diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4662 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9988649262202043,
+  "eval_steps": 500,
+  "global_step": 660,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003026863412788498,
+      "grad_norm": 3.235436458649722,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 0.6633,
+      "step": 1
+    },
+    {
+      "epoch": 0.006053726825576996,
+      "grad_norm": 3.5645935283996213,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 0.7281,
+      "step": 2
+    },
+    {
+      "epoch": 0.009080590238365494,
+      "grad_norm": 3.093392616778869,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.6619,
+      "step": 3
+    },
+    {
+      "epoch": 0.012107453651153992,
+      "grad_norm": 3.179784613260367,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 0.7157,
+      "step": 4
+    },
+    {
+      "epoch": 0.01513431706394249,
+      "grad_norm": 3.46258699147839,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 0.6966,
+      "step": 5
+    },
+    {
+      "epoch": 0.018161180476730987,
+      "grad_norm": 2.9974449421523177,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 0.69,
+      "step": 6
+    },
+    {
+      "epoch": 0.021188043889519486,
+      "grad_norm": 2.927738436003254,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 0.6768,
+      "step": 7
+    },
+    {
+      "epoch": 0.024214907302307985,
+      "grad_norm": 2.5627230161070766,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.6258,
+      "step": 8
+    },
+    {
+      "epoch": 0.02724177071509648,
+      "grad_norm": 1.9533304515860135,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 0.6542,
+      "step": 9
+    },
+    {
+      "epoch": 0.03026863412788498,
+      "grad_norm": 1.6031670956646584,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.6031,
+      "step": 10
+    },
+    {
+      "epoch": 0.03329549754067348,
+      "grad_norm": 1.582263600875127,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.5973,
+      "step": 11
+    },
+    {
+      "epoch": 0.036322360953461974,
+      "grad_norm": 1.8187361836534968,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 0.56,
+      "step": 12
+    },
+    {
+      "epoch": 0.03934922436625047,
+      "grad_norm": 2.6328924517759726,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 0.5889,
+      "step": 13
+    },
+    {
+      "epoch": 0.04237608777903897,
+      "grad_norm": 2.297455235478784,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 0.567,
+      "step": 14
+    },
+    {
+      "epoch": 0.04540295119182747,
+      "grad_norm": 1.6940276525975617,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.539,
+      "step": 15
+    },
+    {
+      "epoch": 0.04842981460461597,
+      "grad_norm": 1.557100492365704,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 0.551,
+      "step": 16
+    },
+    {
+      "epoch": 0.051456678017404466,
+      "grad_norm": 1.484912717744596,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 0.561,
+      "step": 17
+    },
+    {
+      "epoch": 0.05448354143019296,
+      "grad_norm": 1.5377546874099912,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.5402,
+      "step": 18
+    },
+    {
+      "epoch": 0.057510404842981463,
+      "grad_norm": 1.3623070346388788,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 0.5315,
+      "step": 19
+    },
+    {
+      "epoch": 0.06053726825576996,
+      "grad_norm": 1.194648507210797,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.5457,
+      "step": 20
+    },
+    {
+      "epoch": 0.06356413166855845,
+      "grad_norm": 1.1096469526390318,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 0.5437,
+      "step": 21
+    },
+    {
+      "epoch": 0.06659099508134696,
+      "grad_norm": 1.1488715164216095,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.5363,
+      "step": 22
+    },
+    {
+      "epoch": 0.06961785849413545,
+      "grad_norm": 1.08655716021847,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 0.5024,
+      "step": 23
+    },
+    {
+      "epoch": 0.07264472190692395,
+      "grad_norm": 0.9854350717024366,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.527,
+      "step": 24
+    },
+    {
+      "epoch": 0.07567158531971245,
+      "grad_norm": 0.9343934321342595,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 0.5013,
+      "step": 25
+    },
+    {
+      "epoch": 0.07869844873250094,
+      "grad_norm": 0.9357695211108015,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 0.4887,
+      "step": 26
+    },
+    {
+      "epoch": 0.08172531214528944,
+      "grad_norm": 0.9545589291967199,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.4864,
+      "step": 27
+    },
+    {
+      "epoch": 0.08475217555807794,
+      "grad_norm": 0.9992711933908768,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 0.5016,
+      "step": 28
+    },
+    {
+      "epoch": 0.08777903897086645,
+      "grad_norm": 0.8724761296961243,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 0.4776,
+      "step": 29
+    },
+    {
+      "epoch": 0.09080590238365494,
+      "grad_norm": 0.9939969896946891,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.5119,
+      "step": 30
+    },
+    {
+      "epoch": 0.09383276579644344,
+      "grad_norm": 0.9721951413329807,
+      "learning_rate": 9.393939393939396e-06,
+      "loss": 0.4876,
+      "step": 31
+    },
+    {
+      "epoch": 0.09685962920923194,
+      "grad_norm": 0.9605804902651769,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.4914,
+      "step": 32
+    },
+    {
+      "epoch": 0.09988649262202043,
+      "grad_norm": 0.8933282406320995,
+      "learning_rate": 1e-05,
+      "loss": 0.4692,
+      "step": 33
+    },
+    {
+      "epoch": 0.10291335603480893,
+      "grad_norm": 0.9103939718766416,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 0.4705,
+      "step": 34
+    },
+    {
+      "epoch": 0.10594021944759743,
+      "grad_norm": 0.9089186839038462,
+      "learning_rate": 1.0606060606060606e-05,
+      "loss": 0.4876,
+      "step": 35
+    },
+    {
+      "epoch": 0.10896708286038592,
+      "grad_norm": 0.8808305054441777,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.5064,
+      "step": 36
+    },
+    {
+      "epoch": 0.11199394627317442,
+      "grad_norm": 0.9202133707568927,
+      "learning_rate": 1.1212121212121212e-05,
+      "loss": 0.5072,
+      "step": 37
+    },
+    {
+      "epoch": 0.11502080968596293,
+      "grad_norm": 0.8482614015957081,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 0.475,
+      "step": 38
+    },
+    {
+      "epoch": 0.11804767309875142,
+      "grad_norm": 0.8474539666938257,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 0.4812,
+      "step": 39
+    },
+    {
+      "epoch": 0.12107453651153992,
+      "grad_norm": 0.8905684341419864,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.472,
+      "step": 40
+    },
+    {
+      "epoch": 0.12410139992432842,
+      "grad_norm": 0.8984945379550433,
+      "learning_rate": 1.2424242424242425e-05,
+      "loss": 0.4645,
+      "step": 41
+    },
+    {
+      "epoch": 0.1271282633371169,
+      "grad_norm": 1.011380245894981,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 0.4644,
+      "step": 42
+    },
+    {
+      "epoch": 0.1301551267499054,
+      "grad_norm": 0.9490157445069504,
+      "learning_rate": 1.3030303030303032e-05,
+      "loss": 0.4849,
+      "step": 43
+    },
+    {
+      "epoch": 0.13318199016269391,
+      "grad_norm": 0.9771201657720733,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.4884,
+      "step": 44
+    },
+    {
+      "epoch": 0.1362088535754824,
+      "grad_norm": 0.9590703124374225,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 0.4535,
+      "step": 45
+    },
+    {
+      "epoch": 0.1392357169882709,
+      "grad_norm": 0.8898615991271187,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 0.4838,
+      "step": 46
+    },
+    {
+      "epoch": 0.1422625804010594,
+      "grad_norm": 1.0535049541973271,
+      "learning_rate": 1.4242424242424245e-05,
+      "loss": 0.4903,
+      "step": 47
+    },
+    {
+      "epoch": 0.1452894438138479,
+      "grad_norm": 0.9437094791917945,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.478,
+      "step": 48
+    },
+    {
+      "epoch": 0.14831630722663638,
+      "grad_norm": 0.8997776335801163,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 0.4797,
+      "step": 49
+    },
+    {
+      "epoch": 0.1513431706394249,
+      "grad_norm": 1.0086849690587059,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.4831,
+      "step": 50
+    },
+    {
+      "epoch": 0.1543700340522134,
+      "grad_norm": 1.134307357014477,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 0.489,
+      "step": 51
+    },
+    {
+      "epoch": 0.15739689746500188,
+      "grad_norm": 1.0213818056555448,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 0.4723,
+      "step": 52
+    },
+    {
+      "epoch": 0.1604237608777904,
+      "grad_norm": 0.9374569620036011,
+      "learning_rate": 1.606060606060606e-05,
+      "loss": 0.4737,
+      "step": 53
+    },
+    {
+      "epoch": 0.16345062429057888,
+      "grad_norm": 0.9877869443702924,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.4814,
+      "step": 54
+    },
+    {
+      "epoch": 0.16647748770336737,
+      "grad_norm": 0.8689881305911301,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4623,
+      "step": 55
+    },
+    {
+      "epoch": 0.1695043511161559,
+      "grad_norm": 0.9661969974601068,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 0.4679,
+      "step": 56
+    },
+    {
+      "epoch": 0.17253121452894438,
+      "grad_norm": 0.9232526663906674,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 0.4993,
+      "step": 57
+    },
+    {
+      "epoch": 0.1755580779417329,
+      "grad_norm": 0.9442789679228032,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 0.4951,
+      "step": 58
+    },
+    {
+      "epoch": 0.17858494135452138,
+      "grad_norm": 0.9631389402769466,
+      "learning_rate": 1.787878787878788e-05,
+      "loss": 0.4912,
+      "step": 59
+    },
+    {
+      "epoch": 0.18161180476730987,
+      "grad_norm": 0.9990868712008387,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.4836,
+      "step": 60
+    },
+    {
+      "epoch": 0.1846386681800984,
+      "grad_norm": 0.9553625858589984,
+      "learning_rate": 1.8484848484848487e-05,
+      "loss": 0.4647,
+      "step": 61
+    },
+    {
+      "epoch": 0.18766553159288688,
+      "grad_norm": 1.0539161489638034,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 0.4874,
+      "step": 62
+    },
+    {
+      "epoch": 0.19069239500567536,
+      "grad_norm": 0.945529567190985,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 0.4724,
+      "step": 63
+    },
+    {
+      "epoch": 0.19371925841846388,
+      "grad_norm": 1.018363798045299,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 0.4746,
+      "step": 64
+    },
+    {
+      "epoch": 0.19674612183125237,
+      "grad_norm": 0.9043620095709098,
+      "learning_rate": 1.96969696969697e-05,
+      "loss": 0.4596,
+      "step": 65
+    },
+    {
+      "epoch": 0.19977298524404086,
+      "grad_norm": 0.8644661702492273,
+      "learning_rate": 2e-05,
+      "loss": 0.4551,
+      "step": 66
+    },
+    {
+      "epoch": 0.20279984865682937,
+      "grad_norm": 1.0451189341512501,
+      "learning_rate": 1.9999860139251737e-05,
+      "loss": 0.4714,
+      "step": 67
+    },
+    {
+      "epoch": 0.20582671206961786,
+      "grad_norm": 1.002626602774765,
+      "learning_rate": 1.9999440560919153e-05,
+      "loss": 0.4781,
+      "step": 68
+    },
+    {
+      "epoch": 0.20885357548240635,
+      "grad_norm": 0.9545348999807897,
+      "learning_rate": 1.9998741276738753e-05,
+      "loss": 0.4592,
+      "step": 69
+    },
+    {
+      "epoch": 0.21188043889519487,
+      "grad_norm": 1.030593546665657,
+      "learning_rate": 1.999776230627102e-05,
+      "loss": 0.4787,
+      "step": 70
+    },
+    {
+      "epoch": 0.21490730230798336,
+      "grad_norm": 0.9634420246177808,
+      "learning_rate": 1.9996503676899863e-05,
+      "loss": 0.4846,
+      "step": 71
+    },
+    {
+      "epoch": 0.21793416572077184,
+      "grad_norm": 0.9467386587915338,
+      "learning_rate": 1.9994965423831853e-05,
+      "loss": 0.4714,
+      "step": 72
+    },
+    {
+      "epoch": 0.22096102913356036,
+      "grad_norm": 0.8924293788346243,
+      "learning_rate": 1.9993147590095232e-05,
+      "loss": 0.4602,
+      "step": 73
+    },
+    {
+      "epoch": 0.22398789254634885,
+      "grad_norm": 0.9380565748757239,
+      "learning_rate": 1.999105022653872e-05,
+      "loss": 0.4574,
+      "step": 74
+    },
+    {
+      "epoch": 0.22701475595913734,
+      "grad_norm": 0.9518581092321344,
+      "learning_rate": 1.9988673391830082e-05,
+      "loss": 0.4839,
+      "step": 75
+    },
+    {
+      "epoch": 0.23004161937192585,
+      "grad_norm": 0.9538524503816821,
+      "learning_rate": 1.9986017152454497e-05,
+      "loss": 0.4733,
+      "step": 76
+    },
+    {
+      "epoch": 0.23306848278471434,
+      "grad_norm": 0.9234948362334984,
+      "learning_rate": 1.9983081582712684e-05,
+      "loss": 0.4584,
+      "step": 77
+    },
+    {
+      "epoch": 0.23609534619750283,
+      "grad_norm": 1.0118006052373039,
+      "learning_rate": 1.9979866764718846e-05,
+      "loss": 0.4352,
+      "step": 78
+    },
+    {
+      "epoch": 0.23912220961029135,
+      "grad_norm": 1.0145236646947924,
+      "learning_rate": 1.997637278839835e-05,
+      "loss": 0.4852,
+      "step": 79
+    },
+    {
+      "epoch": 0.24214907302307984,
+      "grad_norm": 0.8513391984670292,
+      "learning_rate": 1.9972599751485225e-05,
+      "loss": 0.4563,
+      "step": 80
+    },
+    {
+      "epoch": 0.24517593643586832,
+      "grad_norm": 1.251985454014456,
+      "learning_rate": 1.9968547759519426e-05,
+      "loss": 0.4585,
+      "step": 81
+    },
+    {
+      "epoch": 0.24820279984865684,
+      "grad_norm": 1.110770916578026,
+      "learning_rate": 1.9964216925843876e-05,
+      "loss": 0.4718,
+      "step": 82
+    },
+    {
+      "epoch": 0.2512296632614453,
+      "grad_norm": 0.9977711604302626,
+      "learning_rate": 1.9959607371601303e-05,
+      "loss": 0.4456,
+      "step": 83
+    },
+    {
+      "epoch": 0.2542565266742338,
+      "grad_norm": 1.103736062284778,
+      "learning_rate": 1.9954719225730847e-05,
+      "loss": 0.4712,
+      "step": 84
+    },
+    {
+      "epoch": 0.25728339008702233,
+      "grad_norm": 0.9312002516164151,
