821163,
+      "grad_norm": 0.1366501897573471,
+      "learning_rate": 9.749372280516611e-05,
+      "loss": 0.0447,
+      "step": 8770
+    },
+    {
+      "epoch": 1.5828375698575807,
+      "grad_norm": 0.1659681797027588,
+      "learning_rate": 9.748510014334097e-05,
+      "loss": 0.0429,
+      "step": 8780
+    },
+    {
+      "epoch": 1.584640346133045,
+      "grad_norm": 0.1764449030160904,
+      "learning_rate": 9.747646305679621e-05,
+      "loss": 0.0459,
+      "step": 8790
+    },
+    {
+      "epoch": 1.5864431224085092,
+      "grad_norm": 0.11895448714494705,
+      "learning_rate": 9.74678115481555e-05,
+      "loss": 0.0413,
+      "step": 8800
+    },
+    {
+      "epoch": 1.5882458986839734,
+      "grad_norm": 0.12485641241073608,
+      "learning_rate": 9.745914562004696e-05,
+      "loss": 0.0378,
+      "step": 8810
+    },
+    {
+      "epoch": 1.5900486749594376,
+      "grad_norm": 0.12429939210414886,
+      "learning_rate": 9.745046527510307e-05,
+      "loss": 0.0409,
+      "step": 8820
+    },
+    {
+      "epoch": 1.5918514512349018,
+      "grad_norm": 0.09228527545928955,
+      "learning_rate": 9.744177051596068e-05,
+      "loss": 0.0352,
+      "step": 8830
+    },
+    {
+      "epoch": 1.593654227510366,
+      "grad_norm": 0.1651102751493454,
+      "learning_rate": 9.743306134526105e-05,
+      "loss": 0.0393,
+      "step": 8840
+    },
+    {
+      "epoch": 1.5954570037858302,
+      "grad_norm": 0.16719184815883636,
+      "learning_rate": 9.742433776564977e-05,
+      "loss": 0.0374,
+      "step": 8850
+    },
+    {
+      "epoch": 1.5972597800612944,
+      "grad_norm": 0.11218594759702682,
+      "learning_rate": 9.741559977977683e-05,
+      "loss": 0.0379,
+      "step": 8860
+    },
+    {
+      "epoch": 1.5990625563367586,
+      "grad_norm": 0.13786685466766357,
+      "learning_rate": 9.740684739029661e-05,
+      "loss": 0.0379,
+      "step": 8870
+    },
+    {
+      "epoch": 1.6008653326122229,
+      "grad_norm": 0.11107755452394485,
+      "learning_rate": 9.739808059986789e-05,
+      "loss": 0.0394,
+      "step": 8880
+    },
+    {
+      "epoch": 1.602668108887687,
+      "grad_norm": 0.14831335842609406,
+      "learning_rate": 9.738929941115373e-05,
+      "loss": 0.0425,
+      "step": 8890
+    },
+    {
+      "epoch": 1.6044708851631513,
+      "grad_norm": 0.15752890706062317,
+      "learning_rate": 9.738050382682167e-05,
+      "loss": 0.0442,
+      "step": 8900
+    },
+    {
+      "epoch": 1.6062736614386155,
+      "grad_norm": 0.15082195401191711,
+      "learning_rate": 9.737169384954355e-05,
+      "loss": 0.0453,
+      "step": 8910
+    },
+    {
+      "epoch": 1.6080764377140797,
+      "grad_norm": 0.13747678697109222,
+      "learning_rate": 9.736286948199562e-05,
+      "loss": 0.0423,
+      "step": 8920
+    },
+    {
+      "epoch": 1.609879213989544,
+      "grad_norm": 0.13459938764572144,
+      "learning_rate": 9.735403072685848e-05,
+      "loss": 0.0417,
+      "step": 8930
+    },
+    {
+      "epoch": 1.6116819902650081,
+      "grad_norm": 0.1419130563735962,
+      "learning_rate": 9.734517758681712e-05,
+      "loss": 0.039,
+      "step": 8940
