diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8749 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 12450,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0024099289070972406,
+      "grad_norm": 2.472470998764038,
+      "learning_rate": 7.228915662650603e-08,
+      "loss": 0.4414,
+      "step": 10
+    },
+    {
+      "epoch": 0.004819857814194481,
+      "grad_norm": 2.5485317707061768,
+      "learning_rate": 1.526104417670683e-07,
+      "loss": 0.4342,
+      "step": 20
+    },
+    {
+      "epoch": 0.007229786721291722,
+      "grad_norm": 2.1985578536987305,
+      "learning_rate": 2.3293172690763053e-07,
+      "loss": 0.4302,
+      "step": 30
+    },
+    {
+      "epoch": 0.009639715628388962,
+      "grad_norm": 1.616897463798523,
+      "learning_rate": 3.1325301204819284e-07,
+      "loss": 0.4214,
+      "step": 40
+    },
+    {
+      "epoch": 0.012049644535486204,
+      "grad_norm": 1.18423593044281,
+      "learning_rate": 3.93574297188755e-07,
+      "loss": 0.4146,
+      "step": 50
+    },
+    {
+      "epoch": 0.014459573442583444,
+      "grad_norm": 0.7588810920715332,
+      "learning_rate": 4.738955823293173e-07,
+      "loss": 0.3957,
+      "step": 60
+    },
+    {
+      "epoch": 0.016869502349680685,
+      "grad_norm": 0.5396777987480164,
+      "learning_rate": 5.542168674698796e-07,
+      "loss": 0.3805,
+      "step": 70
+    },
+    {
+      "epoch": 0.019279431256777925,
+      "grad_norm": 0.37596839666366577,
+      "learning_rate": 6.345381526104419e-07,
+      "loss": 0.3605,
+      "step": 80
+    },
+    {
+      "epoch": 0.021689360163875165,
+      "grad_norm": 0.29817959666252136,
+      "learning_rate": 7.14859437751004e-07,
+      "loss": 0.3513,
+      "step": 90
+    },
+    {
+      "epoch": 0.024099289070972408,
+      "grad_norm": 0.25247910618782043,
+      "learning_rate": 7.951807228915663e-07,
+      "loss": 0.3382,
+      "step": 100
+    },
+    {
+      "epoch": 0.026509217978069648,
+      "grad_norm": 0.4067523181438446,
+      "learning_rate": 8.755020080321286e-07,
+      "loss": 0.3405,
+      "step": 110
+    },
+    {
+      "epoch": 0.028919146885166887,
+      "grad_norm": 0.18429981172084808,
+      "learning_rate": 9.558232931726909e-07,
+      "loss": 0.3298,
+      "step": 120
+    },
+    {
+      "epoch": 0.03132907579226413,
+      "grad_norm": 0.24616770446300507,
+      "learning_rate": 1.0361445783132532e-06,
+      "loss": 0.3259,
+      "step": 130
+    },
+    {
+      "epoch": 0.03373900469936137,
+      "grad_norm": 0.17981302738189697,
+      "learning_rate": 1.1164658634538152e-06,
+      "loss": 0.3171,
+      "step": 140
+    },
+    {
+      "epoch": 0.03614893360645861,
+      "grad_norm": 0.17655853927135468,
+      "learning_rate": 1.1967871485943775e-06,
+      "loss": 0.3173,
+      "step": 150
+    },
+    {
+      "epoch": 0.03855886251355585,
+      "grad_norm": 0.15255206823349,
+      "learning_rate": 1.2771084337349398e-06,
+      "loss": 0.3153,
+      "step": 160
+    },
+    {
+      "epoch": 0.04096879142065309,
+      "grad_norm": 0.1549864560365677,
+      "learning_rate": 1.357429718875502e-06,
+      "loss": 0.3071,
+      "step": 170
+    },
+    {
+      "epoch": 0.04337872032775033,
+      "grad_norm": 0.16035476326942444,
+      "learning_rate": 1.4377510040160644e-06,
+      "loss": 0.3071,
+      "step": 180
+    },
+    {
+      "epoch": 0.04578864923484757,
+      "grad_norm": 0.17461898922920227,
+      "learning_rate": 1.5180722891566266e-06,
+      "loss": 0.3056,
+      "step": 190
+    },
+    {
+      "epoch": 0.048198578141944816,
+      "grad_norm": 0.14300084114074707,
+      "learning_rate": 1.598393574297189e-06,
+      "loss": 0.3018,
+      "step": 200
+    },
+    {
+      "epoch": 0.05060850704904205,
+      "grad_norm": 0.1381385326385498,
+      "learning_rate": 1.6787148594377512e-06,
+      "loss": 0.2994,
+      "step": 210
+    },
+    {
+      "epoch": 0.053018435956139295,
+      "grad_norm": 0.1574939638376236,
+      "learning_rate": 1.7590361445783133e-06,
+      "loss": 0.3013,
+      "step": 220
+    },
+    {
+      "epoch": 0.05542836486323653,
+      "grad_norm": 0.15499268472194672,
+      "learning_rate": 1.8393574297188758e-06,
+      "loss": 0.2931,
+      "step": 230
+    },
+    {
+      "epoch": 0.057838293770333775,
+      "grad_norm": 0.1661788523197174,
+      "learning_rate": 1.919678714859438e-06,
+      "loss": 0.2936,
+      "step": 240
+    },
+    {
+      "epoch": 0.06024822267743102,
+      "grad_norm": 0.16183660924434662,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.2913,
+      "step": 250
+    },
+    {
+      "epoch": 0.06265815158452825,
+      "grad_norm": 0.15030546486377716,
+      "learning_rate": 2.0803212851405624e-06,
+      "loss": 0.2914,
+      "step": 260
+    },
+    {
+      "epoch": 0.06506808049162549,
+      "grad_norm": 0.16651834547519684,
+      "learning_rate": 2.1606425702811245e-06,
+      "loss": 0.2896,
+      "step": 270
+    },
+    {
+      "epoch": 0.06747800939872274,
+      "grad_norm": 0.1691359132528305,
+      "learning_rate": 2.240963855421687e-06,
+      "loss": 0.2855,
+      "step": 280
+    },
+    {
+      "epoch": 0.06988793830581998,
+      "grad_norm": 0.15761148929595947,
+      "learning_rate": 2.321285140562249e-06,
+      "loss": 0.2865,
+      "step": 290
+    },
+    {
+      "epoch": 0.07229786721291721,
+      "grad_norm": 0.17686566710472107,
+      "learning_rate": 2.4016064257028115e-06,
+      "loss": 0.2854,
+      "step": 300
+    },
+    {
+      "epoch": 0.07470779612001446,
+      "grad_norm": 0.15508554875850677,
+      "learning_rate": 2.4819277108433736e-06,
+      "loss": 0.2792,
+      "step": 310
+    },
+    {
+      "epoch": 0.0771177250271117,
+      "grad_norm": 0.15847276151180267,
+      "learning_rate": 2.5622489959839357e-06,
+      "loss": 0.2821,
+      "step": 320
+    },
+    {
+      "epoch": 0.07952765393420894,
+      "grad_norm": 0.18257629871368408,
+      "learning_rate": 2.642570281124498e-06,
+      "loss": 0.2771,
+      "step": 330
+    },
+    {
+      "epoch": 0.08193758284130619,
+      "grad_norm": 0.173823282122612,
+      "learning_rate": 2.7228915662650607e-06,
+      "loss": 0.277,
+      "step": 340
+    },
+    {
+      "epoch": 0.08434751174840342,
+      "grad_norm": 0.16445614397525787,
+      "learning_rate": 2.8032128514056227e-06,
+      "loss": 0.2771,
+      "step": 350
+    },
+    {
+      "epoch": 0.08675744065550066,
+      "grad_norm": 0.1713678389787674,
+      "learning_rate": 2.883534136546185e-06,
+      "loss": 0.2758,
+      "step": 360
+    },
+    {
+      "epoch": 0.08916736956259791,
+      "grad_norm": 0.1728232353925705,
+      "learning_rate": 2.9638554216867473e-06,
+      "loss": 0.2744,
+      "step": 370
+    },
+    {
+      "epoch": 0.09157729846969515,
+      "grad_norm": 0.15372931957244873,
+      "learning_rate": 3.0441767068273094e-06,
+      "loss": 0.2742,
+      "step": 380
+    },
+    {
+      "epoch": 0.09398722737679238,
+      "grad_norm": 0.23547357320785522,
+      "learning_rate": 3.124497991967872e-06,
+      "loss": 0.2737,
+      "step": 390
+    },
+    {
+      "epoch": 0.09639715628388963,
+      "grad_norm": 0.19922538101673126,
+      "learning_rate": 3.204819277108434e-06,
+      "loss": 0.2741,
+      "step": 400
+    },
+    {
+      "epoch": 0.09880708519098687,
+      "grad_norm": 0.1566275954246521,
+      "learning_rate": 3.2851405622489964e-06,
+      "loss": 0.267,
+      "step": 410
+    },
+    {
+      "epoch": 0.1012170140980841,
+      "grad_norm": 0.18829286098480225,
+      "learning_rate": 3.3654618473895585e-06,
+      "loss": 0.2697,
+      "step": 420
+    },
+    {
+      "epoch": 0.10362694300518134,
+      "grad_norm": 0.16823740303516388,
+      "learning_rate": 3.4457831325301206e-06,
+      "loss": 0.2736,
+      "step": 430
+    },
+    {
+      "epoch": 0.10603687191227859,
+      "grad_norm": 0.16835574805736542,
+      "learning_rate": 3.526104417670683e-06,
+      "loss": 0.2667,
+      "step": 440
+    },
+    {
+      "epoch": 0.10844680081937583,
+      "grad_norm": 0.16501298546791077,
+      "learning_rate": 3.6064257028112455e-06,
+      "loss": 0.2681,
+      "step": 450
+    },
+    {
+      "epoch": 0.11085672972647306,
+      "grad_norm": 0.16061000525951385,
+      "learning_rate": 3.6867469879518076e-06,
+      "loss": 0.2702,
+      "step": 460
+    },
+    {
+      "epoch": 0.11326665863357031,
+      "grad_norm": 0.16026341915130615,
+      "learning_rate": 3.7670682730923697e-06,
+      "loss": 0.2666,
+      "step": 470
+    },
+    {
+      "epoch": 0.11567658754066755,
+      "grad_norm": 0.16079480946063995,
+      "learning_rate": 3.847389558232932e-06,
+      "loss": 0.2667,
+      "step": 480
+    },
+    {
+      "epoch": 0.11808651644776479,
+      "grad_norm": 0.16454647481441498,
+      "learning_rate": 3.927710843373494e-06,
+      "loss": 0.2646,
+      "step": 490
+    },
+    {
+      "epoch": 0.12049644535486204,
+      "grad_norm": 0.20878982543945312,
+      "learning_rate": 4.008032128514057e-06,
+      "loss": 0.2686,
+      "step": 500
+    },
+    {
+      "epoch": 0.12290637426195927,
+      "grad_norm": 0.22473016381263733,
+      "learning_rate": 4.088353413654618e-06,
+      "loss": 0.264,
+      "step": 510
+    },
+    {
+      "epoch": 0.1253163031690565,
+      "grad_norm": 0.16910991072654724,
+      "learning_rate": 4.168674698795181e-06,
+      "loss": 0.2685,
+      "step": 520
+    },
+    {
+      "epoch": 0.12772623207615375,
+      "grad_norm": 0.16423989832401276,
+      "learning_rate": 4.248995983935743e-06,
+      "loss": 0.265,
+      "step": 530
+    },
+    {
+      "epoch": 0.13013616098325098,
+      "grad_norm": 0.16456137597560883,
+      "learning_rate": 4.329317269076306e-06,
+      "loss": 0.264,
+      "step": 540
+    },
+    {
+      "epoch": 0.13254608989034825,
+      "grad_norm": 0.22010430693626404,
+      "learning_rate": 4.4096385542168675e-06,
+      "loss": 0.2629,
+      "step": 550
+    },
+    {
+      "epoch": 0.13495601879744548,
+      "grad_norm": 0.186540886759758,
+      "learning_rate": 4.48995983935743e-06,
+      "loss": 0.2606,
+      "step": 560
+    },
+    {
+      "epoch": 0.13736594770454272,
+      "grad_norm": 0.1970025897026062,
+      "learning_rate": 4.5702811244979925e-06,
+      "loss": 0.2649,
+      "step": 570
+    },
+    {
+      "epoch": 0.13977587661163995,
+      "grad_norm": 0.18762168288230896,
+      "learning_rate": 4.650602409638554e-06,
+      "loss": 0.2628,
+      "step": 580
+    },
+    {
+      "epoch": 0.1421858055187372,
+      "grad_norm": 0.18925753235816956,
+      "learning_rate": 4.730923694779117e-06,
+      "loss": 0.2636,
+      "step": 590
+    },
+    {
+      "epoch": 0.14459573442583443,
+      "grad_norm": 0.24714338779449463,
+      "learning_rate": 4.811244979919679e-06,
+      "loss": 0.2587,
+      "step": 600
+    },
+    {
+      "epoch": 0.1470056633329317,
+      "grad_norm": 0.18089915812015533,
+      "learning_rate": 4.891566265060242e-06,
+      "loss": 0.2586,
+      "step": 610
+    },
+    {
+      "epoch": 0.14941559224002893,
+      "grad_norm": 0.21872669458389282,
+      "learning_rate": 4.971887550200803e-06,
+      "loss": 0.2596,
+      "step": 620
+    },
+    {
+      "epoch": 0.15182552114712616,
+      "grad_norm": 0.23077884316444397,
+      "learning_rate": 5.052208835341366e-06,
+      "loss": 0.2592,
+      "step": 630
+    },
+    {
+      "epoch": 0.1542354500542234,
+      "grad_norm": 0.20345240831375122,
+      "learning_rate": 5.132530120481927e-06,
+      "loss": 0.2582,
+      "step": 640
+    },
+    {
+      "epoch": 0.15664537896132064,
+      "grad_norm": 0.1861467957496643,
+      "learning_rate": 5.21285140562249e-06,
+      "loss": 0.2584,
+      "step": 650
+    },
+    {
+      "epoch": 0.15905530786841787,
+      "grad_norm": 0.18403823673725128,
+      "learning_rate": 5.293172690763053e-06,
+      "loss": 0.2572,
+      "step": 660
+    },
+    {
+      "epoch": 0.1614652367755151,
+      "grad_norm": 0.19485147297382355,
+      "learning_rate": 5.373493975903615e-06,
+      "loss": 0.2596,
+      "step": 670
+    },
+    {
+      "epoch": 0.16387516568261237,
+      "grad_norm": 0.26093626022338867,
+      "learning_rate": 5.453815261044177e-06,
+      "loss": 0.2566,
+      "step": 680
+    },
+    {
+      "epoch": 0.1662850945897096,
+      "grad_norm": 0.19746626913547516,
+      "learning_rate": 5.53413654618474e-06,
+      "loss": 0.2585,
+      "step": 690
+    },
+    {
+      "epoch": 0.16869502349680685,
+      "grad_norm": 0.19504830241203308,
+      "learning_rate": 5.6144578313253015e-06,
+      "loss": 0.2528,
+      "step": 700
+    },
+    {
+      "epoch": 0.17110495240390408,
+      "grad_norm": 0.25125887989997864,
+      "learning_rate": 5.694779116465864e-06,
+      "loss": 0.2533,
+      "step": 710
+    },
+    {
+      "epoch": 0.17351488131100132,
+      "grad_norm": 0.1774957925081253,
+      "learning_rate": 5.775100401606426e-06,
+      "loss": 0.2547,
+      "step": 720
+    },
+    {
+      "epoch": 0.17592481021809855,
+      "grad_norm": 0.2207845002412796,
+      "learning_rate": 5.855421686746988e-06,
+      "loss": 0.259,
+      "step": 730
+    },
+    {
+      "epoch": 0.17833473912519582,
+      "grad_norm": 0.20615459978580475,
+      "learning_rate": 5.935742971887551e-06,
+      "loss": 0.2585,
+      "step": 740
+    },
+    {
+      "epoch": 0.18074466803229305,
+      "grad_norm": 0.19467566907405853,
+      "learning_rate": 6.016064257028112e-06,
+      "loss": 0.2561,
+      "step": 750
+    },
+    {
+      "epoch": 0.1831545969393903,
+      "grad_norm": 0.18339978158473969,
+      "learning_rate": 6.096385542168676e-06,
+      "loss": 0.2502,
+      "step": 760
+    },
+    {
+      "epoch": 0.18556452584648753,
+      "grad_norm": 0.20017538964748383,
+      "learning_rate": 6.176706827309238e-06,
+      "loss": 0.2546,
+      "step": 770
+    },
+    {
+      "epoch": 0.18797445475358476,
+      "grad_norm": 0.17474955320358276,
+      "learning_rate": 6.2570281124498e-06,
+      "loss": 0.2515,
+      "step": 780
+    },
+    {
+      "epoch": 0.190384383660682,
+      "grad_norm": 0.21440750360488892,
+      "learning_rate": 6.337349397590362e-06,
+      "loss": 0.2564,
+      "step": 790
+    },
+    {
+      "epoch": 0.19279431256777926,
+      "grad_norm": 0.2097756266593933,
+      "learning_rate": 6.417670682730924e-06,
+      "loss": 0.254,
+      "step": 800
+    },
+    {
+      "epoch": 0.1952042414748765,
+      "grad_norm": 0.1950419396162033,
+      "learning_rate": 6.4979919678714864e-06,
+      "loss": 0.2523,
+      "step": 810
+    },
+    {
+      "epoch": 0.19761417038197374,
+      "grad_norm": 0.22887657582759857,
+      "learning_rate": 6.578313253012049e-06,
+      "loss": 0.2537,
+      "step": 820
+    },
+    {
+      "epoch": 0.20002409928907097,
+      "grad_norm": 0.27371925115585327,
+      "learning_rate": 6.6586345381526106e-06,
+      "loss": 0.2552,
+      "step": 830
+    },
+    {
+      "epoch": 0.2024340281961682,
+      "grad_norm": 0.23672780394554138,
+      "learning_rate": 6.738955823293173e-06,
+      "loss": 0.2513,
+      "step": 840
+    },
+    {
+      "epoch": 0.20484395710326544,
+      "grad_norm": 0.19304253160953522,
+      "learning_rate": 6.819277108433735e-06,
+      "loss": 0.2533,
+      "step": 850
+    },
+    {
+      "epoch": 0.20725388601036268,
+      "grad_norm": 0.199726864695549,
+      "learning_rate": 6.899598393574298e-06,
+      "loss": 0.251,
+      "step": 860
+    },
+    {
+      "epoch": 0.20966381491745995,
+      "grad_norm": 0.19144225120544434,
+      "learning_rate": 6.9799196787148605e-06,
+      "loss": 0.2525,
+      "step": 870
+    },
+    {
+      "epoch": 0.21207374382455718,
+      "grad_norm": 0.20240966975688934,
+      "learning_rate": 7.060240963855422e-06,
+      "loss": 0.2515,
+      "step": 880
+    },
+    {
+      "epoch": 0.21448367273165442,
+      "grad_norm": 0.22919943928718567,
+      "learning_rate": 7.140562248995985e-06,
+      "loss": 0.251,
+      "step": 890
+    },
+    {
+      "epoch": 0.21689360163875165,
+      "grad_norm": 0.23874454200267792,
+      "learning_rate": 7.220883534136547e-06,
+      "loss": 0.2475,
+      "step": 900
+    },
+    {
+      "epoch": 0.2193035305458489,
+      "grad_norm": 0.19525328278541565,
+      "learning_rate": 7.301204819277109e-06,
+      "loss": 0.2493,
+      "step": 910
+    },
+    {
+      "epoch": 0.22171345945294613,
+      "grad_norm": 0.2014315128326416,
+      "learning_rate": 7.381526104417671e-06,
+      "loss": 0.252,
+      "step": 920
+    },
+    {
+      "epoch": 0.2241233883600434,
+      "grad_norm": 0.19596624374389648,
+      "learning_rate": 7.461847389558233e-06,
+      "loss": 0.2501,
+      "step": 930
+    },
+    {
+      "epoch": 0.22653331726714063,
+      "grad_norm": 0.19004347920417786,
+      "learning_rate": 7.5421686746987955e-06,
+      "loss": 0.2498,
+      "step": 940
+    },
+    {
+      "epoch": 0.22894324617423786,
+      "grad_norm": 0.23779860138893127,
+      "learning_rate": 7.622489959839358e-06,
+      "loss": 0.2497,
+      "step": 950
+    },
+    {
+      "epoch": 0.2313531750813351,
+      "grad_norm": 0.23027803003787994,
+      "learning_rate": 7.702811244979921e-06,
+      "loss": 0.25,
+      "step": 960
+    },
+    {
+      "epoch": 0.23376310398843234,
+      "grad_norm": 0.21898239850997925,
+      "learning_rate": 7.783132530120484e-06,
+      "loss": 0.2493,
+      "step": 970
+    },
+    {
+      "epoch": 0.23617303289552957,
+      "grad_norm": 0.18939456343650818,
+      "learning_rate": 7.863453815261045e-06,
+      "loss": 0.2457,
+      "step": 980
+    },
+    {
+      "epoch": 0.23858296180262684,
+      "grad_norm": 0.19975551962852478,
+      "learning_rate": 7.943775100401607e-06,
+      "loss": 0.2479,
+      "step": 990
+    },
+    {
+      "epoch": 0.24099289070972407,
+      "grad_norm": 0.1990450918674469,
+      "learning_rate": 8.02409638554217e-06,
+      "loss": 0.2502,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2434028196168213,
+      "grad_norm": 0.21345728635787964,
+      "learning_rate": 8.104417670682732e-06,
+      "loss": 0.2486,
+      "step": 1010
+    },
+    {
+      "epoch": 0.24581274852391855,
+      "grad_norm": 0.24549713730812073,
+      "learning_rate": 8.184738955823293e-06,
+      "loss": 0.2492,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24822267743101578,
+      "grad_norm": 0.20228305459022522,
+      "learning_rate": 8.265060240963855e-06,
+      "loss": 0.2473,
+      "step": 1030
+    },
+    {
+      "epoch": 0.250632606338113,
+      "grad_norm": 0.1969956010580063,
+      "learning_rate": 8.345381526104418e-06,
+      "loss": 0.2477,
+      "step": 1040
+    },
+    {
+      "epoch": 0.25304253524521025,
+      "grad_norm": 0.23591509461402893,
+      "learning_rate": 8.42570281124498e-06,
+      "loss": 0.2468,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2554524641523075,
+      "grad_norm": 0.21046389639377594,
+      "learning_rate": 8.506024096385543e-06,
+      "loss": 0.2479,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2578623930594047,
+      "grad_norm": 0.2074204385280609,
+      "learning_rate": 8.586345381526105e-06,
+      "loss": 0.2452,
+      "step": 1070
+    },
+    {
+      "epoch": 0.26027232196650196,
+      "grad_norm": 0.3702162206172943,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.2477,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26268225087359925,
+      "grad_norm": 0.21479126811027527,
+      "learning_rate": 8.74698795180723e-06,
+      "loss": 0.2486,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2650921797806965,
+      "grad_norm": 0.20660068094730377,
+      "learning_rate": 8.827309236947791e-06,
+      "loss": 0.2465,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2675021086877937,
+      "grad_norm": 0.2068599909543991,
+      "learning_rate": 8.907630522088354e-06,
+      "loss": 0.2512,
+      "step": 1110
+    },
+    {
+      "epoch": 0.26991203759489096,
+      "grad_norm": 0.2575608789920807,
+      "learning_rate": 8.987951807228916e-06,
+      "loss": 0.2476,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2723219665019882,
+      "grad_norm": 0.19399215281009674,
+      "learning_rate": 9.068273092369479e-06,
+      "loss": 0.2435,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27473189540908544,
+      "grad_norm": 0.2270069122314453,
+      "learning_rate": 9.148594377510041e-06,
+      "loss": 0.248,
+      "step": 1140
+    },
+    {
+      "epoch": 0.27714182431618267,
+      "grad_norm": 0.22409838438034058,
+      "learning_rate": 9.228915662650602e-06,
+      "loss": 0.2431,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2795517532232799,
+      "grad_norm": 0.24383105337619781,
+      "learning_rate": 9.309236947791166e-06,
+      "loss": 0.2457,
+      "step": 1160
+    },
+    {
+      "epoch": 0.28196168213037714,
+      "grad_norm": 0.2236703783273697,
+      "learning_rate": 9.389558232931729e-06,
+      "loss": 0.243,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2843716110374744,
+      "grad_norm": 0.22612161934375763,
+      "learning_rate": 9.46987951807229e-06,
+      "loss": 0.2448,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2867815399445716,
+      "grad_norm": 0.23215730488300323,
+      "learning_rate": 9.550200803212852e-06,
+      "loss": 0.2418,
+      "step": 1190
+    },
+    {
+      "epoch": 0.28919146885166885,
+      "grad_norm": 0.22533060610294342,
+      "learning_rate": 9.630522088353414e-06,
+      "loss": 0.2425,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2916013977587661,
+      "grad_norm": 0.22571276128292084,
+      "learning_rate": 9.710843373493977e-06,
+      "loss": 0.2455,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2940113266658634,
+      "grad_norm": 0.24537992477416992,
+      "learning_rate": 9.79116465863454e-06,
+      "loss": 0.2394,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2964212555729606,
+      "grad_norm": 0.21820850670337677,
+      "learning_rate": 9.8714859437751e-06,
+      "loss": 0.2464,
+      "step": 1230
+    },
+    {
+      "epoch": 0.29883118448005785,
+      "grad_norm": 0.2275010347366333,
+      "learning_rate": 9.951807228915663e-06,
+      "loss": 0.2427,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3012411133871551,
+      "grad_norm": 0.1905357539653778,
+      "learning_rate": 9.999996855613166e-06,
+      "loss": 0.243,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3036510422942523,
+      "grad_norm": 0.2797210216522217,
+      "learning_rate": 9.999961481306676e-06,
+      "loss": 0.2425,
+      "step": 1260
+    },
+    {
+      "epoch": 0.30606097120134956,
+      "grad_norm": 0.1980619579553604,
+      "learning_rate": 9.999886802489159e-06,
+      "loss": 0.2403,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3084709001084468,
+      "grad_norm": 0.20793509483337402,
+      "learning_rate": 9.999772819747658e-06,
+      "loss": 0.2436,
+      "step": 1280
+    },
+    {
+      "epoch": 0.31088082901554404,
+      "grad_norm": 0.20385341346263885,
+      "learning_rate": 9.99961953397819e-06,
+      "loss": 0.2465,
+      "step": 1290
+    },
+    {
+      "epoch": 0.31329075792264127,
+      "grad_norm": 0.20672109723091125,
+      "learning_rate": 9.999426946385727e-06,
+      "loss": 0.2467,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3157006868297385,
+      "grad_norm": 0.1893812119960785,
+      "learning_rate": 9.999195058484192e-06,
+      "loss": 0.2435,
+      "step": 1310
+    },
+    {
+      "epoch": 0.31811061573683574,
+      "grad_norm": 0.19826377928256989,
+      "learning_rate": 9.998923872096449e-06,
+      "loss": 0.2444,
+      "step": 1320
+    },
+    {
+      "epoch": 0.320520544643933,
+      "grad_norm": 0.25412455201148987,
+      "learning_rate": 9.998613389354283e-06,
+      "loss": 0.2404,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3229304735510302,
+      "grad_norm": 0.22612008452415466,
+      "learning_rate": 9.998263612698386e-06,
+      "loss": 0.2457,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3253404024581275,
+      "grad_norm": 0.2398287057876587,
+      "learning_rate": 9.997874544878343e-06,
+      "loss": 0.2417,
+      "step": 1350
+    },
+    {
+      "epoch": 0.32775033136522475,
+      "grad_norm": 0.26140671968460083,
+      "learning_rate": 9.997446188952599e-06,
+      "loss": 0.2368,
+      "step": 1360
+    },
+    {
+      "epoch": 0.330160260272322,
+      "grad_norm": 0.19120003283023834,
+      "learning_rate": 9.996978548288446e-06,
+      "loss": 0.245,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3325701891794192,
+      "grad_norm": 0.21241794526576996,
+      "learning_rate": 9.996471626561988e-06,
+      "loss": 0.2424,
+      "step": 1380
+    },
+    {
+      "epoch": 0.33498011808651645,
+      "grad_norm": 0.20897141098976135,
+      "learning_rate": 9.995925427758117e-06,
+      "loss": 0.2424,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3373900469936137,
+      "grad_norm": 0.2229437381029129,
+      "learning_rate": 9.995339956170482e-06,
+      "loss": 0.2404,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3397999759007109,
+      "grad_norm": 0.23609474301338196,
+      "learning_rate": 9.994715216401457e-06,
+      "loss": 0.2392,
+      "step": 1410
+    },
+    {
+      "epoch": 0.34220990480780816,
+      "grad_norm": 0.20246021449565887,
+      "learning_rate": 9.994051213362091e-06,
+      "loss": 0.2401,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3446198337149054,
+      "grad_norm": 0.18765322864055634,
+      "learning_rate": 9.993347952272095e-06,
+      "loss": 0.2351,
+      "step": 1430
+    },
+    {
+      "epoch": 0.34702976262200264,
+      "grad_norm": 0.23499734699726105,
+      "learning_rate": 9.992605438659773e-06,
+      "loss": 0.2392,
+      "step": 1440
+    },
+    {
+      "epoch": 0.34943969152909987,
+      "grad_norm": 0.2348964661359787,
+      "learning_rate": 9.991823678361997e-06,
+      "loss": 0.2405,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3518496204361971,
+      "grad_norm": 0.21696138381958008,
+      "learning_rate": 9.991002677524158e-06,
+      "loss": 0.2385,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3542595493432944,
+      "grad_norm": 0.2489856481552124,
+      "learning_rate": 9.990142442600113e-06,
+      "loss": 0.2426,
+      "step": 1470
+    },
+    {
+      "epoch": 0.35666947825039164,
+      "grad_norm": 0.20439158380031586,
+      "learning_rate": 9.989242980352134e-06,
+      "loss": 0.2422,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3590794071574889,
+      "grad_norm": 0.23717837035655975,
+      "learning_rate": 9.988304297850864e-06,
+      "loss": 0.2405,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3614893360645861,
+      "grad_norm": 0.22787103056907654,
+      "learning_rate": 9.987326402475246e-06,
+      "loss": 0.2417,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36389926497168334,
+      "grad_norm": 0.2548515796661377,
+      "learning_rate": 9.986309301912484e-06,
+      "loss": 0.2365,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3663091938787806,
+      "grad_norm": 0.2120496928691864,
+      "learning_rate": 9.985253004157967e-06,
+      "loss": 0.2382,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3687191227858778,
+      "grad_norm": 0.22003258764743805,
+      "learning_rate": 9.984157517515209e-06,
+      "loss": 0.2358,
+      "step": 1530
+    },
+    {
+      "epoch": 0.37112905169297505,
+      "grad_norm": 0.24021808803081512,
+      "learning_rate": 9.983022850595794e-06,
+      "loss": 0.2379,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3735389806000723,
+      "grad_norm": 0.17676447331905365,
+      "learning_rate": 9.981849012319294e-06,
+      "loss": 0.2374,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3759489095071695,
