diff --git "a/checkpoint-2907/trainer_state.json" "b/checkpoint-2907/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2907/trainer_state.json"
@@ -0,0 +1,20466 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9610724413508636,
+  "eval_steps": 243,
+  "global_step": 2907,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.08921732753515244,
+      "learning_rate": 2e-05,
+      "loss": 0.8982,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 0.910159707069397,
+      "eval_runtime": 111.5727,
+      "eval_samples_per_second": 89.072,
+      "eval_steps_per_second": 22.272,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.07926193624734879,
+      "learning_rate": 4e-05,
+      "loss": 0.8904,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.09000936150550842,
+      "learning_rate": 6e-05,
+      "loss": 0.8817,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.08947216719388962,
+      "learning_rate": 8e-05,
+      "loss": 0.8865,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.09538354724645615,
+      "learning_rate": 0.0001,
+      "loss": 0.7988,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.11288489401340485,
+      "learning_rate": 0.00012,
+      "loss": 0.8345,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.08290982991456985,
+      "learning_rate": 0.00014,
+      "loss": 0.8137,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.06841357797384262,
+      "learning_rate": 0.00016,
+      "loss": 0.7308,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.07010111957788467,
+      "learning_rate": 0.00018,
+      "loss": 0.8572,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.07711824029684067,
+      "learning_rate": 0.0002,
+      "loss": 0.7942,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.06440314650535583,
+      "learning_rate": 0.00019999994120062405,
+      "loss": 0.7049,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.0699019655585289,
+      "learning_rate": 0.0001999997648025654,
+      "loss": 0.8176,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.07182443886995316,
+      "learning_rate": 0.00019999947080603144,
+      "loss": 0.8583,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.057579606771469116,
+      "learning_rate": 0.00019999905921136795,
+      "loss": 0.8021,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.051665112376213074,
+      "learning_rate": 0.0001999985300190589,
+      "loss": 0.7698,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.05774515122175217,
+      "learning_rate": 0.00019999788322972668,
+      "loss": 0.7913,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.054730940610170364,
+      "learning_rate": 0.0001999971188441319,
+      "loss": 0.7205,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.05652638524770737,
+      "learning_rate": 0.00019999623686317342,
+      "loss": 0.7844,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.05296637862920761,
+      "learning_rate": 0.00019999523728788846,
+      "loss": 0.7022,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.05956929177045822,
+      "learning_rate": 0.00019999412011945252,
+      "loss": 0.7541,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.050489749759435654,
+      "learning_rate": 0.00019999288535917938,
+      "loss": 0.7155,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.048119645565748215,
+      "learning_rate": 0.00019999153300852106,
+      "loss": 0.757,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.050988584756851196,
+      "learning_rate": 0.00019999006306906797,
+      "loss": 0.7282,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.04979432374238968,
+      "learning_rate": 0.00019998847554254868,
+      "loss": 0.79,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.04835788160562515,
+      "learning_rate": 0.00019998677043083017,
+      "loss": 0.7097,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.05280015245079994,
+      "learning_rate": 0.00019998494773591758,
+      "loss": 0.8246,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.05030938982963562,
+      "learning_rate": 0.00019998300745995438,
+      "loss": 0.7865,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.050140462815761566,
+      "learning_rate": 0.00019998094960522233,
+      "loss": 0.7679,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.053974639624357224,
+      "learning_rate": 0.00019997877417414142,
+      "loss": 0.7476,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.048135802149772644,
+      "learning_rate": 0.00019997648116926994,
+      "loss": 0.7894,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.05285784229636192,
+      "learning_rate": 0.00019997407059330443,
+      "loss": 0.7967,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.06759033352136612,
+      "learning_rate": 0.00019997154244907973,
+      "loss": 0.8675,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.05007735639810562,
+      "learning_rate": 0.00019996889673956885,
+      "loss": 0.8439,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.044428203254938126,
+      "learning_rate": 0.00019996613346788316,
+      "loss": 0.7767,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.04795186221599579,
+      "learning_rate": 0.00019996325263727218,
+      "loss": 0.7618,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.045297641307115555,
+      "learning_rate": 0.00019996025425112377,
+      "loss": 0.7379,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.04418836906552315,
+      "learning_rate": 0.000199957138312964,
+      "loss": 0.7092,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.05495867133140564,
+      "learning_rate": 0.0001999539048264572,
+      "loss": 0.7708,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.047341588884592056,
+      "learning_rate": 0.0001999505537954058,
+      "loss": 0.7829,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.05130796134471893,
+      "learning_rate": 0.0001999470852237507,
+      "loss": 0.701,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.04843367636203766,
+      "learning_rate": 0.00019994349911557082,
+      "loss": 0.7624,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.05179371312260628,
+      "learning_rate": 0.0001999397954750834,
+      "loss": 0.7432,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.052478544414043427,
+      "learning_rate": 0.00019993597430664387,
+      "loss": 0.7313,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.05324641987681389,
+      "learning_rate": 0.00019993203561474588,
+      "loss": 0.8092,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.051216114312410355,
+      "learning_rate": 0.0001999279794040213,
+      "loss": 0.6828,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.0454455241560936,
+      "learning_rate": 0.0001999238056792401,
+      "loss": 0.7963,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.04751123487949371,
+      "learning_rate": 0.00019991951444531066,
+      "loss": 0.8264,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.05125482380390167,
+      "learning_rate": 0.00019991510570727933,
+      "loss": 0.7282,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.05229455232620239,
+      "learning_rate": 0.00019991057947033074,
+      "loss": 0.7409,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.046746984124183655,
+      "learning_rate": 0.00019990593573978772,
+      "loss": 0.8096,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.0461801253259182,
+      "learning_rate": 0.00019990117452111124,
+      "loss": 0.7995,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.04826509580016136,
+      "learning_rate": 0.00019989629581990037,
+      "loss": 0.795,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.04601113870739937,
+      "learning_rate": 0.00019989129964189246,
+      "loss": 0.7805,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.05406435951590538,
+      "learning_rate": 0.00019988618599296295,
+      "loss": 0.7585,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.054005216807127,
+      "learning_rate": 0.00019988095487912539,
+      "loss": 0.7165,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.04769807681441307,
+      "learning_rate": 0.00019987560630653156,
+      "loss": 0.7695,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.04815264418721199,
+      "learning_rate": 0.00019987014028147125,
+      "loss": 0.6944,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.051092248409986496,
+      "learning_rate": 0.00019986455681037248,
+      "loss": 0.7619,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.048955272883176804,
+      "learning_rate": 0.00019985885589980136,
+      "loss": 0.7827,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.049241162836551666,
+      "learning_rate": 0.00019985303755646206,
+      "loss": 0.6925,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.05324079468846321,
+      "learning_rate": 0.00019984710178719685,
+      "loss": 0.7269,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.05057024210691452,
+      "learning_rate": 0.0001998410485989862,
+      "loss": 0.7598,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.05710586905479431,
+      "learning_rate": 0.0001998348779989485,
+      "loss": 0.8153,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.05069264769554138,
+      "learning_rate": 0.00019982858999434034,
+      "loss": 0.6319,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.0459558442234993,
+      "learning_rate": 0.00019982218459255635,
+      "loss": 0.7189,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.054699912667274475,
+      "learning_rate": 0.00019981566180112917,
+      "loss": 0.7845,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.045664116740226746,
+      "learning_rate": 0.0001998090216277295,
+      "loss": 0.7308,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.046552181243896484,
+      "learning_rate": 0.0001998022640801662,
+      "loss": 0.6392,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.048938997089862823,
+      "learning_rate": 0.00019979538916638597,
+      "loss": 0.7591,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.051963839679956436,
+      "learning_rate": 0.00019978839689447365,
+      "loss": 0.7054,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.05242505297064781,
+      "learning_rate": 0.00019978128727265207,
+      "loss": 0.7467,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.04921518638730049,
+      "learning_rate": 0.00019977406030928204,
+      "loss": 0.7793,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.052262112498283386,
+      "learning_rate": 0.00019976671601286237,
+      "loss": 0.7193,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.053219135850667953,
+      "learning_rate": 0.0001997592543920299,
+      "loss": 0.7483,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.049095865339040756,
+      "learning_rate": 0.00019975167545555942,
+      "loss": 0.7544,
+      "step": 75
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.052104122936725616,
+      "learning_rate": 0.0001997439792123636,
+      "loss": 0.7227,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.05156683921813965,
+      "learning_rate": 0.00019973616567149313,
+      "loss": 0.7752,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.050344448536634445,
+      "learning_rate": 0.00019972823484213669,
+      "loss": 0.7515,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.04634309560060501,
+      "learning_rate": 0.00019972018673362077,
+      "loss": 0.7709,
+      "step": 79
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.05174633115530014,
+      "learning_rate": 0.00019971202135540993,
+      "loss": 0.7293,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.05256710946559906,
+      "learning_rate": 0.00019970373871710647,
+      "loss": 0.7085,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.07102134823799133,
+      "learning_rate": 0.00019969533882845076,
+      "loss": 0.668,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.05783531814813614,
+      "learning_rate": 0.00019968682169932083,
+      "loss": 0.7066,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.04908255860209465,
+      "learning_rate": 0.00019967818733973287,
+      "loss": 0.7456,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.04938043653964996,
+      "learning_rate": 0.00019966943575984068,
+      "loss": 0.7156,
+      "step": 85
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.04811069369316101,
+      "learning_rate": 0.000199660566969936,
+      "loss": 0.7725,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.05829324200749397,
+      "learning_rate": 0.0001996515809804485,
+      "loss": 0.7384,
+      "step": 87
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.05200457572937012,
+      "learning_rate": 0.00019964247780194548,
+      "loss": 0.7739,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.05518084019422531,
+      "learning_rate": 0.00019963325744513227,
+      "loss": 0.6794,
+      "step": 89
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.07461239397525787,
+      "learning_rate": 0.00019962391992085185,
+      "loss": 0.7231,
+      "step": 90
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.04882181063294411,
+      "learning_rate": 0.00019961446524008504,
+      "loss": 0.7539,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.05256763473153114,
+      "learning_rate": 0.0001996048934139504,
+      "loss": 0.7796,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.04832598939538002,
+      "learning_rate": 0.0001995952044537043,
+      "loss": 0.8047,
+      "step": 93
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.05217672884464264,
+      "learning_rate": 0.00019958539837074084,
+      "loss": 0.7766,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.049672115594148636,
+      "learning_rate": 0.00019957547517659183,
+      "loss": 0.7328,
+      "step": 95
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.0518164187669754,
+      "learning_rate": 0.00019956543488292686,
+      "loss": 0.6766,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.04681951552629471,
+      "learning_rate": 0.00019955527750155315,
+      "loss": 0.7647,
+      "step": 97
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.048327744007110596,
+      "learning_rate": 0.0001995450030444157,
+      "loss": 0.7428,
+      "step": 98
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.05234162509441376,
+      "learning_rate": 0.00019953461152359708,
+      "loss": 0.7551,
+      "step": 99
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.04491112753748894,
+      "learning_rate": 0.0001995241029513176,
+      "loss": 0.6876,
+      "step": 100
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.050490718334913254,
+      "learning_rate": 0.00019951347733993528,
+      "loss": 0.832,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.05716025456786156,
+      "learning_rate": 0.00019950273470194566,
+      "loss": 0.7393,
+      "step": 102
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.048837609589099884,
+      "learning_rate": 0.00019949187504998193,
+      "loss": 0.7305,
+      "step": 103
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.04992825165390968,
+      "learning_rate": 0.00019948089839681493,
+      "loss": 0.7371,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.05300760269165039,
+      "learning_rate": 0.00019946980475535307,
+      "loss": 0.7529,
+      "step": 105
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.049852970987558365,
+      "learning_rate": 0.0001994585941386423,
+      "loss": 0.7111,
+      "step": 106
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.05310487747192383,
+      "learning_rate": 0.0001994472665598662,
+      "loss": 0.6912,
+      "step": 107
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.05248330160975456,
+      "learning_rate": 0.00019943582203234584,
+      "loss": 0.7445,
+      "step": 108
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.05639020353555679,
+      "learning_rate": 0.00019942426056953987,
+      "loss": 0.7556,
+      "step": 109
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.050534311681985855,
+      "learning_rate": 0.00019941258218504436,
+      "loss": 0.7028,
+      "step": 110
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.05137235298752785,
+      "learning_rate": 0.00019940078689259305,
+      "loss": 0.7306,
+      "step": 111
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.05214603245258331,
+      "learning_rate": 0.00019938887470605698,
+      "loss": 0.7153,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.05015432462096214,
+      "learning_rate": 0.00019937684563944475,
+      "loss": 0.7393,
+      "step": 113
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.05170932784676552,
+      "learning_rate": 0.0001993646997069024,
+      "loss": 0.7947,
+      "step": 114
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.050847794860601425,
+      "learning_rate": 0.00019935243692271337,
+      "loss": 0.7787,
+      "step": 115
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.04788447543978691,
+      "learning_rate": 0.00019934005730129858,
+      "loss": 0.714,
+      "step": 116
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.05097344517707825,
+      "learning_rate": 0.00019932756085721629,
+      "loss": 0.8083,
+      "step": 117
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.04976595938205719,
+      "learning_rate": 0.00019931494760516217,
+      "loss": 0.8157,
+      "step": 118
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.04821658134460449,
+      "learning_rate": 0.0001993022175599692,
+      "loss": 0.6772,
+      "step": 119
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.054969336837530136,
+      "learning_rate": 0.00019928937073660782,
+      "loss": 0.7019,
+      "step": 120
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.05310753360390663,
+      "learning_rate": 0.0001992764071501857,
+      "loss": 0.7266,
+      "step": 121
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.04759340360760689,
+      "learning_rate": 0.00019926332681594787,
+      "loss": 0.7549,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.05405697599053383,
+      "learning_rate": 0.00019925012974927664,
+      "loss": 0.7416,
+      "step": 123
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.07087159901857376,
+      "learning_rate": 0.0001992368159656916,
+      "loss": 0.7354,
+      "step": 124
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.05469256639480591,
+      "learning_rate": 0.00019922338548084955,
+      "loss": 0.7082,
+      "step": 125
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.05295870825648308,
+      "learning_rate": 0.00019920983831054462,
+      "loss": 0.7852,
+      "step": 126
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.05017589032649994,
+      "learning_rate": 0.0001991961744707081,
+      "loss": 0.7362,
+      "step": 127
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.052979759871959686,
+      "learning_rate": 0.0001991823939774085,
+      "loss": 0.7946,
+      "step": 128
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.053128018975257874,
+      "learning_rate": 0.00019916849684685148,
+      "loss": 0.7697,
+      "step": 129
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.05207785964012146,
+      "learning_rate": 0.00019915448309537994,
+      "loss": 0.7327,
+      "step": 130
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.05230351909995079,
+      "learning_rate": 0.00019914035273947383,
+      "loss": 0.73,
+      "step": 131
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.049084827303886414,
+      "learning_rate": 0.0001991261057957503,
+      "loss": 0.7928,
+      "step": 132
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.05596386641263962,
+      "learning_rate": 0.00019911174228096355,
+      "loss": 0.7639,
+      "step": 133
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.052596963942050934,
+      "learning_rate": 0.00019909726221200495,
+      "loss": 0.6955,
+      "step": 134
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.05625372752547264,
+      "learning_rate": 0.00019908266560590283,
+      "loss": 0.7469,
+      "step": 135
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.05166260153055191,
+      "learning_rate": 0.00019906795247982264,
+      "loss": 0.7589,
+      "step": 136
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.052134498953819275,
+      "learning_rate": 0.0001990531228510668,
+      "loss": 0.7128,
+      "step": 137
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.051448870450258255,
+      "learning_rate": 0.0001990381767370748,
+      "loss": 0.7148,
+      "step": 138
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.053590189665555954,
+      "learning_rate": 0.00019902311415542307,
+      "loss": 0.7676,
+      "step": 139
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.05264313891530037,
+      "learning_rate": 0.00019900793512382504,
+      "loss": 0.7393,
+      "step": 140
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.05209501087665558,
+      "learning_rate": 0.00019899263966013106,
+      "loss": 0.7095,
+      "step": 141
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.050406355410814285,
+      "learning_rate": 0.00019897722778232838,
+      "loss": 0.7533,
+      "step": 142
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.053870897740125656,
+      "learning_rate": 0.00019896169950854117,
+      "loss": 0.7563,
+      "step": 143
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.06269948929548264,
+      "learning_rate": 0.00019894605485703051,
+      "loss": 0.795,
+      "step": 144
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.05105578899383545,
+      "learning_rate": 0.00019893029384619428,
+      "loss": 0.7268,
+      "step": 145
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.049749474972486496,
+      "learning_rate": 0.0001989144164945673,
+      "loss": 0.6783,
+      "step": 146
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.051559630781412125,
+      "learning_rate": 0.00019889842282082109,
+      "loss": 0.7204,
+      "step": 147
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.051908280700445175,
+      "learning_rate": 0.00019888231284376402,
+      "loss": 0.7048,
+      "step": 148
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.05648699402809143,
+      "learning_rate": 0.0001988660865823412,
+      "loss": 0.7145,
+      "step": 149
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.05817189812660217,
+      "learning_rate": 0.00019884974405563454,
+      "loss": 0.6883,
+      "step": 150
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05215545743703842,
+      "learning_rate": 0.0001988332852828626,
+      "loss": 0.7226,
+      "step": 151
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05281899496912956,
+      "learning_rate": 0.0001988167102833808,
+      "loss": 0.7174,
+      "step": 152
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05761328712105751,
+      "learning_rate": 0.00019880001907668102,
+      "loss": 0.7779,
+      "step": 153
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05430774390697479,
+      "learning_rate": 0.00019878321168239195,
+      "loss": 0.7061,
+      "step": 154
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05135492607951164,
+      "learning_rate": 0.0001987662881202789,
+      "loss": 0.7317,
+      "step": 155
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05098530277609825,
