diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15538 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 50,
+  "global_step": 18031,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.37098968029022217,
+      "learning_rate": 2.2172949002217297e-08,
+      "loss": 0.9723,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.38433295488357544,
+      "learning_rate": 2.2172949002217298e-07,
+      "loss": 1.0622,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.45640310645103455,
+      "learning_rate": 4.4345898004434597e-07,
+      "loss": 1.0429,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.4661204218864441,
+      "learning_rate": 6.651884700665189e-07,
+      "loss": 1.0408,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.4835126996040344,
+      "learning_rate": 8.869179600886919e-07,
+      "loss": 1.0878,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.4311816394329071,
+      "learning_rate": 1.1086474501108648e-06,
+      "loss": 1.0378,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.0920552015304565,
+      "eval_runtime": 0.606,
+      "eval_samples_per_second": 16.502,
+      "eval_steps_per_second": 3.3,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.4773673415184021,
+      "learning_rate": 1.3303769401330377e-06,
+      "loss": 1.0761,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.44414222240448,
+      "learning_rate": 1.5521064301552107e-06,
+      "loss": 1.0436,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5979101061820984,
+      "learning_rate": 1.7738359201773839e-06,
+      "loss": 1.0419,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.5221405625343323,
+      "learning_rate": 1.995565410199557e-06,
+      "loss": 1.0558,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.42636510729789734,
+      "learning_rate": 2.2172949002217296e-06,
+      "loss": 1.0321,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 1.0590953826904297,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.414,
+      "eval_steps_per_second": 3.283,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4288226366043091,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 0.989,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.45129188895225525,
+      "learning_rate": 2.6607538802660755e-06,
+      "loss": 0.9782,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.48875483870506287,
+      "learning_rate": 2.8824833702882487e-06,
+      "loss": 1.0032,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.3884626030921936,
+      "learning_rate": 3.1042128603104214e-06,
+      "loss": 0.9348,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.3806435465812683,
+      "learning_rate": 3.3259423503325946e-06,
+      "loss": 0.9863,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 0.998333752155304,
+      "eval_runtime": 0.6128,
+      "eval_samples_per_second": 16.32,
+      "eval_steps_per_second": 3.264,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.3922533690929413,
+      "learning_rate": 3.5476718403547677e-06,
+      "loss": 0.9788,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.41704419255256653,
+      "learning_rate": 3.7694013303769405e-06,
+      "loss": 0.9197,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.41639530658721924,
+      "learning_rate": 3.991130820399114e-06,
+      "loss": 0.9335,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.37787139415740967,
+      "learning_rate": 4.212860310421286e-06,
+      "loss": 0.9618,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4257109761238098,
+      "learning_rate": 4.434589800443459e-06,
+      "loss": 0.9689,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 0.9614308476448059,
+      "eval_runtime": 0.6193,
+      "eval_samples_per_second": 16.148,
+      "eval_steps_per_second": 3.23,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4102950692176819,
+      "learning_rate": 4.656319290465633e-06,
+      "loss": 0.9467,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.41053393483161926,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.9058,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4410107135772705,
+      "learning_rate": 5.099778270509978e-06,
+      "loss": 0.8812,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.42213955521583557,
+      "learning_rate": 5.321507760532151e-06,
+      "loss": 0.9381,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.4775751829147339,
+      "learning_rate": 5.5432372505543246e-06,
+      "loss": 0.926,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 0.9444061517715454,
+      "eval_runtime": 0.6147,
+      "eval_samples_per_second": 16.269,
+      "eval_steps_per_second": 3.254,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.5134453773498535,
+      "learning_rate": 5.764966740576497e-06,
+      "loss": 0.9114,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.45892617106437683,
+      "learning_rate": 5.98669623059867e-06,
+      "loss": 0.8781,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.49861735105514526,
+      "learning_rate": 6.208425720620843e-06,
+      "loss": 0.9096,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5129882097244263,
+      "learning_rate": 6.430155210643016e-06,
+      "loss": 0.9248,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5301437973976135,
+      "learning_rate": 6.651884700665189e-06,
+      "loss": 0.908,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 0.9324199557304382,
+      "eval_runtime": 0.615,
+      "eval_samples_per_second": 16.261,
+      "eval_steps_per_second": 3.252,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5101982355117798,
+      "learning_rate": 6.873614190687362e-06,
+      "loss": 0.878,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5797512531280518,
+      "learning_rate": 7.0953436807095355e-06,
+      "loss": 0.875,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5401256084442139,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.865,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5543472170829773,
+      "learning_rate": 7.538802660753881e-06,
+      "loss": 0.8994,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5431869029998779,
+      "learning_rate": 7.760532150776053e-06,
+      "loss": 0.8611,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 0.923511803150177,
+      "eval_runtime": 0.6162,
+      "eval_samples_per_second": 16.229,
+      "eval_steps_per_second": 3.246,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5307214856147766,
+      "learning_rate": 7.982261640798227e-06,
+      "loss": 0.8456,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6495651006698608,
+      "learning_rate": 8.2039911308204e-06,
+      "loss": 0.886,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.611789882183075,
+      "learning_rate": 8.425720620842573e-06,
+      "loss": 0.9192,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6726333498954773,
+      "learning_rate": 8.647450110864746e-06,
+      "loss": 0.8623,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.615410327911377,
+      "learning_rate": 8.869179600886918e-06,
+      "loss": 0.8677,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 0.9155791401863098,
+      "eval_runtime": 0.6162,
+      "eval_samples_per_second": 16.229,
+      "eval_steps_per_second": 3.246,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6737934947013855,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.921,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6643930673599243,
+      "learning_rate": 9.312638580931265e-06,
+      "loss": 0.854,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6588521003723145,
+      "learning_rate": 9.534368070953438e-06,
+      "loss": 0.8759,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7499340772628784,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.8897,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8111746907234192,
+      "learning_rate": 9.977827050997784e-06,
+      "loss": 0.8908,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 0.9106427431106567,
+      "eval_runtime": 0.6174,
+      "eval_samples_per_second": 16.198,
+      "eval_steps_per_second": 3.24,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7945922017097473,
+      "learning_rate": 1.0199556541019956e-05,
+      "loss": 0.8462,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7537086009979248,
+      "learning_rate": 1.0421286031042131e-05,
+      "loss": 0.8737,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.6655195951461792,
+      "learning_rate": 1.0643015521064302e-05,
+      "loss": 0.8546,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7690728306770325,
+      "learning_rate": 1.0864745011086475e-05,
+      "loss": 0.8589,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8440932631492615,
+      "learning_rate": 1.1086474501108649e-05,
+      "loss": 0.8805,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 0.9038592576980591,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.34,
+      "eval_steps_per_second": 3.268,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8415856957435608,
+      "learning_rate": 1.1308203991130822e-05,
+      "loss": 0.8713,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8858282566070557,
+      "learning_rate": 1.1529933481152995e-05,
+      "loss": 0.8467,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8213603496551514,
+      "learning_rate": 1.1751662971175166e-05,
+      "loss": 0.8514,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8467979431152344,
+      "learning_rate": 1.197339246119734e-05,
+      "loss": 0.8515,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.765421986579895,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.8599,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 0.8986900448799133,
+      "eval_runtime": 0.6179,
+      "eval_samples_per_second": 16.185,
+      "eval_steps_per_second": 3.237,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7823172807693481,
+      "learning_rate": 1.2416851441241686e-05,
+      "loss": 0.8704,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7606292963027954,
+      "learning_rate": 1.263858093126386e-05,
+      "loss": 0.8583,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9138412475585938,
+      "learning_rate": 1.2860310421286033e-05,
+      "loss": 0.8513,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7301716804504395,
+      "learning_rate": 1.3082039911308204e-05,
+      "loss": 0.9023,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8641847968101501,
+      "learning_rate": 1.3303769401330378e-05,
+      "loss": 0.8486,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 0.8891505002975464,
+      "eval_runtime": 0.615,
+      "eval_samples_per_second": 16.26,
+      "eval_steps_per_second": 3.252,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.881588876247406,
+      "learning_rate": 1.3525498891352551e-05,
+      "loss": 0.846,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8803728818893433,
+      "learning_rate": 1.3747228381374724e-05,
+      "loss": 0.8178,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9609507322311401,
+      "learning_rate": 1.3968957871396898e-05,
+      "loss": 0.8463,
+      "step": 630
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9375690817832947,
+      "learning_rate": 1.4190687361419071e-05,
+      "loss": 0.882,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8671866655349731,
+      "learning_rate": 1.4412416851441242e-05,
+      "loss": 0.8134,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.883123517036438,
+      "eval_runtime": 0.6118,
+      "eval_samples_per_second": 16.346,
+      "eval_steps_per_second": 3.269,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8724412322044373,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.8658,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0153032541275024,
+      "learning_rate": 1.485587583148559e-05,
+      "loss": 0.8642,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9495956301689148,
+      "learning_rate": 1.5077605321507762e-05,
+      "loss": 0.8475,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9556280970573425,
+      "learning_rate": 1.5299334811529935e-05,
+      "loss": 0.8267,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0413883924484253,
+      "learning_rate": 1.5521064301552106e-05,
+      "loss": 0.8231,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.8770906329154968,
+      "eval_runtime": 0.6154,
+      "eval_samples_per_second": 16.248,
+      "eval_steps_per_second": 3.25,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0282937288284302,
+      "learning_rate": 1.574279379157428e-05,
+      "loss": 0.8305,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0459904670715332,
+      "learning_rate": 1.5964523281596455e-05,
+      "loss": 0.8456,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0485682487487793,
+      "learning_rate": 1.6186252771618626e-05,
+      "loss": 0.8188,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0812604427337646,
+      "learning_rate": 1.64079822616408e-05,
+      "loss": 0.8585,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9856111407279968,
+      "learning_rate": 1.6629711751662975e-05,
+      "loss": 0.7947,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.8732670545578003,
+      "eval_runtime": 0.614,
+      "eval_samples_per_second": 16.287,
+      "eval_steps_per_second": 3.257,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0509120225906372,
+      "learning_rate": 1.6851441241685146e-05,
+      "loss": 0.8627,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0772370100021362,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 0.8172,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0859620571136475,
+      "learning_rate": 1.729490022172949e-05,
+      "loss": 0.8218,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0991852283477783,
+      "learning_rate": 1.7516629711751666e-05,
+      "loss": 0.831,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0878872871398926,
+      "learning_rate": 1.7738359201773837e-05,
+      "loss": 0.8282,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.867023766040802,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.391,
+      "eval_steps_per_second": 3.278,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.1412290334701538,
+      "learning_rate": 1.796008869179601e-05,
+      "loss": 0.8506,
+      "step": 810
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1338917016983032,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.8039,
+      "step": 820
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1018694639205933,
+      "learning_rate": 1.8403547671840357e-05,
+      "loss": 0.8222,
+      "step": 830
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0673363208770752,
+      "learning_rate": 1.862527716186253e-05,
+      "loss": 0.8226,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.049584150314331,
+      "learning_rate": 1.8847006651884702e-05,
+      "loss": 0.838,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.8649986386299133,
+      "eval_runtime": 0.6123,
+      "eval_samples_per_second": 16.332,
+      "eval_steps_per_second": 3.266,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1148349046707153,
+      "learning_rate": 1.9068736141906876e-05,
+      "loss": 0.8082,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2069787979125977,
+      "learning_rate": 1.9290465631929047e-05,
+      "loss": 0.834,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0948930978775024,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.8263,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0878373384475708,
+      "learning_rate": 1.9733924611973393e-05,
+      "loss": 0.8134,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1251206398010254,
+      "learning_rate": 1.9955654101995567e-05,
+      "loss": 0.7834,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.8653883934020996,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.365,
+      "eval_steps_per_second": 3.273,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1090091466903687,
+      "learning_rate": 1.9990659116118865e-05,
+      "loss": 0.8003,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1569527387619019,
+      "learning_rate": 1.9978983011267442e-05,
+      "loss": 0.8635,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1255342960357666,
+      "learning_rate": 1.9967306906416022e-05,
+      "loss": 0.8376,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1842678785324097,
+      "learning_rate": 1.99556308015646e-05,
+      "loss": 0.8854,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.161680817604065,
+      "learning_rate": 1.994395469671318e-05,
+      "loss": 0.8096,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.8573818206787109,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.36,
+      "eval_steps_per_second": 3.272,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.172326683998108,
+      "learning_rate": 1.9932278591861757e-05,
+      "loss": 0.8035,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.2169458866119385,
+      "learning_rate": 1.9920602487010334e-05,
+      "loss": 0.8412,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0622872114181519,
+      "learning_rate": 1.9908926382158914e-05,
+      "loss": 0.8144,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.1667189598083496,
+      "learning_rate": 1.989725027730749e-05,
+      "loss": 0.8307,
+      "step": 990
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1407010555267334,
+      "learning_rate": 1.988557417245607e-05,
+      "loss": 0.8782,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.8587248921394348,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.364,
+      "eval_steps_per_second": 3.273,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1451735496520996,
+      "learning_rate": 1.9873898067604648e-05,
+      "loss": 0.8011,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0524930953979492,
+      "learning_rate": 1.9862221962753228e-05,
+      "loss": 0.8442,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1778239011764526,
+      "learning_rate": 1.9850545857901805e-05,
+      "loss": 0.8708,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1116106510162354,
+      "learning_rate": 1.9838869753050386e-05,
+      "loss": 0.779,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.154364824295044,
+      "learning_rate": 1.9827193648198962e-05,
+      "loss": 0.8302,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.8541529774665833,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.371,
+      "eval_steps_per_second": 3.274,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1855803728103638,
+      "learning_rate": 1.981551754334754e-05,
+      "loss": 0.8179,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2805979251861572,
+      "learning_rate": 1.980384143849612e-05,
+      "loss": 0.7959,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.257548213005066,
+      "learning_rate": 1.9792165333644697e-05,
+      "loss": 0.8374,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1202774047851562,
+      "learning_rate": 1.9780489228793277e-05,
+      "loss": 0.8024,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2020343542099,
+      "learning_rate": 1.9768813123941854e-05,
+      "loss": 0.7952,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.8500730395317078,
+      "eval_runtime": 0.6177,
+      "eval_samples_per_second": 16.19,
+      "eval_steps_per_second": 3.238,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2609871625900269,
+      "learning_rate": 1.9757137019090434e-05,
+      "loss": 0.8019,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1995282173156738,
+      "learning_rate": 1.974546091423901e-05,
+      "loss": 0.8205,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2330576181411743,
+      "learning_rate": 1.973378480938759e-05,
+      "loss": 0.8439,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2292983531951904,
+      "learning_rate": 1.9722108704536168e-05,
+      "loss": 0.7909,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2863200902938843,
+      "learning_rate": 1.971043259968475e-05,
+      "loss": 0.799,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.8546978235244751,
+      "eval_runtime": 0.6153,
+      "eval_samples_per_second": 16.252,
+      "eval_steps_per_second": 3.25,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.3344646692276,
+      "learning_rate": 1.9698756494833326e-05,
+      "loss": 0.8124,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.2979106903076172,
+      "learning_rate": 1.9687080389981902e-05,
+      "loss": 0.8119,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2481484413146973,
+      "learning_rate": 1.9675404285130483e-05,
+      "loss": 0.787,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.215266466140747,
+      "learning_rate": 1.966372818027906e-05,
+      "loss": 0.8124,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3916720151901245,
+      "learning_rate": 1.965205207542764e-05,
+      "loss": 0.7919,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.8512755632400513,
+      "eval_runtime": 0.617,
+      "eval_samples_per_second": 16.207,
+      "eval_steps_per_second": 3.241,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3035218715667725,
+      "learning_rate": 1.9640375970576217e-05,
+      "loss": 0.8001,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3856993913650513,
+      "learning_rate": 1.9628699865724797e-05,
+      "loss": 0.8562,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2390185594558716,
+      "learning_rate": 1.9617023760873374e-05,
+      "loss": 0.8204,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.310853362083435,
+      "learning_rate": 1.9605347656021954e-05,
+      "loss": 0.7849,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4000093936920166,
+      "learning_rate": 1.959367155117053e-05,
+      "loss": 0.7954,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.8510808944702148,
+      "eval_runtime": 0.6104,
+      "eval_samples_per_second": 16.382,
+      "eval_steps_per_second": 3.276,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3155741691589355,
+      "learning_rate": 1.958199544631911e-05,
+      "loss": 0.8005,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3227680921554565,
+      "learning_rate": 1.957031934146769e-05,
+      "loss": 0.8173,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3040021657943726,
+      "learning_rate": 1.9558643236616266e-05,
+      "loss": 0.7948,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.34744131565094,
+      "learning_rate": 1.9546967131764846e-05,
+      "loss": 0.7901,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4661809206008911,
+      "learning_rate": 1.9535291026913423e-05,
+      "loss": 0.8089,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.8488125801086426,
+      "eval_runtime": 0.6151,
+      "eval_samples_per_second": 16.258,
+      "eval_steps_per_second": 3.252,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2757831811904907,
+      "learning_rate": 1.9523614922062003e-05,
+      "loss": 0.8561,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3130873441696167,
+      "learning_rate": 1.951193881721058e-05,
+      "loss": 0.8294,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2937477827072144,
+      "learning_rate": 1.950026271235916e-05,
+      "loss": 0.8065,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.2813125848770142,
+      "learning_rate": 1.9488586607507737e-05,
+      "loss": 0.8073,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3467236757278442,
+      "learning_rate": 1.9476910502656318e-05,
+      "loss": 0.8319,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.8497697710990906,
+      "eval_runtime": 0.6158,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 3.248,
+      "step": 1350
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3060228824615479,
+      "learning_rate": 1.9465234397804894e-05,
+      "loss": 0.8307,
+      "step": 1360
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2668310403823853,
+      "learning_rate": 1.945355829295347e-05,
+      "loss": 0.7585,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.358108401298523,
+      "learning_rate": 1.9441882188102052e-05,
+      "loss": 0.8276,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3149278163909912,
+      "learning_rate": 1.943020608325063e-05,
+      "loss": 0.7926,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2571593523025513,
+      "learning_rate": 1.941852997839921e-05,
+      "loss": 0.7852,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 0.848658561706543,
+      "eval_runtime": 0.6112,
+      "eval_samples_per_second": 16.36,
+      "eval_steps_per_second": 3.272,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.421294927597046,
+      "learning_rate": 1.9406853873547786e-05,
+      "loss": 0.7992,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.373687982559204,
+      "learning_rate": 1.9395177768696366e-05,
+      "loss": 0.8362,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4148705005645752,
+      "learning_rate": 1.9383501663844943e-05,
+      "loss": 0.7837,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.540602445602417,
+      "learning_rate": 1.9371825558993523e-05,
+      "loss": 0.7805,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4312440156936646,
+      "learning_rate": 1.93601494541421e-05,
+      "loss": 0.7951,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 0.8483745455741882,
+      "eval_runtime": 0.6215,
+      "eval_samples_per_second": 16.09,
+      "eval_steps_per_second": 3.218,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4073846340179443,
+      "learning_rate": 1.9348473349290677e-05,
+      "loss": 0.809,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3141839504241943,
+      "learning_rate": 1.9336797244439258e-05,
+      "loss": 0.8142,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4639538526535034,
+      "learning_rate": 1.9325121139587834e-05,
+      "loss": 0.7867,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3674744367599487,
+      "learning_rate": 1.9313445034736415e-05,
+      "loss": 0.8085,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.5070226192474365,
+      "learning_rate": 1.9301768929884992e-05,
+      "loss": 0.7653,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 0.8452969789505005,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.391,
+      "eval_steps_per_second": 3.278,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.346683144569397,
+      "learning_rate": 1.9290092825033572e-05,
+      "loss": 0.79,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3689582347869873,
+      "learning_rate": 1.927841672018215e-05,
+      "loss": 0.7857,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.4211719036102295,
+      "learning_rate": 1.926674061533073e-05,
+      "loss": 0.774,
+      "step": 1530
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.365173578262329,
+      "learning_rate": 1.9255064510479306e-05,
+      "loss": 0.7955,
+      "step": 1540
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5663175582885742,
+      "learning_rate": 1.9243388405627883e-05,
+      "loss": 0.8075,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.8408949971199036,
+      "eval_runtime": 0.6128,
+      "eval_samples_per_second": 16.319,
+      "eval_steps_per_second": 3.264,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4302055835723877,
+      "learning_rate": 1.923171230077646e-05,
+      "loss": 0.8078,
+      "step": 1560
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3450196981430054,
+      "learning_rate": 1.922003619592504e-05,
+      "loss": 0.7556,
+      "step": 1570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6568788290023804,
+      "learning_rate": 1.920836009107362e-05,
+      "loss": 0.7635,
+      "step": 1580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.479311466217041,
+      "learning_rate": 1.9196683986222198e-05,
+      "loss": 0.7985,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.60139000415802,
+      "learning_rate": 1.9185007881370778e-05,
+      "loss": 0.7974,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.8401718139648438,
+      "eval_runtime": 0.6136,
+      "eval_samples_per_second": 16.298,
+      "eval_steps_per_second": 3.26,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4208217859268188,
+      "learning_rate": 1.9173331776519355e-05,
+      "loss": 0.7742,
+      "step": 1610
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4855490922927856,
+      "learning_rate": 1.9161655671667935e-05,
+      "loss": 0.8059,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4890614748001099,
+      "learning_rate": 1.9149979566816512e-05,
+      "loss": 0.8049,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3954768180847168,
+      "learning_rate": 1.9138303461965092e-05,
+      "loss": 0.7785,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.45125150680542,
+      "learning_rate": 1.9126627357113666e-05,
+      "loss": 0.7849,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.8411121368408203,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.335,
+      "eval_steps_per_second": 3.267,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3477277755737305,
+      "learning_rate": 1.9114951252262246e-05,
+      "loss": 0.8265,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4179579019546509,
+      "learning_rate": 1.9103275147410823e-05,
+      "loss": 0.8022,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4127483367919922,
+      "learning_rate": 1.9091599042559403e-05,
+      "loss": 0.7836,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.4985295534133911,
+      "learning_rate": 1.907992293770798e-05,
+      "loss": 0.809,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5186188220977783,
+      "learning_rate": 1.906824683285656e-05,
+      "loss": 0.7767,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.8316742777824402,
+      "eval_runtime": 0.6136,
+      "eval_samples_per_second": 16.296,
+      "eval_steps_per_second": 3.259,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.5827696323394775,
+      "learning_rate": 1.905657072800514e-05,
+      "loss": 0.7906,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4132195711135864,
+      "learning_rate": 1.9044894623153718e-05,
+      "loss": 0.789,
+      "step": 1720
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5781434774398804,
+      "learning_rate": 1.9033218518302298e-05,
+      "loss": 0.7476,
+      "step": 1730
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.2523741722106934,
+      "learning_rate": 1.9021542413450875e-05,
+      "loss": 0.7854,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.7323291301727295,
+      "learning_rate": 1.9009866308599452e-05,
+      "loss": 0.7968,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.8321050405502319,
+      "eval_runtime": 0.6128,
+      "eval_samples_per_second": 16.319,
+      "eval_steps_per_second": 3.264,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4369182586669922,
+      "learning_rate": 1.899819020374803e-05,
+      "loss": 0.7783,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6335262060165405,
+      "learning_rate": 1.898651409889661e-05,
+      "loss": 0.7902,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.514256238937378,
+      "learning_rate": 1.8974837994045186e-05,
+      "loss": 0.7733,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.374267339706421,
+      "learning_rate": 1.8963161889193767e-05,
+      "loss": 0.7773,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.62778639793396,
+      "learning_rate": 1.8951485784342343e-05,
+      "loss": 0.7642,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.827767014503479,
+      "eval_runtime": 0.614,
+      "eval_samples_per_second": 16.287,
+      "eval_steps_per_second": 3.257,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4110387563705444,
+      "learning_rate": 1.8939809679490924e-05,
+      "loss": 0.7552,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.457823634147644,
+      "learning_rate": 1.89281335746395e-05,
+      "loss": 0.8399,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4737167358398438,
+      "learning_rate": 1.891645746978808e-05,
+      "loss": 0.7657,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.6507104635238647,
+      "learning_rate": 1.890478136493666e-05,
+      "loss": 0.8233,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5327750444412231,
+      "learning_rate": 1.8893105260085235e-05,
+      "loss": 0.7833,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.8298091888427734,
+      "eval_runtime": 0.6212,
+      "eval_samples_per_second": 16.098,
+      "eval_steps_per_second": 3.22,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4576866626739502,
+      "learning_rate": 1.8881429155233815e-05,
+      "loss": 0.7549,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.5197103023529053,
+      "learning_rate": 1.8869753050382392e-05,
+      "loss": 0.7615,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.4350595474243164,
+      "learning_rate": 1.8858076945530972e-05,
+      "loss": 0.7821,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.542031168937683,
+      "learning_rate": 1.884640084067955e-05,
+      "loss": 0.8095,
+      "step": 1890
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5905065536499023,
+      "learning_rate": 1.883472473582813e-05,
+      "loss": 0.7699,
+      "step": 1900
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.8264826536178589,
+      "eval_runtime": 0.6146,
+      "eval_samples_per_second": 16.271,
+      "eval_steps_per_second": 3.254,
+      "step": 1900
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.7026077508926392,
+      "learning_rate": 1.8823048630976707e-05,
+      "loss": 0.7875,
+      "step": 1910
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5972000360488892,
+      "learning_rate": 1.8811372526125287e-05,
+      "loss": 0.795,
+      "step": 1920
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5831798315048218,
+      "learning_rate": 1.8799696421273864e-05,
+      "loss": 0.8382,
+      "step": 1930
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4751758575439453,
+      "learning_rate": 1.8788020316422444e-05,
+      "loss": 0.7591,
+      "step": 1940
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.554112434387207,
+      "learning_rate": 1.877634421157102e-05,
+      "loss": 0.8334,
+      "step": 1950
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.8250280618667603,
+      "eval_runtime": 0.6162,
+      "eval_samples_per_second": 16.229,
+      "eval_steps_per_second": 3.246,
+      "step": 1950
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5446873903274536,
+      "learning_rate": 1.8764668106719598e-05,
+      "loss": 0.7913,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5793495178222656,
+      "learning_rate": 1.8752992001868178e-05,
+      "loss": 0.7897,
+      "step": 1970
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5823038816452026,
+      "learning_rate": 1.8741315897016755e-05,
+      "loss": 0.7892,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4952340126037598,
+      "learning_rate": 1.8729639792165335e-05,
+      "loss": 0.7672,
+      "step": 1990
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5737073421478271,
+      "learning_rate": 1.8717963687313912e-05,
+      "loss": 0.791,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.8253053426742554,
+      "eval_runtime": 0.6151,
+      "eval_samples_per_second": 16.258,
+      "eval_steps_per_second": 3.252,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5520979166030884,
+      "learning_rate": 1.8706287582462493e-05,
+      "loss": 0.7838,
+      "step": 2010
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6264978647232056,
+      "learning_rate": 1.869461147761107e-05,
+      "loss": 0.7706,
+      "step": 2020
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5793561935424805,
+      "learning_rate": 1.868293537275965e-05,
+      "loss": 0.763,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4953360557556152,
+      "learning_rate": 1.8671259267908227e-05,
+      "loss": 0.8411,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.6294291019439697,
+      "learning_rate": 1.8659583163056804e-05,
+      "loss": 0.8076,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.8228891491889954,
+      "eval_runtime": 0.6155,
+      "eval_samples_per_second": 16.246,
+      "eval_steps_per_second": 3.249,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.621170997619629,
+      "learning_rate": 1.8647907058205384e-05,
+      "loss": 0.777,
+      "step": 2060
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5768071413040161,
+      "learning_rate": 1.863623095335396e-05,
+      "loss": 0.7893,
+      "step": 2070
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6094276905059814,
+      "learning_rate": 1.862455484850254e-05,
+      "loss": 0.8557,
+      "step": 2080
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.608544111251831,
+      "learning_rate": 1.8612878743651118e-05,
+      "loss": 0.7656,
+      "step": 2090
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6395914554595947,
+      "learning_rate": 1.86012026387997e-05,
+      "loss": 0.8,
+      "step": 2100
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.8171369433403015,
+      "eval_runtime": 0.6107,
+      "eval_samples_per_second": 16.375,
+      "eval_steps_per_second": 3.275,
+      "step": 2100
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5605717897415161,
+      "learning_rate": 1.8589526533948275e-05,
+      "loss": 0.7719,
+      "step": 2110
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6394354104995728,
+      "learning_rate": 1.8577850429096856e-05,
+      "loss": 0.7661,
+      "step": 2120
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.718377947807312,
+      "learning_rate": 1.8566174324245433e-05,
+      "loss": 0.7924,
+      "step": 2130
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5379204750061035,
+      "learning_rate": 1.8554498219394013e-05,
+      "loss": 0.7618,
+      "step": 2140
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6493401527404785,
+      "learning_rate": 1.854282211454259e-05,
+      "loss": 0.775,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.8139251470565796,
+      "eval_runtime": 0.6107,
+      "eval_samples_per_second": 16.374,
+      "eval_steps_per_second": 3.275,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5666680335998535,
+      "learning_rate": 1.8531146009691167e-05,
+      "loss": 0.7462,
+      "step": 2160
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.605649709701538,
+      "learning_rate": 1.8519469904839747e-05,
+      "loss": 0.7753,
+      "step": 2170
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.5899513959884644,
+      "learning_rate": 1.8507793799988324e-05,
+      "loss": 0.7713,
+      "step": 2180
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.716153621673584,
+      "learning_rate": 1.8496117695136904e-05,
+      "loss": 0.767,
+      "step": 2190
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6608442068099976,
+      "learning_rate": 1.848444159028548e-05,
+      "loss": 0.7593,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.812643826007843,
+      "eval_runtime": 0.7629,
+      "eval_samples_per_second": 13.107,
+      "eval_steps_per_second": 2.621,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6444684267044067,
+      "learning_rate": 1.847276548543406e-05,
+      "loss": 0.7848,
+      "step": 2210
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.601854681968689,
+      "learning_rate": 1.846108938058264e-05,
+      "loss": 0.7899,
+      "step": 2220
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.605835199356079,
+      "learning_rate": 1.844941327573122e-05,
+      "loss": 0.8019,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.608300805091858,
+      "learning_rate": 1.8437737170879796e-05,
+      "loss": 0.7878,
+      "step": 2240
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6795828342437744,
+      "learning_rate": 1.8426061066028373e-05,
+      "loss": 0.7391,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.8170574307441711,
+      "eval_runtime": 0.6201,
+      "eval_samples_per_second": 16.125,
+      "eval_steps_per_second": 3.225,
+      "step": 2250
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6474111080169678,
+      "learning_rate": 1.8414384961176953e-05,
+      "loss": 0.7732,
+      "step": 2260
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.548788070678711,
+      "learning_rate": 1.840270885632553e-05,
+      "loss": 0.7604,
+      "step": 2270
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7148107290267944,
+      "learning_rate": 1.839103275147411e-05,
+      "loss": 0.7535,
+      "step": 2280
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6238055229187012,
+      "learning_rate": 1.8379356646622687e-05,
+      "loss": 0.8102,
+      "step": 2290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6822603940963745,
+      "learning_rate": 1.8367680541771267e-05,
+      "loss": 0.7778,
+      "step": 2300
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.8137375116348267,
+      "eval_runtime": 0.6144,
+      "eval_samples_per_second": 16.275,
+      "eval_steps_per_second": 3.255,
+      "step": 2300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6084645986557007,
+      "learning_rate": 1.8356004436919844e-05,
+      "loss": 0.7863,
+      "step": 2310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8630409240722656,
+      "learning_rate": 1.8344328332068425e-05,
+      "loss": 0.7676,
+      "step": 2320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.804220199584961,
+      "learning_rate": 1.8332652227217e-05,
+      "loss": 0.7725,
+      "step": 2330
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6440900564193726,
+      "learning_rate": 1.8320976122365582e-05,
+      "loss": 0.794,
+      "step": 2340
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6076515913009644,
+      "learning_rate": 1.830930001751416e-05,
+      "loss": 0.766,
+      "step": 2350
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.8126281499862671,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.335,
+      "eval_steps_per_second": 3.267,
+      "step": 2350
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.7760066986083984,
+      "learning_rate": 1.8297623912662736e-05,
+      "loss": 0.7772,
+      "step": 2360
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.717779278755188,
+      "learning_rate": 1.8285947807811316e-05,
+      "loss": 0.7996,
+      "step": 2370
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.758806586265564,
+      "learning_rate": 1.8274271702959893e-05,
+      "loss": 0.7929,
+      "step": 2380
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.6087369918823242,
+      "learning_rate": 1.8262595598108473e-05,
+      "loss": 0.733,
+      "step": 2390
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5837440490722656,
+      "learning_rate": 1.825091949325705e-05,
+      "loss": 0.7877,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 0.8130624890327454,
+      "eval_runtime": 0.6477,
+      "eval_samples_per_second": 15.44,
+      "eval_steps_per_second": 3.088,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5343356132507324,
+      "learning_rate": 1.823924338840563e-05,
+      "loss": 0.7795,
+      "step": 2410
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.788965106010437,
+      "learning_rate": 1.8227567283554208e-05,
+      "loss": 0.7919,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5707054138183594,
+      "learning_rate": 1.8215891178702788e-05,
+      "loss": 0.8115,
+      "step": 2430
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.549182415008545,
+      "learning_rate": 1.8204215073851365e-05,
+      "loss": 0.808,
+      "step": 2440
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4891955852508545,
+      "learning_rate": 1.819253896899994e-05,
+      "loss": 0.7587,
+      "step": 2450
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.8115557432174683,
+      "eval_runtime": 0.6139,
+      "eval_samples_per_second": 16.289,
+      "eval_steps_per_second": 3.258,
+      "step": 2450
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6633095741271973,
