diff --git "a/checkpoint-9375/trainer_state.json" "b/checkpoint-9375/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-9375/trainer_state.json"
@@ -0,0 +1,6592 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 9375,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0032,
+      "grad_norm": 10.0,
+      "learning_rate": 2.132196162046908e-06,
+      "loss": 28.3701,
+      "step": 10
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 10.0,
+      "learning_rate": 4.264392324093816e-06,
+      "loss": 24.5207,
+      "step": 20
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 10.0,
+      "learning_rate": 6.396588486140726e-06,
+      "loss": 22.6885,
+      "step": 30
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 10.0,
+      "learning_rate": 8.528784648187633e-06,
+      "loss": 25.2666,
+      "step": 40
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 10.0,
+      "learning_rate": 1.0660980810234541e-05,
+      "loss": 23.0791,
+      "step": 50
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 1.2793176972281452e-05,
+      "loss": 23.3554,
+      "step": 60
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 10.000000953674316,
+      "learning_rate": 1.4925373134328357e-05,
+      "loss": 23.0232,
+      "step": 70
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 10.0,
+      "learning_rate": 1.7057569296375266e-05,
+      "loss": 21.8077,
+      "step": 80
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 1.9189765458422178e-05,
+      "loss": 20.0271,
+      "step": 90
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 2.1321961620469083e-05,
+      "loss": 17.0629,
+      "step": 100
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 2.345415778251599e-05,
+      "loss": 18.8398,
+      "step": 110
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 2.5586353944562904e-05,
+      "loss": 15.8197,
+      "step": 120
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 2.771855010660981e-05,
+      "loss": 14.2088,
+      "step": 130
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 10.0,
+      "learning_rate": 2.9850746268656714e-05,
+      "loss": 14.0394,
+      "step": 140
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 10.0,
+      "learning_rate": 3.1982942430703626e-05,
+      "loss": 14.3635,
+      "step": 150
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 10.0,
+      "learning_rate": 3.411513859275053e-05,
+      "loss": 11.0378,
+      "step": 160
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 3.624733475479744e-05,
+      "loss": 10.8982,
+      "step": 170
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 3.8379530916844355e-05,
+      "loss": 9.9566,
+      "step": 180
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 4.051172707889126e-05,
+      "loss": 10.2806,
+      "step": 190
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 4.2643923240938166e-05,
+      "loss": 9.8932,
+      "step": 200
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 10.0,
+      "learning_rate": 4.477611940298508e-05,
+      "loss": 8.1511,
+      "step": 210
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 4.690831556503198e-05,
+      "loss": 8.07,
+      "step": 220
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 4.904051172707889e-05,
+      "loss": 8.0589,
+      "step": 230
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 5.117270788912581e-05,
+      "loss": 6.8857,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 5.330490405117271e-05,
+      "loss": 6.7318,
+      "step": 250
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 10.0,
+      "learning_rate": 5.543710021321962e-05,
+      "loss": 7.0872,
+      "step": 260
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 5.756929637526652e-05,
+      "loss": 6.5861,
+      "step": 270
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 10.0,
+      "learning_rate": 5.970149253731343e-05,
+      "loss": 6.3828,
+      "step": 280
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 10.0,
+      "learning_rate": 6.183368869936035e-05,
+      "loss": 6.0198,
+      "step": 290
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 6.396588486140725e-05,
+      "loss": 6.2239,
+      "step": 300
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 6.609808102345416e-05,
+      "loss": 5.7589,
+      "step": 310
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 6.823027718550106e-05,
+      "loss": 5.4196,
+      "step": 320
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 7.036247334754798e-05,
+      "loss": 5.6412,
+      "step": 330
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 10.0,
+      "learning_rate": 7.249466950959489e-05,
+      "loss": 5.2791,
+      "step": 340
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 10.0,
+      "learning_rate": 7.46268656716418e-05,
+      "loss": 5.5189,
+      "step": 350
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 10.0,
+      "learning_rate": 7.675906183368871e-05,
+      "loss": 5.0969,
+      "step": 360
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 7.889125799573562e-05,
+      "loss": 5.2089,
+      "step": 370
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 8.102345415778252e-05,
+      "loss": 6.2735,
+      "step": 380
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 8.315565031982943e-05,
+      "loss": 5.1972,
+      "step": 390
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 8.528784648187633e-05,
+      "loss": 4.791,
+      "step": 400
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 10.0,
+      "learning_rate": 8.742004264392325e-05,
+      "loss": 5.139,
+      "step": 410
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 8.955223880597016e-05,
+      "loss": 5.1234,
+      "step": 420
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 9.726828575134277,
+      "learning_rate": 9.168443496801706e-05,
+      "loss": 4.763,
+      "step": 430
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 10.0,
+      "learning_rate": 9.381663113006397e-05,
+      "loss": 5.0725,
+      "step": 440
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 10.0,
+      "learning_rate": 9.594882729211087e-05,
+      "loss": 4.708,
+      "step": 450
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.808102345415778e-05,
+      "loss": 4.3924,
+      "step": 460
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 8.861834526062012,
+      "learning_rate": 9.998877161464182e-05,
+      "loss": 4.3606,
+      "step": 470
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.987648776105997e-05,
+      "loss": 4.2736,
+      "step": 480
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.97642039074781e-05,
+      "loss": 4.5541,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 8.72944450378418,
+      "learning_rate": 9.965192005389625e-05,
+      "loss": 4.3459,
+      "step": 500
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.95396362003144e-05,
+      "loss": 4.0927,
+      "step": 510
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 9.396402359008789,
+      "learning_rate": 9.942735234673256e-05,
+      "loss": 4.1877,
+      "step": 520
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 6.486421585083008,
+      "learning_rate": 9.931506849315069e-05,
+      "loss": 3.9887,
+      "step": 530
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.920278463956883e-05,
+      "loss": 4.1643,
+      "step": 540
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 8.012758255004883,
+      "learning_rate": 9.909050078598698e-05,
+      "loss": 4.0236,
+      "step": 550
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.897821693240512e-05,
+      "loss": 3.9494,
+      "step": 560
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 9.886593307882327e-05,
+      "loss": 3.7742,
+      "step": 570
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.875364922524142e-05,
+      "loss": 3.817,
+      "step": 580
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 10.0,
+      "learning_rate": 9.864136537165956e-05,
+      "loss": 3.8002,
+      "step": 590
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 9.852908151807771e-05,
+      "loss": 3.5841,
+      "step": 600
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 9.841679766449586e-05,
+      "loss": 3.7045,
+      "step": 610
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.8304513810914e-05,
+      "loss": 3.3623,
+      "step": 620
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 9.819222995733213e-05,
+      "loss": 3.3953,
+      "step": 630
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 7.71815824508667,
+      "learning_rate": 9.807994610375028e-05,
+      "loss": 3.5513,
+      "step": 640
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 7.819033145904541,
+      "learning_rate": 9.796766225016843e-05,
+      "loss": 3.3845,
+      "step": 650
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 9.44756031036377,
+      "learning_rate": 9.785537839658657e-05,
+      "loss": 3.2571,
+      "step": 660
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 5.626587390899658,
+      "learning_rate": 9.774309454300472e-05,
+      "loss": 3.1949,
+      "step": 670
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 9.968693733215332,
+      "learning_rate": 9.763081068942287e-05,
+      "loss": 3.05,
+      "step": 680
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 9.028199195861816,
+      "learning_rate": 9.751852683584101e-05,
+      "loss": 3.378,
+      "step": 690
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.740624298225916e-05,
+      "loss": 3.2383,
+      "step": 700
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 10.0,
+      "learning_rate": 9.729395912867731e-05,
+      "loss": 3.2572,
+      "step": 710
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 9.906242370605469,
+      "learning_rate": 9.718167527509545e-05,
+      "loss": 3.2422,
+      "step": 720
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 7.099668025970459,
+      "learning_rate": 9.706939142151358e-05,
+      "loss": 3.199,
+      "step": 730
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 6.1059136390686035,
+      "learning_rate": 9.695710756793174e-05,
+      "loss": 2.9136,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 8.184073448181152,
+      "learning_rate": 9.684482371434989e-05,
+      "loss": 3.0956,
+      "step": 750
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 7.790270805358887,
+      "learning_rate": 9.673253986076802e-05,
+      "loss": 2.9784,
+      "step": 760
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 9.385482788085938,
+      "learning_rate": 9.662025600718617e-05,
+      "loss": 3.042,
+      "step": 770
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 9.105142593383789,
+      "learning_rate": 9.650797215360432e-05,
+      "loss": 3.0201,
+      "step": 780
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.639568830002246e-05,
+      "loss": 2.8601,
+      "step": 790
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.628340444644061e-05,
+      "loss": 3.0558,
+      "step": 800
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 7.617279052734375,
+      "learning_rate": 9.617112059285875e-05,
+      "loss": 2.9531,
+      "step": 810
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 7.272448539733887,
+      "learning_rate": 9.605883673927689e-05,
+      "loss": 2.9163,
+      "step": 820
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 10.0,
+      "learning_rate": 9.594655288569504e-05,
+      "loss": 3.0416,
+      "step": 830
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 6.786662578582764,
+      "learning_rate": 9.583426903211319e-05,
+      "loss": 3.0136,
+      "step": 840
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 6.918086051940918,
+      "learning_rate": 9.572198517853134e-05,
+      "loss": 2.8705,
+      "step": 850
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 9.991483688354492,
+      "learning_rate": 9.560970132494948e-05,
+      "loss": 2.9016,
+      "step": 860
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 6.9515581130981445,
+      "learning_rate": 9.549741747136763e-05,
+      "loss": 2.7265,
+      "step": 870
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 8.731250762939453,
+      "learning_rate": 9.538513361778578e-05,
+      "loss": 2.8442,
+      "step": 880
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 7.565758228302002,
+      "learning_rate": 9.527284976420391e-05,
+      "loss": 2.7234,
+      "step": 890
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 7.732306003570557,
+      "learning_rate": 9.516056591062205e-05,
+      "loss": 3.0467,
+      "step": 900
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.50482820570402e-05,
+      "loss": 2.9938,
+      "step": 910
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 7.414172649383545,
+      "learning_rate": 9.493599820345834e-05,
+      "loss": 2.7438,
+      "step": 920
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 8.1228609085083,
+      "learning_rate": 9.482371434987649e-05,
+      "loss": 2.8815,
+      "step": 930
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 6.716073513031006,
+      "learning_rate": 9.471143049629464e-05,
+      "loss": 2.6688,
+      "step": 940
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 5.223897933959961,
+      "learning_rate": 9.459914664271278e-05,
+      "loss": 2.7251,
+      "step": 950
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 7.797822952270508,
+      "learning_rate": 9.448686278913093e-05,
+      "loss": 2.5746,
+      "step": 960
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 6.63670539855957,
+      "learning_rate": 9.437457893554908e-05,
+      "loss": 2.6461,
+      "step": 970
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 10.0,
+      "learning_rate": 9.426229508196722e-05,
+      "loss": 2.7993,
+      "step": 980
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 7.738500595092773,
+      "learning_rate": 9.415001122838537e-05,
+      "loss": 2.7401,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 8.710482597351074,
+      "learning_rate": 9.40377273748035e-05,
+      "loss": 2.6068,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 6.2432684898376465,
+      "learning_rate": 9.392544352122165e-05,
+      "loss": 2.6246,
+      "step": 1010
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 7.530930519104004,
+      "learning_rate": 9.381315966763979e-05,
+      "loss": 2.7303,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 9.89444637298584,
+      "learning_rate": 9.370087581405794e-05,
+      "loss": 2.7558,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 9.358859196047609e-05,
+      "loss": 2.8216,
+      "step": 1040
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 5.699310302734375,
+      "learning_rate": 9.347630810689423e-05,
+      "loss": 2.81,
+      "step": 1050
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.336402425331238e-05,
+      "loss": 2.797,
+      "step": 1060
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 5.669271469116211,
+      "learning_rate": 9.325174039973053e-05,
+      "loss": 2.6244,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 9.576983451843262,
+      "learning_rate": 9.313945654614867e-05,
+      "loss": 2.7266,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 5.313910007476807,
+      "learning_rate": 9.30271726925668e-05,
+      "loss": 2.8692,
+      "step": 1090
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 8.139154434204102,
+      "learning_rate": 9.291488883898496e-05,
+      "loss": 2.6439,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 6.6432600021362305,
+      "learning_rate": 9.280260498540311e-05,
+      "loss": 2.6079,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 9.269032113182124e-05,
+      "loss": 2.7581,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 4.576703071594238,
+      "learning_rate": 9.25780372782394e-05,
+      "loss": 2.6304,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 5.5572285652160645,
+      "learning_rate": 9.246575342465755e-05,
+      "loss": 2.5335,
+      "step": 1140
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 7.863306045532227,
+      "learning_rate": 9.235346957107568e-05,
+      "loss": 2.4998,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 8.850831031799316,
+      "learning_rate": 9.224118571749383e-05,
+      "loss": 2.6284,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 6.468632221221924,
+      "learning_rate": 9.212890186391197e-05,
+      "loss": 2.6882,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 5.677201271057129,
+      "learning_rate": 9.201661801033011e-05,
+      "loss": 2.4775,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 4.348922252655029,
+      "learning_rate": 9.190433415674826e-05,
+      "loss": 2.7046,
+      "step": 1190
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 7.337803363800049,
+      "learning_rate": 9.179205030316641e-05,
+      "loss": 2.5728,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 9.881601333618164,
+      "learning_rate": 9.167976644958456e-05,
+      "loss": 2.5674,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 7.198265552520752,
+      "learning_rate": 9.15674825960027e-05,
+      "loss": 2.545,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 4.9246907234191895,
+      "learning_rate": 9.145519874242085e-05,
+      "loss": 2.7013,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 5.395358085632324,
+      "learning_rate": 9.1342914888839e-05,
+      "loss": 2.5742,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 6.851541042327881,
+      "learning_rate": 9.123063103525713e-05,
+      "loss": 2.6272,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 6.056896209716797,
+      "learning_rate": 9.111834718167527e-05,
+      "loss": 2.5673,
+      "step": 1260
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 6.445649147033691,
+      "learning_rate": 9.100606332809342e-05,
+      "loss": 2.8856,
+      "step": 1270
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 7.691575527191162,
+      "learning_rate": 9.089377947451156e-05,
+      "loss": 2.6714,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 5.955436706542969,
+      "learning_rate": 9.078149562092971e-05,
+      "loss": 2.6481,
+      "step": 1290
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 9.62521743774414,
+      "learning_rate": 9.066921176734786e-05,
+      "loss": 2.5761,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 4.644749641418457,
+      "learning_rate": 9.055692791376601e-05,
+      "loss": 2.5133,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 4.176416397094727,
+      "learning_rate": 9.044464406018415e-05,
+      "loss": 2.4743,
+      "step": 1320
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 6.192687511444092,
+      "learning_rate": 9.03323602066023e-05,
+      "loss": 2.5556,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 4.459692001342773,