+      "learning_rate": 1.994955262496446e-05,
+      "loss": 0.4631,
+      "step": 85
+    },
+    {
+      "epoch": 0.2603102534998108,
+      "grad_norm": 0.9678544247788788,
+      "learning_rate": 1.9944107713823068e-05,
+      "loss": 0.472,
+      "step": 86
+    },
+    {
+      "epoch": 0.2633371169125993,
+      "grad_norm": 0.8587164979946262,
+      "learning_rate": 1.9938384644612542e-05,
+      "loss": 0.4728,
+      "step": 87
+    },
+    {
+      "epoch": 0.26636398032538783,
+      "grad_norm": 0.9493569562876263,
+      "learning_rate": 1.9932383577419432e-05,
+      "loss": 0.46,
+      "step": 88
+    },
+    {
+      "epoch": 0.2693908437381763,
+      "grad_norm": 0.9528631855943044,
+      "learning_rate": 1.9926104680106484e-05,
+      "loss": 0.4746,
+      "step": 89
+    },
+    {
+      "epoch": 0.2724177071509648,
+      "grad_norm": 1.0221463667936372,
+      "learning_rate": 1.9919548128307954e-05,
+      "loss": 0.4841,
+      "step": 90
+    },
+    {
+      "epoch": 0.2754445705637533,
+      "grad_norm": 0.880456179465352,
+      "learning_rate": 1.9912714105424694e-05,
+      "loss": 0.456,
+      "step": 91
+    },
+    {
+      "epoch": 0.2784714339765418,
+      "grad_norm": 1.007384572340114,
+      "learning_rate": 1.990560280261901e-05,
+      "loss": 0.4884,
+      "step": 92
+    },
+    {
+      "epoch": 0.2814982973893303,
+      "grad_norm": 0.8533142414315109,
+      "learning_rate": 1.989821441880933e-05,
+      "loss": 0.4322,
+      "step": 93
+    },
+    {
+      "epoch": 0.2845251608021188,
+      "grad_norm": 0.9114622713395024,
+      "learning_rate": 1.9890549160664633e-05,
+      "loss": 0.4751,
+      "step": 94
+    },
+    {
+      "epoch": 0.2875520242149073,
+      "grad_norm": 0.9326624170732905,
+      "learning_rate": 1.9882607242598663e-05,
+      "loss": 0.4507,
+      "step": 95
+    },
+    {
+      "epoch": 0.2905788876276958,
+      "grad_norm": 0.8529005087231366,
+      "learning_rate": 1.9874388886763944e-05,
+      "loss": 0.4514,
+      "step": 96
+    },
+    {
+      "epoch": 0.2936057510404843,
+      "grad_norm": 8.965430832156894,
+      "learning_rate": 1.9865894323045558e-05,
+      "loss": 0.4974,
+      "step": 97
+    },
+    {
+      "epoch": 0.29663261445327277,
+      "grad_norm": 12.904180683674056,
+      "learning_rate": 1.9857123789054707e-05,
+      "loss": 0.5255,
+      "step": 98
+    },
+    {
+      "epoch": 0.2996594778660613,
+      "grad_norm": 1.3225451438349816,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.4819,
+      "step": 99
+    },
+    {
+      "epoch": 0.3026863412788498,
+      "grad_norm": 1.0162073741038125,
+      "learning_rate": 1.9838755799290993e-05,
+      "loss": 0.475,
+      "step": 100
+    },
+    {
+      "epoch": 0.30571320469163826,
+      "grad_norm": 1.0484189507403519,
+      "learning_rate": 1.9829158857310288e-05,
+      "loss": 0.4663,
+      "step": 101
+    },
+    {
+      "epoch": 0.3087400681044268,
+      "grad_norm": 1.0536659287160743,
+      "learning_rate": 1.9819286972627066e-05,
+      "loss": 0.4969,
+      "step": 102
+    },
+    {
+      "epoch": 0.3117669315172153,
+      "grad_norm": 1.1267796297604018,
+      "learning_rate": 1.9809140421379168e-05,
+      "loss": 0.4808,
+      "step": 103
+    },
+    {
+      "epoch": 0.31479379493000376,
+      "grad_norm": 0.9124226717828057,
+      "learning_rate": 1.979871948738743e-05,
+      "loss": 0.4632,
+      "step": 104
+    },
+    {
+      "epoch": 0.3178206583427923,
+      "grad_norm": 1.0202922501989256,
+      "learning_rate": 1.978802446214779e-05,
+      "loss": 0.4712,
+      "step": 105
+    },
+    {
+      "epoch": 0.3208475217555808,
+      "grad_norm": 0.9541637843124078,
+      "learning_rate": 1.9777055644823087e-05,
+      "loss": 0.448,
+      "step": 106
+    },
+    {
+      "epoch": 0.32387438516836925,
+      "grad_norm": 0.9904713396707905,
+      "learning_rate": 1.9765813342234726e-05,
+      "loss": 0.4805,
+      "step": 107
+    },
+    {
+      "epoch": 0.32690124858115777,
+      "grad_norm": 0.9698707772276001,
+      "learning_rate": 1.9754297868854075e-05,
+      "loss": 0.476,
+      "step": 108
+    },
+    {
+      "epoch": 0.3299281119939463,
+      "grad_norm": 0.9354237855295647,
+      "learning_rate": 1.9742509546793673e-05,
+      "loss": 0.4468,
+      "step": 109
+    },
+    {
+      "epoch": 0.33295497540673474,
+      "grad_norm": 0.9230396561919654,
+      "learning_rate": 1.973044870579824e-05,
+      "loss": 0.4589,
+      "step": 110
+    },
+    {
+      "epoch": 0.33598183881952326,
+      "grad_norm": 1.0501558478916118,
+      "learning_rate": 1.9718115683235418e-05,
+      "loss": 0.4458,
+      "step": 111
+    },
+    {
+      "epoch": 0.3390087022323118,
+      "grad_norm": 0.8140014657502624,
+      "learning_rate": 1.970551082408636e-05,
+      "loss": 0.4456,
+      "step": 112
+    },
+    {
+      "epoch": 0.3420355656451003,
+      "grad_norm": 0.9197656914259078,
+      "learning_rate": 1.969263448093608e-05,
+      "loss": 0.4574,
+      "step": 113
+    },
+    {
+      "epoch": 0.34506242905788875,
+      "grad_norm": 0.9466028010229838,
+      "learning_rate": 1.9679487013963566e-05,
+      "loss": 0.4562,
+      "step": 114
+    },
+    {
+      "epoch": 0.34808929247067727,
+      "grad_norm": 0.7859343506423083,
+      "learning_rate": 1.9666068790931733e-05,
+      "loss": 0.4652,
+      "step": 115
+    },
+    {
+      "epoch": 0.3511161558834658,
+      "grad_norm": 0.842199823964043,
+      "learning_rate": 1.9652380187177128e-05,
+      "loss": 0.4511,
+      "step": 116
+    },
+    {
+      "epoch": 0.35414301929625425,
+      "grad_norm": 0.8678047904641017,
+      "learning_rate": 1.9638421585599422e-05,
+      "loss": 0.4685,
+      "step": 117
+    },
+    {
+      "epoch": 0.35716988270904276,
+      "grad_norm": 0.8935927032846372,
+      "learning_rate": 1.9624193376650708e-05,
+      "loss": 0.4547,
+      "step": 118
+    },
+    {
+      "epoch": 0.3601967461218313,
+      "grad_norm": 0.8090398363404913,
+      "learning_rate": 1.960969595832457e-05,
+      "loss": 0.4489,
+      "step": 119
+    },
+    {
+      "epoch": 0.36322360953461974,
+      "grad_norm": 0.8571063903270532,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.4687,
+      "step": 120
+    },
+    {
+      "epoch": 0.36625047294740826,
+      "grad_norm": 0.8347566205280457,
+      "learning_rate": 1.957989512315489e-05,
+      "loss": 0.4362,
+      "step": 121
+    },
+    {
+      "epoch": 0.3692773363601968,
+      "grad_norm": 0.8496125169402892,
+      "learning_rate": 1.956459253990476e-05,
+      "loss": 0.4494,
+      "step": 122
+    },
+    {
+      "epoch": 0.37230419977298523,
+      "grad_norm": 0.8607909235741252,
+      "learning_rate": 1.9549022414440738e-05,
+      "loss": 0.4587,
+      "step": 123
+    },
+    {
+      "epoch": 0.37533106318577375,
+      "grad_norm": 0.8377688613673809,
+      "learning_rate": 1.9533185182292705e-05,
+      "loss": 0.4649,
+      "step": 124
+    },
+    {
+      "epoch": 0.37835792659856227,
+      "grad_norm": 0.8956718022996314,
+      "learning_rate": 1.9517081286462082e-05,
+      "loss": 0.4691,
+      "step": 125
+    },
+    {
+      "epoch": 0.3813847900113507,
+      "grad_norm": 0.8619607962579379,
+      "learning_rate": 1.9500711177409456e-05,
+      "loss": 0.4632,
+      "step": 126
+    },
+    {
+      "epoch": 0.38441165342413924,
+      "grad_norm": 0.7922147893885945,
+      "learning_rate": 1.9484075313041968e-05,
+      "loss": 0.4493,
+      "step": 127
+    },
+    {
+      "epoch": 0.38743851683692776,
+      "grad_norm": 0.8228714306966575,
+      "learning_rate": 1.9467174158700507e-05,
+      "loss": 0.4371,
+      "step": 128
+    },
+    {
+      "epoch": 0.3904653802497162,
+      "grad_norm": 0.8104019279961472,
+      "learning_rate": 1.9450008187146685e-05,
+      "loss": 0.4391,
+      "step": 129
+    },
+    {
+      "epoch": 0.39349224366250474,
+      "grad_norm": 0.7838993189333445,
+      "learning_rate": 1.9432577878549635e-05,
+      "loss": 0.4688,
+      "step": 130
+    },
+    {
+      "epoch": 0.39651910707529325,
+      "grad_norm": 0.8756965482353155,
+      "learning_rate": 1.9414883720472557e-05,
+      "loss": 0.4598,
+      "step": 131
+    },
+    {
+      "epoch": 0.3995459704880817,
+      "grad_norm": 0.953360061737771,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.4733,
+      "step": 132
+    },
+    {
+      "epoch": 0.40257283390087023,
+      "grad_norm": 0.8308443396762537,
+      "learning_rate": 1.937870584301945e-05,
+      "loss": 0.4671,
+      "step": 133
+    },
+    {
+      "epoch": 0.40559969731365875,
+      "grad_norm": 0.817588513292344,
+      "learning_rate": 1.9360223135616423e-05,
+      "loss": 0.4617,
+      "step": 134
+    },
+    {
+      "epoch": 0.4086265607264472,
+      "grad_norm": 0.8327012262765101,
+      "learning_rate": 1.9341478602651068e-05,
+      "loss": 0.4564,
+      "step": 135
+    },
+    {
+      "epoch": 0.4116534241392357,
+      "grad_norm": 0.8538273356717329,
+      "learning_rate": 1.932247276844826e-05,
+      "loss": 0.4402,
+      "step": 136
+    },
+    {
+      "epoch": 0.41468028755202424,
+      "grad_norm": 0.759617785498127,
+      "learning_rate": 1.9303206164642037e-05,
+      "loss": 0.4601,
+      "step": 137
+    },
+    {
+      "epoch": 0.4177071509648127,
+      "grad_norm": 0.8194134193241404,
+      "learning_rate": 1.9283679330160726e-05,
+      "loss": 0.46,
+      "step": 138
+    },
+    {
+      "epoch": 0.4207340143776012,
+      "grad_norm": 0.8739952351533794,
+      "learning_rate": 1.9263892811211865e-05,
+      "loss": 0.4624,
+      "step": 139
+    },
+    {
+      "epoch": 0.42376087779038973,
+      "grad_norm": 0.7767141630388571,
+      "learning_rate": 1.9243847161266924e-05,
+      "loss": 0.4507,
+      "step": 140
+    },
+    {
+      "epoch": 0.4267877412031782,
+      "grad_norm": 0.9201821660105389,
+      "learning_rate": 1.9223542941045817e-05,
+      "loss": 0.4676,
+      "step": 141
+    },
+    {
+      "epoch": 0.4298146046159667,
+      "grad_norm": 0.7722345893954266,
+      "learning_rate": 1.920298071850123e-05,
+      "loss": 0.4478,
+      "step": 142
+    },
+    {
+      "epoch": 0.4328414680287552,
+      "grad_norm": 0.7741229952829096,
+      "learning_rate": 1.9182161068802742e-05,
+      "loss": 0.4431,
+      "step": 143
+    },
+    {
+      "epoch": 0.4358683314415437,
+      "grad_norm": 0.7981544009870728,
+      "learning_rate": 1.9161084574320696e-05,
+      "loss": 0.4436,
+      "step": 144
+    },
+    {
+      "epoch": 0.4388951948543322,
+      "grad_norm": 0.8242168829362365,
+      "learning_rate": 1.913975182460996e-05,
+      "loss": 0.4557,
+      "step": 145
+    },
+    {
+      "epoch": 0.4419220582671207,
+      "grad_norm": 0.7706535357047114,
+      "learning_rate": 1.9118163416393392e-05,
+      "loss": 0.4422,
+      "step": 146
+    },
+    {
+      "epoch": 0.4449489216799092,
+      "grad_norm": 0.8419257040272804,
+      "learning_rate": 1.9096319953545186e-05,
+      "loss": 0.4554,
+      "step": 147
+    },
+    {
+      "epoch": 0.4479757850926977,
+      "grad_norm": 0.8378837909589202,
+      "learning_rate": 1.9074222047073945e-05,
+      "loss": 0.4762,
+      "step": 148
+    },
+    {
+      "epoch": 0.4510026485054862,
+      "grad_norm": 0.8385547498874574,
+      "learning_rate": 1.9051870315105626e-05,
+      "loss": 0.459,
+      "step": 149
+    },
+    {
+      "epoch": 0.4540295119182747,
+      "grad_norm": 0.7759374570558363,
+      "learning_rate": 1.9029265382866216e-05,
+      "loss": 0.4522,
+      "step": 150
+    },
+    {
+      "epoch": 0.4570563753310632,
+      "grad_norm": 0.7759289530084589,
+      "learning_rate": 1.9006407882664256e-05,
+      "loss": 0.4214,
+      "step": 151
+    },
+    {
+      "epoch": 0.4600832387438517,
+      "grad_norm": 0.7709877456302019,
+      "learning_rate": 1.8983298453873172e-05,
+      "loss": 0.4507,
+      "step": 152
+    },
+    {
+      "epoch": 0.46311010215664017,
+      "grad_norm": 0.8010872699905442,
+      "learning_rate": 1.895993774291336e-05,
+      "loss": 0.4583,
+      "step": 153
+    },
+    {
+      "epoch": 0.4661369655694287,
+      "grad_norm": 0.7442741212452282,
+      "learning_rate": 1.8936326403234125e-05,
+      "loss": 0.4357,
+      "step": 154
+    },
+    {
+      "epoch": 0.4691638289822172,
+      "grad_norm": 0.8124936265412964,
+      "learning_rate": 1.891246509529539e-05,
+      "loss": 0.4632,
+      "step": 155
+    },
+    {
+      "epoch": 0.47219069239500566,
+      "grad_norm": 0.7966106026514826,
+      "learning_rate": 1.8888354486549238e-05,
+      "loss": 0.4562,
+      "step": 156
+    },
+    {
+      "epoch": 0.4752175558077942,
+      "grad_norm": 0.7676758497760634,
+      "learning_rate": 1.886399525142122e-05,
+      "loss": 0.4343,
+      "step": 157
+    },
+    {
+      "epoch": 0.4782444192205827,
+      "grad_norm": 0.7601518615733768,
+      "learning_rate": 1.8839388071291506e-05,
+      "loss": 0.4454,
+      "step": 158
+    },
+    {
+      "epoch": 0.48127128263337116,
+      "grad_norm": 0.7744639740374917,
+      "learning_rate": 1.881453363447582e-05,
+      "loss": 0.4688,
+      "step": 159
+    },
+    {
+      "epoch": 0.48429814604615967,
+      "grad_norm": 0.8663350945527193,
+      "learning_rate": 1.8789432636206197e-05,
+      "loss": 0.4383,
+      "step": 160
+    },
+    {
+      "epoch": 0.4873250094589482,
+      "grad_norm": 0.8573400399489379,
+      "learning_rate": 1.8764085778611507e-05,
+      "loss": 0.477,
+      "step": 161
+    },
+    {
+      "epoch": 0.49035187287173665,
+      "grad_norm": 0.776800144650827,
+      "learning_rate": 1.873849377069785e-05,
+      "loss": 0.4279,
+      "step": 162
+    },
+    {
+      "epoch": 0.49337873628452517,
+      "grad_norm": 0.8162914307039698,
+      "learning_rate": 1.87126573283287e-05,
+      "loss": 0.4498,
+      "step": 163
+    },
+    {
+      "epoch": 0.4964055996973137,
+      "grad_norm": 0.7793488477804008,
+      "learning_rate": 1.8686577174204887e-05,
+      "loss": 0.4507,
+      "step": 164
+    },
+    {
+      "epoch": 0.49943246311010214,
+      "grad_norm": 0.7581955869362403,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.436,
+      "step": 165
+    },
+    {
+      "epoch": 0.5024593265228906,
+      "grad_norm": 0.8428076101491118,
+      "learning_rate": 1.863368865556191e-05,
+      "loss": 0.4412,
+      "step": 166
+    },
+    {
+      "epoch": 0.5054861899356792,
+      "grad_norm": 0.9637638201950827,
+      "learning_rate": 1.8606881770448305e-05,