+      "learning_rate": 1.8180862864148522e-05,
+      "loss": 0.8224,
+      "step": 2460
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7882791757583618,
+      "learning_rate": 1.81691867592971e-05,
+      "loss": 0.7974,
+      "step": 2470
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.600311040878296,
+      "learning_rate": 1.815751065444568e-05,
+      "loss": 0.8012,
+      "step": 2480
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8840075731277466,
+      "learning_rate": 1.8145834549594256e-05,
+      "loss": 0.7886,
+      "step": 2490
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7567533254623413,
+      "learning_rate": 1.8134158444742836e-05,
+      "loss": 0.7815,
+      "step": 2500
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.8141907453536987,
+      "eval_runtime": 0.6165,
+      "eval_samples_per_second": 16.22,
+      "eval_steps_per_second": 3.244,
+      "step": 2500
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5910779237747192,
+      "learning_rate": 1.8122482339891413e-05,
+      "loss": 0.7727,
+      "step": 2510
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.172549843788147,
+      "learning_rate": 1.8110806235039994e-05,
+      "loss": 0.747,
+      "step": 2520
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7534321546554565,
+      "learning_rate": 1.809913013018857e-05,
+      "loss": 0.7726,
+      "step": 2530
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6567786931991577,
+      "learning_rate": 1.808745402533715e-05,
+      "loss": 0.7421,
+      "step": 2540
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.4687511920928955,
+      "learning_rate": 1.8075777920485728e-05,
+      "loss": 0.7773,
+      "step": 2550
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.8118950724601746,
+      "eval_runtime": 0.6131,
+      "eval_samples_per_second": 16.31,
+      "eval_steps_per_second": 3.262,
+      "step": 2550
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.598311424255371,
+      "learning_rate": 1.8064101815634305e-05,
+      "loss": 0.7818,
+      "step": 2560
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.8767908811569214,
+      "learning_rate": 1.8052425710782885e-05,
+      "loss": 0.7533,
+      "step": 2570
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.7389650344848633,
+      "learning_rate": 1.8040749605931462e-05,
+      "loss": 0.7443,
+      "step": 2580
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6800880432128906,
+      "learning_rate": 1.8029073501080042e-05,
+      "loss": 0.7625,
+      "step": 2590
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.5652203559875488,
+      "learning_rate": 1.801739739622862e-05,
+      "loss": 0.7649,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.8082802891731262,
+      "eval_runtime": 0.616,
+      "eval_samples_per_second": 16.233,
+      "eval_steps_per_second": 3.247,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.804059386253357,
+      "learning_rate": 1.80057212913772e-05,
+      "loss": 0.7685,
+      "step": 2610
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6646842956542969,
+      "learning_rate": 1.7994045186525776e-05,
+      "loss": 0.8561,
+      "step": 2620
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6391340494155884,
+      "learning_rate": 1.7982369081674357e-05,
+      "loss": 0.7448,
+      "step": 2630
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.651447057723999,
+      "learning_rate": 1.7970692976822934e-05,
+      "loss": 0.7804,
+      "step": 2640
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6947163343429565,
+      "learning_rate": 1.795901687197151e-05,
+      "loss": 0.7909,
+      "step": 2650
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.8095752000808716,
+      "eval_runtime": 0.6881,
+      "eval_samples_per_second": 14.533,
+      "eval_steps_per_second": 2.907,
+      "step": 2650
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6480817794799805,
+      "learning_rate": 1.794734076712009e-05,
+      "loss": 0.7713,
+      "step": 2660
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.725573182106018,
+      "learning_rate": 1.7935664662268668e-05,
+      "loss": 0.7576,
+      "step": 2670
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6866319179534912,
+      "learning_rate": 1.7923988557417248e-05,
+      "loss": 0.7762,
+      "step": 2680
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7244880199432373,
+      "learning_rate": 1.7912312452565825e-05,
+      "loss": 0.7998,
+      "step": 2690
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.696043610572815,
+      "learning_rate": 1.7900636347714405e-05,
+      "loss": 0.7448,
+      "step": 2700
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.8121166229248047,
+      "eval_runtime": 0.6105,
+      "eval_samples_per_second": 16.379,
+      "eval_steps_per_second": 3.276,
+      "step": 2700
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7087675333023071,
+      "learning_rate": 1.7888960242862982e-05,
+      "loss": 0.7584,
+      "step": 2710
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5538440942764282,
+      "learning_rate": 1.7877284138011563e-05,
+      "loss": 0.7146,
+      "step": 2720
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6454079151153564,
+      "learning_rate": 1.786560803316014e-05,
+      "loss": 0.828,
+      "step": 2730
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.700996994972229,
+      "learning_rate": 1.785393192830872e-05,
+      "loss": 0.7632,
+      "step": 2740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6457117795944214,
+      "learning_rate": 1.7842255823457297e-05,
+      "loss": 0.7759,
+      "step": 2750
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.8119804263114929,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.335,
+      "eval_steps_per_second": 3.267,
+      "step": 2750
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.9449173212051392,
+      "learning_rate": 1.7830579718605874e-05,
+      "loss": 0.7653,
+      "step": 2760
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.6828150749206543,
+      "learning_rate": 1.7818903613754454e-05,
+      "loss": 0.814,
+      "step": 2770
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8183623552322388,
+      "learning_rate": 1.780722750890303e-05,
+      "loss": 0.7572,
+      "step": 2780
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7264996767044067,
+      "learning_rate": 1.779555140405161e-05,
+      "loss": 0.763,
+      "step": 2790
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6708654165267944,
+      "learning_rate": 1.7783875299200188e-05,
+      "loss": 0.7457,
+      "step": 2800
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.8132955431938171,
+      "eval_runtime": 0.6128,
+      "eval_samples_per_second": 16.319,
+      "eval_steps_per_second": 3.264,
+      "step": 2800
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5656832456588745,
+      "learning_rate": 1.777219919434877e-05,
+      "loss": 0.782,
+      "step": 2810
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8406404256820679,
+      "learning_rate": 1.7760523089497345e-05,
+      "loss": 0.7683,
+      "step": 2820
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.9548797607421875,
+      "learning_rate": 1.7748846984645926e-05,
+      "loss": 0.728,
+      "step": 2830
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8788783550262451,
+      "learning_rate": 1.7737170879794503e-05,
+      "loss": 0.7907,
+      "step": 2840
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.8320709466934204,
+      "learning_rate": 1.772549477494308e-05,
+      "loss": 0.7578,
+      "step": 2850
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.811928391456604,
+      "eval_runtime": 0.6141,
+      "eval_samples_per_second": 16.283,
+      "eval_steps_per_second": 3.257,
+      "step": 2850
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.698197364807129,
+      "learning_rate": 1.7713818670091656e-05,
+      "loss": 0.7665,
+      "step": 2860
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.683192253112793,
+      "learning_rate": 1.7702142565240237e-05,
+      "loss": 0.7553,
+      "step": 2870
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7601021528244019,
+      "learning_rate": 1.7690466460388817e-05,
+      "loss": 0.7687,
+      "step": 2880
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.751502513885498,
+      "learning_rate": 1.7678790355537394e-05,
+      "loss": 0.7332,
+      "step": 2890
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.917421817779541,
+      "learning_rate": 1.7667114250685974e-05,
+      "loss": 0.7626,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.8111085891723633,
+      "eval_runtime": 0.6151,
+      "eval_samples_per_second": 16.258,
+      "eval_steps_per_second": 3.252,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6795481443405151,
+      "learning_rate": 1.765543814583455e-05,
+      "loss": 0.7806,
+      "step": 2910
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7080681324005127,
+      "learning_rate": 1.764376204098313e-05,
+      "loss": 0.7404,
+      "step": 2920
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7981574535369873,
+      "learning_rate": 1.763208593613171e-05,
+      "loss": 0.7521,
+      "step": 2930
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.675347924232483,
+      "learning_rate": 1.762040983128029e-05,
+      "loss": 0.7973,
+      "step": 2940
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7315988540649414,
+      "learning_rate": 1.7608733726428862e-05,
+      "loss": 0.7663,
+      "step": 2950
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.8106160163879395,
+      "eval_runtime": 0.6117,
+      "eval_samples_per_second": 16.348,
+      "eval_steps_per_second": 3.27,
+      "step": 2950
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.626943588256836,
+      "learning_rate": 1.7597057621577443e-05,
+      "loss": 0.784,
+      "step": 2960
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.863278865814209,
+      "learning_rate": 1.758538151672602e-05,
+      "loss": 0.7554,
+      "step": 2970
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6058528423309326,
+      "learning_rate": 1.75737054118746e-05,
+      "loss": 0.7783,
+      "step": 2980
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.558549404144287,
+      "learning_rate": 1.7562029307023177e-05,
+      "loss": 0.7657,
+      "step": 2990
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8426812887191772,
+      "learning_rate": 1.7550353202171757e-05,
+      "loss": 0.781,
+      "step": 3000
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 0.8102399110794067,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.343,
+      "eval_steps_per_second": 3.269,
+      "step": 3000
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7108243703842163,
+      "learning_rate": 1.7538677097320337e-05,
+      "loss": 0.8273,
+      "step": 3010
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8647005558013916,
+      "learning_rate": 1.7527000992468914e-05,
+      "loss": 0.7349,
+      "step": 3020
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6504108905792236,
+      "learning_rate": 1.7515324887617495e-05,
+      "loss": 0.7728,
+      "step": 3030
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8317971229553223,
+      "learning_rate": 1.750364878276607e-05,
+      "loss": 0.803,
+      "step": 3040
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6233679056167603,
+      "learning_rate": 1.749197267791465e-05,
+      "loss": 0.7539,
+      "step": 3050
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 0.8070241808891296,
+      "eval_runtime": 0.6134,
+      "eval_samples_per_second": 16.301,
+      "eval_steps_per_second": 3.26,
+      "step": 3050
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.923634648323059,
+      "learning_rate": 1.7480296573063225e-05,
+      "loss": 0.7614,
+      "step": 3060
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7152570486068726,
+      "learning_rate": 1.7468620468211806e-05,
+      "loss": 0.7634,
+      "step": 3070
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6512693166732788,
+      "learning_rate": 1.7456944363360383e-05,
+      "loss": 0.7604,
+      "step": 3080
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.801382064819336,
+      "learning_rate": 1.7445268258508963e-05,
+      "loss": 0.7269,
+      "step": 3090
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.9989584684371948,
+      "learning_rate": 1.743359215365754e-05,
+      "loss": 0.7587,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 0.8063490986824036,
+      "eval_runtime": 0.609,
+      "eval_samples_per_second": 16.421,
+      "eval_steps_per_second": 3.284,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7113401889801025,
+      "learning_rate": 1.742191604880612e-05,
+      "loss": 0.7784,
+      "step": 3110
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.664608120918274,
+      "learning_rate": 1.7410239943954697e-05,
+      "loss": 0.7589,
+      "step": 3120
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.677135944366455,
+      "learning_rate": 1.7398563839103277e-05,
+      "loss": 0.7428,
+      "step": 3130
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.7286300659179688,
+      "learning_rate": 1.7386887734251858e-05,
+      "loss": 0.769,
+      "step": 3140
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.841322660446167,
+      "learning_rate": 1.737521162940043e-05,
+      "loss": 0.7783,
+      "step": 3150
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 0.807155430316925,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.335,
+      "eval_steps_per_second": 3.267,
+      "step": 3150
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7234498262405396,
+      "learning_rate": 1.736353552454901e-05,
+      "loss": 0.763,
+      "step": 3160
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.870957612991333,
+      "learning_rate": 1.735185941969759e-05,
+      "loss": 0.781,
+      "step": 3170
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.0361716747283936,
+      "learning_rate": 1.734018331484617e-05,
+      "loss": 0.7541,
+      "step": 3180
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7071706056594849,
+      "learning_rate": 1.7328507209994746e-05,
+      "loss": 0.7713,
+      "step": 3190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7161351442337036,
+      "learning_rate": 1.7316831105143326e-05,
+      "loss": 0.7374,
+      "step": 3200
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.8112953901290894,
+      "eval_runtime": 0.6171,
+      "eval_samples_per_second": 16.205,
+      "eval_steps_per_second": 3.241,
+      "step": 3200
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.5581424236297607,
+      "learning_rate": 1.7305155000291903e-05,
+      "loss": 0.7929,
+      "step": 3210
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6825488805770874,
+      "learning_rate": 1.7293478895440483e-05,
+      "loss": 0.7851,
+      "step": 3220
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.6949360370635986,
+      "learning_rate": 1.728180279058906e-05,
+      "loss": 0.758,
+      "step": 3230
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7868378162384033,
+      "learning_rate": 1.727012668573764e-05,
+      "loss": 0.8034,
+      "step": 3240
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.796143889427185,
+      "learning_rate": 1.7258450580886217e-05,
+      "loss": 0.7783,
+      "step": 3250
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.808771014213562,
+      "eval_runtime": 0.6087,
+      "eval_samples_per_second": 16.428,
+      "eval_steps_per_second": 3.286,
+      "step": 3250
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9562104940414429,
+      "learning_rate": 1.7246774476034794e-05,
+      "loss": 0.7654,
+      "step": 3260
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.970877766609192,
+      "learning_rate": 1.7235098371183375e-05,
+      "loss": 0.8121,
+      "step": 3270
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.9556795358657837,
+      "learning_rate": 1.722342226633195e-05,
+      "loss": 0.7763,
+      "step": 3280
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.622677206993103,
+      "learning_rate": 1.7211746161480532e-05,
+      "loss": 0.7379,
+      "step": 3290
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8337496519088745,
+      "learning_rate": 1.720007005662911e-05,
+      "loss": 0.7911,
+      "step": 3300
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.8071421384811401,
+      "eval_runtime": 0.6171,
+      "eval_samples_per_second": 16.204,
+      "eval_steps_per_second": 3.241,
+      "step": 3300
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.7908275127410889,
+      "learning_rate": 1.718839395177769e-05,
+      "loss": 0.76,
+      "step": 3310
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8745039701461792,
+      "learning_rate": 1.7176717846926266e-05,
+      "loss": 0.7381,
+      "step": 3320
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8303120136260986,
+      "learning_rate": 1.7165041742074846e-05,
+      "loss": 0.7196,
+      "step": 3330
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7722407579421997,
+      "learning_rate": 1.7153365637223423e-05,
+      "loss": 0.7727,
+      "step": 3340
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7016830444335938,
+      "learning_rate": 1.7141689532372e-05,
+      "loss": 0.7624,
+      "step": 3350
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.8055642247200012,
+      "eval_runtime": 0.6082,
+      "eval_samples_per_second": 16.441,
+      "eval_steps_per_second": 3.288,
+      "step": 3350
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7147846221923828,
+      "learning_rate": 1.713001342752058e-05,
+      "loss": 0.7408,
+      "step": 3360
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8144420385360718,
+      "learning_rate": 1.7118337322669157e-05,
+      "loss": 0.7385,
+      "step": 3370
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8798972368240356,
+      "learning_rate": 1.7106661217817738e-05,
+      "loss": 0.7808,
+      "step": 3380
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9121612310409546,
+      "learning_rate": 1.7094985112966315e-05,
+      "loss": 0.7847,
+      "step": 3390
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8080029487609863,
+      "learning_rate": 1.7083309008114895e-05,
+      "loss": 0.7553,
+      "step": 3400
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.8064807653427124,
+      "eval_runtime": 0.6137,
+      "eval_samples_per_second": 16.296,
+      "eval_steps_per_second": 3.259,
+      "step": 3400
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7861032485961914,
+      "learning_rate": 1.7071632903263472e-05,
+      "loss": 0.7416,
+      "step": 3410
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8840128183364868,
+      "learning_rate": 1.7059956798412052e-05,
+      "loss": 0.7179,
+      "step": 3420
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8472384214401245,
+      "learning_rate": 1.704828069356063e-05,
+      "loss": 0.7285,
+      "step": 3430
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8753427267074585,
+      "learning_rate": 1.703660458870921e-05,
+      "loss": 0.7597,
+      "step": 3440
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7295902967453003,
+      "learning_rate": 1.7024928483857786e-05,
+      "loss": 0.7715,
+      "step": 3450
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.8035610318183899,
+      "eval_runtime": 0.615,
+      "eval_samples_per_second": 16.261,
+      "eval_steps_per_second": 3.252,
+      "step": 3450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7846031188964844,
+      "learning_rate": 1.7013252379006363e-05,
+      "loss": 0.71,
+      "step": 3460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.828118085861206,
+      "learning_rate": 1.7001576274154944e-05,
+      "loss": 0.7265,
+      "step": 3470
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.9868234395980835,
+      "learning_rate": 1.698990016930352e-05,
+      "loss": 0.7682,
+      "step": 3480
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6777920722961426,
+      "learning_rate": 1.69782240644521e-05,
+      "loss": 0.7879,
+      "step": 3490
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.833130955696106,
+      "learning_rate": 1.6966547959600678e-05,
+      "loss": 0.7577,
+      "step": 3500
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.807073712348938,
+      "eval_runtime": 0.613,
+      "eval_samples_per_second": 16.313,
+      "eval_steps_per_second": 3.263,
+      "step": 3500
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8520170450210571,
+      "learning_rate": 1.6954871854749258e-05,
+      "loss": 0.7894,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7053380012512207,
+      "learning_rate": 1.6943195749897835e-05,
+      "loss": 0.7647,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8893325328826904,
+      "learning_rate": 1.6931519645046415e-05,
+      "loss": 0.7474,
+      "step": 3530
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7591818571090698,
+      "learning_rate": 1.6919843540194992e-05,
+      "loss": 0.7347,
+      "step": 3540
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.767617106437683,
+      "learning_rate": 1.690816743534357e-05,
+      "loss": 0.7454,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.8065764307975769,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.339,
+      "eval_steps_per_second": 3.268,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.865065336227417,
+      "learning_rate": 1.689649133049215e-05,
+      "loss": 0.7242,
+      "step": 3560
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.776489496231079,
+      "learning_rate": 1.6884815225640726e-05,
+      "loss": 0.8146,
+      "step": 3570
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7679094076156616,
+      "learning_rate": 1.6873139120789307e-05,
+      "loss": 0.7349,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7521227598190308,
+      "learning_rate": 1.6861463015937884e-05,
+      "loss": 0.7327,
+      "step": 3590
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8284246921539307,
+      "learning_rate": 1.6849786911086464e-05,
+      "loss": 0.7076,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.8036044836044312,
+      "eval_runtime": 0.6125,
+      "eval_samples_per_second": 16.327,
+      "eval_steps_per_second": 3.265,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.60626220703125,
+      "learning_rate": 1.683811080623504e-05,
+      "loss": 0.7698,
+      "step": 3610
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.8618712425231934,
+      "learning_rate": 1.682643470138362e-05,
+      "loss": 0.7435,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.859926462173462,
+      "learning_rate": 1.6814758596532198e-05,
+      "loss": 0.7647,
+      "step": 3630
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.889540195465088,
+      "learning_rate": 1.680308249168078e-05,
+      "loss": 0.7379,
+      "step": 3640
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9317718744277954,
+      "learning_rate": 1.6791406386829355e-05,
+      "loss": 0.7791,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 0.8025843501091003,
+      "eval_runtime": 0.6105,
+      "eval_samples_per_second": 16.38,
+      "eval_steps_per_second": 3.276,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6960883140563965,
+      "learning_rate": 1.6779730281977932e-05,
+      "loss": 0.7336,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.854125738143921,
+      "learning_rate": 1.6768054177126513e-05,
+      "loss": 0.7657,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.7691715955734253,
+      "learning_rate": 1.675637807227509e-05,
+      "loss": 0.751,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.1935060024261475,
+      "learning_rate": 1.674470196742367e-05,
+      "loss": 0.7588,
+      "step": 3690
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8579378128051758,
+      "learning_rate": 1.6733025862572247e-05,
+      "loss": 0.767,
+      "step": 3700
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.7996485829353333,
+      "eval_runtime": 0.6091,
+      "eval_samples_per_second": 16.418,
+      "eval_steps_per_second": 3.284,
+      "step": 3700
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8725074529647827,
+      "learning_rate": 1.6721349757720827e-05,
+      "loss": 0.7332,
+      "step": 3710
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8373311758041382,
+      "learning_rate": 1.6709673652869404e-05,
+      "loss": 0.7289,
+      "step": 3720
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.866402268409729,
+      "learning_rate": 1.6697997548017984e-05,
+      "loss": 0.7562,
+      "step": 3730
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7181967496871948,
+      "learning_rate": 1.668632144316656e-05,
+      "loss": 0.7266,
+      "step": 3740
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8862179517745972,
+      "learning_rate": 1.6674645338315138e-05,
+      "loss": 0.7694,
+      "step": 3750
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.8013283014297485,
+      "eval_runtime": 0.6126,
+      "eval_samples_per_second": 16.323,
+      "eval_steps_per_second": 3.265,
+      "step": 3750
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7550722360610962,
+      "learning_rate": 1.666296923346372e-05,
+      "loss": 0.7848,
+      "step": 3760
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7283313274383545,
+      "learning_rate": 1.6651293128612295e-05,
+      "loss": 0.7839,
+      "step": 3770
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9237864017486572,
+      "learning_rate": 1.6639617023760876e-05,
+      "loss": 0.7568,
+      "step": 3780
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.104057550430298,
+      "learning_rate": 1.6627940918909453e-05,
+      "loss": 0.7351,
+      "step": 3790
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8582379817962646,
+      "learning_rate": 1.6616264814058033e-05,
+      "loss": 0.7685,
+      "step": 3800
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.8003425598144531,
+      "eval_runtime": 0.6116,
+      "eval_samples_per_second": 16.352,
+      "eval_steps_per_second": 3.27,
+      "step": 3800
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7590450048446655,
+      "learning_rate": 1.660458870920661e-05,
+      "loss": 0.7658,
+      "step": 3810
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.827046275138855,
+      "learning_rate": 1.659291260435519e-05,
+      "loss": 0.7918,
+      "step": 3820
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.9223333597183228,
+      "learning_rate": 1.6581236499503767e-05,
+      "loss": 0.7826,
+      "step": 3830
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.8338439464569092,
+      "learning_rate": 1.6569560394652344e-05,
+      "loss": 0.7209,
+      "step": 3840
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7311135530471802,
+      "learning_rate": 1.6557884289800924e-05,
+      "loss": 0.7659,
+      "step": 3850
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.7991049885749817,
+      "eval_runtime": 0.6104,
+      "eval_samples_per_second": 16.383,
+      "eval_steps_per_second": 3.277,
+      "step": 3850
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 2.0367937088012695,
+      "learning_rate": 1.65462081849495e-05,
+      "loss": 0.7699,
+      "step": 3860
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7700647115707397,
+      "learning_rate": 1.653453208009808e-05,
+      "loss": 0.7868,
+      "step": 3870
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9972193241119385,
+      "learning_rate": 1.652285597524666e-05,
+      "loss": 0.7479,
+      "step": 3880
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.837267279624939,
+      "learning_rate": 1.651117987039524e-05,
+      "loss": 0.721,
+      "step": 3890
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.946642518043518,
+      "learning_rate": 1.6499503765543816e-05,
+      "loss": 0.745,
+      "step": 3900
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.800929069519043,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.335,
+      "eval_steps_per_second": 3.267,
+      "step": 3900
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0362648963928223,
+      "learning_rate": 1.6487827660692396e-05,
+      "loss": 0.7465,
+      "step": 3910
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8187445402145386,
+      "learning_rate": 1.6476151555840973e-05,
+      "loss": 0.7415,
+      "step": 3920
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8124090433120728,
+      "learning_rate": 1.6464475450989553e-05,
+      "loss": 0.7343,
+      "step": 3930
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8152122497558594,
+      "learning_rate": 1.645279934613813e-05,
+      "loss": 0.7567,
+      "step": 3940
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7911946773529053,
+      "learning_rate": 1.6441123241286707e-05,
+      "loss": 0.7347,
+      "step": 3950
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.8003100156784058,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.358,
+      "eval_steps_per_second": 3.272,
+      "step": 3950
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.837650179862976,
+      "learning_rate": 1.6429447136435287e-05,
+      "loss": 0.7676,
+      "step": 3960
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9903982877731323,
+      "learning_rate": 1.6417771031583864e-05,
+      "loss": 0.7048,
+      "step": 3970
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.0357728004455566,
+      "learning_rate": 1.6406094926732445e-05,
+      "loss": 0.7656,
+      "step": 3980
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9021623134613037,
+      "learning_rate": 1.639441882188102e-05,
+      "loss": 0.7593,
+      "step": 3990
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7145459651947021,
+      "learning_rate": 1.6382742717029602e-05,
+      "loss": 0.7667,
+      "step": 4000
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.7987167239189148,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.36,
+      "eval_steps_per_second": 3.272,
+      "step": 4000
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7253038883209229,
+      "learning_rate": 1.637106661217818e-05,
+      "loss": 0.7533,
+      "step": 4010
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.7158074378967285,
+      "learning_rate": 1.635939050732676e-05,
+      "loss": 0.751,
+      "step": 4020
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9340077638626099,
+      "learning_rate": 1.6347714402475336e-05,
+      "loss": 0.7678,
+      "step": 4030
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.8820750713348389,
+      "learning_rate": 1.6336038297623913e-05,
+      "loss": 0.8206,
+      "step": 4040
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9036202430725098,
+      "learning_rate": 1.6324362192772493e-05,
+      "loss": 0.7142,
+      "step": 4050
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 0.7943220138549805,
+      "eval_runtime": 0.6021,
+      "eval_samples_per_second": 16.61,
+      "eval_steps_per_second": 3.322,
+      "step": 4050
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9611130952835083,
+      "learning_rate": 1.631268608792107e-05,
+      "loss": 0.7616,
+      "step": 4060
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8484268188476562,
+      "learning_rate": 1.630100998306965e-05,
+      "loss": 0.7383,
+      "step": 4070
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8817882537841797,
+      "learning_rate": 1.6289333878218227e-05,
+      "loss": 0.7335,
+      "step": 4080
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.116612434387207,
+      "learning_rate": 1.6277657773366808e-05,
+      "loss": 0.7579,
+      "step": 4090
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9358896017074585,
+      "learning_rate": 1.6265981668515385e-05,
+      "loss": 0.7267,
+      "step": 4100
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.7959241271018982,
+      "eval_runtime": 0.603,
+      "eval_samples_per_second": 16.583,
+      "eval_steps_per_second": 3.317,
+      "step": 4100
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8836027383804321,
+      "learning_rate": 1.6254305563663965e-05,
+      "loss": 0.7683,
+      "step": 4110
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7520395517349243,
+      "learning_rate": 1.6242629458812542e-05,
+      "loss": 0.7439,
+      "step": 4120
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.917356252670288,
+      "learning_rate": 1.6230953353961122e-05,
+      "loss": 0.7457,
+      "step": 4130
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.779251217842102,
+      "learning_rate": 1.6219277249109696e-05,
+      "loss": 0.7653,
+      "step": 4140
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.066404104232788,
+      "learning_rate": 1.6207601144258276e-05,
+      "loss": 0.7358,
+      "step": 4150
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.7986497282981873,
+      "eval_runtime": 0.6151,
+      "eval_samples_per_second": 16.259,
+      "eval_steps_per_second": 3.252,
+      "step": 4150
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.930268406867981,
+      "learning_rate": 1.6195925039406853e-05,
+      "loss": 0.7427,
+      "step": 4160
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.0457441806793213,
+      "learning_rate": 1.6184248934555433e-05,
+      "loss": 0.773,
+      "step": 4170
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.812563180923462,
+      "learning_rate": 1.6172572829704014e-05,
+      "loss": 0.743,
+      "step": 4180
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9552372694015503,
+      "learning_rate": 1.616089672485259e-05,
+      "loss": 0.759,
+      "step": 4190
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9468024969100952,
+      "learning_rate": 1.614922062000117e-05,
+      "loss": 0.7312,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.7997319102287292,
+      "eval_runtime": 0.6128,
+      "eval_samples_per_second": 16.318,
+      "eval_steps_per_second": 3.264,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8385539054870605,
+      "learning_rate": 1.6137544515149748e-05,
+      "loss": 0.6876,
+      "step": 4210
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.9853817224502563,
+      "learning_rate": 1.6125868410298328e-05,
+      "loss": 0.7623,
+      "step": 4220
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.7769476175308228,
+      "learning_rate": 1.6114192305446905e-05,
+      "loss": 0.738,
+      "step": 4230
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7060681581497192,
+      "learning_rate": 1.6102516200595482e-05,
+      "loss": 0.7328,
+      "step": 4240
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.3390636444091797,
+      "learning_rate": 1.609084009574406e-05,
+      "loss": 0.7452,
+      "step": 4250
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.8025315999984741,
+      "eval_runtime": 0.6106,
+      "eval_samples_per_second": 16.378,
+      "eval_steps_per_second": 3.276,
+      "step": 4250
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9849003553390503,
+      "learning_rate": 1.607916399089264e-05,
+      "loss": 0.7362,
+      "step": 4260
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.805030345916748,
+      "learning_rate": 1.6067487886041216e-05,
+      "loss": 0.7721,
+      "step": 4270
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9208256006240845,
+      "learning_rate": 1.6055811781189796e-05,
+      "loss": 0.7585,
+      "step": 4280
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.06467866897583,
+      "learning_rate": 1.6044135676338373e-05,
+      "loss": 0.7433,
+      "step": 4290
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7585352659225464,
+      "learning_rate": 1.6032459571486954e-05,
+      "loss": 0.7394,
+      "step": 4300
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.7972410917282104,
+      "eval_runtime": 0.6141,
+      "eval_samples_per_second": 16.283,
+      "eval_steps_per_second": 3.257,
+      "step": 4300
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8073714971542358,
+      "learning_rate": 1.6020783466635534e-05,
+      "loss": 0.7551,
+      "step": 4310
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9544780254364014,
+      "learning_rate": 1.600910736178411e-05,
+      "loss": 0.7408,
+      "step": 4320
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9564666748046875,
+      "learning_rate": 1.599743125693269e-05,
+      "loss": 0.7379,
+      "step": 4330
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6817543506622314,
+      "learning_rate": 1.5985755152081265e-05,
+      "loss": 0.7406,
+      "step": 4340
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6767189502716064,
+      "learning_rate": 1.5974079047229845e-05,
+      "loss": 0.7267,
+      "step": 4350
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.7988241314888,
+      "eval_runtime": 0.613,
+      "eval_samples_per_second": 16.313,
+      "eval_steps_per_second": 3.263,
+      "step": 4350
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.971882939338684,
+      "learning_rate": 1.5962402942378422e-05,
+      "loss": 0.7545,
+      "step": 4360
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8934246301651,
+      "learning_rate": 1.5950726837527002e-05,
+      "loss": 0.7742,
+      "step": 4370
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.8758774995803833,
+      "learning_rate": 1.593905073267558e-05,
+      "loss": 0.7334,
+      "step": 4380
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.903163194656372,
+      "learning_rate": 1.592737462782416e-05,
+      "loss": 0.7091,
+      "step": 4390
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 2.2816100120544434,
+      "learning_rate": 1.5915698522972736e-05,
+      "loss": 0.7597,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.7970215082168579,
+      "eval_runtime": 0.6117,
+      "eval_samples_per_second": 16.347,
+      "eval_steps_per_second": 3.269,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.9306708574295044,
+      "learning_rate": 1.5904022418121317e-05,
+      "loss": 0.7323,
+      "step": 4410
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.6347606182098389,
+      "learning_rate": 1.5892346313269894e-05,
+      "loss": 0.7784,
+      "step": 4420
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.767989993095398,
+      "learning_rate": 1.5880670208418474e-05,
+      "loss": 0.7389,
+      "step": 4430
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9894226789474487,
+      "learning_rate": 1.586899410356705e-05,
+      "loss": 0.7779,
+      "step": 4440
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8943524360656738,
+      "learning_rate": 1.5857317998715628e-05,
+      "loss": 0.7675,
+      "step": 4450
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.7969298958778381,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.365,
+      "eval_steps_per_second": 3.273,
+      "step": 4450
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7042205333709717,
+      "learning_rate": 1.5845641893864208e-05,
+      "loss": 0.7392,
+      "step": 4460
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.732508659362793,
+      "learning_rate": 1.5833965789012785e-05,
+      "loss": 0.7912,
+      "step": 4470
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.956009030342102,
+      "learning_rate": 1.5822289684161365e-05,
+      "loss": 0.7945,
+      "step": 4480
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.815978765487671,
+      "learning_rate": 1.5810613579309942e-05,
+      "loss": 0.758,
+      "step": 4490
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.1841483116149902,
+      "learning_rate": 1.5798937474458522e-05,
+      "loss": 0.7607,
+      "step": 4500
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.7960728406906128,
+      "eval_runtime": 0.6115,
+      "eval_samples_per_second": 16.354,
+      "eval_steps_per_second": 3.271,
+      "step": 4500
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8473882675170898,
+      "learning_rate": 1.57872613696071e-05,
+      "loss": 0.7831,
+      "step": 4510
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8496242761611938,
+      "learning_rate": 1.577558526475568e-05,
+      "loss": 0.7456,
+      "step": 4520
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.91486656665802,
+      "learning_rate": 1.5763909159904257e-05,
+      "loss": 0.7427,
+      "step": 4530
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9782878160476685,
+      "learning_rate": 1.5752233055052834e-05,
+      "loss": 0.7568,
+      "step": 4540
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.855936050415039,
+      "learning_rate": 1.5741724560686555e-05,
+      "loss": 0.7416,
+      "step": 4550
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.7999919652938843,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.339,
+      "eval_steps_per_second": 3.268,
+      "step": 4550
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9870688915252686,
+      "learning_rate": 1.5730048455835132e-05,
+      "loss": 0.725,
+      "step": 4560
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.7163158655166626,
+      "learning_rate": 1.5718372350983713e-05,
+      "loss": 0.7211,
+      "step": 4570
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.9278556108474731,
+      "learning_rate": 1.5706696246132293e-05,
+      "loss": 0.7337,
+      "step": 4580
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.8344335556030273,
+      "learning_rate": 1.569502014128087e-05,
+      "loss": 0.7303,
+      "step": 4590
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.940553903579712,
+      "learning_rate": 1.568334403642945e-05,
+      "loss": 0.7803,
+      "step": 4600
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.793971061706543,
+      "eval_runtime": 0.6158,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 3.248,
+      "step": 4600
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8804936408996582,
+      "learning_rate": 1.5671667931578027e-05,
+      "loss": 0.7388,
+      "step": 4610
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8510812520980835,
+      "learning_rate": 1.5659991826726608e-05,
+      "loss": 0.7316,
+      "step": 4620
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.037454128265381,
+      "learning_rate": 1.5648315721875184e-05,
+      "loss": 0.7308,
+      "step": 4630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.0108394622802734,
+      "learning_rate": 1.563663961702376e-05,
+      "loss": 0.7276,
+      "step": 4640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9425954818725586,
+      "learning_rate": 1.5624963512172338e-05,
+      "loss": 0.721,
+      "step": 4650
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.7969173192977905,
+      "eval_runtime": 0.6152,
+      "eval_samples_per_second": 16.256,
+      "eval_steps_per_second": 3.251,
+      "step": 4650
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8366847038269043,
+      "learning_rate": 1.561328740732092e-05,
+      "loss": 0.755,
+      "step": 4660
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8646974563598633,
+      "learning_rate": 1.5601611302469496e-05,
+      "loss": 0.7324,
+      "step": 4670
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8410062789916992,
+      "learning_rate": 1.5589935197618076e-05,
+      "loss": 0.7264,
+      "step": 4680
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8647485971450806,
+      "learning_rate": 1.5578259092766653e-05,
+      "loss": 0.8073,
+      "step": 4690
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9440714120864868,
+      "learning_rate": 1.5566582987915233e-05,
+      "loss": 0.7605,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.7980841398239136,
+      "eval_runtime": 0.6166,
+      "eval_samples_per_second": 16.218,
+      "eval_steps_per_second": 3.244,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.8777036666870117,
+      "learning_rate": 1.5554906883063813e-05,
+      "loss": 0.7803,
+      "step": 4710