+      "learning_rate": 9.022007635302045e-05,
+      "loss": 2.4436,
+      "step": 1340
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 6.155325412750244,
+      "learning_rate": 9.010779249943859e-05,
+      "loss": 2.4914,
+      "step": 1350
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 4.706230163574219,
+      "learning_rate": 8.999550864585672e-05,
+      "loss": 2.4705,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 6.316796779632568,
+      "learning_rate": 8.988322479227488e-05,
+      "loss": 2.4537,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 7.529996395111084,
+      "learning_rate": 8.977094093869301e-05,
+      "loss": 2.3581,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 7.215747356414795,
+      "learning_rate": 8.965865708511116e-05,
+      "loss": 2.4923,
+      "step": 1390
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 6.232388973236084,
+      "learning_rate": 8.954637323152931e-05,
+      "loss": 2.423,
+      "step": 1400
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 4.451657772064209,
+      "learning_rate": 8.943408937794746e-05,
+      "loss": 2.567,
+      "step": 1410
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 5.643525123596191,
+      "learning_rate": 8.93218055243656e-05,
+      "loss": 2.4042,
+      "step": 1420
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 8.920952167078375e-05,
+      "loss": 2.4591,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 7.8325514793396,
+      "learning_rate": 8.909723781720189e-05,
+      "loss": 2.5524,
+      "step": 1440
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 7.601844310760498,
+      "learning_rate": 8.898495396362003e-05,
+      "loss": 2.4915,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 10.0,
+      "learning_rate": 8.887267011003818e-05,
+      "loss": 2.5315,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 5.353204250335693,
+      "learning_rate": 8.876038625645633e-05,
+      "loss": 2.4169,
+      "step": 1470
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 9.473418235778809,
+      "learning_rate": 8.864810240287447e-05,
+      "loss": 2.4273,
+      "step": 1480
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 9.468120574951172,
+      "learning_rate": 8.853581854929262e-05,
+      "loss": 2.6286,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.575021743774414,
+      "learning_rate": 8.842353469571077e-05,
+      "loss": 2.6011,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 7.234686374664307,
+      "learning_rate": 8.83112508421289e-05,
+      "loss": 2.3283,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 4.9356184005737305,
+      "learning_rate": 8.819896698854705e-05,
+      "loss": 2.3356,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 5.09841775894165,
+      "learning_rate": 8.808668313496519e-05,
+      "loss": 2.3931,
+      "step": 1530
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 5.874563694000244,
+      "learning_rate": 8.797439928138334e-05,
+      "loss": 2.3976,
+      "step": 1540
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 6.544667720794678,
+      "learning_rate": 8.786211542780148e-05,
+      "loss": 2.5013,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 7.496267795562744,
+      "learning_rate": 8.774983157421963e-05,
+      "loss": 2.4145,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 6.886484146118164,
+      "learning_rate": 8.763754772063778e-05,
+      "loss": 2.5783,
+      "step": 1570
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 5.7106547355651855,
+      "learning_rate": 8.752526386705592e-05,
+      "loss": 2.3237,
+      "step": 1580
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 4.9011125564575195,
+      "learning_rate": 8.741298001347407e-05,
+      "loss": 2.4852,
+      "step": 1590
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 4.599292755126953,
+      "learning_rate": 8.730069615989222e-05,
+      "loss": 2.4273,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 5.276413440704346,
+      "learning_rate": 8.718841230631036e-05,
+      "loss": 2.3425,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 9.41219425201416,
+      "learning_rate": 8.70761284527285e-05,
+      "loss": 2.4555,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 8.96435260772705,
+      "learning_rate": 8.696384459914664e-05,
+      "loss": 2.3372,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 6.9553985595703125,
+      "learning_rate": 8.68515607455648e-05,
+      "loss": 2.5085,
+      "step": 1640
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 5.708377838134766,
+      "learning_rate": 8.673927689198293e-05,
+      "loss": 2.4528,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 8.115442276000977,
+      "learning_rate": 8.662699303840108e-05,
+      "loss": 2.3779,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 4.635824680328369,
+      "learning_rate": 8.651470918481923e-05,
+      "loss": 2.2828,
+      "step": 1670
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 3.9847660064697266,
+      "learning_rate": 8.640242533123737e-05,
+      "loss": 2.4229,
+      "step": 1680
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 7.811954975128174,
+      "learning_rate": 8.629014147765552e-05,
+      "loss": 2.2189,
+      "step": 1690
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 6.775632858276367,
+      "learning_rate": 8.617785762407367e-05,
+      "loss": 2.3237,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 8.606557377049181e-05,
+      "loss": 2.4138,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 5.420886039733887,
+      "learning_rate": 8.595328991690995e-05,
+      "loss": 2.4619,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 5.865475177764893,
+      "learning_rate": 8.58410060633281e-05,
+      "loss": 2.4588,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 6.2092509269714355,
+      "learning_rate": 8.572872220974623e-05,
+      "loss": 2.4464,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 7.776535511016846,
+      "learning_rate": 8.561643835616438e-05,
+      "loss": 2.4011,
+      "step": 1750
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 6.631662368774414,
+      "learning_rate": 8.550415450258253e-05,
+      "loss": 2.355,
+      "step": 1760
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 6.007630825042725,
+      "learning_rate": 8.539187064900069e-05,
+      "loss": 2.5012,
+      "step": 1770
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 8.527958679541882e-05,
+      "loss": 2.5772,
+      "step": 1780
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 5.767541408538818,
+      "learning_rate": 8.516730294183697e-05,
+      "loss": 2.5464,
+      "step": 1790
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 7.417142868041992,
+      "learning_rate": 8.505501908825511e-05,
+      "loss": 2.453,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 5.6994733810424805,
+      "learning_rate": 8.494273523467325e-05,
+      "loss": 2.5931,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 7.309426784515381,
+      "learning_rate": 8.48304513810914e-05,
+      "loss": 2.4631,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 5.572476863861084,
+      "learning_rate": 8.471816752750955e-05,
+      "loss": 2.5555,
+      "step": 1830
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 6.397329330444336,
+      "learning_rate": 8.460588367392769e-05,
+      "loss": 2.2597,
+      "step": 1840
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 4.919052600860596,
+      "learning_rate": 8.449359982034584e-05,
+      "loss": 2.3108,
+      "step": 1850
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 8.084444999694824,
+      "learning_rate": 8.438131596676399e-05,
+      "loss": 2.3109,
+      "step": 1860
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 6.542473316192627,
+      "learning_rate": 8.426903211318214e-05,
+      "loss": 2.523,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 8.415674825960028e-05,
+      "loss": 2.591,
+      "step": 1880
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 10.0,
+      "learning_rate": 8.404446440601843e-05,
+      "loss": 2.1992,
+      "step": 1890
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 8.659969329833984,
+      "learning_rate": 8.393218055243656e-05,
+      "loss": 2.2673,
+      "step": 1900
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 6.26581335067749,
+      "learning_rate": 8.38198966988547e-05,
+      "loss": 2.2571,
+      "step": 1910
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 4.025020122528076,
+      "learning_rate": 8.370761284527285e-05,
+      "loss": 2.4292,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 8.125853538513184,
+      "learning_rate": 8.3595328991691e-05,
+      "loss": 2.4198,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 9.907950401306152,
+      "learning_rate": 8.348304513810914e-05,
+      "loss": 2.3154,
+      "step": 1940
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 4.589335918426514,
+      "learning_rate": 8.337076128452729e-05,
+      "loss": 2.2819,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 3.842802047729492,
+      "learning_rate": 8.325847743094544e-05,
+      "loss": 2.2328,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 9.061366081237793,
+      "learning_rate": 8.314619357736358e-05,
+      "loss": 2.3952,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 3.815833330154419,
+      "learning_rate": 8.303390972378173e-05,
+      "loss": 2.4474,
+      "step": 1980
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 8.364211082458496,
+      "learning_rate": 8.292162587019986e-05,
+      "loss": 2.4021,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.37713623046875,
+      "learning_rate": 8.280934201661802e-05,
+      "loss": 2.5195,
+      "step": 2000
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 8.178314208984375,
+      "learning_rate": 8.269705816303615e-05,
+      "loss": 2.3142,
+      "step": 2010
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 7.809556007385254,
+      "learning_rate": 8.25847743094543e-05,
+      "loss": 2.1985,
+      "step": 2020
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 9.732132911682129,
+      "learning_rate": 8.247249045587245e-05,
+      "loss": 2.2352,
+      "step": 2030
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 6.2440571784973145,
+      "learning_rate": 8.236020660229059e-05,
+      "loss": 2.1969,
+      "step": 2040
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 5.272789478302002,
+      "learning_rate": 8.224792274870874e-05,
+      "loss": 2.3256,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 9.74461555480957,
+      "learning_rate": 8.213563889512689e-05,
+      "loss": 2.4389,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 4.720157146453857,
+      "learning_rate": 8.202335504154503e-05,
+      "loss": 2.4185,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 6.901200294494629,
+      "learning_rate": 8.191107118796317e-05,
+      "loss": 2.4403,
+      "step": 2080
+    },
+    {
+      "epoch": 0.6688,
+      "grad_norm": 6.6285624504089355,
+      "learning_rate": 8.179878733438132e-05,
+      "loss": 2.3017,
+      "step": 2090
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 5.480203628540039,
+      "learning_rate": 8.168650348079947e-05,
+      "loss": 2.4988,
+      "step": 2100
+    },
+    {
+      "epoch": 0.6752,
+      "grad_norm": 6.834201335906982,
+      "learning_rate": 8.15742196272176e-05,
+      "loss": 2.355,
+      "step": 2110
+    },
+    {
+      "epoch": 0.6784,
+      "grad_norm": 5.478670120239258,
+      "learning_rate": 8.146193577363576e-05,
+      "loss": 2.5405,
+      "step": 2120
+    },
+    {
+      "epoch": 0.6816,
+      "grad_norm": 8.395478248596191,
+      "learning_rate": 8.13496519200539e-05,
+      "loss": 2.2928,
+      "step": 2130
+    },
+    {
+      "epoch": 0.6848,
+      "grad_norm": 7.398046016693115,
+      "learning_rate": 8.123736806647204e-05,
+      "loss": 2.2259,
+      "step": 2140
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 8.656116485595703,
+      "learning_rate": 8.11250842128902e-05,
+      "loss": 2.2555,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6912,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 8.101280035930835e-05,
+      "loss": 2.4798,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6944,
+      "grad_norm": 4.557366847991943,
+      "learning_rate": 8.090051650572648e-05,
+      "loss": 2.5535,
+      "step": 2170
+    },
+    {
+      "epoch": 0.6976,
+      "grad_norm": 4.364936351776123,
+      "learning_rate": 8.078823265214462e-05,
+      "loss": 2.1993,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7008,
+      "grad_norm": 6.322412014007568,
+      "learning_rate": 8.067594879856277e-05,
+      "loss": 2.3593,
+      "step": 2190
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 6.9465789794921875,
+      "learning_rate": 8.056366494498092e-05,
+      "loss": 2.3772,
+      "step": 2200
+    },
+    {
+      "epoch": 0.7072,
+      "grad_norm": 6.708092212677002,
+      "learning_rate": 8.045138109139906e-05,
+      "loss": 2.332,
+      "step": 2210
+    },
+    {
+      "epoch": 0.7104,
+      "grad_norm": 5.4256768226623535,
+      "learning_rate": 8.033909723781721e-05,
+      "loss": 2.3298,
+      "step": 2220
+    },
+    {
+      "epoch": 0.7136,
+      "grad_norm": 4.8214850425720215,
+      "learning_rate": 8.022681338423536e-05,
+      "loss": 2.244,
+      "step": 2230
+    },
+    {
+      "epoch": 0.7168,
+      "grad_norm": 4.757294654846191,
+      "learning_rate": 8.01145295306535e-05,
+      "loss": 2.516,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 8.969782829284668,
+      "learning_rate": 8.000224567707165e-05,
+      "loss": 2.5106,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7232,
+      "grad_norm": 5.573437690734863,
+      "learning_rate": 7.988996182348978e-05,
+      "loss": 2.5074,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7264,
+      "grad_norm": 6.883465766906738,
+      "learning_rate": 7.977767796990792e-05,
+      "loss": 2.2634,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7296,
+      "grad_norm": 6.035622596740723,
+      "learning_rate": 7.966539411632607e-05,
+      "loss": 2.3059,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7328,
+      "grad_norm": 6.905811309814453,
+      "learning_rate": 7.955311026274422e-05,
+      "loss": 2.3145,
+      "step": 2290
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 6.611968517303467,
+      "learning_rate": 7.944082640916236e-05,
+      "loss": 2.3089,
+      "step": 2300
+    },
+    {
+      "epoch": 0.7392,
+      "grad_norm": 8.46692180633545,
+      "learning_rate": 7.932854255558051e-05,
+      "loss": 2.3344,
+      "step": 2310
+    },
+    {
+      "epoch": 0.7424,
+      "grad_norm": 7.151182174682617,
+      "learning_rate": 7.921625870199866e-05,
+      "loss": 2.2796,
+      "step": 2320
+    },
+    {
+      "epoch": 0.7456,
+      "grad_norm": 3.971759080886841,
+      "learning_rate": 7.910397484841681e-05,
+      "loss": 2.2586,
+      "step": 2330
+    },
+    {
+      "epoch": 0.7488,
+      "grad_norm": 6.936121940612793,
+      "learning_rate": 7.899169099483495e-05,
+      "loss": 2.2539,
+      "step": 2340
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 4.994493007659912,
+      "learning_rate": 7.887940714125309e-05,
+      "loss": 2.1336,
+      "step": 2350
+    },
+    {
+      "epoch": 0.7552,
+      "grad_norm": 5.026124000549316,
+      "learning_rate": 7.876712328767124e-05,
+      "loss": 2.2772,
+      "step": 2360
+    },
+    {
+      "epoch": 0.7584,
+      "grad_norm": 8.548295974731445,
+      "learning_rate": 7.865483943408937e-05,
+      "loss": 2.2297,
+      "step": 2370
+    },
+    {
+      "epoch": 0.7616,
+      "grad_norm": 6.2121901512146,
+      "learning_rate": 7.854255558050752e-05,
+      "loss": 2.2928,
+      "step": 2380
+    },
+    {
+      "epoch": 0.7648,
+      "grad_norm": 6.6148810386657715,
+      "learning_rate": 7.843027172692568e-05,
+      "loss": 2.3334,
+      "step": 2390
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 4.6633758544921875,
+      "learning_rate": 7.831798787334381e-05,
+      "loss": 2.4213,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7712,
+      "grad_norm": 6.86946439743042,
+      "learning_rate": 7.820570401976196e-05,
+      "loss": 2.2479,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7744,
+      "grad_norm": 7.258929252624512,
+      "learning_rate": 7.809342016618011e-05,
+      "loss": 2.1996,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7776,
+      "grad_norm": 6.279372215270996,
+      "learning_rate": 7.798113631259825e-05,
+      "loss": 2.2665,
+      "step": 2430
+    },
+    {
+      "epoch": 0.7808,
+      "grad_norm": 8.62110710144043,
+      "learning_rate": 7.78688524590164e-05,
+      "loss": 2.4074,
+      "step": 2440
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 10.000000953674316,
+      "learning_rate": 7.775656860543454e-05,
+      "loss": 3.6339,
+      "step": 2450
+    },
+    {
+      "epoch": 0.7872,
+      "grad_norm": 4.590153217315674,
+      "learning_rate": 7.764428475185269e-05,
+      "loss": 2.3271,
+      "step": 2460
+    },
+    {
+      "epoch": 0.7904,
+      "grad_norm": 5.067539691925049,
+      "learning_rate": 7.753200089827083e-05,
+      "loss": 2.2559,
+      "step": 2470
+    },
+    {
+      "epoch": 0.7936,
+      "grad_norm": 4.031103610992432,
+      "learning_rate": 7.741971704468898e-05,
+      "loss": 2.3426,
+      "step": 2480
+    },
+    {
+      "epoch": 0.7968,
+      "grad_norm": 8.196564674377441,
+      "learning_rate": 7.730743319110713e-05,
+      "loss": 2.1718,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 4.67139196395874,
+      "learning_rate": 7.719514933752526e-05,
+      "loss": 2.5061,
+      "step": 2500
+    },
+    {
+      "epoch": 0.8032,
+      "grad_norm": 6.368116855621338,
+      "learning_rate": 7.708286548394342e-05,
+      "loss": 2.2068,
+      "step": 2510
+    },
+    {
+      "epoch": 0.8064,
+      "grad_norm": 8.849568367004395,
+      "learning_rate": 7.697058163036157e-05,
+      "loss": 2.3294,
+      "step": 2520
+    },
+    {
+      "epoch": 0.8096,
+      "grad_norm": 5.521136283874512,
+      "learning_rate": 7.68582977767797e-05,
+      "loss": 2.3293,
+      "step": 2530
+    },
+    {
+      "epoch": 0.8128,
+      "grad_norm": 7.696922302246094,
+      "learning_rate": 7.674601392319784e-05,
+      "loss": 2.3748,
+      "step": 2540
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 7.695867538452148,
+      "learning_rate": 7.663373006961599e-05,
+      "loss": 2.0888,
+      "step": 2550
+    },
+    {
+      "epoch": 0.8192,
+      "grad_norm": 4.007026672363281,
+      "learning_rate": 7.652144621603414e-05,
+      "loss": 2.2417,
+      "step": 2560
+    },
+    {
+      "epoch": 0.8224,