+      "loss": 0.4351,
+      "step": 167
+    },
+    {
+      "epoch": 0.5085130533484676,
+      "grad_norm": 0.8169262851281399,
+      "learning_rate": 1.8579834132349773e-05,
+      "loss": 0.4442,
+      "step": 168
+    },
+    {
+      "epoch": 0.5115399167612561,
+      "grad_norm": 0.8167322292597452,
+      "learning_rate": 1.8552546497846893e-05,
+      "loss": 0.4341,
+      "step": 169
+    },
+    {
+      "epoch": 0.5145667801740447,
+      "grad_norm": 0.8770780917772608,
+      "learning_rate": 1.8525019630233463e-05,
+      "loss": 0.4743,
+      "step": 170
+    },
+    {
+      "epoch": 0.5175936435868331,
+      "grad_norm": 1.0567304218943345,
+      "learning_rate": 1.8497254299495147e-05,
+      "loss": 0.4371,
+      "step": 171
+    },
+    {
+      "epoch": 0.5206205069996216,
+      "grad_norm": 0.8148740362632707,
+      "learning_rate": 1.8469251282287925e-05,
+      "loss": 0.4435,
+      "step": 172
+    },
+    {
+      "epoch": 0.5236473704124102,
+      "grad_norm": 0.7933230558237211,
+      "learning_rate": 1.8441011361916387e-05,
+      "loss": 0.4423,
+      "step": 173
+    },
+    {
+      "epoch": 0.5266742338251986,
+      "grad_norm": 0.8256694445635196,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.4433,
+      "step": 174
+    },
+    {
+      "epoch": 0.5297010972379871,
+      "grad_norm": 0.7420630428123556,
+      "learning_rate": 1.8383823978010077e-05,
+      "loss": 0.4519,
+      "step": 175
+    },
+    {
+      "epoch": 0.5327279606507757,
+      "grad_norm": 0.7933055150071372,
+      "learning_rate": 1.8354878114129368e-05,
+      "loss": 0.417,
+      "step": 176
+    },
+    {
+      "epoch": 0.5357548240635641,
+      "grad_norm": 0.7520339336260935,
+      "learning_rate": 1.8325698546347714e-05,
+      "loss": 0.4623,
+      "step": 177
+    },
+    {
+      "epoch": 0.5387816874763526,
+      "grad_norm": 0.7619546315139389,
+      "learning_rate": 1.8296286090880362e-05,
+      "loss": 0.4408,
+      "step": 178
+    },
+    {
+      "epoch": 0.5418085508891411,
+      "grad_norm": 0.7958562166380587,
+      "learning_rate": 1.8266641570456915e-05,
+      "loss": 0.4473,
+      "step": 179
+    },
+    {
+      "epoch": 0.5448354143019296,
+      "grad_norm": 0.8414500343837285,
+      "learning_rate": 1.8236765814298328e-05,
+      "loss": 0.4447,
+      "step": 180
+    },
+    {
+      "epoch": 0.5478622777147181,
+      "grad_norm": 0.9080616373812107,
+      "learning_rate": 1.820665965809373e-05,
+      "loss": 0.4456,
+      "step": 181
+    },
+    {
+      "epoch": 0.5508891411275066,
+      "grad_norm": 0.754258431179921,
+      "learning_rate": 1.8176323943977034e-05,
+      "loss": 0.4351,
+      "step": 182
+    },
+    {
+      "epoch": 0.5539160045402951,
+      "grad_norm": 0.8098503106780682,
+      "learning_rate": 1.814575952050336e-05,
+      "loss": 0.441,
+      "step": 183
+    },
+    {
+      "epoch": 0.5569428679530836,
+      "grad_norm": 0.8352289414660538,
+      "learning_rate": 1.8114967242625342e-05,
+      "loss": 0.4494,
+      "step": 184
+    },
+    {
+      "epoch": 0.5599697313658721,
+      "grad_norm": 0.7681163692296273,
+      "learning_rate": 1.808394797166919e-05,
+      "loss": 0.4384,
+      "step": 185
+    },
+    {
+      "epoch": 0.5629965947786606,
+      "grad_norm": 0.747127957915616,
+      "learning_rate": 1.8052702575310588e-05,
+      "loss": 0.4121,
+      "step": 186
+    },
+    {
+      "epoch": 0.5660234581914491,
+      "grad_norm": 0.8218478679852107,
+      "learning_rate": 1.802123192755044e-05,
+      "loss": 0.4409,
+      "step": 187
+    },
+    {
+      "epoch": 0.5690503216042376,
+      "grad_norm": 0.8152328898758183,
+      "learning_rate": 1.7989536908690413e-05,
+      "loss": 0.4327,
+      "step": 188
+    },
+    {
+      "epoch": 0.5720771850170261,
+      "grad_norm": 0.8711371168887442,
+      "learning_rate": 1.7957618405308323e-05,
+      "loss": 0.4378,
+      "step": 189
+    },
+    {
+      "epoch": 0.5751040484298146,
+      "grad_norm": 0.8638263265536331,
+      "learning_rate": 1.792547731023332e-05,
+      "loss": 0.4146,
+      "step": 190
+    },
+    {
+      "epoch": 0.5781309118426031,
+      "grad_norm": 0.8002603800846039,
+      "learning_rate": 1.789311452252092e-05,
+      "loss": 0.4268,
+      "step": 191
+    },
+    {
+      "epoch": 0.5811577752553916,
+      "grad_norm": 0.8573803414316463,
+      "learning_rate": 1.7860530947427878e-05,
+      "loss": 0.4385,
+      "step": 192
+    },
+    {
+      "epoch": 0.58418463866818,
+      "grad_norm": 0.8073779924473132,
+      "learning_rate": 1.782772749638682e-05,
+      "loss": 0.4336,
+      "step": 193
+    },
+    {
+      "epoch": 0.5872115020809686,
+      "grad_norm": 0.7608108526849635,
+      "learning_rate": 1.779470508698079e-05,
+      "loss": 0.4255,
+      "step": 194
+    },
+    {
+      "epoch": 0.5902383654937571,
+      "grad_norm": 0.8482995521517027,
+      "learning_rate": 1.776146464291757e-05,
+      "loss": 0.425,
+      "step": 195
+    },
+    {
+      "epoch": 0.5932652289065455,
+      "grad_norm": 0.8916144337271275,
+      "learning_rate": 1.772800709400383e-05,
+      "loss": 0.4411,
+      "step": 196
+    },
+    {
+      "epoch": 0.5962920923193341,
+      "grad_norm": 0.7417144652621136,
+      "learning_rate": 1.7694333376119144e-05,
+      "loss": 0.4382,
+      "step": 197
+    },
+    {
+      "epoch": 0.5993189557321226,
+      "grad_norm": 0.8300238664862641,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.4334,
+      "step": 198
+    },
+    {
+      "epoch": 0.602345819144911,
+      "grad_norm": 0.7299705378025234,
+      "learning_rate": 1.762634120716238e-05,
+      "loss": 0.4235,
+      "step": 199
+    },
+    {
+      "epoch": 0.6053726825576996,
+      "grad_norm": 0.7044753738932356,
+      "learning_rate": 1.7592024657977432e-05,
+      "loss": 0.4299,
+      "step": 200
+    },
+    {
+      "epoch": 0.6083995459704881,
+      "grad_norm": 0.7642157289476394,
+      "learning_rate": 1.7557495743542586e-05,
+      "loss": 0.4458,
+      "step": 201
+    },
+    {
+      "epoch": 0.6114264093832765,
+      "grad_norm": 0.8428239024660611,
+      "learning_rate": 1.75227554297058e-05,
+      "loss": 0.4304,
+      "step": 202
+    },
+    {
+      "epoch": 0.6144532727960651,
+      "grad_norm": 0.8077485956789754,
+      "learning_rate": 1.7487804688228327e-05,
+      "loss": 0.4367,
+      "step": 203
+    },
+    {
+      "epoch": 0.6174801362088536,
+      "grad_norm": 0.9063559367377406,
+      "learning_rate": 1.745264449675755e-05,
+      "loss": 0.4453,
+      "step": 204
+    },
+    {
+      "epoch": 0.620506999621642,
+      "grad_norm": 0.8043034850106837,
+      "learning_rate": 1.7417275838799596e-05,
+      "loss": 0.4514,
+      "step": 205
+    },
+    {
+      "epoch": 0.6235338630344306,
+      "grad_norm": 0.8144260157371154,
+      "learning_rate": 1.7381699703691866e-05,
+      "loss": 0.4459,
+      "step": 206
+    },
+    {
+      "epoch": 0.626560726447219,
+      "grad_norm": 0.9563709608799079,
+      "learning_rate": 1.734591708657533e-05,
+      "loss": 0.4258,
+      "step": 207
+    },
+    {
+      "epoch": 0.6295875898600075,
+      "grad_norm": 0.7479430675820818,
+      "learning_rate": 1.730992898836672e-05,
+      "loss": 0.4352,
+      "step": 208
+    },
+    {
+      "epoch": 0.6326144532727961,
+      "grad_norm": 1.092865824271307,
+      "learning_rate": 1.7273736415730488e-05,
+      "loss": 0.4365,
+      "step": 209
+    },
+    {
+      "epoch": 0.6356413166855845,
+      "grad_norm": 0.8375699585075692,
+      "learning_rate": 1.72373403810507e-05,
+      "loss": 0.4599,
+      "step": 210
+    },
+    {
+      "epoch": 0.638668180098373,
+      "grad_norm": 0.8254247763199031,
+      "learning_rate": 1.720074190240269e-05,
+      "loss": 0.4331,
+      "step": 211
+    },
+    {
+      "epoch": 0.6416950435111616,
+      "grad_norm": 0.8155654342341622,
+      "learning_rate": 1.7163942003524574e-05,
+      "loss": 0.441,
+      "step": 212
+    },
+    {
+      "epoch": 0.64472190692395,
+      "grad_norm": 0.8363163686812076,
+      "learning_rate": 1.7126941713788633e-05,
+      "loss": 0.4572,
+      "step": 213
+    },
+    {
+      "epoch": 0.6477487703367385,
+      "grad_norm": 0.8065614536681267,
+      "learning_rate": 1.70897420681725e-05,
+      "loss": 0.4229,
+      "step": 214
+    },
+    {
+      "epoch": 0.6507756337495271,
+      "grad_norm": 0.8798391293864685,
+      "learning_rate": 1.7052344107230244e-05,
+      "loss": 0.4375,
+      "step": 215
+    },
+    {
+      "epoch": 0.6538024971623155,
+      "grad_norm": 0.816194442972985,
+      "learning_rate": 1.7014748877063212e-05,
+      "loss": 0.4505,
+      "step": 216
+    },
+    {
+      "epoch": 0.656829360575104,
+      "grad_norm": 0.763640038718994,
+      "learning_rate": 1.697695742929082e-05,
+      "loss": 0.4285,
+      "step": 217
+    },
+    {
+      "epoch": 0.6598562239878926,
+      "grad_norm": 0.8665799367816605,
+      "learning_rate": 1.693897082102109e-05,
+      "loss": 0.4327,
+      "step": 218
+    },
+    {
+      "epoch": 0.662883087400681,
+      "grad_norm": 0.7632749371911016,
+      "learning_rate": 1.6900790114821122e-05,
+      "loss": 0.419,
+      "step": 219
+    },
+    {
+      "epoch": 0.6659099508134695,
+      "grad_norm": 0.82972578035047,
+      "learning_rate": 1.686241637868734e-05,
+      "loss": 0.4227,
+      "step": 220
+    },
+    {
+      "epoch": 0.6689368142262581,
+      "grad_norm": 0.726025867928948,
+      "learning_rate": 1.682385068601563e-05,
+      "loss": 0.4299,
+      "step": 221
+    },
+    {
+      "epoch": 0.6719636776390465,
+      "grad_norm": 0.8050613150824316,
+      "learning_rate": 1.6785094115571323e-05,
+      "loss": 0.4341,
+      "step": 222
+    },
+    {
+      "epoch": 0.6749905410518351,
+      "grad_norm": 0.7668645524154176,
+      "learning_rate": 1.674614775145901e-05,
+      "loss": 0.434,
+      "step": 223
+    },
+    {
+      "epoch": 0.6780174044646236,
+      "grad_norm": 0.8473114078768379,
+      "learning_rate": 1.670701268309221e-05,
+      "loss": 0.4166,
+      "step": 224
+    },
+    {
+      "epoch": 0.681044267877412,
+      "grad_norm": 0.8383790645003683,
+      "learning_rate": 1.666769000516292e-05,
+      "loss": 0.428,
+      "step": 225
+    },
+    {
+      "epoch": 0.6840711312902006,
+      "grad_norm": 0.7693649578515219,
+      "learning_rate": 1.6628180817610963e-05,
+      "loss": 0.4309,
+      "step": 226
+    },
+    {
+      "epoch": 0.687097994702989,
+      "grad_norm": 0.8320533148447656,
+      "learning_rate": 1.658848622559325e-05,
+      "loss": 0.4212,
+      "step": 227
+    },
+    {
+      "epoch": 0.6901248581157775,
+      "grad_norm": 0.7260198892087389,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.4251,
+      "step": 228
+    },
+    {
+      "epoch": 0.6931517215285661,
+      "grad_norm": 0.9029577261157358,
+      "learning_rate": 1.6508545274687936e-05,
+      "loss": 0.4572,
+      "step": 229
+    },
+    {
+      "epoch": 0.6961785849413545,
+      "grad_norm": 0.7197466238499969,
+      "learning_rate": 1.6468301151920576e-05,
+      "loss": 0.4242,
+      "step": 230
+    },
+    {
+      "epoch": 0.699205448354143,
+      "grad_norm": 0.7742847644397802,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.4394,
+      "step": 231
+    },
+    {
+      "epoch": 0.7022323117669316,
+      "grad_norm": 0.8380011329459673,
+      "learning_rate": 1.6387271240298082e-05,
+      "loss": 0.4371,
+      "step": 232
+    },
+    {
+      "epoch": 0.70525917517972,
+      "grad_norm": 0.7107929674639848,
+      "learning_rate": 1.6346487718023762e-05,
+      "loss": 0.4366,
+      "step": 233
+    },
+    {
+      "epoch": 0.7082860385925085,
+      "grad_norm": 0.9619734201764684,
+      "learning_rate": 1.6305526670845225e-05,
+      "loss": 0.4353,
+      "step": 234
+    },
+    {
+      "epoch": 0.7113129020052971,
+      "grad_norm": 0.751949350016426,
+      "learning_rate": 1.6264389244531015e-05,
+      "loss": 0.4483,
+      "step": 235
+    },
+    {
+      "epoch": 0.7143397654180855,
+      "grad_norm": 0.9579595471430041,
+      "learning_rate": 1.6223076589783368e-05,
+      "loss": 0.4419,
+      "step": 236
+    },
+    {
+      "epoch": 0.717366628830874,
+      "grad_norm": 0.8814789311643189,
+      "learning_rate": 1.6181589862206053e-05,
+      "loss": 0.4412,
+      "step": 237
+    },
+    {
+      "epoch": 0.7203934922436626,
+      "grad_norm": 0.7890367977878839,
+      "learning_rate": 1.613993022227202e-05,
+      "loss": 0.4711,
+      "step": 238
+    },
+    {
+      "epoch": 0.723420355656451,
+      "grad_norm": 0.8725786601277864,
+      "learning_rate": 1.6098098835290955e-05,
+      "loss": 0.4097,
+      "step": 239
+    },
+    {
+      "epoch": 0.7264472190692395,
+      "grad_norm": 0.7535459533804848,
+      "learning_rate": 1.6056096871376667e-05,
+      "loss": 0.405,
+      "step": 240
+    },
+    {
+      "epoch": 0.729474082482028,
+      "grad_norm": 0.8151725329306759,
+      "learning_rate": 1.6013925505414386e-05,
+      "loss": 0.4443,
+      "step": 241
+    },
+    {
+      "epoch": 0.7325009458948165,
+      "grad_norm": 0.6986284239125755,
+      "learning_rate": 1.5971585917027864e-05,
+      "loss": 0.4387,
+      "step": 242
+    },
+    {
+      "epoch": 0.735527809307605,
+      "grad_norm": 0.711160824465759,
+      "learning_rate": 1.5929079290546408e-05,
+      "loss": 0.4313,
+      "step": 243
+    },
+    {
+      "epoch": 0.7385546727203935,
+      "grad_norm": 0.7747993590917854,
+      "learning_rate": 1.5886406814971728e-05,
+      "loss": 0.4345,
+      "step": 244
+    },
+    {
+      "epoch": 0.741581536133182,
+      "grad_norm": 0.6998573640035735,
+      "learning_rate": 1.584356968394471e-05,
+      "loss": 0.4191,
+      "step": 245
+    },
+    {
+      "epoch": 0.7446083995459705,
+      "grad_norm": 0.7828113851190492,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 0.4584,
+      "step": 246
+    },
+    {
+      "epoch": 0.747635262958759,
+      "grad_norm": 0.8208386890894316,
+      "learning_rate": 1.575740625309244e-05,
+      "loss": 0.4687,
+      "step": 247
+    },
+    {
+      "epoch": 0.7506621263715475,
+      "grad_norm": 0.6749431645261781,
+      "learning_rate": 1.5714082363443576e-05,
+      "loss": 0.4134,
+      "step": 248
+    },
+    {
+      "epoch": 0.753688989784336,
+      "grad_norm": 0.7318896821137213,
+      "learning_rate": 1.5670598638627707e-05,
+      "loss": 0.4495,
+      "step": 249
+    },
+    {
+      "epoch": 0.7567158531971245,