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.919523000717163,
+      "learning_rate": 1.554323077821239e-05,
+      "loss": 0.7427,
+      "step": 4720
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9593130350112915,
+      "learning_rate": 1.553155467336097e-05,
+      "loss": 0.741,
+      "step": 4730
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9281760454177856,
+      "learning_rate": 1.5519878568509544e-05,
+      "loss": 0.7488,
+      "step": 4740
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9199045896530151,
+      "learning_rate": 1.5508202463658124e-05,
+      "loss": 0.7326,
+      "step": 4750
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.7964409589767456,
+      "eval_runtime": 0.616,
+      "eval_samples_per_second": 16.233,
+      "eval_steps_per_second": 3.247,
+      "step": 4750
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.027524709701538,
+      "learning_rate": 1.54965263588067e-05,
+      "loss": 0.7259,
+      "step": 4760
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 2.067150354385376,
+      "learning_rate": 1.548485025395528e-05,
+      "loss": 0.7475,
+      "step": 4770
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9131090641021729,
+      "learning_rate": 1.547317414910386e-05,
+      "loss": 0.7694,
+      "step": 4780
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8171725273132324,
+      "learning_rate": 1.546149804425244e-05,
+      "loss": 0.7501,
+      "step": 4790
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.037759780883789,
+      "learning_rate": 1.5449821939401016e-05,
+      "loss": 0.75,
+      "step": 4800
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.7951205968856812,
+      "eval_runtime": 0.6132,
+      "eval_samples_per_second": 16.309,
+      "eval_steps_per_second": 3.262,
+      "step": 4800
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7844374179840088,
+      "learning_rate": 1.5438145834549596e-05,
+      "loss": 0.7287,
+      "step": 4810
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0128531455993652,
+      "learning_rate": 1.5426469729698173e-05,
+      "loss": 0.695,
+      "step": 4820
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7967103719711304,
+      "learning_rate": 1.5414793624846753e-05,
+      "loss": 0.7419,
+      "step": 4830
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8948699235916138,
+      "learning_rate": 1.540311751999533e-05,
+      "loss": 0.7575,
+      "step": 4840
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8341615200042725,
+      "learning_rate": 1.5391441415143907e-05,
+      "loss": 0.7325,
+      "step": 4850
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.7953310012817383,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.359,
+      "eval_steps_per_second": 3.272,
+      "step": 4850
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.0162975788116455,
+      "learning_rate": 1.5379765310292488e-05,
+      "loss": 0.748,
+      "step": 4860
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8275803327560425,
+      "learning_rate": 1.5368089205441064e-05,
+      "loss": 0.7276,
+      "step": 4870
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9825547933578491,
+      "learning_rate": 1.5356413100589645e-05,
+      "loss": 0.7276,
+      "step": 4880
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7838776111602783,
+      "learning_rate": 1.534473699573822e-05,
+      "loss": 0.7318,
+      "step": 4890
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8920131921768188,
+      "learning_rate": 1.5333060890886802e-05,
+      "loss": 0.7245,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.7934932708740234,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.415,
+      "eval_steps_per_second": 3.283,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.9621186256408691,
+      "learning_rate": 1.532138478603538e-05,
+      "loss": 0.7191,
+      "step": 4910
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.011176586151123,
+      "learning_rate": 1.530970868118396e-05,
+      "loss": 0.7288,
+      "step": 4920
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.889427900314331,
+      "learning_rate": 1.5298032576332536e-05,
+      "loss": 0.7298,
+      "step": 4930
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7516738176345825,
+      "learning_rate": 1.5286356471481113e-05,
+      "loss": 0.7242,
+      "step": 4940
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.872968316078186,
+      "learning_rate": 1.5274680366629693e-05,
+      "loss": 0.7309,
+      "step": 4950
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.7972428202629089,
+      "eval_runtime": 0.6094,
+      "eval_samples_per_second": 16.41,
+      "eval_steps_per_second": 3.282,
+      "step": 4950
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9194343090057373,
+      "learning_rate": 1.526300426177827e-05,
+      "loss": 0.7592,
+      "step": 4960
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2093405723571777,
+      "learning_rate": 1.525132815692685e-05,
+      "loss": 0.7437,
+      "step": 4970
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7437870502471924,
+      "learning_rate": 1.523965205207543e-05,
+      "loss": 0.7414,
+      "step": 4980
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9983580112457275,
+      "learning_rate": 1.5227975947224008e-05,
+      "loss": 0.7384,
+      "step": 4990
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7982289791107178,
+      "learning_rate": 1.5216299842372586e-05,
+      "loss": 0.7452,
+      "step": 5000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.7945663928985596,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.343,
+      "eval_steps_per_second": 3.269,
+      "step": 5000
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7225558757781982,
+      "learning_rate": 1.5204623737521165e-05,
+      "loss": 0.7632,
+      "step": 5010
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9274910688400269,
+      "learning_rate": 1.5192947632669744e-05,
+      "loss": 0.7248,
+      "step": 5020
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6656044721603394,
+      "learning_rate": 1.5181271527818322e-05,
+      "loss": 0.7307,
+      "step": 5030
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9163330793380737,
+      "learning_rate": 1.51695954229669e-05,
+      "loss": 0.767,
+      "step": 5040
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.03719425201416,
+      "learning_rate": 1.5157919318115478e-05,
+      "loss": 0.7702,
+      "step": 5050
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.793404757976532,
+      "eval_runtime": 0.6072,
+      "eval_samples_per_second": 16.47,
+      "eval_steps_per_second": 3.294,
+      "step": 5050
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.2201104164123535,
+      "learning_rate": 1.5146243213264056e-05,
+      "loss": 0.7603,
+      "step": 5060
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8825465440750122,
+      "learning_rate": 1.5134567108412635e-05,
+      "loss": 0.7466,
+      "step": 5070
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.9366422891616821,
+      "learning_rate": 1.5122891003561214e-05,
+      "loss": 0.7319,
+      "step": 5080
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.581455945968628,
+      "learning_rate": 1.5111214898709792e-05,
+      "loss": 0.7494,
+      "step": 5090
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.6154475212097168,
+      "learning_rate": 1.5099538793858371e-05,
+      "loss": 0.7581,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.7927392721176147,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.368,
+      "eval_steps_per_second": 3.274,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0091278553009033,
+      "learning_rate": 1.508786268900695e-05,
+      "loss": 0.7089,
+      "step": 5110
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.7709681987762451,
+      "learning_rate": 1.5076186584155528e-05,
+      "loss": 0.782,
+      "step": 5120
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.8411822319030762,
+      "learning_rate": 1.5064510479304107e-05,
+      "loss": 0.7437,
+      "step": 5130
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9285340309143066,
+      "learning_rate": 1.5052834374452682e-05,
+      "loss": 0.7592,
+      "step": 5140
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8929240703582764,
+      "learning_rate": 1.504115826960126e-05,
+      "loss": 0.7274,
+      "step": 5150
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 0.7955920100212097,
+      "eval_runtime": 0.6114,
+      "eval_samples_per_second": 16.357,
+      "eval_steps_per_second": 3.271,
+      "step": 5150
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9914500713348389,
+      "learning_rate": 1.5029482164749841e-05,
+      "loss": 0.7463,
+      "step": 5160
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.850766658782959,
+      "learning_rate": 1.501780605989842e-05,
+      "loss": 0.7159,
+      "step": 5170
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0711920261383057,
+      "learning_rate": 1.5006129955046998e-05,
+      "loss": 0.7154,
+      "step": 5180
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.101351022720337,
+      "learning_rate": 1.4994453850195577e-05,
+      "loss": 0.7359,
+      "step": 5190
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.094176769256592,
+      "learning_rate": 1.4982777745344155e-05,
+      "loss": 0.7453,
+      "step": 5200
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 0.7972155809402466,
+      "eval_runtime": 0.6102,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 3.277,
+      "step": 5200
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0258727073669434,
+      "learning_rate": 1.4971101640492734e-05,
+      "loss": 0.7418,
+      "step": 5210
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8117177486419678,
+      "learning_rate": 1.4959425535641313e-05,
+      "loss": 0.7383,
+      "step": 5220
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9444230794906616,
+      "learning_rate": 1.4947749430789891e-05,
+      "loss": 0.7621,
+      "step": 5230
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7746025323867798,
+      "learning_rate": 1.4936073325938466e-05,
+      "loss": 0.7313,
+      "step": 5240
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.221085548400879,
+      "learning_rate": 1.4924397221087045e-05,
+      "loss": 0.7377,
+      "step": 5250
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 0.7950941920280457,
+      "eval_runtime": 0.6271,
+      "eval_samples_per_second": 15.947,
+      "eval_steps_per_second": 3.189,
+      "step": 5250
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9495569467544556,
+      "learning_rate": 1.4912721116235624e-05,
+      "loss": 0.8164,
+      "step": 5260
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 2.0034453868865967,
+      "learning_rate": 1.4901045011384202e-05,
+      "loss": 0.7388,
+      "step": 5270
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.8646876811981201,
+      "learning_rate": 1.4889368906532781e-05,
+      "loss": 0.7251,
+      "step": 5280
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.790398359298706,
+      "learning_rate": 1.4877692801681361e-05,
+      "loss": 0.7275,
+      "step": 5290
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7589030265808105,
+      "learning_rate": 1.486601669682994e-05,
+      "loss": 0.7471,
+      "step": 5300
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 0.7960600256919861,
+      "eval_runtime": 0.6135,
+      "eval_samples_per_second": 16.299,
+      "eval_steps_per_second": 3.26,
+      "step": 5300
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.7687097787857056,
+      "learning_rate": 1.4854340591978518e-05,
+      "loss": 0.7384,
+      "step": 5310
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9463614225387573,
+      "learning_rate": 1.4842664487127097e-05,
+      "loss": 0.7375,
+      "step": 5320
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9503099918365479,
+      "learning_rate": 1.4830988382275676e-05,
+      "loss": 0.7551,
+      "step": 5330
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.7840229272842407,
+      "learning_rate": 1.4819312277424251e-05,
+      "loss": 0.7329,
+      "step": 5340
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.87968909740448,
+      "learning_rate": 1.480763617257283e-05,
+      "loss": 0.765,
+      "step": 5350
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.7926689386367798,
+      "eval_runtime": 0.6098,
+      "eval_samples_per_second": 16.4,
+      "eval_steps_per_second": 3.28,
+      "step": 5350
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9858055114746094,
+      "learning_rate": 1.4795960067721408e-05,
+      "loss": 0.7616,
+      "step": 5360
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9157699346542358,
+      "learning_rate": 1.4784283962869987e-05,
+      "loss": 0.7226,
+      "step": 5370
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.979188323020935,
+      "learning_rate": 1.4772607858018565e-05,
+      "loss": 0.7609,
+      "step": 5380
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8780100345611572,
+      "learning_rate": 1.4760931753167144e-05,
+      "loss": 0.7482,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.907871127128601,
+      "learning_rate": 1.4749255648315723e-05,
+      "loss": 0.7342,
+      "step": 5400
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.7913875579833984,
+      "eval_runtime": 0.6135,
+      "eval_samples_per_second": 16.301,
+      "eval_steps_per_second": 3.26,
+      "step": 5400
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.91660475730896,
+      "learning_rate": 1.4737579543464301e-05,
+      "loss": 0.7386,
+      "step": 5410
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9849096536636353,
+      "learning_rate": 1.4725903438612882e-05,
+      "loss": 0.7568,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 2.1076576709747314,
+      "learning_rate": 1.471422733376146e-05,
+      "loss": 0.7091,
+      "step": 5430
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8458292484283447,
+      "learning_rate": 1.4702551228910035e-05,
+      "loss": 0.7668,
+      "step": 5440
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9444243907928467,
+      "learning_rate": 1.4690875124058614e-05,
+      "loss": 0.7276,
+      "step": 5450
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.7948039174079895,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.335,
+      "eval_steps_per_second": 3.267,
+      "step": 5450
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8873387575149536,
+      "learning_rate": 1.4679199019207193e-05,
+      "loss": 0.7143,
+      "step": 5460
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.736655831336975,
+      "learning_rate": 1.4667522914355771e-05,
+      "loss": 0.8109,
+      "step": 5470
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.8737461566925049,
+      "learning_rate": 1.465584680950435e-05,
+      "loss": 0.7399,
+      "step": 5480
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.94661545753479,
+      "learning_rate": 1.4644170704652929e-05,
+      "loss": 0.8257,
+      "step": 5490
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8770204782485962,
+      "learning_rate": 1.4632494599801507e-05,
+      "loss": 0.7039,
+      "step": 5500
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.7911434173583984,
+      "eval_runtime": 0.6098,
+      "eval_samples_per_second": 16.4,
+      "eval_steps_per_second": 3.28,
+      "step": 5500
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9144564867019653,
+      "learning_rate": 1.4620818494950086e-05,
+      "loss": 0.7496,
+      "step": 5510
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8862963914871216,
+      "learning_rate": 1.4609142390098664e-05,
+      "loss": 0.7562,
+      "step": 5520
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1573410034179688,
+      "learning_rate": 1.4597466285247243e-05,
+      "loss": 0.7096,
+      "step": 5530
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.043396234512329,
+      "learning_rate": 1.458579018039582e-05,
+      "loss": 0.7314,
+      "step": 5540
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8541291952133179,
+      "learning_rate": 1.4574114075544399e-05,
+      "loss": 0.775,
+      "step": 5550
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.7938019633293152,
+      "eval_runtime": 0.6161,
+      "eval_samples_per_second": 16.232,
+      "eval_steps_per_second": 3.246,
+      "step": 5550
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7729216814041138,
+      "learning_rate": 1.4562437970692977e-05,
+      "loss": 0.7322,
+      "step": 5560
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.097247362136841,
+      "learning_rate": 1.4550761865841556e-05,
+      "loss": 0.7348,
+      "step": 5570
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9859389066696167,
+      "learning_rate": 1.4539085760990134e-05,
+      "loss": 0.7394,
+      "step": 5580
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2787526845932007,
+      "learning_rate": 1.4527409656138713e-05,
+      "loss": 0.7393,
+      "step": 5590
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.17509126663208,
+      "learning_rate": 1.4515733551287292e-05,
+      "loss": 0.7539,
+      "step": 5600
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.7940343618392944,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.34,
+      "eval_steps_per_second": 3.268,
+      "step": 5600
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9021981954574585,
+      "learning_rate": 1.450405744643587e-05,
+      "loss": 0.7399,
+      "step": 5610
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.0535192489624023,
+      "learning_rate": 1.4492381341584449e-05,
+      "loss": 0.7617,
+      "step": 5620
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.739254355430603,
+      "learning_rate": 1.4480705236733027e-05,
+      "loss": 0.7467,
+      "step": 5630
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.7759631872177124,
+      "learning_rate": 1.4469029131881604e-05,
+      "loss": 0.7224,
+      "step": 5640
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 2.1736414432525635,
+      "learning_rate": 1.4457353027030183e-05,
+      "loss": 0.7213,
+      "step": 5650
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 0.7901352047920227,
+      "eval_runtime": 0.6126,
+      "eval_samples_per_second": 16.323,
+      "eval_steps_per_second": 3.265,
+      "step": 5650
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.873023271560669,
+      "learning_rate": 1.4445676922178762e-05,
+      "loss": 0.7256,
+      "step": 5660
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.9268872737884521,
+      "learning_rate": 1.443400081732734e-05,
+      "loss": 0.7908,
+      "step": 5670
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8685206174850464,
+      "learning_rate": 1.4422324712475919e-05,
+      "loss": 0.7396,
+      "step": 5680
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7836887836456299,
+      "learning_rate": 1.4410648607624497e-05,
+      "loss": 0.7509,
+      "step": 5690
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9234646558761597,
+      "learning_rate": 1.4398972502773076e-05,
+      "loss": 0.7409,
+      "step": 5700
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.7892717123031616,
+      "eval_runtime": 0.615,
+      "eval_samples_per_second": 16.259,
+      "eval_steps_per_second": 3.252,
+      "step": 5700
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.842178463935852,
+      "learning_rate": 1.4387296397921655e-05,
+      "loss": 0.7146,
+      "step": 5710
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7722291946411133,
+      "learning_rate": 1.4375620293070233e-05,
+      "loss": 0.7371,
+      "step": 5720
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8781352043151855,
+      "learning_rate": 1.4363944188218812e-05,
+      "loss": 0.713,
+      "step": 5730
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9482977390289307,
+      "learning_rate": 1.4352268083367389e-05,
+      "loss": 0.7241,
+      "step": 5740
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1493382453918457,
+      "learning_rate": 1.4340591978515967e-05,
+      "loss": 0.7318,
+      "step": 5750
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.7901837229728699,
+      "eval_runtime": 0.6136,
+      "eval_samples_per_second": 16.297,
+      "eval_steps_per_second": 3.259,
+      "step": 5750
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.1078696250915527,
+      "learning_rate": 1.4328915873664546e-05,
+      "loss": 0.742,
+      "step": 5760
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8728898763656616,
+      "learning_rate": 1.4317239768813125e-05,
+      "loss": 0.7239,
+      "step": 5770
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0316765308380127,
+      "learning_rate": 1.4305563663961703e-05,
+      "loss": 0.7558,
+      "step": 5780
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9217534065246582,
+      "learning_rate": 1.4293887559110282e-05,
+      "loss": 0.7423,
+      "step": 5790
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.804758071899414,
+      "learning_rate": 1.428221145425886e-05,
+      "loss": 0.7897,
+      "step": 5800
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.790118932723999,
+      "eval_runtime": 0.6121,
+      "eval_samples_per_second": 16.336,
+      "eval_steps_per_second": 3.267,
+      "step": 5800
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9590510129928589,
+      "learning_rate": 1.427053534940744e-05,
+      "loss": 0.7273,
+      "step": 5810
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0691580772399902,
+      "learning_rate": 1.4258859244556018e-05,
+      "loss": 0.7269,
+      "step": 5820
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.014873504638672,
+      "learning_rate": 1.4247183139704596e-05,
+      "loss": 0.7623,
+      "step": 5830
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.0224273204803467,
+      "learning_rate": 1.4235507034853173e-05,
+      "loss": 0.7241,
+      "step": 5840
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.9337172508239746,
+      "learning_rate": 1.4223830930001752e-05,
+      "loss": 0.7362,
+      "step": 5850
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.7900481224060059,
+      "eval_runtime": 0.6135,
+      "eval_samples_per_second": 16.301,
+      "eval_steps_per_second": 3.26,
+      "step": 5850
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.7371723651885986,
+      "learning_rate": 1.421215482515033e-05,
+      "loss": 0.8025,
+      "step": 5860
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1891839504241943,
+      "learning_rate": 1.420047872029891e-05,
+      "loss": 0.7235,
+      "step": 5870
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.7742581367492676,
+      "learning_rate": 1.4188802615447488e-05,
+      "loss": 0.7027,
+      "step": 5880
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9490573406219482,
+      "learning_rate": 1.4177126510596066e-05,
+      "loss": 0.7333,
+      "step": 5890
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.034620523452759,
+      "learning_rate": 1.4165450405744645e-05,
+      "loss": 0.7302,
+      "step": 5900
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.7889585494995117,
+      "eval_runtime": 0.6121,
+      "eval_samples_per_second": 16.338,
+      "eval_steps_per_second": 3.268,
+      "step": 5900
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9658241271972656,
+      "learning_rate": 1.4153774300893224e-05,
+      "loss": 0.7412,
+      "step": 5910
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1861014366149902,
+      "learning_rate": 1.4142098196041802e-05,
+      "loss": 0.7226,
+      "step": 5920
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9467813968658447,
+      "learning_rate": 1.4130422091190381e-05,
+      "loss": 0.8106,
+      "step": 5930
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8656258583068848,
+      "learning_rate": 1.4118745986338958e-05,
+      "loss": 0.7227,
+      "step": 5940
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.9497514963150024,
+      "learning_rate": 1.4107069881487536e-05,
+      "loss": 0.7288,
+      "step": 5950
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.7929795384407043,
+      "eval_runtime": 0.6124,
+      "eval_samples_per_second": 16.328,
+      "eval_steps_per_second": 3.266,
+      "step": 5950
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8500863313674927,
+      "learning_rate": 1.4095393776636115e-05,
+      "loss": 0.737,
+      "step": 5960
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.1630983352661133,
+      "learning_rate": 1.4083717671784694e-05,
+      "loss": 0.7259,
+      "step": 5970
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8160234689712524,
+      "learning_rate": 1.4072041566933272e-05,
+      "loss": 0.7627,
+      "step": 5980
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.870741844177246,
+      "learning_rate": 1.4060365462081851e-05,
+      "loss": 0.7089,
+      "step": 5990
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0655198097229004,
+      "learning_rate": 1.404868935723043e-05,
+      "loss": 0.7195,
+      "step": 6000
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.7853182554244995,
+      "eval_runtime": 0.6171,
+      "eval_samples_per_second": 16.205,
+      "eval_steps_per_second": 3.241,
+      "step": 6000
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.170337200164795,
+      "learning_rate": 1.4037013252379008e-05,
+      "loss": 0.7494,
+      "step": 6010
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8251579999923706,
+      "learning_rate": 1.4025337147527587e-05,
+      "loss": 0.7111,
+      "step": 6020
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0090854167938232,
+      "learning_rate": 1.4013661042676165e-05,
+      "loss": 0.7038,
+      "step": 6030
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0172417163848877,
+      "learning_rate": 1.4001984937824742e-05,
+      "loss": 0.73,
+      "step": 6040
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7745598554611206,
+      "learning_rate": 1.3990308832973321e-05,
+      "loss": 0.7234,
+      "step": 6050
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.7892107963562012,
+      "eval_runtime": 0.613,
+      "eval_samples_per_second": 16.313,
+      "eval_steps_per_second": 3.263,
+      "step": 6050
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0899953842163086,
+      "learning_rate": 1.39786327281219e-05,
+      "loss": 0.7541,
+      "step": 6060
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.011955738067627,
+      "learning_rate": 1.3966956623270478e-05,
+      "loss": 0.7455,
+      "step": 6070
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.2581675052642822,
+      "learning_rate": 1.3955280518419057e-05,
+      "loss": 0.744,
+      "step": 6080
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.8642923831939697,
+      "learning_rate": 1.3943604413567635e-05,
+      "loss": 0.7834,
+      "step": 6090
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9759089946746826,
+      "learning_rate": 1.3931928308716214e-05,
+      "loss": 0.7512,
+      "step": 6100
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.7899565696716309,
+      "eval_runtime": 0.609,
+      "eval_samples_per_second": 16.421,
+      "eval_steps_per_second": 3.284,
+      "step": 6100
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.076939821243286,
+      "learning_rate": 1.3920252203864793e-05,
+      "loss": 0.7263,
+      "step": 6110
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0767054557800293,
+      "learning_rate": 1.3908576099013371e-05,
+      "loss": 0.7295,
+      "step": 6120
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9546929597854614,
+      "learning_rate": 1.389689999416195e-05,
+      "loss": 0.749,
+      "step": 6130
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9209812879562378,
+      "learning_rate": 1.3885223889310527e-05,
+      "loss": 0.7229,
+      "step": 6140
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1199162006378174,
+      "learning_rate": 1.3873547784459105e-05,
+      "loss": 0.738,
+      "step": 6150
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.788119912147522,
+      "eval_runtime": 0.6124,
+      "eval_samples_per_second": 16.328,
+      "eval_steps_per_second": 3.266,
+      "step": 6150
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9958100318908691,
+      "learning_rate": 1.3861871679607684e-05,
+      "loss": 0.73,
+      "step": 6160
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0326905250549316,
+      "learning_rate": 1.3850195574756263e-05,
+      "loss": 0.7217,
+      "step": 6170
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.943064570426941,
+      "learning_rate": 1.3838519469904841e-05,
+      "loss": 0.7195,
+      "step": 6180
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.0280864238739014,
+      "learning_rate": 1.382684336505342e-05,
+      "loss": 0.7159,
+      "step": 6190
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.7285759449005127,
+      "learning_rate": 1.3815167260201998e-05,
+      "loss": 0.7987,
+      "step": 6200
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 0.7844266295433044,
+      "eval_runtime": 0.6137,
+      "eval_samples_per_second": 16.293,
+      "eval_steps_per_second": 3.259,
+      "step": 6200
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.9906049966812134,
+      "learning_rate": 1.3803491155350577e-05,
+      "loss": 0.734,
+      "step": 6210
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.1304662227630615,
+      "learning_rate": 1.3791815050499156e-05,
+      "loss": 0.7425,
+      "step": 6220
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8910713195800781,
+      "learning_rate": 1.3780138945647734e-05,
+      "loss": 0.7158,
+      "step": 6230
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.3210713863372803,
+      "learning_rate": 1.3768462840796311e-05,
+      "loss": 0.7283,
+      "step": 6240
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9663628339767456,
+      "learning_rate": 1.375678673594489e-05,
+      "loss": 0.7746,
+      "step": 6250
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.7855362296104431,
+      "eval_runtime": 0.6099,
+      "eval_samples_per_second": 16.396,
+      "eval_steps_per_second": 3.279,
+      "step": 6250
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0221545696258545,
+      "learning_rate": 1.3745110631093468e-05,
+      "loss": 0.7254,
+      "step": 6260
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0435986518859863,
+      "learning_rate": 1.3733434526242047e-05,
+      "loss": 0.73,
+      "step": 6270
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9049526453018188,
+      "learning_rate": 1.3721758421390626e-05,
+      "loss": 0.7134,
+      "step": 6280
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0307445526123047,
+      "learning_rate": 1.3710082316539204e-05,
+      "loss": 0.7113,
+      "step": 6290
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.973630666732788,
+      "learning_rate": 1.3698406211687783e-05,
+      "loss": 0.7323,
+      "step": 6300
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.7841699719429016,
+      "eval_runtime": 0.6089,
+      "eval_samples_per_second": 16.423,
+      "eval_steps_per_second": 3.285,
+      "step": 6300
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.013610363006592,
+      "learning_rate": 1.3686730106836362e-05,
+      "loss": 0.72,
+      "step": 6310
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8538728952407837,
+      "learning_rate": 1.367505400198494e-05,
+      "loss": 0.7493,
+      "step": 6320
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1160643100738525,
+      "learning_rate": 1.3663377897133519e-05,
+      "loss": 0.7627,
+      "step": 6330
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0380024909973145,
+      "learning_rate": 1.3651701792282096e-05,
+      "loss": 0.7531,
+      "step": 6340
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.109076499938965,
+      "learning_rate": 1.3640025687430674e-05,
+      "loss": 0.7749,
+      "step": 6350
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.7839874029159546,
+      "eval_runtime": 0.6134,
+      "eval_samples_per_second": 16.303,
+      "eval_steps_per_second": 3.261,
+      "step": 6350
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0254173278808594,
+      "learning_rate": 1.3628349582579253e-05,
+      "loss": 0.7456,
+      "step": 6360
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.9714245796203613,
+      "learning_rate": 1.3616673477727832e-05,
+      "loss": 0.7089,
+      "step": 6370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.1261284351348877,
+      "learning_rate": 1.360499737287641e-05,
+      "loss": 0.7087,
+      "step": 6380
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.0249807834625244,
+      "learning_rate": 1.3593321268024989e-05,
+      "loss": 0.7406,
+      "step": 6390
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.8258070945739746,
+      "learning_rate": 1.3581645163173567e-05,
+      "loss": 0.7401,
+      "step": 6400
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.790998101234436,
+      "eval_runtime": 0.6095,
+      "eval_samples_per_second": 16.406,
+      "eval_steps_per_second": 3.281,
+      "step": 6400
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.063620090484619,
+      "learning_rate": 1.3569969058322146e-05,
+      "loss": 0.7682,
+      "step": 6410
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.921318531036377,
+      "learning_rate": 1.3558292953470725e-05,
+      "loss": 0.7368,
+      "step": 6420
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7165294885635376,
+      "learning_rate": 1.3546616848619303e-05,
+      "loss": 0.7652,
+      "step": 6430
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.026031494140625,
+      "learning_rate": 1.3534940743767878e-05,
+      "loss": 0.7233,
+      "step": 6440
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.903433918952942,
+      "learning_rate": 1.3523264638916457e-05,
+      "loss": 0.7232,
+      "step": 6450
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.7895861864089966,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.334,
+      "eval_steps_per_second": 3.267,
+      "step": 6450
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8935394287109375,
+      "learning_rate": 1.3511588534065037e-05,
+      "loss": 0.7347,
+      "step": 6460
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.125434398651123,
+      "learning_rate": 1.3499912429213616e-05,
+      "loss": 0.7253,
+      "step": 6470
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8065615892410278,
+      "learning_rate": 1.3488236324362195e-05,
+      "loss": 0.7324,
+      "step": 6480
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8938298225402832,
+      "learning_rate": 1.3476560219510773e-05,
+      "loss": 0.7415,
+      "step": 6490
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.2024078369140625,
+      "learning_rate": 1.3464884114659352e-05,
+      "loss": 0.7461,
+      "step": 6500
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.7891114950180054,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.342,
+      "eval_steps_per_second": 3.268,
+      "step": 6500
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8333817720413208,
+      "learning_rate": 1.345320800980793e-05,
+      "loss": 0.7762,
+      "step": 6510
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9520988464355469,
+      "learning_rate": 1.3441531904956509e-05,
+      "loss": 0.7137,
+      "step": 6520
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8310378789901733,
+      "learning_rate": 1.3429855800105084e-05,
+      "loss": 0.7762,
+      "step": 6530
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8594748973846436,
+      "learning_rate": 1.3418179695253663e-05,
+      "loss": 0.7085,
+      "step": 6540
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.03914475440979,
+      "learning_rate": 1.3406503590402242e-05,
+      "loss": 0.7376,
+      "step": 6550
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.7881189584732056,
+      "eval_runtime": 0.61,
+      "eval_samples_per_second": 16.392,
+      "eval_steps_per_second": 3.278,
+      "step": 6550
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 2.0291972160339355,
+      "learning_rate": 1.339482748555082e-05,
+      "loss": 0.7105,
+      "step": 6560
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7613359689712524,
+      "learning_rate": 1.3383151380699399e-05,
+      "loss": 0.7173,
+      "step": 6570
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.9951395988464355,
+      "learning_rate": 1.3371475275847977e-05,
+      "loss": 0.7253,
+      "step": 6580
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9363685846328735,
+      "learning_rate": 1.3359799170996558e-05,
+      "loss": 0.7003,
+      "step": 6590
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.873198390007019,
+      "learning_rate": 1.3348123066145136e-05,
+      "loss": 0.7453,
+      "step": 6600
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.7913551330566406,
+      "eval_runtime": 0.6133,
+      "eval_samples_per_second": 16.305,
+      "eval_steps_per_second": 3.261,
+      "step": 6600
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.997471570968628,
+      "learning_rate": 1.3336446961293715e-05,
+      "loss": 0.831,
+      "step": 6610
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1266872882843018,
+      "learning_rate": 1.3325938466927435e-05,
+      "loss": 0.741,
+      "step": 6620
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.055396795272827,
+      "learning_rate": 1.3314262362076014e-05,
+      "loss": 0.739,
+      "step": 6630
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9011343717575073,
+      "learning_rate": 1.330258625722459e-05,
+      "loss": 0.7476,
+      "step": 6640
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.9498547315597534,
+      "learning_rate": 1.329091015237317e-05,
+      "loss": 0.7635,
+      "step": 6650
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.7874776721000671,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.369,
+      "eval_steps_per_second": 3.274,
+      "step": 6650
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.5744065046310425,
+      "learning_rate": 1.3279234047521748e-05,
+      "loss": 0.7198,
+      "step": 6660
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.6557278633117676,
+      "learning_rate": 1.3267557942670327e-05,
+      "loss": 0.7451,
+      "step": 6670
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8324756622314453,
+      "learning_rate": 1.3255881837818905e-05,
+      "loss": 0.7203,
+      "step": 6680
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.082050323486328,
+      "learning_rate": 1.3244205732967484e-05,
+      "loss": 0.711,
+      "step": 6690
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1044728755950928,
+      "learning_rate": 1.3232529628116062e-05,
+      "loss": 0.6869,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.7892599701881409,
+      "eval_runtime": 0.6104,
+      "eval_samples_per_second": 16.381,
+      "eval_steps_per_second": 3.276,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.0905628204345703,
+      "learning_rate": 1.3220853523264641e-05,
+      "loss": 0.7434,
+      "step": 6710
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8691200017929077,
+      "learning_rate": 1.320917741841322e-05,
+      "loss": 0.7845,
+      "step": 6720
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.050682306289673,
+      "learning_rate": 1.3197501313561797e-05,
+      "loss": 0.7415,
+      "step": 6730
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.00516414642334,
+      "learning_rate": 1.3185825208710375e-05,
+      "loss": 0.709,
+      "step": 6740
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 2.1088707447052,
+      "learning_rate": 1.3174149103858954e-05,
+      "loss": 0.7141,
+      "step": 6750
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.7866424322128296,
+      "eval_runtime": 0.6141,
+      "eval_samples_per_second": 16.283,
+      "eval_steps_per_second": 3.257,
+      "step": 6750
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.8866454362869263,
+      "learning_rate": 1.3162472999007532e-05,
+      "loss": 0.7164,
+      "step": 6760
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1154236793518066,
+      "learning_rate": 1.3150796894156111e-05,
+      "loss": 0.7723,
+      "step": 6770
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0821170806884766,
+      "learning_rate": 1.313912078930469e-05,
+      "loss": 0.7114,
+      "step": 6780
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.1243057250976562,
+      "learning_rate": 1.3127444684453268e-05,
+      "loss": 0.7143,
+      "step": 6790
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.6204321384429932,
+      "learning_rate": 1.3115768579601847e-05,
+      "loss": 0.771,
+      "step": 6800
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.7841218113899231,
+      "eval_runtime": 0.6096,
+      "eval_samples_per_second": 16.405,
+      "eval_steps_per_second": 3.281,
+      "step": 6800
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.185901165008545,
+      "learning_rate": 1.3104092474750426e-05,
+      "loss": 0.7451,
+      "step": 6810
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.8085384368896484,
+      "learning_rate": 1.3092416369899004e-05,
+      "loss": 0.722,
+      "step": 6820
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.844062089920044,
+      "learning_rate": 1.308074026504758e-05,
+      "loss": 0.7217,
+      "step": 6830
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9088714122772217,
+      "learning_rate": 1.3069064160196158e-05,
+      "loss": 0.7174,
+      "step": 6840
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0905704498291016,
+      "learning_rate": 1.3057388055344737e-05,
+      "loss": 0.7182,
+      "step": 6850
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.788231372833252,
+      "eval_runtime": 0.6096,
+      "eval_samples_per_second": 16.404,
+      "eval_steps_per_second": 3.281,
+      "step": 6850
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9573746919631958,
+      "learning_rate": 1.3045711950493317e-05,
+      "loss": 0.735,
+      "step": 6860
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.011361598968506,
+      "learning_rate": 1.3034035845641896e-05,
+      "loss": 0.7299,
+      "step": 6870
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.025827407836914,
+      "learning_rate": 1.3022359740790474e-05,
+      "loss": 0.7338,
+      "step": 6880
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0000534057617188,
+      "learning_rate": 1.3010683635939053e-05,
+      "loss": 0.7616,
+      "step": 6890
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.99836266040802,
+      "learning_rate": 1.2999007531087631e-05,
+      "loss": 0.739,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 0.7846401929855347,
+      "eval_runtime": 0.6107,
+      "eval_samples_per_second": 16.374,
+      "eval_steps_per_second": 3.275,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.046915292739868,
+      "learning_rate": 1.298733142623621e-05,
+      "loss": 0.7544,
+      "step": 6910
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9967786073684692,
+      "learning_rate": 1.2975655321384789e-05,
+      "loss": 0.6849,
+      "step": 6920
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 2.0040202140808105,
+      "learning_rate": 1.2963979216533364e-05,
+      "loss": 0.6998,
+      "step": 6930
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.9282623529434204,
+      "learning_rate": 1.2952303111681942e-05,
+      "loss": 0.7291,
+      "step": 6940
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2685484886169434,
+      "learning_rate": 1.2940627006830521e-05,
+      "loss": 0.7345,
+      "step": 6950
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.7853933572769165,