+      "grad_norm": 5.373133659362793,
+      "learning_rate": 7.640916236245228e-05,
+      "loss": 2.397,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8256,
+      "grad_norm": 7.045199871063232,
+      "learning_rate": 7.629687850887043e-05,
+      "loss": 2.3619,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8288,
+      "grad_norm": 6.271018028259277,
+      "learning_rate": 7.618459465528858e-05,
+      "loss": 2.2644,
+      "step": 2590
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 8.043970108032227,
+      "learning_rate": 7.607231080170672e-05,
+      "loss": 2.5374,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8352,
+      "grad_norm": 6.158629894256592,
+      "learning_rate": 7.596002694812487e-05,
+      "loss": 2.2779,
+      "step": 2610
+    },
+    {
+      "epoch": 0.8384,
+      "grad_norm": 3.6692025661468506,
+      "learning_rate": 7.5847743094543e-05,
+      "loss": 2.2478,
+      "step": 2620
+    },
+    {
+      "epoch": 0.8416,
+      "grad_norm": 6.483317852020264,
+      "learning_rate": 7.573545924096114e-05,
+      "loss": 2.152,
+      "step": 2630
+    },
+    {
+      "epoch": 0.8448,
+      "grad_norm": 5.592394828796387,
+      "learning_rate": 7.562317538737929e-05,
+      "loss": 2.4605,
+      "step": 2640
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 5.773263931274414,
+      "learning_rate": 7.551089153379744e-05,
+      "loss": 2.4734,
+      "step": 2650
+    },
+    {
+      "epoch": 0.8512,
+      "grad_norm": 5.897200584411621,
+      "learning_rate": 7.53986076802156e-05,
+      "loss": 2.2845,
+      "step": 2660
+    },
+    {
+      "epoch": 0.8544,
+      "grad_norm": 6.476402759552002,
+      "learning_rate": 7.528632382663373e-05,
+      "loss": 2.1883,
+      "step": 2670
+    },
+    {
+      "epoch": 0.8576,
+      "grad_norm": 5.920806884765625,
+      "learning_rate": 7.517403997305188e-05,
+      "loss": 2.4448,
+      "step": 2680
+    },
+    {
+      "epoch": 0.8608,
+      "grad_norm": 4.258580207824707,
+      "learning_rate": 7.506175611947003e-05,
+      "loss": 2.1579,
+      "step": 2690
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 4.463003635406494,
+      "learning_rate": 7.494947226588817e-05,
+      "loss": 2.2095,
+      "step": 2700
+    },
+    {
+      "epoch": 0.8672,
+      "grad_norm": 8.6676607131958,
+      "learning_rate": 7.483718841230631e-05,
+      "loss": 2.3664,
+      "step": 2710
+    },
+    {
+      "epoch": 0.8704,
+      "grad_norm": 8.912688255310059,
+      "learning_rate": 7.472490455872446e-05,
+      "loss": 2.4029,
+      "step": 2720
+    },
+    {
+      "epoch": 0.8736,
+      "grad_norm": 6.20881986618042,
+      "learning_rate": 7.46126207051426e-05,
+      "loss": 2.2843,
+      "step": 2730
+    },
+    {
+      "epoch": 0.8768,
+      "grad_norm": 7.998341083526611,
+      "learning_rate": 7.450033685156075e-05,
+      "loss": 2.2651,
+      "step": 2740
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 9.811906814575195,
+      "learning_rate": 7.43880529979789e-05,
+      "loss": 2.3149,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8832,
+      "grad_norm": 8.817770957946777,
+      "learning_rate": 7.427576914439703e-05,
+      "loss": 2.4565,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8864,
+      "grad_norm": 4.727315902709961,
+      "learning_rate": 7.416348529081518e-05,
+      "loss": 2.2516,
+      "step": 2770
+    },
+    {
+      "epoch": 0.8896,
+      "grad_norm": 6.4811859130859375,
+      "learning_rate": 7.405120143723333e-05,
+      "loss": 2.0967,
+      "step": 2780
+    },
+    {
+      "epoch": 0.8928,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 7.393891758365149e-05,
+      "loss": 2.3187,
+      "step": 2790
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 9.609201431274414,
+      "learning_rate": 7.382663373006962e-05,
+      "loss": 2.2366,
+      "step": 2800
+    },
+    {
+      "epoch": 0.8992,
+      "grad_norm": 6.191689968109131,
+      "learning_rate": 7.371434987648776e-05,
+      "loss": 2.2946,
+      "step": 2810
+    },
+    {
+      "epoch": 0.9024,
+      "grad_norm": 9.036121368408203,
+      "learning_rate": 7.360206602290591e-05,
+      "loss": 2.2624,
+      "step": 2820
+    },
+    {
+      "epoch": 0.9056,
+      "grad_norm": 3.915534257888794,
+      "learning_rate": 7.348978216932405e-05,
+      "loss": 2.4489,
+      "step": 2830
+    },
+    {
+      "epoch": 0.9088,
+      "grad_norm": 7.408352851867676,
+      "learning_rate": 7.33774983157422e-05,
+      "loss": 2.1291,
+      "step": 2840
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 8.331685066223145,
+      "learning_rate": 7.326521446216035e-05,
+      "loss": 2.322,
+      "step": 2850
+    },
+    {
+      "epoch": 0.9152,
+      "grad_norm": 5.713507175445557,
+      "learning_rate": 7.315293060857849e-05,
+      "loss": 2.4402,
+      "step": 2860
+    },
+    {
+      "epoch": 0.9184,
+      "grad_norm": 5.220091342926025,
+      "learning_rate": 7.304064675499664e-05,
+      "loss": 2.1427,
+      "step": 2870
+    },
+    {
+      "epoch": 0.9216,
+      "grad_norm": 5.63717794418335,
+      "learning_rate": 7.292836290141479e-05,
+      "loss": 2.2149,
+      "step": 2880
+    },
+    {
+      "epoch": 0.9248,
+      "grad_norm": 8.326608657836914,
+      "learning_rate": 7.281607904783292e-05,
+      "loss": 2.4356,
+      "step": 2890
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 6.667220592498779,
+      "learning_rate": 7.270379519425106e-05,
+      "loss": 2.1894,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9312,
+      "grad_norm": 7.291802883148193,
+      "learning_rate": 7.259151134066921e-05,
+      "loss": 2.2346,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9344,
+      "grad_norm": 5.355696678161621,
+      "learning_rate": 7.247922748708736e-05,
+      "loss": 2.3652,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9376,
+      "grad_norm": 7.320453643798828,
+      "learning_rate": 7.23669436335055e-05,
+      "loss": 2.1491,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9408,
+      "grad_norm": 6.482708930969238,
+      "learning_rate": 7.225465977992365e-05,
+      "loss": 2.1797,
+      "step": 2940
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 8.68603515625,
+      "learning_rate": 7.21423759263418e-05,
+      "loss": 2.186,
+      "step": 2950
+    },
+    {
+      "epoch": 0.9472,
+      "grad_norm": 7.641050815582275,
+      "learning_rate": 7.203009207275994e-05,
+      "loss": 2.2185,
+      "step": 2960
+    },
+    {
+      "epoch": 0.9504,
+      "grad_norm": 7.222169876098633,
+      "learning_rate": 7.191780821917809e-05,
+      "loss": 2.3927,
+      "step": 2970
+    },
+    {
+      "epoch": 0.9536,
+      "grad_norm": 4.246700763702393,
+      "learning_rate": 7.180552436559623e-05,
+      "loss": 2.3365,
+      "step": 2980
+    },
+    {
+      "epoch": 0.9568,
+      "grad_norm": 6.096432685852051,
+      "learning_rate": 7.169324051201438e-05,
+      "loss": 2.1232,
+      "step": 2990
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.851985454559326,
+      "learning_rate": 7.158095665843251e-05,
+      "loss": 2.2176,
+      "step": 3000
+    },
+    {
+      "epoch": 0.9632,
+      "grad_norm": 7.797232627868652,
+      "learning_rate": 7.146867280485066e-05,
+      "loss": 2.4043,
+      "step": 3010
+    },
+    {
+      "epoch": 0.9664,
+      "grad_norm": 5.3702521324157715,
+      "learning_rate": 7.135638895126882e-05,
+      "loss": 2.2836,
+      "step": 3020
+    },
+    {
+      "epoch": 0.9696,
+      "grad_norm": 8.7493257522583,
+      "learning_rate": 7.124410509768695e-05,
+      "loss": 2.2188,
+      "step": 3030
+    },
+    {
+      "epoch": 0.9728,
+      "grad_norm": 10.0,
+      "learning_rate": 7.11318212441051e-05,
+      "loss": 2.3706,
+      "step": 3040
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 8.834953308105469,
+      "learning_rate": 7.101953739052325e-05,
+      "loss": 2.442,
+      "step": 3050
+    },
+    {
+      "epoch": 0.9792,
+      "grad_norm": 7.613494873046875,
+      "learning_rate": 7.090725353694139e-05,
+      "loss": 2.3874,
+      "step": 3060
+    },
+    {
+      "epoch": 0.9824,
+      "grad_norm": 5.66098165512085,
+      "learning_rate": 7.079496968335954e-05,
+      "loss": 2.1254,
+      "step": 3070
+    },
+    {
+      "epoch": 0.9856,
+      "grad_norm": 8.746124267578125,
+      "learning_rate": 7.068268582977768e-05,
+      "loss": 2.2545,
+      "step": 3080
+    },
+    {
+      "epoch": 0.9888,
+      "grad_norm": 9.96347713470459,
+      "learning_rate": 7.057040197619582e-05,
+      "loss": 2.1841,
+      "step": 3090
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 8.428536415100098,
+      "learning_rate": 7.045811812261397e-05,
+      "loss": 2.3754,
+      "step": 3100
+    },
+    {
+      "epoch": 0.9952,
+      "grad_norm": 6.405836582183838,
+      "learning_rate": 7.034583426903212e-05,
+      "loss": 2.2303,
+      "step": 3110
+    },
+    {
+      "epoch": 0.9984,
+      "grad_norm": 7.094132423400879,
+      "learning_rate": 7.023355041545027e-05,
+      "loss": 2.1851,
+      "step": 3120
+    },
+    {
+      "epoch": 1.0016,
+      "grad_norm": 5.770329475402832,
+      "learning_rate": 7.01212665618684e-05,
+      "loss": 2.2898,
+      "step": 3130
+    },
+    {
+      "epoch": 1.0048,
+      "grad_norm": 4.412771701812744,
+      "learning_rate": 7.000898270828656e-05,
+      "loss": 2.3988,
+      "step": 3140
+    },
+    {
+      "epoch": 1.008,
+      "grad_norm": 9.633313179016113,
+      "learning_rate": 6.98966988547047e-05,
+      "loss": 2.35,
+      "step": 3150
+    },
+    {
+      "epoch": 1.0112,
+      "grad_norm": 7.720678806304932,
+      "learning_rate": 6.978441500112284e-05,
+      "loss": 2.1658,
+      "step": 3160
+    },
+    {
+      "epoch": 1.0144,
+      "grad_norm": 7.196460723876953,
+      "learning_rate": 6.967213114754098e-05,
+      "loss": 2.1582,
+      "step": 3170
+    },
+    {
+      "epoch": 1.0176,
+      "grad_norm": 3.8286609649658203,
+      "learning_rate": 6.955984729395913e-05,
+      "loss": 2.3164,
+      "step": 3180
+    },
+    {
+      "epoch": 1.0208,
+      "grad_norm": 7.099523067474365,
+      "learning_rate": 6.944756344037727e-05,
+      "loss": 2.2672,
+      "step": 3190
+    },
+    {
+      "epoch": 1.024,
+      "grad_norm": 6.283836364746094,
+      "learning_rate": 6.933527958679542e-05,
+      "loss": 2.2812,
+      "step": 3200
+    },
+    {
+      "epoch": 1.0272,
+      "grad_norm": 4.959673881530762,
+      "learning_rate": 6.922299573321357e-05,
+      "loss": 2.1724,
+      "step": 3210
+    },
+    {
+      "epoch": 1.0304,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 6.911071187963172e-05,
+      "loss": 2.159,
+      "step": 3220
+    },
+    {
+      "epoch": 1.0336,
+      "grad_norm": 4.327045440673828,
+      "learning_rate": 6.899842802604986e-05,
+      "loss": 2.3596,
+      "step": 3230
+    },
+    {
+      "epoch": 1.0368,
+      "grad_norm": 7.3624653816223145,
+      "learning_rate": 6.888614417246801e-05,
+      "loss": 2.3693,
+      "step": 3240
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 8.922405242919922,
+      "learning_rate": 6.877386031888615e-05,
+      "loss": 2.3983,
+      "step": 3250
+    },
+    {
+      "epoch": 1.0432,
+      "grad_norm": 6.19892692565918,
+      "learning_rate": 6.866157646530428e-05,
+      "loss": 2.1969,
+      "step": 3260
+    },
+    {
+      "epoch": 1.0464,
+      "grad_norm": 6.3894524574279785,
+      "learning_rate": 6.854929261172243e-05,
+      "loss": 2.2934,
+      "step": 3270
+    },
+    {
+      "epoch": 1.0496,
+      "grad_norm": 7.240560054779053,
+      "learning_rate": 6.843700875814058e-05,
+      "loss": 2.2858,
+      "step": 3280
+    },
+    {
+      "epoch": 1.0528,
+      "grad_norm": 6.6486992835998535,
+      "learning_rate": 6.832472490455872e-05,
+      "loss": 2.2498,
+      "step": 3290
+    },
+    {
+      "epoch": 1.056,
+      "grad_norm": 7.461024761199951,
+      "learning_rate": 6.821244105097687e-05,
+      "loss": 2.2005,
+      "step": 3300
+    },
+    {
+      "epoch": 1.0592,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 6.810015719739502e-05,
+      "loss": 2.2171,
+      "step": 3310
+    },
+    {
+      "epoch": 1.0624,
+      "grad_norm": 6.346460342407227,
+      "learning_rate": 6.798787334381316e-05,
+      "loss": 2.0921,
+      "step": 3320
+    },
+    {
+      "epoch": 1.0656,
+      "grad_norm": 6.42893648147583,
+      "learning_rate": 6.787558949023131e-05,
+      "loss": 2.2411,
+      "step": 3330
+    },
+    {
+      "epoch": 1.0688,
+      "grad_norm": 5.788027286529541,
+      "learning_rate": 6.776330563664946e-05,
+      "loss": 2.0989,
+      "step": 3340
+    },
+    {
+      "epoch": 1.072,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 6.76510217830676e-05,
+      "loss": 2.3168,
+      "step": 3350
+    },
+    {
+      "epoch": 1.0752,
+      "grad_norm": 5.730334758758545,
+      "learning_rate": 6.753873792948574e-05,
+      "loss": 2.2532,
+      "step": 3360
+    },
+    {
+      "epoch": 1.0784,
+      "grad_norm": 7.052060127258301,
+      "learning_rate": 6.742645407590389e-05,
+      "loss": 2.1161,
+      "step": 3370
+    },
+    {
+      "epoch": 1.0816,
+      "grad_norm": 4.694394588470459,
+      "learning_rate": 6.731417022232204e-05,
+      "loss": 2.2371,
+      "step": 3380
+    },
+    {
+      "epoch": 1.0848,
+      "grad_norm": 6.626163482666016,
+      "learning_rate": 6.720188636874017e-05,
+      "loss": 2.5154,
+      "step": 3390
+    },
+    {
+      "epoch": 1.088,
+      "grad_norm": 7.122054576873779,
+      "learning_rate": 6.708960251515832e-05,
+      "loss": 2.1627,
+      "step": 3400
+    },
+    {
+      "epoch": 1.0912,
+      "grad_norm": 6.109677791595459,
+      "learning_rate": 6.697731866157648e-05,
+      "loss": 2.072,
+      "step": 3410
+    },
+    {
+      "epoch": 1.0944,
+      "grad_norm": 7.214818000793457,
+      "learning_rate": 6.686503480799461e-05,
+      "loss": 2.354,
+      "step": 3420
+    },
+    {
+      "epoch": 1.0976,
+      "grad_norm": 9.6759672164917,
+      "learning_rate": 6.675275095441276e-05,
+      "loss": 2.3704,
+      "step": 3430
+    },
+    {
+      "epoch": 1.1008,
+      "grad_norm": 4.210315227508545,
+      "learning_rate": 6.66404671008309e-05,
+      "loss": 2.3059,
+      "step": 3440
+    },
+    {
+      "epoch": 1.104,
+      "grad_norm": 6.809922218322754,
+      "learning_rate": 6.652818324724905e-05,
+      "loss": 2.3811,
+      "step": 3450
+    },
+    {
+      "epoch": 1.1072,
+      "grad_norm": 5.466648578643799,
+      "learning_rate": 6.641589939366719e-05,
+      "loss": 2.4183,
+      "step": 3460
+    },
+    {
+      "epoch": 1.1104,
+      "grad_norm": 8.649759292602539,
+      "learning_rate": 6.630361554008534e-05,
+      "loss": 2.2508,
+      "step": 3470
+    },
+    {
+      "epoch": 1.1136,
+      "grad_norm": 4.219826698303223,
+      "learning_rate": 6.619133168650349e-05,
+      "loss": 2.1897,
+      "step": 3480
+    },
+    {
+      "epoch": 1.1168,
+      "grad_norm": 6.436397552490234,
+      "learning_rate": 6.607904783292163e-05,
+      "loss": 2.1246,
+      "step": 3490
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 6.596676397933978e-05,
+      "loss": 2.5233,
+      "step": 3500
+    },
+    {
+      "epoch": 1.1232,
+      "grad_norm": 10.0,
+      "learning_rate": 6.585448012575793e-05,
+      "loss": 2.3533,
+      "step": 3510
+    },
+    {
+      "epoch": 1.1264,
+      "grad_norm": 3.651031732559204,
+      "learning_rate": 6.574219627217606e-05,
+      "loss": 2.2338,
+      "step": 3520
+    },
+    {
+      "epoch": 1.1296,
+      "grad_norm": 5.527792453765869,
+      "learning_rate": 6.56299124185942e-05,
+      "loss": 2.2797,
+      "step": 3530
+    },
+    {
+      "epoch": 1.1328,
+      "grad_norm": 4.227158546447754,
+      "learning_rate": 6.551762856501235e-05,
+      "loss": 2.1877,
+      "step": 3540
+    },
+    {
+      "epoch": 1.1360000000000001,
+      "grad_norm": 6.878480434417725,
+      "learning_rate": 6.540534471143049e-05,
+      "loss": 2.1618,
+      "step": 3550
+    },
+    {
+      "epoch": 1.1392,
+      "grad_norm": 5.831515789031982,
+      "learning_rate": 6.529306085784864e-05,
+      "loss": 2.1245,
+      "step": 3560
+    },
+    {
+      "epoch": 1.1424,
+      "grad_norm": 3.441094160079956,
+      "learning_rate": 6.518077700426679e-05,
+      "loss": 2.1778,
+      "step": 3570
+    },
+    {
+      "epoch": 1.1456,
+      "grad_norm": 7.330443382263184,
+      "learning_rate": 6.506849315068494e-05,
+      "loss": 1.9831,
+      "step": 3580
+    },
+    {
+      "epoch": 1.1488,
+      "grad_norm": 8.98618221282959,
+      "learning_rate": 6.495620929710308e-05,
+      "loss": 2.1796,
+      "step": 3590
+    },
+    {
+      "epoch": 1.152,
+      "grad_norm": 4.757387638092041,
+      "learning_rate": 6.484392544352123e-05,
+      "loss": 2.3529,
+      "step": 3600
+    },
+    {
+      "epoch": 1.1552,
+      "grad_norm": 3.5186734199523926,
+      "learning_rate": 6.473164158993938e-05,
+      "loss": 2.3361,
+      "step": 3610
+    },
+    {
+      "epoch": 1.1584,
+      "grad_norm": 6.991964340209961,
+      "learning_rate": 6.461935773635752e-05,
+      "loss": 2.1439,
+      "step": 3620
+    },
+    {
+      "epoch": 1.1616,
+      "grad_norm": 6.133205413818359,
+      "learning_rate": 6.450707388277565e-05,
+      "loss": 2.1227,
+      "step": 3630
+    },
+    {
+      "epoch": 1.1648,
+      "grad_norm": 4.490849018096924,
+      "learning_rate": 6.43947900291938e-05,
+      "loss": 2.2649,
+      "step": 3640
+    },
+    {
+      "epoch": 1.168,
+      "grad_norm": 6.143466949462891,
+      "learning_rate": 6.428250617561194e-05,
+      "loss": 2.2646,
+      "step": 3650
+    },
+    {
+      "epoch": 1.1712,
+      "grad_norm": 4.09475040435791,
+      "learning_rate": 6.417022232203009e-05,
+      "loss": 2.2631,
+      "step": 3660
+    },
+    {
+      "epoch": 1.1743999999999999,
+      "grad_norm": 9.874448776245117,
+      "learning_rate": 6.405793846844824e-05,
+      "loss": 2.2013,
+      "step": 3670
+    },
+    {
+      "epoch": 1.1776,
+      "grad_norm": 7.732521057128906,
+      "learning_rate": 6.39456546148664e-05,
+      "loss": 2.2698,
+      "step": 3680
+    },
+    {
+      "epoch": 1.1808,
+      "grad_norm": 5.697376251220703,
+      "learning_rate": 6.383337076128453e-05,
+      "loss": 2.1787,
+      "step": 3690
+    },
+    {
+      "epoch": 1.184,
+      "grad_norm": 6.074872970581055,
+      "learning_rate": 6.372108690770268e-05,
+      "loss": 2.1337,
+      "step": 3700
+    },
+    {
+      "epoch": 1.1872,
+      "grad_norm": 7.495859622955322,
+      "learning_rate": 6.360880305412082e-05,
+      "loss": 2.1082,
+      "step": 3710
+    },
+    {
+      "epoch": 1.1904,
+      "grad_norm": 5.88323450088501,
+      "learning_rate": 6.349651920053896e-05,