+      "grad_norm": 0.21659384667873383,
+      "learning_rate": 9.980636011913207e-06,
+      "loss": 0.2378,
+      "step": 1560
+    },
+    {
+      "epoch": 0.37835883841426676,
+      "grad_norm": 0.21923330426216125,
+      "learning_rate": 9.979383858912886e-06,
+      "loss": 0.2383,
+      "step": 1570
+    },
+    {
+      "epoch": 0.380768767321364,
+      "grad_norm": 0.1923467069864273,
+      "learning_rate": 9.97809256316146e-06,
+      "loss": 0.2358,
+      "step": 1580
+    },
+    {
+      "epoch": 0.38317869622846124,
+      "grad_norm": 0.2098626345396042,
+      "learning_rate": 9.976762134809752e-06,
+      "loss": 0.2391,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3855886251355585,
+      "grad_norm": 0.22998303174972534,
+      "learning_rate": 9.975392584316215e-06,
+      "loss": 0.2356,
+      "step": 1600
+    },
+    {
+      "epoch": 0.38799855404265576,
+      "grad_norm": 0.21946196258068085,
+      "learning_rate": 9.973983922446832e-06,
+      "loss": 0.2389,
+      "step": 1610
+    },
+    {
+      "epoch": 0.390408482949753,
+      "grad_norm": 0.2061765342950821,
+      "learning_rate": 9.972536160275042e-06,
+      "loss": 0.2352,
+      "step": 1620
+    },
+    {
+      "epoch": 0.39281841185685024,
+      "grad_norm": 0.24974577128887177,
+      "learning_rate": 9.971049309181648e-06,
+      "loss": 0.237,
+      "step": 1630
+    },
+    {
+      "epoch": 0.39522834076394747,
+      "grad_norm": 0.2364003211259842,
+      "learning_rate": 9.969523380854736e-06,
+      "loss": 0.239,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3976382696710447,
+      "grad_norm": 0.23062963783740997,
+      "learning_rate": 9.967958387289564e-06,
+      "loss": 0.2396,
+      "step": 1650
+    },
+    {
+      "epoch": 0.40004819857814194,
+      "grad_norm": 0.17892813682556152,
+      "learning_rate": 9.966354340788496e-06,
+      "loss": 0.2352,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4024581274852392,
+      "grad_norm": 0.19819405674934387,
+      "learning_rate": 9.964711253960877e-06,
+      "loss": 0.2364,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4048680563923364,
+      "grad_norm": 0.2314508855342865,
+      "learning_rate": 9.963029139722952e-06,
+      "loss": 0.2371,
+      "step": 1680
+    },
+    {
+      "epoch": 0.40727798529943365,
+      "grad_norm": 0.21803642809391022,
+      "learning_rate": 9.96130801129776e-06,
+      "loss": 0.2354,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4096879142065309,
+      "grad_norm": 0.23534069955348969,
+      "learning_rate": 9.959547882215025e-06,
+      "loss": 0.2366,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4120978431136281,
+      "grad_norm": 0.22902339696884155,
+      "learning_rate": 9.957748766311059e-06,
+      "loss": 0.2393,
+      "step": 1710
+    },
+    {
+      "epoch": 0.41450777202072536,
+      "grad_norm": 0.2218971848487854,
+      "learning_rate": 9.955910677728643e-06,
+      "loss": 0.2377,
+      "step": 1720
+    },
+    {
+      "epoch": 0.41691770092782265,
+      "grad_norm": 0.1902938187122345,
+      "learning_rate": 9.954033630916926e-06,
+      "loss": 0.2339,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4193276298349199,
+      "grad_norm": 0.16372597217559814,
+      "learning_rate": 9.952117640631298e-06,
+      "loss": 0.2336,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4217375587420171,
+      "grad_norm": 0.24245886504650116,
+      "learning_rate": 9.950162721933292e-06,
+      "loss": 0.2362,
+      "step": 1750
+    },
+    {
+      "epoch": 0.42414748764911436,
+      "grad_norm": 0.19669395685195923,
+      "learning_rate": 9.948168890190448e-06,
+      "loss": 0.2332,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4265574165562116,
+      "grad_norm": 0.1758386343717575,
+      "learning_rate": 9.946136161076205e-06,
+      "loss": 0.235,
+      "step": 1770
+    },
+    {
+      "epoch": 0.42896734546330884,
+      "grad_norm": 0.1818431317806244,
+      "learning_rate": 9.94406455056977e-06,
+      "loss": 0.2368,
+      "step": 1780
+    },
+    {
+      "epoch": 0.43137727437040607,
+      "grad_norm": 0.1829291433095932,
+      "learning_rate": 9.941954074955995e-06,
+      "loss": 0.2307,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4337872032775033,
+      "grad_norm": 0.18251778185367584,
+      "learning_rate": 9.939804750825253e-06,
+      "loss": 0.2329,
+      "step": 1800
+    },
+    {
+      "epoch": 0.43619713218460054,
+      "grad_norm": 0.19176249206066132,
+      "learning_rate": 9.937616595073299e-06,
+      "loss": 0.2348,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4386070610916978,
+      "grad_norm": 0.19189012050628662,
+      "learning_rate": 9.935389624901143e-06,
+      "loss": 0.2368,
+      "step": 1820
+    },
+    {
+      "epoch": 0.441016989998795,
+      "grad_norm": 0.1790163666009903,
+      "learning_rate": 9.933123857814917e-06,
+      "loss": 0.2354,
+      "step": 1830
+    },
+    {
+      "epoch": 0.44342691890589225,
+      "grad_norm": 0.2602790892124176,
+      "learning_rate": 9.93081931162573e-06,
+      "loss": 0.2341,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4458368478129895,
+      "grad_norm": 0.16906730830669403,
+      "learning_rate": 9.928476004449534e-06,
+      "loss": 0.2334,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4482467767200868,
+      "grad_norm": 0.1677951216697693,
+      "learning_rate": 9.926093954706982e-06,
+      "loss": 0.2338,
+      "step": 1860
+    },
+    {
+      "epoch": 0.450656705627184,
+      "grad_norm": 0.1938229352235794,
+      "learning_rate": 9.923673181123273e-06,
+      "loss": 0.2369,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45306663453428125,
+      "grad_norm": 0.18625344336032867,
+      "learning_rate": 9.921213702728023e-06,
+      "loss": 0.2324,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4554765634413785,
+      "grad_norm": 0.20547117292881012,
+      "learning_rate": 9.918715538855098e-06,
+      "loss": 0.2332,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4578864923484757,
+      "grad_norm": 0.20020659267902374,
+      "learning_rate": 9.916178709142472e-06,
+      "loss": 0.233,
+      "step": 1900
+    },
+    {
+      "epoch": 0.46029642125557296,
+      "grad_norm": 0.24178935587406158,
+      "learning_rate": 9.913603233532067e-06,
+      "loss": 0.2341,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4627063501626702,
+      "grad_norm": 0.17368467152118683,
+      "learning_rate": 9.910989132269604e-06,
+      "loss": 0.2339,
+      "step": 1920
+    },
+    {
+      "epoch": 0.46511627906976744,
+      "grad_norm": 0.19401304423809052,
+      "learning_rate": 9.908336425904432e-06,
+      "loss": 0.2341,
+      "step": 1930
+    },
+    {
+      "epoch": 0.46752620797686467,
+      "grad_norm": 0.19002994894981384,
+      "learning_rate": 9.905645135289378e-06,
+      "loss": 0.2342,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4699361368839619,
+      "grad_norm": 0.21070018410682678,
+      "learning_rate": 9.902915281580581e-06,
+      "loss": 0.2356,
+      "step": 1950
+    },
+    {
+      "epoch": 0.47234606579105914,
+      "grad_norm": 0.19492904841899872,
+      "learning_rate": 9.900146886237316e-06,
+      "loss": 0.2341,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4747559946981564,
+      "grad_norm": 0.2291557788848877,
+      "learning_rate": 9.897339971021836e-06,
+      "loss": 0.2332,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4771659236052537,
+      "grad_norm": 0.21163547039031982,
+      "learning_rate": 9.894494557999195e-06,
+      "loss": 0.232,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4795758525123509,
+      "grad_norm": 0.30296486616134644,
+      "learning_rate": 9.891610669537084e-06,
+      "loss": 0.2293,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48198578141944814,
+      "grad_norm": 0.1773892492055893,
+      "learning_rate": 9.888688328305638e-06,
+      "loss": 0.2324,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4843957103265454,
+      "grad_norm": 0.17265821993350983,
+      "learning_rate": 9.885727557277275e-06,
+      "loss": 0.2322,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4868056392336426,
+      "grad_norm": 0.18319979310035706,
+      "learning_rate": 9.882728379726506e-06,
+      "loss": 0.2334,
+      "step": 2020
+    },
+    {
+      "epoch": 0.48921556814073985,
+      "grad_norm": 0.17308953404426575,
+      "learning_rate": 9.879690819229752e-06,
+      "loss": 0.2269,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4916254970478371,
+      "grad_norm": 0.24003373086452484,
+      "learning_rate": 9.876614899665167e-06,
+      "loss": 0.2328,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4940354259549343,
+      "grad_norm": 0.22242967784404755,
+      "learning_rate": 9.873500645212434e-06,
+      "loss": 0.2328,
+      "step": 2050
+    },
+    {
+      "epoch": 0.49644535486203156,
+      "grad_norm": 0.185471773147583,
+      "learning_rate": 9.870348080352597e-06,
+      "loss": 0.2324,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4988552837691288,
+      "grad_norm": 0.17182640731334686,
+      "learning_rate": 9.867157229867847e-06,
+      "loss": 0.2301,
+      "step": 2070
+    },
+    {
+      "epoch": 0.501265212676226,
+      "grad_norm": 0.19547314941883087,
+      "learning_rate": 9.863928118841344e-06,
+      "loss": 0.2342,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5036751415833233,
+      "grad_norm": 0.2110988199710846,
+      "learning_rate": 9.860660772657008e-06,
+      "loss": 0.2313,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5060850704904205,
+      "grad_norm": 0.2130996584892273,
+      "learning_rate": 9.857355216999324e-06,
+      "loss": 0.229,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5084949993975177,
+      "grad_norm": 0.20759201049804688,
+      "learning_rate": 9.854011477853147e-06,
+      "loss": 0.2325,
+      "step": 2110
+    },
+    {
+      "epoch": 0.510904928304615,
+      "grad_norm": 0.19392450153827667,
+      "learning_rate": 9.850629581503481e-06,
+      "loss": 0.2315,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5133148572117122,
+      "grad_norm": 0.17622078955173492,
+      "learning_rate": 9.847209554535288e-06,
+      "loss": 0.2314,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5157247861188095,
+      "grad_norm": 0.1835789829492569,
+      "learning_rate": 9.843751423833274e-06,
+      "loss": 0.2297,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5181347150259067,
+      "grad_norm": 0.20619632303714752,
+      "learning_rate": 9.840255216581676e-06,
+      "loss": 0.2283,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5205446439330039,
+      "grad_norm": 0.1713174283504486,
+      "learning_rate": 9.836720960264049e-06,
+      "loss": 0.2341,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5229545728401012,
+      "grad_norm": 0.162379190325737,
+      "learning_rate": 9.833148682663048e-06,
+      "loss": 0.2328,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5253645017471985,
+      "grad_norm": 0.16697606444358826,
+      "learning_rate": 9.829538411860218e-06,
+      "loss": 0.231,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5277744306542957,
+      "grad_norm": 0.17996534705162048,
+      "learning_rate": 9.82589017623576e-06,
+      "loss": 0.2326,
+      "step": 2190
+    },
+    {
+      "epoch": 0.530184359561393,
+      "grad_norm": 0.18227118253707886,
+      "learning_rate": 9.822204004468319e-06,
+      "loss": 0.2322,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5325942884684902,
+      "grad_norm": 0.168010875582695,
+      "learning_rate": 9.818479925534755e-06,
+      "loss": 0.2269,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5350042173755875,
+      "grad_norm": 0.1992473155260086,
+      "learning_rate": 9.814717968709912e-06,
+      "loss": 0.2334,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5374141462826847,
+      "grad_norm": 0.22966599464416504,
+      "learning_rate": 9.810918163566396e-06,
+      "loss": 0.2296,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5398240751897819,
+      "grad_norm": 0.1932167261838913,
+      "learning_rate": 9.80708053997433e-06,
+      "loss": 0.23,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5422340040968792,
+      "grad_norm": 0.20305748283863068,
+      "learning_rate": 9.803205128101134e-06,
+      "loss": 0.2325,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5446439330039764,
+      "grad_norm": 0.1645287424325943,
+      "learning_rate": 9.799291958411273e-06,
+      "loss": 0.2311,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5470538619110736,
+      "grad_norm": 0.1746455430984497,
+      "learning_rate": 9.795341061666031e-06,
+      "loss": 0.2302,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5494637908181709,
+      "grad_norm": 0.2108176350593567,
+      "learning_rate": 9.791352468923257e-06,
+      "loss": 0.2325,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5518737197252681,
+      "grad_norm": 0.1855156570672989,
+      "learning_rate": 9.787326211537132e-06,
+      "loss": 0.2269,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5542836486323653,
+      "grad_norm": 0.1825191229581833,
+      "learning_rate": 9.783262321157915e-06,
+      "loss": 0.2314,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5566935775394626,
+      "grad_norm": 0.18101802468299866,
+      "learning_rate": 9.779160829731698e-06,
+      "loss": 0.2311,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5591035064465598,
+      "grad_norm": 0.1984725147485733,
+      "learning_rate": 9.77502176950015e-06,
+      "loss": 0.2296,
+      "step": 2320
+    },
+    {
+      "epoch": 0.561513435353657,
+      "grad_norm": 0.16921648383140564,
+      "learning_rate": 9.770845173000272e-06,
+      "loss": 0.2302,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5639233642607543,
+      "grad_norm": 0.20288076996803284,
+      "learning_rate": 9.766631073064132e-06,
+      "loss": 0.2281,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5663332931678515,
+      "grad_norm": 0.1872728168964386,
+      "learning_rate": 9.762379502818613e-06,
+      "loss": 0.2297,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5687432220749488,
+      "grad_norm": 0.1912074089050293,
+      "learning_rate": 9.758090495685151e-06,
+      "loss": 0.2294,
+      "step": 2360
+    },
+    {
+      "epoch": 0.571153150982046,
+      "grad_norm": 0.20468741655349731,
+      "learning_rate": 9.75376408537947e-06,
+      "loss": 0.2308,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5735630798891432,
+      "grad_norm": 0.19299040734767914,
+      "learning_rate": 9.749400305911323e-06,
+      "loss": 0.2299,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5759730087962405,
+      "grad_norm": 0.1859942078590393,
+      "learning_rate": 9.744999191584214e-06,
+      "loss": 0.2288,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5783829377033377,
+      "grad_norm": 0.17618857324123383,
+      "learning_rate": 9.740560776995142e-06,
+      "loss": 0.232,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5807928666104349,
+      "grad_norm": 0.20461530983448029,
+      "learning_rate": 9.736085097034318e-06,
+      "loss": 0.2292,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5832027955175322,
+      "grad_norm": 0.19141855835914612,
+      "learning_rate": 9.731572186884894e-06,
+      "loss": 0.2281,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5856127244246295,
+      "grad_norm": 0.19507476687431335,
+      "learning_rate": 9.727022082022692e-06,
+      "loss": 0.2249,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5880226533317268,
+      "grad_norm": 0.1954670399427414,
+      "learning_rate": 9.722434818215914e-06,
+      "loss": 0.2315,
+      "step": 2440
+    },
+    {
+      "epoch": 0.590432582238824,
+      "grad_norm": 0.1806354522705078,
+      "learning_rate": 9.71781043152487e-06,
+      "loss": 0.2295,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5928425111459212,
+      "grad_norm": 0.1709500104188919,
+      "learning_rate": 9.713148958301692e-06,
+      "loss": 0.2284,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5952524400530185,
+      "grad_norm": 0.18373407423496246,
+      "learning_rate": 9.708450435190048e-06,
+      "loss": 0.2311,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5976623689601157,
+      "grad_norm": 0.16200312972068787,
+      "learning_rate": 9.703714899124853e-06,
+      "loss": 0.2267,
+      "step": 2480
+    },
+    {
+      "epoch": 0.600072297867213,
+      "grad_norm": 0.19261179864406586,
+      "learning_rate": 9.698942387331983e-06,
+      "loss": 0.2285,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6024822267743102,
+      "grad_norm": 0.18697507679462433,
+      "learning_rate": 9.694132937327969e-06,
+      "loss": 0.2292,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6048921556814074,
+      "grad_norm": 0.19558601081371307,
+      "learning_rate": 9.689286586919721e-06,
+      "loss": 0.231,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6073020845885047,
+      "grad_norm": 0.1981854885816574,
+      "learning_rate": 9.684403374204223e-06,
+      "loss": 0.2322,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6097120134956019,
+      "grad_norm": 0.1691233515739441,
+      "learning_rate": 9.679483337568223e-06,
+      "loss": 0.2303,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6121219424026991,
+      "grad_norm": 0.18886698782444,
+      "learning_rate": 9.674526515687947e-06,
+      "loss": 0.2312,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6145318713097964,
+      "grad_norm": 0.1817430853843689,
+      "learning_rate": 9.669532947528789e-06,
+      "loss": 0.2256,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6169418002168936,
+      "grad_norm": 0.19985781610012054,
+      "learning_rate": 9.664502672345002e-06,
+      "loss": 0.2307,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6193517291239908,
+      "grad_norm": 0.2230524867773056,
+      "learning_rate": 9.65943572967939e-06,
+      "loss": 0.2282,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6217616580310881,
+      "grad_norm": 0.17067363858222961,
+      "learning_rate": 9.654332159363004e-06,
+      "loss": 0.2321,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6241715869381853,
+      "grad_norm": 0.17747369408607483,
+      "learning_rate": 9.649192001514817e-06,
+      "loss": 0.2258,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6265815158452825,
+      "grad_norm": 0.21526795625686646,
+      "learning_rate": 9.64401529654142e-06,
+      "loss": 0.2263,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6289914447523798,
+      "grad_norm": 0.1777113825082779,
+      "learning_rate": 9.638802085136698e-06,
+      "loss": 0.2265,
+      "step": 2610
+    },
+    {
+      "epoch": 0.631401373659477,
+      "grad_norm": 0.1751803159713745,
+      "learning_rate": 9.63355240828151e-06,
+      "loss": 0.2273,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6338113025665743,
+      "grad_norm": 0.19355806708335876,
+      "learning_rate": 9.628266307243373e-06,
+      "loss": 0.2263,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6362212314736715,
+      "grad_norm": 0.22096408903598785,
+      "learning_rate": 9.62294382357613e-06,
+      "loss": 0.2272,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6386311603807687,
+      "grad_norm": 0.17918968200683594,
+      "learning_rate": 9.617584999119624e-06,
+      "loss": 0.2246,
+      "step": 2650
+    },
+    {
+      "epoch": 0.641041089287866,
+      "grad_norm": 0.17709515988826752,
+      "learning_rate": 9.612189875999378e-06,
+      "loss": 0.2243,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6434510181949632,
+      "grad_norm": 0.20072241127490997,
+      "learning_rate": 9.606758496626252e-06,
+      "loss": 0.2283,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6458609471020604,
+      "grad_norm": 0.19501695036888123,
+      "learning_rate": 9.60129090369612e-06,
+      "loss": 0.2283,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6482708760091578,
+      "grad_norm": 0.19082598388195038,
+      "learning_rate": 9.59578714018952e-06,
+      "loss": 0.2278,
+      "step": 2690
+    },
+    {
+      "epoch": 0.650680804916255,
+      "grad_norm": 0.1871589571237564,
+      "learning_rate": 9.590247249371338e-06,
+      "loss": 0.2301,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6530907338233523,
+      "grad_norm": 0.20353522896766663,
+      "learning_rate": 9.584671274790447e-06,
+      "loss": 0.223,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6555006627304495,
+      "grad_norm": 0.18655039370059967,
+      "learning_rate": 9.579059260279376e-06,
+      "loss": 0.2285,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6579105916375467,
+      "grad_norm": 0.2111136019229889,
+      "learning_rate": 9.573411249953963e-06,
+      "loss": 0.2278,
+      "step": 2730
+    },
+    {
+      "epoch": 0.660320520544644,
+      "grad_norm": 0.19638030230998993,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 0.2283,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6627304494517412,
+      "grad_norm": 0.18445482850074768,
+      "learning_rate": 9.562007419737916e-06,
+      "loss": 0.2282,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6651403783588384,
+      "grad_norm": 0.22625213861465454,
+      "learning_rate": 9.556251689492366e-06,
+      "loss": 0.2296,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6675503072659357,
+      "grad_norm": 0.23736052215099335,
+      "learning_rate": 9.550460142721938e-06,
+      "loss": 0.2256,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6699602361730329,
+      "grad_norm": 0.16888026893138885,
+      "learning_rate": 9.544632824953767e-06,
+      "loss": 0.2233,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6723701650801301,
+      "grad_norm": 0.20066499710083008,
+      "learning_rate": 9.538769781996178e-06,
+      "loss": 0.2256,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6747800939872274,
+      "grad_norm": 0.18064813315868378,
+      "learning_rate": 9.532871059938335e-06,
+      "loss": 0.2266,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6771900228943246,
+      "grad_norm": 0.17394624650478363,
+      "learning_rate": 9.526936705149872e-06,
+      "loss": 0.2273,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6795999518014219,
+      "grad_norm": 0.1744656264781952,
+      "learning_rate": 9.520966764280532e-06,
+      "loss": 0.2258,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6820098807085191,
+      "grad_norm": 0.1700519323348999,
+      "learning_rate": 9.514961284259796e-06,
+      "loss": 0.229,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6844198096156163,
+      "grad_norm": 0.20003622770309448,
+      "learning_rate": 9.50892031229652e-06,
+      "loss": 0.2251,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6868297385227136,
+      "grad_norm": 0.19110295176506042,
+      "learning_rate": 9.50284389587856e-06,
+      "loss": 0.2283,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6892396674298108,
+      "grad_norm": 0.18319687247276306,
+      "learning_rate": 9.4967320827724e-06,
+      "loss": 0.2251,
+      "step": 2860
+    },
+    {
+      "epoch": 0.691649596336908,
+      "grad_norm": 0.19341939687728882,
+      "learning_rate": 9.490584921022773e-06,
+      "loss": 0.2262,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6940595252440053,
+      "grad_norm": 0.17186424136161804,
+      "learning_rate": 9.484402458952289e-06,
+      "loss": 0.2253,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6964694541511025,
+      "grad_norm": 0.16283217072486877,
+      "learning_rate": 9.478184745161052e-06,
+      "loss": 0.2263,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6988793830581997,
+      "grad_norm": 0.17181184887886047,
+      "learning_rate": 9.471931828526282e-06,
+      "loss": 0.2288,
+      "step": 2900
+    },
+    {
+      "epoch": 0.701289311965297,
+      "grad_norm": 0.151957705616951,
+      "learning_rate": 9.46564375820192e-06,
+      "loss": 0.2237,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7036992408723942,
+      "grad_norm": 0.21640506386756897,
+      "learning_rate": 9.459320583618253e-06,
+      "loss": 0.226,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7061091697794915,
+      "grad_norm": 0.17349959909915924,
+      "learning_rate": 9.452962354481523e-06,
+      "loss": 0.226,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7085190986865888,
+      "grad_norm": 0.16324348747730255,
+      "learning_rate": 9.44656912077353e-06,
+      "loss": 0.2264,
+      "step": 2940
+    },
+    {
+      "epoch": 0.710929027593686,
+      "grad_norm": 0.17272070050239563,
+      "learning_rate": 9.440140932751249e-06,
+      "loss": 0.2278,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7133389565007833,
+      "grad_norm": 0.17296038568019867,
+      "learning_rate": 9.433677840946424e-06,
+      "loss": 0.2248,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7157488854078805,
+      "grad_norm": 0.22695279121398926,
+      "learning_rate": 9.427179896165182e-06,
+      "loss": 0.2252,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7181588143149777,
+      "grad_norm": 0.1850307136774063,
+      "learning_rate": 9.420647149487622e-06,
+      "loss": 0.2272,
+      "step": 2980
+    },
+    {
+      "epoch": 0.720568743222075,
+      "grad_norm": 0.17052899301052094,
+      "learning_rate": 9.414079652267422e-06,
+      "loss": 0.224,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7229786721291722,
+      "grad_norm": 0.17390076816082,
+      "learning_rate": 9.407477456131438e-06,
+      "loss": 0.2263,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7253886010362695,
+      "grad_norm": 0.19219738245010376,
+      "learning_rate": 9.400840612979283e-06,
+      "loss": 0.2289,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7277985299433667,
+      "grad_norm": 0.1580750048160553,
+      "learning_rate": 9.394169174982935e-06,
+      "loss": 0.2254,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7302084588504639,
+      "grad_norm": 0.17882807552814484,
+      "learning_rate": 9.387463194586321e-06,
+      "loss": 0.2239,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7326183877575612,
+      "grad_norm": 0.19159093499183655,
+      "learning_rate": 9.380722724504902e-06,
+      "loss": 0.2246,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7350283166646584,
+      "grad_norm": 0.17110764980316162,
+      "learning_rate": 9.373947817725262e-06,
+      "loss": 0.2267,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7374382455717556,
+      "grad_norm": 0.17714989185333252,
+      "learning_rate": 9.367138527504694e-06,
+      "loss": 0.2248,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7398481744788529,
+      "grad_norm": 0.19953234493732452,
+      "learning_rate": 9.36029490737077e-06,
+      "loss": 0.2249,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7422581033859501,
+      "grad_norm": 0.24448458850383759,
+      "learning_rate": 9.353417011120937e-06,
+      "loss": 0.2237,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7446680322930473,
+      "grad_norm": 0.20036014914512634,
+      "learning_rate": 9.34650489282208e-06,
+      "loss": 0.227,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7470779612001446,
+      "grad_norm": 0.18248037993907928,
+      "learning_rate": 9.339558606810102e-06,
+      "loss": 0.2218,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7494878901072418,
+      "grad_norm": 0.19500508904457092,
+      "learning_rate": 9.332578207689501e-06,
+      "loss": 0.2235,
+      "step": 3110
+    },
+    {
+      "epoch": 0.751897819014339,
+      "grad_norm": 0.2017701268196106,
+      "learning_rate": 9.325563750332935e-06,