+    },
+    {
+      "epoch": 1.6134847665404723,
+      "grad_norm": 0.1557479202747345,
+      "learning_rate": 9.733631006456088e-05,
+      "loss": 0.0357,
+      "step": 8950
+    },
+    {
+      "epoch": 1.6152875428159366,
+      "grad_norm": 0.1271543949842453,
+      "learning_rate": 9.732742816278348e-05,
+      "loss": 0.0417,
+      "step": 8960
+    },
+    {
+      "epoch": 1.6170903190914008,
+      "grad_norm": 0.1513073593378067,
+      "learning_rate": 9.731853188418302e-05,
+      "loss": 0.0414,
+      "step": 8970
+    },
+    {
+      "epoch": 1.618893095366865,
+      "grad_norm": 0.10693933814764023,
+      "learning_rate": 9.730962123146194e-05,
+      "loss": 0.0389,
+      "step": 8980
+    },
+    {
+      "epoch": 1.6206958716423292,
+      "grad_norm": 0.133611798286438,
+      "learning_rate": 9.730069620732709e-05,
+      "loss": 0.0411,
+      "step": 8990
+    },
+    {
+      "epoch": 1.6224986479177934,
+      "grad_norm": 0.15051952004432678,
+      "learning_rate": 9.72917568144896e-05,
+      "loss": 0.0431,
+      "step": 9000
+    },
+    {
+      "epoch": 1.6243014241932576,
+      "grad_norm": 0.21103809773921967,
+      "learning_rate": 9.728280305566509e-05,
+      "loss": 0.0435,
+      "step": 9010
+    },
+    {
+      "epoch": 1.6261042004687218,
+      "grad_norm": 0.08502990752458572,
+      "learning_rate": 9.727383493357343e-05,
+      "loss": 0.0394,
+      "step": 9020
+    },
+    {
+      "epoch": 1.627906976744186,
+      "grad_norm": 0.1304304301738739,
+      "learning_rate": 9.726485245093891e-05,
+      "loss": 0.042,
+      "step": 9030
+    },
+    {
+      "epoch": 1.6297097530196503,
+      "grad_norm": 0.15076182782649994,
+      "learning_rate": 9.725585561049018e-05,
+      "loss": 0.0434,
+      "step": 9040
+    },
+    {
+      "epoch": 1.6315125292951145,
+      "grad_norm": 0.12900467216968536,
+      "learning_rate": 9.724684441496022e-05,
+      "loss": 0.0382,
+      "step": 9050
+    },
+    {
+      "epoch": 1.6333153055705787,
+      "grad_norm": 0.14301662147045135,
+      "learning_rate": 9.72378188670864e-05,
+      "loss": 0.0404,
+      "step": 9060
+    },
+    {
+      "epoch": 1.635118081846043,
+      "grad_norm": 0.12155018746852875,
+      "learning_rate": 9.722877896961047e-05,
+      "loss": 0.0418,
+      "step": 9070
+    },
+    {
+      "epoch": 1.6369208581215071,
+      "grad_norm": 0.12534725666046143,
+      "learning_rate": 9.721972472527848e-05,
+      "loss": 0.0448,
+      "step": 9080
+    },
+    {
+      "epoch": 1.6387236343969713,
+      "grad_norm": 0.1575479507446289,
+      "learning_rate": 9.721065613684089e-05,
+      "loss": 0.0425,
+      "step": 9090
+    },
+    {
+      "epoch": 1.6405264106724355,
+      "grad_norm": 0.132240891456604,
+      "learning_rate": 9.72015732070525e-05,
+      "loss": 0.0414,
+      "step": 9100
+    },
+    {
+      "epoch": 1.6423291869478998,
+      "grad_norm": 0.16293112933635712,
+      "learning_rate": 9.719247593867244e-05,
+      "loss": 0.0438,
+      "step": 9110
+    },
+    {
+      "epoch": 1.644131963223364,
+      "grad_norm": 0.18946024775505066,
+      "learning_rate": 9.718336433446423e-05,