+      "loss": 2.2119,
+      "step": 3720
+    },
+    {
+      "epoch": 1.1936,
+      "grad_norm": 4.20490837097168,
+      "learning_rate": 6.338423534695711e-05,
+      "loss": 2.1251,
+      "step": 3730
+    },
+    {
+      "epoch": 1.1968,
+      "grad_norm": 7.4095540046691895,
+      "learning_rate": 6.327195149337526e-05,
+      "loss": 2.3263,
+      "step": 3740
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 5.438231468200684,
+      "learning_rate": 6.31596676397934e-05,
+      "loss": 2.0824,
+      "step": 3750
+    },
+    {
+      "epoch": 1.2032,
+      "grad_norm": 7.119965553283691,
+      "learning_rate": 6.304738378621155e-05,
+      "loss": 2.37,
+      "step": 3760
+    },
+    {
+      "epoch": 1.2064,
+      "grad_norm": 4.426470756530762,
+      "learning_rate": 6.29350999326297e-05,
+      "loss": 2.3092,
+      "step": 3770
+    },
+    {
+      "epoch": 1.2096,
+      "grad_norm": 5.602468490600586,
+      "learning_rate": 6.282281607904783e-05,
+      "loss": 2.1568,
+      "step": 3780
+    },
+    {
+      "epoch": 1.2128,
+      "grad_norm": 4.94999361038208,
+      "learning_rate": 6.271053222546598e-05,
+      "loss": 2.0438,
+      "step": 3790
+    },
+    {
+      "epoch": 1.216,
+      "grad_norm": 6.560863494873047,
+      "learning_rate": 6.259824837188412e-05,
+      "loss": 2.2386,
+      "step": 3800
+    },
+    {
+      "epoch": 1.2192,
+      "grad_norm": 9.445927619934082,
+      "learning_rate": 6.248596451830227e-05,
+      "loss": 3.1694,
+      "step": 3810
+    },
+    {
+      "epoch": 1.2224,
+      "grad_norm": 6.8221564292907715,
+      "learning_rate": 6.237368066472041e-05,
+      "loss": 2.2535,
+      "step": 3820
+    },
+    {
+      "epoch": 1.2256,
+      "grad_norm": 6.22130012512207,
+      "learning_rate": 6.226139681113856e-05,
+      "loss": 2.1744,
+      "step": 3830
+    },
+    {
+      "epoch": 1.2288000000000001,
+      "grad_norm": 5.471611499786377,
+      "learning_rate": 6.214911295755671e-05,
+      "loss": 2.5386,
+      "step": 3840
+    },
+    {
+      "epoch": 1.232,
+      "grad_norm": 3.5220141410827637,
+      "learning_rate": 6.203682910397485e-05,
+      "loss": 2.1352,
+      "step": 3850
+    },
+    {
+      "epoch": 1.2352,
+      "grad_norm": 6.965565204620361,
+      "learning_rate": 6.1924545250393e-05,
+      "loss": 2.2617,
+      "step": 3860
+    },
+    {
+      "epoch": 1.2384,
+      "grad_norm": 5.471559047698975,
+      "learning_rate": 6.181226139681115e-05,
+      "loss": 2.1913,
+      "step": 3870
+    },
+    {
+      "epoch": 1.2416,
+      "grad_norm": 7.852587699890137,
+      "learning_rate": 6.169997754322929e-05,
+      "loss": 2.2588,
+      "step": 3880
+    },
+    {
+      "epoch": 1.2448,
+      "grad_norm": 3.5346343517303467,
+      "learning_rate": 6.158769368964744e-05,
+      "loss": 2.1421,
+      "step": 3890
+    },
+    {
+      "epoch": 1.248,
+      "grad_norm": 8.021695137023926,
+      "learning_rate": 6.147540983606557e-05,
+      "loss": 2.3086,
+      "step": 3900
+    },
+    {
+      "epoch": 1.2511999999999999,
+      "grad_norm": 5.215954780578613,
+      "learning_rate": 6.136312598248372e-05,
+      "loss": 2.1955,
+      "step": 3910
+    },
+    {
+      "epoch": 1.2544,
+      "grad_norm": 6.547055244445801,
+      "learning_rate": 6.125084212890186e-05,
+      "loss": 2.0465,
+      "step": 3920
+    },
+    {
+      "epoch": 1.2576,
+      "grad_norm": 8.973469734191895,
+      "learning_rate": 6.113855827532001e-05,
+      "loss": 2.2372,
+      "step": 3930
+    },
+    {
+      "epoch": 1.2608,
+      "grad_norm": 8.85340404510498,
+      "learning_rate": 6.102627442173816e-05,
+      "loss": 2.2969,
+      "step": 3940
+    },
+    {
+      "epoch": 1.264,
+      "grad_norm": 8.8905029296875,
+      "learning_rate": 6.09139905681563e-05,
+      "loss": 2.1088,
+      "step": 3950
+    },
+    {
+      "epoch": 1.2671999999999999,
+      "grad_norm": 3.574281930923462,
+      "learning_rate": 6.0801706714574444e-05,
+      "loss": 2.2229,
+      "step": 3960
+    },
+    {
+      "epoch": 1.2704,
+      "grad_norm": 6.460089683532715,
+      "learning_rate": 6.0689422860992595e-05,
+      "loss": 2.0212,
+      "step": 3970
+    },
+    {
+      "epoch": 1.2736,
+      "grad_norm": 6.969429969787598,
+      "learning_rate": 6.057713900741073e-05,
+      "loss": 2.2653,
+      "step": 3980
+    },
+    {
+      "epoch": 1.2768,
+      "grad_norm": 6.06675386428833,
+      "learning_rate": 6.046485515382888e-05,
+      "loss": 2.2088,
+      "step": 3990
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 9.980952262878418,
+      "learning_rate": 6.0352571300247026e-05,
+      "loss": 2.1276,
+      "step": 4000
+    },
+    {
+      "epoch": 1.2832,
+      "grad_norm": 5.178830146789551,
+      "learning_rate": 6.024028744666518e-05,
+      "loss": 2.2661,
+      "step": 4010
+    },
+    {
+      "epoch": 1.2864,
+      "grad_norm": 9.019587516784668,
+      "learning_rate": 6.0128003593083314e-05,
+      "loss": 2.1305,
+      "step": 4020
+    },
+    {
+      "epoch": 1.2896,
+      "grad_norm": 6.0943074226379395,
+      "learning_rate": 6.0015719739501465e-05,
+      "loss": 2.1621,
+      "step": 4030
+    },
+    {
+      "epoch": 1.2928,
+      "grad_norm": 3.6441543102264404,
+      "learning_rate": 5.9903435885919615e-05,
+      "loss": 2.0759,
+      "step": 4040
+    },
+    {
+      "epoch": 1.296,
+      "grad_norm": 4.56536865234375,
+      "learning_rate": 5.979115203233775e-05,
+      "loss": 2.1728,
+      "step": 4050
+    },
+    {
+      "epoch": 1.2992,
+      "grad_norm": 4.460301876068115,
+      "learning_rate": 5.9678868178755896e-05,
+      "loss": 2.1825,
+      "step": 4060
+    },
+    {
+      "epoch": 1.3024,
+      "grad_norm": 4.983593940734863,
+      "learning_rate": 5.956658432517405e-05,
+      "loss": 2.2135,
+      "step": 4070
+    },
+    {
+      "epoch": 1.3056,
+      "grad_norm": 6.907173156738281,
+      "learning_rate": 5.9454300471592184e-05,
+      "loss": 2.3566,
+      "step": 4080
+    },
+    {
+      "epoch": 1.3088,
+      "grad_norm": 5.130218982696533,
+      "learning_rate": 5.9342016618010335e-05,
+      "loss": 2.2541,
+      "step": 4090
+    },
+    {
+      "epoch": 1.312,
+      "grad_norm": 4.747898578643799,
+      "learning_rate": 5.922973276442848e-05,
+      "loss": 1.9409,
+      "step": 4100
+    },
+    {
+      "epoch": 1.3152,
+      "grad_norm": 4.745423793792725,
+      "learning_rate": 5.9117448910846616e-05,
+      "loss": 2.2585,
+      "step": 4110
+    },
+    {
+      "epoch": 1.3184,
+      "grad_norm": 6.524007320404053,
+      "learning_rate": 5.900516505726477e-05,
+      "loss": 2.0822,
+      "step": 4120
+    },
+    {
+      "epoch": 1.3216,
+      "grad_norm": 8.787076950073242,
+      "learning_rate": 5.889288120368292e-05,
+      "loss": 2.0697,
+      "step": 4130
+    },
+    {
+      "epoch": 1.3248,
+      "grad_norm": 3.5528957843780518,
+      "learning_rate": 5.878059735010106e-05,
+      "loss": 2.1111,
+      "step": 4140
+    },
+    {
+      "epoch": 1.328,
+      "grad_norm": 5.02815580368042,
+      "learning_rate": 5.86683134965192e-05,
+      "loss": 2.1446,
+      "step": 4150
+    },
+    {
+      "epoch": 1.3312,
+      "grad_norm": 3.4902503490448,
+      "learning_rate": 5.855602964293735e-05,
+      "loss": 2.0923,
+      "step": 4160
+    },
+    {
+      "epoch": 1.3344,
+      "grad_norm": 4.021839141845703,
+      "learning_rate": 5.84437457893555e-05,
+      "loss": 2.291,
+      "step": 4170
+    },
+    {
+      "epoch": 1.3376000000000001,
+      "grad_norm": 5.555225372314453,
+      "learning_rate": 5.833146193577364e-05,
+      "loss": 2.1877,
+      "step": 4180
+    },
+    {
+      "epoch": 1.3408,
+      "grad_norm": 6.962709426879883,
+      "learning_rate": 5.821917808219178e-05,
+      "loss": 2.0687,
+      "step": 4190
+    },
+    {
+      "epoch": 1.3439999999999999,
+      "grad_norm": 10.0,
+      "learning_rate": 5.810689422860993e-05,
+      "loss": 2.3351,
+      "step": 4200
+    },
+    {
+      "epoch": 1.3472,
+      "grad_norm": 4.848272800445557,
+      "learning_rate": 5.799461037502807e-05,
+      "loss": 2.2501,
+      "step": 4210
+    },
+    {
+      "epoch": 1.3504,
+      "grad_norm": 5.333714962005615,
+      "learning_rate": 5.788232652144622e-05,
+      "loss": 2.1061,
+      "step": 4220
+    },
+    {
+      "epoch": 1.3536000000000001,
+      "grad_norm": 6.211556434631348,
+      "learning_rate": 5.777004266786436e-05,
+      "loss": 2.1438,
+      "step": 4230
+    },
+    {
+      "epoch": 1.3568,
+      "grad_norm": 8.62783145904541,
+      "learning_rate": 5.7657758814282514e-05,
+      "loss": 2.1886,
+      "step": 4240
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 7.506083965301514,
+      "learning_rate": 5.754547496070065e-05,
+      "loss": 2.2546,
+      "step": 4250
+    },
+    {
+      "epoch": 1.3632,
+      "grad_norm": 5.266780376434326,
+      "learning_rate": 5.74331911071188e-05,
+      "loss": 2.3712,
+      "step": 4260
+    },
+    {
+      "epoch": 1.3664,
+      "grad_norm": 9.855100631713867,
+      "learning_rate": 5.7320907253536945e-05,
+      "loss": 2.1088,
+      "step": 4270
+    },
+    {
+      "epoch": 1.3696,
+      "grad_norm": 3.677947759628296,
+      "learning_rate": 5.720862339995508e-05,
+      "loss": 2.347,
+      "step": 4280
+    },
+    {
+      "epoch": 1.3728,
+      "grad_norm": 7.6731276512146,
+      "learning_rate": 5.709633954637323e-05,
+      "loss": 2.2568,
+      "step": 4290
+    },
+    {
+      "epoch": 1.376,
+      "grad_norm": 5.236011505126953,
+      "learning_rate": 5.6984055692791384e-05,
+      "loss": 2.0683,
+      "step": 4300
+    },
+    {
+      "epoch": 1.3792,
+      "grad_norm": 6.686437606811523,
+      "learning_rate": 5.687177183920952e-05,
+      "loss": 2.1171,
+      "step": 4310
+    },
+    {
+      "epoch": 1.3824,
+      "grad_norm": 10.0,
+      "learning_rate": 5.675948798562767e-05,
+      "loss": 2.2386,
+      "step": 4320
+    },
+    {
+      "epoch": 1.3856,
+      "grad_norm": 4.5010986328125,
+      "learning_rate": 5.6647204132045816e-05,
+      "loss": 2.1134,
+      "step": 4330
+    },
+    {
+      "epoch": 1.3888,
+      "grad_norm": 3.935786724090576,
+      "learning_rate": 5.653492027846395e-05,
+      "loss": 2.4407,
+      "step": 4340
+    },
+    {
+      "epoch": 1.392,
+      "grad_norm": 6.398930072784424,
+      "learning_rate": 5.6422636424882103e-05,
+      "loss": 2.2234,
+      "step": 4350
+    },
+    {
+      "epoch": 1.3952,
+      "grad_norm": 8.995603561401367,
+      "learning_rate": 5.6310352571300254e-05,
+      "loss": 2.1862,
+      "step": 4360
+    },
+    {
+      "epoch": 1.3984,
+      "grad_norm": 8.50732421875,
+      "learning_rate": 5.61980687177184e-05,
+      "loss": 2.1446,
+      "step": 4370
+    },
+    {
+      "epoch": 1.4016,
+      "grad_norm": 8.399739265441895,
+      "learning_rate": 5.6085784864136535e-05,
+      "loss": 2.1897,
+      "step": 4380
+    },
+    {
+      "epoch": 1.4048,
+      "grad_norm": 9.738068580627441,
+      "learning_rate": 5.5973501010554686e-05,
+      "loss": 2.1621,
+      "step": 4390
+    },
+    {
+      "epoch": 1.408,
+      "grad_norm": 7.87959098815918,
+      "learning_rate": 5.5861217156972837e-05,
+      "loss": 2.2949,
+      "step": 4400
+    },
+    {
+      "epoch": 1.4112,
+      "grad_norm": 6.368355751037598,
+      "learning_rate": 5.5748933303390974e-05,
+      "loss": 2.0071,
+      "step": 4410
+    },
+    {
+      "epoch": 1.4144,
+      "grad_norm": 5.002744197845459,
+      "learning_rate": 5.563664944980912e-05,
+      "loss": 2.2448,
+      "step": 4420
+    },
+    {
+      "epoch": 1.4176,
+      "grad_norm": 4.510824203491211,
+      "learning_rate": 5.552436559622727e-05,
+      "loss": 2.3409,
+      "step": 4430
+    },
+    {
+      "epoch": 1.4208,
+      "grad_norm": 6.529369831085205,
+      "learning_rate": 5.5412081742645405e-05,
+      "loss": 2.1176,
+      "step": 4440
+    },
+    {
+      "epoch": 1.424,
+      "grad_norm": 9.94469165802002,
+      "learning_rate": 5.5299797889063556e-05,
+      "loss": 2.0862,
+      "step": 4450
+    },
+    {
+      "epoch": 1.4272,
+      "grad_norm": 5.885842323303223,
+      "learning_rate": 5.51875140354817e-05,
+      "loss": 2.4251,
+      "step": 4460
+    },
+    {
+      "epoch": 1.4304000000000001,
+      "grad_norm": 6.890707492828369,
+      "learning_rate": 5.507523018189985e-05,
+      "loss": 1.9864,
+      "step": 4470
+    },
+    {
+      "epoch": 1.4336,
+      "grad_norm": 4.47102689743042,
+      "learning_rate": 5.496294632831799e-05,
+      "loss": 2.1435,
+      "step": 4480
+    },
+    {
+      "epoch": 1.4368,
+      "grad_norm": 4.225493431091309,
+      "learning_rate": 5.485066247473614e-05,
+      "loss": 2.0326,
+      "step": 4490
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 4.37125825881958,
+      "learning_rate": 5.473837862115428e-05,
+      "loss": 2.1564,
+      "step": 4500
+    },
+    {
+      "epoch": 1.4432,
+      "grad_norm": 8.290303230285645,
+      "learning_rate": 5.462609476757242e-05,
+      "loss": 2.0743,
+      "step": 4510
+    },
+    {
+      "epoch": 1.4464000000000001,
+      "grad_norm": 5.776940822601318,
+      "learning_rate": 5.451381091399057e-05,
+      "loss": 2.2523,
+      "step": 4520
+    },
+    {
+      "epoch": 1.4496,
+      "grad_norm": 4.822579383850098,
+      "learning_rate": 5.440152706040872e-05,
+      "loss": 2.1078,
+      "step": 4530
+    },
+    {
+      "epoch": 1.4527999999999999,
+      "grad_norm": 5.790931224822998,
+      "learning_rate": 5.428924320682686e-05,
+      "loss": 2.0797,
+      "step": 4540
+    },
+    {
+      "epoch": 1.456,
+      "grad_norm": 5.75389289855957,
+      "learning_rate": 5.4176959353245e-05,
+      "loss": 2.1237,
+      "step": 4550
+    },
+    {
+      "epoch": 1.4592,
+      "grad_norm": 7.504802227020264,
+      "learning_rate": 5.406467549966315e-05,
+      "loss": 2.2125,
+      "step": 4560
+    },
+    {
+      "epoch": 1.4624,
+      "grad_norm": 4.340479850769043,
+      "learning_rate": 5.395239164608129e-05,
+      "loss": 2.1611,
+      "step": 4570
+    },
+    {
+      "epoch": 1.4656,
+      "grad_norm": 8.178547859191895,
+      "learning_rate": 5.384010779249944e-05,
+      "loss": 2.2023,
+      "step": 4580
+    },
+    {
+      "epoch": 1.4687999999999999,
+      "grad_norm": 9.241583824157715,
+      "learning_rate": 5.3727823938917584e-05,
+      "loss": 2.1268,
+      "step": 4590
+    },
+    {
+      "epoch": 1.472,
+      "grad_norm": 4.36149263381958,
+      "learning_rate": 5.3615540085335735e-05,
+      "loss": 2.1976,
+      "step": 4600
+    },
+    {
+      "epoch": 1.4752,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 5.350325623175387e-05,
+      "loss": 2.1717,
+      "step": 4610
+    },
+    {
+      "epoch": 1.4784,
+      "grad_norm": 5.395240306854248,
+      "learning_rate": 5.339097237817202e-05,
+      "loss": 1.9805,
+      "step": 4620
+    },
+    {
+      "epoch": 1.4816,
+      "grad_norm": 5.555797576904297,
+      "learning_rate": 5.327868852459017e-05,
+      "loss": 2.2378,
+      "step": 4630
+    },
+    {
+      "epoch": 1.4848,
+      "grad_norm": 5.219169616699219,
+      "learning_rate": 5.316640467100831e-05,
+      "loss": 2.1956,
+      "step": 4640
+    },
+    {
+      "epoch": 1.488,
+      "grad_norm": 9.0369234085083,
+      "learning_rate": 5.3054120817426454e-05,
+      "loss": 2.0657,
+      "step": 4650
+    },
+    {
+      "epoch": 1.4912,
+      "grad_norm": 7.064931392669678,
+      "learning_rate": 5.2941836963844605e-05,
+      "loss": 1.9879,
+      "step": 4660
+    },
+    {
+      "epoch": 1.4944,
+      "grad_norm": 6.1723809242248535,
+      "learning_rate": 5.282955311026274e-05,
+      "loss": 2.2089,
+      "step": 4670
+    },
+    {
+      "epoch": 1.4976,
+      "grad_norm": 9.340553283691406,
+      "learning_rate": 5.271726925668089e-05,
+      "loss": 2.1378,
+      "step": 4680
+    },
+    {
+      "epoch": 1.5008,
+      "grad_norm": 7.710911750793457,
+      "learning_rate": 5.260498540309904e-05,
+      "loss": 2.0892,
+      "step": 4690
+    },
+    {
+      "epoch": 1.504,
+      "grad_norm": 8.593657493591309,
+      "learning_rate": 5.249270154951719e-05,
+      "loss": 2.421,
+      "step": 4700
+    },
+    {
+      "epoch": 1.5072,
+      "grad_norm": 3.763519763946533,
+      "learning_rate": 5.2380417695935325e-05,
+      "loss": 2.0054,
+      "step": 4710
+    },
+    {
+      "epoch": 1.5104,
+      "grad_norm": 5.541579723358154,
+      "learning_rate": 5.2268133842353475e-05,
+      "loss": 2.2029,
+      "step": 4720
+    },
+    {
+      "epoch": 1.5135999999999998,
+      "grad_norm": 4.48792839050293,
+      "learning_rate": 5.215584998877162e-05,
+      "loss": 2.0555,
+      "step": 4730
+    },
+    {
+      "epoch": 1.5168,
+      "grad_norm": 4.694746494293213,
+      "learning_rate": 5.2043566135189756e-05,
+      "loss": 2.1252,
+      "step": 4740
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 3.626134157180786,
+      "learning_rate": 5.193128228160791e-05,
+      "loss": 2.2292,
+      "step": 4750
+    },
+    {
+      "epoch": 1.5232,
+      "grad_norm": 7.184067249298096,
+      "learning_rate": 5.181899842802606e-05,
+      "loss": 2.138,
+      "step": 4760
+    },
+    {
+      "epoch": 1.5264,
+      "grad_norm": 5.250986576080322,
+      "learning_rate": 5.1706714574444195e-05,
+      "loss": 2.2403,
+      "step": 4770
+    },
+    {
+      "epoch": 1.5295999999999998,
+      "grad_norm": 6.30639123916626,
+      "learning_rate": 5.159443072086234e-05,
+      "loss": 2.1862,
+      "step": 4780
+    },
+    {
+      "epoch": 1.5328,
+      "grad_norm": 5.524565696716309,
+      "learning_rate": 5.148214686728049e-05,
+      "loss": 2.1111,
+      "step": 4790
+    },
+    {
+      "epoch": 1.536,
+      "grad_norm": 4.2406697273254395,
+      "learning_rate": 5.136986301369864e-05,
+      "loss": 2.1711,
+      "step": 4800
+    },
+    {
+      "epoch": 1.5392000000000001,
+      "grad_norm": 4.823990821838379,
+      "learning_rate": 5.125757916011678e-05,
+      "loss": 2.057,
+      "step": 4810
+    },
+    {
+      "epoch": 1.5424,
+      "grad_norm": 6.957513809204102,
+      "learning_rate": 5.114529530653492e-05,
+      "loss": 2.0207,
+      "step": 4820
+    },
+    {
+      "epoch": 1.5455999999999999,
+      "grad_norm": 5.449457168579102,
+      "learning_rate": 5.103301145295307e-05,
+      "loss": 2.222,
+      "step": 4830
+    },
+    {
+      "epoch": 1.5488,
+      "grad_norm": 4.320427894592285,
+      "learning_rate": 5.092072759937121e-05,
+      "loss": 2.2214,
+      "step": 4840
+    },
+    {
+      "epoch": 1.552,
+      "grad_norm": 4.92500638961792,
+      "learning_rate": 5.080844374578936e-05,
+      "loss": 2.2,
+      "step": 4850
+    },
+    {
+      "epoch": 1.5552000000000001,