+      "loss": 0.2257,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7543077479214363,
+      "grad_norm": 0.1895110011100769,
+      "learning_rate": 9.31851528988079e-06,
+      "loss": 0.2237,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7567176768285335,
+      "grad_norm": 0.17268790304660797,
+      "learning_rate": 9.311432881740752e-06,
+      "loss": 0.2237,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7591276057356308,
+      "grad_norm": 0.1882033497095108,
+      "learning_rate": 9.304316581587367e-06,
+      "loss": 0.2215,
+      "step": 3150
+    },
+    {
+      "epoch": 0.761537534642728,
+      "grad_norm": 0.1775064915418625,
+      "learning_rate": 9.297166445361608e-06,
+      "loss": 0.2209,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7639474635498252,
+      "grad_norm": 0.21683911979198456,
+      "learning_rate": 9.289982529270424e-06,
+      "loss": 0.2279,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7663573924569225,
+      "grad_norm": 0.19911475479602814,
+      "learning_rate": 9.28276488978632e-06,
+      "loss": 0.2211,
+      "step": 3180
+    },
+    {
+      "epoch": 0.7687673213640197,
+      "grad_norm": 0.18233388662338257,
+      "learning_rate": 9.275513583646885e-06,
+      "loss": 0.2245,
+      "step": 3190
+    },
+    {
+      "epoch": 0.771177250271117,
+      "grad_norm": 0.19964131712913513,
+      "learning_rate": 9.26822866785437e-06,
+      "loss": 0.2255,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7735871791782143,
+      "grad_norm": 0.1638319343328476,
+      "learning_rate": 9.260910199675224e-06,
+      "loss": 0.2225,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7759971080853115,
+      "grad_norm": 0.18763862550258636,
+      "learning_rate": 9.253558236639654e-06,
+      "loss": 0.2265,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7784070369924088,
+      "grad_norm": 0.20708943903446198,
+      "learning_rate": 9.246172836541167e-06,
+      "loss": 0.2246,
+      "step": 3230
+    },
+    {
+      "epoch": 0.780816965899506,
+      "grad_norm": 0.16020740568637848,
+      "learning_rate": 9.238754057436121e-06,
+      "loss": 0.2274,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7832268948066032,
+      "grad_norm": 0.19551053643226624,
+      "learning_rate": 9.23130195764326e-06,
+      "loss": 0.223,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7856368237137005,
+      "grad_norm": 0.18796652555465698,
+      "learning_rate": 9.223816595743263e-06,
+      "loss": 0.2206,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7880467526207977,
+      "grad_norm": 0.18281373381614685,
+      "learning_rate": 9.21629803057828e-06,
+      "loss": 0.2282,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7904566815278949,
+      "grad_norm": 0.19269907474517822,
+      "learning_rate": 9.208746321251477e-06,
+      "loss": 0.2258,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7928666104349922,
+      "grad_norm": 0.18990552425384521,
+      "learning_rate": 9.201161527126554e-06,
+      "loss": 0.2207,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7952765393420894,
+      "grad_norm": 0.2274039387702942,
+      "learning_rate": 9.193543707827297e-06,
+      "loss": 0.2239,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7976864682491867,
+      "grad_norm": 0.23829886317253113,
+      "learning_rate": 9.185892923237101e-06,
+      "loss": 0.2208,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8000963971562839,
+      "grad_norm": 0.18112841248512268,
+      "learning_rate": 9.178209233498497e-06,
+      "loss": 0.2232,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8025063260633811,
+      "grad_norm": 0.1845228374004364,
+      "learning_rate": 9.170492699012686e-06,
+      "loss": 0.2258,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8049162549704784,
+      "grad_norm": 0.16812431812286377,
+      "learning_rate": 9.162743380439057e-06,
+      "loss": 0.2256,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8073261838775756,
+      "grad_norm": 0.19715647399425507,
+      "learning_rate": 9.154961338694714e-06,
+      "loss": 0.2188,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8097361127846728,
+      "grad_norm": 0.16696830093860626,
+      "learning_rate": 9.147146634954e-06,
+      "loss": 0.224,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8121460416917701,
+      "grad_norm": 0.20943832397460938,
+      "learning_rate": 9.139299330648006e-06,
+      "loss": 0.2243,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8145559705988673,
+      "grad_norm": 0.18432581424713135,
+      "learning_rate": 9.131419487464104e-06,
+      "loss": 0.2224,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8169658995059645,
+      "grad_norm": 0.1849818378686905,
+      "learning_rate": 9.123507167345444e-06,
+      "loss": 0.2257,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8193758284130618,
+      "grad_norm": 0.18599388003349304,
+      "learning_rate": 9.115562432490482e-06,
+      "loss": 0.2252,
+      "step": 3400
+    },
+    {
+      "epoch": 0.821785757320159,
+      "grad_norm": 0.17823128402233124,
+      "learning_rate": 9.107585345352481e-06,
+      "loss": 0.2207,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8241956862272563,
+      "grad_norm": 0.1734575480222702,
+      "learning_rate": 9.099575968639028e-06,
+      "loss": 0.2235,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8266056151343535,
+      "grad_norm": 0.1771286129951477,
+      "learning_rate": 9.091534365311531e-06,
+      "loss": 0.2234,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8290155440414507,
+      "grad_norm": 0.17510093748569489,
+      "learning_rate": 9.08346059858474e-06,
+      "loss": 0.2208,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8314254729485481,
+      "grad_norm": 0.16973088681697845,
+      "learning_rate": 9.075354731926232e-06,
+      "loss": 0.2217,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8338354018556453,
+      "grad_norm": 0.1625513732433319,
+      "learning_rate": 9.067216829055922e-06,
+      "loss": 0.224,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8362453307627425,
+      "grad_norm": 0.17390358448028564,
+      "learning_rate": 9.059046953945563e-06,
+      "loss": 0.2226,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8386552596698398,
+      "grad_norm": 0.1795208603143692,
+      "learning_rate": 9.050845170818239e-06,
+      "loss": 0.2219,
+      "step": 3480
+    },
+    {
+      "epoch": 0.841065188576937,
+      "grad_norm": 0.1884290724992752,
+      "learning_rate": 9.04261154414786e-06,
+      "loss": 0.2231,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8434751174840343,
+      "grad_norm": 0.16878260672092438,
+      "learning_rate": 9.03434613865866e-06,
+      "loss": 0.2233,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8458850463911315,
+      "grad_norm": 0.17098242044448853,
+      "learning_rate": 9.026049019324686e-06,
+      "loss": 0.2263,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8482949752982287,
+      "grad_norm": 0.18065747618675232,
+      "learning_rate": 9.01772025136928e-06,
+      "loss": 0.218,
+      "step": 3520
+    },
+    {
+      "epoch": 0.850704904205326,
+      "grad_norm": 0.15557970106601715,
+      "learning_rate": 9.009359900264579e-06,
+      "loss": 0.2226,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8531148331124232,
+      "grad_norm": 0.18261565268039703,
+      "learning_rate": 9.00096803173099e-06,
+      "loss": 0.2177,
+      "step": 3540
+    },
+    {
+      "epoch": 0.8555247620195204,
+      "grad_norm": 0.1887233406305313,
+      "learning_rate": 8.992544711736682e-06,
+      "loss": 0.2202,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8579346909266177,
+      "grad_norm": 0.18528446555137634,
+      "learning_rate": 8.984090006497056e-06,
+      "loss": 0.2206,
+      "step": 3560
+    },
+    {
+      "epoch": 0.8603446198337149,
+      "grad_norm": 0.24964436888694763,
+      "learning_rate": 8.97560398247424e-06,
+      "loss": 0.2195,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8627545487408121,
+      "grad_norm": 0.18638278543949127,
+      "learning_rate": 8.96708670637655e-06,
+      "loss": 0.2197,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8651644776479094,
+      "grad_norm": 0.16028571128845215,
+      "learning_rate": 8.958538245157975e-06,
+      "loss": 0.2231,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8675744065550066,
+      "grad_norm": 0.1577351987361908,
+      "learning_rate": 8.949958666017652e-06,
+      "loss": 0.2221,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8699843354621039,
+      "grad_norm": 0.26832443475723267,
+      "learning_rate": 8.941348036399333e-06,
+      "loss": 0.2224,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8723942643692011,
+      "grad_norm": 0.19743038713932037,
+      "learning_rate": 8.932706423990856e-06,
+      "loss": 0.224,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8748041932762983,
+      "grad_norm": 0.17211681604385376,
+      "learning_rate": 8.924033896723617e-06,
+      "loss": 0.2243,
+      "step": 3630
+    },
+    {
+      "epoch": 0.8772141221833956,
+      "grad_norm": 0.18818245828151703,
+      "learning_rate": 8.915330522772028e-06,
+      "loss": 0.2229,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8796240510904928,
+      "grad_norm": 0.18545369803905487,
+      "learning_rate": 8.906596370552985e-06,
+      "loss": 0.2204,
+      "step": 3650
+    },
+    {
+      "epoch": 0.88203397999759,
+      "grad_norm": 0.18533460795879364,
+      "learning_rate": 8.897831508725338e-06,
+      "loss": 0.2204,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8844439089046873,
+      "grad_norm": 0.1828158050775528,
+      "learning_rate": 8.889036006189338e-06,
+      "loss": 0.221,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8868538378117845,
+      "grad_norm": 0.19318468868732452,
+      "learning_rate": 8.880209932086102e-06,
+      "loss": 0.2238,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8892637667188817,
+      "grad_norm": 0.16859950125217438,
+      "learning_rate": 8.87135335579707e-06,
+      "loss": 0.2179,
+      "step": 3690
+    },
+    {
+      "epoch": 0.891673695625979,
+      "grad_norm": 0.16465112566947937,
+      "learning_rate": 8.862466346943457e-06,
+      "loss": 0.2243,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8940836245330763,
+      "grad_norm": 0.19712437689304352,
+      "learning_rate": 8.853548975385714e-06,
+      "loss": 0.2211,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8964935534401736,
+      "grad_norm": 0.1706128865480423,
+      "learning_rate": 8.84460131122296e-06,
+      "loss": 0.2209,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8989034823472708,
+      "grad_norm": 0.20529857277870178,
+      "learning_rate": 8.835623424792453e-06,
+      "loss": 0.2242,
+      "step": 3730
+    },
+    {
+      "epoch": 0.901313411254368,
+      "grad_norm": 0.16499772667884827,
+      "learning_rate": 8.826615386669025e-06,
+      "loss": 0.2209,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9037233401614653,
+      "grad_norm": 0.16960477828979492,
+      "learning_rate": 8.817577267664528e-06,
+      "loss": 0.222,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9061332690685625,
+      "grad_norm": 0.1796937882900238,
+      "learning_rate": 8.808509138827275e-06,
+      "loss": 0.2209,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9085431979756597,
+      "grad_norm": 0.16342346370220184,
+      "learning_rate": 8.799411071441496e-06,
+      "loss": 0.2204,
+      "step": 3770
+    },
+    {
+      "epoch": 0.910953126882757,
+      "grad_norm": 0.1790073662996292,
+      "learning_rate": 8.790283137026754e-06,
+      "loss": 0.2203,
+      "step": 3780
+    },
+    {
+      "epoch": 0.9133630557898542,
+      "grad_norm": 0.17574553191661835,
+      "learning_rate": 8.781125407337405e-06,
+      "loss": 0.2221,
+      "step": 3790
+    },
+    {
+      "epoch": 0.9157729846969515,
+      "grad_norm": 0.15836398303508759,
+      "learning_rate": 8.77193795436202e-06,
+      "loss": 0.219,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9181829136040487,
+      "grad_norm": 0.17478342354297638,
+      "learning_rate": 8.762720850322823e-06,
+      "loss": 0.2203,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9205928425111459,
+      "grad_norm": 0.15719293057918549,
+      "learning_rate": 8.753474167675128e-06,
+      "loss": 0.2227,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9230027714182432,
+      "grad_norm": 0.1693807989358902,
+      "learning_rate": 8.744197979106763e-06,
+      "loss": 0.2236,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9254127003253404,
+      "grad_norm": 0.21582992374897003,
+      "learning_rate": 8.7348923575375e-06,
+      "loss": 0.2221,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9278226292324376,
+      "grad_norm": 0.21359696984291077,
+      "learning_rate": 8.725557376118482e-06,
+      "loss": 0.2222,
+      "step": 3850
+    },
+    {
+      "epoch": 0.9302325581395349,
+      "grad_norm": 0.17620986700057983,
+      "learning_rate": 8.716193108231655e-06,
+      "loss": 0.2209,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9326424870466321,
+      "grad_norm": 0.16482968628406525,
+      "learning_rate": 8.706799627489175e-06,
+      "loss": 0.2212,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9350524159537293,
+      "grad_norm": 0.16502827405929565,
+      "learning_rate": 8.697377007732848e-06,
+      "loss": 0.2204,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9374623448608266,
+      "grad_norm": 0.21359188854694366,
+      "learning_rate": 8.687925323033536e-06,
+      "loss": 0.2206,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9398722737679238,
+      "grad_norm": 0.19615745544433594,
+      "learning_rate": 8.67844464769058e-06,
+      "loss": 0.2201,
+      "step": 3900
+    },
+    {
+      "epoch": 0.942282202675021,
+      "grad_norm": 0.17519152164459229,
+      "learning_rate": 8.668935056231216e-06,
+      "loss": 0.2182,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9446921315821183,
+      "grad_norm": 0.1906772404909134,
+      "learning_rate": 8.659396623409987e-06,
+      "loss": 0.2239,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9471020604892155,
+      "grad_norm": 0.19051863253116608,
+      "learning_rate": 8.649829424208163e-06,
+      "loss": 0.2224,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9495119893963128,
+      "grad_norm": 0.22634129226207733,
+      "learning_rate": 8.640233533833136e-06,
+      "loss": 0.2187,
+      "step": 3940
+    },
+    {
+      "epoch": 0.95192191830341,
+      "grad_norm": 0.1962852030992508,
+      "learning_rate": 8.630609027717843e-06,
+      "loss": 0.2196,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9543318472105073,
+      "grad_norm": 0.185867577791214,
+      "learning_rate": 8.620955981520171e-06,
+      "loss": 0.2194,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9567417761176046,
+      "grad_norm": 0.17657962441444397,
+      "learning_rate": 8.611274471122355e-06,
+      "loss": 0.2229,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9591517050247018,
+      "grad_norm": 0.15567971765995026,
+      "learning_rate": 8.601564572630387e-06,
+      "loss": 0.2193,
+      "step": 3980
+    },
+    {
+      "epoch": 0.961561633931799,
+      "grad_norm": 0.16838258504867554,
+      "learning_rate": 8.591826362373421e-06,
+      "loss": 0.2236,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9639715628388963,
+      "grad_norm": 0.18067292869091034,
+      "learning_rate": 8.58205991690316e-06,
+      "loss": 0.2227,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9663814917459935,
+      "grad_norm": 0.16675949096679688,
+      "learning_rate": 8.572265312993274e-06,
+      "loss": 0.2197,
+      "step": 4010
+    },
+    {
+      "epoch": 0.9687914206530908,
+      "grad_norm": 0.17282336950302124,
+      "learning_rate": 8.562442627638774e-06,
+      "loss": 0.2213,
+      "step": 4020
+    },
+    {
+      "epoch": 0.971201349560188,
+      "grad_norm": 0.192217618227005,
+      "learning_rate": 8.552591938055425e-06,
+      "loss": 0.2184,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9736112784672852,
+      "grad_norm": 0.1581621617078781,
+      "learning_rate": 8.542713321679137e-06,
+      "loss": 0.2221,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9760212073743825,
+      "grad_norm": 0.1765250712633133,
+      "learning_rate": 8.532806856165337e-06,
+      "loss": 0.2188,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9784311362814797,
+      "grad_norm": 0.17979812622070312,
+      "learning_rate": 8.522872619388387e-06,
+      "loss": 0.222,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9808410651885769,
+      "grad_norm": 0.1699352264404297,
+      "learning_rate": 8.512910689440951e-06,
+      "loss": 0.2201,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9832509940956742,
+      "grad_norm": 0.17940454185009003,
+      "learning_rate": 8.50292114463339e-06,
+      "loss": 0.2163,
+      "step": 4080
+    },
+    {
+      "epoch": 0.9856609230027714,
+      "grad_norm": 0.16519592702388763,
+      "learning_rate": 8.49290406349314e-06,
+      "loss": 0.2203,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9880708519098687,
+      "grad_norm": 0.16589729487895966,
+      "learning_rate": 8.482859524764108e-06,
+      "loss": 0.2224,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9904807808169659,
+      "grad_norm": 0.15325430035591125,
+      "learning_rate": 8.472787607406036e-06,
+      "loss": 0.2187,
+      "step": 4110
+    },
+    {
+      "epoch": 0.9928907097240631,
+      "grad_norm": 0.16159406304359436,
+      "learning_rate": 8.462688390593894e-06,
+      "loss": 0.2247,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9953006386311604,
+      "grad_norm": 0.1637933999300003,
+      "learning_rate": 8.452561953717246e-06,
+      "loss": 0.2178,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9977105675382576,
+      "grad_norm": 0.16510990262031555,
+      "learning_rate": 8.442408376379637e-06,
+      "loss": 0.2219,
+      "step": 4140
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.19598262012004852,
+      "learning_rate": 8.43222773839796e-06,
+      "loss": 0.2169,
+      "step": 4150
+    },
+    {
+      "epoch": 1.0024099289070973,
+      "grad_norm": 0.22410273551940918,
+      "learning_rate": 8.422020119801831e-06,
+      "loss": 0.2121,
+      "step": 4160
+    },
+    {
+      "epoch": 1.0048198578141945,
+      "grad_norm": 0.1659049689769745,
+      "learning_rate": 8.411785600832959e-06,
+      "loss": 0.2159,
+      "step": 4170
+    },
+    {
+      "epoch": 1.0072297867212918,
+      "grad_norm": 0.1600600630044937,
+      "learning_rate": 8.401524261944519e-06,
+      "loss": 0.2142,
+      "step": 4180
+    },
+    {
+      "epoch": 1.009639715628389,
+      "grad_norm": 0.17050403356552124,
+      "learning_rate": 8.39123618380051e-06,
+      "loss": 0.2133,
+      "step": 4190
+    },
+    {
+      "epoch": 1.0120496445354863,
+      "grad_norm": 0.1873362809419632,
+      "learning_rate": 8.380921447275137e-06,
+      "loss": 0.2161,
+      "step": 4200
+    },
+    {
+      "epoch": 1.0144595734425834,
+      "grad_norm": 0.17461583018302917,
+      "learning_rate": 8.370580133452153e-06,
+      "loss": 0.2161,
+      "step": 4210
+    },
+    {
+      "epoch": 1.0168695023496808,
+      "grad_norm": 0.17326101660728455,
+      "learning_rate": 8.360212323624246e-06,
+      "loss": 0.2137,
+      "step": 4220
+    },
+    {
+      "epoch": 1.019279431256778,
+      "grad_norm": 0.16447477042675018,
+      "learning_rate": 8.349818099292379e-06,
+      "loss": 0.2164,
+      "step": 4230
+    },
+    {
+      "epoch": 1.0216893601638752,
+      "grad_norm": 0.16297782957553864,
+      "learning_rate": 8.339397542165166e-06,
+      "loss": 0.213,
+      "step": 4240
+    },
+    {
+      "epoch": 1.0240992890709724,
+      "grad_norm": 0.1830231249332428,
+      "learning_rate": 8.328950734158219e-06,
+      "loss": 0.214,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0265092179780697,
+      "grad_norm": 0.17782536149024963,
+      "learning_rate": 8.318477757393502e-06,
+      "loss": 0.2119,
+      "step": 4260
+    },
+    {
+      "epoch": 1.0289191468851668,
+      "grad_norm": 0.16562552750110626,
+      "learning_rate": 8.3079786941987e-06,
+      "loss": 0.2129,
+      "step": 4270
+    },
+    {
+      "epoch": 1.0313290757922642,
+      "grad_norm": 0.1697404384613037,
+      "learning_rate": 8.297453627106556e-06,
+      "loss": 0.2111,
+      "step": 4280
+    },
+    {
+      "epoch": 1.0337390046993613,
+      "grad_norm": 0.17114974558353424,
+      "learning_rate": 8.28690263885423e-06,
+      "loss": 0.2123,
+      "step": 4290
+    },
+    {
+      "epoch": 1.0361489336064587,
+      "grad_norm": 0.16717548668384552,
+      "learning_rate": 8.276325812382648e-06,
+      "loss": 0.2128,
+      "step": 4300
+    },
+    {
+      "epoch": 1.0385588625135558,
+      "grad_norm": 0.16737718880176544,
+      "learning_rate": 8.265723230835852e-06,
+      "loss": 0.2107,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0409687914206531,
+      "grad_norm": 0.15507322549819946,
+      "learning_rate": 8.255094977560335e-06,
+      "loss": 0.2127,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0433787203277503,
+      "grad_norm": 0.1764843910932541,
+      "learning_rate": 8.244441136104406e-06,
+      "loss": 0.212,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0457886492348476,
+      "grad_norm": 0.17232689261436462,
+      "learning_rate": 8.233761790217512e-06,
+      "loss": 0.2136,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0481985781419447,
+      "grad_norm": 0.1621418595314026,
+      "learning_rate": 8.223057023849595e-06,
+      "loss": 0.2128,
+      "step": 4350
+    },
+    {
+      "epoch": 1.050608507049042,
+      "grad_norm": 0.16862477362155914,
+      "learning_rate": 8.212326921150426e-06,
+      "loss": 0.2155,
+      "step": 4360
+    },
+    {
+      "epoch": 1.0530184359561392,
+      "grad_norm": 0.16127009689807892,
+      "learning_rate": 8.20157156646894e-06,
+      "loss": 0.2158,
+      "step": 4370
+    },
+    {
+      "epoch": 1.0554283648632365,
+      "grad_norm": 0.16644121706485748,
+      "learning_rate": 8.190791044352581e-06,
+      "loss": 0.2111,
+      "step": 4380
+    },
+    {
+      "epoch": 1.0578382937703337,
+      "grad_norm": 0.17563723027706146,
+      "learning_rate": 8.179985439546633e-06,
+      "loss": 0.2114,
+      "step": 4390
+    },
+    {
+      "epoch": 1.060248222677431,
+      "grad_norm": 0.1750829517841339,
+      "learning_rate": 8.16915483699355e-06,
+      "loss": 0.2117,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0626581515845284,
+      "grad_norm": 0.17254221439361572,
+      "learning_rate": 8.158299321832301e-06,
+      "loss": 0.2118,
+      "step": 4410
+    },
+    {
+      "epoch": 1.0650680804916255,
+      "grad_norm": 0.17799583077430725,
+      "learning_rate": 8.147418979397682e-06,
+      "loss": 0.2187,
+      "step": 4420
+    },
+    {
+      "epoch": 1.0674780093987228,
+      "grad_norm": 0.17555776238441467,
+      "learning_rate": 8.13651389521966e-06,
+      "loss": 0.2155,
+      "step": 4430
+    },
+    {
+      "epoch": 1.06988793830582,
+      "grad_norm": 0.1640000343322754,
+      "learning_rate": 8.125584155022696e-06,
+      "loss": 0.2134,
+      "step": 4440
+    },
+    {
+      "epoch": 1.0722978672129173,
+      "grad_norm": 0.17578564584255219,
+      "learning_rate": 8.114629844725073e-06,
+      "loss": 0.2119,
+      "step": 4450
+    },
+    {
+      "epoch": 1.0747077961200144,
+      "grad_norm": 0.1732892245054245,
+      "learning_rate": 8.103651050438213e-06,
+      "loss": 0.2124,
+      "step": 4460
+    },
+    {
+      "epoch": 1.0771177250271118,
+      "grad_norm": 0.15404583513736725,
+      "learning_rate": 8.09264785846601e-06,
+      "loss": 0.2149,
+      "step": 4470
+    },
+    {
+      "epoch": 1.079527653934209,
+      "grad_norm": 0.18235298991203308,
+      "learning_rate": 8.081620355304147e-06,
+      "loss": 0.2106,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0819375828413063,
+      "grad_norm": 0.1770295798778534,
+      "learning_rate": 8.070568627639418e-06,
+      "loss": 0.2141,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0843475117484034,
+      "grad_norm": 0.1517912745475769,
+      "learning_rate": 8.059492762349037e-06,
+      "loss": 0.2125,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0867574406555007,
+      "grad_norm": 0.17619790136814117,
+      "learning_rate": 8.048392846499974e-06,
+      "loss": 0.2135,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0891673695625979,
+      "grad_norm": 0.1587211936712265,
+      "learning_rate": 8.037268967348252e-06,
+      "loss": 0.2152,
+      "step": 4520
+    },
+    {
+      "epoch": 1.0915772984696952,
+      "grad_norm": 0.22725209593772888,
+      "learning_rate": 8.026121212338271e-06,
+      "loss": 0.2122,
+      "step": 4530
+    },
+    {
+      "epoch": 1.0939872273767923,
+      "grad_norm": 0.18444053828716278,
+      "learning_rate": 8.014949669102117e-06,
+      "loss": 0.2099,
+      "step": 4540
+    },
+    {
+      "epoch": 1.0963971562838897,
+      "grad_norm": 0.1482313573360443,
+      "learning_rate": 8.003754425458878e-06,
+      "loss": 0.2138,
+      "step": 4550
+    },
+    {
+      "epoch": 1.0988070851909868,
+      "grad_norm": 0.16575364768505096,
+      "learning_rate": 7.992535569413944e-06,
+      "loss": 0.2094,
+      "step": 4560
+    },
+    {
+      "epoch": 1.1012170140980841,
+      "grad_norm": 0.19779005646705627,
+      "learning_rate": 7.981293189158327e-06,
+      "loss": 0.2097,
+      "step": 4570
+    },
+    {
+      "epoch": 1.1036269430051813,
+      "grad_norm": 0.16325879096984863,
+      "learning_rate": 7.970027373067961e-06,
+      "loss": 0.2128,
+      "step": 4580
+    },
+    {
+      "epoch": 1.1060368719122786,
+      "grad_norm": 0.18840157985687256,
+      "learning_rate": 7.958738209703004e-06,
+      "loss": 0.2114,
+      "step": 4590
+    },
+    {
+      "epoch": 1.1084468008193757,
+      "grad_norm": 0.1780056655406952,
+      "learning_rate": 7.94742578780715e-06,
+      "loss": 0.211,
+      "step": 4600
+    },
+    {
+      "epoch": 1.110856729726473,
+      "grad_norm": 0.15295591950416565,
+      "learning_rate": 7.936090196306925e-06,
+      "loss": 0.2125,
+      "step": 4610
+    },
+    {
+      "epoch": 1.1132666586335702,
+      "grad_norm": 0.17244797945022583,
+      "learning_rate": 7.924731524310993e-06,
+      "loss": 0.2121,
+      "step": 4620
+    },
+    {
+      "epoch": 1.1156765875406676,
+      "grad_norm": 0.16484029591083527,
+      "learning_rate": 7.91334986110945e-06,
+      "loss": 0.2161,
+      "step": 4630
+    },
+    {
+      "epoch": 1.1180865164477647,
+      "grad_norm": 0.16274769604206085,
+      "learning_rate": 7.90194529617313e-06,
+      "loss": 0.2127,
+      "step": 4640
+    },
+    {
+      "epoch": 1.120496445354862,
+      "grad_norm": 0.16935664415359497,
+      "learning_rate": 7.890517919152892e-06,
+      "loss": 0.2137,
+      "step": 4650
+    },
+    {
+      "epoch": 1.1229063742619592,
+      "grad_norm": 0.15090295672416687,
+      "learning_rate": 7.879067819878918e-06,
+      "loss": 0.213,
+      "step": 4660
+    },
+    {
+      "epoch": 1.1253163031690565,
+      "grad_norm": 0.16716746985912323,
+      "learning_rate": 7.867595088360016e-06,
+      "loss": 0.2115,
+      "step": 4670
+    },
+    {
+      "epoch": 1.1277262320761539,
+      "grad_norm": 0.20338410139083862,
+      "learning_rate": 7.856099814782901e-06,
+      "loss": 0.2133,
+      "step": 4680
+    },
+    {
+      "epoch": 1.130136160983251,
+      "grad_norm": 0.1768144816160202,