+      "learning_rate": 0.00019874924841024373,
+      "loss": 0.7592,
+      "step": 156
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.051871709525585175,
+      "learning_rate": 0.00019873209257232494,
+      "loss": 0.8104,
+      "step": 157
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.0578281544148922,
+      "learning_rate": 0.0001987148206266976,
+      "loss": 0.7072,
+      "step": 158
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05245187506079674,
+      "learning_rate": 0.00019869743259367327,
+      "loss": 0.7621,
+      "step": 159
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.05832192674279213,
+      "learning_rate": 0.00019867992849370008,
+      "loss": 0.7617,
+      "step": 160
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.0534343346953392,
+      "learning_rate": 0.00019866230834736262,
+      "loss": 0.7869,
+      "step": 161
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.061366647481918335,
+      "learning_rate": 0.00019864457217538196,
+      "loss": 0.7412,
+      "step": 162
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.05365344136953354,
+      "learning_rate": 0.00019862671999861567,
+      "loss": 0.6683,
+      "step": 163
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.05422106385231018,
+      "learning_rate": 0.0001986087518380576,
+      "loss": 0.7618,
+      "step": 164
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.05549204722046852,
+      "learning_rate": 0.00019859066771483815,
+      "loss": 0.7674,
+      "step": 165
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.053906746208667755,
+      "learning_rate": 0.000198572467650224,
+      "loss": 0.8516,
+      "step": 166
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.05335883051156998,
+      "learning_rate": 0.0001985541516656182,
+      "loss": 0.7301,
+      "step": 167
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.056282807141542435,
+      "learning_rate": 0.00019853571978256014,
+      "loss": 0.723,
+      "step": 168
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.0541800893843174,
+      "learning_rate": 0.0001985171720227254,
+      "loss": 0.7342,
+      "step": 169
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.053833167999982834,
+      "learning_rate": 0.00019849850840792605,
+      "loss": 0.742,
+      "step": 170
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.0539311021566391,
+      "learning_rate": 0.00019847972896011012,
+      "loss": 0.733,
+      "step": 171
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.055201999843120575,
+      "learning_rate": 0.00019846083370136217,
+      "loss": 0.7608,
+      "step": 172
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.05684024840593338,
+      "learning_rate": 0.00019844182265390265,
+      "loss": 0.7475,
+      "step": 173
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.05470721423625946,
+      "learning_rate": 0.00019842269584008837,
+      "loss": 0.7758,
+      "step": 174
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.05426143482327461,
+      "learning_rate": 0.00019840345328241222,
+      "loss": 0.8154,
+      "step": 175
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.05676954984664917,
+      "learning_rate": 0.00019838409500350318,
+      "loss": 0.6831,
+      "step": 176
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.06125069782137871,
+      "learning_rate": 0.0001983646210261264,
+      "loss": 0.8139,
+      "step": 177
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.055671993643045425,
+      "learning_rate": 0.000198345031373183,
+      "loss": 0.7354,
+      "step": 178
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.05255590006709099,
+      "learning_rate": 0.00019832532606771017,
+      "loss": 0.772,
+      "step": 179
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.05209804326295853,
+      "learning_rate": 0.0001983055051328811,
+      "loss": 0.702,
+      "step": 180
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.051347844302654266,
+      "learning_rate": 0.000198285568592005,
+      "loss": 0.7941,
+      "step": 181
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.049175627529621124,
+      "learning_rate": 0.00019826551646852692,
+      "loss": 0.7156,
+      "step": 182
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.06953848898410797,
+      "learning_rate": 0.000198245348786028,
+      "loss": 0.6817,
+      "step": 183
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.05407484620809555,
+      "learning_rate": 0.00019822506556822506,
+      "loss": 0.7818,
+      "step": 184
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.053331997245550156,
+      "learning_rate": 0.00019820466683897103,
+      "loss": 0.721,
+      "step": 185
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.05798467993736267,
+      "learning_rate": 0.0001981841526222545,
+      "loss": 0.7407,
+      "step": 186
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.05653946474194527,
+      "learning_rate": 0.00019816352294219994,
+      "loss": 0.7092,
+      "step": 187
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.05470096692442894,
+      "learning_rate": 0.0001981427778230676,
+      "loss": 0.7337,
+      "step": 188
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.06434475630521774,
+      "learning_rate": 0.0001981219172892535,
+      "loss": 0.6977,
+      "step": 189
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.052794575691223145,
+      "learning_rate": 0.00019810094136528934,
+      "loss": 0.7601,
+      "step": 190
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.05482126399874687,
+      "learning_rate": 0.00019807985007584255,
+      "loss": 0.7619,
+      "step": 191
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.06532236188650131,
+      "learning_rate": 0.00019805864344571626,
+      "loss": 0.7934,
+      "step": 192
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.0744403600692749,
+      "learning_rate": 0.00019803732149984912,
+      "loss": 0.6666,
+      "step": 193
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.05643633008003235,
+      "learning_rate": 0.00019801588426331557,
+      "loss": 0.7519,
+      "step": 194
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.05338543280959129,
+      "learning_rate": 0.00019799433176132544,
+      "loss": 0.7072,
+      "step": 195
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.05368245020508766,
+      "learning_rate": 0.00019797266401922427,
+      "loss": 0.7325,
+      "step": 196
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.05443282797932625,
+      "learning_rate": 0.00019795088106249305,
+      "loss": 0.8166,
+      "step": 197
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.052494440227746964,
+      "learning_rate": 0.00019792898291674823,
+      "loss": 0.7339,
+      "step": 198
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.06454979628324509,
+      "learning_rate": 0.0001979069696077418,
+      "loss": 0.803,
+      "step": 199
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.054398063570261,
+      "learning_rate": 0.00019788484116136108,
+      "loss": 0.7857,
+      "step": 200
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.05177140235900879,
+      "learning_rate": 0.0001978625976036289,
+      "loss": 0.7605,
+      "step": 201
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.05473088100552559,
+      "learning_rate": 0.00019784023896070337,
+      "loss": 0.7145,
+      "step": 202
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.06038124859333038,
+      "learning_rate": 0.00019781776525887797,
+      "loss": 0.7154,
+      "step": 203
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.054728008806705475,
+      "learning_rate": 0.00019779517652458156,
+      "loss": 0.7575,
+      "step": 204
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.05367952585220337,
+      "learning_rate": 0.0001977724727843781,
+      "loss": 0.7259,
+      "step": 205
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.056090373545885086,
+      "learning_rate": 0.00019774965406496703,
+      "loss": 0.796,
+      "step": 206
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.056659117341041565,
+      "learning_rate": 0.00019772672039318278,
+      "loss": 0.7572,
+      "step": 207
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.061686623841524124,
+      "learning_rate": 0.0001977036717959951,
+      "loss": 0.7514,
+      "step": 208
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.05229324847459793,
+      "learning_rate": 0.00019768050830050888,
+      "loss": 0.7923,
+      "step": 209
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.05652037635445595,
+      "learning_rate": 0.00019765722993396402,
+      "loss": 0.7452,
+      "step": 210
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.0566401481628418,
+      "learning_rate": 0.00019763383672373566,
+      "loss": 0.6641,
+      "step": 211
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.061378199607133865,
+      "learning_rate": 0.00019761032869733395,
+      "loss": 0.7042,
+      "step": 212
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.05892565846443176,
+      "learning_rate": 0.00019758670588240396,
+      "loss": 0.7209,
+      "step": 213
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.054732099175453186,
+      "learning_rate": 0.00019756296830672585,
+      "loss": 0.7396,
+      "step": 214
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.053043294697999954,
+      "learning_rate": 0.0001975391159982147,
+      "loss": 0.7441,
+      "step": 215
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.058401089161634445,
+      "learning_rate": 0.00019751514898492056,
+      "loss": 0.7553,
+      "step": 216
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.05779936537146568,
+      "learning_rate": 0.0001974910672950283,
+      "loss": 0.6659,
+      "step": 217
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.0589526928961277,
+      "learning_rate": 0.00019746687095685775,
+      "loss": 0.7177,
+      "step": 218
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.05505611002445221,
+      "learning_rate": 0.00019744255999886345,
+      "loss": 0.702,
+      "step": 219
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.053444795310497284,
+      "learning_rate": 0.00019741813444963476,
+      "loss": 0.7458,
+      "step": 220
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.05537053942680359,
+      "learning_rate": 0.0001973935943378959,
+      "loss": 0.7178,
+      "step": 221
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.05319499969482422,
+      "learning_rate": 0.00019736893969250568,
+      "loss": 0.6376,
+      "step": 222
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.05316654592752457,
+      "learning_rate": 0.0001973441705424577,
+      "loss": 0.7649,
+      "step": 223
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.06086993217468262,
+      "learning_rate": 0.00019731928691688006,
+      "loss": 0.7171,
+      "step": 224
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.058167632669210434,
+      "learning_rate": 0.0001972942888450357,
+      "loss": 0.7636,
+      "step": 225
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.057685092091560364,
+      "learning_rate": 0.00019726917635632204,
+      "loss": 0.7487,
+      "step": 226
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.056136585772037506,
+      "learning_rate": 0.000197243949480271,
+      "loss": 0.7456,
+      "step": 227
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.052869562059640884,
+      "learning_rate": 0.0001972186082465491,
+      "loss": 0.693,
+      "step": 228
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.05286895111203194,
+      "learning_rate": 0.0001971931526849573,
+      "loss": 0.713,
+      "step": 229
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.05643904581665993,
+      "learning_rate": 0.00019716758282543106,
+      "loss": 0.7467,
+      "step": 230
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.05756821483373642,
+      "learning_rate": 0.00019714189869804018,
+      "loss": 0.675,
+      "step": 231
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.05546725168824196,
+      "learning_rate": 0.00019711610033298888,
+      "loss": 0.7998,
+      "step": 232
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.06143364682793617,
+      "learning_rate": 0.0001970901877606157,
+      "loss": 0.728,
+      "step": 233
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.06431245058774948,
+      "learning_rate": 0.00019706416101139353,
+      "loss": 0.7068,
+      "step": 234
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.054505374282598495,
+      "learning_rate": 0.0001970380201159295,
+      "loss": 0.6696,
+      "step": 235
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.054913364350795746,
+      "learning_rate": 0.00019701176510496495,
+      "loss": 0.7246,
+      "step": 236
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.06259609013795853,
+      "learning_rate": 0.0001969853960093755,
+      "loss": 0.7613,
+      "step": 237
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.0802953839302063,
+      "learning_rate": 0.00019695891286017082,
+      "loss": 0.7843,
+      "step": 238
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.054023925215005875,
+      "learning_rate": 0.00019693231568849475,
+      "loss": 0.7848,
+      "step": 239
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.05291634425520897,
+      "learning_rate": 0.0001969056045256253,
+      "loss": 0.5979,
+      "step": 240
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.052885495126247406,
+      "learning_rate": 0.00019687877940297442,
+      "loss": 0.715,
+      "step": 241
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.054459016770124435,
+      "learning_rate": 0.00019685184035208813,
+      "loss": 0.7413,
+      "step": 242
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.058789074420928955,
+      "learning_rate": 0.0001968247874046464,
+      "loss": 0.7637,
+      "step": 243
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.8141337633132935,
+      "eval_runtime": 111.9734,
+      "eval_samples_per_second": 88.753,
+      "eval_steps_per_second": 22.193,
+      "step": 243
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.054029788821935654,
+      "learning_rate": 0.0001967976205924632,
+      "loss": 0.7184,
+      "step": 244
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.055858369916677475,
+      "learning_rate": 0.00019677033994748631,
+      "loss": 0.7238,
+      "step": 245
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.05345359072089195,
+      "learning_rate": 0.0001967429455017975,
+      "loss": 0.6889,
+      "step": 246
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.06066269427537918,
+      "learning_rate": 0.00019671543728761224,
+      "loss": 0.8698,
+      "step": 247
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.058042511343955994,
+      "learning_rate": 0.00019668781533727984,
+      "loss": 0.6563,
+      "step": 248
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.056662410497665405,
+      "learning_rate": 0.00019666007968328336,
+      "loss": 0.6852,
+      "step": 249
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.05772121623158455,
+      "learning_rate": 0.00019663223035823964,
+      "loss": 0.7117,
+      "step": 250
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.0612647645175457,
+      "learning_rate": 0.00019660426739489913,
+      "loss": 0.7662,
+      "step": 251
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.06314199417829514,
+      "learning_rate": 0.00019657619082614586,
+      "loss": 0.7101,
+      "step": 252
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.05658784881234169,
+      "learning_rate": 0.0001965480006849976,
+      "loss": 0.7685,
+      "step": 253
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.05959884822368622,
+      "learning_rate": 0.00019651969700460557,
+      "loss": 0.7267,
+      "step": 254
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.058939650654792786,
+      "learning_rate": 0.00019649127981825456,
+      "loss": 0.6825,
+      "step": 255
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.060037851333618164,
+      "learning_rate": 0.00019646274915936278,
+      "loss": 0.7476,
+      "step": 256
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.06029137596487999,
+      "learning_rate": 0.00019643410506148197,
+      "loss": 0.7497,
+      "step": 257
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.05577164515852928,
+      "learning_rate": 0.00019640534755829725,
+      "loss": 0.7159,
+      "step": 258
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.05618567392230034,
+      "learning_rate": 0.00019637647668362704,
+      "loss": 0.7257,
+      "step": 259
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.06020184978842735,
+      "learning_rate": 0.00019634749247142313,
+      "loss": 0.6776,
+      "step": 260
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.06419654935598373,
+      "learning_rate": 0.00019631839495577066,
+      "loss": 0.6984,
+      "step": 261
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.057507649064064026,
+      "learning_rate": 0.00019628918417088784,
+      "loss": 0.6914,
+      "step": 262
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.05764051154255867,
+      "learning_rate": 0.00019625986015112628,
+      "loss": 0.7786,
+      "step": 263
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.056871023029088974,
+      "learning_rate": 0.0001962304229309706,
+      "loss": 0.7259,
+      "step": 264
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.05893797427415848,
+      "learning_rate": 0.00019620087254503866,
+      "loss": 0.7435,
+      "step": 265
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.0583423487842083,
+      "learning_rate": 0.00019617120902808125,
+      "loss": 0.7597,
+      "step": 266
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.058257438242435455,
+      "learning_rate": 0.00019614143241498242,
+      "loss": 0.6764,
+      "step": 267
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.06315512955188751,
+      "learning_rate": 0.00019611154274075903,
+      "loss": 0.6298,
+      "step": 268
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.06553982943296432,
+      "learning_rate": 0.00019608154004056093,
+      "loss": 0.6789,
+      "step": 269
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.05806327983736992,
+      "learning_rate": 0.000196051424349671,
+      "loss": 0.7067,
+      "step": 270
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.057144396007061005,
+      "learning_rate": 0.00019602119570350483,
+      "loss": 0.6802,
+      "step": 271
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.05949268490076065,
+      "learning_rate": 0.000195990854137611,
+      "loss": 0.7046,
+      "step": 272
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.05610249564051628,
+      "learning_rate": 0.0001959603996876708,
+      "loss": 0.7682,
+      "step": 273
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.056379836052656174,
+      "learning_rate": 0.00019592983238949824,
+      "loss": 0.714,
+      "step": 274
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.05497708171606064,
+      "learning_rate": 0.00019589915227904016,
+      "loss": 0.6559,
+      "step": 275
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.06712428480386734,
+      "learning_rate": 0.0001958683593923759,
+      "loss": 0.772,
+      "step": 276
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.054381344467401505,
+      "learning_rate": 0.00019583745376571755,
+      "loss": 0.7686,
+      "step": 277
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.059507984668016434,
+      "learning_rate": 0.00019580643543540976,
+      "loss": 0.7504,
+      "step": 278
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.06055927649140358,
+      "learning_rate": 0.00019577530443792967,
+      "loss": 0.7746,
+      "step": 279
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.05875448137521744,
+      "learning_rate": 0.00019574406080988697,
+      "loss": 0.7587,
+      "step": 280
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.05514727905392647,
+      "learning_rate": 0.00019571270458802373,
+      "loss": 0.7401,
+      "step": 281
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.05958930030465126,
+      "learning_rate": 0.00019568123580921453,
+      "loss": 0.7265,
+      "step": 282
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.05936933308839798,
+      "learning_rate": 0.00019564965451046625,
+      "loss": 0.7312,
+      "step": 283
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.06500967592000961,
+      "learning_rate": 0.00019561796072891806,
+      "loss": 0.744,
+      "step": 284
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.05779334157705307,
+      "learning_rate": 0.0001955861545018415,
+      "loss": 0.7134,
+      "step": 285
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.05654512718319893,
+      "learning_rate": 0.00019555423586664028,
+      "loss": 0.7576,
+      "step": 286
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.08160107582807541,
+      "learning_rate": 0.00019552220486085032,
+      "loss": 0.7471,
+      "step": 287
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.05500080808997154,
+      "learning_rate": 0.0001954900615221397,
+      "loss": 0.7135,
+      "step": 288
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.05892986059188843,
+      "learning_rate": 0.00019545780588830854,
+      "loss": 0.7841,
+      "step": 289
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.05923021584749222,
+      "learning_rate": 0.00019542543799728907,
+      "loss": 0.7111,
+      "step": 290
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.059189844876527786,
+      "learning_rate": 0.00019539295788714554,
+      "loss": 0.7892,
+      "step": 291
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.0577772781252861,
+      "learning_rate": 0.00019536036559607417,
+      "loss": 0.773,
+      "step": 292
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.06281998008489609,
+      "learning_rate": 0.0001953276611624031,
+      "loss": 0.791,
+      "step": 293
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.061750974506139755,
+      "learning_rate": 0.00019529484462459226,
+      "loss": 0.7482,
+      "step": 294
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.06235501915216446,
+      "learning_rate": 0.00019526191602123357,
+      "loss": 0.7399,
+      "step": 295
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.057336192578077316,
+      "learning_rate": 0.00019522887539105064,
+      "loss": 0.7354,
+      "step": 296
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.05565524473786354,
+      "learning_rate": 0.0001951957227728988,
+      "loss": 0.6847,
+      "step": 297
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.06352953612804413,
+      "learning_rate": 0.00019516245820576513,
+      "loss": 0.7753,
+      "step": 298
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.054871831089258194,
+      "learning_rate": 0.0001951290817287684,
+      "loss": 0.7619,
+      "step": 299
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.05734177678823471,
+      "learning_rate": 0.00019509559338115883,
+      "loss": 0.7775,
+      "step": 300