+      "loss": 0.04,
+      "step": 9120
+    },
+    {
+      "epoch": 1.6459347394988282,
+      "grad_norm": 0.1284693032503128,
+      "learning_rate": 9.717423839719574e-05,
+      "loss": 0.043,
+      "step": 9130
+    },
+    {
+      "epoch": 1.6477375157742924,
+      "grad_norm": 0.20048880577087402,
+      "learning_rate": 9.71650981296392e-05,
+      "loss": 0.0397,
+      "step": 9140
+    },
+    {
+      "epoch": 1.6495402920497566,
+      "grad_norm": 0.1629897505044937,
+      "learning_rate": 9.715594353457118e-05,
+      "loss": 0.0431,
+      "step": 9150
+    },
+    {
+      "epoch": 1.6513430683252208,
+      "grad_norm": 0.17216765880584717,
+      "learning_rate": 9.714677461477257e-05,
+      "loss": 0.0455,
+      "step": 9160
+    },
+    {
+      "epoch": 1.653145844600685,
+      "grad_norm": 0.12067907303571701,
+      "learning_rate": 9.713759137302869e-05,
+      "loss": 0.0435,
+      "step": 9170
+    },
+    {
+      "epoch": 1.6549486208761492,
+      "grad_norm": 0.17365257441997528,
+      "learning_rate": 9.712839381212914e-05,
+      "loss": 0.0446,
+      "step": 9180
+    },
+    {
+      "epoch": 1.6567513971516135,
+      "grad_norm": 0.12032567709684372,
+      "learning_rate": 9.71191819348679e-05,
+      "loss": 0.0421,
+      "step": 9190
+    },
+    {
+      "epoch": 1.6585541734270777,
+      "grad_norm": 0.1644958257675171,
+      "learning_rate": 9.710995574404331e-05,
+      "loss": 0.0452,
+      "step": 9200
+    },
+    {
+      "epoch": 1.6603569497025419,
+      "grad_norm": 0.1338106095790863,
+      "learning_rate": 9.710071524245802e-05,
+      "loss": 0.0417,
+      "step": 9210
+    },
+    {
+      "epoch": 1.662159725978006,
+      "grad_norm": 0.1877439022064209,
+      "learning_rate": 9.709146043291906e-05,
+      "loss": 0.0373,
+      "step": 9220
+    },
+    {
+      "epoch": 1.6639625022534703,
+      "grad_norm": 0.09826911240816116,
+      "learning_rate": 9.70821913182378e-05,
+      "loss": 0.0414,
+      "step": 9230
+    },
+    {
+      "epoch": 1.6657652785289345,
+      "grad_norm": 0.13114754855632782,
+      "learning_rate": 9.707290790122995e-05,
+      "loss": 0.0417,
+      "step": 9240
+    },
+    {
+      "epoch": 1.6675680548043987,
+      "grad_norm": 0.17269818484783173,
+      "learning_rate": 9.706361018471557e-05,
+      "loss": 0.0411,
+      "step": 9250
+    },
+    {
+      "epoch": 1.669370831079863,
+      "grad_norm": 0.1072874590754509,
+      "learning_rate": 9.705429817151906e-05,
+      "loss": 0.0438,
+      "step": 9260
+    },
+    {
+      "epoch": 1.6711736073553272,
+      "grad_norm": 0.1763143688440323,
+      "learning_rate": 9.704497186446917e-05,
+      "loss": 0.0432,
+      "step": 9270
+    },
+    {
+      "epoch": 1.6729763836307914,
+      "grad_norm": 0.12969639897346497,
+      "learning_rate": 9.703563126639896e-05,
+      "loss": 0.0443,
+      "step": 9280
+    },
+    {
+      "epoch": 1.6747791599062556,
+      "grad_norm": 0.22380030155181885,
+      "learning_rate": 9.70262763801459e-05,
+      "loss": 0.0425,
+      "step": 9290
+    },
+    {
+      "epoch": 1.6765819361817198,
+      "grad_norm": 0.14714008569717407,