+      "learning_rate": 7.844582089511486e-06,
+      "loss": 0.2112,
+      "step": 4690
+    },
+    {
+      "epoch": 1.1325460898903483,
+      "grad_norm": 0.16909371316432953,
+      "learning_rate": 7.833042003086186e-06,
+      "loss": 0.2113,
+      "step": 4700
+    },
+    {
+      "epoch": 1.1349560187974455,
+      "grad_norm": 0.1611366719007492,
+      "learning_rate": 7.82147964622319e-06,
+      "loss": 0.2126,
+      "step": 4710
+    },
+    {
+      "epoch": 1.1373659477045428,
+      "grad_norm": 0.19154107570648193,
+      "learning_rate": 7.809895109813752e-06,
+      "loss": 0.2109,
+      "step": 4720
+    },
+    {
+      "epoch": 1.13977587661164,
+      "grad_norm": 0.14273270964622498,
+      "learning_rate": 7.798288484923482e-06,
+      "loss": 0.2117,
+      "step": 4730
+    },
+    {
+      "epoch": 1.1421858055187373,
+      "grad_norm": 0.19387304782867432,
+      "learning_rate": 7.786659862791628e-06,
+      "loss": 0.212,
+      "step": 4740
+    },
+    {
+      "epoch": 1.1445957344258344,
+      "grad_norm": 0.16093966364860535,
+      "learning_rate": 7.775009334830354e-06,
+      "loss": 0.2148,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1470056633329317,
+      "grad_norm": 0.16083253920078278,
+      "learning_rate": 7.763336992624027e-06,
+      "loss": 0.2132,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1494155922400289,
+      "grad_norm": 0.18043479323387146,
+      "learning_rate": 7.751642927928495e-06,
+      "loss": 0.2125,
+      "step": 4770
+    },
+    {
+      "epoch": 1.1518255211471262,
+      "grad_norm": 0.168365940451622,
+      "learning_rate": 7.739927232670363e-06,
+      "loss": 0.2138,
+      "step": 4780
+    },
+    {
+      "epoch": 1.1542354500542233,
+      "grad_norm": 0.1817869246006012,
+      "learning_rate": 7.728189998946278e-06,
+      "loss": 0.2123,
+      "step": 4790
+    },
+    {
+      "epoch": 1.1566453789613207,
+      "grad_norm": 0.18804657459259033,
+      "learning_rate": 7.716431319022197e-06,
+      "loss": 0.2142,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1590553078684178,
+      "grad_norm": 0.18389475345611572,
+      "learning_rate": 7.704651285332662e-06,
+      "loss": 0.2123,
+      "step": 4810
+    },
+    {
+      "epoch": 1.1614652367755152,
+      "grad_norm": 0.20771625638008118,
+      "learning_rate": 7.692849990480082e-06,
+      "loss": 0.214,
+      "step": 4820
+    },
+    {
+      "epoch": 1.1638751656826123,
+      "grad_norm": 0.18531720340251923,
+      "learning_rate": 7.681027527233995e-06,
+      "loss": 0.2105,
+      "step": 4830
+    },
+    {
+      "epoch": 1.1662850945897096,
+      "grad_norm": 0.16333264112472534,
+      "learning_rate": 7.669183988530346e-06,
+      "loss": 0.214,
+      "step": 4840
+    },
+    {
+      "epoch": 1.1686950234968068,
+      "grad_norm": 0.17973899841308594,
+      "learning_rate": 7.65731946747075e-06,
+      "loss": 0.2142,
+      "step": 4850
+    },
+    {
+      "epoch": 1.171104952403904,
+      "grad_norm": 0.16610458493232727,
+      "learning_rate": 7.645434057321765e-06,
+      "loss": 0.211,
+      "step": 4860
+    },
+    {
+      "epoch": 1.1735148813110012,
+      "grad_norm": 0.14925527572631836,
+      "learning_rate": 7.633527851514163e-06,
+      "loss": 0.2113,
+      "step": 4870
+    },
+    {
+      "epoch": 1.1759248102180986,
+      "grad_norm": 0.16800275444984436,
+      "learning_rate": 7.621600943642175e-06,
+      "loss": 0.2128,
+      "step": 4880
+    },
+    {
+      "epoch": 1.1783347391251957,
+      "grad_norm": 0.1554495096206665,
+      "learning_rate": 7.609653427462789e-06,
+      "loss": 0.2144,
+      "step": 4890
+    },
+    {
+      "epoch": 1.180744668032293,
+      "grad_norm": 0.16693717241287231,
+      "learning_rate": 7.5976853968949785e-06,
+      "loss": 0.2125,
+      "step": 4900
+    },
+    {
+      "epoch": 1.1831545969393904,
+      "grad_norm": 0.1670588105916977,
+      "learning_rate": 7.585696946018988e-06,
+      "loss": 0.2127,
+      "step": 4910
+    },
+    {
+      "epoch": 1.1855645258464875,
+      "grad_norm": 0.14736416935920715,
+      "learning_rate": 7.573688169075584e-06,
+      "loss": 0.2088,
+      "step": 4920
+    },
+    {
+      "epoch": 1.1879744547535847,
+      "grad_norm": 0.1936376839876175,
+      "learning_rate": 7.561659160465314e-06,
+      "loss": 0.2092,
+      "step": 4930
+    },
+    {
+      "epoch": 1.190384383660682,
+      "grad_norm": 0.16551412642002106,
+      "learning_rate": 7.549610014747769e-06,
+      "loss": 0.2126,
+      "step": 4940
+    },
+    {
+      "epoch": 1.1927943125677793,
+      "grad_norm": 0.15562906861305237,
+      "learning_rate": 7.537540826640834e-06,
+      "loss": 0.2091,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1952042414748765,
+      "grad_norm": 0.14535550773143768,
+      "learning_rate": 7.525451691019945e-06,
+      "loss": 0.21,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1976141703819738,
+      "grad_norm": 0.157695010304451,
+      "learning_rate": 7.513342702917349e-06,
+      "loss": 0.2127,
+      "step": 4970
+    },
+    {
+      "epoch": 1.200024099289071,
+      "grad_norm": 0.15610983967781067,
+      "learning_rate": 7.5012139575213505e-06,
+      "loss": 0.2129,
+      "step": 4980
+    },
+    {
+      "epoch": 1.2024340281961683,
+      "grad_norm": 0.14861977100372314,
+      "learning_rate": 7.4890655501755634e-06,
+      "loss": 0.2127,
+      "step": 4990
+    },
+    {
+      "epoch": 1.2048439571032654,
+      "grad_norm": 0.1682192087173462,
+      "learning_rate": 7.476897576378169e-06,
+      "loss": 0.2116,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2072538860103628,
+      "grad_norm": 0.1472558081150055,
+      "learning_rate": 7.464710131781154e-06,
+      "loss": 0.2124,
+      "step": 5010
+    },
+    {
+      "epoch": 1.20966381491746,
+      "grad_norm": 0.15624482929706573,
+      "learning_rate": 7.452503312189567e-06,
+      "loss": 0.2084,
+      "step": 5020
+    },
+    {
+      "epoch": 1.2120737438245572,
+      "grad_norm": 0.17334376275539398,
+      "learning_rate": 7.440277213560763e-06,
+      "loss": 0.2103,
+      "step": 5030
+    },
+    {
+      "epoch": 1.2144836727316544,
+      "grad_norm": 0.14388589560985565,
+      "learning_rate": 7.428031932003647e-06,
+      "loss": 0.2107,
+      "step": 5040
+    },
+    {
+      "epoch": 1.2168936016387517,
+      "grad_norm": 0.1735595464706421,
+      "learning_rate": 7.415767563777922e-06,
+      "loss": 0.2118,
+      "step": 5050
+    },
+    {
+      "epoch": 1.2193035305458488,
+      "grad_norm": 0.14402276277542114,
+      "learning_rate": 7.40348420529333e-06,
+      "loss": 0.2115,
+      "step": 5060
+    },
+    {
+      "epoch": 1.2217134594529462,
+      "grad_norm": 0.20903652906417847,
+      "learning_rate": 7.3911819531088926e-06,
+      "loss": 0.2125,
+      "step": 5070
+    },
+    {
+      "epoch": 1.2241233883600433,
+      "grad_norm": 0.16466741263866425,
+      "learning_rate": 7.378860903932159e-06,
+      "loss": 0.2103,
+      "step": 5080
+    },
+    {
+      "epoch": 1.2265333172671407,
+      "grad_norm": 0.15621258318424225,
+      "learning_rate": 7.366521154618438e-06,
+      "loss": 0.2116,
+      "step": 5090
+    },
+    {
+      "epoch": 1.2289432461742378,
+      "grad_norm": 0.14338956773281097,
+      "learning_rate": 7.354162802170037e-06,
+      "loss": 0.2119,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2313531750813351,
+      "grad_norm": 0.14769229292869568,
+      "learning_rate": 7.341785943735507e-06,
+      "loss": 0.212,
+      "step": 5110
+    },
+    {
+      "epoch": 1.2337631039884323,
+      "grad_norm": 0.16111035645008087,
+      "learning_rate": 7.3293906766088694e-06,
+      "loss": 0.2108,
+      "step": 5120
+    },
+    {
+      "epoch": 1.2361730328955296,
+      "grad_norm": 0.16381020843982697,
+      "learning_rate": 7.316977098228858e-06,
+      "loss": 0.2147,
+      "step": 5130
+    },
+    {
+      "epoch": 1.238582961802627,
+      "grad_norm": 0.15808707475662231,
+      "learning_rate": 7.3045453061781504e-06,
+      "loss": 0.2143,
+      "step": 5140
+    },
+    {
+      "epoch": 1.240992890709724,
+      "grad_norm": 0.18159683048725128,
+      "learning_rate": 7.292095398182601e-06,
+      "loss": 0.2136,
+      "step": 5150
+    },
+    {
+      "epoch": 1.2434028196168212,
+      "grad_norm": 0.17652052640914917,
+      "learning_rate": 7.2796274721104745e-06,
+      "loss": 0.2104,
+      "step": 5160
+    },
+    {
+      "epoch": 1.2458127485239185,
+      "grad_norm": 0.21559830009937286,
+      "learning_rate": 7.267141625971672e-06,
+      "loss": 0.2114,
+      "step": 5170
+    },
+    {
+      "epoch": 1.248222677431016,
+      "grad_norm": 0.15372782945632935,
+      "learning_rate": 7.254637957916964e-06,
+      "loss": 0.2087,
+      "step": 5180
+    },
+    {
+      "epoch": 1.250632606338113,
+      "grad_norm": 0.16764594614505768,
+      "learning_rate": 7.2421165662372216e-06,
+      "loss": 0.2112,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2530425352452101,
+      "grad_norm": 0.16227737069129944,
+      "learning_rate": 7.229577549362638e-06,
+      "loss": 0.211,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2554524641523075,
+      "grad_norm": 0.15372371673583984,
+      "learning_rate": 7.217021005861957e-06,
+      "loss": 0.2098,
+      "step": 5210
+    },
+    {
+      "epoch": 1.2578623930594048,
+      "grad_norm": 0.14599712193012238,
+      "learning_rate": 7.204447034441699e-06,
+      "loss": 0.2128,
+      "step": 5220
+    },
+    {
+      "epoch": 1.260272321966502,
+      "grad_norm": 0.1904706507921219,
+      "learning_rate": 7.191855733945388e-06,
+      "loss": 0.2123,
+      "step": 5230
+    },
+    {
+      "epoch": 1.2626822508735993,
+      "grad_norm": 0.16215404868125916,
+      "learning_rate": 7.179247203352766e-06,
+      "loss": 0.211,
+      "step": 5240
+    },
+    {
+      "epoch": 1.2650921797806964,
+      "grad_norm": 0.1625434160232544,
+      "learning_rate": 7.166621541779023e-06,
+      "loss": 0.2127,
+      "step": 5250
+    },
+    {
+      "epoch": 1.2675021086877938,
+      "grad_norm": 0.15905945003032684,
+      "learning_rate": 7.153978848474015e-06,
+      "loss": 0.2093,
+      "step": 5260
+    },
+    {
+      "epoch": 1.269912037594891,
+      "grad_norm": 0.17562928795814514,
+      "learning_rate": 7.141319222821483e-06,
+      "loss": 0.2104,
+      "step": 5270
+    },
+    {
+      "epoch": 1.2723219665019883,
+      "grad_norm": 0.15171091258525848,
+      "learning_rate": 7.128642764338273e-06,
+      "loss": 0.213,
+      "step": 5280
+    },
+    {
+      "epoch": 1.2747318954090854,
+      "grad_norm": 0.17683686316013336,
+      "learning_rate": 7.115949572673552e-06,
+      "loss": 0.2084,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2771418243161827,
+      "grad_norm": 0.16825149953365326,
+      "learning_rate": 7.1032397476080285e-06,
+      "loss": 0.21,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2795517532232799,
+      "grad_norm": 0.16288043558597565,
+      "learning_rate": 7.090513389053164e-06,
+      "loss": 0.2093,
+      "step": 5310
+    },
+    {
+      "epoch": 1.2819616821303772,
+      "grad_norm": 0.1785936951637268,
+      "learning_rate": 7.0777705970503885e-06,
+      "loss": 0.2121,
+      "step": 5320
+    },
+    {
+      "epoch": 1.2843716110374743,
+      "grad_norm": 0.178307443857193,
+      "learning_rate": 7.065011471770316e-06,
+      "loss": 0.2094,
+      "step": 5330
+    },
+    {
+      "epoch": 1.2867815399445717,
+      "grad_norm": 0.176238015294075,
+      "learning_rate": 7.052236113511955e-06,
+      "loss": 0.2069,
+      "step": 5340
+    },
+    {
+      "epoch": 1.2891914688516688,
+      "grad_norm": 0.16577781736850739,
+      "learning_rate": 7.039444622701922e-06,
+      "loss": 0.2118,
+      "step": 5350
+    },
+    {
+      "epoch": 1.2916013977587661,
+      "grad_norm": 0.16853252053260803,
+      "learning_rate": 7.0266370998936475e-06,
+      "loss": 0.2095,
+      "step": 5360
+    },
+    {
+      "epoch": 1.2940113266658635,
+      "grad_norm": 0.17209801077842712,
+      "learning_rate": 7.013813645766593e-06,
+      "loss": 0.2083,
+      "step": 5370
+    },
+    {
+      "epoch": 1.2964212555729606,
+      "grad_norm": 0.15391361713409424,
+      "learning_rate": 7.000974361125454e-06,
+      "loss": 0.2097,
+      "step": 5380
+    },
+    {
+      "epoch": 1.2988311844800577,
+      "grad_norm": 0.15086984634399414,
+      "learning_rate": 6.98811934689937e-06,
+      "loss": 0.2119,
+      "step": 5390
+    },
+    {
+      "epoch": 1.301241113387155,
+      "grad_norm": 0.1588638871908188,
+      "learning_rate": 6.975248704141128e-06,
+      "loss": 0.2097,
+      "step": 5400
+    },
+    {
+      "epoch": 1.3036510422942524,
+      "grad_norm": 0.15231099724769592,
+      "learning_rate": 6.96236253402637e-06,
+      "loss": 0.2126,
+      "step": 5410
+    },
+    {
+      "epoch": 1.3060609712013496,
+      "grad_norm": 0.1609620451927185,
+      "learning_rate": 6.949460937852803e-06,
+      "loss": 0.2101,
+      "step": 5420
+    },
+    {
+      "epoch": 1.3084709001084467,
+      "grad_norm": 0.15570048987865448,
+      "learning_rate": 6.936544017039391e-06,
+      "loss": 0.2142,
+      "step": 5430
+    },
+    {
+      "epoch": 1.310880829015544,
+      "grad_norm": 0.158262237906456,
+      "learning_rate": 6.923611873125568e-06,
+      "loss": 0.2099,
+      "step": 5440
+    },
+    {
+      "epoch": 1.3132907579226414,
+      "grad_norm": 0.17083002626895905,
+      "learning_rate": 6.910664607770436e-06,
+      "loss": 0.2107,
+      "step": 5450
+    },
+    {
+      "epoch": 1.3157006868297385,
+      "grad_norm": 0.14634591341018677,
+      "learning_rate": 6.897702322751968e-06,
+      "loss": 0.211,
+      "step": 5460
+    },
+    {
+      "epoch": 1.3181106157368356,
+      "grad_norm": 0.16394425928592682,
+      "learning_rate": 6.8847251199662025e-06,
+      "loss": 0.2124,
+      "step": 5470
+    },
+    {
+      "epoch": 1.320520544643933,
+      "grad_norm": 0.14441823959350586,
+      "learning_rate": 6.871733101426446e-06,
+      "loss": 0.2063,
+      "step": 5480
+    },
+    {
+      "epoch": 1.3229304735510303,
+      "grad_norm": 0.14914725720882416,
+      "learning_rate": 6.858726369262474e-06,
+      "loss": 0.2086,
+      "step": 5490
+    },
+    {
+      "epoch": 1.3253404024581275,
+      "grad_norm": 0.15426112711429596,
+      "learning_rate": 6.8457050257197225e-06,
+      "loss": 0.2104,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3277503313652248,
+      "grad_norm": 0.15343551337718964,
+      "learning_rate": 6.832669173158488e-06,
+      "loss": 0.2109,
+      "step": 5510
+    },
+    {
+      "epoch": 1.330160260272322,
+      "grad_norm": 0.16048264503479004,
+      "learning_rate": 6.819618914053126e-06,
+      "loss": 0.2082,
+      "step": 5520
+    },
+    {
+      "epoch": 1.3325701891794193,
+      "grad_norm": 0.17848889529705048,
+      "learning_rate": 6.806554350991233e-06,
+      "loss": 0.2125,
+      "step": 5530
+    },
+    {
+      "epoch": 1.3349801180865164,
+      "grad_norm": 0.16506092250347137,
+      "learning_rate": 6.793475586672853e-06,
+      "loss": 0.2116,
+      "step": 5540
+    },
+    {
+      "epoch": 1.3373900469936137,
+      "grad_norm": 0.16847744584083557,
+      "learning_rate": 6.780382723909669e-06,
+      "loss": 0.21,
+      "step": 5550
+    },
+    {
+      "epoch": 1.3397999759007109,
+      "grad_norm": 0.1667633205652237,
+      "learning_rate": 6.767275865624183e-06,
+      "loss": 0.212,
+      "step": 5560
+    },
+    {
+      "epoch": 1.3422099048078082,
+      "grad_norm": 0.1553545743227005,
+      "learning_rate": 6.754155114848924e-06,
+      "loss": 0.2121,
+      "step": 5570
+    },
+    {
+      "epoch": 1.3446198337149053,
+      "grad_norm": 0.19598965346813202,
+      "learning_rate": 6.741020574725622e-06,
+      "loss": 0.2088,
+      "step": 5580
+    },
+    {
+      "epoch": 1.3470297626220027,
+      "grad_norm": 0.16720835864543915,
+      "learning_rate": 6.72787234850441e-06,
+      "loss": 0.2128,
+      "step": 5590
+    },
+    {
+      "epoch": 1.3494396915290998,
+      "grad_norm": 0.181742861866951,
+      "learning_rate": 6.7147105395430045e-06,
+      "loss": 0.2104,
+      "step": 5600
+    },
+    {
+      "epoch": 1.3518496204361972,
+      "grad_norm": 0.17101141810417175,
+      "learning_rate": 6.701535251305895e-06,
+      "loss": 0.2082,
+      "step": 5610
+    },
+    {
+      "epoch": 1.3542595493432943,
+      "grad_norm": 0.15846946835517883,
+      "learning_rate": 6.688346587363533e-06,
+      "loss": 0.2097,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3566694782503916,
+      "grad_norm": 0.16642266511917114,
+      "learning_rate": 6.675144651391511e-06,
+      "loss": 0.2115,
+      "step": 5630
+    },
+    {
+      "epoch": 1.359079407157489,
+      "grad_norm": 0.16766521334648132,
+      "learning_rate": 6.661929547169761e-06,
+      "loss": 0.2076,
+      "step": 5640
+    },
+    {
+      "epoch": 1.361489336064586,
+      "grad_norm": 0.16784071922302246,
+      "learning_rate": 6.648701378581722e-06,
+      "loss": 0.2117,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3638992649716832,
+      "grad_norm": 0.14052754640579224,
+      "learning_rate": 6.635460249613537e-06,
+      "loss": 0.2068,
+      "step": 5660
+    },
+    {
+      "epoch": 1.3663091938787806,
+      "grad_norm": 0.16746129095554352,
+      "learning_rate": 6.622206264353225e-06,
+      "loss": 0.2069,
+      "step": 5670
+    },
+    {
+      "epoch": 1.368719122785878,
+      "grad_norm": 0.1509017050266266,
+      "learning_rate": 6.6089395269898715e-06,
+      "loss": 0.2066,
+      "step": 5680
+    },
+    {
+      "epoch": 1.371129051692975,
+      "grad_norm": 0.15285679697990417,
+      "learning_rate": 6.595660141812806e-06,
+      "loss": 0.2102,
+      "step": 5690
+    },
+    {
+      "epoch": 1.3735389806000722,
+      "grad_norm": 0.15320274233818054,
+      "learning_rate": 6.582368213210781e-06,
+      "loss": 0.2088,
+      "step": 5700
+    },
+    {
+      "epoch": 1.3759489095071695,
+      "grad_norm": 0.16169145703315735,
+      "learning_rate": 6.569063845671153e-06,
+      "loss": 0.2102,
+      "step": 5710
+    },
+    {
+      "epoch": 1.3783588384142669,
+      "grad_norm": 0.16042403876781464,
+      "learning_rate": 6.555747143779058e-06,
+      "loss": 0.2108,
+      "step": 5720
+    },
+    {
+      "epoch": 1.380768767321364,
+      "grad_norm": 0.1627410501241684,
+      "learning_rate": 6.542418212216592e-06,
+      "loss": 0.2097,
+      "step": 5730
+    },
+    {
+      "epoch": 1.3831786962284611,
+      "grad_norm": 0.16954395174980164,
+      "learning_rate": 6.5290771557619935e-06,
+      "loss": 0.2102,
+      "step": 5740
+    },
+    {
+      "epoch": 1.3855886251355585,
+      "grad_norm": 0.16062310338020325,
+      "learning_rate": 6.51572407928881e-06,
+      "loss": 0.2099,
+      "step": 5750
+    },
+    {
+      "epoch": 1.3879985540426558,
+      "grad_norm": 0.1502860188484192,
+      "learning_rate": 6.502359087765077e-06,
+      "loss": 0.2131,
+      "step": 5760
+    },
+    {
+      "epoch": 1.390408482949753,
+      "grad_norm": 0.15315794944763184,
+      "learning_rate": 6.488982286252495e-06,
+      "loss": 0.2114,
+      "step": 5770
+    },
+    {
+      "epoch": 1.3928184118568503,
+      "grad_norm": 0.14878347516059875,
+      "learning_rate": 6.475593779905604e-06,
+      "loss": 0.2071,
+      "step": 5780
+    },
+    {
+      "epoch": 1.3952283407639474,
+      "grad_norm": 0.15133234858512878,
+      "learning_rate": 6.462193673970954e-06,
+      "loss": 0.2095,
+      "step": 5790
+    },
+    {
+      "epoch": 1.3976382696710448,
+      "grad_norm": 0.15413503348827362,
+      "learning_rate": 6.448782073786276e-06,
+      "loss": 0.2091,
+      "step": 5800
+    },
+    {
+      "epoch": 1.400048198578142,
+      "grad_norm": 0.14477072656154633,
+      "learning_rate": 6.435359084779663e-06,
+      "loss": 0.2085,
+      "step": 5810
+    },
+    {
+      "epoch": 1.4024581274852392,
+      "grad_norm": 0.14902052283287048,
+      "learning_rate": 6.4219248124687295e-06,
+      "loss": 0.2137,
+      "step": 5820
+    },
+    {
+      "epoch": 1.4048680563923364,
+      "grad_norm": 0.1455923467874527,
+      "learning_rate": 6.408479362459791e-06,
+      "loss": 0.2069,
+      "step": 5830
+    },
+    {
+      "epoch": 1.4072779852994337,
+      "grad_norm": 0.1630076915025711,
+      "learning_rate": 6.39502284044703e-06,
+      "loss": 0.209,
+      "step": 5840
+    },
+    {
+      "epoch": 1.4096879142065308,
+      "grad_norm": 0.16327600181102753,
+      "learning_rate": 6.381555352211663e-06,
+      "loss": 0.2078,
+      "step": 5850
+    },
+    {
+      "epoch": 1.4120978431136282,
+      "grad_norm": 0.16105572879314423,
+      "learning_rate": 6.368077003621116e-06,
+      "loss": 0.2061,
+      "step": 5860
+    },
+    {
+      "epoch": 1.4145077720207253,
+      "grad_norm": 0.1474512666463852,
+      "learning_rate": 6.354587900628184e-06,
+      "loss": 0.2116,
+      "step": 5870
+    },
+    {
+      "epoch": 1.4169177009278227,
+      "grad_norm": 0.1535140872001648,
+      "learning_rate": 6.341088149270204e-06,
+      "loss": 0.2121,
+      "step": 5880
+    },
+    {
+      "epoch": 1.4193276298349198,
+      "grad_norm": 0.16071055829524994,
+      "learning_rate": 6.327577855668216e-06,
+      "loss": 0.2099,
+      "step": 5890
+    },
+    {
+      "epoch": 1.4217375587420171,
+      "grad_norm": 0.15555599331855774,
+      "learning_rate": 6.3140571260261385e-06,
+      "loss": 0.2119,
+      "step": 5900
+    },
+    {
+      "epoch": 1.4241474876491145,
+      "grad_norm": 0.1541549116373062,
+      "learning_rate": 6.300526066629923e-06,
+      "loss": 0.2087,
+      "step": 5910
+    },
+    {
+      "epoch": 1.4265574165562116,
+      "grad_norm": 0.1483973264694214,
+      "learning_rate": 6.286984783846723e-06,
+      "loss": 0.2126,
+      "step": 5920
+    },
+    {
+      "epoch": 1.4289673454633087,
+      "grad_norm": 0.1526423990726471,
+      "learning_rate": 6.273433384124058e-06,
+      "loss": 0.211,
+      "step": 5930
+    },
+    {
+      "epoch": 1.431377274370406,
+      "grad_norm": 0.18809592723846436,
+      "learning_rate": 6.259871973988977e-06,
+      "loss": 0.211,
+      "step": 5940
+    },
+    {
+      "epoch": 1.4337872032775034,
+      "grad_norm": 0.19463743269443512,
+      "learning_rate": 6.24630066004722e-06,
+      "loss": 0.2114,
+      "step": 5950
+    },
+    {
+      "epoch": 1.4361971321846005,
+      "grad_norm": 0.16021324694156647,
+      "learning_rate": 6.232719548982381e-06,
+      "loss": 0.2111,
+      "step": 5960
+    },
+    {
+      "epoch": 1.4386070610916977,
+      "grad_norm": 0.19545216858386993,
+      "learning_rate": 6.219128747555066e-06,
+      "loss": 0.2087,
+      "step": 5970
+    },
+    {
+      "epoch": 1.441016989998795,
+      "grad_norm": 0.16862092912197113,
+      "learning_rate": 6.205528362602064e-06,
+      "loss": 0.2073,
+      "step": 5980
+    },
+    {
+      "epoch": 1.4434269189058924,
+      "grad_norm": 0.1454867571592331,
+      "learning_rate": 6.19191850103549e-06,
+      "loss": 0.2089,
+      "step": 5990
+    },
+    {
+      "epoch": 1.4458368478129895,
+      "grad_norm": 0.14500007033348083,
+      "learning_rate": 6.1782992698419605e-06,
+      "loss": 0.2096,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4482467767200868,
+      "grad_norm": 0.14368648827075958,
+      "learning_rate": 6.164670776081746e-06,
+      "loss": 0.209,
+      "step": 6010
+    },
+    {
+      "epoch": 1.450656705627184,
+      "grad_norm": 0.15274104475975037,
+      "learning_rate": 6.151033126887928e-06,
+      "loss": 0.2098,
+      "step": 6020
+    },
+    {
+      "epoch": 1.4530666345342813,
+      "grad_norm": 0.15271003544330597,
+      "learning_rate": 6.137386429465557e-06,
+      "loss": 0.2093,
+      "step": 6030
+    },
+    {
+      "epoch": 1.4554765634413784,
+      "grad_norm": 0.15823811292648315,
+      "learning_rate": 6.123730791090814e-06,
+      "loss": 0.2079,
+      "step": 6040
+    },
+    {
+      "epoch": 1.4578864923484758,
+      "grad_norm": 0.15441280603408813,
+      "learning_rate": 6.1100663191101664e-06,
+      "loss": 0.2106,
+      "step": 6050
+    },
+    {
+      "epoch": 1.460296421255573,
+      "grad_norm": 0.15077108144760132,
+      "learning_rate": 6.0963931209395165e-06,
+      "loss": 0.2088,
+      "step": 6060
+    },
+    {
+      "epoch": 1.4627063501626703,
+      "grad_norm": 0.1825905740261078,
+      "learning_rate": 6.082711304063369e-06,
+      "loss": 0.209,
+      "step": 6070
+    },
+    {
+      "epoch": 1.4651162790697674,
+      "grad_norm": 0.18124468624591827,
+      "learning_rate": 6.069020976033973e-06,
+      "loss": 0.2088,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4675262079768647,
+      "grad_norm": 0.1527973711490631,
+      "learning_rate": 6.055322244470492e-06,
+      "loss": 0.212,
+      "step": 6090
+    },
+    {
+      "epoch": 1.4699361368839619,
+      "grad_norm": 0.17023658752441406,
+      "learning_rate": 6.041615217058141e-06,
+      "loss": 0.214,
+      "step": 6100
+    },
+    {
+      "epoch": 1.4723460657910592,
+      "grad_norm": 0.14895488321781158,
+      "learning_rate": 6.027900001547354e-06,
+      "loss": 0.2078,
+      "step": 6110
+    },
+    {
+      "epoch": 1.4747559946981563,
+      "grad_norm": 0.14877694845199585,
+      "learning_rate": 6.014176705752928e-06,
+      "loss": 0.2079,
+      "step": 6120
+    },
+    {
+      "epoch": 1.4771659236052537,
+      "grad_norm": 0.17328816652297974,
+      "learning_rate": 6.000445437553182e-06,
+      "loss": 0.2122,
+      "step": 6130
+    },
+    {
+      "epoch": 1.479575852512351,
+      "grad_norm": 0.14150387048721313,
+      "learning_rate": 5.986706304889103e-06,
+      "loss": 0.2093,
+      "step": 6140
+    },
+    {
+      "epoch": 1.4819857814194481,
+      "grad_norm": 0.14140060544013977,
+      "learning_rate": 5.9729594157635e-06,
+      "loss": 0.2076,
+      "step": 6150
+    },
+    {
+      "epoch": 1.4843957103265453,
+      "grad_norm": 0.145376518368721,
+      "learning_rate": 5.95920487824016e-06,
+      "loss": 0.2084,
+      "step": 6160
+    },
+    {
+      "epoch": 1.4868056392336426,
+      "grad_norm": 0.15817610919475555,
+      "learning_rate": 5.945442800442989e-06,
+      "loss": 0.2091,
+      "step": 6170
+    },
+    {
+      "epoch": 1.48921556814074,
+      "grad_norm": 0.16702431440353394,
+      "learning_rate": 5.9316732905551655e-06,
+      "loss": 0.2096,
+      "step": 6180
+    },
+    {
+      "epoch": 1.491625497047837,
+      "grad_norm": 0.1363145262002945,
+      "learning_rate": 5.917896456818296e-06,
+      "loss": 0.2051,
+      "step": 6190
+    },
+    {
+      "epoch": 1.4940354259549342,
+      "grad_norm": 0.1454513818025589,
+      "learning_rate": 5.904112407531558e-06,
+      "loss": 0.2101,
+      "step": 6200
+    },
+    {
+      "epoch": 1.4964453548620316,
+      "grad_norm": 0.14361591637134552,
+      "learning_rate": 5.89032125105085e-06,
+      "loss": 0.2086,
+      "step": 6210
+    },
+    {
+      "epoch": 1.498855283769129,
+      "grad_norm": 0.15663976967334747,
+      "learning_rate": 5.876523095787938e-06,
+      "loss": 0.2079,
+      "step": 6220
+    },
+    {
+      "epoch": 1.501265212676226,
+      "grad_norm": 0.16403771936893463,
+      "learning_rate": 5.862718050209608e-06,
+      "loss": 0.2077,
+      "step": 6230
+    },
+    {
+      "epoch": 1.5036751415833232,
+      "grad_norm": 0.14956094324588776,
+      "learning_rate": 5.848906222836808e-06,
+      "loss": 0.2102,
+      "step": 6240
+    },
+    {
+      "epoch": 1.5060850704904205,
+      "grad_norm": 0.1433073878288269,