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.05894226208329201,
+      "learning_rate": 0.00019506199320231838,
+      "loss": 0.7182,
+      "step": 301
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.05706853047013283,
+      "learning_rate": 0.00019502828123176043,
+      "loss": 0.795,
+      "step": 302
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.06002698838710785,
+      "learning_rate": 0.00019499445750912982,
+      "loss": 0.8106,
+      "step": 303
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.05632604658603668,
+      "learning_rate": 0.0001949605220742028,
+      "loss": 0.7415,
+      "step": 304
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.05873245373368263,
+      "learning_rate": 0.00019492647496688705,
+      "loss": 0.7293,
+      "step": 305
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.060128260403871536,
+      "learning_rate": 0.00019489231622722155,
+      "loss": 0.7364,
+      "step": 306
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.05893580988049507,
+      "learning_rate": 0.00019485804589537654,
+      "loss": 0.7152,
+      "step": 307
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.07257378846406937,
+      "learning_rate": 0.0001948236640116535,
+      "loss": 0.7262,
+      "step": 308
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.057697609066963196,
+      "learning_rate": 0.0001947891706164851,
+      "loss": 0.6915,
+      "step": 309
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.05808861181139946,
+      "learning_rate": 0.00019475456575043513,
+      "loss": 0.6977,
+      "step": 310
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.060822539031505585,
+      "learning_rate": 0.0001947198494541985,
+      "loss": 0.7963,
+      "step": 311
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.05885370075702667,
+      "learning_rate": 0.00019468502176860116,
+      "loss": 0.782,
+      "step": 312
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.05855882540345192,
+      "learning_rate": 0.00019465008273459997,
+      "loss": 0.7248,
+      "step": 313
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.06043313443660736,
+      "learning_rate": 0.00019461503239328283,
+      "loss": 0.7003,
+      "step": 314
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.0554848350584507,
+      "learning_rate": 0.00019457987078586852,
+      "loss": 0.6923,
+      "step": 315
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.05844718590378761,
+      "learning_rate": 0.00019454459795370665,
+      "loss": 0.7486,
+      "step": 316
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.059286076575517654,
+      "learning_rate": 0.0001945092139382776,
+      "loss": 0.7273,
+      "step": 317
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.07302168011665344,
+      "learning_rate": 0.00019447371878119257,
+      "loss": 0.7323,
+      "step": 318
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.0687604621052742,
+      "learning_rate": 0.00019443811252419337,
+      "loss": 0.7803,
+      "step": 319
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.057997170835733414,
+      "learning_rate": 0.00019440239520915258,
+      "loss": 0.7097,
+      "step": 320
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.05903468281030655,
+      "learning_rate": 0.00019436656687807328,
+      "loss": 0.7934,
+      "step": 321
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.06016958877444267,
+      "learning_rate": 0.00019433062757308914,
+      "loss": 0.6503,
+      "step": 322
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.06088041886687279,
+      "learning_rate": 0.00019429457733646433,
+      "loss": 0.7536,
+      "step": 323
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.057344887405633926,
+      "learning_rate": 0.00019425841621059347,
+      "loss": 0.6832,
+      "step": 324
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.061634745448827744,
+      "learning_rate": 0.0001942221442380016,
+      "loss": 0.6935,
+      "step": 325
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.06309819221496582,
+      "learning_rate": 0.00019418576146134419,
+      "loss": 0.7277,
+      "step": 326
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.05775715038180351,
+      "learning_rate": 0.0001941492679234068,
+      "loss": 0.6798,
+      "step": 327
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.05710741505026817,
+      "learning_rate": 0.00019411266366710543,
+      "loss": 0.6822,
+      "step": 328
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.05784869194030762,
+      "learning_rate": 0.0001940759487354862,
+      "loss": 0.6763,
+      "step": 329
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.05700847879052162,
+      "learning_rate": 0.0001940391231717255,
+      "loss": 0.6696,
+      "step": 330
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.06073322892189026,
+      "learning_rate": 0.00019400218701912965,
+      "loss": 0.7032,
+      "step": 331
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.06225202605128288,
+      "learning_rate": 0.00019396514032113513,
+      "loss": 0.7258,
+      "step": 332
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.06155748292803764,
+      "learning_rate": 0.00019392798312130836,
+      "loss": 0.7372,
+      "step": 333
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.06383652240037918,
+      "learning_rate": 0.00019389071546334585,
+      "loss": 0.7572,
+      "step": 334
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.057412583380937576,
+      "learning_rate": 0.00019385333739107375,
+      "loss": 0.6642,
+      "step": 335
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.061736416071653366,
+      "learning_rate": 0.00019381584894844834,
+      "loss": 0.7309,
+      "step": 336
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.060744259506464005,
+      "learning_rate": 0.00019377825017955549,
+      "loss": 0.7161,
+      "step": 337
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.05964282155036926,
+      "learning_rate": 0.00019374054112861089,
+      "loss": 0.7405,
+      "step": 338
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.06293050944805145,
+      "learning_rate": 0.00019370272183995993,
+      "loss": 0.645,
+      "step": 339
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.05713023245334625,
+      "learning_rate": 0.0001936647923580776,
+      "loss": 0.7232,
+      "step": 340
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.05848432704806328,
+      "learning_rate": 0.0001936267527275685,
+      "loss": 0.735,
+      "step": 341
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.05630767345428467,
+      "learning_rate": 0.00019358860299316678,
+      "loss": 0.7797,
+      "step": 342
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.06115993112325668,
+      "learning_rate": 0.00019355034319973603,
+      "loss": 0.6872,
+      "step": 343
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.06437361240386963,
+      "learning_rate": 0.00019351197339226935,
+      "loss": 0.7728,
+      "step": 344
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.06171271950006485,
+      "learning_rate": 0.00019347349361588907,
+      "loss": 0.7873,
+      "step": 345
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.06096939742565155,
+      "learning_rate": 0.00019343490391584696,
+      "loss": 0.702,
+      "step": 346
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.05415042117238045,
+      "learning_rate": 0.000193396204337524,
+      "loss": 0.6321,
+      "step": 347
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.05933550372719765,
+      "learning_rate": 0.0001933573949264305,
+      "loss": 0.7682,
+      "step": 348
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.05992317944765091,
+      "learning_rate": 0.00019331847572820574,
+      "loss": 0.7141,
+      "step": 349
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.061700768768787384,
+      "learning_rate": 0.00019327944678861826,
+      "loss": 0.787,
+      "step": 350
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.06429409980773926,
+      "learning_rate": 0.0001932403081535656,
+      "loss": 0.8566,
+      "step": 351
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.060838643461465836,
+      "learning_rate": 0.00019320105986907432,
+      "loss": 0.7432,
+      "step": 352
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.06311419606208801,
+      "learning_rate": 0.0001931617019812999,
+      "loss": 0.7504,
+      "step": 353
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.05914825573563576,
+      "learning_rate": 0.00019312223453652673,
+      "loss": 0.7791,
+      "step": 354
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.06227264553308487,
+      "learning_rate": 0.000193082657581168,
+      "loss": 0.7584,
+      "step": 355
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.05977202579379082,
+      "learning_rate": 0.00019304297116176574,
+      "loss": 0.6649,
+      "step": 356
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.059180788695812225,
+      "learning_rate": 0.00019300317532499069,
+      "loss": 0.6559,
+      "step": 357
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.06028233468532562,
+      "learning_rate": 0.00019296327011764225,
+      "loss": 0.7214,
+      "step": 358
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.06343667954206467,
+      "learning_rate": 0.00019292325558664843,
+      "loss": 0.6716,
+      "step": 359
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.059275731444358826,
+      "learning_rate": 0.00019288313177906588,
+      "loss": 0.7321,
+      "step": 360
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.06691042333841324,
+      "learning_rate": 0.00019284289874207963,
+      "loss": 0.7095,
+      "step": 361
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.06491021066904068,
+      "learning_rate": 0.00019280255652300325,
+      "loss": 0.6874,
+      "step": 362
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.06535612046718597,
+      "learning_rate": 0.00019276210516927872,
+      "loss": 0.7175,
+      "step": 363
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06091107428073883,
+      "learning_rate": 0.00019272154472847625,
+      "loss": 0.6964,
+      "step": 364
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06472720205783844,
+      "learning_rate": 0.00019268087524829447,
+      "loss": 0.7771,
+      "step": 365
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06106911227107048,
+      "learning_rate": 0.0001926400967765602,
+      "loss": 0.7023,
+      "step": 366
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06196704879403114,
+      "learning_rate": 0.00019259920936122836,
+      "loss": 0.7171,
+      "step": 367
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06340812891721725,
+      "learning_rate": 0.00019255821305038206,
+      "loss": 0.6776,
+      "step": 368
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.05937547609210014,
+      "learning_rate": 0.0001925171078922325,
+      "loss": 0.6742,
+      "step": 369
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.05785318836569786,
+      "learning_rate": 0.00019247589393511872,
+      "loss": 0.6348,
+      "step": 370
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.060800664126873016,
+      "learning_rate": 0.00019243457122750793,
+      "loss": 0.7789,
+      "step": 371
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06566699594259262,
+      "learning_rate": 0.00019239313981799506,
+      "loss": 0.7235,
+      "step": 372
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.06358596682548523,
+      "learning_rate": 0.00019235159975530293,
+      "loss": 0.6262,
+      "step": 373
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.06445778906345367,
+      "learning_rate": 0.00019230995108828215,
+      "loss": 0.6495,
+      "step": 374
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.059928618371486664,
+      "learning_rate": 0.00019226819386591106,
+      "loss": 0.6459,
+      "step": 375
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.0632234737277031,
+      "learning_rate": 0.0001922263281372956,
+      "loss": 0.6777,
+      "step": 376
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.06647662818431854,
+      "learning_rate": 0.00019218435395166933,
+      "loss": 0.6579,
+      "step": 377
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.07134454697370529,
+      "learning_rate": 0.0001921422713583934,
+      "loss": 0.7326,
+      "step": 378
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.056324154138565063,
+      "learning_rate": 0.00019210008040695638,
+      "loss": 0.6974,
+      "step": 379
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.06590214371681213,
+      "learning_rate": 0.00019205778114697437,
+      "loss": 0.6706,
+      "step": 380
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.05920105054974556,
+      "learning_rate": 0.0001920153736281907,
+      "loss": 0.8247,
+      "step": 381
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.06643356382846832,
+      "learning_rate": 0.00019197285790047608,
+      "loss": 0.7184,
+      "step": 382
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.07437850534915924,
+      "learning_rate": 0.0001919302340138285,
+      "loss": 0.8304,
+      "step": 383
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.059930771589279175,
+      "learning_rate": 0.0001918875020183732,
+      "loss": 0.6705,
+      "step": 384
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.05600350350141525,
+      "learning_rate": 0.00019184466196436236,
+      "loss": 0.6985,
+      "step": 385
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.05550659820437431,
+      "learning_rate": 0.00019180171390217537,
+      "loss": 0.7603,
+      "step": 386
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.0702609270811081,
+      "learning_rate": 0.00019175865788231863,
+      "loss": 0.6904,
+      "step": 387
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.07218141108751297,
+      "learning_rate": 0.00019171549395542546,
+      "loss": 0.7817,
+      "step": 388
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.06199540197849274,
+      "learning_rate": 0.00019167222217225617,
+      "loss": 0.7508,
+      "step": 389
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.06544966995716095,
+      "learning_rate": 0.00019162884258369775,
+      "loss": 0.7264,
+      "step": 390
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.06334829330444336,
+      "learning_rate": 0.0001915853552407641,
+      "loss": 0.7282,
+      "step": 391
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.06337782740592957,
+      "learning_rate": 0.0001915417601945958,
+      "loss": 0.7331,
+      "step": 392
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.06350013613700867,
+      "learning_rate": 0.00019149805749646006,
+      "loss": 0.6549,
+      "step": 393
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.05926767736673355,
+      "learning_rate": 0.0001914542471977507,
+      "loss": 0.7421,
+      "step": 394
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.06383328139781952,
+      "learning_rate": 0.00019141032934998813,
+      "loss": 0.7322,
+      "step": 395
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.07084285467863083,
+      "learning_rate": 0.00019136630400481912,
+      "loss": 0.7155,
+      "step": 396
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.06484265625476837,
+      "learning_rate": 0.00019132217121401696,
+      "loss": 0.6974,
+      "step": 397
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.06470082700252533,
+      "learning_rate": 0.0001912779310294813,
+      "loss": 0.8072,
+      "step": 398
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.0596977099776268,
+      "learning_rate": 0.000191233583503238,
+      "loss": 0.6591,
+      "step": 399
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.057743631303310394,
+      "learning_rate": 0.0001911891286874392,
+      "loss": 0.7119,
+      "step": 400
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.06098794564604759,
+      "learning_rate": 0.00019114456663436324,
+      "loss": 0.6982,
+      "step": 401
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.05964438244700432,
+      "learning_rate": 0.00019109989739641447,
+      "loss": 0.6741,
+      "step": 402
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06789212673902512,
+      "learning_rate": 0.00019105512102612343,
+      "loss": 0.7152,
+      "step": 403
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06945686042308807,
+      "learning_rate": 0.00019101023757614656,
+      "loss": 0.7184,
+      "step": 404
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06108655408024788,
+      "learning_rate": 0.0001909652470992662,
+      "loss": 0.7476,
+      "step": 405
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06126492843031883,
+      "learning_rate": 0.0001909201496483906,
+      "loss": 0.7669,
+      "step": 406
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06494157016277313,
+      "learning_rate": 0.00019087494527655384,
+      "loss": 0.7191,
+      "step": 407
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06480975449085236,
+      "learning_rate": 0.00019082963403691563,
+      "loss": 0.7215,
+      "step": 408
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06445412337779999,
+      "learning_rate": 0.00019078421598276148,
+      "loss": 0.7358,
+      "step": 409
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06461311876773834,
+      "learning_rate": 0.00019073869116750244,
+      "loss": 0.7462,
+      "step": 410
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.06038064882159233,
+      "learning_rate": 0.00019069305964467508,
+      "loss": 0.6975,
+      "step": 411
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.058075543493032455,
+      "learning_rate": 0.00019064732146794157,
+      "loss": 0.7348,
+      "step": 412
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06547833234071732,
+      "learning_rate": 0.00019060147669108942,
+      "loss": 0.6846,
+      "step": 413
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.05910010635852814,
+      "learning_rate": 0.00019055552536803146,
+      "loss": 0.6978,
+      "step": 414
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.061262935400009155,
+      "learning_rate": 0.00019050946755280594,
+      "loss": 0.7925,
+      "step": 415
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06281925737857819,
+      "learning_rate": 0.00019046330329957622,
+      "loss": 0.7537,
+      "step": 416
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06479451060295105,
+      "learning_rate": 0.00019041703266263094,
+      "loss": 0.7462,
+      "step": 417
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06690587103366852,
+      "learning_rate": 0.00019037065569638375,
+      "loss": 0.7118,
+      "step": 418
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06481601297855377,
+      "learning_rate": 0.00019032417245537343,
+      "loss": 0.6746,
+      "step": 419
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.06519322842359543,
+      "learning_rate": 0.0001902775829942636,
+      "loss": 0.7062,
+      "step": 420
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.058651816099882126,
+      "learning_rate": 0.000190230887367843,
+      "loss": 0.7479,
+      "step": 421
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06371847540140152,
+      "learning_rate": 0.00019018408563102504,
+      "loss": 0.7954,
+      "step": 422
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.061327967792749405,
+      "learning_rate": 0.00019013717783884798,
+      "loss": 0.6943,
+      "step": 423
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06613918393850327,
+      "learning_rate": 0.00019009016404647477,
+      "loss": 0.7395,
+      "step": 424
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06330893188714981,
+      "learning_rate": 0.00019004304430919312,
+      "loss": 0.6978,
+      "step": 425
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.061978746205568314,
+      "learning_rate": 0.0001899958186824152,
+      "loss": 0.6608,
+      "step": 426
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06049848347902298,
+      "learning_rate": 0.00018994848722167778,
+      "loss": 0.8073,
+      "step": 427
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.061442065984010696,
+      "learning_rate": 0.00018990104998264205,
+      "loss": 0.6818,
+      "step": 428
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.0643540471792221,
+      "learning_rate": 0.0001898535070210936,
+      "loss": 0.7317,
+      "step": 429
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.06444810330867767,
+      "learning_rate": 0.00018980585839294237,
+      "loss": 0.6947,
+      "step": 430
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.063243567943573,
+      "learning_rate": 0.00018975810415422257,
+      "loss": 0.8226,
+      "step": 431
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.062445830553770065,
+      "learning_rate": 0.00018971024436109258,
+      "loss": 0.6881,
+      "step": 432
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.06391721218824387,
+      "learning_rate": 0.00018966227906983492,
+      "loss": 0.7531,
+      "step": 433
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.07556512951850891,
+      "learning_rate": 0.00018961420833685618,
+      "loss": 0.6901,
+      "step": 434
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.07024349272251129,
+      "learning_rate": 0.0001895660322186869,
+      "loss": 0.6927,
+      "step": 435
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.062289077788591385,
+      "learning_rate": 0.00018951775077198163,
+      "loss": 0.6386,
+      "step": 436
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.0685746967792511,
+      "learning_rate": 0.00018946936405351875,
+      "loss": 0.7119,
+      "step": 437
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.06907238811254501,
+      "learning_rate": 0.00018942087212020045,
+      "loss": 0.767,
+      "step": 438
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.05862850695848465,
+      "learning_rate": 0.0001893722750290526,
+      "loss": 0.7121,
+      "step": 439
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.06181567534804344,
+      "learning_rate": 0.00018932357283722482,
+      "loss": 0.709,
+      "step": 440
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.06529933214187622,
+      "learning_rate": 0.00018927476560199022,
+      "loss": 0.7013,
+      "step": 441
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.06098626181483269,
+      "learning_rate": 0.00018922585338074555,
+      "loss": 0.6641,
+      "step": 442
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.06008550152182579,
+      "learning_rate": 0.00018917683623101094,
+      "loss": 0.7429,
+      "step": 443
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.06048089265823364,
+      "learning_rate": 0.00018912771421042995,
+      "loss": 0.7312,
+      "step": 444
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.06655411422252655,
+      "learning_rate": 0.0001890784873767695,
+      "loss": 0.7395,
+      "step": 445
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.07008501142263412,
+      "learning_rate": 0.00018902915578791968,
+      "loss": 0.7715,
+      "step": 446
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.07080001384019852,