+      "grad_norm": 4.991371154785156,
+      "learning_rate": 5.0696159892207503e-05,
+      "loss": 2.0129,
+      "step": 4860
+    },
+    {
+      "epoch": 1.5584,
+      "grad_norm": 6.521137237548828,
+      "learning_rate": 5.058387603862564e-05,
+      "loss": 2.2163,
+      "step": 4870
+    },
+    {
+      "epoch": 1.5615999999999999,
+      "grad_norm": 7.385705947875977,
+      "learning_rate": 5.047159218504379e-05,
+      "loss": 2.0385,
+      "step": 4880
+    },
+    {
+      "epoch": 1.5648,
+      "grad_norm": 8.725979804992676,
+      "learning_rate": 5.035930833146194e-05,
+      "loss": 2.1222,
+      "step": 4890
+    },
+    {
+      "epoch": 1.568,
+      "grad_norm": 4.52073335647583,
+      "learning_rate": 5.024702447788008e-05,
+      "loss": 2.0741,
+      "step": 4900
+    },
+    {
+      "epoch": 1.5712000000000002,
+      "grad_norm": 6.499835014343262,
+      "learning_rate": 5.013474062429823e-05,
+      "loss": 2.1543,
+      "step": 4910
+    },
+    {
+      "epoch": 1.5744,
+      "grad_norm": 7.410144329071045,
+      "learning_rate": 5.0022456770716374e-05,
+      "loss": 2.2229,
+      "step": 4920
+    },
+    {
+      "epoch": 1.5776,
+      "grad_norm": 4.372899055480957,
+      "learning_rate": 4.991017291713452e-05,
+      "loss": 2.1517,
+      "step": 4930
+    },
+    {
+      "epoch": 1.5808,
+      "grad_norm": 3.7120919227600098,
+      "learning_rate": 4.979788906355267e-05,
+      "loss": 2.1163,
+      "step": 4940
+    },
+    {
+      "epoch": 1.584,
+      "grad_norm": 7.357861518859863,
+      "learning_rate": 4.968560520997081e-05,
+      "loss": 2.1247,
+      "step": 4950
+    },
+    {
+      "epoch": 1.5872000000000002,
+      "grad_norm": 6.083400726318359,
+      "learning_rate": 4.957332135638895e-05,
+      "loss": 2.1781,
+      "step": 4960
+    },
+    {
+      "epoch": 1.5904,
+      "grad_norm": 6.298897743225098,
+      "learning_rate": 4.94610375028071e-05,
+      "loss": 1.9974,
+      "step": 4970
+    },
+    {
+      "epoch": 1.5936,
+      "grad_norm": 6.194458484649658,
+      "learning_rate": 4.9348753649225244e-05,
+      "loss": 2.1545,
+      "step": 4980
+    },
+    {
+      "epoch": 1.5968,
+      "grad_norm": 5.1978278160095215,
+      "learning_rate": 4.923646979564339e-05,
+      "loss": 2.1107,
+      "step": 4990
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 4.76450777053833,
+      "learning_rate": 4.912418594206153e-05,
+      "loss": 2.3348,
+      "step": 5000
+    },
+    {
+      "epoch": 1.6032,
+      "grad_norm": 4.51431131362915,
+      "learning_rate": 4.9011902088479676e-05,
+      "loss": 2.0502,
+      "step": 5010
+    },
+    {
+      "epoch": 1.6064,
+      "grad_norm": 6.624210834503174,
+      "learning_rate": 4.8899618234897826e-05,
+      "loss": 2.1568,
+      "step": 5020
+    },
+    {
+      "epoch": 1.6096,
+      "grad_norm": 5.511273384094238,
+      "learning_rate": 4.878733438131597e-05,
+      "loss": 2.0227,
+      "step": 5030
+    },
+    {
+      "epoch": 1.6128,
+      "grad_norm": 3.3496336936950684,
+      "learning_rate": 4.8675050527734114e-05,
+      "loss": 2.1213,
+      "step": 5040
+    },
+    {
+      "epoch": 1.616,
+      "grad_norm": 7.776872158050537,
+      "learning_rate": 4.856276667415226e-05,
+      "loss": 2.1931,
+      "step": 5050
+    },
+    {
+      "epoch": 1.6192,
+      "grad_norm": 6.206504821777344,
+      "learning_rate": 4.84504828205704e-05,
+      "loss": 2.1976,
+      "step": 5060
+    },
+    {
+      "epoch": 1.6223999999999998,
+      "grad_norm": 9.433921813964844,
+      "learning_rate": 4.833819896698855e-05,
+      "loss": 2.2875,
+      "step": 5070
+    },
+    {
+      "epoch": 1.6256,
+      "grad_norm": 5.77447509765625,
+      "learning_rate": 4.8225915113406696e-05,
+      "loss": 2.2109,
+      "step": 5080
+    },
+    {
+      "epoch": 1.6288,
+      "grad_norm": 3.713836193084717,
+      "learning_rate": 4.811363125982484e-05,
+      "loss": 2.1561,
+      "step": 5090
+    },
+    {
+      "epoch": 1.6320000000000001,
+      "grad_norm": 3.5686769485473633,
+      "learning_rate": 4.8001347406242984e-05,
+      "loss": 2.1125,
+      "step": 5100
+    },
+    {
+      "epoch": 1.6352,
+      "grad_norm": 7.770430564880371,
+      "learning_rate": 4.788906355266113e-05,
+      "loss": 2.2041,
+      "step": 5110
+    },
+    {
+      "epoch": 1.6383999999999999,
+      "grad_norm": 5.841362953186035,
+      "learning_rate": 4.777677969907928e-05,
+      "loss": 2.1188,
+      "step": 5120
+    },
+    {
+      "epoch": 1.6416,
+      "grad_norm": 3.36885142326355,
+      "learning_rate": 4.766449584549742e-05,
+      "loss": 2.0692,
+      "step": 5130
+    },
+    {
+      "epoch": 1.6448,
+      "grad_norm": 7.515949249267578,
+      "learning_rate": 4.755221199191556e-05,
+      "loss": 2.0279,
+      "step": 5140
+    },
+    {
+      "epoch": 1.6480000000000001,
+      "grad_norm": 4.221615314483643,
+      "learning_rate": 4.743992813833371e-05,
+      "loss": 2.0905,
+      "step": 5150
+    },
+    {
+      "epoch": 1.6512,
+      "grad_norm": 6.048422336578369,
+      "learning_rate": 4.7327644284751854e-05,
+      "loss": 2.1089,
+      "step": 5160
+    },
+    {
+      "epoch": 1.6543999999999999,
+      "grad_norm": 5.11595344543457,
+      "learning_rate": 4.7215360431170005e-05,
+      "loss": 2.2271,
+      "step": 5170
+    },
+    {
+      "epoch": 1.6576,
+      "grad_norm": 5.2255449295043945,
+      "learning_rate": 4.710307657758814e-05,
+      "loss": 2.1604,
+      "step": 5180
+    },
+    {
+      "epoch": 1.6608,
+      "grad_norm": 5.745364189147949,
+      "learning_rate": 4.6990792724006286e-05,
+      "loss": 2.0856,
+      "step": 5190
+    },
+    {
+      "epoch": 1.6640000000000001,
+      "grad_norm": 4.490666389465332,
+      "learning_rate": 4.687850887042444e-05,
+      "loss": 2.1981,
+      "step": 5200
+    },
+    {
+      "epoch": 1.6672,
+      "grad_norm": 5.564126968383789,
+      "learning_rate": 4.676622501684258e-05,
+      "loss": 2.0797,
+      "step": 5210
+    },
+    {
+      "epoch": 1.6703999999999999,
+      "grad_norm": 5.2047247886657715,
+      "learning_rate": 4.6653941163260725e-05,
+      "loss": 2.163,
+      "step": 5220
+    },
+    {
+      "epoch": 1.6736,
+      "grad_norm": 8.44140625,
+      "learning_rate": 4.654165730967887e-05,
+      "loss": 2.2305,
+      "step": 5230
+    },
+    {
+      "epoch": 1.6768,
+      "grad_norm": 8.154293060302734,
+      "learning_rate": 4.642937345609701e-05,
+      "loss": 2.1215,
+      "step": 5240
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 5.150942802429199,
+      "learning_rate": 4.631708960251516e-05,
+      "loss": 2.1661,
+      "step": 5250
+    },
+    {
+      "epoch": 1.6832,
+      "grad_norm": 6.649692058563232,
+      "learning_rate": 4.620480574893331e-05,
+      "loss": 2.2111,
+      "step": 5260
+    },
+    {
+      "epoch": 1.6864,
+      "grad_norm": 8.254748344421387,
+      "learning_rate": 4.609252189535145e-05,
+      "loss": 2.0773,
+      "step": 5270
+    },
+    {
+      "epoch": 1.6896,
+      "grad_norm": 2.958098888397217,
+      "learning_rate": 4.5980238041769595e-05,
+      "loss": 2.101,
+      "step": 5280
+    },
+    {
+      "epoch": 1.6928,
+      "grad_norm": 7.841224670410156,
+      "learning_rate": 4.586795418818774e-05,
+      "loss": 2.4461,
+      "step": 5290
+    },
+    {
+      "epoch": 1.696,
+      "grad_norm": 3.867351531982422,
+      "learning_rate": 4.575567033460589e-05,
+      "loss": 2.1184,
+      "step": 5300
+    },
+    {
+      "epoch": 1.6992,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 4.564338648102403e-05,
+      "loss": 2.2102,
+      "step": 5310
+    },
+    {
+      "epoch": 1.7024,
+      "grad_norm": 5.966880798339844,
+      "learning_rate": 4.553110262744217e-05,
+      "loss": 2.1324,
+      "step": 5320
+    },
+    {
+      "epoch": 1.7056,
+      "grad_norm": 5.331724643707275,
+      "learning_rate": 4.541881877386032e-05,
+      "loss": 2.2293,
+      "step": 5330
+    },
+    {
+      "epoch": 1.7088,
+      "grad_norm": 5.222733497619629,
+      "learning_rate": 4.5306534920278465e-05,
+      "loss": 2.044,
+      "step": 5340
+    },
+    {
+      "epoch": 1.712,
+      "grad_norm": 3.4903883934020996,
+      "learning_rate": 4.5194251066696616e-05,
+      "loss": 2.1925,
+      "step": 5350
+    },
+    {
+      "epoch": 1.7151999999999998,
+      "grad_norm": 9.891901969909668,
+      "learning_rate": 4.508196721311476e-05,
+      "loss": 2.1628,
+      "step": 5360
+    },
+    {
+      "epoch": 1.7184,
+      "grad_norm": 3.6791141033172607,
+      "learning_rate": 4.4969683359532897e-05,
+      "loss": 2.2044,
+      "step": 5370
+    },
+    {
+      "epoch": 1.7216,
+      "grad_norm": 3.284372091293335,
+      "learning_rate": 4.485739950595105e-05,
+      "loss": 2.0421,
+      "step": 5380
+    },
+    {
+      "epoch": 1.7248,
+      "grad_norm": 2.7818946838378906,
+      "learning_rate": 4.474511565236919e-05,
+      "loss": 2.0564,
+      "step": 5390
+    },
+    {
+      "epoch": 1.728,
+      "grad_norm": 9.54232406616211,
+      "learning_rate": 4.463283179878734e-05,
+      "loss": 2.1152,
+      "step": 5400
+    },
+    {
+      "epoch": 1.7311999999999999,
+      "grad_norm": 6.284071445465088,
+      "learning_rate": 4.452054794520548e-05,
+      "loss": 2.1527,
+      "step": 5410
+    },
+    {
+      "epoch": 1.7344,
+      "grad_norm": 4.312674045562744,
+      "learning_rate": 4.440826409162362e-05,
+      "loss": 2.0515,
+      "step": 5420
+    },
+    {
+      "epoch": 1.7376,
+      "grad_norm": 4.142785549163818,
+      "learning_rate": 4.4295980238041774e-05,
+      "loss": 2.2114,
+      "step": 5430
+    },
+    {
+      "epoch": 1.7408000000000001,
+      "grad_norm": 7.880631446838379,
+      "learning_rate": 4.418369638445992e-05,
+      "loss": 2.2083,
+      "step": 5440
+    },
+    {
+      "epoch": 1.744,
+      "grad_norm": 4.2559709548950195,
+      "learning_rate": 4.407141253087806e-05,
+      "loss": 2.2029,
+      "step": 5450
+    },
+    {
+      "epoch": 1.7471999999999999,
+      "grad_norm": 6.414773941040039,
+      "learning_rate": 4.3959128677296205e-05,
+      "loss": 2.2198,
+      "step": 5460
+    },
+    {
+      "epoch": 1.7504,
+      "grad_norm": 4.784211158752441,
+      "learning_rate": 4.384684482371435e-05,
+      "loss": 1.9202,
+      "step": 5470
+    },
+    {
+      "epoch": 1.7536,
+      "grad_norm": 8.340254783630371,
+      "learning_rate": 4.37345609701325e-05,
+      "loss": 2.2925,
+      "step": 5480
+    },
+    {
+      "epoch": 1.7568000000000001,
+      "grad_norm": 3.7721922397613525,
+      "learning_rate": 4.3622277116550644e-05,
+      "loss": 2.1257,
+      "step": 5490
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 5.6239776611328125,
+      "learning_rate": 4.350999326296879e-05,
+      "loss": 2.1088,
+      "step": 5500
+    },
+    {
+      "epoch": 1.7631999999999999,
+      "grad_norm": 6.1071295738220215,
+      "learning_rate": 4.339770940938693e-05,
+      "loss": 2.1422,
+      "step": 5510
+    },
+    {
+      "epoch": 1.7664,
+      "grad_norm": 8.268501281738281,
+      "learning_rate": 4.3285425555805075e-05,
+      "loss": 2.0904,
+      "step": 5520
+    },
+    {
+      "epoch": 1.7696,
+      "grad_norm": 8.720447540283203,
+      "learning_rate": 4.3173141702223226e-05,
+      "loss": 2.2511,
+      "step": 5530
+    },
+    {
+      "epoch": 1.7728000000000002,
+      "grad_norm": 9.763360023498535,
+      "learning_rate": 4.306085784864137e-05,
+      "loss": 2.1065,
+      "step": 5540
+    },
+    {
+      "epoch": 1.776,
+      "grad_norm": 9.336665153503418,
+      "learning_rate": 4.294857399505951e-05,
+      "loss": 2.0478,
+      "step": 5550
+    },
+    {
+      "epoch": 1.7792,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 4.283629014147766e-05,
+      "loss": 2.07,
+      "step": 5560
+    },
+    {
+      "epoch": 1.7824,
+      "grad_norm": 5.811784744262695,
+      "learning_rate": 4.27240062878958e-05,
+      "loss": 2.0205,
+      "step": 5570
+    },
+    {
+      "epoch": 1.7856,
+      "grad_norm": 8.127388000488281,
+      "learning_rate": 4.261172243431395e-05,
+      "loss": 2.1649,
+      "step": 5580
+    },
+    {
+      "epoch": 1.7888,
+      "grad_norm": 5.085277080535889,
+      "learning_rate": 4.249943858073209e-05,
+      "loss": 2.1256,
+      "step": 5590
+    },
+    {
+      "epoch": 1.792,
+      "grad_norm": 5.166441440582275,
+      "learning_rate": 4.2387154727150233e-05,
+      "loss": 2.1498,
+      "step": 5600
+    },
+    {
+      "epoch": 1.7952,
+      "grad_norm": 6.641541481018066,
+      "learning_rate": 4.2274870873568384e-05,
+      "loss": 2.1661,
+      "step": 5610
+    },
+    {
+      "epoch": 1.7984,
+      "grad_norm": 3.5791046619415283,
+      "learning_rate": 4.216258701998653e-05,
+      "loss": 2.2342,
+      "step": 5620
+    },
+    {
+      "epoch": 1.8016,
+      "grad_norm": 4.0631279945373535,
+      "learning_rate": 4.205030316640468e-05,
+      "loss": 2.136,
+      "step": 5630
+    },
+    {
+      "epoch": 1.8048,
+      "grad_norm": 4.964014530181885,
+      "learning_rate": 4.1938019312822816e-05,
+      "loss": 2.4784,
+      "step": 5640
+    },
+    {
+      "epoch": 1.808,
+      "grad_norm": 7.552999019622803,
+      "learning_rate": 4.182573545924096e-05,
+      "loss": 2.3117,
+      "step": 5650
+    },
+    {
+      "epoch": 1.8112,
+      "grad_norm": 4.7818284034729,
+      "learning_rate": 4.171345160565911e-05,
+      "loss": 2.1387,
+      "step": 5660
+    },
+    {
+      "epoch": 1.8144,
+      "grad_norm": 7.117303848266602,
+      "learning_rate": 4.1601167752077254e-05,
+      "loss": 2.1771,
+      "step": 5670
+    },
+    {
+      "epoch": 1.8176,
+      "grad_norm": 6.260792255401611,
+      "learning_rate": 4.14888838984954e-05,
+      "loss": 1.9511,
+      "step": 5680
+    },
+    {
+      "epoch": 1.8208,
+      "grad_norm": 6.069239139556885,
+      "learning_rate": 4.137660004491354e-05,
+      "loss": 2.3494,
+      "step": 5690
+    },
+    {
+      "epoch": 1.8239999999999998,
+      "grad_norm": 3.6275076866149902,
+      "learning_rate": 4.1264316191331686e-05,
+      "loss": 2.037,
+      "step": 5700
+    },
+    {
+      "epoch": 1.8272,
+      "grad_norm": 5.257174491882324,
+      "learning_rate": 4.115203233774984e-05,
+      "loss": 2.1981,
+      "step": 5710
+    },
+    {
+      "epoch": 1.8304,
+      "grad_norm": 9.999999046325684,
+      "learning_rate": 4.103974848416798e-05,
+      "loss": 2.1808,
+      "step": 5720
+    },
+    {
+      "epoch": 1.8336000000000001,
+      "grad_norm": 6.373122215270996,
+      "learning_rate": 4.092746463058612e-05,
+      "loss": 2.1297,
+      "step": 5730
+    },
+    {
+      "epoch": 1.8368,
+      "grad_norm": 6.5584940910339355,
+      "learning_rate": 4.081518077700427e-05,
+      "loss": 2.1768,
+      "step": 5740
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 8.723784446716309,
+      "learning_rate": 4.070289692342241e-05,
+      "loss": 2.0283,
+      "step": 5750
+    },
+    {
+      "epoch": 1.8432,
+      "grad_norm": 5.2506537437438965,
+      "learning_rate": 4.059061306984056e-05,
+      "loss": 2.2787,
+      "step": 5760
+    },
+    {
+      "epoch": 1.8464,
+      "grad_norm": 6.952129364013672,
+      "learning_rate": 4.047832921625871e-05,
+      "loss": 2.0738,
+      "step": 5770
+    },
+    {
+      "epoch": 1.8496000000000001,
+      "grad_norm": 7.82886266708374,
+      "learning_rate": 4.0366045362676844e-05,
+      "loss": 1.9776,
+      "step": 5780
+    },
+    {
+      "epoch": 1.8528,
+      "grad_norm": 8.772919654846191,
+      "learning_rate": 4.0253761509094995e-05,
+      "loss": 2.137,
+      "step": 5790
+    },
+    {
+      "epoch": 1.8559999999999999,
+      "grad_norm": 4.685011386871338,
+      "learning_rate": 4.014147765551314e-05,
+      "loss": 2.3822,
+      "step": 5800
+    },
+    {
+      "epoch": 1.8592,
+      "grad_norm": 6.074718952178955,
+      "learning_rate": 4.002919380193129e-05,
+      "loss": 2.2003,
+      "step": 5810
+    },
+    {
+      "epoch": 1.8624,
+      "grad_norm": 6.719532489776611,
+      "learning_rate": 3.9916909948349426e-05,
+      "loss": 1.978,
+      "step": 5820
+    },
+    {
+      "epoch": 1.8656000000000001,
+      "grad_norm": 6.288917064666748,
+      "learning_rate": 3.980462609476757e-05,
+      "loss": 2.0413,
+      "step": 5830
+    },
+    {
+      "epoch": 1.8688,
+      "grad_norm": 3.5031559467315674,
+      "learning_rate": 3.969234224118572e-05,
+      "loss": 2.0542,
+      "step": 5840
+    },
+    {
+      "epoch": 1.8719999999999999,
+      "grad_norm": 7.236006259918213,
+      "learning_rate": 3.9580058387603865e-05,
+      "loss": 2.1693,
+      "step": 5850
+    },
+    {
+      "epoch": 1.8752,
+      "grad_norm": 5.247088432312012,
+      "learning_rate": 3.946777453402201e-05,
+      "loss": 2.2759,
+      "step": 5860
+    },
+    {
+      "epoch": 1.8784,
+      "grad_norm": 4.443241596221924,
+      "learning_rate": 3.935549068044015e-05,
+      "loss": 2.1932,
+      "step": 5870
+    },
+    {
+      "epoch": 1.8816000000000002,
+      "grad_norm": 3.587862968444824,
+      "learning_rate": 3.9243206826858297e-05,
+      "loss": 2.2292,
+      "step": 5880
+    },
+    {
+      "epoch": 1.8848,
+      "grad_norm": 2.6182422637939453,
+      "learning_rate": 3.913092297327645e-05,
+      "loss": 1.9304,
+      "step": 5890
+    },
+    {
+      "epoch": 1.888,
+      "grad_norm": 3.457829713821411,
+      "learning_rate": 3.901863911969459e-05,
+      "loss": 2.1348,
+      "step": 5900
+    },
+    {
+      "epoch": 1.8912,
+      "grad_norm": 4.102409839630127,
+      "learning_rate": 3.8906355266112735e-05,
+      "loss": 2.0977,
+      "step": 5910
+    },
+    {
+      "epoch": 1.8944,
+      "grad_norm": 6.1073994636535645,
+      "learning_rate": 3.879407141253088e-05,
+      "loss": 2.171,
+      "step": 5920
+    },
+    {
+      "epoch": 1.8976,
+      "grad_norm": 4.545835494995117,
+      "learning_rate": 3.868178755894902e-05,
+      "loss": 2.1105,
+      "step": 5930
+    },
+    {
+      "epoch": 1.9008,
+      "grad_norm": 6.679617881774902,
+      "learning_rate": 3.8569503705367174e-05,
+      "loss": 2.0686,
+      "step": 5940
+    },
+    {
+      "epoch": 1.904,
+      "grad_norm": 8.60380744934082,
+      "learning_rate": 3.845721985178532e-05,
+      "loss": 2.1866,
+      "step": 5950
+    },
+    {
+      "epoch": 1.9072,
+      "grad_norm": 5.819077014923096,
+      "learning_rate": 3.834493599820346e-05,
+      "loss": 2.1823,
+      "step": 5960
+    },
+    {
+      "epoch": 1.9104,
+      "grad_norm": 3.7871108055114746,
+      "learning_rate": 3.8232652144621605e-05,
+      "loss": 2.1523,
+      "step": 5970
+    },
+    {
+      "epoch": 1.9136,
+      "grad_norm": 5.970479965209961,
+      "learning_rate": 3.812036829103975e-05,