+      "learning_rate": 9.701690720855171e-05,
+      "loss": 0.0367,
+      "step": 9300
+    },
+    {
+      "epoch": 1.678384712457184,
+      "grad_norm": 0.16367602348327637,
+      "learning_rate": 9.700752375446253e-05,
+      "loss": 0.0466,
+      "step": 9310
+    },
+    {
+      "epoch": 1.6801874887326482,
+      "grad_norm": 0.11947980523109436,
+      "learning_rate": 9.69981260207288e-05,
+      "loss": 0.0406,
+      "step": 9320
+    },
+    {
+      "epoch": 1.6819902650081124,
+      "grad_norm": 0.171796977519989,
+      "learning_rate": 9.698871401020529e-05,
+      "loss": 0.0413,
+      "step": 9330
+    },
+    {
+      "epoch": 1.6837930412835767,
+      "grad_norm": 0.13770607113838196,
+      "learning_rate": 9.697928772575112e-05,
+      "loss": 0.0452,
+      "step": 9340
+    },
+    {
+      "epoch": 1.685595817559041,
+      "grad_norm": 0.12719982862472534,
+      "learning_rate": 9.696984717022976e-05,
+      "loss": 0.0492,
+      "step": 9350
+    },
+    {
+      "epoch": 1.687398593834505,
+      "grad_norm": 0.1812325119972229,
+      "learning_rate": 9.6960392346509e-05,
+      "loss": 0.0407,
+      "step": 9360
+    },
+    {
+      "epoch": 1.6892013701099695,
+      "grad_norm": 0.15962675213813782,
+      "learning_rate": 9.695092325746097e-05,
+      "loss": 0.0425,
+      "step": 9370
+    },
+    {
+      "epoch": 1.6910041463854335,
+      "grad_norm": 0.16723452508449554,
+      "learning_rate": 9.694143990596211e-05,
+      "loss": 0.0371,
+      "step": 9380
+    },
+    {
+      "epoch": 1.692806922660898,
+      "grad_norm": 0.13127018511295319,
+      "learning_rate": 9.693194229489325e-05,
+      "loss": 0.0399,
+      "step": 9390
+    },
+    {
+      "epoch": 1.694609698936362,
+      "grad_norm": 0.1530247926712036,
+      "learning_rate": 9.692243042713944e-05,
+      "loss": 0.0372,
+      "step": 9400
+    },
+    {
+      "epoch": 1.6964124752118264,
+      "grad_norm": 0.1293390542268753,
+      "learning_rate": 9.691290430559022e-05,
+      "loss": 0.0424,
+      "step": 9410
+    },
+    {
+      "epoch": 1.6982152514872904,
+      "grad_norm": 0.12290593981742859,
+      "learning_rate": 9.690336393313932e-05,
+      "loss": 0.0406,
+      "step": 9420
+    },
+    {
+      "epoch": 1.7000180277627548,
+      "grad_norm": 0.13951396942138672,
+      "learning_rate": 9.689380931268487e-05,
+      "loss": 0.041,
+      "step": 9430
+    },
+    {
+      "epoch": 1.7018208040382188,
+      "grad_norm": 0.14184926450252533,
+      "learning_rate": 9.688424044712932e-05,
+      "loss": 0.0432,
+      "step": 9440
+    },
+    {
+      "epoch": 1.7036235803136832,
+      "grad_norm": 0.12445221841335297,
+      "learning_rate": 9.687465733937942e-05,
+      "loss": 0.0419,
+      "step": 9450
+    },
+    {
+      "epoch": 1.7054263565891472,
+      "grad_norm": 0.12526944279670715,
+      "learning_rate": 9.686505999234627e-05,
+      "loss": 0.042,
+      "step": 9460
+    },
+    {
+      "epoch": 1.7072291328646116,
+      "grad_norm": 0.11837741732597351,
+      "learning_rate": 9.685544840894529e-05,
+      "loss": 0.0436,
+      "step": 9470
+    },
+    {
+      "epoch": 1.7090319091400756,
+      "grad_norm": 0.15270712971687317,