+      "eval_runtime": 0.6132,
+      "eval_samples_per_second": 16.309,
+      "eval_steps_per_second": 3.262,
+      "step": 6950
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9175786972045898,
+      "learning_rate": 1.29289509019791e-05,
+      "loss": 0.7174,
+      "step": 6960
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.5591322183609009,
+      "learning_rate": 1.2917274797127678e-05,
+      "loss": 0.7761,
+      "step": 6970
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9000513553619385,
+      "learning_rate": 1.2905598692276257e-05,
+      "loss": 0.7298,
+      "step": 6980
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2343156337738037,
+      "learning_rate": 1.2893922587424837e-05,
+      "loss": 0.702,
+      "step": 6990
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9918251037597656,
+      "learning_rate": 1.2882246482573416e-05,
+      "loss": 0.7697,
+      "step": 7000
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.788036048412323,
+      "eval_runtime": 0.6139,
+      "eval_samples_per_second": 16.29,
+      "eval_steps_per_second": 3.258,
+      "step": 7000
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.184072971343994,
+      "learning_rate": 1.2870570377721994e-05,
+      "loss": 0.7521,
+      "step": 7010
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.884344220161438,
+      "learning_rate": 1.2858894272870573e-05,
+      "loss": 0.7572,
+      "step": 7020
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.9404270648956299,
+      "learning_rate": 1.2847218168019148e-05,
+      "loss": 0.7627,
+      "step": 7030
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1148927211761475,
+      "learning_rate": 1.2835542063167727e-05,
+      "loss": 0.7198,
+      "step": 7040
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.055178165435791,
+      "learning_rate": 1.2823865958316306e-05,
+      "loss": 0.743,
+      "step": 7050
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.7888015508651733,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.363,
+      "eval_steps_per_second": 3.273,
+      "step": 7050
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.1209557056427,
+      "learning_rate": 1.2812189853464884e-05,
+      "loss": 0.7082,
+      "step": 7060
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0028674602508545,
+      "learning_rate": 1.2800513748613463e-05,
+      "loss": 0.7489,
+      "step": 7070
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.436264991760254,
+      "learning_rate": 1.2788837643762041e-05,
+      "loss": 0.7244,
+      "step": 7080
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.0268330574035645,
+      "learning_rate": 1.277716153891062e-05,
+      "loss": 0.7051,
+      "step": 7090
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.153724431991577,
+      "learning_rate": 1.2765485434059199e-05,
+      "loss": 0.7445,
+      "step": 7100
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.7899669408798218,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.339,
+      "eval_steps_per_second": 3.268,
+      "step": 7100
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.8381693363189697,
+      "learning_rate": 1.2753809329207777e-05,
+      "loss": 0.7353,
+      "step": 7110
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.93001389503479,
+      "learning_rate": 1.2742133224356358e-05,
+      "loss": 0.7392,
+      "step": 7120
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.994533896446228,
+      "learning_rate": 1.2730457119504933e-05,
+      "loss": 0.7749,
+      "step": 7130
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.035572052001953,
+      "learning_rate": 1.2718781014653511e-05,
+      "loss": 0.7314,
+      "step": 7140
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0678281784057617,
+      "learning_rate": 1.270710490980209e-05,
+      "loss": 0.7506,
+      "step": 7150
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.786938488483429,
+      "eval_runtime": 0.6103,
+      "eval_samples_per_second": 16.385,
+      "eval_steps_per_second": 3.277,
+      "step": 7150
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.145230293273926,
+      "learning_rate": 1.2695428804950669e-05,
+      "loss": 0.7245,
+      "step": 7160
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.047818899154663,
+      "learning_rate": 1.2683752700099247e-05,
+      "loss": 0.7195,
+      "step": 7170
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.8724055290222168,
+      "learning_rate": 1.2672076595247826e-05,
+      "loss": 0.7289,
+      "step": 7180
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.784937858581543,
+      "learning_rate": 1.2660400490396404e-05,
+      "loss": 0.766,
+      "step": 7190
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0545976161956787,
+      "learning_rate": 1.2648724385544983e-05,
+      "loss": 0.7274,
+      "step": 7200
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.7896177768707275,
+      "eval_runtime": 0.6129,
+      "eval_samples_per_second": 16.315,
+      "eval_steps_per_second": 3.263,
+      "step": 7200
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.06841778755188,
+      "learning_rate": 1.2637048280693562e-05,
+      "loss": 0.7414,
+      "step": 7210
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9431219100952148,
+      "learning_rate": 1.262537217584214e-05,
+      "loss": 0.6723,
+      "step": 7220
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.1300158500671387,
+      "learning_rate": 1.2613696070990717e-05,
+      "loss": 0.7215,
+      "step": 7230
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9715406894683838,
+      "learning_rate": 1.2602019966139296e-05,
+      "loss": 0.7388,
+      "step": 7240
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9708106517791748,
+      "learning_rate": 1.2590343861287874e-05,
+      "loss": 0.7188,
+      "step": 7250
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.7906878590583801,
+      "eval_runtime": 0.6087,
+      "eval_samples_per_second": 16.427,
+      "eval_steps_per_second": 3.285,
+      "step": 7250
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.333930730819702,
+      "learning_rate": 1.2578667756436453e-05,
+      "loss": 0.7351,
+      "step": 7260
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0958805084228516,
+      "learning_rate": 1.2566991651585032e-05,
+      "loss": 0.7259,
+      "step": 7270
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9661331176757812,
+      "learning_rate": 1.255531554673361e-05,
+      "loss": 0.7584,
+      "step": 7280
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0475594997406006,
+      "learning_rate": 1.2543639441882189e-05,
+      "loss": 0.7352,
+      "step": 7290
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.0641934871673584,
+      "learning_rate": 1.2531963337030768e-05,
+      "loss": 0.7333,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.7860339879989624,
+      "eval_runtime": 0.6064,
+      "eval_samples_per_second": 16.491,
+      "eval_steps_per_second": 3.298,
+      "step": 7300
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.7937456369400024,
+      "learning_rate": 1.2520287232179346e-05,
+      "loss": 0.7767,
+      "step": 7310
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2383811473846436,
+      "learning_rate": 1.2508611127327925e-05,
+      "loss": 0.7361,
+      "step": 7320
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.103365182876587,
+      "learning_rate": 1.2496935022476502e-05,
+      "loss": 0.7024,
+      "step": 7330
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.005986452102661,
+      "learning_rate": 1.248525891762508e-05,
+      "loss": 0.7215,
+      "step": 7340
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8555827140808105,
+      "learning_rate": 1.2473582812773659e-05,
+      "loss": 0.7329,
+      "step": 7350
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 0.7881810069084167,
+      "eval_runtime": 0.6118,
+      "eval_samples_per_second": 16.344,
+      "eval_steps_per_second": 3.269,
+      "step": 7350
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0253376960754395,
+      "learning_rate": 1.2461906707922238e-05,
+      "loss": 0.7572,
+      "step": 7360
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.927605390548706,
+      "learning_rate": 1.2450230603070816e-05,
+      "loss": 0.7422,
+      "step": 7370
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9607007503509521,
+      "learning_rate": 1.2438554498219395e-05,
+      "loss": 0.6981,
+      "step": 7380
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.2108051776885986,
+      "learning_rate": 1.2426878393367973e-05,
+      "loss": 0.7092,
+      "step": 7390
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.8442574739456177,
+      "learning_rate": 1.2415202288516552e-05,
+      "loss": 0.7533,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 0.789718747138977,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.334,
+      "eval_steps_per_second": 3.267,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.97207772731781,
+      "learning_rate": 1.240352618366513e-05,
+      "loss": 0.7018,
+      "step": 7410
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.012571096420288,
+      "learning_rate": 1.239185007881371e-05,
+      "loss": 0.7102,
+      "step": 7420
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.075765371322632,
+      "learning_rate": 1.2380173973962286e-05,
+      "loss": 0.7135,
+      "step": 7430
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9438190460205078,
+      "learning_rate": 1.2368497869110865e-05,
+      "loss": 0.7301,
+      "step": 7440
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.9945006370544434,
+      "learning_rate": 1.2356821764259443e-05,
+      "loss": 0.6878,
+      "step": 7450
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 0.7870052456855774,
+      "eval_runtime": 0.6126,
+      "eval_samples_per_second": 16.324,
+      "eval_steps_per_second": 3.265,
+      "step": 7450
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.162912130355835,
+      "learning_rate": 1.2345145659408022e-05,
+      "loss": 0.7047,
+      "step": 7460
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.1406214237213135,
+      "learning_rate": 1.23334695545566e-05,
+      "loss": 0.7382,
+      "step": 7470
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 2.0262863636016846,
+      "learning_rate": 1.232179344970518e-05,
+      "loss": 0.7371,
+      "step": 7480
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.917303442955017,
+      "learning_rate": 1.2310117344853758e-05,
+      "loss": 0.7096,
+      "step": 7490
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9172418117523193,
+      "learning_rate": 1.2298441240002336e-05,
+      "loss": 0.7193,
+      "step": 7500
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.7885881066322327,
+      "eval_runtime": 0.6098,
+      "eval_samples_per_second": 16.398,
+      "eval_steps_per_second": 3.28,
+      "step": 7500
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.21590518951416,
+      "learning_rate": 1.2286765135150915e-05,
+      "loss": 0.7253,
+      "step": 7510
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.076140880584717,
+      "learning_rate": 1.2275089030299494e-05,
+      "loss": 0.7808,
+      "step": 7520
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9610168933868408,
+      "learning_rate": 1.226341292544807e-05,
+      "loss": 0.7482,
+      "step": 7530
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0242979526519775,
+      "learning_rate": 1.225173682059665e-05,
+      "loss": 0.7439,
+      "step": 7540
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1023123264312744,
+      "learning_rate": 1.2240060715745228e-05,
+      "loss": 0.7436,
+      "step": 7550
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.7864001393318176,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.392,
+      "eval_steps_per_second": 3.278,
+      "step": 7550
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.149568796157837,
+      "learning_rate": 1.2228384610893806e-05,
+      "loss": 0.7247,
+      "step": 7560
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.146484375,
+      "learning_rate": 1.2216708506042385e-05,
+      "loss": 0.71,
+      "step": 7570
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0798139572143555,
+      "learning_rate": 1.2205032401190964e-05,
+      "loss": 0.6941,
+      "step": 7580
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.887951374053955,
+      "learning_rate": 1.2193356296339542e-05,
+      "loss": 0.7358,
+      "step": 7590
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.3733999729156494,
+      "learning_rate": 1.2181680191488121e-05,
+      "loss": 0.7234,
+      "step": 7600
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.7819582223892212,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.373,
+      "eval_steps_per_second": 3.275,
+      "step": 7600
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.094634771347046,
+      "learning_rate": 1.21700040866367e-05,
+      "loss": 0.7588,
+      "step": 7610
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.7685121297836304,
+      "learning_rate": 1.2158327981785278e-05,
+      "loss": 0.742,
+      "step": 7620
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.0696113109588623,
+      "learning_rate": 1.2146651876933855e-05,
+      "loss": 0.701,
+      "step": 7630
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.2344696521759033,
+      "learning_rate": 1.2134975772082434e-05,
+      "loss": 0.698,
+      "step": 7640
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.9609960317611694,
+      "learning_rate": 1.2123299667231012e-05,
+      "loss": 0.7269,
+      "step": 7650
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.785152792930603,
+      "eval_runtime": 0.6116,
+      "eval_samples_per_second": 16.351,
+      "eval_steps_per_second": 3.27,
+      "step": 7650
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.8363230228424072,
+      "learning_rate": 1.2111623562379591e-05,
+      "loss": 0.7688,
+      "step": 7660
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.3206257820129395,
+      "learning_rate": 1.209994745752817e-05,
+      "loss": 0.766,
+      "step": 7670
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1884496212005615,
+      "learning_rate": 1.2088271352676748e-05,
+      "loss": 0.7994,
+      "step": 7680
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.021320104598999,
+      "learning_rate": 1.2076595247825327e-05,
+      "loss": 0.6964,
+      "step": 7690
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.080582618713379,
+      "learning_rate": 1.2064919142973905e-05,
+      "loss": 0.7179,
+      "step": 7700
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.7822645306587219,
+      "eval_runtime": 0.607,
+      "eval_samples_per_second": 16.475,
+      "eval_steps_per_second": 3.295,
+      "step": 7700
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.065843105316162,
+      "learning_rate": 1.2053243038122484e-05,
+      "loss": 0.7115,
+      "step": 7710
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.199220657348633,
+      "learning_rate": 1.2041566933271063e-05,
+      "loss": 0.7429,
+      "step": 7720
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9715982675552368,
+      "learning_rate": 1.202989082841964e-05,
+      "loss": 0.7116,
+      "step": 7730
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.04953670501709,
+      "learning_rate": 1.2018214723568218e-05,
+      "loss": 0.7576,
+      "step": 7740
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.195939064025879,
+      "learning_rate": 1.2006538618716797e-05,
+      "loss": 0.736,
+      "step": 7750
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.7818417549133301,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.339,
+      "eval_steps_per_second": 3.268,
+      "step": 7750
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.263132095336914,
+      "learning_rate": 1.1994862513865375e-05,
+      "loss": 0.7495,
+      "step": 7760
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.170241594314575,
+      "learning_rate": 1.1983186409013954e-05,
+      "loss": 0.7299,
+      "step": 7770
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9383941888809204,
+      "learning_rate": 1.1971510304162533e-05,
+      "loss": 0.7116,
+      "step": 7780
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.9345946311950684,
+      "learning_rate": 1.1959834199311111e-05,
+      "loss": 0.7405,
+      "step": 7790
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0908350944519043,
+      "learning_rate": 1.194815809445969e-05,
+      "loss": 0.7537,
+      "step": 7800
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 0.7828075885772705,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.358,
+      "eval_steps_per_second": 3.272,
+      "step": 7800
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.0219149589538574,
+      "learning_rate": 1.1936481989608269e-05,
+      "loss": 0.7706,
+      "step": 7810
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1752734184265137,
+      "learning_rate": 1.1924805884756847e-05,
+      "loss": 0.7498,
+      "step": 7820
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1564393043518066,
+      "learning_rate": 1.1913129779905424e-05,
+      "loss": 0.7367,
+      "step": 7830
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.1459696292877197,
+      "learning_rate": 1.1901453675054003e-05,
+      "loss": 0.727,
+      "step": 7840
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7214521169662476,
+      "learning_rate": 1.1889777570202581e-05,
+      "loss": 0.7245,
+      "step": 7850
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.7822232246398926,
+      "eval_runtime": 0.609,
+      "eval_samples_per_second": 16.42,
+      "eval_steps_per_second": 3.284,
+      "step": 7850
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1773509979248047,
+      "learning_rate": 1.187810146535116e-05,
+      "loss": 0.7068,
+      "step": 7860
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0010464191436768,
+      "learning_rate": 1.1866425360499739e-05,
+      "loss": 0.7189,
+      "step": 7870
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1782948970794678,
+      "learning_rate": 1.1854749255648317e-05,
+      "loss": 0.7823,
+      "step": 7880
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1901357173919678,
+      "learning_rate": 1.1843073150796896e-05,
+      "loss": 0.7282,
+      "step": 7890
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.964917778968811,
+      "learning_rate": 1.1831397045945474e-05,
+      "loss": 0.7746,
+      "step": 7900
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.7793163061141968,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.414,
+      "eval_steps_per_second": 3.283,
+      "step": 7900
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4999061822891235,
+      "learning_rate": 1.1819720941094053e-05,
+      "loss": 0.7854,
+      "step": 7910
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1158761978149414,
+      "learning_rate": 1.1808044836242632e-05,
+      "loss": 0.7038,
+      "step": 7920
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1527624130249023,
+      "learning_rate": 1.1796368731391209e-05,
+      "loss": 0.7297,
+      "step": 7930
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.1434409618377686,
+      "learning_rate": 1.1784692626539787e-05,
+      "loss": 0.7407,
+      "step": 7940
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7658051252365112,
+      "learning_rate": 1.1773016521688366e-05,
+      "loss": 0.7192,
+      "step": 7950
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.7799476385116577,
+      "eval_runtime": 0.6082,
+      "eval_samples_per_second": 16.443,
+      "eval_steps_per_second": 3.289,
+      "step": 7950
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9787997007369995,
+      "learning_rate": 1.1761340416836944e-05,
+      "loss": 0.6973,
+      "step": 7960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.075730562210083,
+      "learning_rate": 1.1749664311985523e-05,
+      "loss": 0.7455,
+      "step": 7970
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.944159746170044,
+      "learning_rate": 1.1737988207134102e-05,
+      "loss": 0.7154,
+      "step": 7980
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 2.0847811698913574,
+      "learning_rate": 1.172631210228268e-05,
+      "loss": 0.6934,
+      "step": 7990
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7052133083343506,
+      "learning_rate": 1.1714635997431259e-05,
+      "loss": 0.7433,
+      "step": 8000
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.7797889709472656,
+      "eval_runtime": 0.6088,
+      "eval_samples_per_second": 16.426,
+      "eval_steps_per_second": 3.285,
+      "step": 8000
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.9447972774505615,
+      "learning_rate": 1.1702959892579837e-05,
+      "loss": 0.7161,
+      "step": 8010
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.974008560180664,
+      "learning_rate": 1.1691283787728416e-05,
+      "loss": 0.6963,
+      "step": 8020
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9179285764694214,
+      "learning_rate": 1.1679607682876993e-05,
+      "loss": 0.6947,
+      "step": 8030
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9631364345550537,
+      "learning_rate": 1.1667931578025572e-05,
+      "loss": 0.7412,
+      "step": 8040
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7791589498519897,
+      "learning_rate": 1.165625547317415e-05,
+      "loss": 0.7309,
+      "step": 8050
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.7787421345710754,
+      "eval_runtime": 0.611,
+      "eval_samples_per_second": 16.366,
+      "eval_steps_per_second": 3.273,
+      "step": 8050
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.0084519386291504,
+      "learning_rate": 1.1644579368322729e-05,
+      "loss": 0.749,
+      "step": 8060
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.290315866470337,
+      "learning_rate": 1.1632903263471307e-05,
+      "loss": 0.7794,
+      "step": 8070
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.024437665939331,
+      "learning_rate": 1.1621227158619886e-05,
+      "loss": 0.7345,
+      "step": 8080
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1918177604675293,
+      "learning_rate": 1.1609551053768465e-05,
+      "loss": 0.7165,
+      "step": 8090
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2443952560424805,
+      "learning_rate": 1.1597874948917043e-05,
+      "loss": 0.7404,
+      "step": 8100
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.7802385091781616,
+      "eval_runtime": 0.613,
+      "eval_samples_per_second": 16.312,
+      "eval_steps_per_second": 3.262,
+      "step": 8100
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.1513216495513916,
+      "learning_rate": 1.1586198844065622e-05,
+      "loss": 0.7336,
+      "step": 8110
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.109541654586792,
+      "learning_rate": 1.15745227392142e-05,
+      "loss": 0.706,
+      "step": 8120
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2077205181121826,
+      "learning_rate": 1.1562846634362776e-05,
+      "loss": 0.7308,
+      "step": 8130
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.779334306716919,
+      "learning_rate": 1.1551170529511354e-05,
+      "loss": 0.7563,
+      "step": 8140
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9262919425964355,
+      "learning_rate": 1.1539494424659933e-05,
+      "loss": 0.723,
+      "step": 8150
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.7818833589553833,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.365,
+      "eval_steps_per_second": 3.273,
+      "step": 8150
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.16867995262146,
+      "learning_rate": 1.1527818319808513e-05,
+      "loss": 0.7291,
+      "step": 8160
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.2457830905914307,
+      "learning_rate": 1.1516142214957092e-05,
+      "loss": 0.7187,
+      "step": 8170
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.240046739578247,
+      "learning_rate": 1.150446611010567e-05,
+      "loss": 0.7555,
+      "step": 8180
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.9781808853149414,
+      "learning_rate": 1.149279000525425e-05,
+      "loss": 0.7115,
+      "step": 8190
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8225785493850708,
+      "learning_rate": 1.1481113900402828e-05,
+      "loss": 0.7913,
+      "step": 8200
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 0.7828117609024048,
+      "eval_runtime": 0.6117,
+      "eval_samples_per_second": 16.348,
+      "eval_steps_per_second": 3.27,
+      "step": 8200
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1072356700897217,
+      "learning_rate": 1.1469437795551406e-05,
+      "loss": 0.7369,
+      "step": 8210
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.001088857650757,
+      "learning_rate": 1.1457761690699985e-05,
+      "loss": 0.7123,
+      "step": 8220
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.240983486175537,
+      "learning_rate": 1.144608558584856e-05,
+      "loss": 0.7296,
+      "step": 8230
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1014692783355713,
+      "learning_rate": 1.1434409480997139e-05,
+      "loss": 0.7295,
+      "step": 8240
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.380607843399048,
+      "learning_rate": 1.1422733376145717e-05,
+      "loss": 0.7159,
+      "step": 8250
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.7802363038063049,
+      "eval_runtime": 0.606,
+      "eval_samples_per_second": 16.502,
+      "eval_steps_per_second": 3.3,
+      "step": 8250
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2307772636413574,
+      "learning_rate": 1.1411057271294296e-05,
+      "loss": 0.736,
+      "step": 8260
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1194093227386475,
+      "learning_rate": 1.1399381166442875e-05,
+      "loss": 0.7007,
+      "step": 8270
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.1304867267608643,
+      "learning_rate": 1.1387705061591453e-05,
+      "loss": 0.6997,
+      "step": 8280
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.2611472606658936,
+      "learning_rate": 1.1376028956740034e-05,
+      "loss": 0.7356,
+      "step": 8290
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8811869621276855,
+      "learning_rate": 1.1364352851888612e-05,
+      "loss": 0.7473,
+      "step": 8300
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.7784644365310669,
+      "eval_runtime": 0.6095,
+      "eval_samples_per_second": 16.408,
+      "eval_steps_per_second": 3.282,
+      "step": 8300
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9575659036636353,
+      "learning_rate": 1.1352676747037191e-05,
+      "loss": 0.7263,
+      "step": 8310
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.259768486022949,
+      "learning_rate": 1.134100064218577e-05,
+      "loss": 0.7721,
+      "step": 8320
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9965535402297974,
+      "learning_rate": 1.1329324537334345e-05,
+      "loss": 0.7318,
+      "step": 8330
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.094132423400879,
+      "learning_rate": 1.1317648432482923e-05,
+      "loss": 0.7386,
+      "step": 8340
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.866145133972168,
+      "learning_rate": 1.1305972327631502e-05,
+      "loss": 0.747,
+      "step": 8350
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.776362419128418,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.371,
+      "eval_steps_per_second": 3.274,
+      "step": 8350
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.9977974891662598,
+      "learning_rate": 1.129429622278008e-05,
+      "loss": 0.7402,
+      "step": 8360
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.8253422975540161,
+      "learning_rate": 1.128262011792866e-05,
+      "loss": 0.7311,
+      "step": 8370
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 2.260749578475952,
+      "learning_rate": 1.1270944013077238e-05,
+      "loss": 0.7144,
+      "step": 8380
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.2299959659576416,
+      "learning_rate": 1.1259267908225816e-05,
+      "loss": 0.6983,
+      "step": 8390
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.166328191757202,
+      "learning_rate": 1.1247591803374395e-05,
+      "loss": 0.732,
+      "step": 8400
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.7773420214653015,
+      "eval_runtime": 0.6103,
+      "eval_samples_per_second": 16.385,
+      "eval_steps_per_second": 3.277,
+      "step": 8400
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.9604012966156006,
+      "learning_rate": 1.1235915698522974e-05,
+      "loss": 0.714,
+      "step": 8410
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.897533893585205,
+      "learning_rate": 1.1224239593671554e-05,
+      "loss": 0.7156,
+      "step": 8420
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.050827980041504,
+      "learning_rate": 1.121256348882013e-05,
+      "loss": 0.732,
+      "step": 8430
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0553033351898193,
+      "learning_rate": 1.1200887383968708e-05,
+      "loss": 0.7212,
+      "step": 8440
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.159214735031128,
+      "learning_rate": 1.1189211279117286e-05,
+      "loss": 0.7261,
+      "step": 8450
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.7806721925735474,
+      "eval_runtime": 0.6085,
+      "eval_samples_per_second": 16.435,
+      "eval_steps_per_second": 3.287,
+      "step": 8450
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.4524309635162354,
+      "learning_rate": 1.1177535174265865e-05,
+      "loss": 0.7221,
+      "step": 8460
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0922157764434814,
+      "learning_rate": 1.1165859069414444e-05,
+      "loss": 0.7363,
+      "step": 8470
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.8383965492248535,
+      "learning_rate": 1.1154182964563022e-05,
+      "loss": 0.6974,
+      "step": 8480
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1609091758728027,
+      "learning_rate": 1.1142506859711601e-05,
+      "loss": 0.7208,
+      "step": 8490
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.259551525115967,
+      "learning_rate": 1.113083075486018e-05,
+      "loss": 0.7149,
+      "step": 8500
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.7764089703559875,
+      "eval_runtime": 0.6132,
+      "eval_samples_per_second": 16.307,
+      "eval_steps_per_second": 3.261,
+      "step": 8500
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.224802255630493,
+      "learning_rate": 1.1119154650008758e-05,
+      "loss": 0.7052,
+      "step": 8510
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0728962421417236,
+      "learning_rate": 1.1107478545157337e-05,
+      "loss": 0.7076,
+      "step": 8520
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.3184080123901367,
+      "learning_rate": 1.1095802440305914e-05,
+      "loss": 0.7094,
+      "step": 8530
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.207123279571533,
+      "learning_rate": 1.1084126335454492e-05,
+      "loss": 0.717,
+      "step": 8540
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.0468602180480957,
+      "learning_rate": 1.1072450230603071e-05,
+      "loss": 0.7393,
+      "step": 8550
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.780467689037323,
+      "eval_runtime": 0.6116,
+      "eval_samples_per_second": 16.349,
+      "eval_steps_per_second": 3.27,
+      "step": 8550
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.1516823768615723,
+      "learning_rate": 1.106077412575165e-05,
+      "loss": 0.7371,
+      "step": 8560
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7321467399597168,
+      "learning_rate": 1.1049098020900228e-05,
+      "loss": 0.7297,
+      "step": 8570
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9031447172164917,
+      "learning_rate": 1.1037421916048807e-05,
+      "loss": 0.7179,
+      "step": 8580
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9976637363433838,
+      "learning_rate": 1.1025745811197385e-05,
+      "loss": 0.7192,
+      "step": 8590
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.119530200958252,
+      "learning_rate": 1.1014069706345964e-05,
+      "loss": 0.7024,
+      "step": 8600
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.7828971743583679,
+      "eval_runtime": 0.6121,
+      "eval_samples_per_second": 16.337,
+      "eval_steps_per_second": 3.267,
+      "step": 8600
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.016681432723999,
+      "learning_rate": 1.1002393601494543e-05,
+      "loss": 0.6887,
+      "step": 8610
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.2014849185943604,
+      "learning_rate": 1.0990717496643121e-05,
+      "loss": 0.7714,
+      "step": 8620
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9347399473190308,
+      "learning_rate": 1.0979041391791698e-05,
+      "loss": 0.8086,
+      "step": 8630
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1500508785247803,
+      "learning_rate": 1.0967365286940277e-05,
+      "loss": 0.7587,
+      "step": 8640
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1254630088806152,
+      "learning_rate": 1.0955689182088855e-05,
+      "loss": 0.739,
+      "step": 8650
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.7817353010177612,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.415,
+      "eval_steps_per_second": 3.283,
+      "step": 8650
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.046421527862549,
+      "learning_rate": 1.0944013077237434e-05,
+      "loss": 0.7572,
+      "step": 8660
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1235926151275635,
+      "learning_rate": 1.0932336972386013e-05,
+      "loss": 0.7057,
+      "step": 8670
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.151928424835205,
+      "learning_rate": 1.0920660867534591e-05,
+      "loss": 0.7143,
+      "step": 8680
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.883812427520752,
+      "learning_rate": 1.090898476268317e-05,
+      "loss": 0.7012,
+      "step": 8690
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1463747024536133,
+      "learning_rate": 1.0897308657831748e-05,
+      "loss": 0.7487,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 0.7811581492424011,
+      "eval_runtime": 0.6146,
+      "eval_samples_per_second": 16.272,
+      "eval_steps_per_second": 3.254,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.20796799659729,
+      "learning_rate": 1.0885632552980327e-05,
+      "loss": 0.7123,
+      "step": 8710
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.0724101066589355,
+      "learning_rate": 1.0873956448128906e-05,
+      "loss": 0.6971,
+      "step": 8720
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9543169736862183,
+      "learning_rate": 1.0862280343277483e-05,
+      "loss": 0.7247,
+      "step": 8730
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.268307685852051,
+      "learning_rate": 1.0850604238426061e-05,
+      "loss": 0.7413,
+      "step": 8740
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.0434577465057373,
+      "learning_rate": 1.083892813357464e-05,
+      "loss": 0.7106,
+      "step": 8750
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.7845470905303955,
+      "eval_runtime": 0.6138,
+      "eval_samples_per_second": 16.291,
+      "eval_steps_per_second": 3.258,
+      "step": 8750
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.100933790206909,
+      "learning_rate": 1.0827252028723218e-05,
+      "loss": 0.7174,
+      "step": 8760
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9558902978897095,
+      "learning_rate": 1.0815575923871797e-05,
+      "loss": 0.7,
+      "step": 8770
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1897521018981934,
+      "learning_rate": 1.0803899819020376e-05,
+      "loss": 0.7517,
+      "step": 8780
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.200106620788574,
+      "learning_rate": 1.0792223714168954e-05,
+      "loss": 0.725,
+      "step": 8790
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1067657470703125,
+      "learning_rate": 1.0780547609317533e-05,
+      "loss": 0.7418,
+      "step": 8800
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.7839337587356567,
+      "eval_runtime": 0.6023,
+      "eval_samples_per_second": 16.604,
+      "eval_steps_per_second": 3.321,
+      "step": 8800
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1722028255462646,
+      "learning_rate": 1.0768871504466112e-05,
+      "loss": 0.7352,
+      "step": 8810
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1227924823760986,
+      "learning_rate": 1.075719539961469e-05,
+      "loss": 0.7133,
+      "step": 8820
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1083593368530273,
+      "learning_rate": 1.0745519294763267e-05,
+      "loss": 0.7006,
+      "step": 8830
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1085896492004395,
+      "learning_rate": 1.0733843189911846e-05,
+      "loss": 0.6987,
+      "step": 8840
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.514268398284912,
+      "learning_rate": 1.0722167085060424e-05,
+      "loss": 0.7692,
+      "step": 8850
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.7809699177742004,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.343,
+      "eval_steps_per_second": 3.269,
+      "step": 8850
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.027231216430664,
+      "learning_rate": 1.0710490980209003e-05,
+      "loss": 0.7322,
+      "step": 8860
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9498951435089111,
+      "learning_rate": 1.0698814875357582e-05,
+      "loss": 0.7116,
+      "step": 8870
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9841359853744507,
+      "learning_rate": 1.068713877050616e-05,
+      "loss": 0.729,
+      "step": 8880
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.094667434692383,
+      "learning_rate": 1.0675462665654739e-05,
+      "loss": 0.7332,
+      "step": 8890
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1220037937164307,
+      "learning_rate": 1.0663786560803317e-05,
+      "loss": 0.6941,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.7812142372131348,
+      "eval_runtime": 0.6075,
+      "eval_samples_per_second": 16.46,
+      "eval_steps_per_second": 3.292,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.1472508907318115,
+      "learning_rate": 1.0652110455951896e-05,
+      "loss": 0.7153,
+      "step": 8910
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 2.250849962234497,
+      "learning_rate": 1.0640434351100475e-05,
+      "loss": 0.706,
+      "step": 8920
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9042150974273682,
+      "learning_rate": 1.0628758246249052e-05,
+      "loss": 0.7019,
+      "step": 8930
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9402118921279907,
+      "learning_rate": 1.061708214139763e-05,
+      "loss": 0.7085,
+      "step": 8940
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.314553737640381,
+      "learning_rate": 1.060657364703135e-05,
+      "loss": 0.7265,
+      "step": 8950
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.7825098037719727,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.415,
+      "eval_steps_per_second": 3.283,
+      "step": 8950
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.117086887359619,
+      "learning_rate": 1.0594897542179929e-05,
+      "loss": 0.6991,
+      "step": 8960
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0292906761169434,
+      "learning_rate": 1.0583221437328508e-05,
+      "loss": 0.754,
+      "step": 8970
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9374511241912842,
+      "learning_rate": 1.0571545332477086e-05,
+      "loss": 0.722,
+      "step": 8980
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0550105571746826,
+      "learning_rate": 1.0559869227625665e-05,
+      "loss": 0.7371,
+      "step": 8990
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9470839500427246,
+      "learning_rate": 1.0548193122774243e-05,
+      "loss": 0.7403,
+      "step": 9000
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.78445965051651,
+      "eval_runtime": 0.6161,
+      "eval_samples_per_second": 16.23,
+      "eval_steps_per_second": 3.246,
+      "step": 9000
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2952184677124023,
+      "learning_rate": 1.0536517017922822e-05,
+      "loss": 0.7223,
+      "step": 9010
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2147388458251953,
+      "learning_rate": 1.05248409130714e-05,
+      "loss": 0.6885,
+      "step": 9020
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.115015983581543,
+      "learning_rate": 1.0513164808219978e-05,
+      "loss": 0.7212,
+      "step": 9030
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.2153196334838867,
+      "learning_rate": 1.0501488703368556e-05,
+      "loss": 0.7607,
+      "step": 9040
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0532212257385254,
+      "learning_rate": 1.0489812598517135e-05,
+      "loss": 0.7183,
+      "step": 9050
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.78429114818573,
+      "eval_runtime": 0.6128,
+      "eval_samples_per_second": 16.317,
+      "eval_steps_per_second": 3.263,
+      "step": 9050
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9363305568695068,
+      "learning_rate": 1.0478136493665714e-05,
+      "loss": 0.694,
+      "step": 9060
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.9207769632339478,
+      "learning_rate": 1.0466460388814292e-05,
+      "loss": 0.722,
+      "step": 9070
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.0400052070617676,
+      "learning_rate": 1.045478428396287e-05,
+      "loss": 0.7016,
+      "step": 9080
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.1604580879211426,
+      "learning_rate": 1.044310817911145e-05,
+      "loss": 0.7292,
+      "step": 9090
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.888636827468872,
+      "learning_rate": 1.0431432074260028e-05,
+      "loss": 0.7004,
+      "step": 9100
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.7830975651741028,
+      "eval_runtime": 0.614,
+      "eval_samples_per_second": 16.287,
+      "eval_steps_per_second": 3.257,
+      "step": 9100
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2612714767456055,
+      "learning_rate": 1.0419755969408607e-05,
+      "loss": 0.7188,
+      "step": 9110
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.994923710823059,
+      "learning_rate": 1.0408079864557185e-05,
+      "loss": 0.7271,
+      "step": 9120
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9237220287322998,
+      "learning_rate": 1.0396403759705762e-05,
+      "loss": 0.7184,
+      "step": 9130
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.99628746509552,
+      "learning_rate": 1.038472765485434e-05,
+      "loss": 0.7087,
+      "step": 9140
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.0244808197021484,
+      "learning_rate": 1.037305155000292e-05,
+      "loss": 0.7134,
+      "step": 9150
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.7820218801498413,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.391,
+      "eval_steps_per_second": 3.278,
+      "step": 9150
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.164515733718872,
+      "learning_rate": 1.0361375445151498e-05,
+      "loss": 0.7324,
+      "step": 9160
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1342873573303223,
+      "learning_rate": 1.0349699340300077e-05,
+      "loss": 0.7102,
+      "step": 9170
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.102984666824341,
+      "learning_rate": 1.0338023235448655e-05,
+      "loss": 0.6808,
+      "step": 9180
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.137730836868286,