+      "learning_rate": 5.835087722243801e-06,
+      "loss": 0.2067,
+      "step": 6250
+    },
+    {
+      "epoch": 1.5084949993975179,
+      "grad_norm": 0.14256806671619415,
+      "learning_rate": 5.821262657057303e-06,
+      "loss": 0.2094,
+      "step": 6260
+    },
+    {
+      "epoch": 1.510904928304615,
+      "grad_norm": 0.15444211661815643,
+      "learning_rate": 5.807431135955637e-06,
+      "loss": 0.2095,
+      "step": 6270
+    },
+    {
+      "epoch": 1.513314857211712,
+      "grad_norm": 0.15259243547916412,
+      "learning_rate": 5.793593267667876e-06,
+      "loss": 0.2086,
+      "step": 6280
+    },
+    {
+      "epoch": 1.5157247861188095,
+      "grad_norm": 0.1586480736732483,
+      "learning_rate": 5.779749160972988e-06,
+      "loss": 0.2105,
+      "step": 6290
+    },
+    {
+      "epoch": 1.5181347150259068,
+      "grad_norm": 0.140699103474617,
+      "learning_rate": 5.76589892469898e-06,
+      "loss": 0.2057,
+      "step": 6300
+    },
+    {
+      "epoch": 1.520544643933004,
+      "grad_norm": 0.1529569923877716,
+      "learning_rate": 5.752042667722044e-06,
+      "loss": 0.2095,
+      "step": 6310
+    },
+    {
+      "epoch": 1.522954572840101,
+      "grad_norm": 0.15028007328510284,
+      "learning_rate": 5.7381804989656995e-06,
+      "loss": 0.2088,
+      "step": 6320
+    },
+    {
+      "epoch": 1.5253645017471986,
+      "grad_norm": 0.15581537783145905,
+      "learning_rate": 5.724312527399939e-06,
+      "loss": 0.2069,
+      "step": 6330
+    },
+    {
+      "epoch": 1.5277744306542957,
+      "grad_norm": 0.15059341490268707,
+      "learning_rate": 5.710438862040368e-06,
+      "loss": 0.205,
+      "step": 6340
+    },
+    {
+      "epoch": 1.5301843595613929,
+      "grad_norm": 0.1440630406141281,
+      "learning_rate": 5.696559611947359e-06,
+      "loss": 0.208,
+      "step": 6350
+    },
+    {
+      "epoch": 1.5325942884684902,
+      "grad_norm": 0.1505700945854187,
+      "learning_rate": 5.682674886225175e-06,
+      "loss": 0.2064,
+      "step": 6360
+    },
+    {
+      "epoch": 1.5350042173755876,
+      "grad_norm": 0.1707584410905838,
+      "learning_rate": 5.6687847940211304e-06,
+      "loss": 0.2074,
+      "step": 6370
+    },
+    {
+      "epoch": 1.5374141462826847,
+      "grad_norm": 0.16003894805908203,
+      "learning_rate": 5.654889444524723e-06,
+      "loss": 0.2063,
+      "step": 6380
+    },
+    {
+      "epoch": 1.5398240751897818,
+      "grad_norm": 0.14350832998752594,
+      "learning_rate": 5.6409889469667765e-06,
+      "loss": 0.2079,
+      "step": 6390
+    },
+    {
+      "epoch": 1.5422340040968792,
+      "grad_norm": 0.1445057988166809,
+      "learning_rate": 5.627083410618585e-06,
+      "loss": 0.2072,
+      "step": 6400
+    },
+    {
+      "epoch": 1.5446439330039765,
+      "grad_norm": 0.1630876064300537,
+      "learning_rate": 5.613172944791053e-06,
+      "loss": 0.2072,
+      "step": 6410
+    },
+    {
+      "epoch": 1.5470538619110736,
+      "grad_norm": 0.14087872207164764,
+      "learning_rate": 5.599257658833833e-06,
+      "loss": 0.2068,
+      "step": 6420
+    },
+    {
+      "epoch": 1.5494637908181708,
+      "grad_norm": 0.14263446629047394,
+      "learning_rate": 5.585337662134471e-06,
+      "loss": 0.2103,
+      "step": 6430
+    },
+    {
+      "epoch": 1.551873719725268,
+      "grad_norm": 0.14731428027153015,
+      "learning_rate": 5.571413064117542e-06,
+      "loss": 0.2094,
+      "step": 6440
+    },
+    {
+      "epoch": 1.5542836486323655,
+      "grad_norm": 0.15938155353069305,
+      "learning_rate": 5.5574839742437924e-06,
+      "loss": 0.2082,
+      "step": 6450
+    },
+    {
+      "epoch": 1.5566935775394626,
+      "grad_norm": 0.14798109233379364,
+      "learning_rate": 5.543550502009279e-06,
+      "loss": 0.2107,
+      "step": 6460
+    },
+    {
+      "epoch": 1.5591035064465597,
+      "grad_norm": 0.16606839001178741,
+      "learning_rate": 5.529612756944509e-06,
+      "loss": 0.2059,
+      "step": 6470
+    },
+    {
+      "epoch": 1.561513435353657,
+      "grad_norm": 0.15381751954555511,
+      "learning_rate": 5.515670848613577e-06,
+      "loss": 0.2078,
+      "step": 6480
+    },
+    {
+      "epoch": 1.5639233642607544,
+      "grad_norm": 0.14584167301654816,
+      "learning_rate": 5.501724886613304e-06,
+      "loss": 0.2104,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5663332931678515,
+      "grad_norm": 0.14175280928611755,
+      "learning_rate": 5.4877749805723805e-06,
+      "loss": 0.2077,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5687432220749487,
+      "grad_norm": 0.15253135561943054,
+      "learning_rate": 5.4738212401504966e-06,
+      "loss": 0.2114,
+      "step": 6510
+    },
+    {
+      "epoch": 1.571153150982046,
+      "grad_norm": 0.14893187582492828,
+      "learning_rate": 5.459863775037486e-06,
+      "loss": 0.2114,
+      "step": 6520
+    },
+    {
+      "epoch": 1.5735630798891433,
+      "grad_norm": 0.1650102734565735,
+      "learning_rate": 5.445902694952464e-06,
+      "loss": 0.208,
+      "step": 6530
+    },
+    {
+      "epoch": 1.5759730087962405,
+      "grad_norm": 0.15241795778274536,
+      "learning_rate": 5.43193810964296e-06,
+      "loss": 0.2086,
+      "step": 6540
+    },
+    {
+      "epoch": 1.5783829377033376,
+      "grad_norm": 0.1766674816608429,
+      "learning_rate": 5.417970128884061e-06,
+      "loss": 0.2091,
+      "step": 6550
+    },
+    {
+      "epoch": 1.580792866610435,
+      "grad_norm": 0.15203633904457092,
+      "learning_rate": 5.403998862477538e-06,
+      "loss": 0.2091,
+      "step": 6560
+    },
+    {
+      "epoch": 1.5832027955175323,
+      "grad_norm": 0.17118015885353088,
+      "learning_rate": 5.390024420251003e-06,
+      "loss": 0.2056,
+      "step": 6570
+    },
+    {
+      "epoch": 1.5856127244246294,
+      "grad_norm": 0.1646401584148407,
+      "learning_rate": 5.376046912057022e-06,
+      "loss": 0.2051,
+      "step": 6580
+    },
+    {
+      "epoch": 1.5880226533317268,
+      "grad_norm": 0.1695377081632614,
+      "learning_rate": 5.3620664477722686e-06,
+      "loss": 0.2105,
+      "step": 6590
+    },
+    {
+      "epoch": 1.590432582238824,
+      "grad_norm": 0.15378722548484802,
+      "learning_rate": 5.34808313729665e-06,
+      "loss": 0.2051,
+      "step": 6600
+    },
+    {
+      "epoch": 1.5928425111459212,
+      "grad_norm": 0.1414097249507904,
+      "learning_rate": 5.3340970905524515e-06,
+      "loss": 0.2099,
+      "step": 6610
+    },
+    {
+      "epoch": 1.5952524400530184,
+      "grad_norm": 0.14073528349399567,
+      "learning_rate": 5.3201084174834615e-06,
+      "loss": 0.2086,
+      "step": 6620
+    },
+    {
+      "epoch": 1.5976623689601157,
+      "grad_norm": 0.14301954209804535,
+      "learning_rate": 5.306117228054123e-06,
+      "loss": 0.2048,
+      "step": 6630
+    },
+    {
+      "epoch": 1.600072297867213,
+      "grad_norm": 0.14161662757396698,
+      "learning_rate": 5.292123632248652e-06,
+      "loss": 0.2085,
+      "step": 6640
+    },
+    {
+      "epoch": 1.6024822267743102,
+      "grad_norm": 0.15665100514888763,
+      "learning_rate": 5.278127740070187e-06,
+      "loss": 0.2081,
+      "step": 6650
+    },
+    {
+      "epoch": 1.6048921556814073,
+      "grad_norm": 0.17283719778060913,
+      "learning_rate": 5.2641296615399116e-06,
+      "loss": 0.2038,
+      "step": 6660
+    },
+    {
+      "epoch": 1.6073020845885047,
+      "grad_norm": 0.15164655447006226,
+      "learning_rate": 5.2501295066962035e-06,
+      "loss": 0.2088,
+      "step": 6670
+    },
+    {
+      "epoch": 1.609712013495602,
+      "grad_norm": 0.14629319310188293,
+      "learning_rate": 5.236127385593754e-06,
+      "loss": 0.2092,
+      "step": 6680
+    },
+    {
+      "epoch": 1.6121219424026991,
+      "grad_norm": 0.1373308300971985,
+      "learning_rate": 5.222123408302722e-06,
+      "loss": 0.2065,
+      "step": 6690
+    },
+    {
+      "epoch": 1.6145318713097963,
+      "grad_norm": 0.14942149817943573,
+      "learning_rate": 5.208117684907846e-06,
+      "loss": 0.2092,
+      "step": 6700
+    },
+    {
+      "epoch": 1.6169418002168936,
+      "grad_norm": 0.15661108493804932,
+      "learning_rate": 5.194110325507599e-06,
+      "loss": 0.2104,
+      "step": 6710
+    },
+    {
+      "epoch": 1.619351729123991,
+      "grad_norm": 0.1510433405637741,
+      "learning_rate": 5.180101440213311e-06,
+      "loss": 0.2076,
+      "step": 6720
+    },
+    {
+      "epoch": 1.621761658031088,
+      "grad_norm": 0.1418112814426422,
+      "learning_rate": 5.166091139148307e-06,
+      "loss": 0.2094,
+      "step": 6730
+    },
+    {
+      "epoch": 1.6241715869381852,
+      "grad_norm": 0.15742062032222748,
+      "learning_rate": 5.152079532447042e-06,
+      "loss": 0.2056,
+      "step": 6740
+    },
+    {
+      "epoch": 1.6265815158452825,
+      "grad_norm": 0.15039823949337006,
+      "learning_rate": 5.138066730254236e-06,
+      "loss": 0.2048,
+      "step": 6750
+    },
+    {
+      "epoch": 1.62899144475238,
+      "grad_norm": 0.1480005532503128,
+      "learning_rate": 5.124052842724005e-06,
+      "loss": 0.2053,
+      "step": 6760
+    },
+    {
+      "epoch": 1.631401373659477,
+      "grad_norm": 0.156424880027771,
+      "learning_rate": 5.110037980018996e-06,
+      "loss": 0.2109,
+      "step": 6770
+    },
+    {
+      "epoch": 1.6338113025665741,
+      "grad_norm": 0.13942967355251312,
+      "learning_rate": 5.0960222523095235e-06,
+      "loss": 0.205,
+      "step": 6780
+    },
+    {
+      "epoch": 1.6362212314736715,
+      "grad_norm": 0.15093407034873962,
+      "learning_rate": 5.0820057697727e-06,
+      "loss": 0.2115,
+      "step": 6790
+    },
+    {
+      "epoch": 1.6386311603807688,
+      "grad_norm": 0.13788051903247833,
+      "learning_rate": 5.067988642591575e-06,
+      "loss": 0.2056,
+      "step": 6800
+    },
+    {
+      "epoch": 1.641041089287866,
+      "grad_norm": 0.14531250298023224,
+      "learning_rate": 5.053970980954263e-06,
+      "loss": 0.2069,
+      "step": 6810
+    },
+    {
+      "epoch": 1.643451018194963,
+      "grad_norm": 0.14479826390743256,
+      "learning_rate": 5.0399528950530776e-06,
+      "loss": 0.2058,
+      "step": 6820
+    },
+    {
+      "epoch": 1.6458609471020604,
+      "grad_norm": 0.14563125371932983,
+      "learning_rate": 5.0259344950836715e-06,
+      "loss": 0.2101,
+      "step": 6830
+    },
+    {
+      "epoch": 1.6482708760091578,
+      "grad_norm": 0.14034052193164825,
+      "learning_rate": 5.011915891244167e-06,
+      "loss": 0.207,
+      "step": 6840
+    },
+    {
+      "epoch": 1.650680804916255,
+      "grad_norm": 0.14122657477855682,
+      "learning_rate": 4.997897193734285e-06,
+      "loss": 0.208,
+      "step": 6850
+    },
+    {
+      "epoch": 1.6530907338233523,
+      "grad_norm": 0.15636655688285828,
+      "learning_rate": 4.9838785127544826e-06,
+      "loss": 0.2093,
+      "step": 6860
+    },
+    {
+      "epoch": 1.6555006627304496,
+      "grad_norm": 0.15074920654296875,
+      "learning_rate": 4.969859958505094e-06,
+      "loss": 0.2052,
+      "step": 6870
+    },
+    {
+      "epoch": 1.6579105916375467,
+      "grad_norm": 0.15180207788944244,
+      "learning_rate": 4.955841641185447e-06,
+      "loss": 0.2109,
+      "step": 6880
+    },
+    {
+      "epoch": 1.6603205205446439,
+      "grad_norm": 0.15867879986763,
+      "learning_rate": 4.941823670993016e-06,
+      "loss": 0.2096,
+      "step": 6890
+    },
+    {
+      "epoch": 1.6627304494517412,
+      "grad_norm": 0.14111950993537903,
+      "learning_rate": 4.92780615812254e-06,
+      "loss": 0.2051,
+      "step": 6900
+    },
+    {
+      "epoch": 1.6651403783588385,
+      "grad_norm": 0.14959976077079773,
+      "learning_rate": 4.913789212765166e-06,
+      "loss": 0.2063,
+      "step": 6910
+    },
+    {
+      "epoch": 1.6675503072659357,
+      "grad_norm": 0.16154243052005768,
+      "learning_rate": 4.899772945107583e-06,
+      "loss": 0.2068,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6699602361730328,
+      "grad_norm": 0.14693604409694672,
+      "learning_rate": 4.885757465331144e-06,
+      "loss": 0.2111,
+      "step": 6930
+    },
+    {
+      "epoch": 1.6723701650801301,
+      "grad_norm": 0.1400395929813385,
+      "learning_rate": 4.871742883611018e-06,
+      "loss": 0.2057,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6747800939872275,
+      "grad_norm": 0.13714446127414703,
+      "learning_rate": 4.857729310115307e-06,
+      "loss": 0.2058,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6771900228943246,
+      "grad_norm": 0.1699269860982895,
+      "learning_rate": 4.843716855004194e-06,
+      "loss": 0.2081,
+      "step": 6960
+    },
+    {
+      "epoch": 1.6795999518014217,
+      "grad_norm": 0.1475098431110382,
+      "learning_rate": 4.829705628429061e-06,
+      "loss": 0.209,
+      "step": 6970
+    },
+    {
+      "epoch": 1.682009880708519,
+      "grad_norm": 0.14293253421783447,
+      "learning_rate": 4.815695740531643e-06,
+      "loss": 0.2068,
+      "step": 6980
+    },
+    {
+      "epoch": 1.6844198096156164,
+      "grad_norm": 0.143426775932312,
+      "learning_rate": 4.801687301443149e-06,
+      "loss": 0.2078,
+      "step": 6990
+    },
+    {
+      "epoch": 1.6868297385227136,
+      "grad_norm": 0.14896607398986816,
+      "learning_rate": 4.787680421283391e-06,
+      "loss": 0.205,
+      "step": 7000
+    },
+    {
+      "epoch": 1.6892396674298107,
+      "grad_norm": 0.14818888902664185,
+      "learning_rate": 4.773675210159938e-06,
+      "loss": 0.2055,
+      "step": 7010
+    },
+    {
+      "epoch": 1.691649596336908,
+      "grad_norm": 0.15228557586669922,
+      "learning_rate": 4.759671778167228e-06,
+      "loss": 0.2057,
+      "step": 7020
+    },
+    {
+      "epoch": 1.6940595252440054,
+      "grad_norm": 0.1329268515110016,
+      "learning_rate": 4.745670235385723e-06,
+      "loss": 0.2097,
+      "step": 7030
+    },
+    {
+      "epoch": 1.6964694541511025,
+      "grad_norm": 0.14906267821788788,
+      "learning_rate": 4.7316706918810265e-06,
+      "loss": 0.2086,
+      "step": 7040
+    },
+    {
+      "epoch": 1.6988793830581996,
+      "grad_norm": 0.16887860000133514,
+      "learning_rate": 4.71767325770303e-06,
+      "loss": 0.2067,
+      "step": 7050
+    },
+    {
+      "epoch": 1.701289311965297,
+      "grad_norm": 0.13959096372127533,
+      "learning_rate": 4.703678042885044e-06,
+      "loss": 0.2075,
+      "step": 7060
+    },
+    {
+      "epoch": 1.7036992408723943,
+      "grad_norm": 0.150526762008667,
+      "learning_rate": 4.689685157442927e-06,
+      "loss": 0.2032,
+      "step": 7070
+    },
+    {
+      "epoch": 1.7061091697794915,
+      "grad_norm": 0.15431441366672516,
+      "learning_rate": 4.675694711374234e-06,
+      "loss": 0.2047,
+      "step": 7080
+    },
+    {
+      "epoch": 1.7085190986865888,
+      "grad_norm": 0.14908158779144287,
+      "learning_rate": 4.661706814657338e-06,
+      "loss": 0.2078,
+      "step": 7090
+    },
+    {
+      "epoch": 1.7109290275936861,
+      "grad_norm": 0.15429182350635529,
+      "learning_rate": 4.647721577250578e-06,
+      "loss": 0.2052,
+      "step": 7100
+    },
+    {
+      "epoch": 1.7133389565007833,
+      "grad_norm": 0.15195661783218384,
+      "learning_rate": 4.633739109091379e-06,
+      "loss": 0.2122,
+      "step": 7110
+    },
+    {
+      "epoch": 1.7157488854078804,
+      "grad_norm": 0.14676164090633392,
+      "learning_rate": 4.61975952009541e-06,
+      "loss": 0.2055,
+      "step": 7120
+    },
+    {
+      "epoch": 1.7181588143149777,
+      "grad_norm": 0.13878756761550903,
+      "learning_rate": 4.6057829201556905e-06,
+      "loss": 0.2073,
+      "step": 7130
+    },
+    {
+      "epoch": 1.720568743222075,
+      "grad_norm": 0.1423376202583313,
+      "learning_rate": 4.591809419141758e-06,
+      "loss": 0.2064,
+      "step": 7140
+    },
+    {
+      "epoch": 1.7229786721291722,
+      "grad_norm": 0.14275546371936798,
+      "learning_rate": 4.577839126898784e-06,
+      "loss": 0.2083,
+      "step": 7150
+    },
+    {
+      "epoch": 1.7253886010362693,
+      "grad_norm": 0.15022709965705872,
+      "learning_rate": 4.5638721532467125e-06,
+      "loss": 0.2068,
+      "step": 7160
+    },
+    {
+      "epoch": 1.7277985299433667,
+      "grad_norm": 0.14701788127422333,
+      "learning_rate": 4.549908607979407e-06,
+      "loss": 0.2068,
+      "step": 7170
+    },
+    {
+      "epoch": 1.730208458850464,
+      "grad_norm": 0.14002536237239838,
+      "learning_rate": 4.535948600863774e-06,
+      "loss": 0.2098,
+      "step": 7180
+    },
+    {
+      "epoch": 1.7326183877575612,
+      "grad_norm": 0.15771852433681488,
+      "learning_rate": 4.521992241638912e-06,
+      "loss": 0.2051,
+      "step": 7190
+    },
+    {
+      "epoch": 1.7350283166646583,
+      "grad_norm": 0.1449091136455536,
+      "learning_rate": 4.508039640015237e-06,
+      "loss": 0.2069,
+      "step": 7200
+    },
+    {
+      "epoch": 1.7374382455717556,
+      "grad_norm": 0.14416846632957458,
+      "learning_rate": 4.494090905673634e-06,
+      "loss": 0.2098,
+      "step": 7210
+    },
+    {
+      "epoch": 1.739848174478853,
+      "grad_norm": 0.1471363753080368,
+      "learning_rate": 4.480146148264586e-06,
+      "loss": 0.207,
+      "step": 7220
+    },
+    {
+      "epoch": 1.74225810338595,
+      "grad_norm": 0.15887276828289032,
+      "learning_rate": 4.466205477407308e-06,
+      "loss": 0.2063,
+      "step": 7230
+    },
+    {
+      "epoch": 1.7446680322930472,
+      "grad_norm": 0.14145879447460175,
+      "learning_rate": 4.452269002688897e-06,
+      "loss": 0.2079,
+      "step": 7240
+    },
+    {
+      "epoch": 1.7470779612001446,
+      "grad_norm": 0.14188285171985626,
+      "learning_rate": 4.438336833663459e-06,
+      "loss": 0.2094,
+      "step": 7250
+    },
+    {
+      "epoch": 1.749487890107242,
+      "grad_norm": 0.12969772517681122,
+      "learning_rate": 4.424409079851262e-06,
+      "loss": 0.2081,
+      "step": 7260
+    },
+    {
+      "epoch": 1.751897819014339,
+      "grad_norm": 0.13423402607440948,
+      "learning_rate": 4.410485850737853e-06,
+      "loss": 0.2081,
+      "step": 7270
+    },
+    {
+      "epoch": 1.7543077479214362,
+      "grad_norm": 0.14299023151397705,
+      "learning_rate": 4.39656725577322e-06,
+      "loss": 0.2064,
+      "step": 7280
+    },
+    {
+      "epoch": 1.7567176768285335,
+      "grad_norm": 0.14568351209163666,
+      "learning_rate": 4.382653404370922e-06,
+      "loss": 0.207,
+      "step": 7290
+    },
+    {
+      "epoch": 1.7591276057356309,
+      "grad_norm": 0.13449597358703613,
+      "learning_rate": 4.368744405907224e-06,
+      "loss": 0.2081,
+      "step": 7300
+    },
+    {
+      "epoch": 1.761537534642728,
+      "grad_norm": 0.1447085440158844,
+      "learning_rate": 4.354840369720249e-06,
+      "loss": 0.2081,
+      "step": 7310
+    },
+    {
+      "epoch": 1.7639474635498251,
+      "grad_norm": 0.13538062572479248,
+      "learning_rate": 4.340941405109102e-06,
+      "loss": 0.2101,
+      "step": 7320
+    },
+    {
+      "epoch": 1.7663573924569225,
+      "grad_norm": 0.16251257061958313,
+      "learning_rate": 4.327047621333031e-06,
+      "loss": 0.2064,
+      "step": 7330
+    },
+    {
+      "epoch": 1.7687673213640198,
+      "grad_norm": 0.1393039971590042,
+      "learning_rate": 4.31315912761055e-06,
+      "loss": 0.2045,
+      "step": 7340
+    },
+    {
+      "epoch": 1.771177250271117,
+      "grad_norm": 0.14810238778591156,
+      "learning_rate": 4.299276033118592e-06,
+      "loss": 0.208,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7735871791782143,
+      "grad_norm": 0.1339663565158844,
+      "learning_rate": 4.285398446991648e-06,
+      "loss": 0.2082,
+      "step": 7360
+    },
+    {
+      "epoch": 1.7759971080853116,
+      "grad_norm": 0.14078187942504883,
+      "learning_rate": 4.271526478320901e-06,
+      "loss": 0.2065,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7784070369924088,
+      "grad_norm": 0.13188710808753967,
+      "learning_rate": 4.257660236153387e-06,
+      "loss": 0.2041,
+      "step": 7380
+    },
+    {
+      "epoch": 1.7808169658995059,
+      "grad_norm": 0.1408006250858307,
+      "learning_rate": 4.2437998294911114e-06,
+      "loss": 0.2082,
+      "step": 7390
+    },
+    {
+      "epoch": 1.7832268948066032,
+      "grad_norm": 0.14175231754779816,
+      "learning_rate": 4.229945367290222e-06,
+      "loss": 0.2076,
+      "step": 7400
+    },
+    {
+      "epoch": 1.7856368237137006,
+      "grad_norm": 0.13607344031333923,
+      "learning_rate": 4.216096958460126e-06,
+      "loss": 0.2059,
+      "step": 7410
+    },
+    {
+      "epoch": 1.7880467526207977,
+      "grad_norm": 0.15421463549137115,
+      "learning_rate": 4.2022547118626515e-06,
+      "loss": 0.208,
+      "step": 7420
+    },
+    {
+      "epoch": 1.7904566815278948,
+      "grad_norm": 0.14106203615665436,
+      "learning_rate": 4.18841873631118e-06,
+      "loss": 0.2079,
+      "step": 7430
+    },
+    {
+      "epoch": 1.7928666104349922,
+      "grad_norm": 0.14798112213611603,
+      "learning_rate": 4.174589140569805e-06,
+      "loss": 0.2073,
+      "step": 7440
+    },
+    {
+      "epoch": 1.7952765393420895,
+      "grad_norm": 0.13187192380428314,
+      "learning_rate": 4.160766033352462e-06,
+      "loss": 0.2066,
+      "step": 7450
+    },
+    {
+      "epoch": 1.7976864682491867,
+      "grad_norm": 0.1424017697572708,
+      "learning_rate": 4.14694952332208e-06,
+      "loss": 0.2046,
+      "step": 7460
+    },
+    {
+      "epoch": 1.8000963971562838,
+      "grad_norm": 0.13720762729644775,
+      "learning_rate": 4.133139719089735e-06,
+      "loss": 0.2032,
+      "step": 7470
+    },
+    {
+      "epoch": 1.8025063260633811,
+      "grad_norm": 0.14268174767494202,
+      "learning_rate": 4.119336729213778e-06,
+      "loss": 0.2097,
+      "step": 7480
+    },
+    {
+      "epoch": 1.8049162549704785,
+      "grad_norm": 0.15912844240665436,
+      "learning_rate": 4.105540662199002e-06,
+      "loss": 0.2082,
+      "step": 7490
+    },
+    {
+      "epoch": 1.8073261838775756,
+      "grad_norm": 0.1520354002714157,
+      "learning_rate": 4.0917516264957735e-06,
+      "loss": 0.2038,
+      "step": 7500
+    },
+    {
+      "epoch": 1.8097361127846727,
+      "grad_norm": 0.13783125579357147,
+      "learning_rate": 4.07796973049919e-06,
+      "loss": 0.2067,
+      "step": 7510
+    },
+    {
+      "epoch": 1.81214604169177,
+      "grad_norm": 0.14950409531593323,
+      "learning_rate": 4.0641950825482265e-06,
+      "loss": 0.2065,
+      "step": 7520
+    },
+    {
+      "epoch": 1.8145559705988674,
+      "grad_norm": 0.14664943516254425,
+      "learning_rate": 4.0504277909248715e-06,
+      "loss": 0.2048,
+      "step": 7530
+    },
+    {
+      "epoch": 1.8169658995059645,
+      "grad_norm": 0.14661647379398346,
+      "learning_rate": 4.036667963853296e-06,
+      "loss": 0.208,
+      "step": 7540
+    },
+    {
+      "epoch": 1.8193758284130617,
+      "grad_norm": 0.14886929094791412,
+      "learning_rate": 4.022915709498985e-06,
+      "loss": 0.2071,
+      "step": 7550
+    },
+    {
+      "epoch": 1.821785757320159,
+      "grad_norm": 0.16615243256092072,
+      "learning_rate": 4.009171135967902e-06,
+      "loss": 0.2066,
+      "step": 7560
+    },
+    {
+      "epoch": 1.8241956862272564,
+      "grad_norm": 0.13197271525859833,
+      "learning_rate": 3.9954343513056236e-06,
+      "loss": 0.207,
+      "step": 7570
+    },
+    {
+      "epoch": 1.8266056151343535,
+      "grad_norm": 0.13960954546928406,
+      "learning_rate": 3.981705463496504e-06,
+      "loss": 0.2045,
+      "step": 7580
+    },
+    {
+      "epoch": 1.8290155440414506,
+      "grad_norm": 0.13621185719966888,
+      "learning_rate": 3.967984580462821e-06,
+      "loss": 0.2079,
+      "step": 7590
+    },
+    {
+      "epoch": 1.8314254729485482,
+      "grad_norm": 0.14133788645267487,
+      "learning_rate": 3.954271810063922e-06,
+      "loss": 0.2087,
+      "step": 7600
+    },
+    {
+      "epoch": 1.8338354018556453,
+      "grad_norm": 0.15335774421691895,
+      "learning_rate": 3.940567260095389e-06,
+      "loss": 0.2075,
+      "step": 7610
+    },
+    {
+      "epoch": 1.8362453307627424,
+      "grad_norm": 0.15637780725955963,
+      "learning_rate": 3.926871038288173e-06,
+      "loss": 0.2054,
+      "step": 7620
+    },
+    {
+      "epoch": 1.8386552596698398,
+      "grad_norm": 0.13300193846225739,
+      "learning_rate": 3.9131832523077685e-06,
+      "loss": 0.207,
+      "step": 7630
+    },
+    {
+      "epoch": 1.8410651885769371,
+      "grad_norm": 0.14508311450481415,
+      "learning_rate": 3.8995040097533495e-06,
+      "loss": 0.2075,
+      "step": 7640
+    },
+    {
+      "epoch": 1.8434751174840343,
+      "grad_norm": 0.13044893741607666,
+      "learning_rate": 3.885833418156932e-06,
+      "loss": 0.2077,
+      "step": 7650
+    },
+    {
+      "epoch": 1.8458850463911314,
+      "grad_norm": 0.14246976375579834,
+      "learning_rate": 3.8721715849825305e-06,
+      "loss": 0.2062,
+      "step": 7660
+    },
+    {
+      "epoch": 1.8482949752982287,
+      "grad_norm": 0.13763968646526337,
+      "learning_rate": 3.858518617625301e-06,
+      "loss": 0.2062,
+      "step": 7670
+    },
+    {
+      "epoch": 1.850704904205326,
+      "grad_norm": 0.12822872400283813,
+      "learning_rate": 3.844874623410718e-06,
+      "loss": 0.2053,
+      "step": 7680
+    },
+    {
+      "epoch": 1.8531148331124232,
+      "grad_norm": 0.14770273864269257,
+      "learning_rate": 3.831239709593707e-06,
+      "loss": 0.2059,
+      "step": 7690
+    },
+    {
+      "epoch": 1.8555247620195203,
+      "grad_norm": 0.15450125932693481,
+      "learning_rate": 3.8176139833578215e-06,
+      "loss": 0.2054,
+      "step": 7700
+    },
+    {
+      "epoch": 1.8579346909266177,
+      "grad_norm": 0.15053947269916534,
+      "learning_rate": 3.8039975518143862e-06,
+      "loss": 0.2059,
+      "step": 7710
+    },
+    {
+      "epoch": 1.860344619833715,
+      "grad_norm": 0.16280502080917358,
+      "learning_rate": 3.790390522001662e-06,
+      "loss": 0.207,
+      "step": 7720
+    },
+    {
+      "epoch": 1.8627545487408121,
+      "grad_norm": 0.13911543786525726,
+      "learning_rate": 3.7767930008840055e-06,
+      "loss": 0.2052,
+      "step": 7730
+    },
+    {
+      "epoch": 1.8651644776479093,
+      "grad_norm": 0.14479966461658478,
+      "learning_rate": 3.763205095351021e-06,
+      "loss": 0.2049,
+      "step": 7740
+    },
+    {
+      "epoch": 1.8675744065550066,
+      "grad_norm": 0.13834457099437714,
+      "learning_rate": 3.7496269122167306e-06,
+      "loss": 0.2071,
+      "step": 7750
+    },
+    {
+      "epoch": 1.869984335462104,
+      "grad_norm": 0.1692366749048233,
+      "learning_rate": 3.7360585582187246e-06,
+      "loss": 0.2093,
+      "step": 7760
+    },
+    {
+      "epoch": 1.872394264369201,
+      "grad_norm": 0.14670221507549286,
+      "learning_rate": 3.7225001400173303e-06,
+      "loss": 0.2057,
+      "step": 7770
+    },
+    {
+      "epoch": 1.8748041932762982,
+      "grad_norm": 0.14623087644577026,
+      "learning_rate": 3.708951764194767e-06,
+      "loss": 0.2052,
+      "step": 7780
+    },
+    {
+      "epoch": 1.8772141221833956,
+      "grad_norm": 0.13682162761688232,
+      "learning_rate": 3.6954135372543133e-06,
+      "loss": 0.2058,
+      "step": 7790
+    },
+    {
+      "epoch": 1.879624051090493,
+      "grad_norm": 0.13509906828403473,
+      "learning_rate": 3.681885565619465e-06,
+      "loss": 0.2061,
+      "step": 7800
+    },
+    {
+      "epoch": 1.88203397999759,