+      "grad_norm": 0.6740851031790088,
+      "learning_rate": 1.5626956294978103e-05,
+      "loss": 0.416,
+      "step": 250
+    },
+    {
+      "epoch": 0.759742716609913,
+      "grad_norm": 0.7420662374601229,
+      "learning_rate": 1.5583156553264923e-05,
+      "loss": 0.4279,
+      "step": 251
+    },
+    {
+      "epoch": 0.7627695800227015,
+      "grad_norm": 0.7244466694674909,
+      "learning_rate": 1.5539200638661106e-05,
+      "loss": 0.4176,
+      "step": 252
+    },
+    {
+      "epoch": 0.76579644343549,
+      "grad_norm": 0.699452393476543,
+      "learning_rate": 1.5495089780708062e-05,
+      "loss": 0.4311,
+      "step": 253
+    },
+    {
+      "epoch": 0.7688233068482785,
+      "grad_norm": 0.7131583405325259,
+      "learning_rate": 1.5450825213281317e-05,
+      "loss": 0.4377,
+      "step": 254
+    },
+    {
+      "epoch": 0.771850170261067,
+      "grad_norm": 0.7998788659541156,
+      "learning_rate": 1.5406408174555978e-05,
+      "loss": 0.4218,
+      "step": 255
+    },
+    {
+      "epoch": 0.7748770336738555,
+      "grad_norm": 0.7187304321390485,
+      "learning_rate": 1.5361839906972095e-05,
+      "loss": 0.4113,
+      "step": 256
+    },
+    {
+      "epoch": 0.777903897086644,
+      "grad_norm": 0.7189215644246775,
+      "learning_rate": 1.531712165719992e-05,
+      "loss": 0.4351,
+      "step": 257
+    },
+    {
+      "epoch": 0.7809307604994324,
+      "grad_norm": 0.912368504828945,
+      "learning_rate": 1.5272254676105026e-05,
+      "loss": 0.4279,
+      "step": 258
+    },
+    {
+      "epoch": 0.783957623912221,
+      "grad_norm": 0.7991126185266673,
+      "learning_rate": 1.5227240218713326e-05,
+      "loss": 0.4158,
+      "step": 259
+    },
+    {
+      "epoch": 0.7869844873250095,
+      "grad_norm": 0.9009696818932866,
+      "learning_rate": 1.5182079544175957e-05,
+      "loss": 0.429,
+      "step": 260
+    },
+    {
+      "epoch": 0.7900113507377979,
+      "grad_norm": 0.7407945230185996,
+      "learning_rate": 1.5136773915734067e-05,
+      "loss": 0.4336,
+      "step": 261
+    },
+    {
+      "epoch": 0.7930382141505865,
+      "grad_norm": 0.7883924763371828,
+      "learning_rate": 1.5091324600683472e-05,
+      "loss": 0.4244,
+      "step": 262
+    },
+    {
+      "epoch": 0.796065077563375,
+      "grad_norm": 0.6813004638104684,
+      "learning_rate": 1.5045732870339213e-05,
+      "loss": 0.3992,
+      "step": 263
+    },
+    {
+      "epoch": 0.7990919409761634,
+      "grad_norm": 0.778562345968094,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4418,
+      "step": 264
+    },
+    {
+      "epoch": 0.802118804388952,
+      "grad_norm": 0.8114511723746026,
+      "learning_rate": 1.4954127268912525e-05,
+      "loss": 0.4168,
+      "step": 265
+    },
+    {
+      "epoch": 0.8051456678017405,
+      "grad_norm": 0.7723328828458176,
+      "learning_rate": 1.4908115960235683e-05,
+      "loss": 0.4449,
+      "step": 266
+    },
+    {
+      "epoch": 0.8081725312145289,
+      "grad_norm": 0.7979142117941048,
+      "learning_rate": 1.4861967361004687e-05,
+      "loss": 0.449,
+      "step": 267
+    },
+    {
+      "epoch": 0.8111993946273175,
+      "grad_norm": 0.7295039417962921,
+      "learning_rate": 1.4815682762095065e-05,
+      "loss": 0.4129,
+      "step": 268
+    },
+    {
+      "epoch": 0.814226258040106,
+      "grad_norm": 0.7664832516504386,
+      "learning_rate": 1.476926345818654e-05,
+      "loss": 0.4387,
+      "step": 269
+    },
+    {
+      "epoch": 0.8172531214528944,
+      "grad_norm": 0.756152513707808,
+      "learning_rate": 1.472271074772683e-05,
+      "loss": 0.4185,
+      "step": 270
+    },
+    {
+      "epoch": 0.820279984865683,
+      "grad_norm": 0.6790362026590407,
+      "learning_rate": 1.4676025932895315e-05,
+      "loss": 0.4066,
+      "step": 271
+    },
+    {
+      "epoch": 0.8233068482784714,
+      "grad_norm": 0.7739279330664569,
+      "learning_rate": 1.4629210319566626e-05,
+      "loss": 0.408,
+      "step": 272
+    },
+    {
+      "epoch": 0.8263337116912599,
+      "grad_norm": 0.6958729561966952,
+      "learning_rate": 1.4582265217274105e-05,
+      "loss": 0.4217,
+      "step": 273
+    },
+    {
+      "epoch": 0.8293605751040485,
+      "grad_norm": 0.7218792900005002,
+      "learning_rate": 1.4535191939173179e-05,
+      "loss": 0.4152,
+      "step": 274
+    },
+    {
+      "epoch": 0.8323874385168369,
+      "grad_norm": 0.7056662139566339,
+      "learning_rate": 1.4487991802004625e-05,
+      "loss": 0.419,
+      "step": 275
+    },
+    {
+      "epoch": 0.8354143019296254,
+      "grad_norm": 0.7243290267095895,
+      "learning_rate": 1.4440666126057743e-05,
+      "loss": 0.4226,
+      "step": 276
+    },
+    {
+      "epoch": 0.838441165342414,
+      "grad_norm": 0.9116703642860973,
+      "learning_rate": 1.4393216235133427e-05,
+      "loss": 0.4167,
+      "step": 277
+    },
+    {
+      "epoch": 0.8414680287552024,
+      "grad_norm": 0.6913104899347299,
+      "learning_rate": 1.4345643456507126e-05,
+      "loss": 0.4166,
+      "step": 278
+    },
+    {
+      "epoch": 0.8444948921679909,
+      "grad_norm": 0.7780194594315958,
+      "learning_rate": 1.4297949120891718e-05,
+      "loss": 0.4017,
+      "step": 279
+    },
+    {
+      "epoch": 0.8475217555807795,
+      "grad_norm": 0.7135874824236577,
+      "learning_rate": 1.4250134562400301e-05,
+      "loss": 0.4296,
+      "step": 280
+    },
+    {
+      "epoch": 0.8505486189935679,
+      "grad_norm": 0.6650307696736724,
+      "learning_rate": 1.4202201118508863e-05,
+      "loss": 0.4106,
+      "step": 281
+    },
+    {
+      "epoch": 0.8535754824063564,
+      "grad_norm": 0.7230912742869076,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.4265,
+      "step": 282
+    },
+    {
+      "epoch": 0.856602345819145,
+      "grad_norm": 0.7872017402026339,
+      "learning_rate": 1.4105982941019751e-05,
+      "loss": 0.4474,
+      "step": 283
+    },
+    {
+      "epoch": 0.8596292092319334,
+      "grad_norm": 0.8170141175509318,
+      "learning_rate": 1.405770089885134e-05,
+      "loss": 0.4351,
+      "step": 284
+    },
+    {
+      "epoch": 0.8626560726447219,
+      "grad_norm": 0.9076333196650805,
+      "learning_rate": 1.4009305354066138e-05,
+      "loss": 0.4433,
+      "step": 285
+    },
+    {
+      "epoch": 0.8656829360575105,
+      "grad_norm": 0.7652824667105684,
+      "learning_rate": 1.396079766039157e-05,
+      "loss": 0.3869,
+      "step": 286
+    },
+    {
+      "epoch": 0.8687097994702989,
+      "grad_norm": 0.9185459500767443,
+      "learning_rate": 1.39121791746921e-05,
+      "loss": 0.4398,
+      "step": 287
+    },
+    {
+      "epoch": 0.8717366628830874,
+      "grad_norm": 0.683856092519685,
+      "learning_rate": 1.3863451256931286e-05,
+      "loss": 0.3957,
+      "step": 288
+    },
+    {
+      "epoch": 0.874763526295876,
+      "grad_norm": 0.87356169351691,
+      "learning_rate": 1.381461527013374e-05,
+      "loss": 0.4219,
+      "step": 289
+    },
+    {
+      "epoch": 0.8777903897086644,
+      "grad_norm": 0.7629226280417597,
+      "learning_rate": 1.3765672580346986e-05,
+      "loss": 0.4352,
+      "step": 290
+    },
+    {
+      "epoch": 0.8808172531214529,
+      "grad_norm": 0.8631834212542296,
+      "learning_rate": 1.3716624556603275e-05,
+      "loss": 0.4347,
+      "step": 291
+    },
+    {
+      "epoch": 0.8838441165342414,
+      "grad_norm": 0.8010682582248608,
+      "learning_rate": 1.3667472570881264e-05,
+      "loss": 0.4358,
+      "step": 292
+    },
+    {
+      "epoch": 0.8868709799470299,
+      "grad_norm": 0.6976119432685854,
+      "learning_rate": 1.361821799806765e-05,
+      "loss": 0.4219,
+      "step": 293
+    },
+    {
+      "epoch": 0.8898978433598184,
+      "grad_norm": 0.8420218111264165,
+      "learning_rate": 1.356886221591872e-05,
+      "loss": 0.4229,
+      "step": 294
+    },
+    {
+      "epoch": 0.8929247067726069,
+      "grad_norm": 0.7605670845613539,
+      "learning_rate": 1.3519406605021797e-05,
+      "loss": 0.413,
+      "step": 295
+    },
+    {
+      "epoch": 0.8959515701853954,
+      "grad_norm": 0.7083126933963982,
+      "learning_rate": 1.3469852548756626e-05,
+      "loss": 0.4132,
+      "step": 296
+    },
+    {
+      "epoch": 0.8989784335981839,
+      "grad_norm": 0.726850081451116,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.4459,
+      "step": 297
+    },
+    {
+      "epoch": 0.9020052970109724,
+      "grad_norm": 0.7566549446693742,
+      "learning_rate": 1.3370454647370418e-05,
+      "loss": 0.4283,
+      "step": 298
+    },
+    {
+      "epoch": 0.9050321604237609,
+      "grad_norm": 0.7463337161504136,
+      "learning_rate": 1.3320613582622354e-05,
+      "loss": 0.4204,
+      "step": 299
+    },
+    {
+      "epoch": 0.9080590238365494,
+      "grad_norm": 0.7097766877131763,
+      "learning_rate": 1.3270679633174219e-05,
+      "loss": 0.4277,
+      "step": 300
+    },
+    {
+      "epoch": 0.9110858872493379,
+      "grad_norm": 0.8176773766544522,
+      "learning_rate": 1.3220654195785917e-05,
+      "loss": 0.4575,
+      "step": 301
+    },
+    {
+      "epoch": 0.9141127506621264,
+      "grad_norm": 0.9219923771505071,
+      "learning_rate": 1.3170538669776469e-05,
+      "loss": 0.4334,
+      "step": 302
+    },
+    {
+      "epoch": 0.9171396140749148,
+      "grad_norm": 0.7351748712320839,
+      "learning_rate": 1.3120334456984871e-05,
+      "loss": 0.413,
+      "step": 303
+    },
+    {
+      "epoch": 0.9201664774877034,
+      "grad_norm": 0.7825100363347536,
+      "learning_rate": 1.3070042961730878e-05,
+      "loss": 0.4206,
+      "step": 304
+    },
+    {
+      "epoch": 0.9231933409004919,
+      "grad_norm": 0.6735083502158334,
+      "learning_rate": 1.3019665590775717e-05,
+      "loss": 0.3941,
+      "step": 305
+    },
+    {
+      "epoch": 0.9262202043132803,
+      "grad_norm": 0.8690681440679482,
+      "learning_rate": 1.296920375328275e-05,
+      "loss": 0.4347,
+      "step": 306
+    },
+    {
+      "epoch": 0.9292470677260689,
+      "grad_norm": 0.752863424640265,
+      "learning_rate": 1.2918658860778046e-05,
+      "loss": 0.4191,
+      "step": 307
+    },
+    {
+      "epoch": 0.9322739311388574,
+      "grad_norm": 0.7336037657454644,
+      "learning_rate": 1.2868032327110904e-05,
+      "loss": 0.4135,
+      "step": 308
+    },
+    {
+      "epoch": 0.9353007945516458,
+      "grad_norm": 0.8033102578684338,
+      "learning_rate": 1.2817325568414299e-05,
+      "loss": 0.4162,
+      "step": 309
+    },
+    {
+      "epoch": 0.9383276579644344,
+      "grad_norm": 0.6947098316627842,
+      "learning_rate": 1.2766540003065272e-05,
+      "loss": 0.4062,
+      "step": 310
+    },
+    {
+      "epoch": 0.9413545213772229,
+      "grad_norm": 0.85057880337841,
+      "learning_rate": 1.2715677051645259e-05,
+      "loss": 0.422,
+      "step": 311
+    },
+    {
+      "epoch": 0.9443813847900113,
+      "grad_norm": 0.7118721401571699,
+      "learning_rate": 1.266473813690035e-05,
+      "loss": 0.4241,
+      "step": 312
+    },
+    {
+      "epoch": 0.9474082482027999,
+      "grad_norm": 0.7197445289742042,
+      "learning_rate": 1.2613724683701491e-05,
+      "loss": 0.4244,
+      "step": 313
+    },
+    {
+      "epoch": 0.9504351116155884,
+      "grad_norm": 0.8341867578235612,
+      "learning_rate": 1.2562638119004627e-05,
+      "loss": 0.4172,
+      "step": 314
+    },
+    {
+      "epoch": 0.9534619750283768,
+      "grad_norm": 0.7179575929786518,
+      "learning_rate": 1.2511479871810792e-05,
+      "loss": 0.4352,
+      "step": 315
+    },
+    {
+      "epoch": 0.9564888384411654,
+      "grad_norm": 0.746800585578556,
+      "learning_rate": 1.2460251373126136e-05,
+      "loss": 0.4001,
+      "step": 316
+    },
+    {
+      "epoch": 0.9595157018539539,
+      "grad_norm": 0.7480516491648973,
+      "learning_rate": 1.2408954055921884e-05,
+      "loss": 0.4219,
+      "step": 317
+    },
+    {
+      "epoch": 0.9625425652667423,
+      "grad_norm": 0.6947227466601075,
+      "learning_rate": 1.2357589355094275e-05,
+      "loss": 0.4088,
+      "step": 318
+    },
+    {
+      "epoch": 0.9655694286795309,
+      "grad_norm": 0.6802161432755544,
+      "learning_rate": 1.2306158707424402e-05,
+      "loss": 0.4126,
+      "step": 319
+    },
+    {
+      "epoch": 0.9685962920923193,
+      "grad_norm": 0.6559511391299027,
+      "learning_rate": 1.2254663551538047e-05,
+      "loss": 0.4069,
+      "step": 320
+    },
+    {
+      "epoch": 0.9716231555051078,
+      "grad_norm": 0.6904309448352038,
+      "learning_rate": 1.2203105327865407e-05,
+      "loss": 0.4434,
+      "step": 321
+    },
+    {
+      "epoch": 0.9746500189178964,
+      "grad_norm": 0.7182868452223119,
+      "learning_rate": 1.215148547860084e-05,
+      "loss": 0.4349,
+      "step": 322
+    },
+    {
+      "epoch": 0.9776768823306848,
+      "grad_norm": 0.7814920803954862,
+      "learning_rate": 1.2099805447662485e-05,
+      "loss": 0.393,
+      "step": 323
+    },
+    {
+      "epoch": 0.9807037457434733,
+      "grad_norm": 0.6949386704179543,
+      "learning_rate": 1.2048066680651908e-05,
+      "loss": 0.411,
+      "step": 324
+    },
+    {
+      "epoch": 0.9837306091562619,
+      "grad_norm": 0.6381648381346965,
+      "learning_rate": 1.1996270624813642e-05,
+      "loss": 0.4098,
+      "step": 325
+    },
+    {
+      "epoch": 0.9867574725690503,
+      "grad_norm": 0.7427466598485836,
+      "learning_rate": 1.194441872899471e-05,
+      "loss": 0.4043,
+      "step": 326
+    },
+    {
+      "epoch": 0.9897843359818388,
+      "grad_norm": 0.6621844709178613,
+      "learning_rate": 1.1892512443604103e-05,
+      "loss": 0.4002,
+      "step": 327
+    },
+    {
+      "epoch": 0.9928111993946274,
+      "grad_norm": 0.7912473266303484,
+      "learning_rate": 1.1840553220572204e-05,
+      "loss": 0.4286,
+      "step": 328
+    },
+    {
+      "epoch": 0.9958380628074158,
+      "grad_norm": 0.7011746504431725,
+      "learning_rate": 1.1788542513310178e-05,
+      "loss": 0.4158,
+      "step": 329
+    },
+    {
+      "epoch": 0.9988649262202043,
+      "grad_norm": 0.6704473383696113,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.4317,
+      "step": 330
+    },
+    {
+      "epoch": 1.0030268634127886,
+      "grad_norm": 0.9056114324326852,
+      "learning_rate": 1.1684372466900306e-05,
+      "loss": 0.3097,
+      "step": 331
+    },
+    {
+      "epoch": 1.006053726825577,
+      "grad_norm": 0.8530657966696973,
+      "learning_rate": 1.1632216041612595e-05,