+      "learning_rate": 1.0326347130597234e-05,
+      "loss": 0.7244,
+      "step": 9190
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.352656602859497,
+      "learning_rate": 1.0314671025745812e-05,
+      "loss": 0.7225,
+      "step": 9200
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.7831799983978271,
+      "eval_runtime": 0.6081,
+      "eval_samples_per_second": 16.443,
+      "eval_steps_per_second": 3.289,
+      "step": 9200
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9977498054504395,
+      "learning_rate": 1.0302994920894391e-05,
+      "loss": 0.6913,
+      "step": 9210
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.000168800354004,
+      "learning_rate": 1.029131881604297e-05,
+      "loss": 0.727,
+      "step": 9220
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.3548800945281982,
+      "learning_rate": 1.0279642711191547e-05,
+      "loss": 0.7046,
+      "step": 9230
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.2504868507385254,
+      "learning_rate": 1.0267966606340125e-05,
+      "loss": 0.7273,
+      "step": 9240
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.320638418197632,
+      "learning_rate": 1.0256290501488704e-05,
+      "loss": 0.7046,
+      "step": 9250
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.7830077409744263,
+      "eval_runtime": 0.6062,
+      "eval_samples_per_second": 16.497,
+      "eval_steps_per_second": 3.299,
+      "step": 9250
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 3.0581424236297607,
+      "learning_rate": 1.0244614396637282e-05,
+      "loss": 0.7026,
+      "step": 9260
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4518903493881226,
+      "learning_rate": 1.0232938291785861e-05,
+      "loss": 0.7421,
+      "step": 9270
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 2.1764919757843018,
+      "learning_rate": 1.022126218693444e-05,
+      "loss": 0.7646,
+      "step": 9280
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.085190534591675,
+      "learning_rate": 1.0209586082083018e-05,
+      "loss": 0.7064,
+      "step": 9290
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.2051942348480225,
+      "learning_rate": 1.0197909977231597e-05,
+      "loss": 0.7128,
+      "step": 9300
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.782359778881073,
+      "eval_runtime": 0.6112,
+      "eval_samples_per_second": 16.36,
+      "eval_steps_per_second": 3.272,
+      "step": 9300
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1654062271118164,
+      "learning_rate": 1.0186233872380176e-05,
+      "loss": 0.7309,
+      "step": 9310
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1760201454162598,
+      "learning_rate": 1.0174557767528754e-05,
+      "loss": 0.7498,
+      "step": 9320
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1647613048553467,
+      "learning_rate": 1.0162881662677331e-05,
+      "loss": 0.7078,
+      "step": 9330
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.234020709991455,
+      "learning_rate": 1.015120555782591e-05,
+      "loss": 0.7459,
+      "step": 9340
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.272688865661621,
+      "learning_rate": 1.0139529452974488e-05,
+      "loss": 0.7151,
+      "step": 9350
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.7783147692680359,
+      "eval_runtime": 0.613,
+      "eval_samples_per_second": 16.312,
+      "eval_steps_per_second": 3.262,
+      "step": 9350
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9827055931091309,
+      "learning_rate": 1.0127853348123067e-05,
+      "loss": 0.729,
+      "step": 9360
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9151545763015747,
+      "learning_rate": 1.0116177243271646e-05,
+      "loss": 0.7234,
+      "step": 9370
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9636876583099365,
+      "learning_rate": 1.0104501138420224e-05,
+      "loss": 0.7449,
+      "step": 9380
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.147494316101074,
+      "learning_rate": 1.0092825033568803e-05,
+      "loss": 0.7028,
+      "step": 9390
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.979257345199585,
+      "learning_rate": 1.0081148928717381e-05,
+      "loss": 0.7029,
+      "step": 9400
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.780539870262146,
+      "eval_runtime": 0.6091,
+      "eval_samples_per_second": 16.418,
+      "eval_steps_per_second": 3.284,
+      "step": 9400
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.0492749214172363,
+      "learning_rate": 1.006947282386596e-05,
+      "loss": 0.7432,
+      "step": 9410
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.9653064012527466,
+      "learning_rate": 1.0057796719014537e-05,
+      "loss": 0.6816,
+      "step": 9420
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.34822416305542,
+      "learning_rate": 1.0046120614163116e-05,
+      "loss": 0.7117,
+      "step": 9430
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.298841714859009,
+      "learning_rate": 1.0034444509311694e-05,
+      "loss": 0.7195,
+      "step": 9440
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8653173446655273,
+      "learning_rate": 1.0022768404460273e-05,
+      "loss": 0.6888,
+      "step": 9450
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.7793318033218384,
+      "eval_runtime": 0.609,
+      "eval_samples_per_second": 16.42,
+      "eval_steps_per_second": 3.284,
+      "step": 9450
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.1603472232818604,
+      "learning_rate": 1.0011092299608851e-05,
+      "loss": 0.7078,
+      "step": 9460
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.33748459815979,
+      "learning_rate": 9.99941619475743e-06,
+      "loss": 0.6994,
+      "step": 9470
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2109758853912354,
+      "learning_rate": 9.987740089906009e-06,
+      "loss": 0.7109,
+      "step": 9480
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.3254356384277344,
+      "learning_rate": 9.976063985054587e-06,
+      "loss": 0.6835,
+      "step": 9490
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.0192439556121826,
+      "learning_rate": 9.964387880203164e-06,
+      "loss": 0.7241,
+      "step": 9500
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.7792420387268066,
+      "eval_runtime": 0.6106,
+      "eval_samples_per_second": 16.377,
+      "eval_steps_per_second": 3.275,
+      "step": 9500
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8903729915618896,
+      "learning_rate": 9.952711775351743e-06,
+      "loss": 0.7387,
+      "step": 9510
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.217461585998535,
+      "learning_rate": 9.941035670500321e-06,
+      "loss": 0.7637,
+      "step": 9520
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.245603084564209,
+      "learning_rate": 9.9293595656489e-06,
+      "loss": 0.7309,
+      "step": 9530
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8404752016067505,
+      "learning_rate": 9.917683460797479e-06,
+      "loss": 0.6825,
+      "step": 9540
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.2425146102905273,
+      "learning_rate": 9.906007355946057e-06,
+      "loss": 0.709,
+      "step": 9550
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.7779631614685059,
+      "eval_runtime": 0.6102,
+      "eval_samples_per_second": 16.388,
+      "eval_steps_per_second": 3.278,
+      "step": 9550
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.234157085418701,
+      "learning_rate": 9.894331251094636e-06,
+      "loss": 0.7205,
+      "step": 9560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.1441380977630615,
+      "learning_rate": 9.882655146243214e-06,
+      "loss": 0.6889,
+      "step": 9570
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.984332799911499,
+      "learning_rate": 9.870979041391793e-06,
+      "loss": 0.6988,
+      "step": 9580
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.7375990152359009,
+      "learning_rate": 9.859302936540372e-06,
+      "loss": 0.7509,
+      "step": 9590
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.198277711868286,
+      "learning_rate": 9.847626831688949e-06,
+      "loss": 0.7315,
+      "step": 9600
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.777324378490448,
+      "eval_runtime": 0.6139,
+      "eval_samples_per_second": 16.289,
+      "eval_steps_per_second": 3.258,
+      "step": 9600
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9727591276168823,
+      "learning_rate": 9.835950726837527e-06,
+      "loss": 0.7024,
+      "step": 9610
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9702403545379639,
+      "learning_rate": 9.824274621986106e-06,
+      "loss": 0.7072,
+      "step": 9620
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.148256301879883,
+      "learning_rate": 9.812598517134684e-06,
+      "loss": 0.733,
+      "step": 9630
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 2.17622971534729,
+      "learning_rate": 9.800922412283263e-06,
+      "loss": 0.752,
+      "step": 9640
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1276326179504395,
+      "learning_rate": 9.789246307431842e-06,
+      "loss": 0.7539,
+      "step": 9650
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.7790389060974121,
+      "eval_runtime": 0.6105,
+      "eval_samples_per_second": 16.38,
+      "eval_steps_per_second": 3.276,
+      "step": 9650
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2252426147460938,
+      "learning_rate": 9.77757020258042e-06,
+      "loss": 0.7048,
+      "step": 9660
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1048762798309326,
+      "learning_rate": 9.765894097728999e-06,
+      "loss": 0.7355,
+      "step": 9670
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.070155620574951,
+      "learning_rate": 9.754217992877578e-06,
+      "loss": 0.7254,
+      "step": 9680
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1020781993865967,
+      "learning_rate": 9.742541888026156e-06,
+      "loss": 0.8118,
+      "step": 9690
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.808031439781189,
+      "learning_rate": 9.730865783174733e-06,
+      "loss": 0.7039,
+      "step": 9700
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.7765337228775024,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.359,
+      "eval_steps_per_second": 3.272,
+      "step": 9700
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.097959280014038,
+      "learning_rate": 9.719189678323312e-06,
+      "loss": 0.7301,
+      "step": 9710
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1885650157928467,
+      "learning_rate": 9.70751357347189e-06,
+      "loss": 0.6932,
+      "step": 9720
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1910619735717773,
+      "learning_rate": 9.695837468620469e-06,
+      "loss": 0.7297,
+      "step": 9730
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.100555181503296,
+      "learning_rate": 9.684161363769048e-06,
+      "loss": 0.8382,
+      "step": 9740
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9188323020935059,
+      "learning_rate": 9.672485258917626e-06,
+      "loss": 0.6687,
+      "step": 9750
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.7769054174423218,
+      "eval_runtime": 0.6114,
+      "eval_samples_per_second": 16.355,
+      "eval_steps_per_second": 3.271,
+      "step": 9750
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9275360107421875,
+      "learning_rate": 9.660809154066205e-06,
+      "loss": 0.712,
+      "step": 9760
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.1616499423980713,
+      "learning_rate": 9.649133049214783e-06,
+      "loss": 0.7172,
+      "step": 9770
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.031280279159546,
+      "learning_rate": 9.637456944363362e-06,
+      "loss": 0.6926,
+      "step": 9780
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.9138914346694946,
+      "learning_rate": 9.62578083951194e-06,
+      "loss": 0.7263,
+      "step": 9790
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.97421133518219,
+      "learning_rate": 9.614104734660518e-06,
+      "loss": 0.7494,
+      "step": 9800
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 0.7773782014846802,
+      "eval_runtime": 0.6116,
+      "eval_samples_per_second": 16.349,
+      "eval_steps_per_second": 3.27,
+      "step": 9800
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.2048544883728027,
+      "learning_rate": 9.602428629809096e-06,
+      "loss": 0.7246,
+      "step": 9810
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.074448347091675,
+      "learning_rate": 9.590752524957675e-06,
+      "loss": 0.7151,
+      "step": 9820
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.257120370864868,
+      "learning_rate": 9.579076420106253e-06,
+      "loss": 0.7079,
+      "step": 9830
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1302804946899414,
+      "learning_rate": 9.567400315254832e-06,
+      "loss": 0.7058,
+      "step": 9840
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2036924362182617,
+      "learning_rate": 9.555724210403409e-06,
+      "loss": 0.6777,
+      "step": 9850
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.7781575322151184,
+      "eval_runtime": 0.6149,
+      "eval_samples_per_second": 16.264,
+      "eval_steps_per_second": 3.253,
+      "step": 9850
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2481439113616943,
+      "learning_rate": 9.54404810555199e-06,
+      "loss": 0.7207,
+      "step": 9860
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.287849187850952,
+      "learning_rate": 9.532372000700568e-06,
+      "loss": 0.7418,
+      "step": 9870
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9974662065505981,
+      "learning_rate": 9.520695895849147e-06,
+      "loss": 0.7137,
+      "step": 9880
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.103928565979004,
+      "learning_rate": 9.509019790997725e-06,
+      "loss": 0.702,
+      "step": 9890
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.186194896697998,
+      "learning_rate": 9.497343686146302e-06,
+      "loss": 0.6724,
+      "step": 9900
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.7782595753669739,
+      "eval_runtime": 0.6105,
+      "eval_samples_per_second": 16.381,
+      "eval_steps_per_second": 3.276,
+      "step": 9900
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.261936902999878,
+      "learning_rate": 9.48566758129488e-06,
+      "loss": 0.7194,
+      "step": 9910
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1588361263275146,
+      "learning_rate": 9.47399147644346e-06,
+      "loss": 0.7284,
+      "step": 9920
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.052314043045044,
+      "learning_rate": 9.462315371592038e-06,
+      "loss": 0.7102,
+      "step": 9930
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.0471248626708984,
+      "learning_rate": 9.450639266740617e-06,
+      "loss": 0.6862,
+      "step": 9940
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1881284713745117,
+      "learning_rate": 9.438963161889193e-06,
+      "loss": 0.7565,
+      "step": 9950
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.7777990102767944,
+      "eval_runtime": 0.6131,
+      "eval_samples_per_second": 16.309,
+      "eval_steps_per_second": 3.262,
+      "step": 9950
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.9974339008331299,
+      "learning_rate": 9.427287057037772e-06,
+      "loss": 0.7774,
+      "step": 9960
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.2147862911224365,
+      "learning_rate": 9.41561095218635e-06,
+      "loss": 0.7077,
+      "step": 9970
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.344430685043335,
+      "learning_rate": 9.40393484733493e-06,
+      "loss": 0.7255,
+      "step": 9980
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.1508092880249023,
+      "learning_rate": 9.39225874248351e-06,
+      "loss": 0.7175,
+      "step": 9990
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 2.049746513366699,
+      "learning_rate": 9.380582637632087e-06,
+      "loss": 0.7155,
+      "step": 10000
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.7731291651725769,
+      "eval_runtime": 0.6099,
+      "eval_samples_per_second": 16.396,
+      "eval_steps_per_second": 3.279,
+      "step": 10000
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.8308905363082886,
+      "learning_rate": 9.368906532780665e-06,
+      "loss": 0.7425,
+      "step": 10010
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.128713846206665,
+      "learning_rate": 9.357230427929244e-06,
+      "loss": 0.7241,
+      "step": 10020
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.6540567874908447,
+      "learning_rate": 9.345554323077822e-06,
+      "loss": 0.7789,
+      "step": 10030
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.205031633377075,
+      "learning_rate": 9.333878218226401e-06,
+      "loss": 0.7138,
+      "step": 10040
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3325910568237305,
+      "learning_rate": 9.322202113374978e-06,
+      "loss": 0.7102,
+      "step": 10050
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.7755321264266968,
+      "eval_runtime": 0.6149,
+      "eval_samples_per_second": 16.262,
+      "eval_steps_per_second": 3.252,
+      "step": 10050
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.0699234008789062,
+      "learning_rate": 9.310526008523557e-06,
+      "loss": 0.7291,
+      "step": 10060
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3096320629119873,
+      "learning_rate": 9.298849903672135e-06,
+      "loss": 0.7115,
+      "step": 10070
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.240403890609741,
+      "learning_rate": 9.287173798820714e-06,
+      "loss": 0.7069,
+      "step": 10080
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1584725379943848,
+      "learning_rate": 9.275497693969292e-06,
+      "loss": 0.6872,
+      "step": 10090
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.3668770790100098,
+      "learning_rate": 9.263821589117871e-06,
+      "loss": 0.6982,
+      "step": 10100
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.7746591567993164,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.34,
+      "eval_steps_per_second": 3.268,
+      "step": 10100
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1024999618530273,
+      "learning_rate": 9.25214548426645e-06,
+      "loss": 0.7244,
+      "step": 10110
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.4041571617126465,
+      "learning_rate": 9.240469379415028e-06,
+      "loss": 0.7075,
+      "step": 10120
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.1590189933776855,
+      "learning_rate": 9.228793274563607e-06,
+      "loss": 0.7286,
+      "step": 10130
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9986599683761597,
+      "learning_rate": 9.217117169712185e-06,
+      "loss": 0.7147,
+      "step": 10140
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.980279564857483,
+      "learning_rate": 9.205441064860762e-06,
+      "loss": 0.7088,
+      "step": 10150
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.775314450263977,
+      "eval_runtime": 0.6118,
+      "eval_samples_per_second": 16.345,
+      "eval_steps_per_second": 3.269,
+      "step": 10150
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.010408401489258,
+      "learning_rate": 9.193764960009341e-06,
+      "loss": 0.6988,
+      "step": 10160
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.123391628265381,
+      "learning_rate": 9.18208885515792e-06,
+      "loss": 0.7482,
+      "step": 10170
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 2.5498523712158203,
+      "learning_rate": 9.170412750306498e-06,
+      "loss": 0.7149,
+      "step": 10180
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0450544357299805,
+      "learning_rate": 9.158736645455077e-06,
+      "loss": 0.7124,
+      "step": 10190
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.429563283920288,
+      "learning_rate": 9.147060540603655e-06,
+      "loss": 0.7425,
+      "step": 10200
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.7775689363479614,
+      "eval_runtime": 0.6121,
+      "eval_samples_per_second": 16.336,
+      "eval_steps_per_second": 3.267,
+      "step": 10200
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.8823789358139038,
+      "learning_rate": 9.135384435752234e-06,
+      "loss": 0.7252,
+      "step": 10210
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0774431228637695,
+      "learning_rate": 9.123708330900813e-06,
+      "loss": 0.7427,
+      "step": 10220
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.215461015701294,
+      "learning_rate": 9.112032226049391e-06,
+      "loss": 0.7312,
+      "step": 10230
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.297828197479248,
+      "learning_rate": 9.10035612119797e-06,
+      "loss": 0.701,
+      "step": 10240
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0988781452178955,
+      "learning_rate": 9.088680016346547e-06,
+      "loss": 0.7028,
+      "step": 10250
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.7762842178344727,
+      "eval_runtime": 0.6104,
+      "eval_samples_per_second": 16.384,
+      "eval_steps_per_second": 3.277,
+      "step": 10250
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.170772075653076,
+      "learning_rate": 9.077003911495125e-06,
+      "loss": 0.7399,
+      "step": 10260
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.9261767864227295,
+      "learning_rate": 9.065327806643704e-06,
+      "loss": 0.719,
+      "step": 10270
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1540753841400146,
+      "learning_rate": 9.053651701792283e-06,
+      "loss": 0.7026,
+      "step": 10280
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1176681518554688,
+      "learning_rate": 9.041975596940861e-06,
+      "loss": 0.7186,
+      "step": 10290
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0018396377563477,
+      "learning_rate": 9.03029949208944e-06,
+      "loss": 0.7282,
+      "step": 10300
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.775932788848877,
+      "eval_runtime": 0.6051,
+      "eval_samples_per_second": 16.525,
+      "eval_steps_per_second": 3.305,
+      "step": 10300
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3435580730438232,
+      "learning_rate": 9.018623387238019e-06,
+      "loss": 0.727,
+      "step": 10310
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0578792095184326,
+      "learning_rate": 9.006947282386597e-06,
+      "loss": 0.7209,
+      "step": 10320
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.3711845874786377,
+      "learning_rate": 8.995271177535176e-06,
+      "loss": 0.7122,
+      "step": 10330
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.069974899291992,
+      "learning_rate": 8.983595072683754e-06,
+      "loss": 0.6916,
+      "step": 10340
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.0944151878356934,
+      "learning_rate": 8.971918967832331e-06,
+      "loss": 0.7084,
+      "step": 10350
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 0.7765060067176819,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.336,
+      "eval_steps_per_second": 3.267,
+      "step": 10350
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 2.1235668659210205,
+      "learning_rate": 8.96024286298091e-06,
+      "loss": 0.7015,
+      "step": 10360
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1395819187164307,
+      "learning_rate": 8.948566758129489e-06,
+      "loss": 0.6737,
+      "step": 10370
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8593639135360718,
+      "learning_rate": 8.936890653278067e-06,
+      "loss": 0.7611,
+      "step": 10380
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.938014030456543,
+      "learning_rate": 8.925214548426646e-06,
+      "loss": 0.7268,
+      "step": 10390
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.107361078262329,
+      "learning_rate": 8.913538443575224e-06,
+      "loss": 0.7085,
+      "step": 10400
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.7757299542427063,
+      "eval_runtime": 0.609,
+      "eval_samples_per_second": 16.419,
+      "eval_steps_per_second": 3.284,
+      "step": 10400
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.167476177215576,
+      "learning_rate": 8.901862338723803e-06,
+      "loss": 0.7197,
+      "step": 10410
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0912516117095947,
+      "learning_rate": 8.890186233872382e-06,
+      "loss": 0.6978,
+      "step": 10420
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.188811779022217,
+      "learning_rate": 8.87851012902096e-06,
+      "loss": 0.6963,
+      "step": 10430
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.053473472595215,
+      "learning_rate": 8.866834024169539e-06,
+      "loss": 0.7293,
+      "step": 10440
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.052640438079834,
+      "learning_rate": 8.855157919318116e-06,
+      "loss": 0.7485,
+      "step": 10450
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.7774114608764648,
+      "eval_runtime": 0.6112,
+      "eval_samples_per_second": 16.362,
+      "eval_steps_per_second": 3.272,
+      "step": 10450
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0890774726867676,
+      "learning_rate": 8.843481814466694e-06,
+      "loss": 0.7111,
+      "step": 10460
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.8550294637680054,
+      "learning_rate": 8.831805709615273e-06,
+      "loss": 0.7621,
+      "step": 10470
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.2536511421203613,
+      "learning_rate": 8.820129604763852e-06,
+      "loss": 0.6978,
+      "step": 10480
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1750214099884033,
+      "learning_rate": 8.80845349991243e-06,
+      "loss": 0.7216,
+      "step": 10490
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0018579959869385,
+      "learning_rate": 8.796777395061009e-06,
+      "loss": 0.7077,
+      "step": 10500
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.7752179503440857,
+      "eval_runtime": 0.6153,
+      "eval_samples_per_second": 16.252,
+      "eval_steps_per_second": 3.25,
+      "step": 10500
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.1800074577331543,
+      "learning_rate": 8.785101290209587e-06,
+      "loss": 0.7073,
+      "step": 10510
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.9102400541305542,
+      "learning_rate": 8.773425185358166e-06,
+      "loss": 0.7382,
+      "step": 10520
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.0987470149993896,
+      "learning_rate": 8.761749080506745e-06,
+      "loss": 0.7053,
+      "step": 10530
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.142873764038086,
+      "learning_rate": 8.750072975655323e-06,
+      "loss": 0.736,
+      "step": 10540
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1113665103912354,
+      "learning_rate": 8.7383968708039e-06,
+      "loss": 0.7618,
+      "step": 10550
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.7756747007369995,
+      "eval_runtime": 0.6096,
+      "eval_samples_per_second": 16.404,
+      "eval_steps_per_second": 3.281,
+      "step": 10550
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3636112213134766,
+      "learning_rate": 8.726720765952479e-06,
+      "loss": 0.7306,
+      "step": 10560
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.042551040649414,
+      "learning_rate": 8.715044661101057e-06,
+      "loss": 0.7491,
+      "step": 10570
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.258361577987671,
+      "learning_rate": 8.703368556249636e-06,
+      "loss": 0.6995,
+      "step": 10580
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3399622440338135,
+      "learning_rate": 8.691692451398213e-06,
+      "loss": 0.7393,
+      "step": 10590
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0389201641082764,
+      "learning_rate": 8.680016346546792e-06,
+      "loss": 0.7209,
+      "step": 10600
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.7761565446853638,
+      "eval_runtime": 0.611,
+      "eval_samples_per_second": 16.366,
+      "eval_steps_per_second": 3.273,
+      "step": 10600
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1324329376220703,
+      "learning_rate": 8.66834024169537e-06,
+      "loss": 0.7755,
+      "step": 10610
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.0502030849456787,
+      "learning_rate": 8.656664136843949e-06,
+      "loss": 0.7563,
+      "step": 10620
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.3421247005462646,
+      "learning_rate": 8.644988031992527e-06,
+      "loss": 0.6662,
+      "step": 10630
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.032641887664795,
+      "learning_rate": 8.633311927141106e-06,
+      "loss": 0.748,
+      "step": 10640
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.017148971557617,
+      "learning_rate": 8.621635822289685e-06,
+      "loss": 0.7006,
+      "step": 10650
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.7791746258735657,
+      "eval_runtime": 0.6175,
+      "eval_samples_per_second": 16.195,
+      "eval_steps_per_second": 3.239,
+      "step": 10650
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.9664469957351685,
+      "learning_rate": 8.609959717438263e-06,
+      "loss": 0.7593,
+      "step": 10660
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.024787425994873,
+      "learning_rate": 8.598283612586842e-06,
+      "loss": 0.7604,
+      "step": 10670
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.957316517829895,
+      "learning_rate": 8.58660750773542e-06,
+      "loss": 0.8104,
+      "step": 10680
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1296300888061523,
+      "learning_rate": 8.574931402883998e-06,
+      "loss": 0.7426,
+      "step": 10690
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.1550159454345703,
+      "learning_rate": 8.563255298032576e-06,
+      "loss": 0.7233,
+      "step": 10700
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 0.7777634859085083,
+      "eval_runtime": 0.6141,
+      "eval_samples_per_second": 16.284,
+      "eval_steps_per_second": 3.257,
+      "step": 10700
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.973652958869934,
+      "learning_rate": 8.551579193181155e-06,
+      "loss": 0.7468,
+      "step": 10710
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 2.056971311569214,
+      "learning_rate": 8.539903088329733e-06,
+      "loss": 0.7018,
+      "step": 10720
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9542721509933472,
+      "learning_rate": 8.528226983478312e-06,
+      "loss": 0.7226,
+      "step": 10730
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.8472495079040527,
+      "learning_rate": 8.51655087862689e-06,
+      "loss": 0.7132,
+      "step": 10740
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.1824088096618652,
+      "learning_rate": 8.50487477377547e-06,
+      "loss": 0.6963,
+      "step": 10750
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.7781248688697815,
+      "eval_runtime": 0.6114,
+      "eval_samples_per_second": 16.357,
+      "eval_steps_per_second": 3.271,
+      "step": 10750
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.41373872756958,
+      "learning_rate": 8.493198668924048e-06,
+      "loss": 0.7118,
+      "step": 10760
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5209472179412842,
+      "learning_rate": 8.481522564072626e-06,
+      "loss": 0.7149,
+      "step": 10770
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2392683029174805,
+      "learning_rate": 8.469846459221205e-06,
+      "loss": 0.749,
+      "step": 10780
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2378783226013184,
+      "learning_rate": 8.458170354369782e-06,
+      "loss": 0.7263,
+      "step": 10790
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4283087253570557,
+      "learning_rate": 8.44649424951836e-06,
+      "loss": 0.7367,
+      "step": 10800
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.7747004628181458,
+      "eval_runtime": 0.6133,
+      "eval_samples_per_second": 16.305,
+      "eval_steps_per_second": 3.261,
+      "step": 10800
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2812881469726562,
+      "learning_rate": 8.43481814466694e-06,
+      "loss": 0.7193,
+      "step": 10810
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.33746600151062,
+      "learning_rate": 8.423142039815518e-06,
+      "loss": 0.7102,
+      "step": 10820
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.323997735977173,
+      "learning_rate": 8.411465934964096e-06,
+      "loss": 0.7034,
+      "step": 10830
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0983424186706543,
+      "learning_rate": 8.399789830112675e-06,
+      "loss": 0.6783,
+      "step": 10840
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.212780237197876,
+      "learning_rate": 8.388113725261254e-06,
+      "loss": 0.6894,
+      "step": 10850
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.7773181200027466,
+      "eval_runtime": 0.6161,
+      "eval_samples_per_second": 16.232,
+      "eval_steps_per_second": 3.246,
+      "step": 10850
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.187725782394409,
+      "learning_rate": 8.376437620409832e-06,
+      "loss": 0.707,
+      "step": 10860
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.9294705390930176,
+      "learning_rate": 8.364761515558411e-06,
+      "loss": 0.7306,
+      "step": 10870
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.2794885635375977,
+      "learning_rate": 8.35308541070699e-06,
+      "loss": 0.7044,
+      "step": 10880
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.4399662017822266,
+      "learning_rate": 8.341409305855566e-06,
+      "loss": 0.7134,
+      "step": 10890
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.094402313232422,
+      "learning_rate": 8.329733201004145e-06,
+      "loss": 0.6783,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.7724976539611816,
+      "eval_runtime": 0.6029,
+      "eval_samples_per_second": 16.588,
+      "eval_steps_per_second": 3.318,
+      "step": 10900
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0956320762634277,
+      "learning_rate": 8.318057096152724e-06,
+      "loss": 0.7146,
+      "step": 10910
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.555417537689209,
+      "learning_rate": 8.306380991301302e-06,
+      "loss": 0.7077,
+      "step": 10920
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0937747955322266,
+      "learning_rate": 8.294704886449881e-06,
+      "loss": 0.7005,
+      "step": 10930
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.099982976913452,
+      "learning_rate": 8.28302878159846e-06,
+      "loss": 0.6971,
+      "step": 10940
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.350389003753662,
+      "learning_rate": 8.271352676747038e-06,
+      "loss": 0.7204,
+      "step": 10950
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.7739883661270142,
+      "eval_runtime": 0.6103,
+      "eval_samples_per_second": 16.387,
+      "eval_steps_per_second": 3.277,
+      "step": 10950
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0243704319000244,
+      "learning_rate": 8.259676571895617e-06,
+      "loss": 0.7393,
+      "step": 10960
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.269660711288452,
+      "learning_rate": 8.248000467044195e-06,
+      "loss": 0.7118,
+      "step": 10970
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.230072259902954,
+      "learning_rate": 8.236324362192774e-06,
+      "loss": 0.6854,
+      "step": 10980
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.255703926086426,
+      "learning_rate": 8.224648257341351e-06,
+      "loss": 0.692,
+      "step": 10990
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.9582380056381226,
+      "learning_rate": 8.21297215248993e-06,
+      "loss": 0.6765,
+      "step": 11000
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.7773922681808472,
+      "eval_runtime": 0.613,
+      "eval_samples_per_second": 16.314,
+      "eval_steps_per_second": 3.263,
+      "step": 11000
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2414042949676514,
+      "learning_rate": 8.201296047638508e-06,
+      "loss": 0.6995,
+      "step": 11010
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.017314910888672,
+      "learning_rate": 8.189619942787087e-06,
+      "loss": 0.743,
+      "step": 11020
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.1185383796691895,
+      "learning_rate": 8.179111448420807e-06,
+      "loss": 0.6924,
+      "step": 11030
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0378611087799072,
+      "learning_rate": 8.167435343569386e-06,
+      "loss": 0.6925,
+      "step": 11040
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0168330669403076,
+      "learning_rate": 8.155759238717964e-06,
+      "loss": 0.7284,
+      "step": 11050
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.77501380443573,
+      "eval_runtime": 0.6114,
+      "eval_samples_per_second": 16.357,
+      "eval_steps_per_second": 3.271,
+      "step": 11050
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.2398998737335205,
+      "learning_rate": 8.144083133866543e-06,
+      "loss": 0.7194,
+      "step": 11060
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.213825225830078,
+      "learning_rate": 8.132407029015121e-06,
+      "loss": 0.6984,
+      "step": 11070
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.027752161026001,
+      "learning_rate": 8.1207309241637e-06,
+      "loss": 0.7442,
+      "step": 11080
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.194708824157715,
+      "learning_rate": 8.109054819312277e-06,
+      "loss": 0.7382,
+      "step": 11090
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2868454456329346,
+      "learning_rate": 8.097378714460856e-06,
+      "loss": 0.6781,
+      "step": 11100
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.7761499285697937,
+      "eval_runtime": 0.6038,
+      "eval_samples_per_second": 16.561,
+      "eval_steps_per_second": 3.312,
+      "step": 11100
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.3066868782043457,
+      "learning_rate": 8.085702609609434e-06,
+      "loss": 0.7142,
+      "step": 11110
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.063345432281494,
+      "learning_rate": 8.074026504758013e-06,
+      "loss": 0.743,
+      "step": 11120
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.070523262023926,
+      "learning_rate": 8.062350399906591e-06,
+      "loss": 0.6714,
+      "step": 11130
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1865298748016357,
+      "learning_rate": 8.05067429505517e-06,
+      "loss": 0.7076,
+      "step": 11140
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9950017929077148,
+      "learning_rate": 8.038998190203749e-06,
+      "loss": 0.7025,
+      "step": 11150
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.7755822539329529,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.391,
+      "eval_steps_per_second": 3.278,
+      "step": 11150
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.348957061767578,
+      "learning_rate": 8.027322085352327e-06,
+      "loss": 0.6963,
+      "step": 11160
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0844287872314453,
+      "learning_rate": 8.015645980500906e-06,
+      "loss": 0.7103,
+      "step": 11170
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1541004180908203,
+      "learning_rate": 8.003969875649485e-06,
+      "loss": 0.7073,
+      "step": 11180
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0058650970458984,
+      "learning_rate": 7.992293770798061e-06,
+      "loss": 0.7063,
+      "step": 11190
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.9685399532318115,
+      "learning_rate": 7.98061766594664e-06,
+      "loss": 0.7237,
+      "step": 11200
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.7765508890151978,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.342,
+      "eval_steps_per_second": 3.268,
+      "step": 11200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2462260723114014,
+      "learning_rate": 7.968941561095219e-06,
+      "loss": 0.8233,
+      "step": 11210
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.1218104362487793,
+      "learning_rate": 7.957265456243797e-06,
+      "loss": 0.6771,
+      "step": 11220
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2539467811584473,
+      "learning_rate": 7.945589351392376e-06,
+      "loss": 0.6838,
+      "step": 11230
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.2374560832977295,
+      "learning_rate": 7.933913246540955e-06,
+      "loss": 0.7924,
+      "step": 11240
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.312767505645752,
+      "learning_rate": 7.922237141689533e-06,
+      "loss": 0.6879,
+      "step": 11250
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.7774962186813354,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.371,
+      "eval_steps_per_second": 3.274,
+      "step": 11250
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.0581116676330566,
+      "learning_rate": 7.910561036838112e-06,
+      "loss": 0.6937,
+      "step": 11260
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2454636096954346,
+      "learning_rate": 7.89888493198669e-06,
+      "loss": 0.7014,
+      "step": 11270
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.6528425216674805,
+      "learning_rate": 7.887208827135269e-06,
+      "loss": 0.7586,
+      "step": 11280
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.4045207500457764,
+      "learning_rate": 7.875532722283846e-06,
+      "loss": 0.7159,
+      "step": 11290
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1636478900909424,
+      "learning_rate": 7.863856617432425e-06,
+      "loss": 0.7576,
+      "step": 11300
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.7759847640991211,
+      "eval_runtime": 0.6076,
+      "eval_samples_per_second": 16.459,
+      "eval_steps_per_second": 3.292,
+      "step": 11300
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3433048725128174,
+      "learning_rate": 7.852180512581003e-06,
+      "loss": 0.6888,
+      "step": 11310
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2523438930511475,
+      "learning_rate": 7.840504407729582e-06,
+      "loss": 0.7211,
+      "step": 11320
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1141107082366943,
+      "learning_rate": 7.82882830287816e-06,
+      "loss": 0.7191,
+      "step": 11330
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3527889251708984,
+      "learning_rate": 7.817152198026739e-06,
+      "loss": 0.7248,
+      "step": 11340
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.0987422466278076,
+      "learning_rate": 7.805476093175318e-06,
+      "loss": 0.7194,
+      "step": 11350
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.7743330597877502,
+      "eval_runtime": 0.6146,
+      "eval_samples_per_second": 16.271,
+      "eval_steps_per_second": 3.254,
+      "step": 11350
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.3022468090057373,
+      "learning_rate": 7.793799988323896e-06,
+      "loss": 0.6784,
+      "step": 11360
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.451298475265503,
+      "learning_rate": 7.782123883472475e-06,
+      "loss": 0.7254,
+      "step": 11370
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.170260190963745,
+      "learning_rate": 7.770447778621054e-06,
+      "loss": 0.724,
+      "step": 11380
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.045337677001953,
+      "learning_rate": 7.75877167376963e-06,
+      "loss": 0.6944,
+      "step": 11390
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2563271522521973,
+      "learning_rate": 7.747095568918209e-06,
+      "loss": 0.7679,
+      "step": 11400
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.7764642834663391,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.414,
+      "eval_steps_per_second": 3.283,