+      "loss": 2.2493,
+      "step": 5980
+    },
+    {
+      "epoch": 1.9167999999999998,
+      "grad_norm": 10.0,
+      "learning_rate": 3.80080844374579e-05,
+      "loss": 4.3472,
+      "step": 5990
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 8.660836219787598,
+      "learning_rate": 3.789580058387604e-05,
+      "loss": 2.2363,
+      "step": 6000
+    },
+    {
+      "epoch": 1.9232,
+      "grad_norm": 7.8099894523620605,
+      "learning_rate": 3.778351673029418e-05,
+      "loss": 2.1017,
+      "step": 6010
+    },
+    {
+      "epoch": 1.9264000000000001,
+      "grad_norm": 5.9949445724487305,
+      "learning_rate": 3.767123287671233e-05,
+      "loss": 2.1996,
+      "step": 6020
+    },
+    {
+      "epoch": 1.9296,
+      "grad_norm": 6.623493671417236,
+      "learning_rate": 3.7558949023130475e-05,
+      "loss": 2.1646,
+      "step": 6030
+    },
+    {
+      "epoch": 1.9327999999999999,
+      "grad_norm": 7.280769348144531,
+      "learning_rate": 3.744666516954862e-05,
+      "loss": 2.1039,
+      "step": 6040
+    },
+    {
+      "epoch": 1.936,
+      "grad_norm": 5.2272257804870605,
+      "learning_rate": 3.733438131596676e-05,
+      "loss": 2.1895,
+      "step": 6050
+    },
+    {
+      "epoch": 1.9392,
+      "grad_norm": 6.899860858917236,
+      "learning_rate": 3.722209746238491e-05,
+      "loss": 2.1616,
+      "step": 6060
+    },
+    {
+      "epoch": 1.9424000000000001,
+      "grad_norm": 7.497065544128418,
+      "learning_rate": 3.710981360880306e-05,
+      "loss": 2.1634,
+      "step": 6070
+    },
+    {
+      "epoch": 1.9456,
+      "grad_norm": 4.0999956130981445,
+      "learning_rate": 3.69975297552212e-05,
+      "loss": 2.0569,
+      "step": 6080
+    },
+    {
+      "epoch": 1.9487999999999999,
+      "grad_norm": 3.794092893600464,
+      "learning_rate": 3.6885245901639346e-05,
+      "loss": 2.1319,
+      "step": 6090
+    },
+    {
+      "epoch": 1.952,
+      "grad_norm": 5.509084224700928,
+      "learning_rate": 3.677296204805749e-05,
+      "loss": 2.1621,
+      "step": 6100
+    },
+    {
+      "epoch": 1.9552,
+      "grad_norm": 6.076055526733398,
+      "learning_rate": 3.666067819447563e-05,
+      "loss": 2.2731,
+      "step": 6110
+    },
+    {
+      "epoch": 1.9584000000000001,
+      "grad_norm": 3.3105826377868652,
+      "learning_rate": 3.6548394340893784e-05,
+      "loss": 1.9689,
+      "step": 6120
+    },
+    {
+      "epoch": 1.9616,
+      "grad_norm": 5.338077068328857,
+      "learning_rate": 3.643611048731193e-05,
+      "loss": 2.1141,
+      "step": 6130
+    },
+    {
+      "epoch": 1.9647999999999999,
+      "grad_norm": 5.527437210083008,
+      "learning_rate": 3.632382663373007e-05,
+      "loss": 1.9314,
+      "step": 6140
+    },
+    {
+      "epoch": 1.968,
+      "grad_norm": 3.655165195465088,
+      "learning_rate": 3.6211542780148216e-05,
+      "loss": 2.0693,
+      "step": 6150
+    },
+    {
+      "epoch": 1.9712,
+      "grad_norm": 9.39285659790039,
+      "learning_rate": 3.609925892656636e-05,
+      "loss": 2.0442,
+      "step": 6160
+    },
+    {
+      "epoch": 1.9744000000000002,
+      "grad_norm": 8.24338150024414,
+      "learning_rate": 3.598697507298451e-05,
+      "loss": 2.3283,
+      "step": 6170
+    },
+    {
+      "epoch": 1.9776,
+      "grad_norm": 4.415241718292236,
+      "learning_rate": 3.587469121940265e-05,
+      "loss": 2.1328,
+      "step": 6180
+    },
+    {
+      "epoch": 1.9808,
+      "grad_norm": 5.860321044921875,
+      "learning_rate": 3.57624073658208e-05,
+      "loss": 2.0253,
+      "step": 6190
+    },
+    {
+      "epoch": 1.984,
+      "grad_norm": 5.472553253173828,
+      "learning_rate": 3.565012351223894e-05,
+      "loss": 2.0622,
+      "step": 6200
+    },
+    {
+      "epoch": 1.9872,
+      "grad_norm": 4.431563854217529,
+      "learning_rate": 3.5537839658657086e-05,
+      "loss": 1.9528,
+      "step": 6210
+    },
+    {
+      "epoch": 1.9904,
+      "grad_norm": 7.67194938659668,
+      "learning_rate": 3.542555580507524e-05,
+      "loss": 2.0531,
+      "step": 6220
+    },
+    {
+      "epoch": 1.9936,
+      "grad_norm": 3.6711692810058594,
+      "learning_rate": 3.5313271951493374e-05,
+      "loss": 2.0059,
+      "step": 6230
+    },
+    {
+      "epoch": 1.9968,
+      "grad_norm": 4.1530890464782715,
+      "learning_rate": 3.5200988097911524e-05,
+      "loss": 2.032,
+      "step": 6240
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 4.8354926109313965,
+      "learning_rate": 3.508870424432967e-05,
+      "loss": 2.0153,
+      "step": 6250
+    },
+    {
+      "epoch": 2.0032,
+      "grad_norm": 4.306488037109375,
+      "learning_rate": 3.497642039074781e-05,
+      "loss": 2.0774,
+      "step": 6260
+    },
+    {
+      "epoch": 2.0064,
+      "grad_norm": 4.352275371551514,
+      "learning_rate": 3.4864136537165956e-05,
+      "loss": 2.0359,
+      "step": 6270
+    },
+    {
+      "epoch": 2.0096,
+      "grad_norm": 7.978499889373779,
+      "learning_rate": 3.47518526835841e-05,
+      "loss": 2.029,
+      "step": 6280
+    },
+    {
+      "epoch": 2.0128,
+      "grad_norm": 6.480661869049072,
+      "learning_rate": 3.4639568830002244e-05,
+      "loss": 2.0506,
+      "step": 6290
+    },
+    {
+      "epoch": 2.016,
+      "grad_norm": 3.58092999458313,
+      "learning_rate": 3.4527284976420395e-05,
+      "loss": 2.0744,
+      "step": 6300
+    },
+    {
+      "epoch": 2.0192,
+      "grad_norm": 5.924591064453125,
+      "learning_rate": 3.441500112283854e-05,
+      "loss": 2.1674,
+      "step": 6310
+    },
+    {
+      "epoch": 2.0224,
+      "grad_norm": 6.986868381500244,
+      "learning_rate": 3.430271726925668e-05,
+      "loss": 2.1415,
+      "step": 6320
+    },
+    {
+      "epoch": 2.0256,
+      "grad_norm": 4.46516752243042,
+      "learning_rate": 3.4190433415674826e-05,
+      "loss": 2.119,
+      "step": 6330
+    },
+    {
+      "epoch": 2.0288,
+      "grad_norm": 5.372344493865967,
+      "learning_rate": 3.407814956209297e-05,
+      "loss": 2.0115,
+      "step": 6340
+    },
+    {
+      "epoch": 2.032,
+      "grad_norm": 5.572484970092773,
+      "learning_rate": 3.396586570851112e-05,
+      "loss": 2.0578,
+      "step": 6350
+    },
+    {
+      "epoch": 2.0352,
+      "grad_norm": 4.962610244750977,
+      "learning_rate": 3.3853581854929265e-05,
+      "loss": 2.2153,
+      "step": 6360
+    },
+    {
+      "epoch": 2.0384,
+      "grad_norm": 4.404867649078369,
+      "learning_rate": 3.374129800134741e-05,
+      "loss": 2.0666,
+      "step": 6370
+    },
+    {
+      "epoch": 2.0416,
+      "grad_norm": 4.287213325500488,
+      "learning_rate": 3.362901414776555e-05,
+      "loss": 2.2205,
+      "step": 6380
+    },
+    {
+      "epoch": 2.0448,
+      "grad_norm": 7.050772666931152,
+      "learning_rate": 3.3516730294183697e-05,
+      "loss": 2.0955,
+      "step": 6390
+    },
+    {
+      "epoch": 2.048,
+      "grad_norm": 6.098924160003662,
+      "learning_rate": 3.340444644060185e-05,
+      "loss": 2.1016,
+      "step": 6400
+    },
+    {
+      "epoch": 2.0512,
+      "grad_norm": 6.557415962219238,
+      "learning_rate": 3.3292162587019984e-05,
+      "loss": 2.2017,
+      "step": 6410
+    },
+    {
+      "epoch": 2.0544,
+      "grad_norm": 6.623813152313232,
+      "learning_rate": 3.3179878733438135e-05,
+      "loss": 2.0959,
+      "step": 6420
+    },
+    {
+      "epoch": 2.0576,
+      "grad_norm": 6.275115966796875,
+      "learning_rate": 3.306759487985628e-05,
+      "loss": 2.16,
+      "step": 6430
+    },
+    {
+      "epoch": 2.0608,
+      "grad_norm": 5.501457214355469,
+      "learning_rate": 3.295531102627442e-05,
+      "loss": 2.063,
+      "step": 6440
+    },
+    {
+      "epoch": 2.064,
+      "grad_norm": 3.797224760055542,
+      "learning_rate": 3.284302717269257e-05,
+      "loss": 1.9966,
+      "step": 6450
+    },
+    {
+      "epoch": 2.0672,
+      "grad_norm": 9.272015571594238,
+      "learning_rate": 3.273074331911071e-05,
+      "loss": 2.1951,
+      "step": 6460
+    },
+    {
+      "epoch": 2.0704,
+      "grad_norm": 3.943631172180176,
+      "learning_rate": 3.261845946552886e-05,
+      "loss": 1.997,
+      "step": 6470
+    },
+    {
+      "epoch": 2.0736,
+      "grad_norm": 7.709688663482666,
+      "learning_rate": 3.2506175611947005e-05,
+      "loss": 2.2323,
+      "step": 6480
+    },
+    {
+      "epoch": 2.0768,
+      "grad_norm": 8.620623588562012,
+      "learning_rate": 3.239389175836515e-05,
+      "loss": 2.2165,
+      "step": 6490
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 4.471000671386719,
+      "learning_rate": 3.228160790478329e-05,
+      "loss": 2.177,
+      "step": 6500
+    },
+    {
+      "epoch": 2.0832,
+      "grad_norm": 4.813187122344971,
+      "learning_rate": 3.216932405120144e-05,
+      "loss": 2.1923,
+      "step": 6510
+    },
+    {
+      "epoch": 2.0864,
+      "grad_norm": 4.149843692779541,
+      "learning_rate": 3.205704019761958e-05,
+      "loss": 2.2597,
+      "step": 6520
+    },
+    {
+      "epoch": 2.0896,
+      "grad_norm": 4.905725002288818,
+      "learning_rate": 3.194475634403773e-05,
+      "loss": 2.0405,
+      "step": 6530
+    },
+    {
+      "epoch": 2.0928,
+      "grad_norm": 6.384262561798096,
+      "learning_rate": 3.1832472490455875e-05,
+      "loss": 2.2226,
+      "step": 6540
+    },
+    {
+      "epoch": 2.096,
+      "grad_norm": 7.396153450012207,
+      "learning_rate": 3.172018863687402e-05,
+      "loss": 2.0731,
+      "step": 6550
+    },
+    {
+      "epoch": 2.0992,
+      "grad_norm": 4.137185096740723,
+      "learning_rate": 3.160790478329216e-05,
+      "loss": 2.049,
+      "step": 6560
+    },
+    {
+      "epoch": 2.1024,
+      "grad_norm": 5.075475215911865,
+      "learning_rate": 3.149562092971031e-05,
+      "loss": 2.1183,
+      "step": 6570
+    },
+    {
+      "epoch": 2.1056,
+      "grad_norm": 5.995041370391846,
+      "learning_rate": 3.138333707612846e-05,
+      "loss": 2.1061,
+      "step": 6580
+    },
+    {
+      "epoch": 2.1088,
+      "grad_norm": 8.12357234954834,
+      "learning_rate": 3.1271053222546595e-05,
+      "loss": 1.9933,
+      "step": 6590
+    },
+    {
+      "epoch": 2.112,
+      "grad_norm": 4.065415382385254,
+      "learning_rate": 3.1158769368964746e-05,
+      "loss": 2.1183,
+      "step": 6600
+    },
+    {
+      "epoch": 2.1152,
+      "grad_norm": 3.1915132999420166,
+      "learning_rate": 3.104648551538289e-05,
+      "loss": 1.9599,
+      "step": 6610
+    },
+    {
+      "epoch": 2.1184,
+      "grad_norm": 6.77773380279541,
+      "learning_rate": 3.093420166180103e-05,
+      "loss": 2.0938,
+      "step": 6620
+    },
+    {
+      "epoch": 2.1216,
+      "grad_norm": 5.450401306152344,
+      "learning_rate": 3.082191780821918e-05,
+      "loss": 2.079,
+      "step": 6630
+    },
+    {
+      "epoch": 2.1248,
+      "grad_norm": 4.6854963302612305,
+      "learning_rate": 3.070963395463732e-05,
+      "loss": 2.1767,
+      "step": 6640
+    },
+    {
+      "epoch": 2.128,
+      "grad_norm": 6.384552955627441,
+      "learning_rate": 3.059735010105547e-05,
+      "loss": 2.0163,
+      "step": 6650
+    },
+    {
+      "epoch": 2.1312,
+      "grad_norm": 4.87563419342041,
+      "learning_rate": 3.0485066247473616e-05,
+      "loss": 2.1001,
+      "step": 6660
+    },
+    {
+      "epoch": 2.1344,
+      "grad_norm": 3.775580406188965,
+      "learning_rate": 3.0372782393891756e-05,
+      "loss": 2.0417,
+      "step": 6670
+    },
+    {
+      "epoch": 2.1376,
+      "grad_norm": 4.963315486907959,
+      "learning_rate": 3.0260498540309907e-05,
+      "loss": 2.1564,
+      "step": 6680
+    },
+    {
+      "epoch": 2.1408,
+      "grad_norm": 7.423795223236084,
+      "learning_rate": 3.0148214686728047e-05,
+      "loss": 2.1124,
+      "step": 6690
+    },
+    {
+      "epoch": 2.144,
+      "grad_norm": 5.808844089508057,
+      "learning_rate": 3.0035930833146198e-05,
+      "loss": 2.0786,
+      "step": 6700
+    },
+    {
+      "epoch": 2.1471999999999998,
+      "grad_norm": 4.398837089538574,
+      "learning_rate": 2.9923646979564342e-05,
+      "loss": 2.1347,
+      "step": 6710
+    },
+    {
+      "epoch": 2.1504,
+      "grad_norm": 9.142531394958496,
+      "learning_rate": 2.9811363125982483e-05,
+      "loss": 1.9574,
+      "step": 6720
+    },
+    {
+      "epoch": 2.1536,
+      "grad_norm": 5.84624719619751,
+      "learning_rate": 2.9699079272400633e-05,
+      "loss": 2.0017,
+      "step": 6730
+    },
+    {
+      "epoch": 2.1568,
+      "grad_norm": 7.081725597381592,
+      "learning_rate": 2.9586795418818774e-05,
+      "loss": 1.959,
+      "step": 6740
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 4.674961566925049,
+      "learning_rate": 2.9474511565236924e-05,
+      "loss": 2.0554,
+      "step": 6750
+    },
+    {
+      "epoch": 2.1632,
+      "grad_norm": 6.768802642822266,
+      "learning_rate": 2.9362227711655065e-05,
+      "loss": 2.0301,
+      "step": 6760
+    },
+    {
+      "epoch": 2.1664,
+      "grad_norm": 3.5873348712921143,
+      "learning_rate": 2.924994385807321e-05,
+      "loss": 2.092,
+      "step": 6770
+    },
+    {
+      "epoch": 2.1696,
+      "grad_norm": 4.120325088500977,
+      "learning_rate": 2.9137660004491356e-05,
+      "loss": 1.9953,
+      "step": 6780
+    },
+    {
+      "epoch": 2.1728,
+      "grad_norm": 7.765623569488525,
+      "learning_rate": 2.90253761509095e-05,
+      "loss": 2.0688,
+      "step": 6790
+    },
+    {
+      "epoch": 2.176,
+      "grad_norm": 4.869895935058594,
+      "learning_rate": 2.8913092297327644e-05,
+      "loss": 2.219,
+      "step": 6800
+    },
+    {
+      "epoch": 2.1792,
+      "grad_norm": 5.694615840911865,
+      "learning_rate": 2.880080844374579e-05,
+      "loss": 2.1398,
+      "step": 6810
+    },
+    {
+      "epoch": 2.1824,
+      "grad_norm": 7.104328632354736,
+      "learning_rate": 2.8688524590163935e-05,
+      "loss": 2.0682,
+      "step": 6820
+    },
+    {
+      "epoch": 2.1856,
+      "grad_norm": 5.0297932624816895,
+      "learning_rate": 2.8576240736582082e-05,
+      "loss": 2.052,
+      "step": 6830
+    },
+    {
+      "epoch": 2.1888,
+      "grad_norm": 4.520491123199463,
+      "learning_rate": 2.8463956883000226e-05,
+      "loss": 2.211,
+      "step": 6840
+    },
+    {
+      "epoch": 2.192,
+      "grad_norm": 3.5239853858947754,
+      "learning_rate": 2.835167302941837e-05,
+      "loss": 1.9884,
+      "step": 6850
+    },
+    {
+      "epoch": 2.1952,
+      "grad_norm": 7.16238260269165,
+      "learning_rate": 2.8239389175836517e-05,
+      "loss": 2.1121,
+      "step": 6860
+    },
+    {
+      "epoch": 2.1984,
+      "grad_norm": 4.042484760284424,
+      "learning_rate": 2.812710532225466e-05,
+      "loss": 2.1276,
+      "step": 6870
+    },
+    {
+      "epoch": 2.2016,
+      "grad_norm": 4.16971492767334,
+      "learning_rate": 2.801482146867281e-05,
+      "loss": 2.0987,
+      "step": 6880
+    },
+    {
+      "epoch": 2.2048,
+      "grad_norm": 7.618093013763428,
+      "learning_rate": 2.7902537615090953e-05,
+      "loss": 2.3154,
+      "step": 6890
+    },
+    {
+      "epoch": 2.208,
+      "grad_norm": 5.255606174468994,
+      "learning_rate": 2.7790253761509093e-05,
+      "loss": 2.1601,
+      "step": 6900
+    },
+    {
+      "epoch": 2.2112,
+      "grad_norm": 4.716766834259033,
+      "learning_rate": 2.7677969907927244e-05,
+      "loss": 1.9644,
+      "step": 6910
+    },
+    {
+      "epoch": 2.2144,
+      "grad_norm": 4.157872676849365,
+      "learning_rate": 2.7565686054345384e-05,
+      "loss": 1.9962,
+      "step": 6920
+    },
+    {
+      "epoch": 2.2176,
+      "grad_norm": 6.823659896850586,
+      "learning_rate": 2.7453402200763535e-05,
+      "loss": 2.1439,
+      "step": 6930
+    },
+    {
+      "epoch": 2.2208,
+      "grad_norm": 5.478399753570557,
+      "learning_rate": 2.7341118347181675e-05,
+      "loss": 1.9356,
+      "step": 6940
+    },
+    {
+      "epoch": 2.224,
+      "grad_norm": 4.689790725708008,
+      "learning_rate": 2.722883449359982e-05,
+      "loss": 2.1821,
+      "step": 6950
+    },
+    {
+      "epoch": 2.2272,
+      "grad_norm": 5.204911231994629,
+      "learning_rate": 2.7116550640017967e-05,
+      "loss": 2.0843,
+      "step": 6960
+    },
+    {
+      "epoch": 2.2304,
+      "grad_norm": 6.285693168640137,
+      "learning_rate": 2.700426678643611e-05,
+      "loss": 2.0767,
+      "step": 6970
+    },
+    {
+      "epoch": 2.2336,
+      "grad_norm": 8.708101272583008,
+      "learning_rate": 2.6891982932854258e-05,
+      "loss": 2.1958,
+      "step": 6980
+    },
+    {
+      "epoch": 2.2368,
+      "grad_norm": 6.604062080383301,
+      "learning_rate": 2.6779699079272402e-05,
+      "loss": 2.1266,
+      "step": 6990
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 5.105891704559326,
+      "learning_rate": 2.6667415225690546e-05,
+      "loss": 2.2054,
+      "step": 7000
+    },
+    {
+      "epoch": 2.2432,
+      "grad_norm": 5.714686393737793,
+      "learning_rate": 2.6555131372108693e-05,
+      "loss": 1.9358,
+      "step": 7010
+    },
+    {
+      "epoch": 2.2464,
+      "grad_norm": 3.433338165283203,
+      "learning_rate": 2.6442847518526837e-05,
+      "loss": 2.076,
+      "step": 7020
+    },
+    {
+      "epoch": 2.2496,
+      "grad_norm": 4.48581600189209,
+      "learning_rate": 2.6330563664944984e-05,
+      "loss": 2.0635,
+      "step": 7030
+    },
+    {
+      "epoch": 2.2528,
+      "grad_norm": 5.338774681091309,
+      "learning_rate": 2.6218279811363128e-05,
+      "loss": 2.1363,
+      "step": 7040
+    },
+    {
+      "epoch": 2.2560000000000002,
+      "grad_norm": 7.590831279754639,
+      "learning_rate": 2.6105995957781272e-05,
+      "loss": 1.9187,
+      "step": 7050
+    },
+    {
+      "epoch": 2.2592,
+      "grad_norm": 5.141582489013672,
+      "learning_rate": 2.599371210419942e-05,
+      "loss": 2.1299,
+      "step": 7060
+    },
+    {
+      "epoch": 2.2624,
+      "grad_norm": 5.08326530456543,
+      "learning_rate": 2.5881428250617563e-05,
+      "loss": 2.1819,
+      "step": 7070
+    },
+    {
+      "epoch": 2.2656,
+      "grad_norm": 5.986049175262451,
+      "learning_rate": 2.5769144397035704e-05,
+      "loss": 2.1374,
+      "step": 7080
+    },
+    {
+      "epoch": 2.2688,
+      "grad_norm": 4.5929999351501465,
+      "learning_rate": 2.5656860543453854e-05,
+      "loss": 1.9291,
+      "step": 7090
+    },
+    {
+      "epoch": 2.2720000000000002,
+      "grad_norm": 3.4489879608154297,
+      "learning_rate": 2.5544576689871995e-05,
+      "loss": 2.0773,
+      "step": 7100
+    },
+    {
+      "epoch": 2.2752,
+      "grad_norm": 4.485833168029785,
+      "learning_rate": 2.5432292836290145e-05,
+      "loss": 2.0115,
+      "step": 7110
+    },
+    {
+      "epoch": 2.2784,