+      "loss": 0.3115,
+      "step": 332
+    },
+    {
+      "epoch": 1.0090805902383655,
+      "grad_norm": 0.7101116681619808,
+      "learning_rate": 1.15800139597335e-05,
+      "loss": 0.3154,
+      "step": 333
+    },
+    {
+      "epoch": 1.012107453651154,
+      "grad_norm": 0.6488846749256443,
+      "learning_rate": 1.1527767681467472e-05,
+      "loss": 0.3083,
+      "step": 334
+    },
+    {
+      "epoch": 1.0151343170639424,
+      "grad_norm": 0.7966833609718857,
+      "learning_rate": 1.1475478668255223e-05,
+      "loss": 0.3025,
+      "step": 335
+    },
+    {
+      "epoch": 1.018161180476731,
+      "grad_norm": 0.7997274017116879,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.3016,
+      "step": 336
+    },
+    {
+      "epoch": 1.0211880438895196,
+      "grad_norm": 0.8706465996902483,
+      "learning_rate": 1.1370778288690947e-05,
+      "loss": 0.3078,
+      "step": 337
+    },
+    {
+      "epoch": 1.024214907302308,
+      "grad_norm": 0.8388383356557727,
+      "learning_rate": 1.1318369851033604e-05,
+      "loss": 0.2942,
+      "step": 338
+    },
+    {
+      "epoch": 1.0272417707150965,
+      "grad_norm": 0.7743256303109989,
+      "learning_rate": 1.1265924535737494e-05,
+      "loss": 0.3026,
+      "step": 339
+    },
+    {
+      "epoch": 1.030268634127885,
+      "grad_norm": 0.8122251757192691,
+      "learning_rate": 1.121344380981082e-05,
+      "loss": 0.3125,
+      "step": 340
+    },
+    {
+      "epoch": 1.0332954975406734,
+      "grad_norm": 0.8351870356710823,
+      "learning_rate": 1.1160929141252303e-05,
+      "loss": 0.3037,
+      "step": 341
+    },
+    {
+      "epoch": 1.036322360953462,
+      "grad_norm": 0.7582036188225919,
+      "learning_rate": 1.1108381999010111e-05,
+      "loss": 0.3067,
+      "step": 342
+    },
+    {
+      "epoch": 1.0393492243662505,
+      "grad_norm": 0.945337804023826,
+      "learning_rate": 1.1055803852940772e-05,
+      "loss": 0.2861,
+      "step": 343
+    },
+    {
+      "epoch": 1.042376087779039,
+      "grad_norm": 0.8048903461691072,
+      "learning_rate": 1.1003196173768051e-05,
+      "loss": 0.2979,
+      "step": 344
+    },
+    {
+      "epoch": 1.0454029511918275,
+      "grad_norm": 0.8167878527427705,
+      "learning_rate": 1.0950560433041825e-05,
+      "loss": 0.3077,
+      "step": 345
+    },
+    {
+      "epoch": 1.048429814604616,
+      "grad_norm": 0.8037778021136479,
+      "learning_rate": 1.0897898103096917e-05,
+      "loss": 0.3012,
+      "step": 346
+    },
+    {
+      "epoch": 1.0514566780174044,
+      "grad_norm": 0.7839978759654623,
+      "learning_rate": 1.0845210657011893e-05,
+      "loss": 0.3031,
+      "step": 347
+    },
+    {
+      "epoch": 1.054483541430193,
+      "grad_norm": 0.8173483126490009,
+      "learning_rate": 1.0792499568567885e-05,
+      "loss": 0.303,
+      "step": 348
+    },
+    {
+      "epoch": 1.0575104048429815,
+      "grad_norm": 0.7894527703382768,
+      "learning_rate": 1.0739766312207344e-05,
+      "loss": 0.3112,
+      "step": 349
+    },
+    {
+      "epoch": 1.0605372682557699,
+      "grad_norm": 0.7667546212318844,
+      "learning_rate": 1.068701236299281e-05,
+      "loss": 0.2999,
+      "step": 350
+    },
+    {
+      "epoch": 1.0635641316685585,
+      "grad_norm": 0.7777270351047659,
+      "learning_rate": 1.0634239196565646e-05,
+      "loss": 0.3059,
+      "step": 351
+    },
+    {
+      "epoch": 1.066590995081347,
+      "grad_norm": 0.7337030677894119,
+      "learning_rate": 1.0581448289104759e-05,
+      "loss": 0.3019,
+      "step": 352
+    },
+    {
+      "epoch": 1.0696178584941354,
+      "grad_norm": 0.744014283833164,
+      "learning_rate": 1.0528641117285315e-05,
+      "loss": 0.3085,
+      "step": 353
+    },
+    {
+      "epoch": 1.072644721906924,
+      "grad_norm": 0.7307665672362731,
+      "learning_rate": 1.0475819158237426e-05,
+      "loss": 0.3031,
+      "step": 354
+    },
+    {
+      "epoch": 1.0756715853197125,
+      "grad_norm": 0.721771354976752,
+      "learning_rate": 1.0422983889504831e-05,
+      "loss": 0.3006,
+      "step": 355
+    },
+    {
+      "epoch": 1.0786984487325009,
+      "grad_norm": 0.7197419718483739,
+      "learning_rate": 1.0370136789003582e-05,
+      "loss": 0.3154,
+      "step": 356
+    },
+    {
+      "epoch": 1.0817253121452894,
+      "grad_norm": 0.6930720363551771,
+      "learning_rate": 1.031727933498068e-05,
+      "loss": 0.292,
+      "step": 357
+    },
+    {
+      "epoch": 1.084752175558078,
+      "grad_norm": 0.756567737097623,
+      "learning_rate": 1.0264413005972736e-05,
+      "loss": 0.302,
+      "step": 358
+    },
+    {
+      "epoch": 1.0877790389708664,
+      "grad_norm": 0.7287513135304579,
+      "learning_rate": 1.0211539280764617e-05,
+      "loss": 0.3097,
+      "step": 359
+    },
+    {
+      "epoch": 1.090805902383655,
+      "grad_norm": 0.7545678474652777,
+      "learning_rate": 1.015865963834808e-05,
+      "loss": 0.3018,
+      "step": 360
+    },
+    {
+      "epoch": 1.0938327657964435,
+      "grad_norm": 0.7362187870202994,
+      "learning_rate": 1.0105775557880398e-05,
+      "loss": 0.3028,
+      "step": 361
+    },
+    {
+      "epoch": 1.0968596292092319,
+      "grad_norm": 0.7738766420819987,
+      "learning_rate": 1.0052888518642978e-05,
+      "loss": 0.3031,
+      "step": 362
+    },
+    {
+      "epoch": 1.0998864926220204,
+      "grad_norm": 0.6483276863743285,
+      "learning_rate": 1e-05,
+      "loss": 0.2788,
+      "step": 363
+    },
+    {
+      "epoch": 1.102913356034809,
+      "grad_norm": 0.7085221964890713,
+      "learning_rate": 9.947111481357023e-06,
+      "loss": 0.2918,
+      "step": 364
+    },
+    {
+      "epoch": 1.1059402194475974,
+      "grad_norm": 0.7625098354588125,
+      "learning_rate": 9.894224442119606e-06,
+      "loss": 0.3123,
+      "step": 365
+    },
+    {
+      "epoch": 1.108967082860386,
+      "grad_norm": 0.719688121865531,
+      "learning_rate": 9.841340361651921e-06,
+      "loss": 0.2933,
+      "step": 366
+    },
+    {
+      "epoch": 1.1119939462731745,
+      "grad_norm": 0.750560822944283,
+      "learning_rate": 9.788460719235386e-06,
+      "loss": 0.2966,
+      "step": 367
+    },
+    {
+      "epoch": 1.1150208096859628,
+      "grad_norm": 0.7447605168139636,
+      "learning_rate": 9.735586994027267e-06,
+      "loss": 0.311,
+      "step": 368
+    },
+    {
+      "epoch": 1.1180476730987514,
+      "grad_norm": 0.7384100418059716,
+      "learning_rate": 9.682720665019325e-06,
+      "loss": 0.2982,
+      "step": 369
+    },
+    {
+      "epoch": 1.12107453651154,
+      "grad_norm": 0.7413671197529755,
+      "learning_rate": 9.62986321099642e-06,
+      "loss": 0.2941,
+      "step": 370
+    },
+    {
+      "epoch": 1.1241013999243283,
+      "grad_norm": 0.7589280728066173,
+      "learning_rate": 9.57701611049517e-06,
+      "loss": 0.2944,
+      "step": 371
+    },
+    {
+      "epoch": 1.127128263337117,
+      "grad_norm": 0.763299391635082,
+      "learning_rate": 9.524180841762577e-06,
+      "loss": 0.306,
+      "step": 372
+    },
+    {
+      "epoch": 1.1301551267499055,
+      "grad_norm": 0.8110476425603581,
+      "learning_rate": 9.471358882714687e-06,
+      "loss": 0.2916,
+      "step": 373
+    },
+    {
+      "epoch": 1.1331819901626938,
+      "grad_norm": 0.754620993409508,
+      "learning_rate": 9.418551710895243e-06,
+      "loss": 0.3094,
+      "step": 374
+    },
+    {
+      "epoch": 1.1362088535754824,
+      "grad_norm": 0.687571159798026,
+      "learning_rate": 9.365760803434356e-06,
+      "loss": 0.2881,
+      "step": 375
+    },
+    {
+      "epoch": 1.139235716988271,
+      "grad_norm": 0.8403887107925709,
+      "learning_rate": 9.312987637007191e-06,
+      "loss": 0.3022,
+      "step": 376
+    },
+    {
+      "epoch": 1.1422625804010593,
+      "grad_norm": 0.7279418281704557,
+      "learning_rate": 9.260233687792657e-06,
+      "loss": 0.2907,
+      "step": 377
+    },
+    {
+      "epoch": 1.145289443813848,
+      "grad_norm": 0.8223954170263438,
+      "learning_rate": 9.207500431432115e-06,
+      "loss": 0.2989,
+      "step": 378
+    },
+    {
+      "epoch": 1.1483163072266365,
+      "grad_norm": 0.792893618619976,
+      "learning_rate": 9.154789342988108e-06,
+      "loss": 0.3071,
+      "step": 379
+    },
+    {
+      "epoch": 1.1513431706394248,
+      "grad_norm": 0.7358869242349388,
+      "learning_rate": 9.102101896903084e-06,
+      "loss": 0.3098,
+      "step": 380
+    },
+    {
+      "epoch": 1.1543700340522134,
+      "grad_norm": 0.8298249010965482,
+      "learning_rate": 9.049439566958176e-06,
+      "loss": 0.3116,
+      "step": 381
+    },
+    {
+      "epoch": 1.157396897465002,
+      "grad_norm": 0.8138247056909561,
+      "learning_rate": 8.99680382623195e-06,
+      "loss": 0.2824,
+      "step": 382
+    },
+    {
+      "epoch": 1.1604237608777903,
+      "grad_norm": 0.7690523126623674,
+      "learning_rate": 8.944196147059233e-06,
+      "loss": 0.3018,
+      "step": 383
+    },
+    {
+      "epoch": 1.1634506242905789,
+      "grad_norm": 0.8008296656488894,
+      "learning_rate": 8.89161800098989e-06,
+      "loss": 0.3003,
+      "step": 384
+    },
+    {
+      "epoch": 1.1664774877033675,
+      "grad_norm": 0.7153608484140768,
+      "learning_rate": 8.839070858747697e-06,
+      "loss": 0.305,
+      "step": 385
+    },
+    {
+      "epoch": 1.1695043511161558,
+      "grad_norm": 0.6816796706501214,
+      "learning_rate": 8.786556190189183e-06,
+      "loss": 0.2788,
+      "step": 386
+    },
+    {
+      "epoch": 1.1725312145289444,
+      "grad_norm": 0.7823169151002916,
+      "learning_rate": 8.734075464262507e-06,
+      "loss": 0.2977,
+      "step": 387
+    },
+    {
+      "epoch": 1.175558077941733,
+      "grad_norm": 0.7261528305196607,
+      "learning_rate": 8.681630148966397e-06,
+      "loss": 0.2886,
+      "step": 388
+    },
+    {
+      "epoch": 1.1785849413545213,
+      "grad_norm": 0.7365784987639206,
+      "learning_rate": 8.629221711309056e-06,
+      "loss": 0.2848,
+      "step": 389
+    },
+    {
+      "epoch": 1.1816118047673099,
+      "grad_norm": 0.7499747278448416,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.2888,
+      "step": 390
+    },
+    {
+      "epoch": 1.1846386681800984,
+      "grad_norm": 0.8219867361742994,
+      "learning_rate": 8.52452133174478e-06,
+      "loss": 0.2762,
+      "step": 391
+    },
+    {
+      "epoch": 1.1876655315928868,
+      "grad_norm": 0.7554920110713274,
+      "learning_rate": 8.472232318532531e-06,
+      "loss": 0.3025,
+      "step": 392
+    },
+    {
+      "epoch": 1.1906923950056754,
+      "grad_norm": 0.8195298248798789,
+      "learning_rate": 8.419986040266502e-06,
+      "loss": 0.291,
+      "step": 393
+    },
+    {
+      "epoch": 1.193719258418464,
+      "grad_norm": 0.7562126836406086,
+      "learning_rate": 8.367783958387407e-06,
+      "loss": 0.2875,
+      "step": 394
+    },
+    {
+      "epoch": 1.1967461218312523,
+      "grad_norm": 0.7394157348514436,
+      "learning_rate": 8.315627533099697e-06,
+      "loss": 0.2943,
+      "step": 395
+    },
+    {
+      "epoch": 1.1997729852440409,
+      "grad_norm": 0.7534833583318963,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.2979,
+      "step": 396
+    },
+    {
+      "epoch": 1.2027998486568294,
+      "grad_norm": 0.7693687927357046,
+      "learning_rate": 8.211457486689829e-06,
+      "loss": 0.3063,
+      "step": 397
+    },
+    {
+      "epoch": 1.2058267120696178,
+      "grad_norm": 0.7067736938862111,
+      "learning_rate": 8.159446779427798e-06,
+      "loss": 0.2946,
+      "step": 398
+    },
+    {
+      "epoch": 1.2088535754824064,
+      "grad_norm": 0.7363955653846156,
+      "learning_rate": 8.107487556395902e-06,
+      "loss": 0.2936,
+      "step": 399
+    },
+    {
+      "epoch": 1.211880438895195,
+      "grad_norm": 0.8480638019377852,
+      "learning_rate": 8.055581271005292e-06,
+      "loss": 0.3138,
+      "step": 400
+    },
+    {
+      "epoch": 1.2149073023079833,
+      "grad_norm": 0.7245448626432517,
+      "learning_rate": 8.00372937518636e-06,
+      "loss": 0.2892,
+      "step": 401
+    },
+    {
+      "epoch": 1.2179341657207718,
+      "grad_norm": 0.7994323884630878,
+      "learning_rate": 7.951933319348095e-06,
+      "loss": 0.2991,
+      "step": 402
+    },
+    {
+      "epoch": 1.2209610291335604,
+      "grad_norm": 0.723140744868228,
+      "learning_rate": 7.900194552337516e-06,
+      "loss": 0.2917,
+      "step": 403
+    },
+    {
+      "epoch": 1.2239878925463488,
+      "grad_norm": 0.7004181106477327,
+      "learning_rate": 7.848514521399167e-06,
+      "loss": 0.2944,
+      "step": 404
+    },
+    {
+      "epoch": 1.2270147559591373,
+      "grad_norm": 0.6786102420677808,
+      "learning_rate": 7.796894672134594e-06,
+      "loss": 0.2844,
+      "step": 405
+    },
+    {
+      "epoch": 1.230041619371926,
+      "grad_norm": 0.7292211657839189,
+      "learning_rate": 7.745336448461958e-06,
+      "loss": 0.2964,
+      "step": 406
+    },
+    {
+      "epoch": 1.2330684827847143,
+      "grad_norm": 0.7673864766297774,
+      "learning_rate": 7.6938412925756e-06,
+      "loss": 0.2938,
+      "step": 407
+    },
+    {
+      "epoch": 1.2360953461975028,
+      "grad_norm": 0.7583643773575292,
+      "learning_rate": 7.642410644905726e-06,
+      "loss": 0.2886,
+      "step": 408
+    },
+    {
+      "epoch": 1.2391222096102914,
+      "grad_norm": 0.7188229966634913,
+      "learning_rate": 7.591045944078119e-06,
+      "loss": 0.283,
+      "step": 409
+    },
+    {
+      "epoch": 1.2421490730230798,
+      "grad_norm": 0.7760179689777242,
+      "learning_rate": 7.539748626873866e-06,
+      "loss": 0.3055,
+      "step": 410
+    },
+    {
+      "epoch": 1.2451759364358683,
+      "grad_norm": 0.8005488545888585,
+      "learning_rate": 7.488520128189209e-06,
+      "loss": 0.3137,
+      "step": 411
+    },
+    {
+      "epoch": 1.248202799848657,
+      "grad_norm": 0.6956489887848328,
+      "learning_rate": 7.4373618809953755e-06,
+      "loss": 0.2856,
+      "step": 412
+    },
+    {
+      "epoch": 1.2512296632614452,
+      "grad_norm": 0.7985830014430589,
+      "learning_rate": 7.386275316298513e-06,
+      "loss": 0.3008,
+      "step": 413
+    },
+    {
+      "epoch": 1.2542565266742338,
+      "grad_norm": 0.7086171033345939,
+      "learning_rate": 7.335261863099652e-06,
+      "loss": 0.2969,
+      "step": 414
+    },
+    {
+      "epoch": 1.2572833900870224,
+      "grad_norm": 0.7824473459075301,