+      "step": 11400
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1152358055114746,
+      "learning_rate": 7.735419464066788e-06,
+      "loss": 0.731,
+      "step": 11410
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.21980619430542,
+      "learning_rate": 7.723743359215366e-06,
+      "loss": 0.7286,
+      "step": 11420
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2817158699035645,
+      "learning_rate": 7.712067254363945e-06,
+      "loss": 0.7146,
+      "step": 11430
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.2119171619415283,
+      "learning_rate": 7.700391149512524e-06,
+      "loss": 0.7079,
+      "step": 11440
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.363447904586792,
+      "learning_rate": 7.688715044661102e-06,
+      "loss": 0.7057,
+      "step": 11450
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.776858925819397,
+      "eval_runtime": 0.6062,
+      "eval_samples_per_second": 16.495,
+      "eval_steps_per_second": 3.299,
+      "step": 11450
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0393240451812744,
+      "learning_rate": 7.67703893980968e-06,
+      "loss": 0.6951,
+      "step": 11460
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4617273807525635,
+      "learning_rate": 7.66536283495826e-06,
+      "loss": 0.742,
+      "step": 11470
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.233792543411255,
+      "learning_rate": 7.653686730106838e-06,
+      "loss": 0.7067,
+      "step": 11480
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9768532514572144,
+      "learning_rate": 7.642010625255415e-06,
+      "loss": 0.7425,
+      "step": 11490
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1606011390686035,
+      "learning_rate": 7.630334520403994e-06,
+      "loss": 0.7185,
+      "step": 11500
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.7778235673904419,
+      "eval_runtime": 0.6122,
+      "eval_samples_per_second": 16.334,
+      "eval_steps_per_second": 3.267,
+      "step": 11500
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6036498546600342,
+      "learning_rate": 7.618658415552572e-06,
+      "loss": 0.7084,
+      "step": 11510
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.7627710103988647,
+      "learning_rate": 7.606982310701151e-06,
+      "loss": 0.7608,
+      "step": 11520
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.252089500427246,
+      "learning_rate": 7.595306205849729e-06,
+      "loss": 0.7093,
+      "step": 11530
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.211738109588623,
+      "learning_rate": 7.583630100998307e-06,
+      "loss": 0.6905,
+      "step": 11540
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.0408599376678467,
+      "learning_rate": 7.571953996146886e-06,
+      "loss": 0.7168,
+      "step": 11550
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.7760379314422607,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.373,
+      "eval_steps_per_second": 3.275,
+      "step": 11550
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.4868359565734863,
+      "learning_rate": 7.560277891295464e-06,
+      "loss": 0.7128,
+      "step": 11560
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1039912700653076,
+      "learning_rate": 7.548601786444043e-06,
+      "loss": 0.7013,
+      "step": 11570
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1972694396972656,
+      "learning_rate": 7.536925681592622e-06,
+      "loss": 0.7042,
+      "step": 11580
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.220156192779541,
+      "learning_rate": 7.525249576741199e-06,
+      "loss": 0.6986,
+      "step": 11590
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.2117905616760254,
+      "learning_rate": 7.513573471889778e-06,
+      "loss": 0.7074,
+      "step": 11600
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.7747154235839844,
+      "eval_runtime": 0.6088,
+      "eval_samples_per_second": 16.425,
+      "eval_steps_per_second": 3.285,
+      "step": 11600
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.3415608406066895,
+      "learning_rate": 7.501897367038357e-06,
+      "loss": 0.6956,
+      "step": 11610
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.1761162281036377,
+      "learning_rate": 7.490221262186935e-06,
+      "loss": 0.7242,
+      "step": 11620
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.3202314376831055,
+      "learning_rate": 7.478545157335514e-06,
+      "loss": 0.7029,
+      "step": 11630
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.282759666442871,
+      "learning_rate": 7.466869052484092e-06,
+      "loss": 0.7309,
+      "step": 11640
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8260728120803833,
+      "learning_rate": 7.45519294763267e-06,
+      "loss": 0.773,
+      "step": 11650
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.7724132537841797,
+      "eval_runtime": 0.61,
+      "eval_samples_per_second": 16.393,
+      "eval_steps_per_second": 3.279,
+      "step": 11650
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0385005474090576,
+      "learning_rate": 7.443516842781249e-06,
+      "loss": 0.7182,
+      "step": 11660
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1878271102905273,
+      "learning_rate": 7.4318407379298275e-06,
+      "loss": 0.7219,
+      "step": 11670
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.084010124206543,
+      "learning_rate": 7.420164633078406e-06,
+      "loss": 0.7292,
+      "step": 11680
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.8747354745864868,
+      "learning_rate": 7.408488528226984e-06,
+      "loss": 0.7223,
+      "step": 11690
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0282881259918213,
+      "learning_rate": 7.3968124233755625e-06,
+      "loss": 0.7042,
+      "step": 11700
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.7713759541511536,
+      "eval_runtime": 0.6103,
+      "eval_samples_per_second": 16.385,
+      "eval_steps_per_second": 3.277,
+      "step": 11700
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2396152019500732,
+      "learning_rate": 7.385136318524141e-06,
+      "loss": 0.7141,
+      "step": 11710
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.403683662414551,
+      "learning_rate": 7.37346021367272e-06,
+      "loss": 0.7056,
+      "step": 11720
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1039319038391113,
+      "learning_rate": 7.361784108821298e-06,
+      "loss": 0.7141,
+      "step": 11730
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.183614730834961,
+      "learning_rate": 7.350108003969876e-06,
+      "loss": 0.7324,
+      "step": 11740
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.344392776489258,
+      "learning_rate": 7.338431899118455e-06,
+      "loss": 0.7452,
+      "step": 11750
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.7719575762748718,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.368,
+      "eval_steps_per_second": 3.274,
+      "step": 11750
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.325392246246338,
+      "learning_rate": 7.326755794267033e-06,
+      "loss": 0.6906,
+      "step": 11760
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.1755056381225586,
+      "learning_rate": 7.315079689415612e-06,
+      "loss": 0.6997,
+      "step": 11770
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.2811532020568848,
+      "learning_rate": 7.3034035845641905e-06,
+      "loss": 0.6942,
+      "step": 11780
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.222635269165039,
+      "learning_rate": 7.291727479712768e-06,
+      "loss": 0.7012,
+      "step": 11790
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.318798542022705,
+      "learning_rate": 7.280051374861347e-06,
+      "loss": 0.7455,
+      "step": 11800
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.7739540338516235,
+      "eval_runtime": 0.6106,
+      "eval_samples_per_second": 16.377,
+      "eval_steps_per_second": 3.275,
+      "step": 11800
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.0053822994232178,
+      "learning_rate": 7.2683752700099256e-06,
+      "loss": 0.7171,
+      "step": 11810
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1134233474731445,
+      "learning_rate": 7.256699165158504e-06,
+      "loss": 0.7463,
+      "step": 11820
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.239508867263794,
+      "learning_rate": 7.245023060307083e-06,
+      "loss": 0.731,
+      "step": 11830
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.495964527130127,
+      "learning_rate": 7.23334695545566e-06,
+      "loss": 0.6984,
+      "step": 11840
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0980846881866455,
+      "learning_rate": 7.221670850604239e-06,
+      "loss": 0.7146,
+      "step": 11850
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.7728438377380371,
+      "eval_runtime": 0.6082,
+      "eval_samples_per_second": 16.442,
+      "eval_steps_per_second": 3.288,
+      "step": 11850
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.081796407699585,
+      "learning_rate": 7.209994745752818e-06,
+      "loss": 0.6965,
+      "step": 11860
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.2473373413085938,
+      "learning_rate": 7.198318640901396e-06,
+      "loss": 0.7088,
+      "step": 11870
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.0884206295013428,
+      "learning_rate": 7.186642536049975e-06,
+      "loss": 0.6632,
+      "step": 11880
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.202669382095337,
+      "learning_rate": 7.174966431198552e-06,
+      "loss": 0.719,
+      "step": 11890
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.127920627593994,
+      "learning_rate": 7.1632903263471306e-06,
+      "loss": 0.7177,
+      "step": 11900
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.7751314043998718,
+      "eval_runtime": 0.6134,
+      "eval_samples_per_second": 16.303,
+      "eval_steps_per_second": 3.261,
+      "step": 11900
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.230912446975708,
+      "learning_rate": 7.15161422149571e-06,
+      "loss": 0.6948,
+      "step": 11910
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1798815727233887,
+      "learning_rate": 7.139938116644289e-06,
+      "loss": 0.7205,
+      "step": 11920
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1466784477233887,
+      "learning_rate": 7.128262011792867e-06,
+      "loss": 0.6878,
+      "step": 11930
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1586403846740723,
+      "learning_rate": 7.116585906941444e-06,
+      "loss": 0.7074,
+      "step": 11940
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1542489528656006,
+      "learning_rate": 7.104909802090023e-06,
+      "loss": 0.717,
+      "step": 11950
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 0.774330735206604,
+      "eval_runtime": 0.6099,
+      "eval_samples_per_second": 16.396,
+      "eval_steps_per_second": 3.279,
+      "step": 11950
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.124574661254883,
+      "learning_rate": 7.093233697238601e-06,
+      "loss": 0.6846,
+      "step": 11960
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1661715507507324,
+      "learning_rate": 7.08155759238718e-06,
+      "loss": 0.7508,
+      "step": 11970
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.181657552719116,
+      "learning_rate": 7.0698814875357595e-06,
+      "loss": 0.6979,
+      "step": 11980
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 2.1909782886505127,
+      "learning_rate": 7.058205382684336e-06,
+      "loss": 0.7057,
+      "step": 11990
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.262730121612549,
+      "learning_rate": 7.046529277832915e-06,
+      "loss": 0.6971,
+      "step": 12000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.7725943922996521,
+      "eval_runtime": 0.6156,
+      "eval_samples_per_second": 16.244,
+      "eval_steps_per_second": 3.249,
+      "step": 12000
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1330387592315674,
+      "learning_rate": 7.034853172981494e-06,
+      "loss": 0.6944,
+      "step": 12010
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.210728168487549,
+      "learning_rate": 7.023177068130072e-06,
+      "loss": 0.7061,
+      "step": 12020
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9364556074142456,
+      "learning_rate": 7.011500963278651e-06,
+      "loss": 0.6972,
+      "step": 12030
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.0470292568206787,
+      "learning_rate": 6.999824858427229e-06,
+      "loss": 0.7237,
+      "step": 12040
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2245473861694336,
+      "learning_rate": 6.988148753575807e-06,
+      "loss": 0.6862,
+      "step": 12050
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.7745167016983032,
+      "eval_runtime": 0.6096,
+      "eval_samples_per_second": 16.404,
+      "eval_steps_per_second": 3.281,
+      "step": 12050
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1520209312438965,
+      "learning_rate": 6.976472648724386e-06,
+      "loss": 0.6932,
+      "step": 12060
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3901636600494385,
+      "learning_rate": 6.9647965438729645e-06,
+      "loss": 0.7025,
+      "step": 12070
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.34367299079895,
+      "learning_rate": 6.953120439021543e-06,
+      "loss": 0.7264,
+      "step": 12080
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.047297954559326,
+      "learning_rate": 6.941444334170121e-06,
+      "loss": 0.684,
+      "step": 12090
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.372681140899658,
+      "learning_rate": 6.9297682293186995e-06,
+      "loss": 0.7295,
+      "step": 12100
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.7738874554634094,
+      "eval_runtime": 0.6192,
+      "eval_samples_per_second": 16.151,
+      "eval_steps_per_second": 3.23,
+      "step": 12100
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2460813522338867,
+      "learning_rate": 6.918092124467278e-06,
+      "loss": 0.6991,
+      "step": 12110
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.9521677494049072,
+      "learning_rate": 6.906416019615857e-06,
+      "loss": 0.7003,
+      "step": 12120
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1929900646209717,
+      "learning_rate": 6.894739914764435e-06,
+      "loss": 0.7304,
+      "step": 12130
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.2654876708984375,
+      "learning_rate": 6.883063809913013e-06,
+      "loss": 0.6787,
+      "step": 12140
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.1039581298828125,
+      "learning_rate": 6.871387705061592e-06,
+      "loss": 0.6842,
+      "step": 12150
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.7760211229324341,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.364,
+      "eval_steps_per_second": 3.273,
+      "step": 12150
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.3137083053588867,
+      "learning_rate": 6.85971160021017e-06,
+      "loss": 0.7084,
+      "step": 12160
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 2.123534679412842,
+      "learning_rate": 6.848035495358749e-06,
+      "loss": 0.708,
+      "step": 12170
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.113889455795288,
+      "learning_rate": 6.836359390507328e-06,
+      "loss": 0.6973,
+      "step": 12180
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.2173068523406982,
+      "learning_rate": 6.824683285655905e-06,
+      "loss": 0.7046,
+      "step": 12190
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3941967487335205,
+      "learning_rate": 6.813007180804484e-06,
+      "loss": 0.738,
+      "step": 12200
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 0.7759495973587036,
+      "eval_runtime": 0.6083,
+      "eval_samples_per_second": 16.439,
+      "eval_steps_per_second": 3.288,
+      "step": 12200
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3543057441711426,
+      "learning_rate": 6.801331075953063e-06,
+      "loss": 0.716,
+      "step": 12210
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.503976821899414,
+      "learning_rate": 6.789654971101641e-06,
+      "loss": 0.7195,
+      "step": 12220
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.305551767349243,
+      "learning_rate": 6.77797886625022e-06,
+      "loss": 0.6834,
+      "step": 12230
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4071240425109863,
+      "learning_rate": 6.766302761398798e-06,
+      "loss": 0.6749,
+      "step": 12240
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0897507667541504,
+      "learning_rate": 6.754626656547376e-06,
+      "loss": 0.6944,
+      "step": 12250
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 0.7751011848449707,
+      "eval_runtime": 0.6143,
+      "eval_samples_per_second": 16.278,
+      "eval_steps_per_second": 3.256,
+      "step": 12250
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.4181439876556396,
+      "learning_rate": 6.742950551695955e-06,
+      "loss": 0.6987,
+      "step": 12260
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.3234472274780273,
+      "learning_rate": 6.7312744468445334e-06,
+      "loss": 0.7128,
+      "step": 12270
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.318398952484131,
+      "learning_rate": 6.719598341993112e-06,
+      "loss": 0.6907,
+      "step": 12280
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.069641590118408,
+      "learning_rate": 6.70792223714169e-06,
+      "loss": 0.7062,
+      "step": 12290
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.288182258605957,
+      "learning_rate": 6.6962461322902684e-06,
+      "loss": 0.7493,
+      "step": 12300
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 0.7724893689155579,
+      "eval_runtime": 0.6062,
+      "eval_samples_per_second": 16.496,
+      "eval_steps_per_second": 3.299,
+      "step": 12300
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.368288516998291,
+      "learning_rate": 6.684570027438847e-06,
+      "loss": 0.7434,
+      "step": 12310
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.0410752296447754,
+      "learning_rate": 6.672893922587426e-06,
+      "loss": 0.6901,
+      "step": 12320
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.380566358566284,
+      "learning_rate": 6.661217817736004e-06,
+      "loss": 0.6755,
+      "step": 12330
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.230620861053467,
+      "learning_rate": 6.649541712884582e-06,
+      "loss": 0.7043,
+      "step": 12340
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.233041286468506,
+      "learning_rate": 6.637865608033161e-06,
+      "loss": 0.7436,
+      "step": 12350
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 0.7700685262680054,
+      "eval_runtime": 0.605,
+      "eval_samples_per_second": 16.528,
+      "eval_steps_per_second": 3.306,
+      "step": 12350
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.0960350036621094,
+      "learning_rate": 6.626189503181739e-06,
+      "loss": 0.6992,
+      "step": 12360
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.196409225463867,
+      "learning_rate": 6.614513398330318e-06,
+      "loss": 0.7023,
+      "step": 12370
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3462181091308594,
+      "learning_rate": 6.6028372934788965e-06,
+      "loss": 0.6904,
+      "step": 12380
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2048490047454834,
+      "learning_rate": 6.591161188627474e-06,
+      "loss": 0.7306,
+      "step": 12390
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.303339719772339,
+      "learning_rate": 6.579485083776053e-06,
+      "loss": 0.687,
+      "step": 12400
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.7708209753036499,
+      "eval_runtime": 0.6118,
+      "eval_samples_per_second": 16.344,
+      "eval_steps_per_second": 3.269,
+      "step": 12400
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1657707691192627,
+      "learning_rate": 6.5678089789246315e-06,
+      "loss": 0.7304,
+      "step": 12410
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.290966749191284,
+      "learning_rate": 6.55613287407321e-06,
+      "loss": 0.7477,
+      "step": 12420
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.229321002960205,
+      "learning_rate": 6.544456769221789e-06,
+      "loss": 0.6867,
+      "step": 12430
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.1718966960906982,
+      "learning_rate": 6.5327806643703665e-06,
+      "loss": 0.7005,
+      "step": 12440
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.8991645574569702,
+      "learning_rate": 6.521104559518945e-06,
+      "loss": 0.7115,
+      "step": 12450
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.7713428735733032,
+      "eval_runtime": 0.6118,
+      "eval_samples_per_second": 16.346,
+      "eval_steps_per_second": 3.269,
+      "step": 12450
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.356743574142456,
+      "learning_rate": 6.509428454667524e-06,
+      "loss": 0.6928,
+      "step": 12460
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.391080856323242,
+      "learning_rate": 6.497752349816102e-06,
+      "loss": 0.7123,
+      "step": 12470
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3524560928344727,
+      "learning_rate": 6.486076244964681e-06,
+      "loss": 0.7228,
+      "step": 12480
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.071388006210327,
+      "learning_rate": 6.474400140113258e-06,
+      "loss": 0.6838,
+      "step": 12490
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2569472789764404,
+      "learning_rate": 6.462724035261837e-06,
+      "loss": 0.6942,
+      "step": 12500
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.7710386514663696,
+      "eval_runtime": 0.6149,
+      "eval_samples_per_second": 16.262,
+      "eval_steps_per_second": 3.252,
+      "step": 12500
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.3969180583953857,
+      "learning_rate": 6.451047930410416e-06,
+      "loss": 0.7192,
+      "step": 12510
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.280691623687744,
+      "learning_rate": 6.439371825558995e-06,
+      "loss": 0.7442,
+      "step": 12520
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 2.2526988983154297,
+      "learning_rate": 6.427695720707573e-06,
+      "loss": 0.6943,
+      "step": 12530
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1527631282806396,
+      "learning_rate": 6.41601961585615e-06,
+      "loss": 0.7216,
+      "step": 12540
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.269937515258789,
+      "learning_rate": 6.404343511004729e-06,
+      "loss": 0.7348,
+      "step": 12550
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.7706794738769531,
+      "eval_runtime": 0.606,
+      "eval_samples_per_second": 16.502,
+      "eval_steps_per_second": 3.3,
+      "step": 12550
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.313560962677002,
+      "learning_rate": 6.392667406153308e-06,
+      "loss": 0.6889,
+      "step": 12560
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2815611362457275,
+      "learning_rate": 6.380991301301887e-06,
+      "loss": 0.7075,
+      "step": 12570
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3174023628234863,
+      "learning_rate": 6.3693151964504655e-06,
+      "loss": 0.7255,
+      "step": 12580
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1393978595733643,
+      "learning_rate": 6.357639091599042e-06,
+      "loss": 0.6871,
+      "step": 12590
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.259615182876587,
+      "learning_rate": 6.345962986747621e-06,
+      "loss": 0.7622,
+      "step": 12600
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.7713140249252319,
+      "eval_runtime": 0.6073,
+      "eval_samples_per_second": 16.466,
+      "eval_steps_per_second": 3.293,
+      "step": 12600
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0011301040649414,
+      "learning_rate": 6.3342868818962e-06,
+      "loss": 0.7565,
+      "step": 12610
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2515501976013184,
+      "learning_rate": 6.322610777044778e-06,
+      "loss": 0.7127,
+      "step": 12620
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1811752319335938,
+      "learning_rate": 6.310934672193358e-06,
+      "loss": 0.6794,
+      "step": 12630
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.523432970046997,
+      "learning_rate": 6.299258567341935e-06,
+      "loss": 0.7168,
+      "step": 12640
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.2452480792999268,
+      "learning_rate": 6.287582462490513e-06,
+      "loss": 0.689,
+      "step": 12650
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.7710505723953247,
+      "eval_runtime": 0.6093,
+      "eval_samples_per_second": 16.412,
+      "eval_steps_per_second": 3.282,
+      "step": 12650
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.106053113937378,
+      "learning_rate": 6.275906357639092e-06,
+      "loss": 0.7786,
+      "step": 12660
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.118978500366211,
+      "learning_rate": 6.2642302527876705e-06,
+      "loss": 0.7151,
+      "step": 12670
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.3811709880828857,
+      "learning_rate": 6.252554147936249e-06,
+      "loss": 0.7047,
+      "step": 12680
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1718153953552246,
+      "learning_rate": 6.240878043084827e-06,
+      "loss": 0.6959,
+      "step": 12690
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.1322410106658936,
+      "learning_rate": 6.2292019382334055e-06,
+      "loss": 0.7623,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.7712836265563965,
+      "eval_runtime": 0.6059,
+      "eval_samples_per_second": 16.504,
+      "eval_steps_per_second": 3.301,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 2.0430803298950195,
+      "learning_rate": 6.217525833381984e-06,
+      "loss": 0.6896,
+      "step": 12710
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.4313571453094482,
+      "learning_rate": 6.205849728530563e-06,
+      "loss": 0.6691,
+      "step": 12720
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.27482271194458,
+      "learning_rate": 6.194173623679141e-06,
+      "loss": 0.6921,
+      "step": 12730
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.138463020324707,
+      "learning_rate": 6.182497518827719e-06,
+      "loss": 0.7264,
+      "step": 12740
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.043959617614746,
+      "learning_rate": 6.170821413976298e-06,
+      "loss": 0.6616,
+      "step": 12750
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.7710797190666199,
+      "eval_runtime": 0.6117,
+      "eval_samples_per_second": 16.348,
+      "eval_steps_per_second": 3.27,
+      "step": 12750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1632416248321533,
+      "learning_rate": 6.159145309124876e-06,
+      "loss": 0.6957,
+      "step": 12760
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.511916160583496,
+      "learning_rate": 6.147469204273455e-06,
+      "loss": 0.7367,
+      "step": 12770
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.2605438232421875,
+      "learning_rate": 6.1357930994220336e-06,
+      "loss": 0.7038,
+      "step": 12780
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.1245529651641846,
+      "learning_rate": 6.124116994570611e-06,
+      "loss": 0.6872,
+      "step": 12790
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.151524782180786,
+      "learning_rate": 6.11244088971919e-06,
+      "loss": 0.7689,
+      "step": 12800
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.7710217833518982,
+      "eval_runtime": 0.6159,
+      "eval_samples_per_second": 16.237,
+      "eval_steps_per_second": 3.247,
+      "step": 12800
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.028862714767456,
+      "learning_rate": 6.1007647848677686e-06,
+      "loss": 0.7026,
+      "step": 12810
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0894641876220703,
+      "learning_rate": 6.089088680016347e-06,
+      "loss": 0.7414,
+      "step": 12820
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.159057855606079,
+      "learning_rate": 6.077412575164926e-06,
+      "loss": 0.6877,
+      "step": 12830
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.3790111541748047,
+      "learning_rate": 6.0657364703135036e-06,
+      "loss": 0.7179,
+      "step": 12840
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.406184434890747,
+      "learning_rate": 6.054060365462082e-06,
+      "loss": 0.679,
+      "step": 12850
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 0.7704259157180786,
+      "eval_runtime": 0.6118,
+      "eval_samples_per_second": 16.345,
+      "eval_steps_per_second": 3.269,
+      "step": 12850
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.222783088684082,
+      "learning_rate": 6.042384260610661e-06,
+      "loss": 0.7387,
+      "step": 12860
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.099741220474243,
+      "learning_rate": 6.0307081557592394e-06,
+      "loss": 0.6864,
+      "step": 12870
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.194932699203491,
+      "learning_rate": 6.019032050907818e-06,
+      "loss": 0.711,
+      "step": 12880
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.0382237434387207,
+      "learning_rate": 6.007355946056396e-06,
+      "loss": 0.7326,
+      "step": 12890
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.1373980045318604,
+      "learning_rate": 5.9956798412049744e-06,
+      "loss": 0.7285,
+      "step": 12900
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.7717138528823853,
+      "eval_runtime": 0.6081,
+      "eval_samples_per_second": 16.445,
+      "eval_steps_per_second": 3.289,
+      "step": 12900
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.0556564331054688,
+      "learning_rate": 5.984003736353553e-06,
+      "loss": 0.7218,
+      "step": 12910
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.093574285507202,
+      "learning_rate": 5.972327631502132e-06,
+      "loss": 0.6991,
+      "step": 12920
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.601991891860962,
+      "learning_rate": 5.96065152665071e-06,
+      "loss": 0.709,
+      "step": 12930
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.408097743988037,
+      "learning_rate": 5.948975421799288e-06,
+      "loss": 0.7813,
+      "step": 12940
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3824334144592285,
+      "learning_rate": 5.937299316947867e-06,
+      "loss": 0.7104,
+      "step": 12950
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.7729339599609375,
+      "eval_runtime": 0.6078,
+      "eval_samples_per_second": 16.452,
+      "eval_steps_per_second": 3.29,
+      "step": 12950
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.087531566619873,
+      "learning_rate": 5.925623212096445e-06,
+      "loss": 0.7228,
+      "step": 12960
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.329350233078003,
+      "learning_rate": 5.913947107245024e-06,
+      "loss": 0.7501,
+      "step": 12970
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.24851655960083,
+      "learning_rate": 5.9022710023936025e-06,
+      "loss": 0.6702,
+      "step": 12980
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.733278751373291,
+      "learning_rate": 5.89059489754218e-06,
+      "loss": 0.7251,
+      "step": 12990
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.375452995300293,
+      "learning_rate": 5.878918792690759e-06,
+      "loss": 0.6788,
+      "step": 13000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.7694094777107239,
+      "eval_runtime": 0.61,
+      "eval_samples_per_second": 16.392,
+      "eval_steps_per_second": 3.278,
+      "step": 13000
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2695107460021973,
+      "learning_rate": 5.8672426878393375e-06,
+      "loss": 0.6994,
+      "step": 13010
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.16701602935791,
+      "learning_rate": 5.855566582987916e-06,
+      "loss": 0.7667,
+      "step": 13020
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.4114081859588623,
+      "learning_rate": 5.843890478136495e-06,
+      "loss": 0.702,
+      "step": 13030
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.242604970932007,
+      "learning_rate": 5.8322143732850725e-06,
+      "loss": 0.6607,
+      "step": 13040
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.153864622116089,
+      "learning_rate": 5.820538268433651e-06,
+      "loss": 0.7443,
+      "step": 13050
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.7701629400253296,
+      "eval_runtime": 0.6072,
+      "eval_samples_per_second": 16.469,
+      "eval_steps_per_second": 3.294,
+      "step": 13050
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.3688347339630127,
+      "learning_rate": 5.80886216358223e-06,
+      "loss": 0.7332,
+      "step": 13060
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.2412540912628174,
+      "learning_rate": 5.797186058730808e-06,
+      "loss": 0.7174,
+      "step": 13070
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.194040298461914,
+      "learning_rate": 5.785509953879387e-06,
+      "loss": 0.712,
+      "step": 13080
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4642629623413086,
+      "learning_rate": 5.773833849027965e-06,
+      "loss": 0.6822,
+      "step": 13090
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4387590885162354,
+      "learning_rate": 5.762157744176543e-06,
+      "loss": 0.7685,
+      "step": 13100
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.7677481174468994,
+      "eval_runtime": 0.6099,
+      "eval_samples_per_second": 16.397,
+      "eval_steps_per_second": 3.279,
+      "step": 13100
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.078892469406128,
+      "learning_rate": 5.750481639325122e-06,
+      "loss": 0.6974,
+      "step": 13110
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1985135078430176,
+      "learning_rate": 5.738805534473701e-06,
+      "loss": 0.7124,
+      "step": 13120
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3607141971588135,
+      "learning_rate": 5.727129429622279e-06,
+      "loss": 0.6914,
+      "step": 13130
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3022196292877197,
+      "learning_rate": 5.716620935255999e-06,
+      "loss": 0.6556,
+      "step": 13140
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.9745125770568848,
+      "learning_rate": 5.704944830404577e-06,
+      "loss": 0.7129,
+      "step": 13150
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.7671664357185364,
+      "eval_runtime": 0.6088,
+      "eval_samples_per_second": 16.425,
+      "eval_steps_per_second": 3.285,
+      "step": 13150
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2131617069244385,
+      "learning_rate": 5.693268725553156e-06,
+      "loss": 0.692,
+      "step": 13160
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.3504748344421387,
+      "learning_rate": 5.6815926207017345e-06,
+      "loss": 0.701,
+      "step": 13170
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.059758186340332,
+      "learning_rate": 5.669916515850313e-06,
+      "loss": 0.709,
+      "step": 13180
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.4684035778045654,
+      "learning_rate": 5.658240410998891e-06,
+      "loss": 0.6764,
+      "step": 13190
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.1126599311828613,
+      "learning_rate": 5.6465643061474695e-06,
+      "loss": 0.7048,
+      "step": 13200
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.7675765156745911,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.369,
+      "eval_steps_per_second": 3.274,
+      "step": 13200
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.5027501583099365,
+      "learning_rate": 5.634888201296048e-06,
+      "loss": 0.7054,
+      "step": 13210
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.2247512340545654,
+      "learning_rate": 5.623212096444627e-06,
+      "loss": 0.7044,
+      "step": 13220
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.0924768447875977,
+      "learning_rate": 5.611535991593205e-06,
+      "loss": 0.717,
+      "step": 13230
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.293689250946045,
+      "learning_rate": 5.599859886741783e-06,
+      "loss": 0.7132,
+      "step": 13240
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 2.292529821395874,
+      "learning_rate": 5.588183781890362e-06,
+      "loss": 0.7077,
+      "step": 13250
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.7669377326965332,
+      "eval_runtime": 0.6088,
+      "eval_samples_per_second": 16.426,
+      "eval_steps_per_second": 3.285,
+      "step": 13250
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.691652774810791,
+      "learning_rate": 5.57650767703894e-06,
+      "loss": 0.7066,
+      "step": 13260
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3305978775024414,
+      "learning_rate": 5.564831572187519e-06,
+      "loss": 0.6641,
+      "step": 13270
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.373014211654663,
+      "learning_rate": 5.5531554673360976e-06,
+      "loss": 0.7207,
+      "step": 13280
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.271946430206299,
+      "learning_rate": 5.541479362484675e-06,
+      "loss": 0.7052,
+      "step": 13290
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.16314697265625,
+      "learning_rate": 5.529803257633254e-06,
+      "loss": 0.7023,
+      "step": 13300
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.768413245677948,
+      "eval_runtime": 0.604,
+      "eval_samples_per_second": 16.557,
+      "eval_steps_per_second": 3.311,
+      "step": 13300
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.201843500137329,
+      "learning_rate": 5.5181271527818326e-06,
+      "loss": 0.6893,
+      "step": 13310
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3157248497009277,
+      "learning_rate": 5.506451047930411e-06,
+      "loss": 0.7158,
+      "step": 13320
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3228063583374023,
+      "learning_rate": 5.49477494307899e-06,
+      "loss": 0.681,
+      "step": 13330
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.250251293182373,
+      "learning_rate": 5.4830988382275676e-06,
+      "loss": 0.7163,
+      "step": 13340
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.1168575286865234,
+      "learning_rate": 5.471422733376146e-06,
+      "loss": 0.6742,
+      "step": 13350
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.7682064771652222,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.364,
+      "eval_steps_per_second": 3.273,
+      "step": 13350
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4983458518981934,
+      "learning_rate": 5.459746628524725e-06,
+      "loss": 0.6906,
+      "step": 13360
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4522924423217773,
+      "learning_rate": 5.448070523673303e-06,
+      "loss": 0.7175,
+      "step": 13370
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.9219380617141724,
+      "learning_rate": 5.436394418821882e-06,
+      "loss": 0.72,
+      "step": 13380
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.585495948791504,
+      "learning_rate": 5.42471831397046e-06,
+      "loss": 0.7607,
+      "step": 13390
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.276994466781616,
+      "learning_rate": 5.413042209119038e-06,
+      "loss": 0.6934,
+      "step": 13400
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.7672905325889587,
+      "eval_runtime": 0.6107,
+      "eval_samples_per_second": 16.376,
+      "eval_steps_per_second": 3.275,
+      "step": 13400
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.3009979724884033,
+      "learning_rate": 5.401366104267617e-06,
+      "loss": 0.7319,
+      "step": 13410
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.263928174972534,
+      "learning_rate": 5.389689999416196e-06,
+      "loss": 0.7062,
+      "step": 13420
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 2.4444870948791504,
+      "learning_rate": 5.378013894564774e-06,
+      "loss": 0.7294,
+      "step": 13430
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.229544162750244,
+      "learning_rate": 5.366337789713352e-06,
+      "loss": 0.6731,
+      "step": 13440
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2096409797668457,
+      "learning_rate": 5.354661684861931e-06,
+      "loss": 0.6924,
+      "step": 13450
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.7676502466201782,
+      "eval_runtime": 0.6106,
+      "eval_samples_per_second": 16.378,
+      "eval_steps_per_second": 3.276,
+      "step": 13450
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.405052661895752,
+      "learning_rate": 5.342985580010509e-06,
+      "loss": 0.7344,
+      "step": 13460
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.253854990005493,
+      "learning_rate": 5.331309475159088e-06,
+      "loss": 0.725,
+      "step": 13470
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.9766682386398315,
+      "learning_rate": 5.319633370307666e-06,
+      "loss": 0.7025,
+      "step": 13480
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.090888023376465,
+      "learning_rate": 5.307957265456244e-06,
+      "loss": 0.6559,
+      "step": 13490
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.5892157554626465,
+      "learning_rate": 5.296281160604823e-06,
+      "loss": 0.6824,
+      "step": 13500
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.7686687111854553,
+      "eval_runtime": 0.6085,
+      "eval_samples_per_second": 16.435,
+      "eval_steps_per_second": 3.287,
+      "step": 13500
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0840067863464355,
+      "learning_rate": 5.2846050557534015e-06,
+      "loss": 0.7096,
+      "step": 13510
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.511174440383911,
+      "learning_rate": 5.27292895090198e-06,
+      "loss": 0.7058,
+      "step": 13520
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.0443196296691895,
+      "learning_rate": 5.261252846050557e-06,
+      "loss": 0.6954,
+      "step": 13530
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2680811882019043,
+      "learning_rate": 5.249576741199136e-06,
+      "loss": 0.6982,
+      "step": 13540
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.2729101181030273,
+      "learning_rate": 5.237900636347715e-06,
+      "loss": 0.7024,
+      "step": 13550
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.7688701748847961,
+      "eval_runtime": 0.6104,
+      "eval_samples_per_second": 16.383,
+      "eval_steps_per_second": 3.277,
+      "step": 13550
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.3392393589019775,
+      "learning_rate": 5.226224531496294e-06,
+      "loss": 0.6944,
+      "step": 13560
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.354621648788452,
+      "learning_rate": 5.214548426644872e-06,
+      "loss": 0.6867,
+      "step": 13570
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.300445079803467,
+      "learning_rate": 5.202872321793449e-06,
+      "loss": 0.7231,
+      "step": 13580
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.367570161819458,
+      "learning_rate": 5.191196216942028e-06,
+      "loss": 0.6867,
+      "step": 13590
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.367478847503662,
+      "learning_rate": 5.1795201120906065e-06,
+      "loss": 0.6934,
+      "step": 13600
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.7668425440788269,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.342,
+      "eval_steps_per_second": 3.268,
+      "step": 13600
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 2.597877025604248,
+      "learning_rate": 5.167844007239186e-06,
+      "loss": 0.7029,
+      "step": 13610
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.173400402069092,
+      "learning_rate": 5.156167902387765e-06,
+      "loss": 0.7204,
+      "step": 13620