+      "grad_norm": 6.017434120178223,
+      "learning_rate": 2.5320008982708286e-05,
+      "loss": 2.0676,
+      "step": 7120
+    },
+    {
+      "epoch": 2.2816,
+      "grad_norm": 6.2123332023620605,
+      "learning_rate": 2.520772512912643e-05,
+      "loss": 2.0369,
+      "step": 7130
+    },
+    {
+      "epoch": 2.2848,
+      "grad_norm": 3.7782444953918457,
+      "learning_rate": 2.509544127554458e-05,
+      "loss": 2.0942,
+      "step": 7140
+    },
+    {
+      "epoch": 2.288,
+      "grad_norm": 7.148087024688721,
+      "learning_rate": 2.498315742196272e-05,
+      "loss": 2.1302,
+      "step": 7150
+    },
+    {
+      "epoch": 2.2912,
+      "grad_norm": 5.7953362464904785,
+      "learning_rate": 2.487087356838087e-05,
+      "loss": 2.1838,
+      "step": 7160
+    },
+    {
+      "epoch": 2.2944,
+      "grad_norm": 3.9319803714752197,
+      "learning_rate": 2.4758589714799012e-05,
+      "loss": 2.0159,
+      "step": 7170
+    },
+    {
+      "epoch": 2.2976,
+      "grad_norm": 4.671140193939209,
+      "learning_rate": 2.464630586121716e-05,
+      "loss": 1.9249,
+      "step": 7180
+    },
+    {
+      "epoch": 2.3008,
+      "grad_norm": 4.661734104156494,
+      "learning_rate": 2.4534022007635303e-05,
+      "loss": 2.0409,
+      "step": 7190
+    },
+    {
+      "epoch": 2.304,
+      "grad_norm": 7.39704704284668,
+      "learning_rate": 2.4421738154053447e-05,
+      "loss": 2.1667,
+      "step": 7200
+    },
+    {
+      "epoch": 2.3072,
+      "grad_norm": 8.355154037475586,
+      "learning_rate": 2.4309454300471595e-05,
+      "loss": 2.1305,
+      "step": 7210
+    },
+    {
+      "epoch": 2.3104,
+      "grad_norm": 5.1650285720825195,
+      "learning_rate": 2.419717044688974e-05,
+      "loss": 2.1391,
+      "step": 7220
+    },
+    {
+      "epoch": 2.3136,
+      "grad_norm": 6.877825736999512,
+      "learning_rate": 2.4084886593307886e-05,
+      "loss": 2.0748,
+      "step": 7230
+    },
+    {
+      "epoch": 2.3168,
+      "grad_norm": 4.656685829162598,
+      "learning_rate": 2.3972602739726026e-05,
+      "loss": 2.0812,
+      "step": 7240
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 5.901729106903076,
+      "learning_rate": 2.3860318886144174e-05,
+      "loss": 2.0358,
+      "step": 7250
+    },
+    {
+      "epoch": 2.3232,
+      "grad_norm": 6.996543884277344,
+      "learning_rate": 2.3748035032562318e-05,
+      "loss": 2.1514,
+      "step": 7260
+    },
+    {
+      "epoch": 2.3264,
+      "grad_norm": 4.831353187561035,
+      "learning_rate": 2.3635751178980465e-05,
+      "loss": 2.0404,
+      "step": 7270
+    },
+    {
+      "epoch": 2.3296,
+      "grad_norm": 7.972875595092773,
+      "learning_rate": 2.352346732539861e-05,
+      "loss": 1.9482,
+      "step": 7280
+    },
+    {
+      "epoch": 2.3327999999999998,
+      "grad_norm": 7.147056579589844,
+      "learning_rate": 2.3411183471816753e-05,
+      "loss": 2.2954,
+      "step": 7290
+    },
+    {
+      "epoch": 2.336,
+      "grad_norm": 5.505180835723877,
+      "learning_rate": 2.32988996182349e-05,
+      "loss": 2.1596,
+      "step": 7300
+    },
+    {
+      "epoch": 2.3392,
+      "grad_norm": 7.347818374633789,
+      "learning_rate": 2.3186615764653044e-05,
+      "loss": 2.1602,
+      "step": 7310
+    },
+    {
+      "epoch": 2.3424,
+      "grad_norm": 3.6011338233947754,
+      "learning_rate": 2.307433191107119e-05,
+      "loss": 2.0548,
+      "step": 7320
+    },
+    {
+      "epoch": 2.3456,
+      "grad_norm": 3.5904648303985596,
+      "learning_rate": 2.2962048057489335e-05,
+      "loss": 2.0395,
+      "step": 7330
+    },
+    {
+      "epoch": 2.3487999999999998,
+      "grad_norm": 5.380495071411133,
+      "learning_rate": 2.284976420390748e-05,
+      "loss": 2.0023,
+      "step": 7340
+    },
+    {
+      "epoch": 2.352,
+      "grad_norm": 4.518589973449707,
+      "learning_rate": 2.2737480350325623e-05,
+      "loss": 1.9918,
+      "step": 7350
+    },
+    {
+      "epoch": 2.3552,
+      "grad_norm": 5.9018235206604,
+      "learning_rate": 2.262519649674377e-05,
+      "loss": 1.9901,
+      "step": 7360
+    },
+    {
+      "epoch": 2.3584,
+      "grad_norm": 3.6925086975097656,
+      "learning_rate": 2.2512912643161914e-05,
+      "loss": 2.0245,
+      "step": 7370
+    },
+    {
+      "epoch": 2.3616,
+      "grad_norm": 9.259016990661621,
+      "learning_rate": 2.2400628789580058e-05,
+      "loss": 2.2619,
+      "step": 7380
+    },
+    {
+      "epoch": 2.3648,
+      "grad_norm": 5.5859808921813965,
+      "learning_rate": 2.2288344935998205e-05,
+      "loss": 2.1797,
+      "step": 7390
+    },
+    {
+      "epoch": 2.368,
+      "grad_norm": 4.7031683921813965,
+      "learning_rate": 2.217606108241635e-05,
+      "loss": 2.0357,
+      "step": 7400
+    },
+    {
+      "epoch": 2.3712,
+      "grad_norm": 6.260189056396484,
+      "learning_rate": 2.2063777228834496e-05,
+      "loss": 2.1377,
+      "step": 7410
+    },
+    {
+      "epoch": 2.3744,
+      "grad_norm": 4.784355163574219,
+      "learning_rate": 2.195149337525264e-05,
+      "loss": 1.9451,
+      "step": 7420
+    },
+    {
+      "epoch": 2.3776,
+      "grad_norm": 8.991378784179688,
+      "learning_rate": 2.1839209521670784e-05,
+      "loss": 2.0388,
+      "step": 7430
+    },
+    {
+      "epoch": 2.3808,
+      "grad_norm": 6.376221656799316,
+      "learning_rate": 2.1726925668088928e-05,
+      "loss": 2.379,
+      "step": 7440
+    },
+    {
+      "epoch": 2.384,
+      "grad_norm": 5.871402740478516,
+      "learning_rate": 2.1614641814507075e-05,
+      "loss": 2.1915,
+      "step": 7450
+    },
+    {
+      "epoch": 2.3872,
+      "grad_norm": 5.426553249359131,
+      "learning_rate": 2.150235796092522e-05,
+      "loss": 2.0817,
+      "step": 7460
+    },
+    {
+      "epoch": 2.3904,
+      "grad_norm": 5.839360237121582,
+      "learning_rate": 2.1390074107343367e-05,
+      "loss": 1.924,
+      "step": 7470
+    },
+    {
+      "epoch": 2.3936,
+      "grad_norm": 2.95749568939209,
+      "learning_rate": 2.127779025376151e-05,
+      "loss": 2.0216,
+      "step": 7480
+    },
+    {
+      "epoch": 2.3968,
+      "grad_norm": 8.544522285461426,
+      "learning_rate": 2.1165506400179654e-05,
+      "loss": 1.905,
+      "step": 7490
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 5.736949443817139,
+      "learning_rate": 2.10532225465978e-05,
+      "loss": 1.9288,
+      "step": 7500
+    },
+    {
+      "epoch": 2.4032,
+      "grad_norm": 4.557583808898926,
+      "learning_rate": 2.0940938693015946e-05,
+      "loss": 1.9417,
+      "step": 7510
+    },
+    {
+      "epoch": 2.4064,
+      "grad_norm": 5.55007266998291,
+      "learning_rate": 2.082865483943409e-05,
+      "loss": 2.3085,
+      "step": 7520
+    },
+    {
+      "epoch": 2.4096,
+      "grad_norm": 4.35857629776001,
+      "learning_rate": 2.0716370985852233e-05,
+      "loss": 1.9185,
+      "step": 7530
+    },
+    {
+      "epoch": 2.4128,
+      "grad_norm": 6.802826404571533,
+      "learning_rate": 2.060408713227038e-05,
+      "loss": 2.1785,
+      "step": 7540
+    },
+    {
+      "epoch": 2.416,
+      "grad_norm": 7.604827880859375,
+      "learning_rate": 2.0491803278688525e-05,
+      "loss": 1.9825,
+      "step": 7550
+    },
+    {
+      "epoch": 2.4192,
+      "grad_norm": 3.024289846420288,
+      "learning_rate": 2.0379519425106672e-05,
+      "loss": 1.9379,
+      "step": 7560
+    },
+    {
+      "epoch": 2.4224,
+      "grad_norm": 4.845406532287598,
+      "learning_rate": 2.0267235571524816e-05,
+      "loss": 2.0376,
+      "step": 7570
+    },
+    {
+      "epoch": 2.4256,
+      "grad_norm": 7.506366729736328,
+      "learning_rate": 2.015495171794296e-05,
+      "loss": 2.0193,
+      "step": 7580
+    },
+    {
+      "epoch": 2.4288,
+      "grad_norm": 7.756492614746094,
+      "learning_rate": 2.0042667864361107e-05,
+      "loss": 2.0873,
+      "step": 7590
+    },
+    {
+      "epoch": 2.432,
+      "grad_norm": 6.46793270111084,
+      "learning_rate": 1.993038401077925e-05,
+      "loss": 2.1243,
+      "step": 7600
+    },
+    {
+      "epoch": 2.4352,
+      "grad_norm": 6.6785359382629395,
+      "learning_rate": 1.9818100157197398e-05,
+      "loss": 2.1875,
+      "step": 7610
+    },
+    {
+      "epoch": 2.4384,
+      "grad_norm": 3.648634433746338,
+      "learning_rate": 1.970581630361554e-05,
+      "loss": 2.1792,
+      "step": 7620
+    },
+    {
+      "epoch": 2.4416,
+      "grad_norm": 5.907644271850586,
+      "learning_rate": 1.9593532450033686e-05,
+      "loss": 2.0782,
+      "step": 7630
+    },
+    {
+      "epoch": 2.4448,
+      "grad_norm": 7.055563926696777,
+      "learning_rate": 1.948124859645183e-05,
+      "loss": 2.2235,
+      "step": 7640
+    },
+    {
+      "epoch": 2.448,
+      "grad_norm": 6.187650203704834,
+      "learning_rate": 1.9368964742869977e-05,
+      "loss": 2.0072,
+      "step": 7650
+    },
+    {
+      "epoch": 2.4512,
+      "grad_norm": 7.704046726226807,
+      "learning_rate": 1.925668088928812e-05,
+      "loss": 2.1009,
+      "step": 7660
+    },
+    {
+      "epoch": 2.4544,
+      "grad_norm": 5.42090368270874,
+      "learning_rate": 1.9144397035706265e-05,
+      "loss": 1.9614,
+      "step": 7670
+    },
+    {
+      "epoch": 2.4576000000000002,
+      "grad_norm": 4.029887676239014,
+      "learning_rate": 1.9032113182124412e-05,
+      "loss": 2.1542,
+      "step": 7680
+    },
+    {
+      "epoch": 2.4608,
+      "grad_norm": 6.325004577636719,
+      "learning_rate": 1.8919829328542556e-05,
+      "loss": 2.0916,
+      "step": 7690
+    },
+    {
+      "epoch": 2.464,
+      "grad_norm": 3.8288917541503906,
+      "learning_rate": 1.8807545474960703e-05,
+      "loss": 2.0036,
+      "step": 7700
+    },
+    {
+      "epoch": 2.4672,
+      "grad_norm": 5.6230645179748535,
+      "learning_rate": 1.8695261621378844e-05,
+      "loss": 2.1405,
+      "step": 7710
+    },
+    {
+      "epoch": 2.4704,
+      "grad_norm": 6.5077667236328125,
+      "learning_rate": 1.858297776779699e-05,
+      "loss": 2.1522,
+      "step": 7720
+    },
+    {
+      "epoch": 2.4736000000000002,
+      "grad_norm": 4.403681755065918,
+      "learning_rate": 1.847069391421514e-05,
+      "loss": 2.1962,
+      "step": 7730
+    },
+    {
+      "epoch": 2.4768,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 1.8358410060633282e-05,
+      "loss": 2.1528,
+      "step": 7740
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 3.427657127380371,
+      "learning_rate": 1.8246126207051426e-05,
+      "loss": 2.0474,
+      "step": 7750
+    },
+    {
+      "epoch": 2.4832,
+      "grad_norm": 8.58780288696289,
+      "learning_rate": 1.813384235346957e-05,
+      "loss": 2.0905,
+      "step": 7760
+    },
+    {
+      "epoch": 2.4864,
+      "grad_norm": 7.026243686676025,
+      "learning_rate": 1.8021558499887718e-05,
+      "loss": 2.1772,
+      "step": 7770
+    },
+    {
+      "epoch": 2.4896,
+      "grad_norm": 5.6110405921936035,
+      "learning_rate": 1.790927464630586e-05,
+      "loss": 2.0321,
+      "step": 7780
+    },
+    {
+      "epoch": 2.4928,
+      "grad_norm": 7.481403350830078,
+      "learning_rate": 1.779699079272401e-05,
+      "loss": 1.8835,
+      "step": 7790
+    },
+    {
+      "epoch": 2.496,
+      "grad_norm": 4.553055286407471,
+      "learning_rate": 1.7684706939142153e-05,
+      "loss": 1.9508,
+      "step": 7800
+    },
+    {
+      "epoch": 2.4992,
+      "grad_norm": 5.52471923828125,
+      "learning_rate": 1.7572423085560296e-05,
+      "loss": 2.2059,
+      "step": 7810
+    },
+    {
+      "epoch": 2.5023999999999997,
+      "grad_norm": 3.0748660564422607,
+      "learning_rate": 1.7460139231978444e-05,
+      "loss": 3.6667,
+      "step": 7820
+    },
+    {
+      "epoch": 2.5056000000000003,
+      "grad_norm": 4.8406982421875,
+      "learning_rate": 1.7347855378396588e-05,
+      "loss": 2.188,
+      "step": 7830
+    },
+    {
+      "epoch": 2.5088,
+      "grad_norm": 6.943801403045654,
+      "learning_rate": 1.7235571524814735e-05,
+      "loss": 1.9868,
+      "step": 7840
+    },
+    {
+      "epoch": 2.512,
+      "grad_norm": 6.901176452636719,
+      "learning_rate": 1.7123287671232875e-05,
+      "loss": 1.9901,
+      "step": 7850
+    },
+    {
+      "epoch": 2.5152,
+      "grad_norm": 5.6528520584106445,
+      "learning_rate": 1.7011003817651023e-05,
+      "loss": 1.9197,
+      "step": 7860
+    },
+    {
+      "epoch": 2.5183999999999997,
+      "grad_norm": 9.42558765411377,
+      "learning_rate": 1.6898719964069167e-05,
+      "loss": 2.2289,
+      "step": 7870
+    },
+    {
+      "epoch": 2.5216,
+      "grad_norm": 6.620875835418701,
+      "learning_rate": 1.6786436110487314e-05,
+      "loss": 2.1969,
+      "step": 7880
+    },
+    {
+      "epoch": 2.5248,
+      "grad_norm": 5.273202896118164,
+      "learning_rate": 1.6674152256905458e-05,
+      "loss": 2.1826,
+      "step": 7890
+    },
+    {
+      "epoch": 2.528,
+      "grad_norm": 6.013164043426514,
+      "learning_rate": 1.6561868403323602e-05,
+      "loss": 2.2718,
+      "step": 7900
+    },
+    {
+      "epoch": 2.5312,
+      "grad_norm": 5.780605792999268,
+      "learning_rate": 1.644958454974175e-05,
+      "loss": 1.9115,
+      "step": 7910
+    },
+    {
+      "epoch": 2.5343999999999998,
+      "grad_norm": 5.694150924682617,
+      "learning_rate": 1.6337300696159893e-05,
+      "loss": 2.2051,
+      "step": 7920
+    },
+    {
+      "epoch": 2.5376,
+      "grad_norm": 6.350783824920654,
+      "learning_rate": 1.622501684257804e-05,
+      "loss": 2.0739,
+      "step": 7930
+    },
+    {
+      "epoch": 2.5408,
+      "grad_norm": 3.847805976867676,
+      "learning_rate": 1.611273298899618e-05,
+      "loss": 2.008,
+      "step": 7940
+    },
+    {
+      "epoch": 2.544,
+      "grad_norm": 5.354320526123047,
+      "learning_rate": 1.6000449135414328e-05,
+      "loss": 1.9905,
+      "step": 7950
+    },
+    {
+      "epoch": 2.5472,
+      "grad_norm": 5.156245231628418,
+      "learning_rate": 1.5888165281832472e-05,
+      "loss": 2.1381,
+      "step": 7960
+    },
+    {
+      "epoch": 2.5504,
+      "grad_norm": 5.384928226470947,
+      "learning_rate": 1.577588142825062e-05,
+      "loss": 2.259,
+      "step": 7970
+    },
+    {
+      "epoch": 2.5536,
+      "grad_norm": 6.737875938415527,
+      "learning_rate": 1.5663597574668763e-05,
+      "loss": 2.206,
+      "step": 7980
+    },
+    {
+      "epoch": 2.5568,
+      "grad_norm": 3.4885711669921875,
+      "learning_rate": 1.5551313721086907e-05,
+      "loss": 2.1322,
+      "step": 7990
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 7.307969093322754,
+      "learning_rate": 1.5439029867505054e-05,
+      "loss": 2.2369,
+      "step": 8000
+    },
+    {
+      "epoch": 2.5632,
+      "grad_norm": 4.461708068847656,
+      "learning_rate": 1.5326746013923198e-05,
+      "loss": 1.9901,
+      "step": 8010
+    },
+    {
+      "epoch": 2.5664,
+      "grad_norm": 4.739271640777588,
+      "learning_rate": 1.5214462160341344e-05,
+      "loss": 2.1096,
+      "step": 8020
+    },
+    {
+      "epoch": 2.5696,
+      "grad_norm": 4.999455451965332,
+      "learning_rate": 1.5102178306759488e-05,
+      "loss": 2.0258,
+      "step": 8030
+    },
+    {
+      "epoch": 2.5728,
+      "grad_norm": 3.893523931503296,
+      "learning_rate": 1.4989894453177633e-05,
+      "loss": 2.0223,
+      "step": 8040
+    },
+    {
+      "epoch": 2.576,
+      "grad_norm": 3.8808035850524902,
+      "learning_rate": 1.4877610599595779e-05,
+      "loss": 2.1654,
+      "step": 8050
+    },
+    {
+      "epoch": 2.5792,
+      "grad_norm": 3.7033116817474365,
+      "learning_rate": 1.4765326746013925e-05,
+      "loss": 2.0741,
+      "step": 8060
+    },
+    {
+      "epoch": 2.5824,
+      "grad_norm": 5.134984493255615,
+      "learning_rate": 1.465304289243207e-05,
+      "loss": 2.1801,
+      "step": 8070
+    },
+    {
+      "epoch": 2.5856,
+      "grad_norm": 5.389072895050049,
+      "learning_rate": 1.4540759038850212e-05,
+      "loss": 2.1353,
+      "step": 8080
+    },
+    {
+      "epoch": 2.5888,
+      "grad_norm": 3.2037174701690674,
+      "learning_rate": 1.4428475185268358e-05,
+      "loss": 2.0169,
+      "step": 8090
+    },
+    {
+      "epoch": 2.592,
+      "grad_norm": 6.975593090057373,
+      "learning_rate": 1.4316191331686504e-05,
+      "loss": 2.0105,
+      "step": 8100
+    },
+    {
+      "epoch": 2.5952,
+      "grad_norm": 4.0726423263549805,
+      "learning_rate": 1.4203907478104649e-05,
+      "loss": 2.07,
+      "step": 8110
+    },
+    {
+      "epoch": 2.5984,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 1.4091623624522796e-05,
+      "loss": 2.1497,
+      "step": 8120
+    },
+    {
+      "epoch": 2.6016,
+      "grad_norm": 3.413611888885498,
+      "learning_rate": 1.3979339770940939e-05,
+      "loss": 1.9266,
+      "step": 8130
+    },
+    {
+      "epoch": 2.6048,
+      "grad_norm": 9.220563888549805,
+      "learning_rate": 1.3867055917359084e-05,
+      "loss": 2.2622,
+      "step": 8140
+    },
+    {
+      "epoch": 2.608,
+      "grad_norm": 4.455341339111328,
+      "learning_rate": 1.375477206377723e-05,
+      "loss": 2.0675,
+      "step": 8150
+    },
+    {
+      "epoch": 2.6112,
+      "grad_norm": 5.176009178161621,
+      "learning_rate": 1.3642488210195375e-05,
+      "loss": 2.171,
+      "step": 8160
+    },
+    {
+      "epoch": 2.6144,
+      "grad_norm": 3.342127561569214,
+      "learning_rate": 1.3530204356613518e-05,
+      "loss": 2.1158,
+      "step": 8170
+    },
+    {
+      "epoch": 2.6176,
+      "grad_norm": 3.67814564704895,
+      "learning_rate": 1.3417920503031663e-05,
+      "loss": 2.1136,
+      "step": 8180
+    },
+    {
+      "epoch": 2.6208,
+      "grad_norm": 7.2132391929626465,
+      "learning_rate": 1.330563664944981e-05,
+      "loss": 2.1383,
+      "step": 8190
+    },
+    {
+      "epoch": 2.624,
+      "grad_norm": 5.090510845184326,
+      "learning_rate": 1.3193352795867956e-05,
+      "loss": 2.0992,
+      "step": 8200
+    },
+    {
+      "epoch": 2.6272,
+      "grad_norm": 4.983635902404785,
+      "learning_rate": 1.3081068942286102e-05,
+      "loss": 2.0596,
+      "step": 8210
+    },
+    {
+      "epoch": 2.6304,
+      "grad_norm": 4.354836463928223,
+      "learning_rate": 1.2968785088704244e-05,
+      "loss": 2.053,
+      "step": 8220
+    },
+    {
+      "epoch": 2.6336,
+      "grad_norm": 4.732203006744385,
+      "learning_rate": 1.285650123512239e-05,
+      "loss": 1.9845,
+      "step": 8230
+    },
+    {
+      "epoch": 2.6368,
+      "grad_norm": 3.1118311882019043,
+      "learning_rate": 1.2744217381540535e-05,
+      "loss": 1.9879,
+      "step": 8240
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 4.70589542388916,
+      "learning_rate": 1.263193352795868e-05,