+      "learning_rate": 0.00018897971950189386,
+      "loss": 0.7745,
+      "step": 447
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.06971325725317001,
+      "learning_rate": 0.00018893017857682848,
+      "loss": 0.7391,
+      "step": 448
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.06647112965583801,
+      "learning_rate": 0.00018888053307098305,
+      "loss": 0.6923,
+      "step": 449
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.06085026636719704,
+      "learning_rate": 0.0001888307830427401,
+      "loss": 0.7561,
+      "step": 450
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.06949692964553833,
+      "learning_rate": 0.00018878092855060497,
+      "loss": 0.7839,
+      "step": 451
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.0794142335653305,
+      "learning_rate": 0.00018873096965320595,
+      "loss": 0.6829,
+      "step": 452
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.07217901945114136,
+      "learning_rate": 0.00018868090640929413,
+      "loss": 0.7016,
+      "step": 453
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.06830142438411713,
+      "learning_rate": 0.00018863073887774318,
+      "loss": 0.7442,
+      "step": 454
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.0632953941822052,
+      "learning_rate": 0.00018858046711754954,
+      "loss": 0.7711,
+      "step": 455
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.0693698301911354,
+      "learning_rate": 0.00018853009118783217,
+      "loss": 0.8282,
+      "step": 456
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.061216652393341064,
+      "learning_rate": 0.00018847961114783256,
+      "loss": 0.7011,
+      "step": 457
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.06573464721441269,
+      "learning_rate": 0.00018842902705691454,
+      "loss": 0.7428,
+      "step": 458
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.06309016048908234,
+      "learning_rate": 0.0001883783389745644,
+      "loss": 0.7338,
+      "step": 459
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.06523136794567108,
+      "learning_rate": 0.00018832754696039073,
+      "loss": 0.7215,
+      "step": 460
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06510775536298752,
+      "learning_rate": 0.00018827665107412424,
+      "loss": 0.7303,
+      "step": 461
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06285244226455688,
+      "learning_rate": 0.0001882256513756179,
+      "loss": 0.8391,
+      "step": 462
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06129281967878342,
+      "learning_rate": 0.0001881745479248467,
+      "loss": 0.651,
+      "step": 463
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06267630308866501,
+      "learning_rate": 0.00018812334078190769,
+      "loss": 0.6906,
+      "step": 464
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06995107978582382,
+      "learning_rate": 0.00018807203000701978,
+      "loss": 0.7636,
+      "step": 465
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06257961690425873,
+      "learning_rate": 0.00018802061566052384,
+      "loss": 0.7336,
+      "step": 466
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.061007607728242874,
+      "learning_rate": 0.00018796909780288248,
+      "loss": 0.6634,
+      "step": 467
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06716372072696686,
+      "learning_rate": 0.0001879174764946801,
+      "loss": 0.7982,
+      "step": 468
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06520219892263412,
+      "learning_rate": 0.00018786575179662265,
+      "loss": 0.7175,
+      "step": 469
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.06266435235738754,
+      "learning_rate": 0.00018781392376953778,
+      "loss": 0.8244,
+      "step": 470
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06289061903953552,
+      "learning_rate": 0.00018776199247437457,
+      "loss": 0.7139,
+      "step": 471
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06526096910238266,
+      "learning_rate": 0.00018770995797220356,
+      "loss": 0.7111,
+      "step": 472
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06748399138450623,
+      "learning_rate": 0.00018765782032421674,
+      "loss": 0.7057,
+      "step": 473
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06067729741334915,
+      "learning_rate": 0.0001876055795917273,
+      "loss": 0.616,
+      "step": 474
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.07784713804721832,
+      "learning_rate": 0.00018755323583616966,
+      "loss": 0.7285,
+      "step": 475
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06458035111427307,
+      "learning_rate": 0.00018750078911909947,
+      "loss": 0.7296,
+      "step": 476
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06379619985818863,
+      "learning_rate": 0.00018744823950219339,
+      "loss": 0.6915,
+      "step": 477
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06064402312040329,
+      "learning_rate": 0.00018739558704724911,
+      "loss": 0.7291,
+      "step": 478
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06831635534763336,
+      "learning_rate": 0.00018734283181618525,
+      "loss": 0.7506,
+      "step": 479
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.06278979033231735,
+      "learning_rate": 0.00018728997387104136,
+      "loss": 0.7389,
+      "step": 480
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.072098508477211,
+      "learning_rate": 0.00018723701327397766,
+      "loss": 0.7571,
+      "step": 481
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.14378315210342407,
+      "learning_rate": 0.00018718395008727518,
+      "loss": 0.7914,
+      "step": 482
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.06290939450263977,
+      "learning_rate": 0.0001871307843733356,
+      "loss": 0.6587,
+      "step": 483
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.06457720696926117,
+      "learning_rate": 0.00018707751619468103,
+      "loss": 0.6835,
+      "step": 484
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.07224012911319733,
+      "learning_rate": 0.0001870241456139543,
+      "loss": 0.6609,
+      "step": 485
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.06716971844434738,
+      "learning_rate": 0.00018697067269391848,
+      "loss": 0.7688,
+      "step": 486
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.8164525628089905,
+      "eval_runtime": 111.6882,
+      "eval_samples_per_second": 88.98,
+      "eval_steps_per_second": 22.249,
+      "step": 486
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.07206602394580841,
+      "learning_rate": 0.00018691709749745706,
+      "loss": 0.7399,
+      "step": 487
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.06269916892051697,
+      "learning_rate": 0.00018686342008757383,
+      "loss": 0.6564,
+      "step": 488
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.06308487802743912,
+      "learning_rate": 0.00018680964052739273,
+      "loss": 0.7436,
+      "step": 489
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06728991121053696,
+      "learning_rate": 0.0001867557588801579,
+      "loss": 0.7415,
+      "step": 490
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06449094414710999,
+      "learning_rate": 0.0001867017752092334,
+      "loss": 0.7578,
+      "step": 491
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.08734392374753952,
+      "learning_rate": 0.0001866476895781034,
+      "loss": 0.7041,
+      "step": 492
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06480298936367035,
+      "learning_rate": 0.0001865935020503719,
+      "loss": 0.7842,
+      "step": 493
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.09008181840181351,
+      "learning_rate": 0.00018653921268976285,
+      "loss": 0.6765,
+      "step": 494
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06579532474279404,
+      "learning_rate": 0.00018648482156011977,
+      "loss": 0.6687,
+      "step": 495
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06645182520151138,
+      "learning_rate": 0.000186430328725406,
+      "loss": 0.7584,
+      "step": 496
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06692402064800262,
+      "learning_rate": 0.00018637573424970435,
+      "loss": 0.7336,
+      "step": 497
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06754589080810547,
+      "learning_rate": 0.00018632103819721731,
+      "loss": 0.7692,
+      "step": 498
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.06759774684906006,
+      "learning_rate": 0.00018626624063226675,
+      "loss": 0.792,
+      "step": 499
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.06528620421886444,
+      "learning_rate": 0.00018621134161929393,
+      "loss": 0.76,
+      "step": 500
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.07810235023498535,
+      "learning_rate": 0.00018615634122285936,
+      "loss": 0.7695,
+      "step": 501
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.06633520126342773,
+      "learning_rate": 0.00018610123950764287,
+      "loss": 0.711,
+      "step": 502
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.0723094642162323,
+      "learning_rate": 0.00018604603653844337,
+      "loss": 0.7693,
+      "step": 503
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.06438109278678894,
+      "learning_rate": 0.00018599073238017888,
+      "loss": 0.6729,
+      "step": 504
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.062286246567964554,
+      "learning_rate": 0.00018593532709788635,
+      "loss": 0.6951,
+      "step": 505
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.06298733502626419,
+      "learning_rate": 0.00018587982075672174,
+      "loss": 0.6774,
+      "step": 506
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.06751766055822372,
+      "learning_rate": 0.00018582421342195978,
+      "loss": 0.7711,
+      "step": 507
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.06741178035736084,
+      "learning_rate": 0.0001857685051589941,
+      "loss": 0.706,
+      "step": 508
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.09620961546897888,
+      "learning_rate": 0.00018571269603333677,
+      "loss": 0.6802,
+      "step": 509
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.07265791296958923,
+      "learning_rate": 0.00018565678611061877,
+      "loss": 0.7392,
+      "step": 510
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.06844795495271683,
+      "learning_rate": 0.0001856007754565894,
+      "loss": 0.7633,
+      "step": 511
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.07247466593980789,
+      "learning_rate": 0.00018554466413711645,
+      "loss": 0.6176,
+      "step": 512
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.06879562884569168,
+      "learning_rate": 0.0001854884522181862,
+      "loss": 0.7368,
+      "step": 513
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.06788138300180435,
+      "learning_rate": 0.0001854321397659031,
+      "loss": 0.7507,
+      "step": 514
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.06438209116458893,
+      "learning_rate": 0.00018537572684649,
+      "loss": 0.7451,
+      "step": 515
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.06962509453296661,
+      "learning_rate": 0.00018531921352628772,
+      "loss": 0.747,
+      "step": 516
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.07272413372993469,
+      "learning_rate": 0.00018526259987175522,
+      "loss": 0.6815,
+      "step": 517
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.07266836613416672,
+      "learning_rate": 0.0001852058859494694,
+      "loss": 0.7337,
+      "step": 518
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.06824107468128204,
+      "learning_rate": 0.00018514907182612525,
+      "loss": 0.7137,
+      "step": 519
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.0668516680598259,
+      "learning_rate": 0.00018509215756853535,
+      "loss": 0.7391,
+      "step": 520
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.06273160874843597,
+      "learning_rate": 0.00018503514324363023,
+      "loss": 0.75,
+      "step": 521
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.06964839994907379,
+      "learning_rate": 0.000184978028918458,
+      "loss": 0.723,
+      "step": 522
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.06386502832174301,
+      "learning_rate": 0.00018492081466018435,
+      "loss": 0.7156,
+      "step": 523
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.06581712514162064,
+      "learning_rate": 0.00018486350053609262,
+      "loss": 0.7569,
+      "step": 524
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.07477328926324844,
+      "learning_rate": 0.00018480608661358344,
+      "loss": 0.8663,
+      "step": 525
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.07748843729496002,
+      "learning_rate": 0.0001847485729601749,
+      "loss": 0.5842,
+      "step": 526
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.06755229830741882,
+      "learning_rate": 0.00018469095964350233,
+      "loss": 0.7617,
+      "step": 527
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.0634811520576477,
+      "learning_rate": 0.00018463324673131823,
+      "loss": 0.7125,
+      "step": 528
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.06521796435117722,
+      "learning_rate": 0.00018457543429149234,
+      "loss": 0.7251,
+      "step": 529
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.07096633315086365,
+      "learning_rate": 0.0001845175223920113,
+      "loss": 0.6871,
+      "step": 530
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.066802479326725,
+      "learning_rate": 0.00018445951110097883,
+      "loss": 0.7262,
+      "step": 531
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.06430657207965851,
+      "learning_rate": 0.00018440140048661547,
+      "loss": 0.7712,
+      "step": 532
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.06193665415048599,
+      "learning_rate": 0.00018434319061725858,
+      "loss": 0.6698,
+      "step": 533
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.062316689640283585,
+      "learning_rate": 0.0001842848815613622,
+      "loss": 0.7413,
+      "step": 534
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.06545069813728333,
+      "learning_rate": 0.00018422647338749707,
+      "loss": 0.6857,
+      "step": 535
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.0639072060585022,
+      "learning_rate": 0.00018416796616435054,
+      "loss": 0.6938,
+      "step": 536
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.06502430140972137,
+      "learning_rate": 0.0001841093599607263,
+      "loss": 0.7396,
+      "step": 537
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.06591713428497314,
+      "learning_rate": 0.00018405065484554452,
+      "loss": 0.6659,
+      "step": 538
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.063933365046978,
+      "learning_rate": 0.0001839918508878417,
+      "loss": 0.7334,
+      "step": 539
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.06484158337116241,
+      "learning_rate": 0.00018393294815677056,
+      "loss": 0.7337,
+      "step": 540
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.066524937748909,
+      "learning_rate": 0.0001838739467216,
+      "loss": 0.7174,
+      "step": 541
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.0775236114859581,
+      "learning_rate": 0.0001838148466517149,
+      "loss": 0.7697,
+      "step": 542
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.06475924700498581,
+      "learning_rate": 0.0001837556480166163,
+      "loss": 0.7568,
+      "step": 543
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.06703387200832367,
+      "learning_rate": 0.000183696350885921,
+      "loss": 0.7124,
+      "step": 544
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.0641862079501152,
+      "learning_rate": 0.00018363695532936164,
+      "loss": 0.707,
+      "step": 545
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.06697754561901093,
+      "learning_rate": 0.00018357746141678676,
+      "loss": 0.7185,
+      "step": 546
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.06705842912197113,
+      "learning_rate": 0.00018351786921816038,
+      "loss": 0.7649,
+      "step": 547
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06511738151311874,
+      "learning_rate": 0.00018345817880356218,
+      "loss": 0.722,
+      "step": 548
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06530781090259552,
+      "learning_rate": 0.0001833983902431874,
+      "loss": 0.7143,
+      "step": 549
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06371057033538818,
+      "learning_rate": 0.00018333850360734662,
+      "loss": 0.6969,
+      "step": 550
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.07265166938304901,
+      "learning_rate": 0.00018327851896646574,
+      "loss": 0.7703,
+      "step": 551
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.07043513655662537,
+      "learning_rate": 0.000183218436391086,
+      "loss": 0.7416,
+      "step": 552
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06820254027843475,
+      "learning_rate": 0.0001831582559518637,
+      "loss": 0.7941,
+      "step": 553
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.0645156055688858,
+      "learning_rate": 0.00018309797771957036,
+      "loss": 0.7385,
+      "step": 554
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06690391153097153,
+      "learning_rate": 0.00018303760176509234,
+      "loss": 0.7509,
+      "step": 555
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06634744256734848,
+      "learning_rate": 0.00018297712815943109,
+      "loss": 0.6547,
+      "step": 556
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.06777124106884003,
+      "learning_rate": 0.00018291655697370276,
+      "loss": 0.7103,
+      "step": 557
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06453371793031693,
+      "learning_rate": 0.00018285588827913834,
+      "loss": 0.7032,
+      "step": 558
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06689512729644775,
+      "learning_rate": 0.00018279512214708344,
+      "loss": 0.7677,
+      "step": 559
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06096369028091431,
+      "learning_rate": 0.00018273425864899829,
+      "loss": 0.5788,
+      "step": 560
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06912747025489807,
+      "learning_rate": 0.0001826732978564576,
+      "loss": 0.7094,
+      "step": 561
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06365373730659485,
+      "learning_rate": 0.0001826122398411505,
+      "loss": 0.6554,
+      "step": 562
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06549493223428726,
+      "learning_rate": 0.00018255108467488045,
+      "loss": 0.6927,
+      "step": 563
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06063194200396538,
+      "learning_rate": 0.00018248983242956515,
+      "loss": 0.7077,
+      "step": 564
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06569315493106842,
+      "learning_rate": 0.00018242848317723647,
+      "loss": 0.704,
+      "step": 565
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.0646178349852562,
+      "learning_rate": 0.0001823670369900404,
+      "loss": 0.6658,
+      "step": 566
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.06432666629552841,
+      "learning_rate": 0.0001823054939402369,
+      "loss": 0.7117,
+      "step": 567
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06953758746385574,
+      "learning_rate": 0.00018224385410019976,
+      "loss": 0.657,
+      "step": 568
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.0662669911980629,
+      "learning_rate": 0.00018218211754241666,
+      "loss": 0.7467,
+      "step": 569
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06994238495826721,
+      "learning_rate": 0.00018212028433948915,
+      "loss": 0.771,
+      "step": 570
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06679581850767136,
+      "learning_rate": 0.0001820583545641322,
+      "loss": 0.7267,
+      "step": 571
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.07098355144262314,
+      "learning_rate": 0.00018199632828917446,
+      "loss": 0.7201,
+      "step": 572
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06492327153682709,
+      "learning_rate": 0.00018193420558755807,
+      "loss": 0.7125,
+      "step": 573
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06887879967689514,
+      "learning_rate": 0.00018187198653233855,
+      "loss": 0.6887,
+      "step": 574
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06757469475269318,
+      "learning_rate": 0.00018180967119668472,
+      "loss": 0.6571,
+      "step": 575
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06290429830551147,
+      "learning_rate": 0.00018174725965387867,
+      "loss": 0.702,
+      "step": 576
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.06352391839027405,
+      "learning_rate": 0.00018168475197731553,
+      "loss": 0.7017,
+      "step": 577
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.06685255467891693,
+      "learning_rate": 0.00018162214824050366,
+      "loss": 0.7533,
+      "step": 578
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.06117778643965721,
+      "learning_rate": 0.00018155944851706415,
+      "loss": 0.72,
+      "step": 579
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.0662195160984993,
+      "learning_rate": 0.00018149665288073115,
+      "loss": 0.7206,
+      "step": 580
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.0689619854092598,
+      "learning_rate": 0.00018143376140535158,
+      "loss": 0.6461,
+      "step": 581
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.06290092319250107,
+      "learning_rate": 0.00018137077416488496,
+      "loss": 0.7368,
+      "step": 582
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.06402301788330078,
+      "learning_rate": 0.00018130769123340351,
+      "loss": 0.6917,
+      "step": 583
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.07769487798213959,
+      "learning_rate": 0.00018124451268509203,
+      "loss": 0.7289,
+      "step": 584
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.06879512220621109,
+      "learning_rate": 0.00018118123859424764,
+      "loss": 0.7492,
+      "step": 585
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.07164003700017929,
+      "learning_rate": 0.00018111786903527994,
+      "loss": 0.6481,
+      "step": 586
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06826422363519669,
+      "learning_rate": 0.00018105440408271067,
+      "loss": 0.7569,
+      "step": 587
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.07842717319726944,
+      "learning_rate": 0.00018099084381117386,
+      "loss": 0.73,
+      "step": 588
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06824488192796707,
+      "learning_rate": 0.0001809271882954156,
+      "loss": 0.6665,
+      "step": 589
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06883325427770615,
+      "learning_rate": 0.000180863437610294,
+      "loss": 0.6761,
+      "step": 590
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06425601989030838,
+      "learning_rate": 0.000180799591830779,
+      "loss": 0.6317,
+      "step": 591
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06907620280981064,
+      "learning_rate": 0.00018073565103195253,