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.024242877960205,
+      "learning_rate": 5.1444917975363415e-06,
+      "loss": 0.7283,
+      "step": 13630
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3758063316345215,
+      "learning_rate": 5.13281569268492e-06,
+      "loss": 0.7129,
+      "step": 13640
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0525763034820557,
+      "learning_rate": 5.121139587833499e-06,
+      "loss": 0.688,
+      "step": 13650
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.7670071721076965,
+      "eval_runtime": 0.6068,
+      "eval_samples_per_second": 16.481,
+      "eval_steps_per_second": 3.296,
+      "step": 13650
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.160485029220581,
+      "learning_rate": 5.109463482982077e-06,
+      "loss": 0.7258,
+      "step": 13660
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.0960161685943604,
+      "learning_rate": 5.097787378130656e-06,
+      "loss": 0.6775,
+      "step": 13670
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 3.0020017623901367,
+      "learning_rate": 5.086111273279234e-06,
+      "loss": 0.719,
+      "step": 13680
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3923168182373047,
+      "learning_rate": 5.074435168427812e-06,
+      "loss": 0.6991,
+      "step": 13690
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.253556489944458,
+      "learning_rate": 5.062759063576391e-06,
+      "loss": 0.7308,
+      "step": 13700
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.7692948579788208,
+      "eval_runtime": 0.6091,
+      "eval_samples_per_second": 16.419,
+      "eval_steps_per_second": 3.284,
+      "step": 13700
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.618200659751892,
+      "learning_rate": 5.05108295872497e-06,
+      "loss": 0.7518,
+      "step": 13710
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.124499797821045,
+      "learning_rate": 5.039406853873548e-06,
+      "loss": 0.7051,
+      "step": 13720
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.234586715698242,
+      "learning_rate": 5.027730749022126e-06,
+      "loss": 0.7409,
+      "step": 13730
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.3138747215270996,
+      "learning_rate": 5.016054644170705e-06,
+      "loss": 0.6924,
+      "step": 13740
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.539057493209839,
+      "learning_rate": 5.004378539319283e-06,
+      "loss": 0.6888,
+      "step": 13750
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.7695175409317017,
+      "eval_runtime": 0.6009,
+      "eval_samples_per_second": 16.642,
+      "eval_steps_per_second": 3.328,
+      "step": 13750
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.5522680282592773,
+      "learning_rate": 4.992702434467862e-06,
+      "loss": 0.7677,
+      "step": 13760
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.1782872676849365,
+      "learning_rate": 4.9810263296164404e-06,
+      "loss": 0.7183,
+      "step": 13770
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.283682346343994,
+      "learning_rate": 4.969350224765019e-06,
+      "loss": 0.7163,
+      "step": 13780
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 2.341571092605591,
+      "learning_rate": 4.957674119913598e-06,
+      "loss": 0.695,
+      "step": 13790
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.033668279647827,
+      "learning_rate": 4.9459980150621755e-06,
+      "loss": 0.7361,
+      "step": 13800
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.769692063331604,
+      "eval_runtime": 0.6075,
+      "eval_samples_per_second": 16.46,
+      "eval_steps_per_second": 3.292,
+      "step": 13800
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.2978110313415527,
+      "learning_rate": 4.934321910210754e-06,
+      "loss": 0.6923,
+      "step": 13810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.347923517227173,
+      "learning_rate": 4.922645805359333e-06,
+      "loss": 0.6993,
+      "step": 13820
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.262248992919922,
+      "learning_rate": 4.910969700507911e-06,
+      "loss": 0.7187,
+      "step": 13830
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.5264322757720947,
+      "learning_rate": 4.89929359565649e-06,
+      "loss": 0.7135,
+      "step": 13840
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1052310466766357,
+      "learning_rate": 4.887617490805068e-06,
+      "loss": 0.7777,
+      "step": 13850
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.7693582773208618,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.339,
+      "eval_steps_per_second": 3.268,
+      "step": 13850
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.252171039581299,
+      "learning_rate": 4.875941385953646e-06,
+      "loss": 0.7334,
+      "step": 13860
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.8138601779937744,
+      "learning_rate": 4.864265281102225e-06,
+      "loss": 0.7068,
+      "step": 13870
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.432004928588867,
+      "learning_rate": 4.8525891762508035e-06,
+      "loss": 0.7003,
+      "step": 13880
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.887652039527893,
+      "learning_rate": 4.840913071399382e-06,
+      "loss": 0.735,
+      "step": 13890
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1489367485046387,
+      "learning_rate": 4.82923696654796e-06,
+      "loss": 0.7355,
+      "step": 13900
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.7675796747207642,
+      "eval_runtime": 0.6147,
+      "eval_samples_per_second": 16.268,
+      "eval_steps_per_second": 3.254,
+      "step": 13900
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.31229829788208,
+      "learning_rate": 4.8175608616965385e-06,
+      "loss": 0.695,
+      "step": 13910
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.1557681560516357,
+      "learning_rate": 4.805884756845116e-06,
+      "loss": 0.7077,
+      "step": 13920
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.128478527069092,
+      "learning_rate": 4.794208651993695e-06,
+      "loss": 0.658,
+      "step": 13930
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.045618772506714,
+      "learning_rate": 4.782532547142274e-06,
+      "loss": 0.7205,
+      "step": 13940
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.4439361095428467,
+      "learning_rate": 4.770856442290852e-06,
+      "loss": 0.697,
+      "step": 13950
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.7679734230041504,
+      "eval_runtime": 0.607,
+      "eval_samples_per_second": 16.475,
+      "eval_steps_per_second": 3.295,
+      "step": 13950
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.213278293609619,
+      "learning_rate": 4.759180337439431e-06,
+      "loss": 0.7461,
+      "step": 13960
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.058229446411133,
+      "learning_rate": 4.7475042325880085e-06,
+      "loss": 0.7141,
+      "step": 13970
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0028069019317627,
+      "learning_rate": 4.735828127736587e-06,
+      "loss": 0.7051,
+      "step": 13980
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0091359615325928,
+      "learning_rate": 4.724152022885166e-06,
+      "loss": 0.7325,
+      "step": 13990
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.373077392578125,
+      "learning_rate": 4.712475918033744e-06,
+      "loss": 0.6879,
+      "step": 14000
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 0.7674081325531006,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.39,
+      "eval_steps_per_second": 3.278,
+      "step": 14000
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.163534641265869,
+      "learning_rate": 4.700799813182323e-06,
+      "loss": 0.7315,
+      "step": 14010
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0593481063842773,
+      "learning_rate": 4.689123708330901e-06,
+      "loss": 0.7354,
+      "step": 14020
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.272758722305298,
+      "learning_rate": 4.677447603479479e-06,
+      "loss": 0.7168,
+      "step": 14030
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.26145076751709,
+      "learning_rate": 4.665771498628058e-06,
+      "loss": 0.6862,
+      "step": 14040
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.1342623233795166,
+      "learning_rate": 4.654095393776637e-06,
+      "loss": 0.6853,
+      "step": 14050
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 0.7673859596252441,
+      "eval_runtime": 0.6081,
+      "eval_samples_per_second": 16.444,
+      "eval_steps_per_second": 3.289,
+      "step": 14050
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.5347769260406494,
+      "learning_rate": 4.642419288925215e-06,
+      "loss": 0.6906,
+      "step": 14060
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.9729682207107544,
+      "learning_rate": 4.630743184073793e-06,
+      "loss": 0.7155,
+      "step": 14070
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.4124653339385986,
+      "learning_rate": 4.619067079222372e-06,
+      "loss": 0.6987,
+      "step": 14080
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.3350820541381836,
+      "learning_rate": 4.60739097437095e-06,
+      "loss": 0.737,
+      "step": 14090
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.0715298652648926,
+      "learning_rate": 4.595714869519529e-06,
+      "loss": 0.6923,
+      "step": 14100
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 0.7669563293457031,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.358,
+      "eval_steps_per_second": 3.272,
+      "step": 14100
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.503970146179199,
+      "learning_rate": 4.5840387646681075e-06,
+      "loss": 0.697,
+      "step": 14110
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.2708277702331543,
+      "learning_rate": 4.572362659816685e-06,
+      "loss": 0.6729,
+      "step": 14120
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.153205633163452,
+      "learning_rate": 4.560686554965264e-06,
+      "loss": 0.6542,
+      "step": 14130
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.460183620452881,
+      "learning_rate": 4.5490104501138425e-06,
+      "loss": 0.712,
+      "step": 14140
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 2.129221200942993,
+      "learning_rate": 4.537334345262421e-06,
+      "loss": 0.7151,
+      "step": 14150
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 0.7688795328140259,
+      "eval_runtime": 0.61,
+      "eval_samples_per_second": 16.392,
+      "eval_steps_per_second": 3.278,
+      "step": 14150
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.360013008117676,
+      "learning_rate": 4.525658240411e-06,
+      "loss": 0.669,
+      "step": 14160
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.220984935760498,
+      "learning_rate": 4.5139821355595775e-06,
+      "loss": 0.6882,
+      "step": 14170
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3665828704833984,
+      "learning_rate": 4.502306030708156e-06,
+      "loss": 0.695,
+      "step": 14180
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3276588916778564,
+      "learning_rate": 4.490629925856734e-06,
+      "loss": 0.7166,
+      "step": 14190
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.1937451362609863,
+      "learning_rate": 4.478953821005313e-06,
+      "loss": 0.6863,
+      "step": 14200
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.7695289850234985,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.368,
+      "eval_steps_per_second": 3.274,
+      "step": 14200
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2716734409332275,
+      "learning_rate": 4.467277716153892e-06,
+      "loss": 0.6889,
+      "step": 14210
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2764086723327637,
+      "learning_rate": 4.45560161130247e-06,
+      "loss": 0.7514,
+      "step": 14220
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.3657469749450684,
+      "learning_rate": 4.443925506451048e-06,
+      "loss": 0.6798,
+      "step": 14230
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0969741344451904,
+      "learning_rate": 4.432249401599626e-06,
+      "loss": 0.7402,
+      "step": 14240
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.34328556060791,
+      "learning_rate": 4.420573296748205e-06,
+      "loss": 0.734,
+      "step": 14250
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.7674993276596069,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.368,
+      "eval_steps_per_second": 3.274,
+      "step": 14250
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.298978090286255,
+      "learning_rate": 4.408897191896784e-06,
+      "loss": 0.7114,
+      "step": 14260
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2818267345428467,
+      "learning_rate": 4.397221087045362e-06,
+      "loss": 0.7058,
+      "step": 14270
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.0929410457611084,
+      "learning_rate": 4.3855449821939406e-06,
+      "loss": 0.7389,
+      "step": 14280
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.476551055908203,
+      "learning_rate": 4.373868877342518e-06,
+      "loss": 0.6844,
+      "step": 14290
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.27970027923584,
+      "learning_rate": 4.362192772491097e-06,
+      "loss": 0.7551,
+      "step": 14300
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.7677081227302551,
+      "eval_runtime": 0.6102,
+      "eval_samples_per_second": 16.388,
+      "eval_steps_per_second": 3.278,
+      "step": 14300
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.2080423831939697,
+      "learning_rate": 4.350516667639676e-06,
+      "loss": 0.7357,
+      "step": 14310
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.055227756500244,
+      "learning_rate": 4.338840562788254e-06,
+      "loss": 0.6946,
+      "step": 14320
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.8364222049713135,
+      "learning_rate": 4.327164457936833e-06,
+      "loss": 0.7187,
+      "step": 14330
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3154220581054688,
+      "learning_rate": 4.315488353085411e-06,
+      "loss": 0.7053,
+      "step": 14340
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.131638765335083,
+      "learning_rate": 4.303812248233989e-06,
+      "loss": 0.6894,
+      "step": 14350
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.7690398097038269,
+      "eval_runtime": 0.6082,
+      "eval_samples_per_second": 16.443,
+      "eval_steps_per_second": 3.289,
+      "step": 14350
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2896642684936523,
+      "learning_rate": 4.292136143382568e-06,
+      "loss": 0.716,
+      "step": 14360
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3948166370391846,
+      "learning_rate": 4.2804600385311464e-06,
+      "loss": 0.6797,
+      "step": 14370
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.265130043029785,
+      "learning_rate": 4.268783933679725e-06,
+      "loss": 0.7376,
+      "step": 14380
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6577965021133423,
+      "learning_rate": 4.257107828828303e-06,
+      "loss": 0.7351,
+      "step": 14390
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.0141537189483643,
+      "learning_rate": 4.2454317239768814e-06,
+      "loss": 0.685,
+      "step": 14400
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.7685635685920715,
+      "eval_runtime": 0.6093,
+      "eval_samples_per_second": 16.412,
+      "eval_steps_per_second": 3.282,
+      "step": 14400
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.147343397140503,
+      "learning_rate": 4.23375561912546e-06,
+      "loss": 0.6838,
+      "step": 14410
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.465099573135376,
+      "learning_rate": 4.222079514274039e-06,
+      "loss": 0.6954,
+      "step": 14420
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.223701238632202,
+      "learning_rate": 4.210403409422617e-06,
+      "loss": 0.6764,
+      "step": 14430
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2101335525512695,
+      "learning_rate": 4.198727304571195e-06,
+      "loss": 0.7489,
+      "step": 14440
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.6780245304107666,
+      "learning_rate": 4.187051199719774e-06,
+      "loss": 0.7371,
+      "step": 14450
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.7686441540718079,
+      "eval_runtime": 0.6077,
+      "eval_samples_per_second": 16.457,
+      "eval_steps_per_second": 3.291,
+      "step": 14450
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.3456320762634277,
+      "learning_rate": 4.175375094868352e-06,
+      "loss": 0.6842,
+      "step": 14460
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4545953273773193,
+      "learning_rate": 4.163698990016931e-06,
+      "loss": 0.7249,
+      "step": 14470
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.4514949321746826,
+      "learning_rate": 4.1520228851655095e-06,
+      "loss": 0.7373,
+      "step": 14480
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.2098615169525146,
+      "learning_rate": 4.140346780314087e-06,
+      "loss": 0.7026,
+      "step": 14490
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.702678918838501,
+      "learning_rate": 4.128670675462666e-06,
+      "loss": 0.701,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 0.7689488530158997,
+      "eval_runtime": 0.611,
+      "eval_samples_per_second": 16.367,
+      "eval_steps_per_second": 3.273,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.467792272567749,
+      "learning_rate": 4.1169945706112445e-06,
+      "loss": 0.6726,
+      "step": 14510
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.1781463623046875,
+      "learning_rate": 4.105318465759823e-06,
+      "loss": 0.7056,
+      "step": 14520
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.408137083053589,
+      "learning_rate": 4.093642360908402e-06,
+      "loss": 0.706,
+      "step": 14530
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5934770107269287,
+      "learning_rate": 4.0819662560569795e-06,
+      "loss": 0.7262,
+      "step": 14540
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4160211086273193,
+      "learning_rate": 4.070290151205558e-06,
+      "loss": 0.7117,
+      "step": 14550
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.7691187858581543,
+      "eval_runtime": 0.6099,
+      "eval_samples_per_second": 16.396,
+      "eval_steps_per_second": 3.279,
+      "step": 14550
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.4870803356170654,
+      "learning_rate": 4.058614046354137e-06,
+      "loss": 0.6927,
+      "step": 14560
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3014864921569824,
+      "learning_rate": 4.0469379415027145e-06,
+      "loss": 0.668,
+      "step": 14570
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.2323975563049316,
+      "learning_rate": 4.035261836651293e-06,
+      "loss": 0.7485,
+      "step": 14580
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3516135215759277,
+      "learning_rate": 4.023585731799872e-06,
+      "loss": 0.7167,
+      "step": 14590
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5607802867889404,
+      "learning_rate": 4.01190962694845e-06,
+      "loss": 0.6691,
+      "step": 14600
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.7692115902900696,
+      "eval_runtime": 0.6081,
+      "eval_samples_per_second": 16.444,
+      "eval_steps_per_second": 3.289,
+      "step": 14600
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3449532985687256,
+      "learning_rate": 4.000233522097029e-06,
+      "loss": 0.6986,
+      "step": 14610
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.5133309364318848,
+      "learning_rate": 3.988557417245607e-06,
+      "loss": 0.6722,
+      "step": 14620
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.326199531555176,
+      "learning_rate": 3.976881312394185e-06,
+      "loss": 0.6952,
+      "step": 14630
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.0668652057647705,
+      "learning_rate": 3.965205207542764e-06,
+      "loss": 0.7065,
+      "step": 14640
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.095052480697632,
+      "learning_rate": 3.953529102691343e-06,
+      "loss": 0.7028,
+      "step": 14650
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.7689952850341797,
+      "eval_runtime": 0.609,
+      "eval_samples_per_second": 16.419,
+      "eval_steps_per_second": 3.284,
+      "step": 14650
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.426452398300171,
+      "learning_rate": 3.941852997839921e-06,
+      "loss": 0.7088,
+      "step": 14660
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.953078031539917,
+      "learning_rate": 3.930176892988499e-06,
+      "loss": 0.7419,
+      "step": 14670
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.3756022453308105,
+      "learning_rate": 3.918500788137078e-06,
+      "loss": 0.6957,
+      "step": 14680
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 2.135246753692627,
+      "learning_rate": 3.906824683285656e-06,
+      "loss": 0.682,
+      "step": 14690
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3287878036499023,
+      "learning_rate": 3.895148578434235e-06,
+      "loss": 0.7078,
+      "step": 14700
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.7697198987007141,
+      "eval_runtime": 0.6094,
+      "eval_samples_per_second": 16.41,
+      "eval_steps_per_second": 3.282,
+      "step": 14700
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3664565086364746,
+      "learning_rate": 3.8834724735828135e-06,
+      "loss": 0.6667,
+      "step": 14710
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4556448459625244,
+      "learning_rate": 3.871796368731391e-06,
+      "loss": 0.7048,
+      "step": 14720
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.407667636871338,
+      "learning_rate": 3.86012026387997e-06,
+      "loss": 0.6818,
+      "step": 14730
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.371582269668579,
+      "learning_rate": 3.8484441590285485e-06,
+      "loss": 0.7394,
+      "step": 14740
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2947959899902344,
+      "learning_rate": 3.836768054177127e-06,
+      "loss": 0.7145,
+      "step": 14750
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.7697685956954956,
+      "eval_runtime": 0.6129,
+      "eval_samples_per_second": 16.315,
+      "eval_steps_per_second": 3.263,
+      "step": 14750
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4175286293029785,
+      "learning_rate": 3.825091949325706e-06,
+      "loss": 0.6726,
+      "step": 14760
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3653016090393066,
+      "learning_rate": 3.8134158444742835e-06,
+      "loss": 0.6981,
+      "step": 14770
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2792763710021973,
+      "learning_rate": 3.801739739622862e-06,
+      "loss": 0.6833,
+      "step": 14780
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3387086391448975,
+      "learning_rate": 3.7900636347714403e-06,
+      "loss": 0.6462,
+      "step": 14790
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.9829654693603516,
+      "learning_rate": 3.778387529920019e-06,
+      "loss": 0.7072,
+      "step": 14800
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.7703598141670227,
+      "eval_runtime": 0.6107,
+      "eval_samples_per_second": 16.375,
+      "eval_steps_per_second": 3.275,
+      "step": 14800
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3530917167663574,
+      "learning_rate": 3.7667114250685975e-06,
+      "loss": 0.6773,
+      "step": 14810
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4435482025146484,
+      "learning_rate": 3.7550353202171757e-06,
+      "loss": 0.6946,
+      "step": 14820
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.2573187351226807,
+      "learning_rate": 3.7433592153657543e-06,
+      "loss": 0.6521,
+      "step": 14830
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.3538472652435303,
+      "learning_rate": 3.7316831105143325e-06,
+      "loss": 0.7039,
+      "step": 14840
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.4549813270568848,
+      "learning_rate": 3.720007005662911e-06,
+      "loss": 0.724,
+      "step": 14850
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.7711056470870972,
+      "eval_runtime": 0.6099,
+      "eval_samples_per_second": 16.397,
+      "eval_steps_per_second": 3.279,
+      "step": 14850
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.092198371887207,
+      "learning_rate": 3.7083309008114897e-06,
+      "loss": 0.738,
+      "step": 14860
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 2.464451313018799,
+      "learning_rate": 3.696654795960068e-06,
+      "loss": 0.7717,
+      "step": 14870
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.5269601345062256,
+      "learning_rate": 3.6849786911086466e-06,
+      "loss": 0.6854,
+      "step": 14880
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.552213191986084,
+      "learning_rate": 3.674470196742367e-06,
+      "loss": 0.708,
+      "step": 14890
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1545231342315674,
+      "learning_rate": 3.6627940918909454e-06,
+      "loss": 0.704,
+      "step": 14900
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.7699400186538696,
+      "eval_runtime": 0.6137,
+      "eval_samples_per_second": 16.294,
+      "eval_steps_per_second": 3.259,
+      "step": 14900
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.291929244995117,
+      "learning_rate": 3.651117987039524e-06,
+      "loss": 0.6639,
+      "step": 14910
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.063206911087036,
+      "learning_rate": 3.6394418821881022e-06,
+      "loss": 0.6956,
+      "step": 14920
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4872522354125977,
+      "learning_rate": 3.627765777336681e-06,
+      "loss": 0.6863,
+      "step": 14930
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2838339805603027,
+      "learning_rate": 3.616089672485259e-06,
+      "loss": 0.6825,
+      "step": 14940
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.437544584274292,
+      "learning_rate": 3.6044135676338377e-06,
+      "loss": 0.7143,
+      "step": 14950
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.7692040205001831,
+      "eval_runtime": 0.6099,
+      "eval_samples_per_second": 16.397,
+      "eval_steps_per_second": 3.279,
+      "step": 14950
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.1030797958374023,
+      "learning_rate": 3.5927374627824163e-06,
+      "loss": 0.7271,
+      "step": 14960
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.315521240234375,
+      "learning_rate": 3.5810613579309945e-06,
+      "loss": 0.7084,
+      "step": 14970
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2675883769989014,
+      "learning_rate": 3.569385253079573e-06,
+      "loss": 0.704,
+      "step": 14980
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.298865556716919,
+      "learning_rate": 3.5577091482281513e-06,
+      "loss": 0.6963,
+      "step": 14990
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4779374599456787,
+      "learning_rate": 3.54603304337673e-06,
+      "loss": 0.6793,
+      "step": 15000
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.7683581113815308,
+      "eval_runtime": 0.6094,
+      "eval_samples_per_second": 16.409,
+      "eval_steps_per_second": 3.282,
+      "step": 15000
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4356255531311035,
+      "learning_rate": 3.5343569385253085e-06,
+      "loss": 0.7427,
+      "step": 15010
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.4436514377593994,
+      "learning_rate": 3.5226808336738867e-06,
+      "loss": 0.6852,
+      "step": 15020
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2228686809539795,
+      "learning_rate": 3.5110047288224653e-06,
+      "loss": 0.6515,
+      "step": 15030
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2448713779449463,
+      "learning_rate": 3.4993286239710435e-06,
+      "loss": 0.6871,
+      "step": 15040
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 2.2487831115722656,
+      "learning_rate": 3.487652519119622e-06,
+      "loss": 0.7034,
+      "step": 15050
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.7676941156387329,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.364,
+      "eval_steps_per_second": 3.273,
+      "step": 15050
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.270303249359131,
+      "learning_rate": 3.4759764142682007e-06,
+      "loss": 0.6832,
+      "step": 15060
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3696608543395996,
+      "learning_rate": 3.464300309416779e-06,
+      "loss": 0.7462,
+      "step": 15070
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.466967821121216,
+      "learning_rate": 3.4526242045653576e-06,
+      "loss": 0.7434,
+      "step": 15080
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.671295642852783,
+      "learning_rate": 3.4409480997139353e-06,
+      "loss": 0.6664,
+      "step": 15090
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.5042426586151123,
+      "learning_rate": 3.4292719948625144e-06,
+      "loss": 0.6814,
+      "step": 15100
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.7681771516799927,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.415,
+      "eval_steps_per_second": 3.283,
+      "step": 15100
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3382232189178467,
+      "learning_rate": 3.417595890011093e-06,
+      "loss": 0.6815,
+      "step": 15110
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.062124013900757,
+      "learning_rate": 3.4059197851596707e-06,
+      "loss": 0.6997,
+      "step": 15120
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2337443828582764,
+      "learning_rate": 3.39424368030825e-06,
+      "loss": 0.6888,
+      "step": 15130
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2030904293060303,
+      "learning_rate": 3.3825675754568276e-06,
+      "loss": 0.6772,
+      "step": 15140
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.14607834815979,
+      "learning_rate": 3.370891470605406e-06,
+      "loss": 0.7192,
+      "step": 15150
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.768622875213623,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.364,
+      "eval_steps_per_second": 3.273,
+      "step": 15150
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.066704034805298,
+      "learning_rate": 3.359215365753985e-06,
+      "loss": 0.7346,
+      "step": 15160
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.2817320823669434,
+      "learning_rate": 3.347539260902563e-06,
+      "loss": 0.6768,
+      "step": 15170
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.3509557247161865,
+      "learning_rate": 3.3358631560511416e-06,
+      "loss": 0.6681,
+      "step": 15180
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.316417932510376,
+      "learning_rate": 3.32418705119972e-06,
+      "loss": 0.6844,
+      "step": 15190
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.323413372039795,
+      "learning_rate": 3.3125109463482984e-06,
+      "loss": 0.7023,
+      "step": 15200
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.7674654722213745,
+      "eval_runtime": 0.6103,
+      "eval_samples_per_second": 16.385,
+      "eval_steps_per_second": 3.277,
+      "step": 15200
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.1907122135162354,
+      "learning_rate": 3.300834841496877e-06,
+      "loss": 0.7083,
+      "step": 15210
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.241126775741577,
+      "learning_rate": 3.2891587366454552e-06,
+      "loss": 0.6875,
+      "step": 15220
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.020413398742676,
+      "learning_rate": 3.277482631794034e-06,
+      "loss": 0.7082,
+      "step": 15230
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3539483547210693,
+      "learning_rate": 3.265806526942612e-06,
+      "loss": 0.6871,
+      "step": 15240
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.592371940612793,
+      "learning_rate": 3.2541304220911906e-06,
+      "loss": 0.709,
+      "step": 15250
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.7685578465461731,
+      "eval_runtime": 0.6091,
+      "eval_samples_per_second": 16.418,
+      "eval_steps_per_second": 3.284,
+      "step": 15250
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.589547872543335,
+      "learning_rate": 3.2424543172397693e-06,
+      "loss": 0.7048,
+      "step": 15260
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.329296588897705,
+      "learning_rate": 3.2307782123883475e-06,
+      "loss": 0.7011,
+      "step": 15270
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2237114906311035,
+      "learning_rate": 3.219102107536926e-06,
+      "loss": 0.7344,
+      "step": 15280
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3468308448791504,
+      "learning_rate": 3.2074260026855043e-06,
+      "loss": 0.7053,
+      "step": 15290
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.513903856277466,
+      "learning_rate": 3.195749897834083e-06,
+      "loss": 0.6701,
+      "step": 15300
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.7681719064712524,
+      "eval_runtime": 0.6115,
+      "eval_samples_per_second": 16.352,
+      "eval_steps_per_second": 3.27,
+      "step": 15300
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2606444358825684,
+      "learning_rate": 3.1840737929826615e-06,
+      "loss": 0.7192,
+      "step": 15310
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2392754554748535,
+      "learning_rate": 3.1723976881312397e-06,
+      "loss": 0.6823,
+      "step": 15320
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2624001502990723,
+      "learning_rate": 3.1607215832798183e-06,
+      "loss": 0.6905,
+      "step": 15330
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.3254597187042236,
+      "learning_rate": 3.1490454784283965e-06,
+      "loss": 0.7417,
+      "step": 15340
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4701178073883057,
+      "learning_rate": 3.137369373576975e-06,
+      "loss": 0.7084,
+      "step": 15350
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.7671188116073608,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.39,
+      "eval_steps_per_second": 3.278,
+      "step": 15350
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1466660499572754,
+      "learning_rate": 3.1256932687255537e-06,
+      "loss": 0.7003,
+      "step": 15360
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.2833430767059326,
+      "learning_rate": 3.114017163874132e-06,
+      "loss": 0.6996,
+      "step": 15370
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.235135078430176,
+      "learning_rate": 3.1023410590227105e-06,
+      "loss": 0.7006,
+      "step": 15380
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.189640998840332,
+      "learning_rate": 3.0906649541712887e-06,
+      "loss": 0.7189,
+      "step": 15390
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.546860456466675,
+      "learning_rate": 3.0789888493198674e-06,
+      "loss": 0.697,
+      "step": 15400
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.7680267095565796,
+      "eval_runtime": 0.6117,
+      "eval_samples_per_second": 16.347,
+      "eval_steps_per_second": 3.269,
+      "step": 15400
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.1008286476135254,
+      "learning_rate": 3.067312744468446e-06,
+      "loss": 0.7025,
+      "step": 15410
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3359453678131104,
+      "learning_rate": 3.0556366396170237e-06,
+      "loss": 0.7897,
+      "step": 15420
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4781839847564697,
+      "learning_rate": 3.0439605347656028e-06,
+      "loss": 0.668,
+      "step": 15430
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.1375813484191895,
+      "learning_rate": 3.0322844299141805e-06,
+      "loss": 0.6799,
+      "step": 15440
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3086702823638916,
+      "learning_rate": 3.020608325062759e-06,
+      "loss": 0.7176,
+      "step": 15450
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.7684356570243835,
+      "eval_runtime": 0.6073,
+      "eval_samples_per_second": 16.466,
+      "eval_steps_per_second": 3.293,
+      "step": 15450
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2935543060302734,
+      "learning_rate": 3.008932220211338e-06,
+      "loss": 0.6903,
+      "step": 15460
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.3429744243621826,
+      "learning_rate": 2.997256115359916e-06,
+      "loss": 0.6758,
+      "step": 15470
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4076807498931885,
+      "learning_rate": 2.9855800105084946e-06,
+      "loss": 0.6775,
+      "step": 15480
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.052307367324829,
+      "learning_rate": 2.9739039056570728e-06,
+      "loss": 0.7042,
+      "step": 15490
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.318582534790039,
+      "learning_rate": 2.9622278008056514e-06,
+      "loss": 0.7033,
+      "step": 15500
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.7693504095077515,
+      "eval_runtime": 0.6089,
+      "eval_samples_per_second": 16.422,
+      "eval_steps_per_second": 3.284,
+      "step": 15500
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2870733737945557,
+      "learning_rate": 2.9505516959542296e-06,
+      "loss": 0.708,
+      "step": 15510
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2908599376678467,
+      "learning_rate": 2.938875591102808e-06,
+      "loss": 0.7038,
+      "step": 15520
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.2046635150909424,
+      "learning_rate": 2.927199486251387e-06,
+      "loss": 0.736,
+      "step": 15530
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.5193517208099365,
+      "learning_rate": 2.915523381399965e-06,
+      "loss": 0.6812,
+      "step": 15540
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.0635945796966553,
+      "learning_rate": 2.9038472765485436e-06,
+      "loss": 0.6998,
+      "step": 15550
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.7685672044754028,
+      "eval_runtime": 0.6106,
+      "eval_samples_per_second": 16.376,
+      "eval_steps_per_second": 3.275,
+      "step": 15550
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.36582612991333,
+      "learning_rate": 2.892171171697122e-06,
+      "loss": 0.7306,
+      "step": 15560
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.4535040855407715,
+      "learning_rate": 2.8804950668457004e-06,
+      "loss": 0.6775,
+      "step": 15570
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.583468198776245,
+      "learning_rate": 2.868818961994279e-06,
+      "loss": 0.6828,
+      "step": 15580
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 2.249070644378662,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.6773,
+      "step": 15590
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.196047306060791,
+      "learning_rate": 2.845466752291436e-06,
+      "loss": 0.7313,
+      "step": 15600
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.7691243290901184,
+      "eval_runtime": 0.6105,
+      "eval_samples_per_second": 16.38,
+      "eval_steps_per_second": 3.276,
+      "step": 15600
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.261632204055786,
+      "learning_rate": 2.833790647440014e-06,
+      "loss": 0.7143,
+      "step": 15610
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.568326234817505,
+      "learning_rate": 2.8221145425885927e-06,
+      "loss": 0.6925,
+      "step": 15620
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.233002185821533,
+      "learning_rate": 2.8104384377371713e-06,
+      "loss": 0.6844,
+      "step": 15630
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3340461254119873,
+      "learning_rate": 2.7987623328857495e-06,
+      "loss": 0.7026,
+      "step": 15640
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.36470365524292,
+      "learning_rate": 2.787086228034328e-06,
+      "loss": 0.7069,
+      "step": 15650
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.7690743207931519,
+      "eval_runtime": 0.6125,
+      "eval_samples_per_second": 16.327,
+      "eval_steps_per_second": 3.265,
+      "step": 15650
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0926809310913086,
+      "learning_rate": 2.7754101231829063e-06,
+      "loss": 0.691,
+      "step": 15660
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.477888345718384,
+      "learning_rate": 2.763734018331485e-06,
+      "loss": 0.7025,
+      "step": 15670
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.447984457015991,
+      "learning_rate": 2.7520579134800635e-06,
+      "loss": 0.6787,
+      "step": 15680
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.0055558681488037,
+      "learning_rate": 2.7403818086286417e-06,
+      "loss": 0.6781,
+      "step": 15690
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1302168369293213,
+      "learning_rate": 2.7287057037772203e-06,
+      "loss": 0.7047,
+      "step": 15700
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.7690717577934265,
+      "eval_runtime": 0.6098,
+      "eval_samples_per_second": 16.399,
+      "eval_steps_per_second": 3.28,
+      "step": 15700
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.3251328468322754,
+      "learning_rate": 2.717029598925798e-06,
+      "loss": 0.6663,
+      "step": 15710
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.612375020980835,
+      "learning_rate": 2.705353494074377e-06,
+      "loss": 0.7182,
+      "step": 15720
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.6945440769195557,
+      "learning_rate": 2.6936773892229558e-06,
+      "loss": 0.684,
+      "step": 15730
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2480788230895996,
+      "learning_rate": 2.6820012843715335e-06,
+      "loss": 0.7145,
+      "step": 15740
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.2266530990600586,
+      "learning_rate": 2.6703251795201126e-06,
+      "loss": 0.694,
+      "step": 15750
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 0.7684869766235352,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.373,
+      "eval_steps_per_second": 3.275,
+      "step": 15750
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.5328357219696045,
+      "learning_rate": 2.6586490746686903e-06,
+      "loss": 0.6943,
+      "step": 15760
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 2.1645123958587646,
+      "learning_rate": 2.646972969817269e-06,
+      "loss": 0.6844,
+      "step": 15770
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.332747459411621,
+      "learning_rate": 2.635296864965848e-06,
+      "loss": 0.7001,
+      "step": 15780
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2053799629211426,
+      "learning_rate": 2.6236207601144258e-06,
+      "loss": 0.7633,
+      "step": 15790
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1534461975097656,
+      "learning_rate": 2.6119446552630044e-06,
+      "loss": 0.7117,
+      "step": 15800
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.768386721611023,
+      "eval_runtime": 0.6104,
+      "eval_samples_per_second": 16.382,
+      "eval_steps_per_second": 3.276,
+      "step": 15800
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6755309104919434,
+      "learning_rate": 2.6002685504115826e-06,
+      "loss": 0.7087,
+      "step": 15810
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.6160128116607666,
+      "learning_rate": 2.588592445560161e-06,
+      "loss": 0.7165,
+      "step": 15820
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.4403111934661865,
+      "learning_rate": 2.57691634070874e-06,
+      "loss": 0.6964,
+      "step": 15830
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.429347515106201,
+      "learning_rate": 2.565240235857318e-06,
+      "loss": 0.6983,
+      "step": 15840
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.1012115478515625,