+      "learning_rate": 9.684582259209624e-05,
+      "loss": 0.0411,
+      "step": 9480
+    },
+    {
+      "epoch": 1.71083468541554,
+      "grad_norm": 0.1271277219057083,
+      "learning_rate": 9.683618254472317e-05,
+      "loss": 0.0374,
+      "step": 9490
+    },
+    {
+      "epoch": 1.712637461691004,
+      "grad_norm": 0.10759492963552475,
+      "learning_rate": 9.682652826975449e-05,
+      "loss": 0.039,
+      "step": 9500
+    },
+    {
+      "epoch": 1.7144402379664685,
+      "grad_norm": 0.1235543042421341,
+      "learning_rate": 9.681685977012291e-05,
+      "loss": 0.0404,
+      "step": 9510
+    },
+    {
+      "epoch": 1.7162430142419325,
+      "grad_norm": 0.12931011617183685,
+      "learning_rate": 9.680717704876546e-05,
+      "loss": 0.0354,
+      "step": 9520
+    },
+    {
+      "epoch": 1.718045790517397,
+      "grad_norm": 0.120416559278965,
+      "learning_rate": 9.679748010862349e-05,
+      "loss": 0.0385,
+      "step": 9530
+    },
+    {
+      "epoch": 1.719848566792861,
+      "grad_norm": 0.1548643559217453,
+      "learning_rate": 9.678776895264267e-05,
+      "loss": 0.042,
+      "step": 9540
+    },
+    {
+      "epoch": 1.7216513430683253,
+      "grad_norm": 0.12035730481147766,
+      "learning_rate": 9.6778043583773e-05,
+      "loss": 0.0439,
+      "step": 9550
+    },
+    {
+      "epoch": 1.7234541193437893,
+      "grad_norm": 0.13684290647506714,
+      "learning_rate": 9.67683040049688e-05,
+      "loss": 0.0428,
+      "step": 9560
+    },
+    {
+      "epoch": 1.7252568956192538,
+      "grad_norm": 0.19395649433135986,
+      "learning_rate": 9.675855021918869e-05,
+      "loss": 0.043,
+      "step": 9570
+    },
+    {
+      "epoch": 1.7270596718947178,
+      "grad_norm": 0.115456722676754,
+      "learning_rate": 9.674878222939561e-05,
+      "loss": 0.0377,
+      "step": 9580
+    },
+    {
+      "epoch": 1.7288624481701822,
+      "grad_norm": 0.1280662566423416,
+      "learning_rate": 9.673900003855681e-05,
+      "loss": 0.0426,
+      "step": 9590
+    },
+    {
+      "epoch": 1.7306652244456462,
+      "grad_norm": 0.20764289796352386,
+      "learning_rate": 9.672920364964389e-05,
+      "loss": 0.0446,
+      "step": 9600
+    },
+    {
+      "epoch": 1.7324680007211106,
+      "grad_norm": 0.16308268904685974,
+      "learning_rate": 9.671939306563269e-05,
+      "loss": 0.0436,
+      "step": 9610
+    },
+    {
+      "epoch": 1.7342707769965746,
+      "grad_norm": 0.14351972937583923,
+      "learning_rate": 9.670956828950345e-05,
+      "loss": 0.0397,
+      "step": 9620
+    },
+    {
+      "epoch": 1.736073553272039,
+      "grad_norm": 0.1471412479877472,
+      "learning_rate": 9.669972932424065e-05,
+      "loss": 0.0388,
+      "step": 9630
+    },
+    {
+      "epoch": 1.737876329547503,
+      "grad_norm": 0.14039361476898193,
+      "learning_rate": 9.668987617283312e-05,
+      "loss": 0.039,
+      "step": 9640
+    },
+    {
+      "epoch": 1.7396791058229675,
+      "grad_norm": 0.18541479110717773,
+      "learning_rate": 9.668000883827397e-05,
+      "loss": 0.035,
+      "step": 9650
+    },
+    {
+      "epoch": 1.7414818820984315,