+      "loss": 0.6956,
+      "step": 592
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06852835416793823,
+      "learning_rate": 0.0001806716152890081,
+      "loss": 0.7079,
+      "step": 593
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.07164844870567322,
+      "learning_rate": 0.00018060748467725097,
+      "loss": 0.7305,
+      "step": 594
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06715549528598785,
+      "learning_rate": 0.00018054325927209795,
+      "loss": 0.6846,
+      "step": 595
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.06505031883716583,
+      "learning_rate": 0.00018047893914907733,
+      "loss": 0.7758,
+      "step": 596
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06611162424087524,
+      "learning_rate": 0.0001804145243838287,
+      "loss": 0.7493,
+      "step": 597
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06780173629522324,
+      "learning_rate": 0.00018035001505210308,
+      "loss": 0.7347,
+      "step": 598
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06760706007480621,
+      "learning_rate": 0.00018028541122976264,
+      "loss": 0.7149,
+      "step": 599
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06549396365880966,
+      "learning_rate": 0.00018022071299278065,
+      "loss": 0.7057,
+      "step": 600
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06889012455940247,
+      "learning_rate": 0.00018015592041724143,
+      "loss": 0.7686,
+      "step": 601
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06646202504634857,
+      "learning_rate": 0.00018009103357934024,
+      "loss": 0.7379,
+      "step": 602
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06753333657979965,
+      "learning_rate": 0.00018002605255538324,
+      "loss": 0.6518,
+      "step": 603
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06427349895238876,
+      "learning_rate": 0.0001799609774217872,
+      "loss": 0.6816,
+      "step": 604
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.0671876072883606,
+      "learning_rate": 0.00017989580825507974,
+      "loss": 0.6166,
+      "step": 605
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.06923870742321014,
+      "learning_rate": 0.00017983054513189898,
+      "loss": 0.7363,
+      "step": 606
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.06566382944583893,
+      "learning_rate": 0.00017976518812899353,
+      "loss": 0.6905,
+      "step": 607
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.06720302253961563,
+      "learning_rate": 0.0001796997373232224,
+      "loss": 0.7466,
+      "step": 608
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.07467524707317352,
+      "learning_rate": 0.00017963419279155493,
+      "loss": 0.6771,
+      "step": 609
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.06185929477214813,
+      "learning_rate": 0.00017956855461107068,
+      "loss": 0.7175,
+      "step": 610
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.08537810295820236,
+      "learning_rate": 0.0001795028228589593,
+      "loss": 0.7694,
+      "step": 611
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.06563472747802734,
+      "learning_rate": 0.00017943699761252055,
+      "loss": 0.7425,
+      "step": 612
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.06966511905193329,
+      "learning_rate": 0.00017937107894916407,
+      "loss": 0.8045,
+      "step": 613
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.0687708631157875,
+      "learning_rate": 0.0001793050669464094,
+      "loss": 0.7076,
+      "step": 614
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.06829386204481125,
+      "learning_rate": 0.00017923896168188584,
+      "loss": 0.6838,
+      "step": 615
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.06806325167417526,
+      "learning_rate": 0.0001791727632333323,
+      "loss": 0.6726,
+      "step": 616
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.06878352910280228,
+      "learning_rate": 0.00017910647167859741,
+      "loss": 0.7273,
+      "step": 617
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.07277524471282959,
+      "learning_rate": 0.00017904008709563915,
+      "loss": 0.6682,
+      "step": 618
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.06648612022399902,
+      "learning_rate": 0.000178973609562525,
+      "loss": 0.7046,
+      "step": 619
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.06936463713645935,
+      "learning_rate": 0.00017890703915743168,
+      "loss": 0.8204,
+      "step": 620
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.07113652676343918,
+      "learning_rate": 0.00017884037595864517,
+      "loss": 0.7009,
+      "step": 621
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.07027497887611389,
+      "learning_rate": 0.00017877362004456058,
+      "loss": 0.7249,
+      "step": 622
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.0664186030626297,
+      "learning_rate": 0.000178706771493682,
+      "loss": 0.7165,
+      "step": 623
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.06796207278966904,
+      "learning_rate": 0.0001786398303846225,
+      "loss": 0.7544,
+      "step": 624
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.06871851533651352,
+      "learning_rate": 0.00017857279679610397,
+      "loss": 0.7431,
+      "step": 625
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.07132145762443542,
+      "learning_rate": 0.0001785056708069571,
+      "loss": 0.7186,
+      "step": 626
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.0689890906214714,
+      "learning_rate": 0.00017843845249612122,
+      "loss": 0.7305,
+      "step": 627
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.06329981237649918,
+      "learning_rate": 0.0001783711419426442,
+      "loss": 0.6951,
+      "step": 628
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.06915485858917236,
+      "learning_rate": 0.00017830373922568245,
+      "loss": 0.6136,
+      "step": 629
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.07033320516347885,
+      "learning_rate": 0.00017823624442450065,
+      "loss": 0.7022,
+      "step": 630
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.07259360700845718,
+      "learning_rate": 0.00017816865761847197,
+      "loss": 0.7394,
+      "step": 631
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.06609155982732773,
+      "learning_rate": 0.0001781009788870775,
+      "loss": 0.6955,
+      "step": 632
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.07294370979070663,
+      "learning_rate": 0.00017803320830990667,
+      "loss": 0.6823,
+      "step": 633
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.0686420202255249,
+      "learning_rate": 0.00017796534596665686,
+      "loss": 0.6778,
+      "step": 634
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.06810775399208069,
+      "learning_rate": 0.00017789739193713324,
+      "loss": 0.7313,
+      "step": 635
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.06998413801193237,
+      "learning_rate": 0.000177829346301249,
+      "loss": 0.6917,
+      "step": 636
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.08018749952316284,
+      "learning_rate": 0.00017776120913902489,
+      "loss": 0.7806,
+      "step": 637
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.07463950663805008,
+      "learning_rate": 0.0001776929805305894,
+      "loss": 0.7387,
+      "step": 638
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.06867985427379608,
+      "learning_rate": 0.0001776246605561785,
+      "loss": 0.7435,
+      "step": 639
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.07185962051153183,
+      "learning_rate": 0.00017755624929613565,
+      "loss": 0.7688,
+      "step": 640
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.0686650425195694,
+      "learning_rate": 0.00017748774683091164,
+      "loss": 0.7115,
+      "step": 641
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.06798288226127625,
+      "learning_rate": 0.00017741915324106445,
+      "loss": 0.7755,
+      "step": 642
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.07337713241577148,
+      "learning_rate": 0.0001773504686072594,
+      "loss": 0.7563,
+      "step": 643
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.07286756485700607,
+      "learning_rate": 0.00017728169301026864,
+      "loss": 0.7242,
+      "step": 644
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.0704449713230133,
+      "learning_rate": 0.00017721282653097148,
+      "loss": 0.7912,
+      "step": 645
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.06909286230802536,
+      "learning_rate": 0.00017714386925035406,
+      "loss": 0.6845,
+      "step": 646
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.07194187492132187,
+      "learning_rate": 0.00017707482124950923,
+      "loss": 0.7388,
+      "step": 647
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.09377000480890274,
+      "learning_rate": 0.00017700568260963658,
+      "loss": 0.7118,
+      "step": 648
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.06767406314611435,
+      "learning_rate": 0.00017693645341204236,
+      "loss": 0.6796,
+      "step": 649
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.07329147309064865,
+      "learning_rate": 0.00017686713373813917,
+      "loss": 0.635,
+      "step": 650
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.06956500560045242,
+      "learning_rate": 0.00017679772366944609,
+      "loss": 0.7316,
+      "step": 651
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.07110968232154846,
+      "learning_rate": 0.00017672822328758853,
+      "loss": 0.7194,
+      "step": 652
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.06750847399234772,
+      "learning_rate": 0.00017665863267429802,
+      "loss": 0.6878,
+      "step": 653
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.0709410011768341,
+      "learning_rate": 0.00017658895191141225,
+      "loss": 0.7374,
+      "step": 654
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.06910226494073868,
+      "learning_rate": 0.00017651918108087502,
+      "loss": 0.7234,
+      "step": 655
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.07859506458044052,
+      "learning_rate": 0.00017644932026473584,
+      "loss": 0.7187,
+      "step": 656
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.06830095499753952,
+      "learning_rate": 0.00017637936954515023,
+      "loss": 0.6305,
+      "step": 657
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.07022988796234131,
+      "learning_rate": 0.00017630932900437936,
+      "loss": 0.6217,
+      "step": 658
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.06744571030139923,
+      "learning_rate": 0.00017623919872479,
+      "loss": 0.6997,
+      "step": 659
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.0686475932598114,
+      "learning_rate": 0.00017616897878885453,
+      "loss": 0.6887,
+      "step": 660
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.06921619921922684,
+      "learning_rate": 0.00017609866927915065,
+      "loss": 0.695,
+      "step": 661
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.07123583555221558,
+      "learning_rate": 0.00017602827027836152,
+      "loss": 0.7236,
+      "step": 662
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.0664835125207901,
+      "learning_rate": 0.00017595778186927546,
+      "loss": 0.692,
+      "step": 663
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.06954433768987656,
+      "learning_rate": 0.00017588720413478596,
+      "loss": 0.68,
+      "step": 664
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.06957016885280609,
+      "learning_rate": 0.00017581653715789157,
+      "loss": 0.7483,
+      "step": 665
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07063329964876175,
+      "learning_rate": 0.00017574578102169577,
+      "loss": 0.7072,
+      "step": 666
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07582032680511475,
+      "learning_rate": 0.0001756749358094069,
+      "loss": 0.7581,
+      "step": 667
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.06948222219944,
+      "learning_rate": 0.00017560400160433802,
+      "loss": 0.7368,
+      "step": 668
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07247303426265717,
+      "learning_rate": 0.00017553297848990688,
+      "loss": 0.743,
+      "step": 669
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07067783176898956,
+      "learning_rate": 0.00017546186654963578,
+      "loss": 0.7624,
+      "step": 670
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07306205481290817,
+      "learning_rate": 0.0001753906658671515,
+      "loss": 0.7256,
+      "step": 671
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.06867087632417679,
+      "learning_rate": 0.00017531937652618513,
+      "loss": 0.7204,
+      "step": 672
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.07339783757925034,
+      "learning_rate": 0.000175247998610572,
+      "loss": 0.752,
+      "step": 673
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.0657355785369873,
+      "learning_rate": 0.00017517653220425173,
+      "loss": 0.7191,
+      "step": 674
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.067570261657238,
+      "learning_rate": 0.00017510497739126786,
+      "loss": 0.7149,
+      "step": 675
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.0657983049750328,
+      "learning_rate": 0.000175033334255768,
+      "loss": 0.6925,
+      "step": 676
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.07179775089025497,
+      "learning_rate": 0.00017496160288200358,
+      "loss": 0.7144,
+      "step": 677
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.06727674603462219,
+      "learning_rate": 0.00017488978335432974,
+      "loss": 0.7607,
+      "step": 678
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.0738164633512497,
+      "learning_rate": 0.00017481787575720543,
+      "loss": 0.7518,
+      "step": 679
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.07027100026607513,
+      "learning_rate": 0.00017474588017519303,
+      "loss": 0.8125,
+      "step": 680
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.06803075224161148,
+      "learning_rate": 0.00017467379669295846,
+      "loss": 0.6404,
+      "step": 681
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.0729600042104721,
+      "learning_rate": 0.00017460162539527103,
+      "loss": 0.7422,
+      "step": 682
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.06913101673126221,
+      "learning_rate": 0.0001745293663670032,
+      "loss": 0.695,
+      "step": 683
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.064353346824646,
+      "learning_rate": 0.0001744570196931308,
+      "loss": 0.7335,
+      "step": 684
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.06986693292856216,
+      "learning_rate": 0.00017438458545873252,
+      "loss": 0.7419,
+      "step": 685
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.07011342793703079,
+      "learning_rate": 0.00017431206374899017,
+      "loss": 0.6893,
+      "step": 686
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.06858572363853455,
+      "learning_rate": 0.00017423945464918833,
+      "loss": 0.6907,
+      "step": 687
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.08274766802787781,
+      "learning_rate": 0.00017416675824471448,
+      "loss": 0.7663,
+      "step": 688
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.07006780803203583,
+      "learning_rate": 0.00017409397462105858,
+      "loss": 0.7254,
+      "step": 689
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.07336114346981049,
+      "learning_rate": 0.0001740211038638133,
+      "loss": 0.6293,
+      "step": 690
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.07286055386066437,
+      "learning_rate": 0.00017394814605867378,
+      "loss": 0.6905,
+      "step": 691
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.06696517765522003,
+      "learning_rate": 0.00017387510129143746,
+      "loss": 0.7297,
+      "step": 692
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.06974443048238754,
+      "learning_rate": 0.00017380196964800405,
+      "loss": 0.7166,
+      "step": 693
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.06771736592054367,
+      "learning_rate": 0.0001737287512143755,
+      "loss": 0.7195,
+      "step": 694
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.06780609488487244,
+      "learning_rate": 0.00017365544607665575,
+      "loss": 0.6517,
+      "step": 695
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.06737010926008224,
+      "learning_rate": 0.0001735820543210507,
+      "loss": 0.6878,
+      "step": 696
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.07415246218442917,
+      "learning_rate": 0.00017350857603386815,
+      "loss": 0.7592,
+      "step": 697
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.06827431917190552,
+      "learning_rate": 0.0001734350113015177,
+      "loss": 0.7171,
+      "step": 698
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.07075396925210953,
+      "learning_rate": 0.0001733613602105105,
+      "loss": 0.7731,
+      "step": 699
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.0783689022064209,
+      "learning_rate": 0.00017328762284745937,
+      "loss": 0.6919,
+      "step": 700
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.06844034790992737,
+      "learning_rate": 0.00017321379929907845,
+      "loss": 0.6868,
+      "step": 701
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.07113159447908401,
+      "learning_rate": 0.00017313988965218335,
+      "loss": 0.7393,
+      "step": 702
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.06837478280067444,
+      "learning_rate": 0.00017306589399369092,
+      "loss": 0.7923,
+      "step": 703
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.06957865506410599,
+      "learning_rate": 0.00017299181241061908,
+      "loss": 0.6984,
+      "step": 704
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.07044649869203568,
+      "learning_rate": 0.0001729176449900869,
+      "loss": 0.6994,
+      "step": 705
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.06768269091844559,
+      "learning_rate": 0.00017284339181931427,
+      "loss": 0.6794,
+      "step": 706
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.07598036527633667,
+      "learning_rate": 0.00017276905298562207,
+      "loss": 0.6755,
+      "step": 707
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.07155588269233704,
+      "learning_rate": 0.00017269462857643183,
+      "loss": 0.7878,
+      "step": 708
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.07248366624116898,
+      "learning_rate": 0.00017262011867926567,
+      "loss": 0.6732,
+      "step": 709
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.07186141610145569,
+      "learning_rate": 0.00017254552338174633,
+      "loss": 0.6999,
+      "step": 710
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.07100434601306915,
+      "learning_rate": 0.00017247084277159696,
+      "loss": 0.7418,
+      "step": 711
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.06756517291069031,
+      "learning_rate": 0.000172396076936641,
+      "loss": 0.6875,
+      "step": 712
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.06727565079927444,
+      "learning_rate": 0.00017232122596480217,
+      "loss": 0.7311,
+      "step": 713
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.06914009153842926,
+      "learning_rate": 0.00017224628994410427,
+      "loss": 0.7064,
+      "step": 714
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.06864607334136963,
+      "learning_rate": 0.0001721712689626711,
+      "loss": 0.7255,
+      "step": 715
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07154612243175507,
+      "learning_rate": 0.00017209616310872642,
+      "loss": 0.7724,
+      "step": 716
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07429131865501404,
+      "learning_rate": 0.00017202097247059382,
+      "loss": 0.6746,
+      "step": 717
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07139010727405548,
+      "learning_rate": 0.00017194569713669644,
+      "loss": 0.6725,
+      "step": 718
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07144646346569061,
+      "learning_rate": 0.00017187033719555722,
+      "loss": 0.7501,
+      "step": 719
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.06815102696418762,
+      "learning_rate": 0.0001717948927357985,
+      "loss": 0.6723,
+      "step": 720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.06780737638473511,
+      "learning_rate": 0.00017171936384614202,
+      "loss": 0.6887,
+      "step": 721
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.07604995369911194,
+      "learning_rate": 0.00017164375061540877,
+      "loss": 0.6663,
+      "step": 722
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.07561536878347397,
+      "learning_rate": 0.00017156805313251904,
+      "loss": 0.7449,
+      "step": 723
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.07025757431983948,
+      "learning_rate": 0.00017149227148649204,
+      "loss": 0.7432,
+      "step": 724
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.07664772123098373,
+      "learning_rate": 0.0001714164057664461,
+      "loss": 0.751,
+      "step": 725
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.07072637230157852,
+      "learning_rate": 0.00017134045606159837,
+      "loss": 0.7884,
+      "step": 726
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.06859525293111801,
+      "learning_rate": 0.0001712644224612647,
+      "loss": 0.7368,
+      "step": 727
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.06537380814552307,
+      "learning_rate": 0.00017118830505485967,
+      "loss": 0.716,
+      "step": 728
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.06966748833656311,
+      "learning_rate": 0.00017111210393189644,
+      "loss": 0.7058,
+      "step": 729
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.8200334310531616,
+      "eval_runtime": 111.8923,
+      "eval_samples_per_second": 88.818,
+      "eval_steps_per_second": 22.209,
+      "step": 729
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.06748418509960175,
+      "learning_rate": 0.00017103581918198656,
+      "loss": 0.6189,
+      "step": 730
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.0700392946600914,
+      "learning_rate": 0.0001709594508948399,
+      "loss": 0.671,
+      "step": 731
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.07225173711776733,
+      "learning_rate": 0.0001708829991602647,
+      "loss": 0.6932,
+      "step": 732
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.07260441780090332,
+      "learning_rate": 0.00017080646406816716,
+      "loss": 0.6667,
+      "step": 733
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.07659890502691269,
+      "learning_rate": 0.00017072984570855164,
+      "loss": 0.7625,
+      "step": 734
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.07129763811826706,
+      "learning_rate": 0.0001706531441715203,
+      "loss": 0.6938,
+      "step": 735
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.07404035329818726,
+      "learning_rate": 0.00017057635954727336,
+      "loss": 0.7061,
+      "step": 736
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.07175761461257935,
+      "learning_rate": 0.00017049949192610844,
+      "loss": 0.7886,
+      "step": 737