+      "learning_rate": 2.5535641310058966e-06,
+      "loss": 0.6941,
+      "step": 15850
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.7683624029159546,
+      "eval_runtime": 0.6094,
+      "eval_samples_per_second": 16.41,
+      "eval_steps_per_second": 3.282,
+      "step": 15850
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2343549728393555,
+      "learning_rate": 2.541888026154475e-06,
+      "loss": 0.701,
+      "step": 15860
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.169816732406616,
+      "learning_rate": 2.5302119213030534e-06,
+      "loss": 0.6983,
+      "step": 15870
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.5199921131134033,
+      "learning_rate": 2.518535816451632e-06,
+      "loss": 0.7317,
+      "step": 15880
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.530402421951294,
+      "learning_rate": 2.5068597116002102e-06,
+      "loss": 0.6753,
+      "step": 15890
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2547996044158936,
+      "learning_rate": 2.495183606748789e-06,
+      "loss": 0.6803,
+      "step": 15900
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.768885612487793,
+      "eval_runtime": 0.6107,
+      "eval_samples_per_second": 16.375,
+      "eval_steps_per_second": 3.275,
+      "step": 15900
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.3491618633270264,
+      "learning_rate": 2.4835075018973675e-06,
+      "loss": 0.6997,
+      "step": 15910
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2919349670410156,
+      "learning_rate": 2.4718313970459457e-06,
+      "loss": 0.713,
+      "step": 15920
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.308943748474121,
+      "learning_rate": 2.460155292194524e-06,
+      "loss": 0.6886,
+      "step": 15930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.2621374130249023,
+      "learning_rate": 2.4484791873431025e-06,
+      "loss": 0.6947,
+      "step": 15940
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.271512269973755,
+      "learning_rate": 2.436803082491681e-06,
+      "loss": 0.6923,
+      "step": 15950
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.7677316665649414,
+      "eval_runtime": 0.6092,
+      "eval_samples_per_second": 16.415,
+      "eval_steps_per_second": 3.283,
+      "step": 15950
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4293153285980225,
+      "learning_rate": 2.4251269776402593e-06,
+      "loss": 0.6775,
+      "step": 15960
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5537006855010986,
+      "learning_rate": 2.413450872788838e-06,
+      "loss": 0.7366,
+      "step": 15970
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.5096888542175293,
+      "learning_rate": 2.401774767937416e-06,
+      "loss": 0.6979,
+      "step": 15980
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3841376304626465,
+      "learning_rate": 2.3900986630859947e-06,
+      "loss": 0.6639,
+      "step": 15990
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3979759216308594,
+      "learning_rate": 2.3784225582345733e-06,
+      "loss": 0.6748,
+      "step": 16000
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.7679833769798279,
+      "eval_runtime": 0.6113,
+      "eval_samples_per_second": 16.359,
+      "eval_steps_per_second": 3.272,
+      "step": 16000
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2421648502349854,
+      "learning_rate": 2.3667464533831515e-06,
+      "loss": 0.7092,
+      "step": 16010
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.484659194946289,
+      "learning_rate": 2.35507034853173e-06,
+      "loss": 0.7843,
+      "step": 16020
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.4099161624908447,
+      "learning_rate": 2.3433942436803083e-06,
+      "loss": 0.6838,
+      "step": 16030
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.6203510761260986,
+      "learning_rate": 2.331718138828887e-06,
+      "loss": 0.6919,
+      "step": 16040
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2019495964050293,
+      "learning_rate": 2.3200420339774656e-06,
+      "loss": 0.7143,
+      "step": 16050
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.7672072649002075,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.342,
+      "eval_steps_per_second": 3.268,
+      "step": 16050
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3565802574157715,
+      "learning_rate": 2.3083659291260438e-06,
+      "loss": 0.6841,
+      "step": 16060
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2576301097869873,
+      "learning_rate": 2.2966898242746224e-06,
+      "loss": 0.6869,
+      "step": 16070
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.367213487625122,
+      "learning_rate": 2.2850137194232006e-06,
+      "loss": 0.7335,
+      "step": 16080
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.3077759742736816,
+      "learning_rate": 2.2733376145717788e-06,
+      "loss": 0.7238,
+      "step": 16090
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.421360731124878,
+      "learning_rate": 2.2616615097203574e-06,
+      "loss": 0.692,
+      "step": 16100
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 0.7670503854751587,
+      "eval_runtime": 0.6054,
+      "eval_samples_per_second": 16.519,
+      "eval_steps_per_second": 3.304,
+      "step": 16100
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.1805195808410645,
+      "learning_rate": 2.249985404868936e-06,
+      "loss": 0.7083,
+      "step": 16110
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.2615694999694824,
+      "learning_rate": 2.238309300017514e-06,
+      "loss": 0.6999,
+      "step": 16120
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 2.57191801071167,
+      "learning_rate": 2.226633195166093e-06,
+      "loss": 0.7028,
+      "step": 16130
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.42756986618042,
+      "learning_rate": 2.214957090314671e-06,
+      "loss": 0.659,
+      "step": 16140
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5150256156921387,
+      "learning_rate": 2.2032809854632496e-06,
+      "loss": 0.6893,
+      "step": 16150
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.7659882307052612,
+      "eval_runtime": 0.6369,
+      "eval_samples_per_second": 15.701,
+      "eval_steps_per_second": 3.14,
+      "step": 16150
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3516571521759033,
+      "learning_rate": 2.1916048806118282e-06,
+      "loss": 0.6884,
+      "step": 16160
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.9907314777374268,
+      "learning_rate": 2.1799287757604064e-06,
+      "loss": 0.7214,
+      "step": 16170
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2102601528167725,
+      "learning_rate": 2.168252670908985e-06,
+      "loss": 0.7201,
+      "step": 16180
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2297379970550537,
+      "learning_rate": 2.1565765660575632e-06,
+      "loss": 0.7258,
+      "step": 16190
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.5669262409210205,
+      "learning_rate": 2.144900461206142e-06,
+      "loss": 0.6861,
+      "step": 16200
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.7660841345787048,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.365,
+      "eval_steps_per_second": 3.273,
+      "step": 16200
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2349135875701904,
+      "learning_rate": 2.1332243563547205e-06,
+      "loss": 0.7668,
+      "step": 16210
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.093426465988159,
+      "learning_rate": 2.1215482515032987e-06,
+      "loss": 0.6802,
+      "step": 16220
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.415769100189209,
+      "learning_rate": 2.109872146651877e-06,
+      "loss": 0.6886,
+      "step": 16230
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3475406169891357,
+      "learning_rate": 2.0981960418004555e-06,
+      "loss": 0.6703,
+      "step": 16240
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3554067611694336,
+      "learning_rate": 2.086519936949034e-06,
+      "loss": 0.6964,
+      "step": 16250
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.7669075727462769,
+      "eval_runtime": 0.6098,
+      "eval_samples_per_second": 16.399,
+      "eval_steps_per_second": 3.28,
+      "step": 16250
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.220348358154297,
+      "learning_rate": 2.0748438320976123e-06,
+      "loss": 0.7182,
+      "step": 16260
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.4331984519958496,
+      "learning_rate": 2.063167727246191e-06,
+      "loss": 0.7021,
+      "step": 16270
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.3163468837738037,
+      "learning_rate": 2.051491622394769e-06,
+      "loss": 0.7142,
+      "step": 16280
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.415635824203491,
+      "learning_rate": 2.0398155175433477e-06,
+      "loss": 0.7435,
+      "step": 16290
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.429462432861328,
+      "learning_rate": 2.0281394126919263e-06,
+      "loss": 0.6906,
+      "step": 16300
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.7668562531471252,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.373,
+      "eval_steps_per_second": 3.275,
+      "step": 16300
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 2.2036139965057373,
+      "learning_rate": 2.0164633078405045e-06,
+      "loss": 0.6977,
+      "step": 16310
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6458094120025635,
+      "learning_rate": 2.004787202989083e-06,
+      "loss": 0.7071,
+      "step": 16320
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2309153079986572,
+      "learning_rate": 1.9931110981376613e-06,
+      "loss": 0.7118,
+      "step": 16330
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.1666347980499268,
+      "learning_rate": 1.98143499328624e-06,
+      "loss": 0.6876,
+      "step": 16340
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.716320514678955,
+      "learning_rate": 1.9697588884348186e-06,
+      "loss": 0.701,
+      "step": 16350
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.7665780782699585,
+      "eval_runtime": 0.6078,
+      "eval_samples_per_second": 16.453,
+      "eval_steps_per_second": 3.291,
+      "step": 16350
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.233903408050537,
+      "learning_rate": 1.9580827835833968e-06,
+      "loss": 0.6812,
+      "step": 16360
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.7495980262756348,
+      "learning_rate": 1.9464066787319754e-06,
+      "loss": 0.6875,
+      "step": 16370
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.6137688159942627,
+      "learning_rate": 1.9347305738805536e-06,
+      "loss": 0.6986,
+      "step": 16380
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2753725051879883,
+      "learning_rate": 1.9230544690291318e-06,
+      "loss": 0.6727,
+      "step": 16390
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.182767868041992,
+      "learning_rate": 1.911378364177711e-06,
+      "loss": 0.6989,
+      "step": 16400
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.7669702768325806,
+      "eval_runtime": 0.6079,
+      "eval_samples_per_second": 16.45,
+      "eval_steps_per_second": 3.29,
+      "step": 16400
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.4709830284118652,
+      "learning_rate": 1.899702259326289e-06,
+      "loss": 0.7117,
+      "step": 16410
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.203662633895874,
+      "learning_rate": 1.8880261544748674e-06,
+      "loss": 0.703,
+      "step": 16420
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.9789013862609863,
+      "learning_rate": 1.8763500496234458e-06,
+      "loss": 0.6701,
+      "step": 16430
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.48000431060791,
+      "learning_rate": 1.8646739447720242e-06,
+      "loss": 0.6735,
+      "step": 16440
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.286405086517334,
+      "learning_rate": 1.8529978399206026e-06,
+      "loss": 0.684,
+      "step": 16450
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.7673357725143433,
+      "eval_runtime": 0.6119,
+      "eval_samples_per_second": 16.341,
+      "eval_steps_per_second": 3.268,
+      "step": 16450
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.2844350337982178,
+      "learning_rate": 1.8413217350691812e-06,
+      "loss": 0.7121,
+      "step": 16460
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.243093490600586,
+      "learning_rate": 1.8296456302177596e-06,
+      "loss": 0.7167,
+      "step": 16470
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.679344654083252,
+      "learning_rate": 1.8179695253663378e-06,
+      "loss": 0.6979,
+      "step": 16480
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 2.410310745239258,
+      "learning_rate": 1.8062934205149162e-06,
+      "loss": 0.7074,
+      "step": 16490
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4391376972198486,
+      "learning_rate": 1.7946173156634946e-06,
+      "loss": 0.7179,
+      "step": 16500
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.7669895887374878,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.389,
+      "eval_steps_per_second": 3.278,
+      "step": 16500
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.0283079147338867,
+      "learning_rate": 1.7829412108120732e-06,
+      "loss": 0.7484,
+      "step": 16510
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4193665981292725,
+      "learning_rate": 1.7712651059606517e-06,
+      "loss": 0.709,
+      "step": 16520
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.322601318359375,
+      "learning_rate": 1.75958900110923e-06,
+      "loss": 0.6955,
+      "step": 16530
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.234459400177002,
+      "learning_rate": 1.7479128962578085e-06,
+      "loss": 0.677,
+      "step": 16540
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.346160650253296,
+      "learning_rate": 1.7362367914063869e-06,
+      "loss": 0.6876,
+      "step": 16550
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.7667087316513062,
+      "eval_runtime": 0.6071,
+      "eval_samples_per_second": 16.471,
+      "eval_steps_per_second": 3.294,
+      "step": 16550
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.324331045150757,
+      "learning_rate": 1.7245606865549655e-06,
+      "loss": 0.6775,
+      "step": 16560
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.126572608947754,
+      "learning_rate": 1.7128845817035439e-06,
+      "loss": 0.704,
+      "step": 16570
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3754642009735107,
+      "learning_rate": 1.7012084768521223e-06,
+      "loss": 0.6914,
+      "step": 16580
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.2358624935150146,
+      "learning_rate": 1.6895323720007007e-06,
+      "loss": 0.7128,
+      "step": 16590
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.4077680110931396,
+      "learning_rate": 1.677856267149279e-06,
+      "loss": 0.6937,
+      "step": 16600
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.7659896612167358,
+      "eval_runtime": 0.6102,
+      "eval_samples_per_second": 16.389,
+      "eval_steps_per_second": 3.278,
+      "step": 16600
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.272256851196289,
+      "learning_rate": 1.6661801622978577e-06,
+      "loss": 0.6997,
+      "step": 16610
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5117475986480713,
+      "learning_rate": 1.6545040574464361e-06,
+      "loss": 0.6882,
+      "step": 16620
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.3137876987457275,
+      "learning_rate": 1.6428279525950145e-06,
+      "loss": 0.7346,
+      "step": 16630
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.284081220626831,
+      "learning_rate": 1.6311518477435927e-06,
+      "loss": 0.6686,
+      "step": 16640
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.1261956691741943,
+      "learning_rate": 1.6194757428921711e-06,
+      "loss": 0.6839,
+      "step": 16650
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 0.7666657567024231,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.369,
+      "eval_steps_per_second": 3.274,
+      "step": 16650
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.478227138519287,
+      "learning_rate": 1.6077996380407497e-06,
+      "loss": 0.713,
+      "step": 16660
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.401261329650879,
+      "learning_rate": 1.5961235331893281e-06,
+      "loss": 0.6693,
+      "step": 16670
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4096415042877197,
+      "learning_rate": 1.5844474283379066e-06,
+      "loss": 0.668,
+      "step": 16680
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3899402618408203,
+      "learning_rate": 1.572771323486485e-06,
+      "loss": 0.6909,
+      "step": 16690
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.240556240081787,
+      "learning_rate": 1.5610952186350634e-06,
+      "loss": 0.6907,
+      "step": 16700
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.7665630578994751,
+      "eval_runtime": 0.613,
+      "eval_samples_per_second": 16.314,
+      "eval_steps_per_second": 3.263,
+      "step": 16700
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.179701328277588,
+      "learning_rate": 1.549419113783642e-06,
+      "loss": 0.7034,
+      "step": 16710
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.506798505783081,
+      "learning_rate": 1.5377430089322204e-06,
+      "loss": 0.7578,
+      "step": 16720
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2184534072875977,
+      "learning_rate": 1.5260669040807988e-06,
+      "loss": 0.6943,
+      "step": 16730
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4832656383514404,
+      "learning_rate": 1.5143907992293772e-06,
+      "loss": 0.6861,
+      "step": 16740
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.340266227722168,
+      "learning_rate": 1.5027146943779556e-06,
+      "loss": 0.6599,
+      "step": 16750
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.7668687701225281,
+      "eval_runtime": 0.6118,
+      "eval_samples_per_second": 16.346,
+      "eval_steps_per_second": 3.269,
+      "step": 16750
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.2938528060913086,
+      "learning_rate": 1.4910385895265342e-06,
+      "loss": 0.6751,
+      "step": 16760
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8623875379562378,
+      "learning_rate": 1.4793624846751126e-06,
+      "loss": 0.7303,
+      "step": 16770
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.567239999771118,
+      "learning_rate": 1.467686379823691e-06,
+      "loss": 0.7067,
+      "step": 16780
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3123650550842285,
+      "learning_rate": 1.4560102749722692e-06,
+      "loss": 0.7221,
+      "step": 16790
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.1059062480926514,
+      "learning_rate": 1.4443341701208476e-06,
+      "loss": 0.656,
+      "step": 16800
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.7666193246841431,
+      "eval_runtime": 0.6101,
+      "eval_samples_per_second": 16.39,
+      "eval_steps_per_second": 3.278,
+      "step": 16800
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.094505786895752,
+      "learning_rate": 1.4326580652694265e-06,
+      "loss": 0.7479,
+      "step": 16810
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.172165632247925,
+      "learning_rate": 1.4209819604180046e-06,
+      "loss": 0.7436,
+      "step": 16820
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.148313283920288,
+      "learning_rate": 1.409305855566583e-06,
+      "loss": 0.6587,
+      "step": 16830
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.3585851192474365,
+      "learning_rate": 1.3976297507151615e-06,
+      "loss": 0.6835,
+      "step": 16840
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 2.4866549968719482,
+      "learning_rate": 1.3859536458637399e-06,
+      "loss": 0.7231,
+      "step": 16850
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.7663007974624634,
+      "eval_runtime": 0.6089,
+      "eval_samples_per_second": 16.422,
+      "eval_steps_per_second": 3.284,
+      "step": 16850
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2792775630950928,
+      "learning_rate": 1.3742775410123185e-06,
+      "loss": 0.6923,
+      "step": 16860
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4036784172058105,
+      "learning_rate": 1.3626014361608969e-06,
+      "loss": 0.7194,
+      "step": 16870
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.423809051513672,
+      "learning_rate": 1.3509253313094753e-06,
+      "loss": 0.6873,
+      "step": 16880
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1228232383728027,
+      "learning_rate": 1.3392492264580537e-06,
+      "loss": 0.6828,
+      "step": 16890
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3673083782196045,
+      "learning_rate": 1.327573121606632e-06,
+      "loss": 0.718,
+      "step": 16900
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.7667369842529297,
+      "eval_runtime": 0.6083,
+      "eval_samples_per_second": 16.441,
+      "eval_steps_per_second": 3.288,
+      "step": 16900
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3858864307403564,
+      "learning_rate": 1.3158970167552107e-06,
+      "loss": 0.7048,
+      "step": 16910
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.157116413116455,
+      "learning_rate": 1.3042209119037891e-06,
+      "loss": 0.6906,
+      "step": 16920
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.4072022438049316,
+      "learning_rate": 1.2925448070523675e-06,
+      "loss": 0.6663,
+      "step": 16930
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.5083460807800293,
+      "learning_rate": 1.280868702200946e-06,
+      "loss": 0.6609,
+      "step": 16940
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1902806758880615,
+      "learning_rate": 1.2691925973495241e-06,
+      "loss": 0.6892,
+      "step": 16950
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.7658802270889282,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.341,
+      "eval_steps_per_second": 3.268,
+      "step": 16950
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3695363998413086,
+      "learning_rate": 1.257516492498103e-06,
+      "loss": 0.7267,
+      "step": 16960
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.319840669631958,
+      "learning_rate": 1.2458403876466811e-06,
+      "loss": 0.7233,
+      "step": 16970
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.3470675945281982,
+      "learning_rate": 1.2341642827952595e-06,
+      "loss": 0.6418,
+      "step": 16980
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.6220788955688477,
+      "learning_rate": 1.222488177943838e-06,
+      "loss": 0.6843,
+      "step": 16990
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.2023866176605225,
+      "learning_rate": 1.2108120730924166e-06,
+      "loss": 0.6979,
+      "step": 17000
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.7656064033508301,
+      "eval_runtime": 0.6104,
+      "eval_samples_per_second": 16.383,
+      "eval_steps_per_second": 3.277,
+      "step": 17000
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.8683066368103027,
+      "learning_rate": 1.199135968240995e-06,
+      "loss": 0.7075,
+      "step": 17010
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.1979622840881348,
+      "learning_rate": 1.1874598633895734e-06,
+      "loss": 0.747,
+      "step": 17020
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 2.120609760284424,
+      "learning_rate": 1.1757837585381518e-06,
+      "loss": 0.6841,
+      "step": 17030
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.330552339553833,
+      "learning_rate": 1.1641076536867302e-06,
+      "loss": 0.6911,
+      "step": 17040
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.504176616668701,
+      "learning_rate": 1.1524315488353086e-06,
+      "loss": 0.7166,
+      "step": 17050
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.765422523021698,
+      "eval_runtime": 0.6073,
+      "eval_samples_per_second": 16.466,
+      "eval_steps_per_second": 3.293,
+      "step": 17050
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.314986228942871,
+      "learning_rate": 1.140755443983887e-06,
+      "loss": 0.7054,
+      "step": 17060
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4601542949676514,
+      "learning_rate": 1.1290793391324654e-06,
+      "loss": 0.6743,
+      "step": 17070
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.451432466506958,
+      "learning_rate": 1.117403234281044e-06,
+      "loss": 0.6712,
+      "step": 17080
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.485808849334717,
+      "learning_rate": 1.1057271294296224e-06,
+      "loss": 0.6779,
+      "step": 17090
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4770805835723877,
+      "learning_rate": 1.0940510245782008e-06,
+      "loss": 0.7321,
+      "step": 17100
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.7653018236160278,
+      "eval_runtime": 0.632,
+      "eval_samples_per_second": 15.824,
+      "eval_steps_per_second": 3.165,
+      "step": 17100
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.318748950958252,
+      "learning_rate": 1.0823749197267792e-06,
+      "loss": 0.6747,
+      "step": 17110
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.9840216636657715,
+      "learning_rate": 1.0706988148753576e-06,
+      "loss": 0.7228,
+      "step": 17120
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.517688274383545,
+      "learning_rate": 1.059022710023936e-06,
+      "loss": 0.6855,
+      "step": 17130
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.166759490966797,
+      "learning_rate": 1.0473466051725144e-06,
+      "loss": 0.7202,
+      "step": 17140
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.3945472240448,
+      "learning_rate": 1.035670500321093e-06,
+      "loss": 0.7212,
+      "step": 17150
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.7650858163833618,
+      "eval_runtime": 0.6157,
+      "eval_samples_per_second": 16.24,
+      "eval_steps_per_second": 3.248,
+      "step": 17150
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4738452434539795,
+      "learning_rate": 1.0239943954696715e-06,
+      "loss": 0.6852,
+      "step": 17160
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.2075397968292236,
+      "learning_rate": 1.0123182906182499e-06,
+      "loss": 0.6963,
+      "step": 17170
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.5720317363739014,
+      "learning_rate": 1.0006421857668283e-06,
+      "loss": 0.7048,
+      "step": 17180
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.4343419075012207,
+      "learning_rate": 9.889660809154067e-07,
+      "loss": 0.7691,
+      "step": 17190
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.435464382171631,
+      "learning_rate": 9.77289976063985e-07,
+      "loss": 0.6953,
+      "step": 17200
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.7652177810668945,
+      "eval_runtime": 0.6069,
+      "eval_samples_per_second": 16.478,
+      "eval_steps_per_second": 3.296,
+      "step": 17200
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 2.6392276287078857,
+      "learning_rate": 9.656138712125635e-07,
+      "loss": 0.7361,
+      "step": 17210
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.188244581222534,
+      "learning_rate": 9.539377663611419e-07,
+      "loss": 0.707,
+      "step": 17220
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.548353910446167,
+      "learning_rate": 9.422616615097205e-07,
+      "loss": 0.6733,
+      "step": 17230
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.831445574760437,
+      "learning_rate": 9.305855566582988e-07,
+      "loss": 0.7141,
+      "step": 17240
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.176208257675171,
+      "learning_rate": 9.200770622920195e-07,
+      "loss": 0.6843,
+      "step": 17250
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.7648450136184692,
+      "eval_runtime": 0.6072,
+      "eval_samples_per_second": 16.468,
+      "eval_steps_per_second": 3.294,
+      "step": 17250
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.8308520317077637,
+      "learning_rate": 9.084009574405979e-07,
+      "loss": 0.7139,
+      "step": 17260
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.235097646713257,
+      "learning_rate": 8.967248525891763e-07,
+      "loss": 0.7178,
+      "step": 17270
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.1307973861694336,
+      "learning_rate": 8.850487477377548e-07,
+      "loss": 0.6942,
+      "step": 17280
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2758638858795166,
+      "learning_rate": 8.733726428863332e-07,
+      "loss": 0.6971,
+      "step": 17290
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.261354923248291,
+      "learning_rate": 8.616965380349117e-07,
+      "loss": 0.6856,
+      "step": 17300
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.7645593881607056,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.364,
+      "eval_steps_per_second": 3.273,
+      "step": 17300
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3220369815826416,
+      "learning_rate": 8.5002043318349e-07,
+      "loss": 0.7182,
+      "step": 17310
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.5134150981903076,
+      "learning_rate": 8.383443283320684e-07,
+      "loss": 0.7045,
+      "step": 17320
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3346173763275146,
+      "learning_rate": 8.266682234806469e-07,
+      "loss": 0.6787,
+      "step": 17330
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.2889230251312256,
+      "learning_rate": 8.149921186292253e-07,
+      "loss": 0.7275,
+      "step": 17340
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.576421022415161,
+      "learning_rate": 8.033160137778038e-07,
+      "loss": 0.6994,
+      "step": 17350
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.7651536464691162,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.373,
+      "eval_steps_per_second": 3.275,
+      "step": 17350
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3312604427337646,
+      "learning_rate": 7.916399089263822e-07,
+      "loss": 0.693,
+      "step": 17360
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.355175495147705,
+      "learning_rate": 7.799638040749607e-07,
+      "loss": 0.6914,
+      "step": 17370
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.3729312419891357,
+      "learning_rate": 7.682876992235392e-07,
+      "loss": 0.7119,
+      "step": 17380
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.260957717895508,
+      "learning_rate": 7.566115943721175e-07,
+      "loss": 0.7166,
+      "step": 17390
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.238769054412842,
+      "learning_rate": 7.44935489520696e-07,
+      "loss": 0.6839,
+      "step": 17400
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.7647945880889893,
+      "eval_runtime": 0.609,
+      "eval_samples_per_second": 16.421,
+      "eval_steps_per_second": 3.284,
+      "step": 17400
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.6119112968444824,
+      "learning_rate": 7.332593846692744e-07,
+      "loss": 0.697,
+      "step": 17410
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3611526489257812,
+      "learning_rate": 7.215832798178528e-07,
+      "loss": 0.7072,
+      "step": 17420
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.184889078140259,
+      "learning_rate": 7.099071749664313e-07,
+      "loss": 0.7353,
+      "step": 17430
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3173468112945557,
+      "learning_rate": 6.982310701150097e-07,
+      "loss": 0.6686,
+      "step": 17440
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3475663661956787,
+      "learning_rate": 6.865549652635881e-07,
+      "loss": 0.6802,
+      "step": 17450
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.7647973299026489,
+      "eval_runtime": 0.6087,
+      "eval_samples_per_second": 16.429,
+      "eval_steps_per_second": 3.286,
+      "step": 17450
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.473719596862793,
+      "learning_rate": 6.748788604121666e-07,
+      "loss": 0.6674,
+      "step": 17460
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.3409502506256104,
+      "learning_rate": 6.632027555607449e-07,
+      "loss": 0.6808,
+      "step": 17470
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.324878454208374,
+      "learning_rate": 6.515266507093234e-07,
+      "loss": 0.6826,
+      "step": 17480
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.429734706878662,
+      "learning_rate": 6.398505458579018e-07,
+      "loss": 0.7202,
+      "step": 17490
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.357304334640503,
+      "learning_rate": 6.281744410064802e-07,
+      "loss": 0.6796,
+      "step": 17500
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.7646846771240234,
+      "eval_runtime": 0.6078,
+      "eval_samples_per_second": 16.452,
+      "eval_steps_per_second": 3.29,
+      "step": 17500
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.462076425552368,
+      "learning_rate": 6.164983361550587e-07,
+      "loss": 0.722,
+      "step": 17510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2138359546661377,
+      "learning_rate": 6.048222313036371e-07,
+      "loss": 0.7001,
+      "step": 17520
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.4367923736572266,
+      "learning_rate": 5.931461264522156e-07,
+      "loss": 0.7315,
+      "step": 17530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.67116117477417,
+      "learning_rate": 5.814700216007941e-07,
+      "loss": 0.6999,
+      "step": 17540
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.1781866550445557,
+      "learning_rate": 5.697939167493725e-07,
+      "loss": 0.6934,
+      "step": 17550
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.7649165987968445,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.34,
+      "eval_steps_per_second": 3.268,
+      "step": 17550
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.444739818572998,
+      "learning_rate": 5.581178118979509e-07,
+      "loss": 0.7157,
+      "step": 17560
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.558717966079712,
+      "learning_rate": 5.464417070465293e-07,
+      "loss": 0.6539,
+      "step": 17570
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 2.2989799976348877,
+      "learning_rate": 5.347656021951078e-07,
+      "loss": 0.6988,
+      "step": 17580
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.8470942974090576,
+      "learning_rate": 5.230894973436862e-07,
+      "loss": 0.7129,
+      "step": 17590
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3580868244171143,
+      "learning_rate": 5.114133924922646e-07,
+      "loss": 0.6935,
+      "step": 17600
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.7648688554763794,
+      "eval_runtime": 0.6108,
+      "eval_samples_per_second": 16.372,
+      "eval_steps_per_second": 3.274,
+      "step": 17600
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2725815773010254,
+      "learning_rate": 4.997372876408431e-07,
+      "loss": 0.6891,
+      "step": 17610
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2336254119873047,
+      "learning_rate": 4.880611827894215e-07,
+      "loss": 0.7433,
+      "step": 17620
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.9831641912460327,
+      "learning_rate": 4.763850779379999e-07,
+      "loss": 0.7134,
+      "step": 17630
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.558650255203247,
+      "learning_rate": 4.647089730865784e-07,
+      "loss": 0.6763,
+      "step": 17640
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4820024967193604,
+      "learning_rate": 4.5303286823515684e-07,
+      "loss": 0.6995,
+      "step": 17650
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.7648004293441772,
+      "eval_runtime": 0.6073,
+      "eval_samples_per_second": 16.465,
+      "eval_steps_per_second": 3.293,
+      "step": 17650
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3149256706237793,
+      "learning_rate": 4.413567633837352e-07,
+      "loss": 0.65,
+      "step": 17660
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.4520041942596436,
+      "learning_rate": 4.2968065853231365e-07,
+      "loss": 0.6816,
+      "step": 17670
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.295974016189575,
+      "learning_rate": 4.180045536808921e-07,
+      "loss": 0.702,
+      "step": 17680
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.3948514461517334,
+      "learning_rate": 4.0632844882947056e-07,
+      "loss": 0.7216,
+      "step": 17690
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2249348163604736,
+      "learning_rate": 3.9465234397804897e-07,
+      "loss": 0.7731,
+      "step": 17700
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.7649565935134888,
+      "eval_runtime": 0.6121,
+      "eval_samples_per_second": 16.337,
+      "eval_steps_per_second": 3.267,
+      "step": 17700
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.394788980484009,
+      "learning_rate": 3.8297623912662737e-07,
+      "loss": 0.7122,
+      "step": 17710
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.2445878982543945,
+      "learning_rate": 3.7130013427520583e-07,
+      "loss": 0.7001,
+      "step": 17720
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.399392604827881,
+      "learning_rate": 3.5962402942378423e-07,
+      "loss": 0.6982,
+      "step": 17730
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1935548782348633,
+      "learning_rate": 3.479479245723627e-07,
+      "loss": 0.6842,
+      "step": 17740
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.1252946853637695,
+      "learning_rate": 3.3627181972094115e-07,
+      "loss": 0.6804,
+      "step": 17750
+    },
+    {
+      "epoch": 0.98,
+      "eval_loss": 0.7648642659187317,
+      "eval_runtime": 0.6111,
+      "eval_samples_per_second": 16.364,
+      "eval_steps_per_second": 3.273,
+      "step": 17750
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 2.217083692550659,
+      "learning_rate": 3.2459571486951955e-07,
+      "loss": 0.7093,
+      "step": 17760
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.216444730758667,
+      "learning_rate": 3.1291961001809796e-07,
+      "loss": 0.7199,
+      "step": 17770
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.284081220626831,
+      "learning_rate": 3.012435051666764e-07,
+      "loss": 0.6697,
+      "step": 17780
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.300384283065796,
+      "learning_rate": 2.895674003152548e-07,
+      "loss": 0.6951,
+      "step": 17790
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.336672782897949,
+      "learning_rate": 2.778912954638333e-07,
+      "loss": 0.661,
+      "step": 17800
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.7648496627807617,
+      "eval_runtime": 0.612,
+      "eval_samples_per_second": 16.341,
+      "eval_steps_per_second": 3.268,
+      "step": 17800
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4574379920959473,
+      "learning_rate": 2.6621519061241174e-07,
+      "loss": 0.6735,
+      "step": 17810
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3473894596099854,
+      "learning_rate": 2.5453908576099014e-07,
+      "loss": 0.7046,
+      "step": 17820
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.3205556869506836,
+      "learning_rate": 2.428629809095686e-07,
+      "loss": 0.7386,
+      "step": 17830
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4533944129943848,
+      "learning_rate": 2.31186876058147e-07,
+      "loss": 0.6531,
+      "step": 17840
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.458087682723999,
+      "learning_rate": 2.1951077120672546e-07,
+      "loss": 0.6851,
+      "step": 17850
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.7649797201156616,
+      "eval_runtime": 0.6106,
+      "eval_samples_per_second": 16.376,
+      "eval_steps_per_second": 3.275,
+      "step": 17850
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.5032155513763428,
+      "learning_rate": 2.0783466635530387e-07,
+      "loss": 0.723,
+      "step": 17860
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.168433904647827,
+      "learning_rate": 1.9615856150388232e-07,
+      "loss": 0.6868,
+      "step": 17870
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.264967441558838,
+      "learning_rate": 1.8448245665246078e-07,
+      "loss": 0.6878,
+      "step": 17880
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1190009117126465,
+      "learning_rate": 1.7280635180103919e-07,
+      "loss": 0.6875,
+      "step": 17890
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.35349178314209,
+      "learning_rate": 1.6113024694961764e-07,
+      "loss": 0.6698,
+      "step": 17900
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.7650014162063599,
+      "eval_runtime": 0.6147,
+      "eval_samples_per_second": 16.268,
+      "eval_steps_per_second": 3.254,
+      "step": 17900
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.433678150177002,
+      "learning_rate": 1.4945414209819607e-07,
+      "loss": 0.6902,
+      "step": 17910
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.1968746185302734,
+      "learning_rate": 1.377780372467745e-07,
+      "loss": 0.689,
+      "step": 17920
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.752901077270508,
+      "learning_rate": 1.261019323953529e-07,
+      "loss": 0.6795,
+      "step": 17930
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.241312026977539,
+      "learning_rate": 1.1442582754393136e-07,
+      "loss": 0.6765,
+      "step": 17940
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.2908291816711426,
+      "learning_rate": 1.0274972269250979e-07,
+      "loss": 0.6955,
+      "step": 17950
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.764830470085144,
+      "eval_runtime": 0.6093,
+      "eval_samples_per_second": 16.411,
+      "eval_steps_per_second": 3.282,
+      "step": 17950
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.2523908615112305,
+      "learning_rate": 9.107361784108822e-08,
+      "loss": 0.6858,
+      "step": 17960
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.356487512588501,
+      "learning_rate": 7.939751298966665e-08,
+      "loss": 0.6892,
+      "step": 17970
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.103383779525757,
+      "learning_rate": 6.772140813824508e-08,
+      "loss": 0.7362,
+      "step": 17980
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.52716064453125,
+      "learning_rate": 5.604530328682352e-08,
+      "loss": 0.6875,
+      "step": 17990
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.3393402099609375,
+      "learning_rate": 4.4369198435401956e-08,
+      "loss": 0.7106,
+      "step": 18000
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.7649039626121521,
+      "eval_runtime": 0.6109,
+      "eval_samples_per_second": 16.369,
+      "eval_steps_per_second": 3.274,
+      "step": 18000
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.457542657852173,
+      "learning_rate": 3.269309358398039e-08,
+      "loss": 0.6811,
+      "step": 18010
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.4412107467651367,
+      "learning_rate": 2.1016988732558822e-08,
+      "loss": 0.6783,
+      "step": 18020
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.611578941345215,
+      "learning_rate": 9.340883881137255e-09,
+      "loss": 0.6962,
+      "step": 18030
+    },
+    {
+      "epoch": 1.0,
+      "step": 18031,
+      "total_flos": 2.80353585038046e+18,
+      "train_loss": 0.7375527136034011,
+      "train_runtime": 18665.3611,
+      "train_samples_per_second": 7.728,
+      "train_steps_per_second": 0.966
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 18031,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 2.80353585038046e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}