+      "learning_rate": 7.2843229483547405e-06,
+      "loss": 0.303,
+      "step": 415
+    },
+    {
+      "epoch": 1.2603102534998107,
+      "grad_norm": 0.709514387223654,
+      "learning_rate": 7.233459996934731e-06,
+      "loss": 0.3077,
+      "step": 416
+    },
+    {
+      "epoch": 1.2633371169125993,
+      "grad_norm": 0.7530320837964829,
+      "learning_rate": 7.182674431585703e-06,
+      "loss": 0.3034,
+      "step": 417
+    },
+    {
+      "epoch": 1.2663639803253879,
+      "grad_norm": 0.7315529183474975,
+      "learning_rate": 7.131967672889101e-06,
+      "loss": 0.2897,
+      "step": 418
+    },
+    {
+      "epoch": 1.2693908437381762,
+      "grad_norm": 0.7420913112141836,
+      "learning_rate": 7.081341139221955e-06,
+      "loss": 0.3221,
+      "step": 419
+    },
+    {
+      "epoch": 1.2724177071509648,
+      "grad_norm": 0.7103051952108638,
+      "learning_rate": 7.0307962467172555e-06,
+      "loss": 0.288,
+      "step": 420
+    },
+    {
+      "epoch": 1.2754445705637534,
+      "grad_norm": 0.8047381899219737,
+      "learning_rate": 6.9803344092242855e-06,
+      "loss": 0.3275,
+      "step": 421
+    },
+    {
+      "epoch": 1.2784714339765417,
+      "grad_norm": 0.7104822153733231,
+      "learning_rate": 6.929957038269123e-06,
+      "loss": 0.297,
+      "step": 422
+    },
+    {
+      "epoch": 1.2814982973893303,
+      "grad_norm": 0.7231783380360892,
+      "learning_rate": 6.87966554301513e-06,
+      "loss": 0.2891,
+      "step": 423
+    },
+    {
+      "epoch": 1.2845251608021189,
+      "grad_norm": 0.7518467188516564,
+      "learning_rate": 6.8294613302235325e-06,
+      "loss": 0.2901,
+      "step": 424
+    },
+    {
+      "epoch": 1.2875520242149072,
+      "grad_norm": 0.7713572679487606,
+      "learning_rate": 6.779345804214088e-06,
+      "loss": 0.3006,
+      "step": 425
+    },
+    {
+      "epoch": 1.2905788876276958,
+      "grad_norm": 0.7859497128457842,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 0.2844,
+      "step": 426
+    },
+    {
+      "epoch": 1.2936057510404844,
+      "grad_norm": 0.7411935925908989,
+      "learning_rate": 6.679386417377649e-06,
+      "loss": 0.3014,
+      "step": 427
+    },
+    {
+      "epoch": 1.2966326144532727,
+      "grad_norm": 0.7337313145909549,
+      "learning_rate": 6.629545352629583e-06,
+      "loss": 0.2895,
+      "step": 428
+    },
+    {
+      "epoch": 1.2996594778660613,
+      "grad_norm": 0.7020629155983285,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.2897,
+      "step": 429
+    },
+    {
+      "epoch": 1.3026863412788499,
+      "grad_norm": 0.7690957535592647,
+      "learning_rate": 6.530147451243377e-06,
+      "loss": 0.3076,
+      "step": 430
+    },
+    {
+      "epoch": 1.3057132046916382,
+      "grad_norm": 0.7363293747525321,
+      "learning_rate": 6.480593394978208e-06,
+      "loss": 0.2953,
+      "step": 431
+    },
+    {
+      "epoch": 1.3087400681044268,
+      "grad_norm": 0.7428197272387507,
+      "learning_rate": 6.431137784081283e-06,
+      "loss": 0.2945,
+      "step": 432
+    },
+    {
+      "epoch": 1.3117669315172154,
+      "grad_norm": 0.6998116149373447,
+      "learning_rate": 6.381782001932352e-06,
+      "loss": 0.2873,
+      "step": 433
+    },
+    {
+      "epoch": 1.3147937949300037,
+      "grad_norm": 0.737160826561709,
+      "learning_rate": 6.33252742911874e-06,
+      "loss": 0.3003,
+      "step": 434
+    },
+    {
+      "epoch": 1.3178206583427923,
+      "grad_norm": 0.7461756730788723,
+      "learning_rate": 6.283375443396726e-06,
+      "loss": 0.3067,
+      "step": 435
+    },
+    {
+      "epoch": 1.3208475217555808,
+      "grad_norm": 0.7381979741046565,
+      "learning_rate": 6.234327419653013e-06,
+      "loss": 0.2998,
+      "step": 436
+    },
+    {
+      "epoch": 1.3238743851683692,
+      "grad_norm": 0.8323458597686157,
+      "learning_rate": 6.185384729866264e-06,
+      "loss": 0.2869,
+      "step": 437
+    },
+    {
+      "epoch": 1.3269012485811578,
+      "grad_norm": 0.7505741136765467,
+      "learning_rate": 6.136548743068713e-06,
+      "loss": 0.3058,
+      "step": 438
+    },
+    {
+      "epoch": 1.3299281119939463,
+      "grad_norm": 0.7255593477792327,
+      "learning_rate": 6.087820825307904e-06,
+      "loss": 0.2913,
+      "step": 439
+    },
+    {
+      "epoch": 1.3329549754067347,
+      "grad_norm": 0.8552522766982229,
+      "learning_rate": 6.039202339608432e-06,
+      "loss": 0.2954,
+      "step": 440
+    },
+    {
+      "epoch": 1.3359818388195233,
+      "grad_norm": 0.7278177679141681,
+      "learning_rate": 5.990694645933866e-06,
+      "loss": 0.2894,
+      "step": 441
+    },
+    {
+      "epoch": 1.3390087022323118,
+      "grad_norm": 0.7286476935215261,
+      "learning_rate": 5.9422991011486635e-06,
+      "loss": 0.2773,
+      "step": 442
+    },
+    {
+      "epoch": 1.3420355656451002,
+      "grad_norm": 1.0203950174650196,
+      "learning_rate": 5.894017058980249e-06,
+      "loss": 0.2998,
+      "step": 443
+    },
+    {
+      "epoch": 1.3450624290578888,
+      "grad_norm": 0.6851268629938885,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.2733,
+      "step": 444
+    },
+    {
+      "epoch": 1.3480892924706773,
+      "grad_norm": 0.7329561855268578,
+      "learning_rate": 5.797798881491138e-06,
+      "loss": 0.2938,
+      "step": 445
+    },
+    {
+      "epoch": 1.351116155883466,
+      "grad_norm": 0.8008789233196687,
+      "learning_rate": 5.749865437599703e-06,
+      "loss": 0.3056,
+      "step": 446
+    },
+    {
+      "epoch": 1.3541430192962542,
+      "grad_norm": 0.7333658987533364,
+      "learning_rate": 5.702050879108284e-06,
+      "loss": 0.2877,
+      "step": 447
+    },
+    {
+      "epoch": 1.3571698827090428,
+      "grad_norm": 0.7118878966067858,
+      "learning_rate": 5.654356543492883e-06,
+      "loss": 0.2854,
+      "step": 448
+    },
+    {
+      "epoch": 1.3601967461218312,
+      "grad_norm": 0.7165509071187739,
+      "learning_rate": 5.606783764866576e-06,
+      "loss": 0.2859,
+      "step": 449
+    },
+    {
+      "epoch": 1.3632236095346197,
+      "grad_norm": 0.8456937277267329,
+      "learning_rate": 5.559333873942259e-06,
+      "loss": 0.3045,
+      "step": 450
+    },
+    {
+      "epoch": 1.3662504729474083,
+      "grad_norm": 0.7800574945408677,
+      "learning_rate": 5.512008197995379e-06,
+      "loss": 0.2729,
+      "step": 451
+    },
+    {
+      "epoch": 1.3692773363601969,
+      "grad_norm": 0.733597268236208,
+      "learning_rate": 5.464808060826825e-06,
+      "loss": 0.3001,
+      "step": 452
+    },
+    {
+      "epoch": 1.3723041997729852,
+      "grad_norm": 0.7234405055039879,
+      "learning_rate": 5.417734782725896e-06,
+      "loss": 0.2889,
+      "step": 453
+    },
+    {
+      "epoch": 1.3753310631857738,
+      "grad_norm": 0.8571673435578278,
+      "learning_rate": 5.370789680433376e-06,
+      "loss": 0.2964,
+      "step": 454
+    },
+    {
+      "epoch": 1.3783579265985622,
+      "grad_norm": 0.7100899160075168,
+      "learning_rate": 5.323974067104687e-06,
+      "loss": 0.2877,
+      "step": 455
+    },
+    {
+      "epoch": 1.3813847900113507,
+      "grad_norm": 0.715614073428981,
+      "learning_rate": 5.277289252273175e-06,
+      "loss": 0.2961,
+      "step": 456
+    },
+    {
+      "epoch": 1.3844116534241393,
+      "grad_norm": 0.7120853571127941,
+      "learning_rate": 5.230736541813463e-06,
+      "loss": 0.2873,
+      "step": 457
+    },
+    {
+      "epoch": 1.3874385168369279,
+      "grad_norm": 0.8162581998562425,
+      "learning_rate": 5.184317237904939e-06,
+      "loss": 0.2966,
+      "step": 458
+    },
+    {
+      "epoch": 1.3904653802497162,
+      "grad_norm": 0.7224629289945345,
+      "learning_rate": 5.138032638995315e-06,
+      "loss": 0.2832,
+      "step": 459
+    },
+    {
+      "epoch": 1.3934922436625048,
+      "grad_norm": 0.7181607300724704,
+      "learning_rate": 5.091884039764321e-06,
+      "loss": 0.2787,
+      "step": 460
+    },
+    {
+      "epoch": 1.3965191070752931,
+      "grad_norm": 0.745235460337963,
+      "learning_rate": 5.045872731087479e-06,
+      "loss": 0.3076,
+      "step": 461
+    },
+    {
+      "epoch": 1.3995459704880817,
+      "grad_norm": 0.7059274832329271,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.2812,
+      "step": 462
+    },
+    {
+      "epoch": 1.4025728339008703,
+      "grad_norm": 0.7261226530275093,
+      "learning_rate": 4.954267129660789e-06,
+      "loss": 0.293,
+      "step": 463
+    },
+    {
+      "epoch": 1.4055996973136589,
+      "grad_norm": 0.7441870710723584,
+      "learning_rate": 4.908675399316534e-06,
+      "loss": 0.2915,
+      "step": 464
+    },
+    {
+      "epoch": 1.4086265607264472,
+      "grad_norm": 0.7016763070099846,
+      "learning_rate": 4.863226084265939e-06,
+      "loss": 0.2888,
+      "step": 465
+    },
+    {
+      "epoch": 1.4116534241392358,
+      "grad_norm": 0.7271051268722979,
+      "learning_rate": 4.817920455824045e-06,
+      "loss": 0.2917,
+      "step": 466
+    },
+    {
+      "epoch": 1.4146802875520241,
+      "grad_norm": 0.8513329715776123,
+      "learning_rate": 4.772759781286679e-06,
+      "loss": 0.2855,
+      "step": 467
+    },
+    {
+      "epoch": 1.4177071509648127,
+      "grad_norm": 0.7377943919805398,
+      "learning_rate": 4.727745323894976e-06,
+      "loss": 0.2752,
+      "step": 468
+    },
+    {
+      "epoch": 1.4207340143776013,
+      "grad_norm": 0.7599910722742809,
+      "learning_rate": 4.682878342800087e-06,
+      "loss": 0.276,
+      "step": 469
+    },
+    {
+      "epoch": 1.4237608777903898,
+      "grad_norm": 0.7456135123769635,
+      "learning_rate": 4.638160093027908e-06,
+      "loss": 0.2943,
+      "step": 470
+    },
+    {
+      "epoch": 1.4267877412031782,
+      "grad_norm": 0.7247590545628322,
+      "learning_rate": 4.593591825444028e-06,
+      "loss": 0.2832,
+      "step": 471
+    },
+    {
+      "epoch": 1.4298146046159668,
+      "grad_norm": 0.8152582694794308,
+      "learning_rate": 4.549174786718684e-06,
+      "loss": 0.2842,
+      "step": 472
+    },
+    {
+      "epoch": 1.4328414680287551,
+      "grad_norm": 0.7628957083900787,
+      "learning_rate": 4.504910219291941e-06,
+      "loss": 0.2986,
+      "step": 473
+    },
+    {
+      "epoch": 1.4358683314415437,
+      "grad_norm": 0.7097385828078482,
+      "learning_rate": 4.460799361338898e-06,
+      "loss": 0.2782,
+      "step": 474
+    },
+    {
+      "epoch": 1.4388951948543323,
+      "grad_norm": 0.6939480313334291,
+      "learning_rate": 4.416843446735077e-06,
+      "loss": 0.2899,
+      "step": 475
+    },
+    {
+      "epoch": 1.4419220582671208,
+      "grad_norm": 0.7601419650428849,
+      "learning_rate": 4.373043705021899e-06,
+      "loss": 0.2733,
+      "step": 476
+    },
+    {
+      "epoch": 1.4449489216799092,
+      "grad_norm": 0.747700547424619,
+      "learning_rate": 4.3294013613722944e-06,
+      "loss": 0.2898,
+      "step": 477
+    },
+    {
+      "epoch": 1.4479757850926978,
+      "grad_norm": 0.7321534176052829,
+      "learning_rate": 4.2859176365564294e-06,
+      "loss": 0.2989,
+      "step": 478
+    },
+    {
+      "epoch": 1.451002648505486,
+      "grad_norm": 0.6988411415930715,
+      "learning_rate": 4.2425937469075626e-06,
+      "loss": 0.2886,
+      "step": 479
+    },
+    {
+      "epoch": 1.4540295119182747,
+      "grad_norm": 0.7761365319117365,
+      "learning_rate": 4.19943090428802e-06,
+      "loss": 0.3139,
+      "step": 480
+    },
+    {
+      "epoch": 1.4570563753310632,
+      "grad_norm": 0.7065579979784431,
+      "learning_rate": 4.1564303160552935e-06,
+      "loss": 0.2917,
+      "step": 481
+    },
+    {
+      "epoch": 1.4600832387438518,
+      "grad_norm": 0.7570160815598899,
+      "learning_rate": 4.113593185028273e-06,
+      "loss": 0.2765,
+      "step": 482
+    },
+    {
+      "epoch": 1.4631101021566402,
+      "grad_norm": 0.7795028589365341,
+      "learning_rate": 4.070920709453597e-06,
+      "loss": 0.2941,
+      "step": 483
+    },
+    {
+      "epoch": 1.4661369655694287,
+      "grad_norm": 0.6828977534724469,
+      "learning_rate": 4.028414082972141e-06,
+      "loss": 0.2768,
+      "step": 484
+    },
+    {
+      "epoch": 1.469163828982217,
+      "grad_norm": 0.7548357869538868,
+      "learning_rate": 3.986074494585619e-06,
+      "loss": 0.3045,
+      "step": 485
+    },
+    {
+      "epoch": 1.4721906923950057,
+      "grad_norm": 0.7010855131732908,
+      "learning_rate": 3.943903128623336e-06,
+      "loss": 0.2759,
+      "step": 486
+    },
+    {
+      "epoch": 1.4752175558077942,
+      "grad_norm": 0.7113138520875217,
+      "learning_rate": 3.9019011647090465e-06,
+      "loss": 0.2968,
+      "step": 487
+    },
+    {
+      "epoch": 1.4782444192205828,
+      "grad_norm": 0.8304273303189595,
+      "learning_rate": 3.860069777727983e-06,
+      "loss": 0.2816,
+      "step": 488
+    },
+    {
+      "epoch": 1.4812712826333712,
+      "grad_norm": 0.6854195217020267,
+      "learning_rate": 3.818410137793947e-06,
+      "loss": 0.2874,
+      "step": 489
+    },
+    {
+      "epoch": 1.4842981460461597,
+      "grad_norm": 0.7782139117998569,
+      "learning_rate": 3.7769234102166365e-06,
+      "loss": 0.2908,
+      "step": 490
+    },
+    {
+      "epoch": 1.487325009458948,
+      "grad_norm": 0.7151336456391801,
+      "learning_rate": 3.735610755468988e-06,
+      "loss": 0.2947,
+      "step": 491
+    },
+    {
+      "epoch": 1.4903518728717366,
+      "grad_norm": 0.7493511967942955,
+      "learning_rate": 3.6944733291547784e-06,
+      "loss": 0.2905,
+      "step": 492
+    },
+    {
+      "epoch": 1.4933787362845252,
+      "grad_norm": 0.7999001632157942,
+      "learning_rate": 3.653512281976238e-06,
+      "loss": 0.2813,
+      "step": 493
+    },
+    {
+      "epoch": 1.4964055996973138,
+      "grad_norm": 0.7780555119581226,
+      "learning_rate": 3.612728759701919e-06,
+      "loss": 0.287,
+      "step": 494
+    },
+    {
+      "epoch": 1.4994324631101021,
+      "grad_norm": 0.6980813019684152,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.271,
+      "step": 495
+    },
+    {
+      "epoch": 1.5024593265228905,
+      "grad_norm": 0.7924304836800483,
+      "learning_rate": 3.5316988480794255e-06,
+      "loss": 0.3006,
+      "step": 496
+    },
+    {
+      "epoch": 1.505486189935679,
+      "grad_norm": 0.7764118578876771,
+      "learning_rate": 3.4914547253120655e-06,
+      "loss": 0.3003,
+      "step": 497
+    },
+    {
+      "epoch": 1.5085130533484676,