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.06855437159538269,
+      "learning_rate": 0.00017042254139842088,
+      "loss": 0.7088,
+      "step": 738
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.07093068957328796,
+      "learning_rate": 0.00017034550805470366,
+      "loss": 0.6564,
+      "step": 739
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.07080741226673126,
+      "learning_rate": 0.00017026839198554692,
+      "loss": 0.6722,
+      "step": 740
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.0733497142791748,
+      "learning_rate": 0.0001701911932816383,
+      "loss": 0.7528,
+      "step": 741
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07160144299268723,
+      "learning_rate": 0.0001701139120337624,
+      "loss": 0.6859,
+      "step": 742
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07589136064052582,
+      "learning_rate": 0.00017003654833280108,
+      "loss": 0.7875,
+      "step": 743
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07213631272315979,
+      "learning_rate": 0.000169959102269733,
+      "loss": 0.6945,
+      "step": 744
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07278650999069214,
+      "learning_rate": 0.00016988157393563392,
+      "loss": 0.7683,
+      "step": 745
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.0748310461640358,
+      "learning_rate": 0.00016980396342167607,
+      "loss": 0.7254,
+      "step": 746
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07376185059547424,
+      "learning_rate": 0.00016972627081912847,
+      "loss": 0.7203,
+      "step": 747
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07464928179979324,
+      "learning_rate": 0.00016964849621935665,
+      "loss": 0.7687,
+      "step": 748
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.06802686303853989,
+      "learning_rate": 0.00016957063971382257,
+      "loss": 0.9044,
+      "step": 749
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07590179145336151,
+      "learning_rate": 0.00016949270139408453,
+      "loss": 0.7479,
+      "step": 750
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.07608579099178314,
+      "learning_rate": 0.00016941468135179698,
+      "loss": 0.7025,
+      "step": 751
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.07457372546195984,
+      "learning_rate": 0.00016933657967871057,
+      "loss": 0.7219,
+      "step": 752
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.07248198240995407,
+      "learning_rate": 0.00016925839646667184,
+      "loss": 0.7408,
+      "step": 753
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.07194027304649353,
+      "learning_rate": 0.00016918013180762333,
+      "loss": 0.7088,
+      "step": 754
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.0747242271900177,
+      "learning_rate": 0.00016910178579360323,
+      "loss": 0.6723,
+      "step": 755
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.07224127650260925,
+      "learning_rate": 0.00016902335851674553,
+      "loss": 0.7307,
+      "step": 756
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.08432527631521225,
+      "learning_rate": 0.00016894485006927973,
+      "loss": 0.6041,
+      "step": 757
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.06535546481609344,
+      "learning_rate": 0.00016886626054353073,
+      "loss": 0.7007,
+      "step": 758
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.0744195505976677,
+      "learning_rate": 0.00016878759003191888,
+      "loss": 0.7543,
+      "step": 759
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.0689319595694542,
+      "learning_rate": 0.0001687088386269597,
+      "loss": 0.6784,
+      "step": 760
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.06821681559085846,
+      "learning_rate": 0.0001686300064212639,
+      "loss": 0.6312,
+      "step": 761
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.07716275751590729,
+      "learning_rate": 0.00016855109350753707,
+      "loss": 0.7033,
+      "step": 762
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.07511652261018753,
+      "learning_rate": 0.00016847209997857996,
+      "loss": 0.7273,
+      "step": 763
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.07603703439235687,
+      "learning_rate": 0.00016839302592728782,
+      "loss": 0.7864,
+      "step": 764
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.0709453672170639,
+      "learning_rate": 0.00016831387144665088,
+      "loss": 0.663,
+      "step": 765
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.07530734688043594,
+      "learning_rate": 0.0001682346366297537,
+      "loss": 0.6986,
+      "step": 766
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.09247121959924698,
+      "learning_rate": 0.00016815532156977555,
+      "loss": 0.687,
+      "step": 767
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.06963692605495453,
+      "learning_rate": 0.00016807592635998986,
+      "loss": 0.6596,
+      "step": 768
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.0712541937828064,
+      "learning_rate": 0.00016799645109376446,
+      "loss": 0.7263,
+      "step": 769
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.07108648121356964,
+      "learning_rate": 0.00016791689586456124,
+      "loss": 0.7563,
+      "step": 770
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07317429780960083,
+      "learning_rate": 0.0001678372607659362,
+      "loss": 0.6971,
+      "step": 771
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07137763500213623,
+      "learning_rate": 0.00016775754589153913,
+      "loss": 0.6823,
+      "step": 772
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07120873034000397,
+      "learning_rate": 0.00016767775133511384,
+      "loss": 0.6653,
+      "step": 773
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07151015102863312,
+      "learning_rate": 0.00016759787719049766,
+      "loss": 0.655,
+      "step": 774
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07391924411058426,
+      "learning_rate": 0.00016751792355162163,
+      "loss": 0.7302,
+      "step": 775
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07449232786893845,
+      "learning_rate": 0.0001674378905125102,
+      "loss": 0.734,
+      "step": 776
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.06854017078876495,
+      "learning_rate": 0.0001673577781672812,
+      "loss": 0.7211,
+      "step": 777
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07284322381019592,
+      "learning_rate": 0.00016727758661014587,
+      "loss": 0.6797,
+      "step": 778
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07225552201271057,
+      "learning_rate": 0.00016719731593540832,
+      "loss": 0.6995,
+      "step": 779
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.07095993310213089,
+      "learning_rate": 0.00016711696623746596,
+      "loss": 0.7113,
+      "step": 780
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07299510389566422,
+      "learning_rate": 0.000167036537610809,
+      "loss": 0.6578,
+      "step": 781
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07212400436401367,
+      "learning_rate": 0.0001669560301500205,
+      "loss": 0.6849,
+      "step": 782
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07263436913490295,
+      "learning_rate": 0.00016687544394977625,
+      "loss": 0.7493,
+      "step": 783
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.06992338597774506,
+      "learning_rate": 0.00016679477910484463,
+      "loss": 0.7447,
+      "step": 784
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07319390028715134,
+      "learning_rate": 0.00016671403571008643,
+      "loss": 0.6864,
+      "step": 785
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07034803926944733,
+      "learning_rate": 0.0001666332138604549,
+      "loss": 0.6968,
+      "step": 786
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.06971286237239838,
+      "learning_rate": 0.00016655231365099557,
+      "loss": 0.642,
+      "step": 787
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07530028373003006,
+      "learning_rate": 0.000166471335176846,
+      "loss": 0.64,
+      "step": 788
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.06948436051607132,
+      "learning_rate": 0.00016639027853323596,
+      "loss": 0.7564,
+      "step": 789
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.07382038235664368,
+      "learning_rate": 0.00016630914381548695,
+      "loss": 0.8079,
+      "step": 790
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.08063971996307373,
+      "learning_rate": 0.00016622793111901245,
+      "loss": 0.6679,
+      "step": 791
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.065667524933815,
+      "learning_rate": 0.00016614664053931757,
+      "loss": 0.6448,
+      "step": 792
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.07058931887149811,
+      "learning_rate": 0.00016606527217199899,
+      "loss": 0.6972,
+      "step": 793
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.06650307774543762,
+      "learning_rate": 0.00016598382611274492,
+      "loss": 0.6352,
+      "step": 794
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.0777771845459938,
+      "learning_rate": 0.0001659023024573349,
+      "loss": 0.6967,
+      "step": 795
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.07293581962585449,
+      "learning_rate": 0.00016582070130163973,
+      "loss": 0.7427,
+      "step": 796
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.0743393748998642,
+      "learning_rate": 0.00016573902274162134,
+      "loss": 0.7339,
+      "step": 797
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.07685486972332001,
+      "learning_rate": 0.00016565726687333275,
+      "loss": 0.7306,
+      "step": 798
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.07364094257354736,
+      "learning_rate": 0.00016557543379291776,
+      "loss": 0.7842,
+      "step": 799
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.07045711576938629,
+      "learning_rate": 0.00016549352359661112,
+      "loss": 0.6929,
+      "step": 800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07396701723337173,
+      "learning_rate": 0.00016541153638073816,
+      "loss": 0.7083,
+      "step": 801
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07448946684598923,
+      "learning_rate": 0.0001653294722417148,
+      "loss": 0.6738,
+      "step": 802
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07139252871274948,
+      "learning_rate": 0.00016524733127604754,
+      "loss": 0.7071,
+      "step": 803
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07060480862855911,
+      "learning_rate": 0.000165165113580333,
+      "loss": 0.6935,
+      "step": 804
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.0778721421957016,
+      "learning_rate": 0.0001650828192512583,
+      "loss": 0.7127,
+      "step": 805
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07079839706420898,
+      "learning_rate": 0.00016500044838560043,
+      "loss": 0.7437,
+      "step": 806
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07176418602466583,
+      "learning_rate": 0.00016491800108022657,
+      "loss": 0.6958,
+      "step": 807
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07123876363039017,
+      "learning_rate": 0.00016483547743209367,
+      "loss": 0.6034,
+      "step": 808
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.07393837720155716,
+      "learning_rate": 0.00016475287753824853,
+      "loss": 0.7054,
+      "step": 809
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.07148581743240356,
+      "learning_rate": 0.00016467020149582763,
+      "loss": 0.7549,
+      "step": 810
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.07241423428058624,
+      "learning_rate": 0.0001645874494020569,
+      "loss": 0.6807,
+      "step": 811
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.067020945250988,
+      "learning_rate": 0.00016450462135425187,
+      "loss": 0.6689,
+      "step": 812
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.07124791294336319,
+      "learning_rate": 0.00016442171744981713,
+      "loss": 0.7972,
+      "step": 813
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.06952305883169174,
+      "learning_rate": 0.00016433873778624682,
+      "loss": 0.6847,
+      "step": 814
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.07224664092063904,
+      "learning_rate": 0.00016425568246112383,
+      "loss": 0.7467,
+      "step": 815
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.07566121965646744,
+      "learning_rate": 0.00016417255157212033,
+      "loss": 0.6625,
+      "step": 816
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.07257838547229767,
+      "learning_rate": 0.00016408934521699707,
+      "loss": 0.6801,
+      "step": 817
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.06923528760671616,
+      "learning_rate": 0.00016400606349360375,
+      "loss": 0.6777,
+      "step": 818
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.07393685728311539,
+      "learning_rate": 0.0001639227064998787,
+      "loss": 0.654,
+      "step": 819
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.07075174897909164,
+      "learning_rate": 0.00016383927433384857,
+      "loss": 0.7084,
+      "step": 820
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.07403064519166946,
+      "learning_rate": 0.00016375576709362868,
+      "loss": 0.6926,
+      "step": 821
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.07139717042446136,
+      "learning_rate": 0.00016367218487742239,
+      "loss": 0.7008,
+      "step": 822
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.07197099924087524,
+      "learning_rate": 0.00016358852778352142,
+      "loss": 0.758,
+      "step": 823
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.07550374418497086,
+      "learning_rate": 0.0001635047959103054,
+      "loss": 0.7896,
+      "step": 824
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.09666828066110611,
+      "learning_rate": 0.00016342098935624204,
+      "loss": 0.6539,
+      "step": 825
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.06977133452892303,
+      "learning_rate": 0.00016333710821988677,
+      "loss": 0.7262,
+      "step": 826
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.06979740411043167,
+      "learning_rate": 0.00016325315259988274,
+      "loss": 0.7076,
+      "step": 827
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.06903263181447983,
+      "learning_rate": 0.00016316912259496075,
+      "loss": 0.7015,
+      "step": 828
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.07616313546895981,
+      "learning_rate": 0.000163085018303939,
+      "loss": 0.7447,
+      "step": 829
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.07552159577608109,
+      "learning_rate": 0.00016300083982572313,
+      "loss": 0.7927,
+      "step": 830
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.07039552181959152,
+      "learning_rate": 0.00016291658725930592,
+      "loss": 0.7102,
+      "step": 831
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.0684932991862297,
+      "learning_rate": 0.00016283226070376737,
+      "loss": 0.6915,
+      "step": 832
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.07504136115312576,
+      "learning_rate": 0.00016274786025827445,
+      "loss": 0.7161,
+      "step": 833
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.07205895334482193,
+      "learning_rate": 0.0001626633860220811,
+      "loss": 0.751,
+      "step": 834
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.07040917873382568,
+      "learning_rate": 0.00016257883809452786,
+      "loss": 0.6841,
+      "step": 835
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.07174846529960632,
+      "learning_rate": 0.00016249421657504209,
+      "loss": 0.6553,
+      "step": 836
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.08638732880353928,
+      "learning_rate": 0.0001624095215631376,
+      "loss": 0.6641,
+      "step": 837
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.06990225613117218,
+      "learning_rate": 0.00016232475315841474,
+      "loss": 0.674,
+      "step": 838
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.0807608887553215,
+      "learning_rate": 0.00016223991146056002,
+      "loss": 0.7398,
+      "step": 839
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.0735122337937355,
+      "learning_rate": 0.0001621549965693463,
+      "loss": 0.7068,
+      "step": 840
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.07527075707912445,
+      "learning_rate": 0.00016207000858463238,
+      "loss": 0.7135,
+      "step": 841
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.07502257823944092,
+      "learning_rate": 0.00016198494760636303,
+      "loss": 0.8198,
+      "step": 842
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.0750960186123848,
+      "learning_rate": 0.00016189981373456897,
+      "loss": 0.6907,
+      "step": 843
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.06994899362325668,
+      "learning_rate": 0.00016181460706936654,
+      "loss": 0.7043,
+      "step": 844
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.0717443972826004,
+      "learning_rate": 0.00016172932771095773,
+      "loss": 0.8281,
+      "step": 845
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.06987167149782181,
+      "learning_rate": 0.00016164397575962997,
+      "loss": 0.7121,
+      "step": 846
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.0782201811671257,
+      "learning_rate": 0.00016155855131575614,
+      "loss": 0.7374,
+      "step": 847
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.07733765989542007,
+      "learning_rate": 0.00016147305447979427,
+      "loss": 0.7465,
+      "step": 848
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.07451386749744415,
+      "learning_rate": 0.00016138748535228758,
+      "loss": 0.6913,
+      "step": 849
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.07971774786710739,
+      "learning_rate": 0.00016130184403386432,
+      "loss": 0.6886,
+      "step": 850
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.07056500762701035,
+      "learning_rate": 0.0001612161306252376,
+      "loss": 0.6443,
+      "step": 851
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.07071709632873535,
+      "learning_rate": 0.00016113034522720533,
+      "loss": 0.6901,
+      "step": 852
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.0736258253455162,
+      "learning_rate": 0.00016104448794065004,
+      "loss": 0.7571,
+      "step": 853
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.08299355953931808,
+      "learning_rate": 0.00016095855886653885,
+      "loss": 0.6946,
+      "step": 854
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.0706866979598999,
+      "learning_rate": 0.00016087255810592327,
+      "loss": 0.7256,
+      "step": 855
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.07274890691041946,
+      "learning_rate": 0.0001607864857599391,
+      "loss": 0.7055,
+      "step": 856
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.07094918936491013,
+      "learning_rate": 0.00016070034192980638,
+      "loss": 0.6784,
+      "step": 857
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.0819023996591568,
+      "learning_rate": 0.00016061412671682917,
+      "loss": 0.6779,
+      "step": 858
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.07570182532072067,
+      "learning_rate": 0.00016052784022239547,
+      "loss": 0.7238,
+      "step": 859
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.07302212715148926,
+      "learning_rate": 0.00016044148254797715,
+      "loss": 0.6488,
+      "step": 860
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.06959427148103714,
+      "learning_rate": 0.00016035505379512975,
+      "loss": 0.683,
+      "step": 861
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.07809410244226456,
+      "learning_rate": 0.00016026855406549238,
+      "loss": 0.6587,
+      "step": 862
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.08332572132349014,
+      "learning_rate": 0.00016018198346078762,
+      "loss": 0.7443,
+      "step": 863
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.07175584137439728,
+      "learning_rate": 0.00016009534208282148,
+      "loss": 0.6607,
+      "step": 864
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.07386159151792526,
+      "learning_rate": 0.00016000863003348315,
+      "loss": 0.7377,
+      "step": 865
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.07763604819774628,
+      "learning_rate": 0.0001599218474147448,
+      "loss": 0.7269,
+      "step": 866
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.08228732645511627,
+      "learning_rate": 0.00015983499432866186,
+      "loss": 0.7337,
+      "step": 867
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.0716525986790657,
+      "learning_rate": 0.0001597480708773724,
+      "loss": 0.6779,
+      "step": 868
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07602430880069733,
+      "learning_rate": 0.00015966107716309727,
+      "loss": 0.6689,
+      "step": 869
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07297798246145248,
+      "learning_rate": 0.00015957401328814007,
+      "loss": 0.7421,
+      "step": 870
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07451054453849792,
+      "learning_rate": 0.0001594868793548868,
+      "loss": 0.7841,
+      "step": 871
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.0786144807934761,
+      "learning_rate": 0.0001593996754658059,
+      "loss": 0.6227,
+      "step": 872
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07301963865756989,
+      "learning_rate": 0.000159312401723448,
+      "loss": 0.7599,
+      "step": 873
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07685115933418274,
+      "learning_rate": 0.00015922505823044597,
+      "loss": 0.7269,
+      "step": 874
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.08059655129909515,
+      "learning_rate": 0.0001591376450895147,
+      "loss": 0.7869,
+      "step": 875
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.07008222490549088,
+      "learning_rate": 0.00015905016240345087,
+      "loss": 0.6547,
+      "step": 876
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.0795569196343422,
+      "learning_rate": 0.0001589626102751331,
+      "loss": 0.7661,
+      "step": 877
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.07241260260343552,
+      "learning_rate": 0.00015887498880752155,
+      "loss": 0.7443,
+      "step": 878
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.07290566712617874,
+      "learning_rate": 0.000158787298103658,
+      "loss": 0.7087,
+      "step": 879
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.07114351540803909,
+      "learning_rate": 0.0001586995382666656,
+      "loss": 0.7609,
+      "step": 880
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.07909577339887619,
+      "learning_rate": 0.0001586117093997489,
+      "loss": 0.7302,
+      "step": 881
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.06962461024522781,
+      "learning_rate": 0.00015852381160619343,
+      "loss": 0.7363,
+      "step": 882
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.07440977543592453,
+      "learning_rate": 0.000158435844989366,
+      "loss": 0.696,
+      "step": 883
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.0998484343290329,