+      "grad_norm": 0.13584059476852417,
+      "learning_rate": 3.668367955633107e-06,
+      "loss": 0.2063,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8844439089046872,
+      "grad_norm": 0.14202377200126648,
+      "learning_rate": 3.654860813556666e-06,
+      "loss": 0.2078,
+      "step": 7820
+    },
+    {
+      "epoch": 1.8868538378117845,
+      "grad_norm": 0.1395299881696701,
+      "learning_rate": 3.6413642455692826e-06,
+      "loss": 0.2022,
+      "step": 7830
+    },
+    {
+      "epoch": 1.8892637667188819,
+      "grad_norm": 0.14222587645053864,
+      "learning_rate": 3.6278783577669762e-06,
+      "loss": 0.2076,
+      "step": 7840
+    },
+    {
+      "epoch": 1.891673695625979,
+      "grad_norm": 0.14933638274669647,
+      "learning_rate": 3.614403256161807e-06,
+      "loss": 0.2047,
+      "step": 7850
+    },
+    {
+      "epoch": 1.8940836245330763,
+      "grad_norm": 0.1349928379058838,
+      "learning_rate": 3.60093904668105e-06,
+      "loss": 0.2057,
+      "step": 7860
+    },
+    {
+      "epoch": 1.8964935534401737,
+      "grad_norm": 0.13887856900691986,
+      "learning_rate": 3.5874858351663513e-06,
+      "loss": 0.2053,
+      "step": 7870
+    },
+    {
+      "epoch": 1.8989034823472708,
+      "grad_norm": 0.13957622647285461,
+      "learning_rate": 3.5740437273729074e-06,
+      "loss": 0.2069,
+      "step": 7880
+    },
+    {
+      "epoch": 1.901313411254368,
+      "grad_norm": 0.149143248796463,
+      "learning_rate": 3.560612828968627e-06,
+      "loss": 0.2056,
+      "step": 7890
+    },
+    {
+      "epoch": 1.9037233401614653,
+      "grad_norm": 0.14313067495822906,
+      "learning_rate": 3.5471932455333013e-06,
+      "loss": 0.2058,
+      "step": 7900
+    },
+    {
+      "epoch": 1.9061332690685626,
+      "grad_norm": 0.13201439380645752,
+      "learning_rate": 3.533785082557779e-06,
+      "loss": 0.2063,
+      "step": 7910
+    },
+    {
+      "epoch": 1.9085431979756597,
+      "grad_norm": 0.14562028646469116,
+      "learning_rate": 3.520388445443126e-06,
+      "loss": 0.207,
+      "step": 7920
+    },
+    {
+      "epoch": 1.9109531268827569,
+      "grad_norm": 0.1508123278617859,
+      "learning_rate": 3.5070034394998108e-06,
+      "loss": 0.2047,
+      "step": 7930
+    },
+    {
+      "epoch": 1.9133630557898542,
+      "grad_norm": 0.1282895803451538,
+      "learning_rate": 3.4936301699468646e-06,
+      "loss": 0.2055,
+      "step": 7940
+    },
+    {
+      "epoch": 1.9157729846969516,
+      "grad_norm": 0.14230744540691376,
+      "learning_rate": 3.4802687419110635e-06,
+      "loss": 0.2032,
+      "step": 7950
+    },
+    {
+      "epoch": 1.9181829136040487,
+      "grad_norm": 0.135861337184906,
+      "learning_rate": 3.466919260426095e-06,
+      "loss": 0.2067,
+      "step": 7960
+    },
+    {
+      "epoch": 1.9205928425111458,
+      "grad_norm": 0.1485607922077179,
+      "learning_rate": 3.4535818304317338e-06,
+      "loss": 0.2041,
+      "step": 7970
+    },
+    {
+      "epoch": 1.9230027714182432,
+      "grad_norm": 0.14034852385520935,
+      "learning_rate": 3.440256556773025e-06,
+      "loss": 0.2034,
+      "step": 7980
+    },
+    {
+      "epoch": 1.9254127003253405,
+      "grad_norm": 0.1325804740190506,
+      "learning_rate": 3.426943544199444e-06,
+      "loss": 0.2061,
+      "step": 7990
+    },
+    {
+      "epoch": 1.9278226292324376,
+      "grad_norm": 0.14449110627174377,
+      "learning_rate": 3.413642897364091e-06,
+      "loss": 0.2063,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9302325581395348,
+      "grad_norm": 0.14102737605571747,
+      "learning_rate": 3.400354720822851e-06,
+      "loss": 0.2072,
+      "step": 8010
+    },
+    {
+      "epoch": 1.932642487046632,
+      "grad_norm": 0.13754191994667053,
+      "learning_rate": 3.38707911903359e-06,
+      "loss": 0.2088,
+      "step": 8020
+    },
+    {
+      "epoch": 1.9350524159537295,
+      "grad_norm": 0.15111030638217926,
+      "learning_rate": 3.373816196355315e-06,
+      "loss": 0.2039,
+      "step": 8030
+    },
+    {
+      "epoch": 1.9374623448608266,
+      "grad_norm": 0.1356048732995987,
+      "learning_rate": 3.3605660570473687e-06,
+      "loss": 0.2055,
+      "step": 8040
+    },
+    {
+      "epoch": 1.9398722737679237,
+      "grad_norm": 0.1311032921075821,
+      "learning_rate": 3.3473288052686055e-06,
+      "loss": 0.2066,
+      "step": 8050
+    },
+    {
+      "epoch": 1.942282202675021,
+      "grad_norm": 0.13652199506759644,
+      "learning_rate": 3.334104545076564e-06,
+      "loss": 0.2046,
+      "step": 8060
+    },
+    {
+      "epoch": 1.9446921315821184,
+      "grad_norm": 0.1403491497039795,
+      "learning_rate": 3.320893380426667e-06,
+      "loss": 0.2069,
+      "step": 8070
+    },
+    {
+      "epoch": 1.9471020604892155,
+      "grad_norm": 0.14957809448242188,
+      "learning_rate": 3.3076954151713815e-06,
+      "loss": 0.2067,
+      "step": 8080
+    },
+    {
+      "epoch": 1.9495119893963127,
+      "grad_norm": 0.1357545703649521,
+      "learning_rate": 3.294510753059427e-06,
+      "loss": 0.2024,
+      "step": 8090
+    },
+    {
+      "epoch": 1.95192191830341,
+      "grad_norm": 0.13906139135360718,
+      "learning_rate": 3.2813394977349356e-06,
+      "loss": 0.2029,
+      "step": 8100
+    },
+    {
+      "epoch": 1.9543318472105073,
+      "grad_norm": 0.1386336386203766,
+      "learning_rate": 3.2681817527366575e-06,
+      "loss": 0.2036,
+      "step": 8110
+    },
+    {
+      "epoch": 1.9567417761176045,
+      "grad_norm": 0.1412561684846878,
+      "learning_rate": 3.2550376214971395e-06,
+      "loss": 0.207,
+      "step": 8120
+    },
+    {
+      "epoch": 1.9591517050247018,
+      "grad_norm": 0.13507452607154846,
+      "learning_rate": 3.241907207341902e-06,
+      "loss": 0.2046,
+      "step": 8130
+    },
+    {
+      "epoch": 1.9615616339317992,
+      "grad_norm": 0.12901359796524048,
+      "learning_rate": 3.2287906134886483e-06,
+      "loss": 0.2033,
+      "step": 8140
+    },
+    {
+      "epoch": 1.9639715628388963,
+      "grad_norm": 0.1413521021604538,
+      "learning_rate": 3.215687943046427e-06,
+      "loss": 0.2042,
+      "step": 8150
+    },
+    {
+      "epoch": 1.9663814917459934,
+      "grad_norm": 0.15636803209781647,
+      "learning_rate": 3.202599299014849e-06,
+      "loss": 0.208,
+      "step": 8160
+    },
+    {
+      "epoch": 1.9687914206530908,
+      "grad_norm": 0.1508229821920395,
+      "learning_rate": 3.1895247842832523e-06,
+      "loss": 0.206,
+      "step": 8170
+    },
+    {
+      "epoch": 1.971201349560188,
+      "grad_norm": 0.14489571750164032,
+      "learning_rate": 3.1764645016299133e-06,
+      "loss": 0.1998,
+      "step": 8180
+    },
+    {
+      "epoch": 1.9736112784672852,
+      "grad_norm": 0.1390523463487625,
+      "learning_rate": 3.163418553721229e-06,
+      "loss": 0.2053,
+      "step": 8190
+    },
+    {
+      "epoch": 1.9760212073743824,
+      "grad_norm": 0.14797191321849823,
+      "learning_rate": 3.1503870431109067e-06,
+      "loss": 0.2079,
+      "step": 8200
+    },
+    {
+      "epoch": 1.9784311362814797,
+      "grad_norm": 0.12731759250164032,
+      "learning_rate": 3.1373700722391696e-06,
+      "loss": 0.203,
+      "step": 8210
+    },
+    {
+      "epoch": 1.980841065188577,
+      "grad_norm": 0.1378549486398697,
+      "learning_rate": 3.1243677434319373e-06,
+      "loss": 0.2066,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9832509940956742,
+      "grad_norm": 0.14989231526851654,
+      "learning_rate": 3.111380158900037e-06,
+      "loss": 0.2032,
+      "step": 8230
+    },
+    {
+      "epoch": 1.9856609230027713,
+      "grad_norm": 0.15311060845851898,
+      "learning_rate": 3.098407420738382e-06,
+      "loss": 0.2049,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9880708519098687,
+      "grad_norm": 0.1495356559753418,
+      "learning_rate": 3.0854496309251857e-06,
+      "loss": 0.2061,
+      "step": 8250
+    },
+    {
+      "epoch": 1.990480780816966,
+      "grad_norm": 0.13795362412929535,
+      "learning_rate": 3.0725068913211546e-06,
+      "loss": 0.2031,
+      "step": 8260
+    },
+    {
+      "epoch": 1.9928907097240631,
+      "grad_norm": 0.12623263895511627,
+      "learning_rate": 3.059579303668678e-06,
+      "loss": 0.207,
+      "step": 8270
+    },
+    {
+      "epoch": 1.9953006386311603,
+      "grad_norm": 0.13744844496250153,
+      "learning_rate": 3.046666969591046e-06,
+      "loss": 0.2053,
+      "step": 8280
+    },
+    {
+      "epoch": 1.9977105675382576,
+      "grad_norm": 0.13788847625255585,
+      "learning_rate": 3.0337699905916308e-06,
+      "loss": 0.2064,
+      "step": 8290
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.1785513460636139,
+      "learning_rate": 3.020888468053109e-06,
+      "loss": 0.2027,
+      "step": 8300
+    },
+    {
+      "epoch": 2.002409928907097,
+      "grad_norm": 0.13940368592739105,
+      "learning_rate": 3.0080225032366443e-06,
+      "loss": 0.1998,
+      "step": 8310
+    },
+    {
+      "epoch": 2.0048198578141947,
+      "grad_norm": 0.1354367583990097,
+      "learning_rate": 2.9951721972811133e-06,
+      "loss": 0.1983,
+      "step": 8320
+    },
+    {
+      "epoch": 2.007229786721292,
+      "grad_norm": 0.1270235925912857,
+      "learning_rate": 2.982337651202286e-06,
+      "loss": 0.1963,
+      "step": 8330
+    },
+    {
+      "epoch": 2.009639715628389,
+      "grad_norm": 0.13577590882778168,
+      "learning_rate": 2.9695189658920555e-06,
+      "loss": 0.1993,
+      "step": 8340
+    },
+    {
+      "epoch": 2.012049644535486,
+      "grad_norm": 0.14152930676937103,
+      "learning_rate": 2.95671624211763e-06,
+      "loss": 0.2001,
+      "step": 8350
+    },
+    {
+      "epoch": 2.0144595734425836,
+      "grad_norm": 0.13332296907901764,
+      "learning_rate": 2.9439295805207415e-06,
+      "loss": 0.1965,
+      "step": 8360
+    },
+    {
+      "epoch": 2.0168695023496808,
+      "grad_norm": 0.14344486594200134,
+      "learning_rate": 2.9311590816168646e-06,
+      "loss": 0.1979,
+      "step": 8370
+    },
+    {
+      "epoch": 2.019279431256778,
+      "grad_norm": 0.14192751049995422,
+      "learning_rate": 2.918404845794411e-06,
+      "loss": 0.1973,
+      "step": 8380
+    },
+    {
+      "epoch": 2.021689360163875,
+      "grad_norm": 0.1400498002767563,
+      "learning_rate": 2.905666973313957e-06,
+      "loss": 0.1979,
+      "step": 8390
+    },
+    {
+      "epoch": 2.0240992890709726,
+      "grad_norm": 0.13859783113002777,
+      "learning_rate": 2.8929455643074433e-06,
+      "loss": 0.1978,
+      "step": 8400
+    },
+    {
+      "epoch": 2.0265092179780697,
+      "grad_norm": 0.13500601053237915,
+      "learning_rate": 2.8802407187773917e-06,
+      "loss": 0.2007,
+      "step": 8410
+    },
+    {
+      "epoch": 2.028919146885167,
+      "grad_norm": 0.14077068865299225,
+      "learning_rate": 2.86755253659612e-06,
+      "loss": 0.1995,
+      "step": 8420
+    },
+    {
+      "epoch": 2.031329075792264,
+      "grad_norm": 0.12985944747924805,
+      "learning_rate": 2.854881117504954e-06,
+      "loss": 0.1997,
+      "step": 8430
+    },
+    {
+      "epoch": 2.0337390046993615,
+      "grad_norm": 0.12818634510040283,
+      "learning_rate": 2.8422265611134535e-06,
+      "loss": 0.2004,
+      "step": 8440
+    },
+    {
+      "epoch": 2.0361489336064587,
+      "grad_norm": 0.13415859639644623,
+      "learning_rate": 2.829588966898607e-06,
+      "loss": 0.1983,
+      "step": 8450
+    },
+    {
+      "epoch": 2.038558862513556,
+      "grad_norm": 0.13348065316677094,
+      "learning_rate": 2.8169684342040802e-06,
+      "loss": 0.1995,
+      "step": 8460
+    },
+    {
+      "epoch": 2.040968791420653,
+      "grad_norm": 0.13919700682163239,
+      "learning_rate": 2.8043650622394023e-06,
+      "loss": 0.1994,
+      "step": 8470
+    },
+    {
+      "epoch": 2.0433787203277505,
+      "grad_norm": 0.12803754210472107,
+      "learning_rate": 2.791778950079217e-06,
+      "loss": 0.1984,
+      "step": 8480
+    },
+    {
+      "epoch": 2.0457886492348476,
+      "grad_norm": 0.12549884617328644,
+      "learning_rate": 2.779210196662482e-06,
+      "loss": 0.1983,
+      "step": 8490
+    },
+    {
+      "epoch": 2.0481985781419447,
+      "grad_norm": 0.13034506142139435,
+      "learning_rate": 2.766658900791699e-06,
+      "loss": 0.2004,
+      "step": 8500
+    },
+    {
+      "epoch": 2.050608507049042,
+      "grad_norm": 0.12898895144462585,
+      "learning_rate": 2.7541251611321385e-06,
+      "loss": 0.2015,
+      "step": 8510
+    },
+    {
+      "epoch": 2.0530184359561394,
+      "grad_norm": 0.1386294662952423,
+      "learning_rate": 2.7416090762110603e-06,
+      "loss": 0.1988,
+      "step": 8520
+    },
+    {
+      "epoch": 2.0554283648632365,
+      "grad_norm": 0.15744347870349884,
+      "learning_rate": 2.729110744416943e-06,
+      "loss": 0.2002,
+      "step": 8530
+    },
+    {
+      "epoch": 2.0578382937703337,
+      "grad_norm": 0.1407720148563385,
+      "learning_rate": 2.716630263998706e-06,
+      "loss": 0.1982,
+      "step": 8540
+    },
+    {
+      "epoch": 2.060248222677431,
+      "grad_norm": 0.14954039454460144,
+      "learning_rate": 2.7041677330649408e-06,
+      "loss": 0.2026,
+      "step": 8550
+    },
+    {
+      "epoch": 2.0626581515845284,
+      "grad_norm": 0.1366991400718689,
+      "learning_rate": 2.6917232495831436e-06,
+      "loss": 0.1978,
+      "step": 8560
+    },
+    {
+      "epoch": 2.0650680804916255,
+      "grad_norm": 0.1298081874847412,
+      "learning_rate": 2.6792969113789285e-06,
+      "loss": 0.2013,
+      "step": 8570
+    },
+    {
+      "epoch": 2.0674780093987226,
+      "grad_norm": 0.13624030351638794,
+      "learning_rate": 2.666888816135285e-06,
+      "loss": 0.1967,
+      "step": 8580
+    },
+    {
+      "epoch": 2.06988793830582,
+      "grad_norm": 0.13076886534690857,
+      "learning_rate": 2.6544990613917803e-06,
+      "loss": 0.1984,
+      "step": 8590
+    },
+    {
+      "epoch": 2.0722978672129173,
+      "grad_norm": 0.1781616061925888,
+      "learning_rate": 2.642127744543823e-06,
+      "loss": 0.2006,
+      "step": 8600
+    },
+    {
+      "epoch": 2.0747077961200144,
+      "grad_norm": 0.1324646919965744,
+      "learning_rate": 2.6297749628418654e-06,
+      "loss": 0.198,
+      "step": 8610
+    },
+    {
+      "epoch": 2.0771177250271116,
+      "grad_norm": 0.1453695148229599,
+      "learning_rate": 2.617440813390674e-06,
+      "loss": 0.1996,
+      "step": 8620
+    },
+    {
+      "epoch": 2.079527653934209,
+      "grad_norm": 0.1317797154188156,
+      "learning_rate": 2.605125393148529e-06,
+      "loss": 0.1976,
+      "step": 8630
+    },
+    {
+      "epoch": 2.0819375828413063,
+      "grad_norm": 0.12967994809150696,
+      "learning_rate": 2.592828798926496e-06,
+      "loss": 0.1948,
+      "step": 8640
+    },
+    {
+      "epoch": 2.0843475117484034,
+      "grad_norm": 0.12639065086841583,
+      "learning_rate": 2.580551127387644e-06,
+      "loss": 0.1995,
+      "step": 8650
+    },
+    {
+      "epoch": 2.0867574406555005,
+      "grad_norm": 0.13363341987133026,
+      "learning_rate": 2.5682924750462907e-06,
+      "loss": 0.1984,
+      "step": 8660
+    },
+    {
+      "epoch": 2.089167369562598,
+      "grad_norm": 0.1281256228685379,
+      "learning_rate": 2.5560529382672462e-06,
+      "loss": 0.1974,
+      "step": 8670
+    },
+    {
+      "epoch": 2.091577298469695,
+      "grad_norm": 0.13677582144737244,
+      "learning_rate": 2.5438326132650524e-06,
+      "loss": 0.1997,
+      "step": 8680
+    },
+    {
+      "epoch": 2.0939872273767923,
+      "grad_norm": 0.13215544819831848,
+      "learning_rate": 2.531631596103231e-06,
+      "loss": 0.1982,
+      "step": 8690
+    },
+    {
+      "epoch": 2.0963971562838895,
+      "grad_norm": 0.1397799551486969,
+      "learning_rate": 2.5194499826935216e-06,
+      "loss": 0.1972,
+      "step": 8700
+    },
+    {
+      "epoch": 2.098807085190987,
+      "grad_norm": 0.12224389612674713,
+      "learning_rate": 2.507287868795133e-06,
+      "loss": 0.1984,
+      "step": 8710
+    },
+    {
+      "epoch": 2.101217014098084,
+      "grad_norm": 0.13411986827850342,
+      "learning_rate": 2.495145350013995e-06,
+      "loss": 0.1975,
+      "step": 8720
+    },
+    {
+      "epoch": 2.1036269430051813,
+      "grad_norm": 0.1322869509458542,
+      "learning_rate": 2.4830225218019878e-06,
+      "loss": 0.1998,
+      "step": 8730
+    },
+    {
+      "epoch": 2.1060368719122784,
+      "grad_norm": 0.12509003281593323,
+      "learning_rate": 2.4709194794562204e-06,
+      "loss": 0.1984,
+      "step": 8740
+    },
+    {
+      "epoch": 2.108446800819376,
+      "grad_norm": 0.12176263332366943,
+      "learning_rate": 2.45883631811825e-06,
+      "loss": 0.1982,
+      "step": 8750
+    },
+    {
+      "epoch": 2.110856729726473,
+      "grad_norm": 0.1365647315979004,
+      "learning_rate": 2.4467731327733665e-06,
+      "loss": 0.1968,
+      "step": 8760
+    },
+    {
+      "epoch": 2.11326665863357,
+      "grad_norm": 0.1353791207075119,
+      "learning_rate": 2.4347300182498116e-06,
+      "loss": 0.1991,
+      "step": 8770
+    },
+    {
+      "epoch": 2.1156765875406673,
+      "grad_norm": 0.1300830841064453,
+      "learning_rate": 2.422707069218068e-06,
+      "loss": 0.1993,
+      "step": 8780
+    },
+    {
+      "epoch": 2.118086516447765,
+      "grad_norm": 0.13076601922512054,
+      "learning_rate": 2.4107043801900863e-06,
+      "loss": 0.1977,
+      "step": 8790
+    },
+    {
+      "epoch": 2.120496445354862,
+      "grad_norm": 0.1522458791732788,
+      "learning_rate": 2.39872204551856e-06,
+      "loss": 0.1974,
+      "step": 8800
+    },
+    {
+      "epoch": 2.122906374261959,
+      "grad_norm": 0.13156403601169586,
+      "learning_rate": 2.3867601593961744e-06,
+      "loss": 0.2021,
+      "step": 8810
+    },
+    {
+      "epoch": 2.1253163031690567,
+      "grad_norm": 0.13702353835105896,
+      "learning_rate": 2.374818815854871e-06,
+      "loss": 0.1971,
+      "step": 8820
+    },
+    {
+      "epoch": 2.127726232076154,
+      "grad_norm": 0.13330979645252228,
+      "learning_rate": 2.3628981087651073e-06,
+      "loss": 0.1968,
+      "step": 8830
+    },
+    {
+      "epoch": 2.130136160983251,
+      "grad_norm": 0.14794397354125977,
+      "learning_rate": 2.350998131835117e-06,
+      "loss": 0.2004,
+      "step": 8840
+    },
+    {
+      "epoch": 2.132546089890348,
+      "grad_norm": 0.14478328824043274,
+      "learning_rate": 2.339118978610175e-06,
+      "loss": 0.2013,
+      "step": 8850
+    },
+    {
+      "epoch": 2.1349560187974457,
+      "grad_norm": 0.130048468708992,
+      "learning_rate": 2.3272607424718675e-06,
+      "loss": 0.197,
+      "step": 8860
+    },
+    {
+      "epoch": 2.137365947704543,
+      "grad_norm": 0.129298135638237,
+      "learning_rate": 2.315423516637339e-06,
+      "loss": 0.1994,
+      "step": 8870
+    },
+    {
+      "epoch": 2.13977587661164,
+      "grad_norm": 0.12726633250713348,
+      "learning_rate": 2.3036073941585898e-06,
+      "loss": 0.1967,
+      "step": 8880
+    },
+    {
+      "epoch": 2.142185805518737,
+      "grad_norm": 0.12661658227443695,
+      "learning_rate": 2.2918124679217106e-06,
+      "loss": 0.1944,
+      "step": 8890
+    },
+    {
+      "epoch": 2.1445957344258346,
+      "grad_norm": 0.1268956959247589,
+      "learning_rate": 2.2800388306461847e-06,
+      "loss": 0.1976,
+      "step": 8900
+    },
+    {
+      "epoch": 2.1470056633329317,
+      "grad_norm": 0.12949858605861664,
+      "learning_rate": 2.2682865748841293e-06,
+      "loss": 0.1973,
+      "step": 8910
+    },
+    {
+      "epoch": 2.149415592240029,
+      "grad_norm": 0.1312812715768814,
+      "learning_rate": 2.2565557930195963e-06,
+      "loss": 0.2005,
+      "step": 8920
+    },
+    {
+      "epoch": 2.151825521147126,
+      "grad_norm": 0.13287796080112457,
+      "learning_rate": 2.244846577267818e-06,
+      "loss": 0.1984,
+      "step": 8930
+    },
+    {
+      "epoch": 2.1542354500542236,
+      "grad_norm": 0.13826043903827667,
+      "learning_rate": 2.2331590196745094e-06,
+      "loss": 0.2,
+      "step": 8940
+    },
+    {
+      "epoch": 2.1566453789613207,
+      "grad_norm": 0.12798218429088593,
+      "learning_rate": 2.221493212115123e-06,
+      "loss": 0.2005,
+      "step": 8950
+    },
+    {
+      "epoch": 2.159055307868418,
+      "grad_norm": 0.17131908237934113,
+      "learning_rate": 2.209849246294138e-06,
+      "loss": 0.1991,
+      "step": 8960
+    },
+    {
+      "epoch": 2.161465236775515,
+      "grad_norm": 0.14354480803012848,
+      "learning_rate": 2.1982272137443356e-06,
+      "loss": 0.1972,
+      "step": 8970
+    },
+    {
+      "epoch": 2.1638751656826125,
+      "grad_norm": 0.13229574263095856,
+      "learning_rate": 2.186627205826082e-06,
+      "loss": 0.199,
+      "step": 8980
+    },
+    {
+      "epoch": 2.1662850945897096,
+      "grad_norm": 0.12601736187934875,
+      "learning_rate": 2.1750493137266064e-06,
+      "loss": 0.202,
+      "step": 8990
+    },
+    {
+      "epoch": 2.1686950234968068,
+      "grad_norm": 0.1347973644733429,
+      "learning_rate": 2.1634936284592882e-06,
+      "loss": 0.2004,
+      "step": 9000
+    },
+    {
+      "epoch": 2.171104952403904,
+      "grad_norm": 0.14534367620944977,
+      "learning_rate": 2.151960240862937e-06,
+      "loss": 0.2009,
+      "step": 9010
+    },
+    {
+      "epoch": 2.1735148813110015,
+      "grad_norm": 0.13880938291549683,
+      "learning_rate": 2.1404492416010885e-06,
+      "loss": 0.1972,
+      "step": 9020
+    },
+    {
+      "epoch": 2.1759248102180986,
+      "grad_norm": 0.12886486947536469,
+      "learning_rate": 2.128960721161273e-06,
+      "loss": 0.1998,
+      "step": 9030
+    },
+    {
+      "epoch": 2.1783347391251957,
+      "grad_norm": 0.13548776507377625,
+      "learning_rate": 2.1174947698543276e-06,
+      "loss": 0.1975,
+      "step": 9040
+    },
+    {
+      "epoch": 2.1807446680322933,
+      "grad_norm": 0.1285417228937149,
+      "learning_rate": 2.106051477813662e-06,
+      "loss": 0.2003,
+      "step": 9050
+    },
+    {
+      "epoch": 2.1831545969393904,
+      "grad_norm": 0.12515997886657715,
+      "learning_rate": 2.0946309349945764e-06,
+      "loss": 0.1992,
+      "step": 9060
+    },
+    {
+      "epoch": 2.1855645258464875,
+      "grad_norm": 0.13670197129249573,
+      "learning_rate": 2.0832332311735255e-06,
+      "loss": 0.1969,
+      "step": 9070
+    },
+    {
+      "epoch": 2.1879744547535847,
+      "grad_norm": 0.13152837753295898,
+      "learning_rate": 2.071858455947439e-06,
+      "loss": 0.2005,
+      "step": 9080
+    },
+    {
+      "epoch": 2.1903843836606818,
+      "grad_norm": 0.12892812490463257,
+      "learning_rate": 2.0605066987330015e-06,
+      "loss": 0.2029,
+      "step": 9090
+    },
+    {
+      "epoch": 2.1927943125677793,
+      "grad_norm": 0.125423401594162,
+      "learning_rate": 2.0491780487659518e-06,
+      "loss": 0.1995,
+      "step": 9100
+    },
+    {
+      "epoch": 2.1952042414748765,
+      "grad_norm": 0.13129930198192596,
+      "learning_rate": 2.0378725951003863e-06,
+      "loss": 0.201,
+      "step": 9110
+    },
+    {
+      "epoch": 2.1976141703819736,
+      "grad_norm": 0.12565447390079498,
+      "learning_rate": 2.0265904266080553e-06,
+      "loss": 0.1973,
+      "step": 9120
+    },
+    {
+      "epoch": 2.200024099289071,
+      "grad_norm": 0.11998391151428223,
+      "learning_rate": 2.0153316319776663e-06,
+      "loss": 0.1969,
+      "step": 9130
+    },
+    {
+      "epoch": 2.2024340281961683,
+      "grad_norm": 0.1322380006313324,
+      "learning_rate": 2.004096299714182e-06,
+      "loss": 0.1952,
+      "step": 9140
+    },
+    {
+      "epoch": 2.2048439571032654,
+      "grad_norm": 0.13540686666965485,
+      "learning_rate": 1.992884518138132e-06,
+      "loss": 0.1982,
+      "step": 9150
+    },
+    {
+      "epoch": 2.2072538860103625,
+      "grad_norm": 0.13474301993846893,
+      "learning_rate": 1.9816963753849173e-06,
+      "loss": 0.1974,
+      "step": 9160
+    },
+    {
+      "epoch": 2.20966381491746,
+      "grad_norm": 0.13281075656414032,
+      "learning_rate": 1.9705319594041055e-06,
+      "loss": 0.2006,
+      "step": 9170
+    },
+    {
+      "epoch": 2.2120737438245572,
+      "grad_norm": 0.13555601239204407,
+      "learning_rate": 1.959391357958761e-06,
+      "loss": 0.199,
+      "step": 9180
+    },
+    {
+      "epoch": 2.2144836727316544,
+      "grad_norm": 0.12954938411712646,
+      "learning_rate": 1.9482746586247307e-06,
+      "loss": 0.1992,
+      "step": 9190
+    },
+    {
+      "epoch": 2.2168936016387515,
+      "grad_norm": 0.1328035593032837,
+      "learning_rate": 1.937181948789979e-06,
+      "loss": 0.1993,
+      "step": 9200
+    },
+    {
+      "epoch": 2.219303530545849,
+      "grad_norm": 0.12565462291240692,
+      "learning_rate": 1.926113315653883e-06,
+      "loss": 0.1973,
+      "step": 9210
+    },
+    {
+      "epoch": 2.221713459452946,
+      "grad_norm": 0.123862624168396,
+      "learning_rate": 1.9150688462265567e-06,
+      "loss": 0.1967,
+      "step": 9220
+    },
+    {
+      "epoch": 2.2241233883600433,
+      "grad_norm": 0.13187842071056366,
+      "learning_rate": 1.9040486273281611e-06,
+      "loss": 0.1977,
+      "step": 9230
+    },
+    {
+      "epoch": 2.2265333172671404,
+      "grad_norm": 0.1303049623966217,
+      "learning_rate": 1.8930527455882285e-06,
+      "loss": 0.1994,
+      "step": 9240
+    },
+    {
+      "epoch": 2.228943246174238,
+      "grad_norm": 0.1302567571401596,
+      "learning_rate": 1.8820812874449745e-06,
+      "loss": 0.2006,
+      "step": 9250
+    },
+    {
+      "epoch": 2.231353175081335,
+      "grad_norm": 0.13377898931503296,
+      "learning_rate": 1.871134339144624e-06,
+      "loss": 0.1992,
+      "step": 9260
+    },
+    {
+      "epoch": 2.2337631039884323,
+      "grad_norm": 0.12116660177707672,
+      "learning_rate": 1.8602119867407293e-06,
+      "loss": 0.1993,
+      "step": 9270
+    },
+    {
+      "epoch": 2.2361730328955294,
+      "grad_norm": 0.11837529391050339,
+      "learning_rate": 1.8493143160934962e-06,
+      "loss": 0.1975,
+      "step": 9280
+    },
+    {
+      "epoch": 2.238582961802627,
+      "grad_norm": 0.1215796023607254,
+      "learning_rate": 1.838441412869108e-06,
+      "loss": 0.1987,
+      "step": 9290
+    },
+    {
+      "epoch": 2.240992890709724,
+      "grad_norm": 0.16470345854759216,
+      "learning_rate": 1.827593362539052e-06,
+      "loss": 0.1992,
+      "step": 9300
+    },
+    {
+      "epoch": 2.243402819616821,
+      "grad_norm": 0.12217085063457489,
+      "learning_rate": 1.816770250379446e-06,
+      "loss": 0.1968,
+      "step": 9310
+    },
+    {
+      "epoch": 2.2458127485239183,
+      "grad_norm": 0.1257430613040924,
+      "learning_rate": 1.8059721614703757e-06,
+      "loss": 0.1999,
+      "step": 9320
+    },
+    {
+      "epoch": 2.248222677431016,
+      "grad_norm": 0.12400317937135696,
+      "learning_rate": 1.7951991806952135e-06,
+      "loss": 0.1999,
+      "step": 9330
+    },
+    {
+      "epoch": 2.250632606338113,
+      "grad_norm": 0.1254672110080719,
+      "learning_rate": 1.7844513927399605e-06,
+      "loss": 0.1971,
+      "step": 9340
+    },
+    {
+      "epoch": 2.25304253524521,
+      "grad_norm": 0.12951840460300446,
+      "learning_rate": 1.7737288820925775e-06,
+      "loss": 0.1994,
+      "step": 9350
+    },
+    {
+      "epoch": 2.2554524641523077,
+      "grad_norm": 0.13221119344234467,
+      "learning_rate": 1.7630317330423213e-06,
+      "loss": 0.1973,
+      "step": 9360
+    },
+    {