+      "grad_norm": 0.7233571636448777,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.2775,
+      "step": 498
+    },
+    {
+      "epoch": 1.5115399167612562,
+      "grad_norm": 0.7715212275906694,
+      "learning_rate": 3.4115137744067516e-06,
+      "loss": 0.2921,
+      "step": 499
+    },
+    {
+      "epoch": 1.5145667801740448,
+      "grad_norm": 0.7585703472864148,
+      "learning_rate": 3.37181918238904e-06,
+      "loss": 0.2988,
+      "step": 500
+    },
+    {
+      "epoch": 1.5175936435868331,
+      "grad_norm": 0.7373584233885836,
+      "learning_rate": 3.3323099948370853e-06,
+      "loss": 0.2947,
+      "step": 501
+    },
+    {
+      "epoch": 1.5206205069996215,
+      "grad_norm": 0.7024793202535262,
+      "learning_rate": 3.292987316907792e-06,
+      "loss": 0.2866,
+      "step": 502
+    },
+    {
+      "epoch": 1.52364737041241,
+      "grad_norm": 0.6866934173104662,
+      "learning_rate": 3.253852248540994e-06,
+      "loss": 0.288,
+      "step": 503
+    },
+    {
+      "epoch": 1.5266742338251986,
+      "grad_norm": 0.7446535452898911,
+      "learning_rate": 3.2149058844286796e-06,
+      "loss": 0.2857,
+      "step": 504
+    },
+    {
+      "epoch": 1.5297010972379872,
+      "grad_norm": 0.7058024107660282,
+      "learning_rate": 3.1761493139843734e-06,
+      "loss": 0.2746,
+      "step": 505
+    },
+    {
+      "epoch": 1.5327279606507758,
+      "grad_norm": 0.7152314604142057,
+      "learning_rate": 3.1375836213126653e-06,
+      "loss": 0.275,
+      "step": 506
+    },
+    {
+      "epoch": 1.5357548240635641,
+      "grad_norm": 0.7361499347573892,
+      "learning_rate": 3.099209885178882e-06,
+      "loss": 0.3049,
+      "step": 507
+    },
+    {
+      "epoch": 1.5387816874763525,
+      "grad_norm": 0.7070178398962562,
+      "learning_rate": 3.0610291789789094e-06,
+      "loss": 0.2754,
+      "step": 508
+    },
+    {
+      "epoch": 1.541808550889141,
+      "grad_norm": 0.7056929264345234,
+      "learning_rate": 3.023042570709185e-06,
+      "loss": 0.2834,
+      "step": 509
+    },
+    {
+      "epoch": 1.5448354143019296,
+      "grad_norm": 0.7358822822235239,
+      "learning_rate": 2.9852511229367862e-06,
+      "loss": 0.2769,
+      "step": 510
+    },
+    {
+      "epoch": 1.5478622777147182,
+      "grad_norm": 0.7269462023969843,
+      "learning_rate": 2.9476558927697605e-06,
+      "loss": 0.2891,
+      "step": 511
+    },
+    {
+      "epoch": 1.5508891411275068,
+      "grad_norm": 0.7267648816561406,
+      "learning_rate": 2.9102579318274994e-06,
+      "loss": 0.284,
+      "step": 512
+    },
+    {
+      "epoch": 1.553916004540295,
+      "grad_norm": 0.81082083342321,
+      "learning_rate": 2.8730582862113743e-06,
+      "loss": 0.29,
+      "step": 513
+    },
+    {
+      "epoch": 1.5569428679530835,
+      "grad_norm": 0.7349299651358615,
+      "learning_rate": 2.8360579964754277e-06,
+      "loss": 0.2956,
+      "step": 514
+    },
+    {
+      "epoch": 1.559969731365872,
+      "grad_norm": 0.7347955385843173,
+      "learning_rate": 2.7992580975973136e-06,
+      "loss": 0.2898,
+      "step": 515
+    },
+    {
+      "epoch": 1.5629965947786606,
+      "grad_norm": 0.7741223951636901,
+      "learning_rate": 2.7626596189492983e-06,
+      "loss": 0.2842,
+      "step": 516
+    },
+    {
+      "epoch": 1.5660234581914492,
+      "grad_norm": 0.7302759349378197,
+      "learning_rate": 2.726263584269513e-06,
+      "loss": 0.2872,
+      "step": 517
+    },
+    {
+      "epoch": 1.5690503216042377,
+      "grad_norm": 0.7068378693126671,
+      "learning_rate": 2.690071011633284e-06,
+      "loss": 0.2691,
+      "step": 518
+    },
+    {
+      "epoch": 1.572077185017026,
+      "grad_norm": 0.763210419698676,
+      "learning_rate": 2.6540829134246683e-06,
+      "loss": 0.2985,
+      "step": 519
+    },
+    {
+      "epoch": 1.5751040484298144,
+      "grad_norm": 0.7458387625331496,
+      "learning_rate": 2.618300296308135e-06,
+      "loss": 0.2921,
+      "step": 520
+    },
+    {
+      "epoch": 1.578130911842603,
+      "grad_norm": 0.7549843645214687,
+      "learning_rate": 2.582724161200405e-06,
+      "loss": 0.2952,
+      "step": 521
+    },
+    {
+      "epoch": 1.5811577752553916,
+      "grad_norm": 0.6708451309727878,
+      "learning_rate": 2.5473555032424534e-06,
+      "loss": 0.2707,
+      "step": 522
+    },
+    {
+      "epoch": 1.5841846386681802,
+      "grad_norm": 0.7921423765859591,
+      "learning_rate": 2.5121953117716744e-06,
+      "loss": 0.3005,
+      "step": 523
+    },
+    {
+      "epoch": 1.5872115020809687,
+      "grad_norm": 0.7156686993032988,
+      "learning_rate": 2.477244570294206e-06,
+      "loss": 0.2928,
+      "step": 524
+    },
+    {
+      "epoch": 1.590238365493757,
+      "grad_norm": 0.741438268896103,
+      "learning_rate": 2.4425042564574186e-06,
+      "loss": 0.2909,
+      "step": 525
+    },
+    {
+      "epoch": 1.5932652289065454,
+      "grad_norm": 0.7126420131911106,
+      "learning_rate": 2.4079753420225694e-06,
+      "loss": 0.2893,
+      "step": 526
+    },
+    {
+      "epoch": 1.596292092319334,
+      "grad_norm": 0.6945180976922313,
+      "learning_rate": 2.3736587928376197e-06,
+      "loss": 0.2717,
+      "step": 527
+    },
+    {
+      "epoch": 1.5993189557321226,
+      "grad_norm": 0.7417501745321297,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.2789,
+      "step": 528
+    },
+    {
+      "epoch": 1.6023458191449111,
+      "grad_norm": 0.7839953653490982,
+      "learning_rate": 2.305666623880858e-06,
+      "loss": 0.2839,
+      "step": 529
+    },
+    {
+      "epoch": 1.6053726825576997,
+      "grad_norm": 0.7551774729545742,
+      "learning_rate": 2.27199290599617e-06,
+      "loss": 0.2976,
+      "step": 530
+    },
+    {
+      "epoch": 1.608399545970488,
+      "grad_norm": 0.7850882748573174,
+      "learning_rate": 2.2385353570824308e-06,
+      "loss": 0.3143,
+      "step": 531
+    },
+    {
+      "epoch": 1.6114264093832764,
+      "grad_norm": 0.6912849718601763,
+      "learning_rate": 2.2052949130192136e-06,
+      "loss": 0.2879,
+      "step": 532
+    },
+    {
+      "epoch": 1.614453272796065,
+      "grad_norm": 0.69305922593856,
+      "learning_rate": 2.172272503613183e-06,
+      "loss": 0.2769,
+      "step": 533
+    },
+    {
+      "epoch": 1.6174801362088536,
+      "grad_norm": 0.7498482055488955,
+      "learning_rate": 2.1394690525721275e-06,
+      "loss": 0.2975,
+      "step": 534
+    },
+    {
+      "epoch": 1.6205069996216421,
+      "grad_norm": 0.7029708904978335,
+      "learning_rate": 2.1068854774790783e-06,
+      "loss": 0.2882,
+      "step": 535
+    },
+    {
+      "epoch": 1.6235338630344307,
+      "grad_norm": 0.7267016274862943,
+      "learning_rate": 2.0745226897666858e-06,
+      "loss": 0.2847,
+      "step": 536
+    },
+    {
+      "epoch": 1.626560726447219,
+      "grad_norm": 0.7343301279256917,
+      "learning_rate": 2.0423815946916783e-06,
+      "loss": 0.2729,
+      "step": 537
+    },
+    {
+      "epoch": 1.6295875898600074,
+      "grad_norm": 0.6933712288775056,
+      "learning_rate": 2.010463091309587e-06,
+      "loss": 0.29,
+      "step": 538
+    },
+    {
+      "epoch": 1.632614453272796,
+      "grad_norm": 0.7128394414147942,
+      "learning_rate": 1.9787680724495617e-06,
+      "loss": 0.2789,
+      "step": 539
+    },
+    {
+      "epoch": 1.6356413166855845,
+      "grad_norm": 0.6806063171358394,
+      "learning_rate": 1.947297424689414e-06,
+      "loss": 0.2845,
+      "step": 540
+    },
+    {
+      "epoch": 1.6386681800983731,
+      "grad_norm": 0.7616271314644175,
+      "learning_rate": 1.9160520283308115e-06,
+      "loss": 0.3032,
+      "step": 541
+    },
+    {
+      "epoch": 1.6416950435111617,
+      "grad_norm": 0.7434964202112795,
+      "learning_rate": 1.8850327573746584e-06,
+      "loss": 0.2892,
+      "step": 542
+    },
+    {
+      "epoch": 1.64472190692395,
+      "grad_norm": 0.6899606630068712,
+      "learning_rate": 1.854240479496643e-06,
+      "loss": 0.2754,
+      "step": 543
+    },
+    {
+      "epoch": 1.6477487703367384,
+      "grad_norm": 0.7403122570498195,
+      "learning_rate": 1.8236760560229715e-06,
+      "loss": 0.3013,
+      "step": 544
+    },
+    {
+      "epoch": 1.650775633749527,
+      "grad_norm": 0.6980924495315326,
+      "learning_rate": 1.7933403419062689e-06,
+      "loss": 0.2954,
+      "step": 545
+    },
+    {
+      "epoch": 1.6538024971623155,
+      "grad_norm": 0.7506213925850654,
+      "learning_rate": 1.7632341857016733e-06,
+      "loss": 0.2967,
+      "step": 546
+    },
+    {
+      "epoch": 1.656829360575104,
+      "grad_norm": 0.7082781511509584,
+      "learning_rate": 1.7333584295430894e-06,
+      "loss": 0.282,
+      "step": 547
+    },
+    {
+      "epoch": 1.6598562239878927,
+      "grad_norm": 0.7245149393135567,
+      "learning_rate": 1.7037139091196396e-06,
+      "loss": 0.285,
+      "step": 548
+    },
+    {
+      "epoch": 1.662883087400681,
+      "grad_norm": 0.7430794268067457,
+      "learning_rate": 1.6743014536522872e-06,
+      "loss": 0.2975,
+      "step": 549
+    },
+    {
+      "epoch": 1.6659099508134694,
+      "grad_norm": 0.7352819171956158,
+      "learning_rate": 1.6451218858706374e-06,
+      "loss": 0.2836,
+      "step": 550
+    },
+    {
+      "epoch": 1.668936814226258,
+      "grad_norm": 0.6899265091515417,
+      "learning_rate": 1.616176021989926e-06,
+      "loss": 0.2731,
+      "step": 551
+    },
+    {
+      "epoch": 1.6719636776390465,
+      "grad_norm": 0.7306946981845605,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.2735,
+      "step": 552
+    },
+    {
+      "epoch": 1.674990541051835,
+      "grad_norm": 0.7537542357900331,
+      "learning_rate": 1.558988638083616e-06,
+      "loss": 0.2889,
+      "step": 553
+    },
+    {
+      "epoch": 1.6780174044646237,
+      "grad_norm": 0.769024927653227,
+      "learning_rate": 1.5307487177120773e-06,
+      "loss": 0.3034,
+      "step": 554
+    },
+    {
+      "epoch": 1.681044267877412,
+      "grad_norm": 0.7969651911826424,
+      "learning_rate": 1.5027457005048573e-06,
+      "loss": 0.2867,
+      "step": 555
+    },
+    {
+      "epoch": 1.6840711312902006,
+      "grad_norm": 0.7495097659385466,
+      "learning_rate": 1.4749803697665366e-06,
+      "loss": 0.2823,
+      "step": 556
+    },
+    {
+      "epoch": 1.687097994702989,
+      "grad_norm": 0.742092530785966,
+      "learning_rate": 1.4474535021531099e-06,
+      "loss": 0.2737,
+      "step": 557
+    },
+    {
+      "epoch": 1.6901248581157775,
+      "grad_norm": 0.7408088895223955,
+      "learning_rate": 1.4201658676502294e-06,
+      "loss": 0.2857,
+      "step": 558
+    },
+    {
+      "epoch": 1.693151721528566,
+      "grad_norm": 0.746135322244573,
+      "learning_rate": 1.3931182295516965e-06,
+      "loss": 0.2818,
+      "step": 559
+    },
+    {
+      "epoch": 1.6961785849413547,
+      "grad_norm": 0.7303274415674746,
+      "learning_rate": 1.3663113444380905e-06,
+      "loss": 0.2842,
+      "step": 560
+    },
+    {
+      "epoch": 1.699205448354143,
+      "grad_norm": 0.7421086064710165,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.2746,
+      "step": 561
+    },
+    {
+      "epoch": 1.7022323117669316,
+      "grad_norm": 0.738010247003656,
+      "learning_rate": 1.3134228257951142e-06,
+      "loss": 0.2816,
+      "step": 562
+    },
+    {
+      "epoch": 1.70525917517972,
+      "grad_norm": 0.653930048735564,
+      "learning_rate": 1.2873426716713012e-06,
+      "loss": 0.2731,
+      "step": 563
+    },
+    {
+      "epoch": 1.7082860385925085,
+      "grad_norm": 0.7291415886800213,
+      "learning_rate": 1.2615062293021508e-06,
+      "loss": 0.2833,
+      "step": 564
+    },
+    {
+      "epoch": 1.711312902005297,
+      "grad_norm": 0.7429149069294742,
+      "learning_rate": 1.2359142213884933e-06,
+      "loss": 0.2898,
+      "step": 565
+    },
+    {
+      "epoch": 1.7143397654180856,
+      "grad_norm": 0.7468563459361297,
+      "learning_rate": 1.2105673637938054e-06,
+      "loss": 0.282,
+      "step": 566
+    },
+    {
+      "epoch": 1.717366628830874,
+      "grad_norm": 0.748235390000226,
+      "learning_rate": 1.1854663655241804e-06,
+      "loss": 0.2868,
+      "step": 567
+    },
+    {
+      "epoch": 1.7203934922436626,
+      "grad_norm": 0.7399831789304407,
+      "learning_rate": 1.1606119287084982e-06,
+      "loss": 0.2995,
+      "step": 568
+    },
+    {
+      "epoch": 1.723420355656451,
+      "grad_norm": 0.70530159421794,
+      "learning_rate": 1.136004748578785e-06,
+      "loss": 0.2901,
+      "step": 569
+    },
+    {
+      "epoch": 1.7264472190692395,
+      "grad_norm": 0.7263729785065444,
+      "learning_rate": 1.1116455134507665e-06,
+      "loss": 0.2804,
+      "step": 570
+    },
+    {
+      "epoch": 1.729474082482028,
+      "grad_norm": 0.7400879965869811,
+      "learning_rate": 1.0875349047046113e-06,
+      "loss": 0.2832,
+      "step": 571
+    },
+    {
+      "epoch": 1.7325009458948166,
+      "grad_norm": 0.6939441719805616,
+      "learning_rate": 1.0636735967658785e-06,
+      "loss": 0.2777,
+      "step": 572
+    },
+    {
+      "epoch": 1.735527809307605,
+      "grad_norm": 0.7187457254469004,
+      "learning_rate": 1.0400622570866426e-06,
+      "loss": 0.2882,
+      "step": 573
+    },
+    {
+      "epoch": 1.7385546727203935,
+      "grad_norm": 0.7129927768099702,
+      "learning_rate": 1.0167015461268303e-06,
+      "loss": 0.2965,
+      "step": 574
+    },
+    {
+      "epoch": 1.741581536133182,
+      "grad_norm": 0.7377020049375957,
+      "learning_rate": 9.935921173357444e-07,
+      "loss": 0.2767,
+      "step": 575
+    },
+    {
+      "epoch": 1.7446083995459705,
+      "grad_norm": 0.7171897972465745,
+      "learning_rate": 9.707346171337895e-07,
+      "loss": 0.2843,
+      "step": 576
+    },
+    {
+      "epoch": 1.747635262958759,
+      "grad_norm": 0.7470525113713176,
+      "learning_rate": 9.481296848943744e-07,
+      "loss": 0.2792,
+      "step": 577
+    },
+    {
+      "epoch": 1.7506621263715476,
+      "grad_norm": 0.6835454968910495,
+      "learning_rate": 9.257779529260558e-07,
+      "loss": 0.2836,
+      "step": 578
+    },
+    {
+      "epoch": 1.753688989784336,
+      "grad_norm": 0.7331058996787356,
+      "learning_rate": 9.036800464548157e-07,
+      "loss": 0.2755,
+      "step": 579
+    },
+    {
+      "epoch": 1.7567158531971245,
+      "grad_norm": 0.7002489691833005,
+      "learning_rate": 8.818365836066101e-07,