+      "learning_rate": 0.0001583478096527142,
+      "loss": 0.7897,
+      "step": 884
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.08223816752433777,
+      "learning_rate": 0.0001582597056997665,
+      "loss": 0.6683,
+      "step": 885
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.07521655410528183,
+      "learning_rate": 0.00015817153323413206,
+      "loss": 0.7025,
+      "step": 886
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.07854367047548294,
+      "learning_rate": 0.0001580832923595006,
+      "loss": 0.6931,
+      "step": 887
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07404212653636932,
+      "learning_rate": 0.00015799498317964225,
+      "loss": 0.7398,
+      "step": 888
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07481147348880768,
+      "learning_rate": 0.00015790660579840753,
+      "loss": 0.6826,
+      "step": 889
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07364954799413681,
+      "learning_rate": 0.00015781816031972717,
+      "loss": 0.6833,
+      "step": 890
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07147948443889618,
+      "learning_rate": 0.00015772964684761186,
+      "loss": 0.6816,
+      "step": 891
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07281176000833511,
+      "learning_rate": 0.00015764106548615242,
+      "loss": 0.7211,
+      "step": 892
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07681546360254288,
+      "learning_rate": 0.0001575524163395194,
+      "loss": 0.6958,
+      "step": 893
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07273909449577332,
+      "learning_rate": 0.0001574636995119631,
+      "loss": 0.7166,
+      "step": 894
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.08109702169895172,
+      "learning_rate": 0.0001573749151078134,
+      "loss": 0.734,
+      "step": 895
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.07343237847089767,
+      "learning_rate": 0.00015728606323147965,
+      "loss": 0.6613,
+      "step": 896
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.06834150850772858,
+      "learning_rate": 0.0001571971439874505,
+      "loss": 0.6678,
+      "step": 897
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.07291509956121445,
+      "learning_rate": 0.00015710815748029396,
+      "loss": 0.682,
+      "step": 898
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.07433850318193436,
+      "learning_rate": 0.00015701910381465695,
+      "loss": 0.6787,
+      "step": 899
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.06953968852758408,
+      "learning_rate": 0.00015692998309526556,
+      "loss": 0.6631,
+      "step": 900
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.07962784171104431,
+      "learning_rate": 0.0001568407954269246,
+      "loss": 0.696,
+      "step": 901
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.07209240645170212,
+      "learning_rate": 0.00015675154091451764,
+      "loss": 0.7159,
+      "step": 902
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.0718846246600151,
+      "learning_rate": 0.00015666221966300693,
+      "loss": 0.7385,
+      "step": 903
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.07707607001066208,
+      "learning_rate": 0.00015657283177743307,
+      "loss": 0.7408,
+      "step": 904
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.0742068812251091,
+      "learning_rate": 0.00015648337736291515,
+      "loss": 0.7541,
+      "step": 905
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.07085248082876205,
+      "learning_rate": 0.00015639385652465047,
+      "loss": 0.6785,
+      "step": 906
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.07351714372634888,
+      "learning_rate": 0.00015630426936791433,
+      "loss": 0.7232,
+      "step": 907
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.0779218003153801,
+      "learning_rate": 0.0001562146159980602,
+      "loss": 0.7737,
+      "step": 908
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.07323803752660751,
+      "learning_rate": 0.00015612489652051928,
+      "loss": 0.6536,
+      "step": 909
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.08168342709541321,
+      "learning_rate": 0.00015603511104080056,
+      "loss": 0.7553,
+      "step": 910
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.08360809832811356,
+      "learning_rate": 0.00015594525966449062,
+      "loss": 0.6522,
+      "step": 911
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.07412074506282806,
+      "learning_rate": 0.00015585534249725359,
+      "loss": 0.7589,
+      "step": 912
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.07159864902496338,
+      "learning_rate": 0.00015576535964483094,
+      "loss": 0.6933,
+      "step": 913
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.0748409703373909,
+      "learning_rate": 0.00015567531121304134,
+      "loss": 0.7603,
+      "step": 914
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.08181232959032059,
+      "learning_rate": 0.00015558519730778068,
+      "loss": 0.6875,
+      "step": 915
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.08348861336708069,
+      "learning_rate": 0.00015549501803502173,
+      "loss": 0.6878,
+      "step": 916
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.07457895576953888,
+      "learning_rate": 0.00015540477350081423,
+      "loss": 0.7035,
+      "step": 917
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.07142025232315063,
+      "learning_rate": 0.00015531446381128464,
+      "loss": 0.6809,
+      "step": 918
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.07224272936582565,
+      "learning_rate": 0.00015522408907263596,
+      "loss": 0.6987,
+      "step": 919
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.07358361780643463,
+      "learning_rate": 0.0001551336493911478,
+      "loss": 0.7223,
+      "step": 920
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.07335005700588226,
+      "learning_rate": 0.00015504314487317612,
+      "loss": 0.6515,
+      "step": 921
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.0723937451839447,
+      "learning_rate": 0.00015495257562515307,
+      "loss": 0.7698,
+      "step": 922
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.06864594668149948,
+      "learning_rate": 0.00015486194175358696,
+      "loss": 0.6483,
+      "step": 923
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.07635841518640518,
+      "learning_rate": 0.00015477124336506207,
+      "loss": 0.6139,
+      "step": 924
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.08424652367830276,
+      "learning_rate": 0.0001546804805662386,
+      "loss": 0.6604,
+      "step": 925
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.07373332232236862,
+      "learning_rate": 0.00015458965346385248,
+      "loss": 0.734,
+      "step": 926
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.07332025468349457,
+      "learning_rate": 0.00015449876216471525,
+      "loss": 0.7765,
+      "step": 927
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.07534317672252655,
+      "learning_rate": 0.00015440780677571388,
+      "loss": 0.7436,
+      "step": 928
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.08624652773141861,
+      "learning_rate": 0.00015431678740381085,
+      "loss": 0.7335,
+      "step": 929
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.07540034502744675,
+      "learning_rate": 0.00015422570415604377,
+      "loss": 0.6808,
+      "step": 930
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.07505679130554199,
+      "learning_rate": 0.00015413455713952538,
+      "loss": 0.7086,
+      "step": 931
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.08109767735004425,
+      "learning_rate": 0.0001540433464614435,
+      "loss": 0.8099,
+      "step": 932
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.07563024014234543,
+      "learning_rate": 0.00015395207222906068,
+      "loss": 0.7273,
+      "step": 933
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.07199395447969437,
+      "learning_rate": 0.00015386073454971432,
+      "loss": 0.7727,
+      "step": 934
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.08543197810649872,
+      "learning_rate": 0.00015376933353081635,
+      "loss": 0.705,
+      "step": 935
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.07872208952903748,
+      "learning_rate": 0.00015367786927985327,
+      "loss": 0.7509,
+      "step": 936
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.0703720971941948,
+      "learning_rate": 0.00015358634190438592,
+      "loss": 0.7214,
+      "step": 937
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.0741022378206253,
+      "learning_rate": 0.0001534947515120493,
+      "loss": 0.7421,
+      "step": 938
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.07758384943008423,
+      "learning_rate": 0.00015340309821055255,
+      "loss": 0.7013,
+      "step": 939
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.07203276455402374,
+      "learning_rate": 0.00015331138210767885,
+      "loss": 0.7268,
+      "step": 940
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.09579942375421524,
+      "learning_rate": 0.0001532196033112852,
+      "loss": 0.6617,
+      "step": 941
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.07915519922971725,
+      "learning_rate": 0.00015312776192930228,
+      "loss": 0.7223,
+      "step": 942
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.07456015795469284,
+      "learning_rate": 0.00015303585806973445,
+      "loss": 0.5984,
+      "step": 943
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.07331335544586182,
+      "learning_rate": 0.0001529438918406595,
+      "loss": 0.6881,
+      "step": 944
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.08498119562864304,
+      "learning_rate": 0.00015285186335022854,
+      "loss": 0.7509,
+      "step": 945
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.08055809885263443,
+      "learning_rate": 0.0001527597727066659,
+      "loss": 0.7643,
+      "step": 946
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07158636301755905,
+      "learning_rate": 0.00015266762001826911,
+      "loss": 0.6522,
+      "step": 947
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.0752343162894249,
+      "learning_rate": 0.00015257540539340852,
+      "loss": 0.6749,
+      "step": 948
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07281875610351562,
+      "learning_rate": 0.0001524831289405274,
+      "loss": 0.6796,
+      "step": 949
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.08019158244132996,
+      "learning_rate": 0.00015239079076814166,
+      "loss": 0.6577,
+      "step": 950
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07102995365858078,
+      "learning_rate": 0.00015229839098483992,
+      "loss": 0.7034,
+      "step": 951
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07253885269165039,
+      "learning_rate": 0.00015220592969928315,
+      "loss": 0.6898,
+      "step": 952
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.08555888384580612,
+      "learning_rate": 0.0001521134070202046,
+      "loss": 0.6885,
+      "step": 953
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.09179149568080902,
+      "learning_rate": 0.00015202082305640984,
+      "loss": 0.6164,
+      "step": 954
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.07192881405353546,
+      "learning_rate": 0.00015192817791677646,
+      "loss": 0.6862,
+      "step": 955
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07520081102848053,
+      "learning_rate": 0.00015183547171025399,
+      "loss": 0.6878,
+      "step": 956
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07251414656639099,
+      "learning_rate": 0.00015174270454586375,
+      "loss": 0.7239,
+      "step": 957
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.0739588811993599,
+      "learning_rate": 0.00015164987653269876,
+      "loss": 0.7359,
+      "step": 958
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07903960347175598,
+      "learning_rate": 0.00015155698777992365,
+      "loss": 0.6744,
+      "step": 959
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07801897823810577,
+      "learning_rate": 0.0001514640383967744,
+      "loss": 0.7226,
+      "step": 960
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.0692887008190155,
+      "learning_rate": 0.0001513710284925583,
+      "loss": 0.7153,
+      "step": 961
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07592754065990448,
+      "learning_rate": 0.0001512779581766539,
+      "loss": 0.7229,
+      "step": 962
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07932160049676895,
+      "learning_rate": 0.00015118482755851069,
+      "loss": 0.6589,
+      "step": 963
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.07948953658342361,
+      "learning_rate": 0.0001510916367476491,
+      "loss": 0.7122,
+      "step": 964
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07656390219926834,
+      "learning_rate": 0.00015099838585366042,
+      "loss": 0.7,
+      "step": 965
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07349363714456558,
+      "learning_rate": 0.00015090507498620649,
+      "loss": 0.688,
+      "step": 966
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07416553050279617,
+      "learning_rate": 0.0001508117042550197,
+      "loss": 0.6698,
+      "step": 967
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07723015546798706,
+      "learning_rate": 0.0001507182737699029,
+      "loss": 0.7014,
+      "step": 968
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07453907281160355,
+      "learning_rate": 0.0001506247836407292,
+      "loss": 0.6841,
+      "step": 969
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.08041400462388992,
+      "learning_rate": 0.0001505312339774418,
+      "loss": 0.714,
+      "step": 970
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.0772874653339386,
+      "learning_rate": 0.0001504376248900539,
+      "loss": 0.6877,
+      "step": 971
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07476682960987091,
+      "learning_rate": 0.0001503439564886487,
+      "loss": 0.6952,
+      "step": 972
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.8217894434928894,
+      "eval_runtime": 111.9926,
+      "eval_samples_per_second": 88.738,
+      "eval_steps_per_second": 22.189,
+      "step": 972
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07859773933887482,
+      "learning_rate": 0.000150250228883379,
+      "loss": 0.7203,
+      "step": 973
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07395817339420319,
+      "learning_rate": 0.00015015644218446732,
+      "loss": 0.6552,
+      "step": 974
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07756183296442032,
+      "learning_rate": 0.00015006259650220564,
+      "loss": 0.7028,
+      "step": 975
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07802776247262955,
+      "learning_rate": 0.00014996869194695533,
+      "loss": 0.7267,
+      "step": 976
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07869012653827667,
+      "learning_rate": 0.00014987472862914696,
+      "loss": 0.7803,
+      "step": 977
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07955232262611389,
+      "learning_rate": 0.0001497807066592802,
+      "loss": 0.7248,
+      "step": 978
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07331852614879608,
+      "learning_rate": 0.00014968662614792377,
+      "loss": 0.7308,
+      "step": 979
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07007934898138046,
+      "learning_rate": 0.0001495924872057151,
+      "loss": 0.7123,
+      "step": 980
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.0728902593255043,
+      "learning_rate": 0.00014949828994336047,
+      "loss": 0.7262,
+      "step": 981
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.08288244158029556,
+      "learning_rate": 0.00014940403447163467,
+      "loss": 0.7176,
+      "step": 982
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07168509066104889,
+      "learning_rate": 0.00014930972090138097,
+      "loss": 0.68,
+      "step": 983
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07297495007514954,
+      "learning_rate": 0.00014921534934351092,
+      "loss": 0.6429,
+      "step": 984
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.07195892184972763,
+      "learning_rate": 0.00014912091990900429,
+      "loss": 0.6569,
+      "step": 985
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.073966845870018,
+      "learning_rate": 0.00014902643270890895,
+      "loss": 0.6855,
+      "step": 986
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.07951618731021881,
+      "learning_rate": 0.00014893188785434068,
+      "loss": 0.7277,
+      "step": 987
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07925249636173248,
+      "learning_rate": 0.000148837285456483,
+      "loss": 0.6752,
+      "step": 988
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07493434846401215,
+      "learning_rate": 0.00014874262562658715,
+      "loss": 0.6942,
+      "step": 989
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07437346130609512,
+      "learning_rate": 0.00014864790847597198,
+      "loss": 0.671,
+      "step": 990
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07806906849145889,
+      "learning_rate": 0.0001485531341160236,
+      "loss": 0.6263,
+      "step": 991
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.07348716259002686,
+      "learning_rate": 0.0001484583026581955,
+      "loss": 0.662,
+      "step": 992
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07612631469964981,
+      "learning_rate": 0.00014836341421400832,
+      "loss": 0.6662,
+      "step": 993
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07551120966672897,
+      "learning_rate": 0.00014826846889504964,
+      "loss": 0.5952,
+      "step": 994
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07422953099012375,
+      "learning_rate": 0.000148173466812974,
+      "loss": 0.6016,
+      "step": 995
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.08229976147413254,
+      "learning_rate": 0.00014807840807950262,
+      "loss": 0.6624,
+      "step": 996
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.0733613446354866,
+      "learning_rate": 0.00014798329280642344,
+      "loss": 0.6454,
+      "step": 997
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07359236478805542,
+      "learning_rate": 0.00014788812110559082,
+      "loss": 0.6386,
+      "step": 998
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.08044279366731644,
+      "learning_rate": 0.00014779289308892546,
+      "loss": 0.6251,
+      "step": 999
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07739145308732986,
+      "learning_rate": 0.00014769760886841436,
+      "loss": 0.6926,
+      "step": 1000
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.0751839205622673,
+      "learning_rate": 0.00014760226855611055,
+      "loss": 0.674,
+      "step": 1001
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.07309143990278244,
+      "learning_rate": 0.00014750687226413304,
+      "loss": 0.637,
+      "step": 1002
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.07492461055517197,
+      "learning_rate": 0.0001474114201046667,
+      "loss": 0.6801,
+      "step": 1003
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.08308755606412888,
+      "learning_rate": 0.0001473159121899621,
+      "loss": 0.6568,
+      "step": 1004
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.07458531856536865,
+      "learning_rate": 0.00014722034863233528,
+      "loss": 0.6841,
+      "step": 1005
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.081693135201931,
+      "learning_rate": 0.00014712472954416788,
+      "loss": 0.6072,
+      "step": 1006
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.08209308981895447,
+      "learning_rate": 0.00014702905503790668,
+      "loss": 0.6537,
+      "step": 1007
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.08696984499692917,
+      "learning_rate": 0.00014693332522606374,
+      "loss": 0.6942,
+      "step": 1008
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.08275436609983444,
+      "learning_rate": 0.00014683754022121613,
+      "loss": 0.6477,
+      "step": 1009
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.07651187479496002,
+      "learning_rate": 0.0001467417001360058,
+      "loss": 0.6399,
+      "step": 1010
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.07590115815401077,
+      "learning_rate": 0.0001466458050831395,
+      "loss": 0.6175,
+      "step": 1011
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.08208300173282623,
+      "learning_rate": 0.00014654985517538864,
+      "loss": 0.7188,
+      "step": 1012
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.0778549537062645,
+      "learning_rate": 0.00014645385052558906,
+      "loss": 0.644,
+      "step": 1013
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.07480096817016602,
+      "learning_rate": 0.00014635779124664108,
+      "loss": 0.6407,
+      "step": 1014
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.07706421613693237,
+      "learning_rate": 0.00014626167745150924,
+      "loss": 0.6488,
+      "step": 1015
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.08114165812730789,
+      "learning_rate": 0.00014616550925322205,
+      "loss": 0.6098,
+      "step": 1016
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.07810718566179276,
+      "learning_rate": 0.0001460692867648722,
+      "loss": 0.6655,
+      "step": 1017
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.07784634083509445,
+      "learning_rate": 0.00014597301009961614,
+      "loss": 0.6166,
+      "step": 1018
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.08003939688205719,
+      "learning_rate": 0.000145876679370674,
+      "loss": 0.6955,
+      "step": 1019
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.0889209434390068,
+      "learning_rate": 0.0001457802946913295,
+      "loss": 0.6096,
+      "step": 1020
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.08176947385072708,
+      "learning_rate": 0.00014568385617492986,
+      "loss": 0.6687,
+      "step": 1021
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.08003947138786316,
+      "learning_rate": 0.0001455873639348855,
+      "loss": 0.6139,
+      "step": 1022
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.07719139754772186,
+      "learning_rate": 0.0001454908180846702,
+      "loss": 0.5775,
+      "step": 1023
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.08436036854982376,
+      "learning_rate": 0.0001453942187378206,
+      "loss": 0.725,
+      "step": 1024
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.07681979238986969,
+      "learning_rate": 0.0001452975660079363,
+      "loss": 0.5902,
+      "step": 1025
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.09957487136125565,
+      "learning_rate": 0.00014520086000867978,
+      "loss": 0.622,
+      "step": 1026
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.08798941224813461,
+      "learning_rate": 0.00014510410085377605,
+      "loss": 0.6735,
+      "step": 1027
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.07842235267162323,
+      "learning_rate": 0.00014500728865701268,
+      "loss": 0.6207,
+      "step": 1028
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.08948861807584763,
+      "learning_rate": 0.00014491042353223955,