+      "epoch": 2.257862393059405,
+      "grad_norm": 0.13148674368858337,
+      "learning_rate": 1.7523600296790827e-06,
+      "loss": 0.1982,
+      "step": 9370
+    },
+    {
+      "epoch": 2.260272321966502,
+      "grad_norm": 0.12214197218418121,
+      "learning_rate": 1.7417138558927244e-06,
+      "loss": 0.2,
+      "step": 9380
+    },
+    {
+      "epoch": 2.262682250873599,
+      "grad_norm": 0.13050705194473267,
+      "learning_rate": 1.731093295372422e-06,
+      "loss": 0.198,
+      "step": 9390
+    },
+    {
+      "epoch": 2.2650921797806967,
+      "grad_norm": 0.12322299927473068,
+      "learning_rate": 1.7204984316060063e-06,
+      "loss": 0.1964,
+      "step": 9400
+    },
+    {
+      "epoch": 2.267502108687794,
+      "grad_norm": 0.1294129341840744,
+      "learning_rate": 1.7099293478793066e-06,
+      "loss": 0.2003,
+      "step": 9410
+    },
+    {
+      "epoch": 2.269912037594891,
+      "grad_norm": 0.13078904151916504,
+      "learning_rate": 1.699386127275497e-06,
+      "loss": 0.1996,
+      "step": 9420
+    },
+    {
+      "epoch": 2.272321966501988,
+      "grad_norm": 0.14058412611484528,
+      "learning_rate": 1.6888688526744419e-06,
+      "loss": 0.1992,
+      "step": 9430
+    },
+    {
+      "epoch": 2.2747318954090856,
+      "grad_norm": 0.12963654100894928,
+      "learning_rate": 1.6783776067520435e-06,
+      "loss": 0.1979,
+      "step": 9440
+    },
+    {
+      "epoch": 2.2771418243161827,
+      "grad_norm": 0.13461080193519592,
+      "learning_rate": 1.667912471979599e-06,
+      "loss": 0.1988,
+      "step": 9450
+    },
+    {
+      "epoch": 2.27955175322328,
+      "grad_norm": 0.1404641717672348,
+      "learning_rate": 1.6574735306231415e-06,
+      "loss": 0.1971,
+      "step": 9460
+    },
+    {
+      "epoch": 2.281961682130377,
+      "grad_norm": 0.13926105201244354,
+      "learning_rate": 1.6470608647427994e-06,
+      "loss": 0.2002,
+      "step": 9470
+    },
+    {
+      "epoch": 2.2843716110374745,
+      "grad_norm": 0.1367444097995758,
+      "learning_rate": 1.6366745561921526e-06,
+      "loss": 0.2011,
+      "step": 9480
+    },
+    {
+      "epoch": 2.2867815399445717,
+      "grad_norm": 0.13203634321689606,
+      "learning_rate": 1.626314686617586e-06,
+      "loss": 0.2018,
+      "step": 9490
+    },
+    {
+      "epoch": 2.289191468851669,
+      "grad_norm": 0.12602578103542328,
+      "learning_rate": 1.6159813374576473e-06,
+      "loss": 0.1984,
+      "step": 9500
+    },
+    {
+      "epoch": 2.291601397758766,
+      "grad_norm": 0.12635314464569092,
+      "learning_rate": 1.605674589942411e-06,
+      "loss": 0.1975,
+      "step": 9510
+    },
+    {
+      "epoch": 2.2940113266658635,
+      "grad_norm": 0.12311098724603653,
+      "learning_rate": 1.5953945250928337e-06,
+      "loss": 0.1997,
+      "step": 9520
+    },
+    {
+      "epoch": 2.2964212555729606,
+      "grad_norm": 0.120974600315094,
+      "learning_rate": 1.5851412237201241e-06,
+      "loss": 0.2014,
+      "step": 9530
+    },
+    {
+      "epoch": 2.2988311844800577,
+      "grad_norm": 0.13310423493385315,
+      "learning_rate": 1.5749147664251008e-06,
+      "loss": 0.197,
+      "step": 9540
+    },
+    {
+      "epoch": 2.301241113387155,
+      "grad_norm": 0.12243101745843887,
+      "learning_rate": 1.5647152335975675e-06,
+      "loss": 0.2,
+      "step": 9550
+    },
+    {
+      "epoch": 2.3036510422942524,
+      "grad_norm": 0.1300317943096161,
+      "learning_rate": 1.5545427054156659e-06,
+      "loss": 0.1967,
+      "step": 9560
+    },
+    {
+      "epoch": 2.3060609712013496,
+      "grad_norm": 0.12345019727945328,
+      "learning_rate": 1.5443972618452685e-06,
+      "loss": 0.1955,
+      "step": 9570
+    },
+    {
+      "epoch": 2.3084709001084467,
+      "grad_norm": 0.126836359500885,
+      "learning_rate": 1.5342789826393223e-06,
+      "loss": 0.1988,
+      "step": 9580
+    },
+    {
+      "epoch": 2.3108808290155443,
+      "grad_norm": 0.11867355555295944,
+      "learning_rate": 1.5241879473372501e-06,
+      "loss": 0.1973,
+      "step": 9590
+    },
+    {
+      "epoch": 2.3132907579226414,
+      "grad_norm": 0.1371241956949234,
+      "learning_rate": 1.5141242352642975e-06,
+      "loss": 0.1969,
+      "step": 9600
+    },
+    {
+      "epoch": 2.3157006868297385,
+      "grad_norm": 0.11983615905046463,
+      "learning_rate": 1.5040879255309366e-06,
+      "loss": 0.1963,
+      "step": 9610
+    },
+    {
+      "epoch": 2.3181106157368356,
+      "grad_norm": 0.12212738394737244,
+      "learning_rate": 1.4940790970322217e-06,
+      "loss": 0.1976,
+      "step": 9620
+    },
+    {
+      "epoch": 2.3205205446439328,
+      "grad_norm": 0.12494116276502609,
+      "learning_rate": 1.4840978284471818e-06,
+      "loss": 0.1955,
+      "step": 9630
+    },
+    {
+      "epoch": 2.3229304735510303,
+      "grad_norm": 0.12898215651512146,
+      "learning_rate": 1.4741441982381965e-06,
+      "loss": 0.1988,
+      "step": 9640
+    },
+    {
+      "epoch": 2.3253404024581275,
+      "grad_norm": 0.13270017504692078,
+      "learning_rate": 1.4642182846503834e-06,
+      "loss": 0.1974,
+      "step": 9650
+    },
+    {
+      "epoch": 2.3277503313652246,
+      "grad_norm": 0.1331421136856079,
+      "learning_rate": 1.454320165710979e-06,
+      "loss": 0.2005,
+      "step": 9660
+    },
+    {
+      "epoch": 2.330160260272322,
+      "grad_norm": 0.12074144184589386,
+      "learning_rate": 1.4444499192287275e-06,
+      "loss": 0.1974,
+      "step": 9670
+    },
+    {
+      "epoch": 2.3325701891794193,
+      "grad_norm": 0.11966565996408463,
+      "learning_rate": 1.434607622793268e-06,
+      "loss": 0.2004,
+      "step": 9680
+    },
+    {
+      "epoch": 2.3349801180865164,
+      "grad_norm": 0.12328942865133286,
+      "learning_rate": 1.4247933537745312e-06,
+      "loss": 0.1984,
+      "step": 9690
+    },
+    {
+      "epoch": 2.3373900469936135,
+      "grad_norm": 0.12541703879833221,
+      "learning_rate": 1.4150071893221134e-06,
+      "loss": 0.1984,
+      "step": 9700
+    },
+    {
+      "epoch": 2.339799975900711,
+      "grad_norm": 0.12745068967342377,
+      "learning_rate": 1.4052492063646954e-06,
+      "loss": 0.1974,
+      "step": 9710
+    },
+    {
+      "epoch": 2.342209904807808,
+      "grad_norm": 0.12669983506202698,
+      "learning_rate": 1.395519481609412e-06,
+      "loss": 0.1968,
+      "step": 9720
+    },
+    {
+      "epoch": 2.3446198337149053,
+      "grad_norm": 0.12797443568706512,
+      "learning_rate": 1.3858180915412733e-06,
+      "loss": 0.1965,
+      "step": 9730
+    },
+    {
+      "epoch": 2.3470297626220025,
+      "grad_norm": 0.13788028061389923,
+      "learning_rate": 1.376145112422539e-06,
+      "loss": 0.2012,
+      "step": 9740
+    },
+    {
+      "epoch": 2.3494396915291,
+      "grad_norm": 0.14654968678951263,
+      "learning_rate": 1.3665006202921422e-06,
+      "loss": 0.1984,
+      "step": 9750
+    },
+    {
+      "epoch": 2.351849620436197,
+      "grad_norm": 0.13153453171253204,
+      "learning_rate": 1.3568846909650757e-06,
+      "loss": 0.1972,
+      "step": 9760
+    },
+    {
+      "epoch": 2.3542595493432943,
+      "grad_norm": 0.1279347985982895,
+      "learning_rate": 1.347297400031801e-06,
+      "loss": 0.1986,
+      "step": 9770
+    },
+    {
+      "epoch": 2.3566694782503914,
+      "grad_norm": 0.1343514323234558,
+      "learning_rate": 1.337738822857656e-06,
+      "loss": 0.197,
+      "step": 9780
+    },
+    {
+      "epoch": 2.359079407157489,
+      "grad_norm": 0.1319359540939331,
+      "learning_rate": 1.3282090345822591e-06,
+      "loss": 0.1999,
+      "step": 9790
+    },
+    {
+      "epoch": 2.361489336064586,
+      "grad_norm": 0.12443242222070694,
+      "learning_rate": 1.3187081101189215e-06,
+      "loss": 0.1973,
+      "step": 9800
+    },
+    {
+      "epoch": 2.3638992649716832,
+      "grad_norm": 0.1242903470993042,
+      "learning_rate": 1.309236124154057e-06,
+      "loss": 0.202,
+      "step": 9810
+    },
+    {
+      "epoch": 2.366309193878781,
+      "grad_norm": 0.12554222345352173,
+      "learning_rate": 1.299793151146594e-06,
+      "loss": 0.1966,
+      "step": 9820
+    },
+    {
+      "epoch": 2.368719122785878,
+      "grad_norm": 0.12424005568027496,
+      "learning_rate": 1.2903792653273916e-06,
+      "loss": 0.1985,
+      "step": 9830
+    },
+    {
+      "epoch": 2.371129051692975,
+      "grad_norm": 0.13209636509418488,
+      "learning_rate": 1.2809945406986546e-06,
+      "loss": 0.1994,
+      "step": 9840
+    },
+    {
+      "epoch": 2.373538980600072,
+      "grad_norm": 0.13022255897521973,
+      "learning_rate": 1.2716390510333586e-06,
+      "loss": 0.1999,
+      "step": 9850
+    },
+    {
+      "epoch": 2.3759489095071693,
+      "grad_norm": 0.12892918288707733,
+      "learning_rate": 1.2623128698746527e-06,
+      "loss": 0.2001,
+      "step": 9860
+    },
+    {
+      "epoch": 2.378358838414267,
+      "grad_norm": 0.1219906210899353,
+      "learning_rate": 1.2530160705353068e-06,
+      "loss": 0.1954,
+      "step": 9870
+    },
+    {
+      "epoch": 2.380768767321364,
+      "grad_norm": 0.12953771650791168,
+      "learning_rate": 1.243748726097107e-06,
+      "loss": 0.1954,
+      "step": 9880
+    },
+    {
+      "epoch": 2.383178696228461,
+      "grad_norm": 0.12258866429328918,
+      "learning_rate": 1.2345109094103102e-06,
+      "loss": 0.1969,
+      "step": 9890
+    },
+    {
+      "epoch": 2.3855886251355587,
+      "grad_norm": 0.1489989310503006,
+      "learning_rate": 1.2253026930930422e-06,
+      "loss": 0.1998,
+      "step": 9900
+    },
+    {
+      "epoch": 2.387998554042656,
+      "grad_norm": 0.12289855629205704,
+      "learning_rate": 1.2161241495307546e-06,
+      "loss": 0.2003,
+      "step": 9910
+    },
+    {
+      "epoch": 2.390408482949753,
+      "grad_norm": 0.12116892635822296,
+      "learning_rate": 1.2069753508756332e-06,
+      "loss": 0.2007,
+      "step": 9920
+    },
+    {
+      "epoch": 2.39281841185685,
+      "grad_norm": 0.12422724068164825,
+      "learning_rate": 1.1978563690460454e-06,
+      "loss": 0.1979,
+      "step": 9930
+    },
+    {
+      "epoch": 2.3952283407639476,
+      "grad_norm": 0.12689027190208435,
+      "learning_rate": 1.188767275725966e-06,
+      "loss": 0.1976,
+      "step": 9940
+    },
+    {
+      "epoch": 2.3976382696710448,
+      "grad_norm": 0.13472682237625122,
+      "learning_rate": 1.1797081423644207e-06,
+      "loss": 0.1992,
+      "step": 9950
+    },
+    {
+      "epoch": 2.400048198578142,
+      "grad_norm": 0.12316633760929108,
+      "learning_rate": 1.1706790401749191e-06,
+      "loss": 0.1964,
+      "step": 9960
+    },
+    {
+      "epoch": 2.402458127485239,
+      "grad_norm": 0.12504875659942627,
+      "learning_rate": 1.161680040134897e-06,
+      "loss": 0.1996,
+      "step": 9970
+    },
+    {
+      "epoch": 2.4048680563923366,
+      "grad_norm": 0.137268528342247,
+      "learning_rate": 1.152711212985157e-06,
+      "loss": 0.2009,
+      "step": 9980
+    },
+    {
+      "epoch": 2.4072779852994337,
+      "grad_norm": 0.1279861330986023,
+      "learning_rate": 1.1437726292293205e-06,
+      "loss": 0.2005,
+      "step": 9990
+    },
+    {
+      "epoch": 2.409687914206531,
+      "grad_norm": 0.12599658966064453,
+      "learning_rate": 1.1348643591332554e-06,
+      "loss": 0.1983,
+      "step": 10000
+    },
+    {
+      "epoch": 2.412097843113628,
+      "grad_norm": 0.12882481515407562,
+      "learning_rate": 1.1259864727245462e-06,
+      "loss": 0.2012,
+      "step": 10010
+    },
+    {
+      "epoch": 2.4145077720207255,
+      "grad_norm": 0.12207888811826706,
+      "learning_rate": 1.117139039791923e-06,
+      "loss": 0.2005,
+      "step": 10020
+    },
+    {
+      "epoch": 2.4169177009278227,
+      "grad_norm": 0.1228935495018959,
+      "learning_rate": 1.1083221298847318e-06,
+      "loss": 0.1992,
+      "step": 10030
+    },
+    {
+      "epoch": 2.41932762983492,
+      "grad_norm": 0.12841002643108368,
+      "learning_rate": 1.0995358123123672e-06,
+      "loss": 0.1982,
+      "step": 10040
+    },
+    {
+      "epoch": 2.4217375587420173,
+      "grad_norm": 0.12162084132432938,
+      "learning_rate": 1.0907801561437487e-06,
+      "loss": 0.1978,
+      "step": 10050
+    },
+    {
+      "epoch": 2.4241474876491145,
+      "grad_norm": 0.12312573939561844,
+      "learning_rate": 1.0820552302067626e-06,
+      "loss": 0.2002,
+      "step": 10060
+    },
+    {
+      "epoch": 2.4265574165562116,
+      "grad_norm": 0.1217411682009697,
+      "learning_rate": 1.0733611030877261e-06,
+      "loss": 0.1965,
+      "step": 10070
+    },
+    {
+      "epoch": 2.4289673454633087,
+      "grad_norm": 0.12311185896396637,
+      "learning_rate": 1.064697843130849e-06,
+      "loss": 0.2001,
+      "step": 10080
+    },
+    {
+      "epoch": 2.431377274370406,
+      "grad_norm": 0.12260051816701889,
+      "learning_rate": 1.0560655184376956e-06,
+      "loss": 0.1971,
+      "step": 10090
+    },
+    {
+      "epoch": 2.4337872032775034,
+      "grad_norm": 0.11911850422620773,
+      "learning_rate": 1.0474641968666482e-06,
+      "loss": 0.1968,
+      "step": 10100
+    },
+    {
+      "epoch": 2.4361971321846005,
+      "grad_norm": 0.1274978220462799,
+      "learning_rate": 1.0388939460323761e-06,
+      "loss": 0.2007,
+      "step": 10110
+    },
+    {
+      "epoch": 2.4386070610916977,
+      "grad_norm": 0.12279874086380005,
+      "learning_rate": 1.030354833305301e-06,
+      "loss": 0.2007,
+      "step": 10120
+    },
+    {
+      "epoch": 2.4410169899987952,
+      "grad_norm": 0.11948572844266891,
+      "learning_rate": 1.0218469258110713e-06,
+      "loss": 0.2028,
+      "step": 10130
+    },
+    {
+      "epoch": 2.4434269189058924,
+      "grad_norm": 0.11698044836521149,
+      "learning_rate": 1.013370290430029e-06,
+      "loss": 0.1985,
+      "step": 10140
+    },
+    {
+      "epoch": 2.4458368478129895,
+      "grad_norm": 0.1203853189945221,
+      "learning_rate": 1.0049249937966938e-06,
+      "loss": 0.2024,
+      "step": 10150
+    },
+    {
+      "epoch": 2.4482467767200866,
+      "grad_norm": 0.1191658154129982,
+      "learning_rate": 9.96511102299222e-07,
+      "loss": 0.198,
+      "step": 10160
+    },
+    {
+      "epoch": 2.450656705627184,
+      "grad_norm": 0.13178017735481262,
+      "learning_rate": 9.881286820789066e-07,
+      "loss": 0.1965,
+      "step": 10170
+    },
+    {
+      "epoch": 2.4530666345342813,
+      "grad_norm": 0.12409935891628265,
+      "learning_rate": 9.797777990296364e-07,
+      "loss": 0.198,
+      "step": 10180
+    },
+    {
+      "epoch": 2.4554765634413784,
+      "grad_norm": 0.12384223937988281,
+      "learning_rate": 9.714585187973962e-07,
+      "loss": 0.1996,
+      "step": 10190
+    },
+    {
+      "epoch": 2.4578864923484756,
+      "grad_norm": 0.12161009758710861,
+      "learning_rate": 9.631709067797346e-07,
+      "loss": 0.1979,
+      "step": 10200
+    },
+    {
+      "epoch": 2.460296421255573,
+      "grad_norm": 0.1231224313378334,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 0.196,
+      "step": 10210
+    },
+    {
+      "epoch": 2.4627063501626703,
+      "grad_norm": 0.12427617609500885,
+      "learning_rate": 9.466909477331365e-07,
+      "loss": 0.1953,
+      "step": 10220
+    },
+    {
+      "epoch": 2.4651162790697674,
+      "grad_norm": 0.12145771086215973,
+      "learning_rate": 9.384987302525439e-07,
+      "loss": 0.1972,
+      "step": 10230
+    },
+    {
+      "epoch": 2.4675262079768645,
+      "grad_norm": 0.15090271830558777,
+      "learning_rate": 9.303384400822019e-07,
+      "loss": 0.1972,
+      "step": 10240
+    },
+    {
+      "epoch": 2.469936136883962,
+      "grad_norm": 0.13022595643997192,
+      "learning_rate": 9.222101413698475e-07,
+      "loss": 0.1994,
+      "step": 10250
+    },
+    {
+      "epoch": 2.472346065791059,
+      "grad_norm": 0.11662918329238892,
+      "learning_rate": 9.141138980117348e-07,
+      "loss": 0.1963,
+      "step": 10260
+    },
+    {
+      "epoch": 2.4747559946981563,
+      "grad_norm": 0.1264001727104187,
+      "learning_rate": 9.060497736521312e-07,
+      "loss": 0.1996,
+      "step": 10270
+    },
+    {
+      "epoch": 2.477165923605254,
+      "grad_norm": 0.12486141920089722,
+      "learning_rate": 8.980178316828158e-07,
+      "loss": 0.1961,
+      "step": 10280
+    },
+    {
+      "epoch": 2.479575852512351,
+      "grad_norm": 0.12094808369874954,
+      "learning_rate": 8.900181352425907e-07,
+      "loss": 0.1987,
+      "step": 10290
+    },
+    {
+      "epoch": 2.481985781419448,
+      "grad_norm": 0.11713796108961105,
+      "learning_rate": 8.82050747216766e-07,
+      "loss": 0.1967,
+      "step": 10300
+    },
+    {
+      "epoch": 2.4843957103265453,
+      "grad_norm": 0.11617062240839005,
+      "learning_rate": 8.741157302366859e-07,
+      "loss": 0.1968,
+      "step": 10310
+    },
+    {
+      "epoch": 2.4868056392336424,
+      "grad_norm": 0.14669108390808105,
+      "learning_rate": 8.662131466792217e-07,
+      "loss": 0.1975,
+      "step": 10320
+    },
+    {
+      "epoch": 2.48921556814074,
+      "grad_norm": 0.11834236234426498,
+      "learning_rate": 8.5834305866629e-07,
+      "loss": 0.1996,
+      "step": 10330
+    },
+    {
+      "epoch": 2.491625497047837,
+      "grad_norm": 0.11807161569595337,
+      "learning_rate": 8.505055280643582e-07,
+      "loss": 0.1959,
+      "step": 10340
+    },
+    {
+      "epoch": 2.494035425954934,
+      "grad_norm": 0.12231793254613876,
+      "learning_rate": 8.42700616483963e-07,
+      "loss": 0.1976,
+      "step": 10350
+    },
+    {
+      "epoch": 2.496445354862032,
+      "grad_norm": 0.12237108498811722,
+      "learning_rate": 8.34928385279224e-07,
+      "loss": 0.1982,
+      "step": 10360
+    },
+    {
+      "epoch": 2.498855283769129,
+      "grad_norm": 0.11896923184394836,
+      "learning_rate": 8.271888955473606e-07,
+      "loss": 0.1985,
+      "step": 10370
+    },
+    {
+      "epoch": 2.501265212676226,
+      "grad_norm": 0.1200416311621666,
+      "learning_rate": 8.194822081282144e-07,
+      "loss": 0.1995,
+      "step": 10380
+    },
+    {
+      "epoch": 2.503675141583323,
+      "grad_norm": 0.1195942834019661,
+      "learning_rate": 8.118083836037677e-07,
+      "loss": 0.198,
+      "step": 10390
+    },
+    {
+      "epoch": 2.5060850704904203,
+      "grad_norm": 0.12644240260124207,
+      "learning_rate": 8.041674822976686e-07,
+      "loss": 0.1988,
+      "step": 10400
+    },
+    {
+      "epoch": 2.508494999397518,
+      "grad_norm": 0.13783343136310577,
+      "learning_rate": 7.965595642747593e-07,
+      "loss": 0.2012,
+      "step": 10410
+    },
+    {
+      "epoch": 2.510904928304615,
+      "grad_norm": 0.11760956048965454,
+      "learning_rate": 7.889846893405978e-07,
+      "loss": 0.1968,
+      "step": 10420
+    },
+    {
+      "epoch": 2.513314857211712,
+      "grad_norm": 0.11873472481966019,
+      "learning_rate": 7.814429170409965e-07,
+      "loss": 0.1992,
+      "step": 10430
+    },
+    {
+      "epoch": 2.5157247861188097,
+      "grad_norm": 0.12854446470737457,
+      "learning_rate": 7.739343066615457e-07,
+      "loss": 0.1968,
+      "step": 10440
+    },
+    {
+      "epoch": 2.518134715025907,
+      "grad_norm": 0.12183453142642975,
+      "learning_rate": 7.664589172271519e-07,
+      "loss": 0.1978,
+      "step": 10450
+    },
+    {
+      "epoch": 2.520544643933004,
+      "grad_norm": 0.12129750847816467,
+      "learning_rate": 7.590168075015725e-07,
+      "loss": 0.1967,
+      "step": 10460
+    },
+    {
+      "epoch": 2.522954572840101,
+      "grad_norm": 0.12526491284370422,
+      "learning_rate": 7.51608035986956e-07,
+      "loss": 0.1966,
+      "step": 10470
+    },
+    {
+      "epoch": 2.5253645017471986,
+      "grad_norm": 0.12419699132442474,
+      "learning_rate": 7.442326609233786e-07,
+      "loss": 0.1978,
+      "step": 10480
+    },
+    {
+      "epoch": 2.5277744306542957,
+      "grad_norm": 0.11712777614593506,
+      "learning_rate": 7.368907402883896e-07,
+      "loss": 0.1982,
+      "step": 10490
+    },
+    {
+      "epoch": 2.530184359561393,
+      "grad_norm": 0.11896039545536041,
+      "learning_rate": 7.295823317965533e-07,
+      "loss": 0.2023,
+      "step": 10500
+    },
+    {
+      "epoch": 2.5325942884684904,
+      "grad_norm": 0.12150632590055466,
+      "learning_rate": 7.223074928989971e-07,
+      "loss": 0.1982,
+      "step": 10510
+    },
+    {
+      "epoch": 2.5350042173755876,
+      "grad_norm": 0.11829578876495361,
+      "learning_rate": 7.150662807829584e-07,
+      "loss": 0.1989,
+      "step": 10520
+    },
+    {
+      "epoch": 2.5374141462826847,
+      "grad_norm": 0.12783962488174438,
+      "learning_rate": 7.07858752371336e-07,
+      "loss": 0.1988,
+      "step": 10530
+    },
+    {
+      "epoch": 2.539824075189782,
+      "grad_norm": 0.11983250826597214,
+      "learning_rate": 7.006849643222425e-07,
+      "loss": 0.1954,
+      "step": 10540
+    },
+    {
+      "epoch": 2.542234004096879,
+      "grad_norm": 0.12253327667713165,
+      "learning_rate": 6.935449730285576e-07,
+      "loss": 0.1995,
+      "step": 10550
+    },
+    {
+      "epoch": 2.5446439330039765,
+      "grad_norm": 0.12103965878486633,
+      "learning_rate": 6.864388346174899e-07,
+      "loss": 0.1964,
+      "step": 10560
+    },
+    {
+      "epoch": 2.5470538619110736,
+      "grad_norm": 0.12630999088287354,
+      "learning_rate": 6.793666049501252e-07,
+      "loss": 0.1984,
+      "step": 10570
+    },
+    {
+      "epoch": 2.5494637908181708,
+      "grad_norm": 0.11937756091356277,
+      "learning_rate": 6.723283396210006e-07,
+      "loss": 0.199,
+      "step": 10580
+    },
+    {
+      "epoch": 2.5518737197252683,
+      "grad_norm": 0.12514546513557434,
+      "learning_rate": 6.65324093957656e-07,
+      "loss": 0.1992,
+      "step": 10590
+    },
+    {
+      "epoch": 2.5542836486323655,
+      "grad_norm": 0.1185721680521965,
+      "learning_rate": 6.583539230202062e-07,
+      "loss": 0.1988,
+      "step": 10600
+    },
+    {
+      "epoch": 2.5566935775394626,
+      "grad_norm": 0.15600068867206573,
+      "learning_rate": 6.514178816009059e-07,
+      "loss": 0.1981,
+      "step": 10610
+    },
+    {
+      "epoch": 2.5591035064465597,
+      "grad_norm": 0.11238270252943039,
+      "learning_rate": 6.445160242237181e-07,
+      "loss": 0.1963,
+      "step": 10620
+    },
+    {
+      "epoch": 2.561513435353657,
+      "grad_norm": 0.1281278282403946,
+      "learning_rate": 6.376484051438864e-07,
+      "loss": 0.1964,
+      "step": 10630
+    },
+    {
+      "epoch": 2.5639233642607544,
+      "grad_norm": 0.12617678940296173,
+      "learning_rate": 6.308150783475086e-07,
+      "loss": 0.2001,
+      "step": 10640
+    },
+    {
+      "epoch": 2.5663332931678515,
+      "grad_norm": 0.13613717257976532,
+      "learning_rate": 6.240160975511117e-07,
+      "loss": 0.197,
+      "step": 10650
+    },
+    {
+      "epoch": 2.5687432220749487,
+      "grad_norm": 0.1211361438035965,
+      "learning_rate": 6.172515162012332e-07,
+      "loss": 0.1978,
+      "step": 10660
+    },
+    {
+      "epoch": 2.571153150982046,
+      "grad_norm": 0.11443440616130829,
+      "learning_rate": 6.105213874739913e-07,
+      "loss": 0.201,
+      "step": 10670
+    },
+    {
+      "epoch": 2.5735630798891433,
+      "grad_norm": 0.1274467259645462,
+      "learning_rate": 6.038257642746815e-07,
+      "loss": 0.198,
+      "step": 10680
+    },
+    {
+      "epoch": 2.5759730087962405,
+      "grad_norm": 0.11638859659433365,
+      "learning_rate": 5.971646992373442e-07,
+      "loss": 0.2009,
+      "step": 10690
+    },
+    {
+      "epoch": 2.5783829377033376,
+      "grad_norm": 0.12000516802072525,
+      "learning_rate": 5.905382447243679e-07,
+      "loss": 0.1972,
+      "step": 10700
+    },
+    {
+      "epoch": 2.5807928666104347,
+      "grad_norm": 0.11631269007921219,
+      "learning_rate": 5.839464528260602e-07,
+      "loss": 0.1953,
+      "step": 10710
+    },
+    {
+      "epoch": 2.5832027955175323,
+      "grad_norm": 0.12172136455774307,
+      "learning_rate": 5.773893753602556e-07,
+      "loss": 0.1983,
+      "step": 10720
+    },
+    {
+      "epoch": 2.5856127244246294,
+      "grad_norm": 0.12186679244041443,
+      "learning_rate": 5.708670638718944e-07,
+      "loss": 0.1975,
+      "step": 10730
+    },
+    {
+      "epoch": 2.588022653331727,
+      "grad_norm": 0.11384265124797821,
+      "learning_rate": 5.643795696326248e-07,
+      "loss": 0.1989,
+      "step": 10740
+    },
+    {
+      "epoch": 2.590432582238824,
+      "grad_norm": 0.12156783789396286,
+      "learning_rate": 5.579269436403967e-07,
+      "loss": 0.2025,
+      "step": 10750
+    },
+    {
+      "epoch": 2.5928425111459212,
+      "grad_norm": 0.11662750691175461,
+      "learning_rate": 5.515092366190633e-07,
+      "loss": 0.1983,
+      "step": 10760
+    },
+    {
+      "epoch": 2.5952524400530184,
+      "grad_norm": 0.1321050077676773,
+      "learning_rate": 5.451264990179806e-07,
+      "loss": 0.1995,
+      "step": 10770
+    },
+    {
+      "epoch": 2.5976623689601155,
+      "grad_norm": 0.11678937822580338,
+      "learning_rate": 5.387787810116107e-07,
+      "loss": 0.1971,
+      "step": 10780
+    },
+    {
+      "epoch": 2.600072297867213,
+      "grad_norm": 0.12364859879016876,
+      "learning_rate": 5.324661324991287e-07,
+      "loss": 0.1974,
+      "step": 10790
+    },
+    {
+      "epoch": 2.60248222677431,
+      "grad_norm": 0.11932826787233353,
+      "learning_rate": 5.261886031040297e-07,
+      "loss": 0.1958,
+      "step": 10800
+    },
+    {
+      "epoch": 2.6048921556814073,
+      "grad_norm": 0.11405885964632034,
+      "learning_rate": 5.199462421737378e-07,
+      "loss": 0.1979,
+      "step": 10810
+    },
+    {
+      "epoch": 2.607302084588505,
+      "grad_norm": 0.11789312213659286,
+      "learning_rate": 5.137390987792224e-07,
+      "loss": 0.1979,
+      "step": 10820
+    },
+    {
+      "epoch": 2.609712013495602,
+      "grad_norm": 0.11750980466604233,
+      "learning_rate": 5.075672217146021e-07,
+      "loss": 0.1968,
+      "step": 10830
+    },
+    {
+      "epoch": 2.612121942402699,
+      "grad_norm": 0.1324010044336319,
+      "learning_rate": 5.014306594967777e-07,
+      "loss": 0.1968,
+      "step": 10840
+    },
+    {
+      "epoch": 2.6145318713097963,
+      "grad_norm": 0.12013759464025497,
+      "learning_rate": 4.953294603650321e-07,
+      "loss": 0.1967,
+      "step": 10850
+    },
+    {
+      "epoch": 2.6169418002168934,
+      "grad_norm": 0.11585763096809387,
+      "learning_rate": 4.892636722806681e-07,
+      "loss": 0.199,
+      "step": 10860
+    },
+    {
+      "epoch": 2.619351729123991,
+      "grad_norm": 0.11899024248123169,
+      "learning_rate": 4.832333429266162e-07,
+      "loss": 0.1985,
+      "step": 10870
+    },
+    {
+      "epoch": 2.621761658031088,
+      "grad_norm": 0.12400858104228973,
+      "learning_rate": 4.772385197070734e-07,
+      "loss": 0.2,
+      "step": 10880
+    },
+    {
+      "epoch": 2.624171586938185,
+      "grad_norm": 0.12234139442443848,
+      "learning_rate": 4.712792497471219e-07,
+      "loss": 0.1991,
+      "step": 10890
+    },
+    {
+      "epoch": 2.6265815158452828,
+      "grad_norm": 0.11881452798843384,
+      "learning_rate": 4.653555798923598e-07,
+      "loss": 0.1969,
+      "step": 10900
+    },
+    {
+      "epoch": 2.62899144475238,
+      "grad_norm": 0.11506690829992294,
+      "learning_rate": 4.59467556708536e-07,
+      "loss": 0.1974,
+      "step": 10910
+    },
+    {
+      "epoch": 2.631401373659477,
+      "grad_norm": 0.12822450697422028,