+      "loss": 0.2877,
+      "step": 580
+    },
+    {
+      "epoch": 1.7597427166099129,
+      "grad_norm": 0.676411565349038,
+      "learning_rate": 8.602481753900427e-07,
+      "loss": 0.2689,
+      "step": 581
+    },
+    {
+      "epoch": 1.7627695800227015,
+      "grad_norm": 0.7212299819251898,
+      "learning_rate": 8.389154256793042e-07,
+      "loss": 0.2921,
+      "step": 582
+    },
+    {
+      "epoch": 1.76579644343549,
+      "grad_norm": 0.6871888810542574,
+      "learning_rate": 8.178389311972612e-07,
+      "loss": 0.2816,
+      "step": 583
+    },
+    {
+      "epoch": 1.7688233068482786,
+      "grad_norm": 0.7189354803567105,
+      "learning_rate": 7.970192814987676e-07,
+      "loss": 0.2876,
+      "step": 584
+    },
+    {
+      "epoch": 1.771850170261067,
+      "grad_norm": 0.6980741561133377,
+      "learning_rate": 7.764570589541876e-07,
+      "loss": 0.289,
+      "step": 585
+    },
+    {
+      "epoch": 1.7748770336738555,
+      "grad_norm": 0.741741361276942,
+      "learning_rate": 7.561528387330797e-07,
+      "loss": 0.2769,
+      "step": 586
+    },
+    {
+      "epoch": 1.7779038970866439,
+      "grad_norm": 0.7323169677735252,
+      "learning_rate": 7.361071887881376e-07,
+      "loss": 0.2805,
+      "step": 587
+    },
+    {
+      "epoch": 1.7809307604994324,
+      "grad_norm": 0.6892002504883342,
+      "learning_rate": 7.163206698392744e-07,
+      "loss": 0.2761,
+      "step": 588
+    },
+    {
+      "epoch": 1.783957623912221,
+      "grad_norm": 0.7653224993989634,
+      "learning_rate": 6.96793835357964e-07,
+      "loss": 0.2831,
+      "step": 589
+    },
+    {
+      "epoch": 1.7869844873250096,
+      "grad_norm": 0.699940441666469,
+      "learning_rate": 6.775272315517423e-07,
+      "loss": 0.2803,
+      "step": 590
+    },
+    {
+      "epoch": 1.790011350737798,
+      "grad_norm": 0.7612862245089655,
+      "learning_rate": 6.585213973489335e-07,
+      "loss": 0.2906,
+      "step": 591
+    },
+    {
+      "epoch": 1.7930382141505865,
+      "grad_norm": 0.802610549538578,
+      "learning_rate": 6.397768643835755e-07,
+      "loss": 0.2908,
+      "step": 592
+    },
+    {
+      "epoch": 1.7960650775633749,
+      "grad_norm": 0.7606915724821676,
+      "learning_rate": 6.212941569805508e-07,
+      "loss": 0.2972,
+      "step": 593
+    },
+    {
+      "epoch": 1.7990919409761634,
+      "grad_norm": 0.7580447676780981,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.306,
+      "step": 594
+    },
+    {
+      "epoch": 1.802118804388952,
+      "grad_norm": 0.7217683294734949,
+      "learning_rate": 5.851162795274445e-07,
+      "loss": 0.2851,
+      "step": 595
+    },
+    {
+      "epoch": 1.8051456678017406,
+      "grad_norm": 0.6646661253396376,
+      "learning_rate": 5.674221214503639e-07,
+      "loss": 0.2622,
+      "step": 596
+    },
+    {
+      "epoch": 1.808172531214529,
+      "grad_norm": 0.7527591491824924,
+      "learning_rate": 5.499918128533155e-07,
+      "loss": 0.2845,
+      "step": 597
+    },
+    {
+      "epoch": 1.8111993946273175,
+      "grad_norm": 0.7093218838484242,
+      "learning_rate": 5.328258412994958e-07,
+      "loss": 0.2813,
+      "step": 598
+    },
+    {
+      "epoch": 1.8142262580401058,
+      "grad_norm": 0.7583654632717032,
+      "learning_rate": 5.159246869580348e-07,
+      "loss": 0.2955,
+      "step": 599
+    },
+    {
+      "epoch": 1.8172531214528944,
+      "grad_norm": 0.7241449879881268,
+      "learning_rate": 4.992888225905467e-07,
+      "loss": 0.2809,
+      "step": 600
+    },
+    {
+      "epoch": 1.820279984865683,
+      "grad_norm": 0.7326374792225369,
+      "learning_rate": 4.829187135379221e-07,
+      "loss": 0.2794,
+      "step": 601
+    },
+    {
+      "epoch": 1.8233068482784716,
+      "grad_norm": 0.7011079746347699,
+      "learning_rate": 4.6681481770729844e-07,
+      "loss": 0.2851,
+      "step": 602
+    },
+    {
+      "epoch": 1.82633371169126,
+      "grad_norm": 0.7339175136770655,
+      "learning_rate": 4.509775855592613e-07,
+      "loss": 0.2932,
+      "step": 603
+    },
+    {
+      "epoch": 1.8293605751040485,
+      "grad_norm": 0.742366044242703,
+      "learning_rate": 4.354074600952407e-07,
+      "loss": 0.2817,
+      "step": 604
+    },
+    {
+      "epoch": 1.8323874385168368,
+      "grad_norm": 0.6960508964119638,
+      "learning_rate": 4.2010487684511105e-07,
+      "loss": 0.2749,
+      "step": 605
+    },
+    {
+      "epoch": 1.8354143019296254,
+      "grad_norm": 0.7177484864169353,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.2669,
+      "step": 606
+    },
+    {
+      "epoch": 1.838441165342414,
+      "grad_norm": 0.698412561144052,
+      "learning_rate": 3.9030404167542777e-07,
+      "loss": 0.2891,
+      "step": 607
+    },
+    {
+      "epoch": 1.8414680287552025,
+      "grad_norm": 0.7134931269829115,
+      "learning_rate": 3.7580662334929517e-07,
+      "loss": 0.2796,
+      "step": 608
+    },
+    {
+      "epoch": 1.844494892167991,
+      "grad_norm": 0.655839335813593,
+      "learning_rate": 3.615784144005796e-07,
+      "loss": 0.2694,
+      "step": 609
+    },
+    {
+      "epoch": 1.8475217555807795,
+      "grad_norm": 0.8102919212572529,
+      "learning_rate": 3.476198128228736e-07,
+      "loss": 0.2998,
+      "step": 610
+    },
+    {
+      "epoch": 1.8505486189935678,
+      "grad_norm": 0.758533730707042,
+      "learning_rate": 3.339312090682689e-07,
+      "loss": 0.2919,
+      "step": 611
+    },
+    {
+      "epoch": 1.8535754824063564,
+      "grad_norm": 0.7257975043920856,
+      "learning_rate": 3.2051298603643754e-07,
+      "loss": 0.2782,
+      "step": 612
+    },
+    {
+      "epoch": 1.856602345819145,
+      "grad_norm": 0.7243900104948473,
+      "learning_rate": 3.0736551906392354e-07,
+      "loss": 0.2841,
+      "step": 613
+    },
+    {
+      "epoch": 1.8596292092319335,
+      "grad_norm": 0.7172858690647557,
+      "learning_rate": 2.9448917591363923e-07,
+      "loss": 0.2802,
+      "step": 614
+    },
+    {
+      "epoch": 1.8626560726447219,
+      "grad_norm": 0.6812343387762557,
+      "learning_rate": 2.818843167645835e-07,
+      "loss": 0.2799,
+      "step": 615
+    },
+    {
+      "epoch": 1.8656829360575105,
+      "grad_norm": 0.7377461951855622,
+      "learning_rate": 2.6955129420176193e-07,
+      "loss": 0.284,
+      "step": 616
+    },
+    {
+      "epoch": 1.8687097994702988,
+      "grad_norm": 0.6826023677217762,
+      "learning_rate": 2.5749045320632824e-07,
+      "loss": 0.268,
+      "step": 617
+    },
+    {
+      "epoch": 1.8717366628830874,
+      "grad_norm": 0.7177902190604383,
+      "learning_rate": 2.4570213114592957e-07,
+      "loss": 0.2919,
+      "step": 618
+    },
+    {
+      "epoch": 1.874763526295876,
+      "grad_norm": 0.7630130041934038,
+      "learning_rate": 2.3418665776527738e-07,
+      "loss": 0.2872,
+      "step": 619
+    },
+    {
+      "epoch": 1.8777903897086645,
+      "grad_norm": 0.6598327695164433,
+      "learning_rate": 2.2294435517691504e-07,
+      "loss": 0.2702,
+      "step": 620
+    },
+    {
+      "epoch": 1.8808172531214529,
+      "grad_norm": 0.6955688252731471,
+      "learning_rate": 2.119755378522137e-07,
+      "loss": 0.2698,
+      "step": 621
+    },
+    {
+      "epoch": 1.8838441165342414,
+      "grad_norm": 0.6719447189782813,
+      "learning_rate": 2.0128051261257165e-07,
+      "loss": 0.2821,
+      "step": 622
+    },
+    {
+      "epoch": 1.8868709799470298,
+      "grad_norm": 0.6896399440887028,
+      "learning_rate": 1.908595786208367e-07,
+      "loss": 0.289,
+      "step": 623
+    },
+    {
+      "epoch": 1.8898978433598184,
+      "grad_norm": 0.7462970074930224,
+      "learning_rate": 1.8071302737293294e-07,
+      "loss": 0.2879,
+      "step": 624
+    },
+    {
+      "epoch": 1.892924706772607,
+      "grad_norm": 0.7185170588257133,
+      "learning_rate": 1.7084114268971275e-07,
+      "loss": 0.2711,
+      "step": 625
+    },
+    {
+      "epoch": 1.8959515701853955,
+      "grad_norm": 0.6975099789039778,
+      "learning_rate": 1.612442007090076e-07,
+      "loss": 0.2738,
+      "step": 626
+    },
+    {
+      "epoch": 1.8989784335981839,
+      "grad_norm": 0.7383859590727406,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.279,
+      "step": 627
+    },
+    {
+      "epoch": 1.9020052970109724,
+      "grad_norm": 0.7351693534254338,
+      "learning_rate": 1.4287621094529524e-07,
+      "loss": 0.3017,
+      "step": 628
+    },
+    {
+      "epoch": 1.9050321604237608,
+      "grad_norm": 0.7343494928664471,
+      "learning_rate": 1.3410567695444576e-07,
+      "loss": 0.3005,
+      "step": 629
+    },
+    {
+      "epoch": 1.9080590238365494,
+      "grad_norm": 0.6962622472427729,
+      "learning_rate": 1.2561111323605714e-07,
+      "loss": 0.2843,
+      "step": 630
+    },
+    {
+      "epoch": 1.911085887249338,
+      "grad_norm": 0.6879715684560946,
+      "learning_rate": 1.1739275740134004e-07,
+      "loss": 0.2798,
+      "step": 631
+    },
+    {
+      "epoch": 1.9141127506621265,
+      "grad_norm": 0.7211784347702246,
+      "learning_rate": 1.0945083933537104e-07,
+      "loss": 0.2871,
+      "step": 632
+    },
+    {
+      "epoch": 1.9171396140749148,
+      "grad_norm": 0.6920713687831043,
+      "learning_rate": 1.0178558119067316e-07,
+      "loss": 0.2685,
+      "step": 633
+    },
+    {
+      "epoch": 1.9201664774877034,
+      "grad_norm": 0.7113552568049702,
+      "learning_rate": 9.439719738099318e-08,
+      "loss": 0.273,
+      "step": 634
+    },
+    {
+      "epoch": 1.9231933409004918,
+      "grad_norm": 0.7230670643939646,
+      "learning_rate": 8.728589457530857e-08,
+      "loss": 0.281,
+      "step": 635
+    },
+    {
+      "epoch": 1.9262202043132803,
+      "grad_norm": 0.67956434417701,
+      "learning_rate": 8.04518716920466e-08,
+      "loss": 0.2726,
+      "step": 636
+    },
+    {
+      "epoch": 1.929247067726069,
+      "grad_norm": 0.686099801645611,
+      "learning_rate": 7.389531989351773e-08,
+      "loss": 0.2735,
+      "step": 637
+    },
+    {
+      "epoch": 1.9322739311388575,
+      "grad_norm": 0.6966486340782885,
+      "learning_rate": 6.761642258056977e-08,
+      "loss": 0.2764,
+      "step": 638
+    },
+    {
+      "epoch": 1.9353007945516458,
+      "grad_norm": 0.7164865927215468,
+      "learning_rate": 6.161535538745877e-08,
+      "loss": 0.2923,
+      "step": 639
+    },
+    {
+      "epoch": 1.9383276579644344,
+      "grad_norm": 0.7439418958100256,
+      "learning_rate": 5.5892286176932875e-08,
+      "loss": 0.2813,
+      "step": 640
+    },
+    {
+      "epoch": 1.9413545213772228,
+      "grad_norm": 0.7288911553256673,
+      "learning_rate": 5.044737503554165e-08,
+      "loss": 0.2937,
+      "step": 641
+    },
+    {
+      "epoch": 1.9443813847900113,
+      "grad_norm": 0.6994801424222791,
+      "learning_rate": 4.528077426915412e-08,
+      "loss": 0.2912,
+      "step": 642
+    },
+    {
+      "epoch": 1.9474082482028,
+      "grad_norm": 0.7341432707743182,
+      "learning_rate": 4.0392628398699954e-08,
+      "loss": 0.2832,
+      "step": 643
+    },
+    {
+      "epoch": 1.9504351116155885,
+      "grad_norm": 0.6979419114448084,
+      "learning_rate": 3.578307415612714e-08,
+      "loss": 0.2701,
+      "step": 644
+    },
+    {
+      "epoch": 1.9534619750283768,
+      "grad_norm": 0.750803155187425,
+      "learning_rate": 3.1452240480577265e-08,
+      "loss": 0.2926,
+      "step": 645
+    },
+    {
+      "epoch": 1.9564888384411654,
+      "grad_norm": 0.6838855971162886,
+      "learning_rate": 2.7400248514776184e-08,
+      "loss": 0.2823,
+      "step": 646
+    },
+    {
+      "epoch": 1.9595157018539537,
+      "grad_norm": 0.7356635176578719,
+      "learning_rate": 2.3627211601651157e-08,
+      "loss": 0.2894,
+      "step": 647
+    },
+    {
+      "epoch": 1.9625425652667423,
+      "grad_norm": 0.7016083406103137,
+      "learning_rate": 2.013323528115674e-08,
+      "loss": 0.2764,
+      "step": 648
+    },
+    {
+      "epoch": 1.9655694286795309,
+      "grad_norm": 0.6923939069789969,
+      "learning_rate": 1.6918417287318245e-08,
+      "loss": 0.273,
+      "step": 649
+    },
+    {
+      "epoch": 1.9685962920923195,
+      "grad_norm": 0.754724684231685,
+      "learning_rate": 1.3982847545507271e-08,
+      "loss": 0.2842,
+      "step": 650
+    },
+    {
+      "epoch": 1.9716231555051078,
+      "grad_norm": 0.7434440805750054,
+      "learning_rate": 1.1326608169920373e-08,
+      "loss": 0.2802,
+      "step": 651
+    },
+    {
+      "epoch": 1.9746500189178964,
+      "grad_norm": 0.7231237117817423,
+      "learning_rate": 8.949773461282008e-09,
+      "loss": 0.288,
+      "step": 652
+    },
+    {
+      "epoch": 1.9776768823306847,
+      "grad_norm": 0.6985415731193907,
+      "learning_rate": 6.8524099047695415e-09,
+      "loss": 0.2867,
+      "step": 653
+    },
+    {
+      "epoch": 1.9807037457434733,
+      "grad_norm": 0.7055975612573275,
+      "learning_rate": 5.034576168149175e-09,
+      "loss": 0.2727,
+      "step": 654
+    },
+    {
+      "epoch": 1.9837306091562619,
+      "grad_norm": 0.6909272774853085,
+      "learning_rate": 3.4963231001383657e-09,
+      "loss": 0.2769,
+      "step": 655
+    },
+    {
+      "epoch": 1.9867574725690504,
+      "grad_norm": 0.7060592973015987,
+      "learning_rate": 2.237693728981416e-09,
+      "loss": 0.2773,
+      "step": 656
+    },
+    {
+      "epoch": 1.9897843359818388,
+      "grad_norm": 0.7319430087975928,
+      "learning_rate": 1.2587232612493172e-09,
+      "loss": 0.3011,
+      "step": 657
+    },
+    {
+      "epoch": 1.9928111993946274,
+      "grad_norm": 0.7167491228344711,
+      "learning_rate": 5.594390808494332e-10,
+      "loss": 0.273,
+      "step": 658
+    },
+    {
+      "epoch": 1.9958380628074157,
+      "grad_norm": 0.7035027842122591,
+      "learning_rate": 1.3986074826388697e-10,
+      "loss": 0.2865,
+      "step": 659
+    },
+    {
+      "epoch": 1.9988649262202043,
+      "grad_norm": 0.7201409937608538,
+      "learning_rate": 0.0,
+      "loss": 0.273,
+      "step": 660
+    },
+    {
+      "epoch": 1.9988649262202043,
+      "step": 660,
+      "total_flos": 64959092490240.0,
+      "train_loss": 0.14502406124815798,
+      "train_runtime": 1228.9078,
+      "train_samples_per_second": 68.799,
+      "train_steps_per_second": 0.537
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 660,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 64959092490240.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}