+      "grad_norm": 0.11093011498451233,
+      "learning_rate": 9.667012732356067e-05,
+      "loss": 0.0421,
+      "step": 9660
+    },
+    {
+      "epoch": 1.743284658373896,
+      "grad_norm": 0.14861413836479187,
+      "learning_rate": 9.666023163169493e-05,
+      "loss": 0.0413,
+      "step": 9670
+    },
+    {
+      "epoch": 1.74508743464936,
+      "grad_norm": 0.16462992131710052,
+      "learning_rate": 9.665032176568281e-05,
+      "loss": 0.0447,
+      "step": 9680
+    },
+    {
+      "epoch": 1.7468902109248243,
+      "grad_norm": 0.18467123806476593,
+      "learning_rate": 9.664039772853469e-05,
+      "loss": 0.0418,
+      "step": 9690
+    },
+    {
+      "epoch": 1.7486929872002883,
+      "grad_norm": 0.15918749570846558,
+      "learning_rate": 9.663045952326518e-05,
+      "loss": 0.0424,
+      "step": 9700
+    },
+    {
+      "epoch": 1.7504957634757528,
+      "grad_norm": 0.12315177917480469,
+      "learning_rate": 9.662050715289328e-05,
+      "loss": 0.0372,
+      "step": 9710
+    },
+    {
+      "epoch": 1.7522985397512167,
+      "grad_norm": 0.18573623895645142,
+      "learning_rate": 9.661054062044226e-05,
+      "loss": 0.0403,
+      "step": 9720
+    },
+    {
+      "epoch": 1.7541013160266812,
+      "grad_norm": 0.17335885763168335,
+      "learning_rate": 9.660055992893968e-05,
+      "loss": 0.0413,
+      "step": 9730
+    },
+    {
+      "epoch": 1.7559040923021452,
+      "grad_norm": 0.14658159017562866,
+      "learning_rate": 9.659056508141739e-05,
+      "loss": 0.0397,
+      "step": 9740
+    },
+    {
+      "epoch": 1.7577068685776096,
+      "grad_norm": 0.19271817803382874,
+      "learning_rate": 9.658055608091161e-05,
+      "loss": 0.036,
+      "step": 9750
+    },
+    {
+      "epoch": 1.7595096448530736,
+      "grad_norm": 0.1798640936613083,
+      "learning_rate": 9.657053293046276e-05,
+      "loss": 0.0409,
+      "step": 9760
+    },
+    {
+      "epoch": 1.761312421128538,
+      "grad_norm": 0.12259628623723984,
+      "learning_rate": 9.656049563311564e-05,
+      "loss": 0.0372,
+      "step": 9770
+    },
+    {
+      "epoch": 1.763115197404002,
+      "grad_norm": 0.15863682329654694,
+      "learning_rate": 9.655044419191929e-05,
+      "loss": 0.0419,
+      "step": 9780
+    },
+    {
+      "epoch": 1.7649179736794665,
+      "grad_norm": 0.11790483444929123,
+      "learning_rate": 9.654037860992711e-05,
+      "loss": 0.0433,
+      "step": 9790
+    },
+    {
+      "epoch": 1.7667207499549304,
+      "grad_norm": 0.13711854815483093,
+      "learning_rate": 9.653029889019672e-05,
+      "loss": 0.0455,
+      "step": 9800
+    },
+    {
+      "epoch": 1.7685235262303949,
+      "grad_norm": 0.11476242542266846,
+      "learning_rate": 9.65202050357901e-05,
+      "loss": 0.0372,
+      "step": 9810
+    },
+    {
+      "epoch": 1.7703263025058589,
+      "grad_norm": 0.1282246857881546,
+      "learning_rate": 9.651009704977347e-05,
+      "loss": 0.0409,
+      "step": 9820
+    },
+    {
+      "epoch": 1.7721290787813233,
+      "grad_norm": 0.16368697583675385,
+      "learning_rate": 9.649997493521738e-05,
+      "loss": 0.04,
+      "step": 9830
+    },