+      "learning_rate": 4.5361522648118163e-07,
+      "loss": 0.1993,
+      "step": 10920
+    },
+    {
+      "epoch": 2.633811302566574,
+      "grad_norm": 0.12246552854776382,
+      "learning_rate": 4.477986352152458e-07,
+      "loss": 0.1993,
+      "step": 10930
+    },
+    {
+      "epoch": 2.6362212314736713,
+      "grad_norm": 0.12038519978523254,
+      "learning_rate": 4.420178286347365e-07,
+      "loss": 0.1966,
+      "step": 10940
+    },
+    {
+      "epoch": 2.638631160380769,
+      "grad_norm": 0.12032466381788254,
+      "learning_rate": 4.3627285218235836e-07,
+      "loss": 0.1977,
+      "step": 10950
+    },
+    {
+      "epoch": 2.641041089287866,
+      "grad_norm": 0.12104138731956482,
+      "learning_rate": 4.305637510191596e-07,
+      "loss": 0.2005,
+      "step": 10960
+    },
+    {
+      "epoch": 2.643451018194963,
+      "grad_norm": 0.11752533912658691,
+      "learning_rate": 4.248905700241679e-07,
+      "loss": 0.1996,
+      "step": 10970
+    },
+    {
+      "epoch": 2.6458609471020607,
+      "grad_norm": 0.12105824798345566,
+      "learning_rate": 4.192533537940524e-07,
+      "loss": 0.1982,
+      "step": 10980
+    },
+    {
+      "epoch": 2.648270876009158,
+      "grad_norm": 0.11677469313144684,
+      "learning_rate": 4.1365214664275624e-07,
+      "loss": 0.1994,
+      "step": 10990
+    },
+    {
+      "epoch": 2.650680804916255,
+      "grad_norm": 0.1191301941871643,
+      "learning_rate": 4.0808699260116267e-07,
+      "loss": 0.1999,
+      "step": 11000
+    },
+    {
+      "epoch": 2.653090733823352,
+      "grad_norm": 0.11744995415210724,
+      "learning_rate": 4.025579354167386e-07,
+      "loss": 0.1979,
+      "step": 11010
+    },
+    {
+      "epoch": 2.6555006627304496,
+      "grad_norm": 0.11751815676689148,
+      "learning_rate": 3.9706501855319767e-07,
+      "loss": 0.1977,
+      "step": 11020
+    },
+    {
+      "epoch": 2.6579105916375467,
+      "grad_norm": 0.11772134900093079,
+      "learning_rate": 3.9160828519015537e-07,
+      "loss": 0.1966,
+      "step": 11030
+    },
+    {
+      "epoch": 2.660320520544644,
+      "grad_norm": 0.12891718745231628,
+      "learning_rate": 3.8618777822278854e-07,
+      "loss": 0.2011,
+      "step": 11040
+    },
+    {
+      "epoch": 2.6627304494517414,
+      "grad_norm": 0.11987189948558807,
+      "learning_rate": 3.8080354026150067e-07,
+      "loss": 0.1969,
+      "step": 11050
+    },
+    {
+      "epoch": 2.6651403783588385,
+      "grad_norm": 0.12312425673007965,
+      "learning_rate": 3.754556136315862e-07,
+      "loss": 0.1974,
+      "step": 11060
+    },
+    {
+      "epoch": 2.6675503072659357,
+      "grad_norm": 0.11463374644517899,
+      "learning_rate": 3.701440403728973e-07,
+      "loss": 0.1966,
+      "step": 11070
+    },
+    {
+      "epoch": 2.669960236173033,
+      "grad_norm": 0.11283261328935623,
+      "learning_rate": 3.6486886223951356e-07,
+      "loss": 0.1964,
+      "step": 11080
+    },
+    {
+      "epoch": 2.67237016508013,
+      "grad_norm": 0.11768094450235367,
+      "learning_rate": 3.596301206994135e-07,
+      "loss": 0.1954,
+      "step": 11090
+    },
+    {
+      "epoch": 2.6747800939872275,
+      "grad_norm": 0.1149783730506897,
+      "learning_rate": 3.5442785693414916e-07,
+      "loss": 0.1971,
+      "step": 11100
+    },
+    {
+      "epoch": 2.6771900228943246,
+      "grad_norm": 0.12526769936084747,
+      "learning_rate": 3.4926211183852257e-07,
+      "loss": 0.1952,
+      "step": 11110
+    },
+    {
+      "epoch": 2.6795999518014217,
+      "grad_norm": 0.12192758172750473,
+      "learning_rate": 3.441329260202647e-07,
+      "loss": 0.1952,
+      "step": 11120
+    },
+    {
+      "epoch": 2.6820098807085193,
+      "grad_norm": 0.11524799466133118,
+      "learning_rate": 3.390403397997116e-07,
+      "loss": 0.1994,
+      "step": 11130
+    },
+    {
+      "epoch": 2.6844198096156164,
+      "grad_norm": 0.11889517307281494,
+      "learning_rate": 3.339843932094977e-07,
+      "loss": 0.1972,
+      "step": 11140
+    },
+    {
+      "epoch": 2.6868297385227136,
+      "grad_norm": 0.11067784577608109,
+      "learning_rate": 3.289651259942267e-07,
+      "loss": 0.1944,
+      "step": 11150
+    },
+    {
+      "epoch": 2.6892396674298107,
+      "grad_norm": 0.12659992277622223,
+      "learning_rate": 3.2398257761017516e-07,
+      "loss": 0.1946,
+      "step": 11160
+    },
+    {
+      "epoch": 2.691649596336908,
+      "grad_norm": 0.11584974825382233,
+      "learning_rate": 3.190367872249672e-07,
+      "loss": 0.1983,
+      "step": 11170
+    },
+    {
+      "epoch": 2.6940595252440054,
+      "grad_norm": 0.12076602131128311,
+      "learning_rate": 3.1412779371727873e-07,
+      "loss": 0.195,
+      "step": 11180
+    },
+    {
+      "epoch": 2.6964694541511025,
+      "grad_norm": 0.12034130841493607,
+      "learning_rate": 3.0925563567652474e-07,
+      "loss": 0.1986,
+      "step": 11190
+    },
+    {
+      "epoch": 2.6988793830581996,
+      "grad_norm": 0.11660711467266083,
+      "learning_rate": 3.044203514025579e-07,
+      "loss": 0.1965,
+      "step": 11200
+    },
+    {
+      "epoch": 2.701289311965297,
+      "grad_norm": 0.12302997708320618,
+      "learning_rate": 2.996219789053678e-07,
+      "loss": 0.1977,
+      "step": 11210
+    },
+    {
+      "epoch": 2.7036992408723943,
+      "grad_norm": 0.11597088724374771,
+      "learning_rate": 2.948605559047818e-07,
+      "loss": 0.1991,
+      "step": 11220
+    },
+    {
+      "epoch": 2.7061091697794915,
+      "grad_norm": 0.11687976866960526,
+      "learning_rate": 2.9013611983016887e-07,
+      "loss": 0.1979,
+      "step": 11230
+    },
+    {
+      "epoch": 2.7085190986865886,
+      "grad_norm": 0.12074779719114304,
+      "learning_rate": 2.8544870782014566e-07,
+      "loss": 0.1994,
+      "step": 11240
+    },
+    {
+      "epoch": 2.710929027593686,
+      "grad_norm": 0.1221262738108635,
+      "learning_rate": 2.807983567222822e-07,
+      "loss": 0.1972,
+      "step": 11250
+    },
+    {
+      "epoch": 2.7133389565007833,
+      "grad_norm": 0.11421184986829758,
+      "learning_rate": 2.7618510309281756e-07,
+      "loss": 0.1979,
+      "step": 11260
+    },
+    {
+      "epoch": 2.7157488854078804,
+      "grad_norm": 0.12446928024291992,
+      "learning_rate": 2.716089831963636e-07,
+      "loss": 0.1964,
+      "step": 11270
+    },
+    {
+      "epoch": 2.718158814314978,
+      "grad_norm": 0.11847683787345886,
+      "learning_rate": 2.6707003300563196e-07,
+      "loss": 0.2021,
+      "step": 11280
+    },
+    {
+      "epoch": 2.720568743222075,
+      "grad_norm": 0.11553952842950821,
+      "learning_rate": 2.6256828820113765e-07,
+      "loss": 0.1975,
+      "step": 11290
+    },
+    {
+      "epoch": 2.722978672129172,
+      "grad_norm": 0.12082428485155106,
+      "learning_rate": 2.581037841709322e-07,
+      "loss": 0.1974,
+      "step": 11300
+    },
+    {
+      "epoch": 2.7253886010362693,
+      "grad_norm": 0.11691971868276596,
+      "learning_rate": 2.536765560103122e-07,
+      "loss": 0.1984,
+      "step": 11310
+    },
+    {
+      "epoch": 2.7277985299433665,
+      "grad_norm": 0.11397454142570496,
+      "learning_rate": 2.492866385215559e-07,
+      "loss": 0.198,
+      "step": 11320
+    },
+    {
+      "epoch": 2.730208458850464,
+      "grad_norm": 0.11686151474714279,
+      "learning_rate": 2.449340662136407e-07,
+      "loss": 0.195,
+      "step": 11330
+    },
+    {
+      "epoch": 2.732618387757561,
+      "grad_norm": 0.11599799245595932,
+      "learning_rate": 2.4061887330197485e-07,
+      "loss": 0.1971,
+      "step": 11340
+    },
+    {
+      "epoch": 2.7350283166646583,
+      "grad_norm": 0.12791751325130463,
+      "learning_rate": 2.3634109370813008e-07,
+      "loss": 0.1985,
+      "step": 11350
+    },
+    {
+      "epoch": 2.737438245571756,
+      "grad_norm": 0.1324101835489273,
+      "learning_rate": 2.3210076105957103e-07,
+      "loss": 0.1998,
+      "step": 11360
+    },
+    {
+      "epoch": 2.739848174478853,
+      "grad_norm": 0.11545724421739578,
+      "learning_rate": 2.278979086893962e-07,
+      "loss": 0.2001,
+      "step": 11370
+    },
+    {
+      "epoch": 2.74225810338595,
+      "grad_norm": 0.11950286477804184,
+      "learning_rate": 2.2373256963607093e-07,
+      "loss": 0.1986,
+      "step": 11380
+    },
+    {
+      "epoch": 2.7446680322930472,
+      "grad_norm": 0.11261560767889023,
+      "learning_rate": 2.1960477664317027e-07,
+      "loss": 0.1991,
+      "step": 11390
+    },
+    {
+      "epoch": 2.7470779612001444,
+      "grad_norm": 0.11667031794786453,
+      "learning_rate": 2.1551456215912147e-07,
+      "loss": 0.1969,
+      "step": 11400
+    },
+    {
+      "epoch": 2.749487890107242,
+      "grad_norm": 0.11386009305715561,
+      "learning_rate": 2.114619583369476e-07,
+      "loss": 0.1995,
+      "step": 11410
+    },
+    {
+      "epoch": 2.751897819014339,
+      "grad_norm": 0.11694635450839996,
+      "learning_rate": 2.0744699703401817e-07,
+      "loss": 0.1954,
+      "step": 11420
+    },
+    {
+      "epoch": 2.754307747921436,
+      "grad_norm": 0.12221164256334305,
+      "learning_rate": 2.034697098117927e-07,
+      "loss": 0.1968,
+      "step": 11430
+    },
+    {
+      "epoch": 2.7567176768285337,
+      "grad_norm": 0.12051833420991898,
+      "learning_rate": 1.995301279355788e-07,
+      "loss": 0.1967,
+      "step": 11440
+    },
+    {
+      "epoch": 2.759127605735631,
+      "grad_norm": 0.1132727712392807,
+      "learning_rate": 1.9562828237428332e-07,
+      "loss": 0.1992,
+      "step": 11450
+    },
+    {
+      "epoch": 2.761537534642728,
+      "grad_norm": 0.11581676453351974,
+      "learning_rate": 1.917642038001677e-07,
+      "loss": 0.201,
+      "step": 11460
+    },
+    {
+      "epoch": 2.763947463549825,
+      "grad_norm": 0.11355219036340714,
+      "learning_rate": 1.8793792258861077e-07,
+      "loss": 0.1972,
+      "step": 11470
+    },
+    {
+      "epoch": 2.7663573924569222,
+      "grad_norm": 0.11676806956529617,
+      "learning_rate": 1.8414946881786634e-07,
+      "loss": 0.1995,
+      "step": 11480
+    },
+    {
+      "epoch": 2.76876732136402,
+      "grad_norm": 0.11301768571138382,
+      "learning_rate": 1.8039887226882823e-07,
+      "loss": 0.1964,
+      "step": 11490
+    },
+    {
+      "epoch": 2.771177250271117,
+      "grad_norm": 0.11706644296646118,
+      "learning_rate": 1.7668616242479618e-07,
+      "loss": 0.1995,
+      "step": 11500
+    },
+    {
+      "epoch": 2.7735871791782145,
+      "grad_norm": 0.1184009313583374,
+      "learning_rate": 1.7301136847124477e-07,
+      "loss": 0.1967,
+      "step": 11510
+    },
+    {
+      "epoch": 2.7759971080853116,
+      "grad_norm": 0.1254088580608368,
+      "learning_rate": 1.6937451929559147e-07,
+      "loss": 0.1985,
+      "step": 11520
+    },
+    {
+      "epoch": 2.7784070369924088,
+      "grad_norm": 0.1315934807062149,
+      "learning_rate": 1.6577564348697284e-07,
+      "loss": 0.1989,
+      "step": 11530
+    },
+    {
+      "epoch": 2.780816965899506,
+      "grad_norm": 0.11553444713354111,
+      "learning_rate": 1.622147693360171e-07,
+      "loss": 0.1979,
+      "step": 11540
+    },
+    {
+      "epoch": 2.783226894806603,
+      "grad_norm": 0.11310292780399323,
+      "learning_rate": 1.586919248346236e-07,
+      "loss": 0.1986,
+      "step": 11550
+    },
+    {
+      "epoch": 2.7856368237137006,
+      "grad_norm": 0.11396036297082901,
+      "learning_rate": 1.5520713767574247e-07,
+      "loss": 0.1987,
+      "step": 11560
+    },
+    {
+      "epoch": 2.7880467526207977,
+      "grad_norm": 0.11298102885484695,
+      "learning_rate": 1.5176043525315543e-07,
+      "loss": 0.1986,
+      "step": 11570
+    },
+    {
+      "epoch": 2.790456681527895,
+      "grad_norm": 0.12029560655355453,
+      "learning_rate": 1.483518446612614e-07,
+      "loss": 0.1967,
+      "step": 11580
+    },
+    {
+      "epoch": 2.7928666104349924,
+      "grad_norm": 0.11798765510320663,
+      "learning_rate": 1.4498139269486455e-07,
+      "loss": 0.1996,
+      "step": 11590
+    },
+    {
+      "epoch": 2.7952765393420895,
+      "grad_norm": 0.1185227632522583,
+      "learning_rate": 1.4164910584896163e-07,
+      "loss": 0.1979,
+      "step": 11600
+    },
+    {
+      "epoch": 2.7976864682491867,
+      "grad_norm": 0.11194013059139252,
+      "learning_rate": 1.383550103185366e-07,
+      "loss": 0.1962,
+      "step": 11610
+    },
+    {
+      "epoch": 2.800096397156284,
+      "grad_norm": 0.11631465703248978,
+      "learning_rate": 1.350991319983508e-07,
+      "loss": 0.1984,
+      "step": 11620
+    },
+    {
+      "epoch": 2.802506326063381,
+      "grad_norm": 0.1182403415441513,
+      "learning_rate": 1.3188149648274307e-07,
+      "loss": 0.1993,
+      "step": 11630
+    },
+    {
+      "epoch": 2.8049162549704785,
+      "grad_norm": 0.15543074905872345,
+      "learning_rate": 1.2870212906542612e-07,
+      "loss": 0.1971,
+      "step": 11640
+    },
+    {
+      "epoch": 2.8073261838775756,
+      "grad_norm": 0.12442866712808609,
+      "learning_rate": 1.2556105473928824e-07,
+      "loss": 0.1983,
+      "step": 11650
+    },
+    {
+      "epoch": 2.8097361127846727,
+      "grad_norm": 0.11325249820947647,
+      "learning_rate": 1.2245829819619858e-07,
+      "loss": 0.1985,
+      "step": 11660
+    },
+    {
+      "epoch": 2.8121460416917703,
+      "grad_norm": 0.11899420619010925,
+      "learning_rate": 1.1939388382681106e-07,
+      "loss": 0.1988,
+      "step": 11670
+    },
+    {
+      "epoch": 2.8145559705988674,
+      "grad_norm": 0.12117994576692581,
+      "learning_rate": 1.163678357203718e-07,
+      "loss": 0.1989,
+      "step": 11680
+    },
+    {
+      "epoch": 2.8169658995059645,
+      "grad_norm": 0.13569891452789307,
+      "learning_rate": 1.133801776645338e-07,
+      "loss": 0.1995,
+      "step": 11690
+    },
+    {
+      "epoch": 2.8193758284130617,
+      "grad_norm": 0.11266092211008072,
+      "learning_rate": 1.1043093314516418e-07,
+      "loss": 0.197,
+      "step": 11700
+    },
+    {
+      "epoch": 2.821785757320159,
+      "grad_norm": 0.11539432406425476,
+      "learning_rate": 1.0752012534616496e-07,
+      "loss": 0.1973,
+      "step": 11710
+    },
+    {
+      "epoch": 2.8241956862272564,
+      "grad_norm": 0.11695801466703415,
+      "learning_rate": 1.046477771492882e-07,
+      "loss": 0.1965,
+      "step": 11720
+    },
+    {
+      "epoch": 2.8266056151343535,
+      "grad_norm": 0.1182837262749672,
+      "learning_rate": 1.0181391113395611e-07,
+      "loss": 0.1992,
+      "step": 11730
+    },
+    {
+      "epoch": 2.8290155440414506,
+      "grad_norm": 0.11519190669059753,
+      "learning_rate": 9.901854957708345e-08,
+      "loss": 0.1979,
+      "step": 11740
+    },
+    {
+      "epoch": 2.831425472948548,
+      "grad_norm": 0.11443960666656494,
+      "learning_rate": 9.626171445290378e-08,
+      "loss": 0.1957,
+      "step": 11750
+    },
+    {
+      "epoch": 2.8338354018556453,
+      "grad_norm": 0.11359517276287079,
+      "learning_rate": 9.354342743279455e-08,
+      "loss": 0.2002,
+      "step": 11760
+    },
+    {
+      "epoch": 2.8362453307627424,
+      "grad_norm": 0.11192893981933594,
+      "learning_rate": 9.086370988511006e-08,
+      "loss": 0.1969,
+      "step": 11770
+    },
+    {
+      "epoch": 2.8386552596698396,
+      "grad_norm": 0.11451409757137299,
+      "learning_rate": 8.822258287500829e-08,
+      "loss": 0.1983,
+      "step": 11780
+    },
+    {
+      "epoch": 2.841065188576937,
+      "grad_norm": 0.11336790770292282,
+      "learning_rate": 8.562006716429316e-08,
+      "loss": 0.1984,
+      "step": 11790
+    },
+    {
+      "epoch": 2.8434751174840343,
+      "grad_norm": 0.11402434855699539,
+      "learning_rate": 8.305618321124087e-08,
+      "loss": 0.1973,
+      "step": 11800
+    },
+    {
+      "epoch": 2.8458850463911314,
+      "grad_norm": 0.11499971151351929,
+      "learning_rate": 8.053095117044995e-08,
+      "loss": 0.1989,
+      "step": 11810
+    },
+    {
+      "epoch": 2.848294975298229,
+      "grad_norm": 0.11966579407453537,
+      "learning_rate": 7.804439089267368e-08,
+      "loss": 0.2011,
+      "step": 11820
+    },
+    {
+      "epoch": 2.850704904205326,
+      "grad_norm": 0.12160076946020126,
+      "learning_rate": 7.559652192467127e-08,
+      "loss": 0.1983,
+      "step": 11830
+    },
+    {
+      "epoch": 2.853114833112423,
+      "grad_norm": 0.11282498389482498,
+      "learning_rate": 7.318736350904798e-08,
+      "loss": 0.1983,
+      "step": 11840
+    },
+    {
+      "epoch": 2.8555247620195203,
+      "grad_norm": 0.1345667690038681,
+      "learning_rate": 7.081693458410977e-08,
+      "loss": 0.1996,
+      "step": 11850
+    },
+    {
+      "epoch": 2.8579346909266174,
+      "grad_norm": 0.11672775447368622,
+      "learning_rate": 6.848525378370995e-08,
+      "loss": 0.1964,
+      "step": 11860
+    },
+    {
+      "epoch": 2.860344619833715,
+      "grad_norm": 0.11396300047636032,
+      "learning_rate": 6.61923394371039e-08,
+      "loss": 0.1973,
+      "step": 11870
+    },
+    {
+      "epoch": 2.862754548740812,
+      "grad_norm": 0.1159951463341713,
+      "learning_rate": 6.393820956880681e-08,
+      "loss": 0.1979,
+      "step": 11880
+    },
+    {
+      "epoch": 2.8651644776479093,
+      "grad_norm": 0.11563079059123993,
+      "learning_rate": 6.172288189844833e-08,
+      "loss": 0.1992,
+      "step": 11890
+    },
+    {
+      "epoch": 2.867574406555007,
+      "grad_norm": 0.1158207431435585,
+      "learning_rate": 5.954637384063766e-08,
+      "loss": 0.2007,
+      "step": 11900
+    },
+    {
+      "epoch": 2.869984335462104,
+      "grad_norm": 0.12424372881650925,
+      "learning_rate": 5.740870250482367e-08,
+      "loss": 0.1985,
+      "step": 11910
+    },
+    {
+      "epoch": 2.872394264369201,
+      "grad_norm": 0.11300027370452881,
+      "learning_rate": 5.530988469516052e-08,
+      "loss": 0.199,
+      "step": 11920
+    },
+    {
+      "epoch": 2.874804193276298,
+      "grad_norm": 0.11593002080917358,
+      "learning_rate": 5.324993691037783e-08,
+      "loss": 0.1977,
+      "step": 11930
+    },
+    {
+      "epoch": 2.8772141221833953,
+      "grad_norm": 0.11310574412345886,
+      "learning_rate": 5.12288753436474e-08,
+      "loss": 0.1971,
+      "step": 11940
+    },
+    {
+      "epoch": 2.879624051090493,
+      "grad_norm": 0.11324238777160645,
+      "learning_rate": 4.924671588246e-08,
+      "loss": 0.2001,
+      "step": 11950
+    },
+    {
+      "epoch": 2.88203397999759,
+      "grad_norm": 0.11741005629301071,
+      "learning_rate": 4.7303474108496e-08,
+      "loss": 0.1979,
+      "step": 11960
+    },
+    {
+      "epoch": 2.884443908904687,
+      "grad_norm": 0.11062411963939667,
+      "learning_rate": 4.539916529750832e-08,
+      "loss": 0.1936,
+      "step": 11970
+    },
+    {
+      "epoch": 2.8868538378117847,
+      "grad_norm": 0.12052568048238754,
+      "learning_rate": 4.353380441919575e-08,
+      "loss": 0.196,
+      "step": 11980
+    },
+    {
+      "epoch": 2.889263766718882,
+      "grad_norm": 0.11128173768520355,
+      "learning_rate": 4.170740613709201e-08,
+      "loss": 0.1982,
+      "step": 11990
+    },
+    {
+      "epoch": 2.891673695625979,
+      "grad_norm": 0.1159716323018074,
+      "learning_rate": 3.9919984808445836e-08,
+      "loss": 0.1975,
+      "step": 12000
+    },
+    {
+      "epoch": 2.894083624533076,
+      "grad_norm": 0.12408029288053513,
+      "learning_rate": 3.817155448410936e-08,
+      "loss": 0.1969,
+      "step": 12010
+    },
+    {
+      "epoch": 2.8964935534401737,
+      "grad_norm": 0.11627837270498276,
+      "learning_rate": 3.6462128908428265e-08,
+      "loss": 0.1982,
+      "step": 12020
+    },
+    {
+      "epoch": 2.898903482347271,
+      "grad_norm": 0.12328115850687027,
+      "learning_rate": 3.479172151913346e-08,
+      "loss": 0.1976,
+      "step": 12030
+    },
+    {
+      "epoch": 2.901313411254368,
+      "grad_norm": 0.11239629238843918,
+      "learning_rate": 3.3160345447235674e-08,
+      "loss": 0.1938,
+      "step": 12040
+    },
+    {
+      "epoch": 2.9037233401614655,
+      "grad_norm": 0.11254783719778061,
+      "learning_rate": 3.156801351692051e-08,
+      "loss": 0.1948,
+      "step": 12050
+    },
+    {
+      "epoch": 2.9061332690685626,
+      "grad_norm": 0.1124378964304924,
+      "learning_rate": 3.0014738245450756e-08,
+      "loss": 0.1987,
+      "step": 12060
+    },
+    {
+      "epoch": 2.9085431979756597,
+      "grad_norm": 0.11649011820554733,
+      "learning_rate": 2.8500531843065893e-08,
+      "loss": 0.196,
+      "step": 12070
+    },
+    {
+      "epoch": 2.910953126882757,
+      "grad_norm": 0.10951084643602371,
+      "learning_rate": 2.702540621288441e-08,
+      "loss": 0.1951,
+      "step": 12080
+    },
+    {
+      "epoch": 2.913363055789854,
+      "grad_norm": 0.11133934557437897,
+      "learning_rate": 2.5589372950815538e-08,
+      "loss": 0.197,
+      "step": 12090
+    },
+    {
+      "epoch": 2.9157729846969516,
+      "grad_norm": 0.11958010494709015,
+      "learning_rate": 2.4192443345462667e-08,
+      "loss": 0.199,
+      "step": 12100
+    },
+    {
+      "epoch": 2.9181829136040487,
+      "grad_norm": 0.11115157604217529,
+      "learning_rate": 2.2834628378037848e-08,
+      "loss": 0.1991,
+      "step": 12110
+    },
+    {
+      "epoch": 2.920592842511146,
+      "grad_norm": 0.10967998206615448,
+      "learning_rate": 2.1515938722272977e-08,
+      "loss": 0.196,
+      "step": 12120
+    },
+    {
+      "epoch": 2.9230027714182434,
+      "grad_norm": 0.1160879135131836,
+      "learning_rate": 2.023638474433931e-08,
+      "loss": 0.202,
+      "step": 12130
+    },
+    {
+      "epoch": 2.9254127003253405,
+      "grad_norm": 0.10980487614870071,
+      "learning_rate": 1.8995976502762526e-08,
+      "loss": 0.1967,
+      "step": 12140
+    },
+    {
+      "epoch": 2.9278226292324376,
+      "grad_norm": 0.11112086474895477,
+      "learning_rate": 1.779472374834612e-08,
+      "loss": 0.1977,
+      "step": 12150
+    },
+    {
+      "epoch": 2.9302325581395348,
+      "grad_norm": 0.1170421689748764,
+      "learning_rate": 1.6632635924092587e-08,
+      "loss": 0.1986,
+      "step": 12160
+    },
+    {
+      "epoch": 2.932642487046632,
+      "grad_norm": 0.11265738308429718,
+      "learning_rate": 1.5509722165131246e-08,
+      "loss": 0.1976,
+      "step": 12170
+    },
+    {
+      "epoch": 2.9350524159537295,
+      "grad_norm": 0.11651454865932465,
+      "learning_rate": 1.4425991298645525e-08,
+      "loss": 0.2013,
+      "step": 12180
+    },
+    {
+      "epoch": 2.9374623448608266,
+      "grad_norm": 0.11379573494195938,
+      "learning_rate": 1.3381451843803572e-08,
+      "loss": 0.1968,
+      "step": 12190
+    },
+    {
+      "epoch": 2.9398722737679237,
+      "grad_norm": 0.12176429480314255,
+      "learning_rate": 1.2376112011691088e-08,
+      "loss": 0.2,
+      "step": 12200
+    },
+    {
+      "epoch": 2.9422822026750213,
+      "grad_norm": 0.11730526387691498,
+      "learning_rate": 1.1409979705246932e-08,
+      "loss": 0.1987,
+      "step": 12210
+    },
+    {
+      "epoch": 2.9446921315821184,
+      "grad_norm": 0.11338816583156586,
+      "learning_rate": 1.0483062519200949e-08,
+      "loss": 0.1966,
+      "step": 12220
+    },
+    {
+      "epoch": 2.9471020604892155,
+      "grad_norm": 0.11472780257463455,
+      "learning_rate": 9.595367740014572e-09,
+      "loss": 0.1988,
+      "step": 12230
+    },
+    {
+      "epoch": 2.9495119893963127,
+      "grad_norm": 0.11212562769651413,
+      "learning_rate": 8.746902345824204e-09,
+      "loss": 0.1981,
+      "step": 12240
+    },
+    {
+      "epoch": 2.9519219183034098,
+      "grad_norm": 0.13362792134284973,
+      "learning_rate": 7.937673006384039e-09,
+      "loss": 0.1995,
+      "step": 12250
+    },
+    {
+      "epoch": 2.9543318472105073,
+      "grad_norm": 0.11798904836177826,
+      "learning_rate": 7.167686083015546e-09,
+      "loss": 0.1954,
+      "step": 12260
+    },
+    {
+      "epoch": 2.9567417761176045,
+      "grad_norm": 0.11756514012813568,
+      "learning_rate": 6.4369476285580656e-09,
+      "loss": 0.197,
+      "step": 12270
+    },
+    {
+      "epoch": 2.959151705024702,
+      "grad_norm": 0.11590098589658737,
+      "learning_rate": 5.7454633873188505e-09,
+      "loss": 0.1981,
+      "step": 12280
+    },
+    {
+      "epoch": 2.961561633931799,
+      "grad_norm": 0.11385900527238846,
+      "learning_rate": 5.09323879503032e-09,
+      "loss": 0.1999,
+      "step": 12290
+    },
+    {
+      "epoch": 2.9639715628388963,
+      "grad_norm": 0.12390254437923431,
+      "learning_rate": 4.480278978804542e-09,
+      "loss": 0.2028,
+      "step": 12300
+    },
+    {
+      "epoch": 2.9663814917459934,
+      "grad_norm": 0.11569690704345703,
+      "learning_rate": 3.906588757097152e-09,
+      "loss": 0.1962,
+      "step": 12310
+    },
+    {
+      "epoch": 2.9687914206530905,
+      "grad_norm": 0.11260288953781128,
+      "learning_rate": 3.372172639664606e-09,
+      "loss": 0.1978,
+      "step": 12320
+    },
+    {
+      "epoch": 2.971201349560188,
+      "grad_norm": 0.11671577394008636,
+      "learning_rate": 2.877034827532543e-09,
+      "loss": 0.1979,
+      "step": 12330
+    },
+    {
+      "epoch": 2.9736112784672852,
+      "grad_norm": 0.11633649468421936,
+      "learning_rate": 2.4211792129608112e-09,
+      "loss": 0.1958,
+      "step": 12340
+    },
+    {
+      "epoch": 2.9760212073743824,
+      "grad_norm": 0.11560673266649246,
+      "learning_rate": 2.004609379413491e-09,
+      "loss": 0.199,
+      "step": 12350
+    },
+    {
+      "epoch": 2.97843113628148,
+      "grad_norm": 0.14515136182308197,
+      "learning_rate": 1.6273286015305866e-09,
+      "loss": 0.1962,
+      "step": 12360
+    },
+    {
+      "epoch": 2.980841065188577,
+      "grad_norm": 0.11362209916114807,
+      "learning_rate": 1.2893398451024886e-09,
+      "loss": 0.2004,
+      "step": 12370
+    },
+    {
+      "epoch": 2.983250994095674,
+      "grad_norm": 0.11587734520435333,
+      "learning_rate": 9.906457670449953e-10,
+      "loss": 0.1933,
+      "step": 12380
+    },
+    {
+      "epoch": 2.9856609230027713,
+      "grad_norm": 0.12775199115276337,
+      "learning_rate": 7.312487153826597e-10,
+      "loss": 0.1976,
+      "step": 12390
+    },
+    {
+      "epoch": 2.9880708519098684,
+      "grad_norm": 0.11507666110992432,
+      "learning_rate": 5.111507292254736e-10,
+      "loss": 0.1981,
+      "step": 12400
+    },
+    {
+      "epoch": 2.990480780816966,
+      "grad_norm": 0.11416053771972656,
+      "learning_rate": 3.3035353875499056e-10,
+      "loss": 0.1952,
+      "step": 12410
+    },
+    {
+      "epoch": 2.992890709724063,
+      "grad_norm": 0.11342688649892807,
+      "learning_rate": 1.8885856521211333e-10,
+      "loss": 0.1975,
+      "step": 12420
+    },
+    {
+      "epoch": 2.9953006386311603,
+      "grad_norm": 0.11560092121362686,
+      "learning_rate": 8.666692088266094e-11,
+      "loss": 0.1997,
+      "step": 12430
+    },
+    {
+      "epoch": 2.997710567538258,
+      "grad_norm": 0.13347730040550232,
+      "learning_rate": 2.377940909237264e-11,
+      "loss": 0.1993,
+      "step": 12440
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.16301463544368744,
+      "learning_rate": 1.9652419636084773e-13,
+      "loss": 0.2003,
+      "step": 12450
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 12450,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 10000000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.276371997398208e+21,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}