+      "loss": 2.0729,
+      "step": 8250
+    },
+    {
+      "epoch": 2.6432,
+      "grad_norm": 5.572944164276123,
+      "learning_rate": 1.2519649674376826e-05,
+      "loss": 2.1209,
+      "step": 8260
+    },
+    {
+      "epoch": 2.6464,
+      "grad_norm": 9.762155532836914,
+      "learning_rate": 1.240736582079497e-05,
+      "loss": 2.2191,
+      "step": 8270
+    },
+    {
+      "epoch": 2.6496,
+      "grad_norm": 4.850767612457275,
+      "learning_rate": 1.2295081967213116e-05,
+      "loss": 2.1546,
+      "step": 8280
+    },
+    {
+      "epoch": 2.6528,
+      "grad_norm": 4.146419525146484,
+      "learning_rate": 1.2182798113631261e-05,
+      "loss": 2.096,
+      "step": 8290
+    },
+    {
+      "epoch": 2.656,
+      "grad_norm": 4.993535995483398,
+      "learning_rate": 1.2070514260049405e-05,
+      "loss": 2.1326,
+      "step": 8300
+    },
+    {
+      "epoch": 2.6592000000000002,
+      "grad_norm": 6.308148384094238,
+      "learning_rate": 1.1958230406467551e-05,
+      "loss": 2.1873,
+      "step": 8310
+    },
+    {
+      "epoch": 2.6624,
+      "grad_norm": 7.768478870391846,
+      "learning_rate": 1.1845946552885695e-05,
+      "loss": 2.1019,
+      "step": 8320
+    },
+    {
+      "epoch": 2.6656,
+      "grad_norm": 3.5128209590911865,
+      "learning_rate": 1.173366269930384e-05,
+      "loss": 2.06,
+      "step": 8330
+    },
+    {
+      "epoch": 2.6688,
+      "grad_norm": 8.1862211227417,
+      "learning_rate": 1.1621378845721986e-05,
+      "loss": 2.0348,
+      "step": 8340
+    },
+    {
+      "epoch": 2.672,
+      "grad_norm": 4.3322625160217285,
+      "learning_rate": 1.150909499214013e-05,
+      "loss": 2.1624,
+      "step": 8350
+    },
+    {
+      "epoch": 2.6752000000000002,
+      "grad_norm": 6.597864627838135,
+      "learning_rate": 1.1396811138558275e-05,
+      "loss": 2.0585,
+      "step": 8360
+    },
+    {
+      "epoch": 2.6784,
+      "grad_norm": 4.030665874481201,
+      "learning_rate": 1.1284527284976421e-05,
+      "loss": 2.0265,
+      "step": 8370
+    },
+    {
+      "epoch": 2.6816,
+      "grad_norm": 5.775618553161621,
+      "learning_rate": 1.1172243431394567e-05,
+      "loss": 2.0382,
+      "step": 8380
+    },
+    {
+      "epoch": 2.6848,
+      "grad_norm": 3.6797897815704346,
+      "learning_rate": 1.105995957781271e-05,
+      "loss": 2.0573,
+      "step": 8390
+    },
+    {
+      "epoch": 2.6879999999999997,
+      "grad_norm": 4.699132919311523,
+      "learning_rate": 1.0947675724230856e-05,
+      "loss": 2.0572,
+      "step": 8400
+    },
+    {
+      "epoch": 2.6912000000000003,
+      "grad_norm": 3.724128246307373,
+      "learning_rate": 1.0835391870649002e-05,
+      "loss": 2.0878,
+      "step": 8410
+    },
+    {
+      "epoch": 2.6944,
+      "grad_norm": 6.061574459075928,
+      "learning_rate": 1.0723108017067146e-05,
+      "loss": 2.3014,
+      "step": 8420
+    },
+    {
+      "epoch": 2.6976,
+      "grad_norm": 7.603193759918213,
+      "learning_rate": 1.0610824163485291e-05,
+      "loss": 2.0776,
+      "step": 8430
+    },
+    {
+      "epoch": 2.7008,
+      "grad_norm": 4.737987041473389,
+      "learning_rate": 1.0498540309903437e-05,
+      "loss": 2.075,
+      "step": 8440
+    },
+    {
+      "epoch": 2.7039999999999997,
+      "grad_norm": 3.1090354919433594,
+      "learning_rate": 1.0386256456321582e-05,
+      "loss": 1.9306,
+      "step": 8450
+    },
+    {
+      "epoch": 2.7072000000000003,
+      "grad_norm": 4.949204921722412,
+      "learning_rate": 1.0273972602739726e-05,
+      "loss": 1.9749,
+      "step": 8460
+    },
+    {
+      "epoch": 2.7104,
+      "grad_norm": 6.295539855957031,
+      "learning_rate": 1.0161688749157872e-05,
+      "loss": 2.1108,
+      "step": 8470
+    },
+    {
+      "epoch": 2.7136,
+      "grad_norm": 5.153507709503174,
+      "learning_rate": 1.0049404895576018e-05,
+      "loss": 2.0563,
+      "step": 8480
+    },
+    {
+      "epoch": 2.7168,
+      "grad_norm": 4.624126434326172,
+      "learning_rate": 9.937121041994161e-06,
+      "loss": 2.1859,
+      "step": 8490
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 4.7635884284973145,
+      "learning_rate": 9.824837188412307e-06,
+      "loss": 2.1191,
+      "step": 8500
+    },
+    {
+      "epoch": 2.7232,
+      "grad_norm": 4.604827404022217,
+      "learning_rate": 9.712553334830451e-06,
+      "loss": 2.0735,
+      "step": 8510
+    },
+    {
+      "epoch": 2.7264,
+      "grad_norm": 6.650067329406738,
+      "learning_rate": 9.600269481248596e-06,
+      "loss": 2.1324,
+      "step": 8520
+    },
+    {
+      "epoch": 2.7296,
+      "grad_norm": 4.415541648864746,
+      "learning_rate": 9.487985627666742e-06,
+      "loss": 1.9815,
+      "step": 8530
+    },
+    {
+      "epoch": 2.7328,
+      "grad_norm": 4.163292407989502,
+      "learning_rate": 9.375701774084888e-06,
+      "loss": 1.9876,
+      "step": 8540
+    },
+    {
+      "epoch": 2.7359999999999998,
+      "grad_norm": 2.965481996536255,
+      "learning_rate": 9.263417920503033e-06,
+      "loss": 2.001,
+      "step": 8550
+    },
+    {
+      "epoch": 2.7392,
+      "grad_norm": 3.5702438354492188,
+      "learning_rate": 9.151134066921177e-06,
+      "loss": 1.9329,
+      "step": 8560
+    },
+    {
+      "epoch": 2.7424,
+      "grad_norm": 6.828572750091553,
+      "learning_rate": 9.038850213339323e-06,
+      "loss": 2.0726,
+      "step": 8570
+    },
+    {
+      "epoch": 2.7456,
+      "grad_norm": 5.8427300453186035,
+      "learning_rate": 8.926566359757467e-06,
+      "loss": 1.9933,
+      "step": 8580
+    },
+    {
+      "epoch": 2.7488,
+      "grad_norm": 7.576897621154785,
+      "learning_rate": 8.814282506175612e-06,
+      "loss": 2.2388,
+      "step": 8590
+    },
+    {
+      "epoch": 2.752,
+      "grad_norm": 3.542816162109375,
+      "learning_rate": 8.701998652593756e-06,
+      "loss": 1.9418,
+      "step": 8600
+    },
+    {
+      "epoch": 2.7552,
+      "grad_norm": 8.087502479553223,
+      "learning_rate": 8.589714799011902e-06,
+      "loss": 2.1486,
+      "step": 8610
+    },
+    {
+      "epoch": 2.7584,
+      "grad_norm": 4.78680419921875,
+      "learning_rate": 8.477430945430049e-06,
+      "loss": 2.06,
+      "step": 8620
+    },
+    {
+      "epoch": 2.7616,
+      "grad_norm": 3.113011598587036,
+      "learning_rate": 8.365147091848193e-06,
+      "loss": 2.1584,
+      "step": 8630
+    },
+    {
+      "epoch": 2.7648,
+      "grad_norm": 3.982513666152954,
+      "learning_rate": 8.252863238266339e-06,
+      "loss": 2.014,
+      "step": 8640
+    },
+    {
+      "epoch": 2.768,
+      "grad_norm": 5.133899211883545,
+      "learning_rate": 8.140579384684482e-06,
+      "loss": 1.811,
+      "step": 8650
+    },
+    {
+      "epoch": 2.7712,
+      "grad_norm": 6.8349609375,
+      "learning_rate": 8.028295531102628e-06,
+      "loss": 2.0902,
+      "step": 8660
+    },
+    {
+      "epoch": 2.7744,
+      "grad_norm": 6.302601337432861,
+      "learning_rate": 7.916011677520772e-06,
+      "loss": 2.1602,
+      "step": 8670
+    },
+    {
+      "epoch": 2.7776,
+      "grad_norm": 3.433237075805664,
+      "learning_rate": 7.803727823938918e-06,
+      "loss": 2.1091,
+      "step": 8680
+    },
+    {
+      "epoch": 2.7808,
+      "grad_norm": 6.184892654418945,
+      "learning_rate": 7.691443970357063e-06,
+      "loss": 2.098,
+      "step": 8690
+    },
+    {
+      "epoch": 2.784,
+      "grad_norm": 5.453632354736328,
+      "learning_rate": 7.579160116775208e-06,
+      "loss": 2.1538,
+      "step": 8700
+    },
+    {
+      "epoch": 2.7872,
+      "grad_norm": 7.197730541229248,
+      "learning_rate": 7.4668762631933535e-06,
+      "loss": 1.9261,
+      "step": 8710
+    },
+    {
+      "epoch": 2.7904,
+      "grad_norm": 8.1513032913208,
+      "learning_rate": 7.354592409611498e-06,
+      "loss": 1.9436,
+      "step": 8720
+    },
+    {
+      "epoch": 2.7936,
+      "grad_norm": 5.156404495239258,
+      "learning_rate": 7.242308556029644e-06,
+      "loss": 2.0882,
+      "step": 8730
+    },
+    {
+      "epoch": 2.7968,
+      "grad_norm": 6.841371059417725,
+      "learning_rate": 7.130024702447788e-06,
+      "loss": 1.9373,
+      "step": 8740
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 5.5138373374938965,
+      "learning_rate": 7.017740848865933e-06,
+      "loss": 2.07,
+      "step": 8750
+    },
+    {
+      "epoch": 2.8032,
+      "grad_norm": 4.273646831512451,
+      "learning_rate": 6.905456995284078e-06,
+      "loss": 2.2179,
+      "step": 8760
+    },
+    {
+      "epoch": 2.8064,
+      "grad_norm": 6.17716121673584,
+      "learning_rate": 6.793173141702224e-06,
+      "loss": 1.9038,
+      "step": 8770
+    },
+    {
+      "epoch": 2.8096,
+      "grad_norm": 5.650930881500244,
+      "learning_rate": 6.680889288120369e-06,
+      "loss": 2.0042,
+      "step": 8780
+    },
+    {
+      "epoch": 2.8128,
+      "grad_norm": 7.444324970245361,
+      "learning_rate": 6.568605434538513e-06,
+      "loss": 2.018,
+      "step": 8790
+    },
+    {
+      "epoch": 2.816,
+      "grad_norm": 5.6876068115234375,
+      "learning_rate": 6.456321580956659e-06,
+      "loss": 2.2024,
+      "step": 8800
+    },
+    {
+      "epoch": 2.8192,
+      "grad_norm": 2.757317543029785,
+      "learning_rate": 6.3440377273748035e-06,
+      "loss": 2.1015,
+      "step": 8810
+    },
+    {
+      "epoch": 2.8224,
+      "grad_norm": 5.419109344482422,
+      "learning_rate": 6.231753873792949e-06,
+      "loss": 2.0039,
+      "step": 8820
+    },
+    {
+      "epoch": 2.8256,
+      "grad_norm": 4.2704877853393555,
+      "learning_rate": 6.119470020211094e-06,
+      "loss": 2.0869,
+      "step": 8830
+    },
+    {
+      "epoch": 2.8288,
+      "grad_norm": 4.7045392990112305,
+      "learning_rate": 6.007186166629239e-06,
+      "loss": 2.0181,
+      "step": 8840
+    },
+    {
+      "epoch": 2.832,
+      "grad_norm": 5.270239353179932,
+      "learning_rate": 5.894902313047384e-06,
+      "loss": 2.0192,
+      "step": 8850
+    },
+    {
+      "epoch": 2.8352,
+      "grad_norm": 4.050564765930176,
+      "learning_rate": 5.782618459465529e-06,
+      "loss": 2.0618,
+      "step": 8860
+    },
+    {
+      "epoch": 2.8384,
+      "grad_norm": 5.236793041229248,
+      "learning_rate": 5.6703346058836745e-06,
+      "loss": 2.0668,
+      "step": 8870
+    },
+    {
+      "epoch": 2.8416,
+      "grad_norm": 6.486008644104004,
+      "learning_rate": 5.558050752301819e-06,
+      "loss": 2.0369,
+      "step": 8880
+    },
+    {
+      "epoch": 2.8448,
+      "grad_norm": 4.113170146942139,
+      "learning_rate": 5.445766898719965e-06,
+      "loss": 1.9495,
+      "step": 8890
+    },
+    {
+      "epoch": 2.848,
+      "grad_norm": 6.932297229766846,
+      "learning_rate": 5.33348304513811e-06,
+      "loss": 1.9701,
+      "step": 8900
+    },
+    {
+      "epoch": 2.8512,
+      "grad_norm": 5.153751373291016,
+      "learning_rate": 5.221199191556254e-06,
+      "loss": 2.1513,
+      "step": 8910
+    },
+    {
+      "epoch": 2.8544,
+      "grad_norm": 3.6570873260498047,
+      "learning_rate": 5.108915337974399e-06,
+      "loss": 2.1062,
+      "step": 8920
+    },
+    {
+      "epoch": 2.8576,
+      "grad_norm": 6.329987525939941,
+      "learning_rate": 4.996631484392545e-06,
+      "loss": 2.0764,
+      "step": 8930
+    },
+    {
+      "epoch": 2.8608000000000002,
+      "grad_norm": 5.174140930175781,
+      "learning_rate": 4.8843476308106895e-06,
+      "loss": 2.0468,
+      "step": 8940
+    },
+    {
+      "epoch": 2.864,
+      "grad_norm": 3.7998836040496826,
+      "learning_rate": 4.772063777228835e-06,
+      "loss": 1.9971,
+      "step": 8950
+    },
+    {
+      "epoch": 2.8672,
+      "grad_norm": 8.548787117004395,
+      "learning_rate": 4.65977992364698e-06,
+      "loss": 2.1847,
+      "step": 8960
+    },
+    {
+      "epoch": 2.8704,
+      "grad_norm": 3.677898645401001,
+      "learning_rate": 4.5474960700651246e-06,
+      "loss": 2.0329,
+      "step": 8970
+    },
+    {
+      "epoch": 2.8736,
+      "grad_norm": 6.614884853363037,
+      "learning_rate": 4.43521221648327e-06,
+      "loss": 2.1371,
+      "step": 8980
+    },
+    {
+      "epoch": 2.8768000000000002,
+      "grad_norm": 6.62887716293335,
+      "learning_rate": 4.322928362901415e-06,
+      "loss": 1.9745,
+      "step": 8990
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 5.070212364196777,
+      "learning_rate": 4.21064450931956e-06,
+      "loss": 2.27,
+      "step": 9000
+    },
+    {
+      "epoch": 2.8832,
+      "grad_norm": 6.215460300445557,
+      "learning_rate": 4.098360655737704e-06,
+      "loss": 2.0103,
+      "step": 9010
+    },
+    {
+      "epoch": 2.8864,
+      "grad_norm": 5.103015899658203,
+      "learning_rate": 3.986076802155851e-06,
+      "loss": 2.1006,
+      "step": 9020
+    },
+    {
+      "epoch": 2.8895999999999997,
+      "grad_norm": 4.409531593322754,
+      "learning_rate": 3.873792948573996e-06,
+      "loss": 1.9379,
+      "step": 9030
+    },
+    {
+      "epoch": 2.8928000000000003,
+      "grad_norm": 6.719906806945801,
+      "learning_rate": 3.7615090949921403e-06,
+      "loss": 2.1983,
+      "step": 9040
+    },
+    {
+      "epoch": 2.896,
+      "grad_norm": 6.215670108795166,
+      "learning_rate": 3.6492252414102855e-06,
+      "loss": 2.1848,
+      "step": 9050
+    },
+    {
+      "epoch": 2.8992,
+      "grad_norm": 6.721705436706543,
+      "learning_rate": 3.5369413878284303e-06,
+      "loss": 2.2883,
+      "step": 9060
+    },
+    {
+      "epoch": 2.9024,
+      "grad_norm": 3.897223472595215,
+      "learning_rate": 3.4246575342465754e-06,
+      "loss": 2.0083,
+      "step": 9070
+    },
+    {
+      "epoch": 2.9055999999999997,
+      "grad_norm": 5.6442461013793945,
+      "learning_rate": 3.31237368066472e-06,
+      "loss": 2.0583,
+      "step": 9080
+    },
+    {
+      "epoch": 2.9088000000000003,
+      "grad_norm": 6.394291877746582,
+      "learning_rate": 3.2000898270828658e-06,
+      "loss": 2.1312,
+      "step": 9090
+    },
+    {
+      "epoch": 2.912,
+      "grad_norm": 2.766772985458374,
+      "learning_rate": 3.0878059735010105e-06,
+      "loss": 1.9678,
+      "step": 9100
+    },
+    {
+      "epoch": 2.9152,
+      "grad_norm": 3.434525966644287,
+      "learning_rate": 2.9755221199191557e-06,
+      "loss": 2.1135,
+      "step": 9110
+    },
+    {
+      "epoch": 2.9184,
+      "grad_norm": 4.80876350402832,
+      "learning_rate": 2.863238266337301e-06,
+      "loss": 2.127,
+      "step": 9120
+    },
+    {
+      "epoch": 2.9215999999999998,
+      "grad_norm": 5.191465377807617,
+      "learning_rate": 2.7509544127554456e-06,
+      "loss": 1.9927,
+      "step": 9130
+    },
+    {
+      "epoch": 2.9248,
+      "grad_norm": 6.8004631996154785,
+      "learning_rate": 2.638670559173591e-06,
+      "loss": 2.0743,
+      "step": 9140
+    },
+    {
+      "epoch": 2.928,
+      "grad_norm": 3.939542531967163,
+      "learning_rate": 2.526386705591736e-06,
+      "loss": 2.0535,
+      "step": 9150
+    },
+    {
+      "epoch": 2.9312,
+      "grad_norm": 3.9124600887298584,
+      "learning_rate": 2.414102852009881e-06,
+      "loss": 2.2046,
+      "step": 9160
+    },
+    {
+      "epoch": 2.9344,
+      "grad_norm": 6.297423362731934,
+      "learning_rate": 2.301818998428026e-06,
+      "loss": 2.0143,
+      "step": 9170
+    },
+    {
+      "epoch": 2.9375999999999998,
+      "grad_norm": 5.202453136444092,
+      "learning_rate": 2.1895351448461715e-06,
+      "loss": 2.0839,
+      "step": 9180
+    },
+    {
+      "epoch": 2.9408,
+      "grad_norm": 6.202855587005615,
+      "learning_rate": 2.0772512912643162e-06,
+      "loss": 1.985,
+      "step": 9190
+    },
+    {
+      "epoch": 2.944,
+      "grad_norm": 4.195956230163574,
+      "learning_rate": 1.9649674376824614e-06,
+      "loss": 2.1004,
+      "step": 9200
+    },
+    {
+      "epoch": 2.9472,
+      "grad_norm": 4.236852645874023,
+      "learning_rate": 1.8526835841006064e-06,
+      "loss": 1.9329,
+      "step": 9210
+    },
+    {
+      "epoch": 2.9504,
+      "grad_norm": 5.220149040222168,
+      "learning_rate": 1.7403997305187515e-06,
+      "loss": 2.2291,
+      "step": 9220
+    },
+    {
+      "epoch": 2.9536,
+      "grad_norm": 9.999998092651367,
+      "learning_rate": 1.6281158769368965e-06,
+      "loss": 2.057,
+      "step": 9230
+    },
+    {
+      "epoch": 2.9568,
+      "grad_norm": 6.199564456939697,
+      "learning_rate": 1.5158320233550417e-06,
+      "loss": 1.9977,
+      "step": 9240
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 5.6598734855651855,
+      "learning_rate": 1.4035481697731866e-06,
+      "loss": 2.0441,
+      "step": 9250
+    },
+    {
+      "epoch": 2.9632,
+      "grad_norm": 3.8681833744049072,
+      "learning_rate": 1.2912643161913318e-06,
+      "loss": 2.1307,
+      "step": 9260
+    },
+    {
+      "epoch": 2.9664,
+      "grad_norm": 4.755743503570557,
+      "learning_rate": 1.1789804626094768e-06,
+      "loss": 2.0349,
+      "step": 9270
+    },
+    {
+      "epoch": 2.9696,
+      "grad_norm": 5.985792636871338,
+      "learning_rate": 1.066696609027622e-06,
+      "loss": 2.0914,
+      "step": 9280
+    },
+    {
+      "epoch": 2.9728,
+      "grad_norm": 4.568302154541016,
+      "learning_rate": 9.544127554457669e-07,
+      "loss": 1.8999,
+      "step": 9290
+    },
+    {
+      "epoch": 2.976,
+      "grad_norm": 3.314087390899658,
+      "learning_rate": 8.421289018639121e-07,
+      "loss": 2.0736,
+      "step": 9300
+    },
+    {
+      "epoch": 2.9792,
+      "grad_norm": 4.549111366271973,
+      "learning_rate": 7.298450482820571e-07,
+      "loss": 1.9899,
+      "step": 9310
+    },
+    {
+      "epoch": 2.9824,
+      "grad_norm": 3.963134288787842,
+      "learning_rate": 6.175611947002022e-07,
+      "loss": 2.0542,
+      "step": 9320
+    },
+    {
+      "epoch": 2.9856,
+      "grad_norm": 4.205166816711426,
+      "learning_rate": 5.052773411183473e-07,
+      "loss": 2.1633,
+      "step": 9330
+    },
+    {
+      "epoch": 2.9888,
+      "grad_norm": 4.5056657791137695,
+      "learning_rate": 3.9299348753649227e-07,
+      "loss": 2.135,
+      "step": 9340
+    },
+    {
+      "epoch": 2.992,
+      "grad_norm": 4.240407466888428,
+      "learning_rate": 2.8070963395463734e-07,
+      "loss": 2.0926,
+      "step": 9350
+    },
+    {
+      "epoch": 2.9952,
+      "grad_norm": 3.3892221450805664,
+      "learning_rate": 1.684257803727824e-07,
+      "loss": 1.8936,
+      "step": 9360
+    },
+    {
+      "epoch": 2.9984,
+      "grad_norm": 5.6661481857299805,
+      "learning_rate": 5.6141926790927474e-08,
+      "loss": 1.9597,
+      "step": 9370
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9375,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}