+    {
+      "epoch": 1.7739318550567873,
+      "grad_norm": 0.1306801289319992,
+      "learning_rate": 9.64898386951967e-05,
+      "loss": 0.0408,
+      "step": 9840
+    },
+    {
+      "epoch": 1.7757346313322517,
+      "grad_norm": 0.14899666607379913,
+      "learning_rate": 9.647968833279049e-05,
+      "loss": 0.0455,
+      "step": 9850
+    },
+    {
+      "epoch": 1.7775374076077157,
+      "grad_norm": 0.13333363831043243,
+      "learning_rate": 9.646952385108218e-05,
+      "loss": 0.039,
+      "step": 9860
+    },
+    {
+      "epoch": 1.7793401838831802,
+      "grad_norm": 0.12403760850429535,
+      "learning_rate": 9.645934525315951e-05,
+      "loss": 0.0395,
+      "step": 9870
+    },
+    {
+      "epoch": 1.7811429601586442,
+      "grad_norm": 0.17959313094615936,
+      "learning_rate": 9.644915254211442e-05,
+      "loss": 0.0428,
+      "step": 9880
+    },
+    {
+      "epoch": 1.7829457364341086,
+      "grad_norm": 0.17752526700496674,
+      "learning_rate": 9.643894572104321e-05,
+      "loss": 0.0455,
+      "step": 9890
+    },
+    {
+      "epoch": 1.7847485127095726,
+      "grad_norm": 0.15523189306259155,
+      "learning_rate": 9.642872479304644e-05,
+      "loss": 0.0414,
+      "step": 9900
+    },
+    {
+      "epoch": 1.786551288985037,
+      "grad_norm": 0.11487815529108047,
+      "learning_rate": 9.641848976122895e-05,
+      "loss": 0.0428,
+      "step": 9910
+    },
+    {
+      "epoch": 1.7883540652605012,
+      "grad_norm": 0.11738132685422897,
+      "learning_rate": 9.64082406286999e-05,
+      "loss": 0.0452,
+      "step": 9920
+    },
+    {
+      "epoch": 1.7901568415359654,
+      "grad_norm": 0.12917788326740265,
+      "learning_rate": 9.639797739857269e-05,
+      "loss": 0.0414,
+      "step": 9930
+    },
+    {
+      "epoch": 1.7919596178114297,
+      "grad_norm": 0.1342395395040512,
+      "learning_rate": 9.638770007396498e-05,
+      "loss": 0.0407,
+      "step": 9940
+    },
+    {
+      "epoch": 1.7937623940868939,
+      "grad_norm": 0.12433820962905884,
+      "learning_rate": 9.63774086579988e-05,
+      "loss": 0.0344,
+      "step": 9950
+    },
+    {
+      "epoch": 1.795565170362358,
+      "grad_norm": 0.16066451370716095,
+      "learning_rate": 9.63671031538004e-05,
+      "loss": 0.0441,
+      "step": 9960
+    },
+    {
+      "epoch": 1.7973679466378223,
+      "grad_norm": 0.14150521159172058,
+      "learning_rate": 9.635678356450031e-05,
+      "loss": 0.0391,
+      "step": 9970
+    },
+    {
+      "epoch": 1.7991707229132865,
+      "grad_norm": 0.1580437272787094,
+      "learning_rate": 9.634644989323336e-05,
+      "loss": 0.0461,
+      "step": 9980
+    },
+    {
+      "epoch": 1.8009734991887507,
+      "grad_norm": 0.11043276637792587,
+      "learning_rate": 9.633610214313861e-05,
+      "loss": 0.0402,
+      "step": 9990
+    },
+    {
+      "epoch": 1.802776275464215,
+      "grad_norm": 0.10205219686031342,
+      "learning_rate": 9.632574031735951e-05,
+      "loss": 0.0378,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 60000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 11,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}