+      "loss": 0.622,
+      "step": 1029
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.08005297183990479,
+      "learning_rate": 0.0001448135055933689,
+      "loss": 0.6869,
+      "step": 1030
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.08026837557554245,
+      "learning_rate": 0.00014471653495437504,
+      "loss": 0.6351,
+      "step": 1031
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.08097610622644424,
+      "learning_rate": 0.00014461951172929419,
+      "loss": 0.6767,
+      "step": 1032
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.08407065272331238,
+      "learning_rate": 0.00014452243603222442,
+      "loss": 0.7305,
+      "step": 1033
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.07626531273126602,
+      "learning_rate": 0.00014442530797732558,
+      "loss": 0.6222,
+      "step": 1034
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.08377376943826675,
+      "learning_rate": 0.00014432812767881906,
+      "loss": 0.6118,
+      "step": 1035
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.09067460894584656,
+      "learning_rate": 0.00014423089525098762,
+      "loss": 0.7058,
+      "step": 1036
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.08191598951816559,
+      "learning_rate": 0.00014413361080817545,
+      "loss": 0.6532,
+      "step": 1037
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.07789404690265656,
+      "learning_rate": 0.0001440362744647878,
+      "loss": 0.6818,
+      "step": 1038
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.08609960973262787,
+      "learning_rate": 0.00014393888633529097,
+      "loss": 0.7429,
+      "step": 1039
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.07990411669015884,
+      "learning_rate": 0.0001438414465342123,
+      "loss": 0.6742,
+      "step": 1040
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.08147672563791275,
+      "learning_rate": 0.00014374395517613965,
+      "loss": 0.651,
+      "step": 1041
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.07946927845478058,
+      "learning_rate": 0.0001436464123757217,
+      "loss": 0.6202,
+      "step": 1042
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.08471754193305969,
+      "learning_rate": 0.00014354881824766758,
+      "loss": 0.6345,
+      "step": 1043
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.08079022914171219,
+      "learning_rate": 0.00014345117290674673,
+      "loss": 0.6403,
+      "step": 1044
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.0815759152173996,
+      "learning_rate": 0.00014335347646778884,
+      "loss": 0.653,
+      "step": 1045
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.08412860333919525,
+      "learning_rate": 0.00014325572904568375,
+      "loss": 0.6698,
+      "step": 1046
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.08337196707725525,
+      "learning_rate": 0.0001431579307553812,
+      "loss": 0.6364,
+      "step": 1047
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.0795850083231926,
+      "learning_rate": 0.00014306008171189073,
+      "loss": 0.6643,
+      "step": 1048
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.08132953196763992,
+      "learning_rate": 0.00014296218203028159,
+      "loss": 0.6957,
+      "step": 1049
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.08360311388969421,
+      "learning_rate": 0.00014286423182568265,
+      "loss": 0.6767,
+      "step": 1050
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.08610215783119202,
+      "learning_rate": 0.00014276623121328206,
+      "loss": 0.5871,
+      "step": 1051
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.07904227823019028,
+      "learning_rate": 0.0001426681803083273,
+      "loss": 0.6672,
+      "step": 1052
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.07742542773485184,
+      "learning_rate": 0.0001425700792261251,
+      "loss": 0.5281,
+      "step": 1053
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.0755574032664299,
+      "learning_rate": 0.00014247192808204107,
+      "loss": 0.6197,
+      "step": 1054
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.08822150528430939,
+      "learning_rate": 0.00014237372699149966,
+      "loss": 0.6049,
+      "step": 1055
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.0828738734126091,
+      "learning_rate": 0.00014227547606998422,
+      "loss": 0.651,
+      "step": 1056
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.07611016184091568,
+      "learning_rate": 0.00014217717543303656,
+      "loss": 0.6297,
+      "step": 1057
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.0859667956829071,
+      "learning_rate": 0.00014207882519625705,
+      "loss": 0.6498,
+      "step": 1058
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.08264627307653427,
+      "learning_rate": 0.00014198042547530423,
+      "loss": 0.7216,
+      "step": 1059
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.08394581079483032,
+      "learning_rate": 0.0001418819763858951,
+      "loss": 0.5874,
+      "step": 1060
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08443103730678558,
+      "learning_rate": 0.00014178347804380442,
+      "loss": 0.7102,
+      "step": 1061
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08027410507202148,
+      "learning_rate": 0.00014168493056486512,
+      "loss": 0.6305,
+      "step": 1062
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08662614971399307,
+      "learning_rate": 0.00014158633406496771,
+      "loss": 0.7375,
+      "step": 1063
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08393966406583786,
+      "learning_rate": 0.0001414876886600605,
+      "loss": 0.7086,
+      "step": 1064
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08305832743644714,
+      "learning_rate": 0.00014138899446614926,
+      "loss": 0.8568,
+      "step": 1065
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.09047157317399979,
+      "learning_rate": 0.00014129025159929708,
+      "loss": 0.6329,
+      "step": 1066
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08572204411029816,
+      "learning_rate": 0.00014119146017562442,
+      "loss": 0.7041,
+      "step": 1067
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08337341994047165,
+      "learning_rate": 0.00014109262031130866,
+      "loss": 0.657,
+      "step": 1068
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.084062360227108,
+      "learning_rate": 0.00014099373212258436,
+      "loss": 0.6787,
+      "step": 1069
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.08505363017320633,
+      "learning_rate": 0.0001408947957257427,
+      "loss": 0.645,
+      "step": 1070
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.07976432144641876,
+      "learning_rate": 0.0001407958112371317,
+      "loss": 0.5915,
+      "step": 1071
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.08692666888237,
+      "learning_rate": 0.00014069677877315585,
+      "loss": 0.6989,
+      "step": 1072
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.09206826239824295,
+      "learning_rate": 0.0001405976984502761,
+      "loss": 0.673,
+      "step": 1073
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.08507773280143738,
+      "learning_rate": 0.0001404985703850097,
+      "loss": 0.6632,
+      "step": 1074
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.08390660583972931,
+      "learning_rate": 0.00014039939469393003,
+      "loss": 0.6248,
+      "step": 1075
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.08538491278886795,
+      "learning_rate": 0.00014030017149366638,
+      "loss": 0.6676,
+      "step": 1076
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.07793683558702469,
+      "learning_rate": 0.00014020090090090408,
+      "loss": 0.7204,
+      "step": 1077
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.08168508857488632,
+      "learning_rate": 0.00014010158303238404,
+      "loss": 0.6617,
+      "step": 1078
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.08362981677055359,
+      "learning_rate": 0.00014000221800490289,
+      "loss": 0.6181,
+      "step": 1079
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.08173540234565735,
+      "learning_rate": 0.0001399028059353127,
+      "loss": 0.6241,
+      "step": 1080
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.08719956874847412,
+      "learning_rate": 0.0001398033469405207,
+      "loss": 0.627,
+      "step": 1081
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.08081477880477905,
+      "learning_rate": 0.00013970384113748952,
+      "loss": 0.5977,
+      "step": 1082
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.07814403623342514,
+      "learning_rate": 0.0001396042886432367,
+      "loss": 0.6981,
+      "step": 1083
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.09654086083173752,
+      "learning_rate": 0.00013950468957483474,
+      "loss": 0.587,
+      "step": 1084
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.07919298857450485,
+      "learning_rate": 0.00013940504404941093,
+      "loss": 0.5708,
+      "step": 1085
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.08666522055864334,
+      "learning_rate": 0.00013930535218414713,
+      "loss": 0.6015,
+      "step": 1086
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.0874529704451561,
+      "learning_rate": 0.00013920561409627973,
+      "loss": 0.6765,
+      "step": 1087
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.08722787350416183,
+      "learning_rate": 0.0001391058299030995,
+      "loss": 0.6217,
+      "step": 1088
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.0871277004480362,
+      "learning_rate": 0.0001390059997219514,
+      "loss": 0.6789,
+      "step": 1089
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.08737975358963013,
+      "learning_rate": 0.00013890612367023445,
+      "loss": 0.6159,
+      "step": 1090
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.0826662927865982,
+      "learning_rate": 0.0001388062018654017,
+      "loss": 0.6609,
+      "step": 1091
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.07992442697286606,
+      "learning_rate": 0.00013870623442495988,
+      "loss": 0.6156,
+      "step": 1092
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.09024856239557266,
+      "learning_rate": 0.00013860622146646948,
+      "loss": 0.6537,
+      "step": 1093
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.08088452368974686,
+      "learning_rate": 0.00013850616310754447,
+      "loss": 0.6601,
+      "step": 1094
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.08854856342077255,
+      "learning_rate": 0.0001384060594658523,
+      "loss": 0.6881,
+      "step": 1095
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.08140786737203598,
+      "learning_rate": 0.0001383059106591135,
+      "loss": 0.5797,
+      "step": 1096
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.08662276715040207,
+      "learning_rate": 0.00013820571680510187,
+      "loss": 0.7146,
+      "step": 1097
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.08150844275951385,
+      "learning_rate": 0.00013810547802164415,
+      "loss": 0.6315,
+      "step": 1098
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.08309885114431381,
+      "learning_rate": 0.00013800519442661987,
+      "loss": 0.7192,
+      "step": 1099
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.13309398293495178,
+      "learning_rate": 0.00013790486613796132,
+      "loss": 0.6507,
+      "step": 1100
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.08919757604598999,
+      "learning_rate": 0.00013780449327365328,
+      "loss": 0.6892,
+      "step": 1101
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.0828130841255188,
+      "learning_rate": 0.000137704075951733,
+      "loss": 0.6033,
+      "step": 1102
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.08281046152114868,
+      "learning_rate": 0.00013760361429028997,
+      "loss": 0.7257,
+      "step": 1103
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.08436671644449234,
+      "learning_rate": 0.00013750310840746592,
+      "loss": 0.6299,
+      "step": 1104
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.08063504844903946,
+      "learning_rate": 0.00013740255842145443,
+      "loss": 0.6513,
+      "step": 1105
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.08479522168636322,
+      "learning_rate": 0.00013730196445050106,
+      "loss": 0.7366,
+      "step": 1106
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.09070105105638504,
+      "learning_rate": 0.0001372013266129031,
+      "loss": 0.7016,
+      "step": 1107
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.09310469776391983,
+      "learning_rate": 0.00013710064502700932,
+      "loss": 0.5898,
+      "step": 1108
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.08128984272480011,
+      "learning_rate": 0.00013699991981122012,
+      "loss": 0.6372,
+      "step": 1109
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.08728478848934174,
+      "learning_rate": 0.00013689915108398696,
+      "loss": 0.6544,
+      "step": 1110
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.0837821215391159,
+      "learning_rate": 0.00013679833896381273,
+      "loss": 0.6459,
+      "step": 1111
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.0833863690495491,
+      "learning_rate": 0.00013669748356925113,
+      "loss": 0.6406,
+      "step": 1112
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.08275511115789413,
+      "learning_rate": 0.00013659658501890688,
+      "loss": 0.6217,
+      "step": 1113
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.10468558967113495,
+      "learning_rate": 0.00013649564343143545,
+      "loss": 0.6292,
+      "step": 1114
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.08720619231462479,
+      "learning_rate": 0.00013639465892554281,
+      "loss": 0.6986,
+      "step": 1115
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.08144842833280563,
+      "learning_rate": 0.00013629363161998557,
+      "loss": 0.6447,
+      "step": 1116
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.08362375944852829,
+      "learning_rate": 0.00013619256163357045,
+      "loss": 0.6619,
+      "step": 1117
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.08400712162256241,
+      "learning_rate": 0.00013609144908515467,
+      "loss": 0.6424,
+      "step": 1118
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08988272398710251,
+      "learning_rate": 0.0001359902940936452,
+      "loss": 0.6418,
+      "step": 1119
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08508805185556412,
+      "learning_rate": 0.0001358890967779991,
+      "loss": 0.6585,
+      "step": 1120
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08776996284723282,
+      "learning_rate": 0.0001357878572572231,
+      "loss": 0.6703,
+      "step": 1121
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08439894020557404,
+      "learning_rate": 0.00013568657565037363,
+      "loss": 0.6661,
+      "step": 1122
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08871582895517349,
+      "learning_rate": 0.00013558525207655665,
+      "loss": 0.6791,
+      "step": 1123
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08871687948703766,
+      "learning_rate": 0.00013548388665492734,
+      "loss": 0.6057,
+      "step": 1124
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08875562995672226,
+      "learning_rate": 0.0001353824795046902,
+      "loss": 0.6164,
+      "step": 1125
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08401204645633698,
+      "learning_rate": 0.0001352810307450988,
+      "loss": 0.6145,
+      "step": 1126
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08974327147006989,
+      "learning_rate": 0.0001351795404954556,
+      "loss": 0.5835,
+      "step": 1127
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.08411118388175964,
+      "learning_rate": 0.00013507800887511184,
+      "loss": 0.6387,
+      "step": 1128
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.08708637207746506,
+      "learning_rate": 0.0001349764360034675,
+      "loss": 0.6695,
+      "step": 1129
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.08665607869625092,
+      "learning_rate": 0.00013487482199997094,
+      "loss": 0.6305,
+      "step": 1130
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.09944334626197815,
+      "learning_rate": 0.000134773166984119,
+      "loss": 0.6732,
+      "step": 1131
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.08623045682907104,
+      "learning_rate": 0.0001346714710754567,
+      "loss": 0.7654,
+      "step": 1132
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.08571859449148178,
+      "learning_rate": 0.00013456973439357714,
+      "loss": 0.6327,
+      "step": 1133
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.08249690383672714,
+      "learning_rate": 0.0001344679570581214,
+      "loss": 0.6726,
+      "step": 1134
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.08060690760612488,
+      "learning_rate": 0.00013436613918877836,
+      "loss": 0.5553,
+      "step": 1135
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.08668998628854752,
+      "learning_rate": 0.0001342642809052846,
+      "loss": 0.6979,
+      "step": 1136
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.10263362526893616,
+      "learning_rate": 0.00013416238232742412,
+      "loss": 0.6737,
+      "step": 1137
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.0849577784538269,
+      "learning_rate": 0.00013406044357502846,
+      "loss": 0.6482,
+      "step": 1138
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.09012065082788467,
+      "learning_rate": 0.00013395846476797621,
+      "loss": 0.6301,
+      "step": 1139
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.08457211405038834,
+      "learning_rate": 0.0001338564460261933,
+      "loss": 0.6857,
+      "step": 1140
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.08212517201900482,
+      "learning_rate": 0.0001337543874696524,
+      "loss": 0.5929,
+      "step": 1141
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.08315583318471909,
+      "learning_rate": 0.00013365228921837313,
+      "loss": 0.6259,
+      "step": 1142
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.08320874720811844,
+      "learning_rate": 0.00013355015139242179,
+      "loss": 0.6509,
+      "step": 1143
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.08782950043678284,
+      "learning_rate": 0.00013344797411191111,
+      "loss": 0.6884,
+      "step": 1144
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.086082823574543,
+      "learning_rate": 0.00013334575749700037,
+      "loss": 0.679,
+      "step": 1145
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.08684473484754562,
+      "learning_rate": 0.000133243501667895,
+      "loss": 0.6226,
+      "step": 1146
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.10077089816331863,
+      "learning_rate": 0.00013314120674484662,
+      "loss": 0.6458,
+      "step": 1147
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08714549988508224,
+      "learning_rate": 0.00013303887284815273,
+      "loss": 0.6945,
+      "step": 1148
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.09142305701971054,
+      "learning_rate": 0.00013293650009815674,
+      "loss": 0.6625,
+      "step": 1149
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.09278015792369843,
+      "learning_rate": 0.0001328340886152477,
+      "loss": 0.6438,
+      "step": 1150
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08436746150255203,
+      "learning_rate": 0.0001327316385198603,
+      "loss": 0.695,
+      "step": 1151
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08299244940280914,
+      "learning_rate": 0.00013262914993247452,
+      "loss": 0.6092,
+      "step": 1152
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08477385342121124,
+      "learning_rate": 0.00013252662297361567,
+      "loss": 0.6172,
+      "step": 1153
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08768115937709808,
+      "learning_rate": 0.00013242405776385416,
+      "loss": 0.6735,
+      "step": 1154
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08845745027065277,
+      "learning_rate": 0.00013232145442380538,
+      "loss": 0.7098,
+      "step": 1155
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08628155291080475,
+      "learning_rate": 0.0001322188130741297,
+      "loss": 0.6166,
+      "step": 1156
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.08622749894857407,
+      "learning_rate": 0.0001321161338355319,
+      "loss": 0.6528,
+      "step": 1157
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08740253746509552,
+      "learning_rate": 0.0001320134168287616,
+      "loss": 0.6969,
+      "step": 1158
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08739987760782242,
+      "learning_rate": 0.00013191066217461264,
+      "loss": 0.6608,
+      "step": 1159
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08395621925592422,
+      "learning_rate": 0.0001318078699939233,
+      "loss": 0.6596,
+      "step": 1160
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.09416764974594116,
+      "learning_rate": 0.0001317050404075758,
+      "loss": 0.6476,
+      "step": 1161
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08487122505903244,
+      "learning_rate": 0.0001316021735364965,
+      "loss": 0.6126,
+      "step": 1162
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08098070323467255,
+      "learning_rate": 0.0001314992695016556,
+      "loss": 0.6175,
+      "step": 1163
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08447708189487457,
+      "learning_rate": 0.00013139632842406687,
+      "loss": 0.6606,
+      "step": 1164
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08934439718723297,
+      "learning_rate": 0.00013129335042478778,
+      "loss": 0.6941,
+      "step": 1165
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.08715163171291351,
+      "learning_rate": 0.00013119033562491918,
+      "loss": 0.6431,
+      "step": 1166
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.0908278077840805,
+      "learning_rate": 0.0001310872841456052,
+      "loss": 0.6351,
+      "step": 1167
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.08657587319612503,
+      "learning_rate": 0.00013098419610803303,
+      "loss": 0.6067,
+      "step": 1168
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.09114498645067215,
+      "learning_rate": 0.00013088107163343303,
+      "loss": 0.7069,
+      "step": 1169
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.08801303058862686,
+      "learning_rate": 0.00013077791084307812,
+      "loss": 0.7103,
+      "step": 1170
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.08879801630973816,
+      "learning_rate": 0.00013067471385828428,
+      "loss": 0.6679,
+      "step": 1171
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.0977342426776886,
+      "learning_rate": 0.0001305714808004098,
+      "loss": 0.6865,
+      "step": 1172
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.08699452131986618,
+      "learning_rate": 0.00013046821179085545,
+      "loss": 0.6377,
+      "step": 1173
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.08781784772872925,
+      "learning_rate": 0.00013036490695106433,
+      "loss": 0.6666,
+      "step": 1174
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.08928069472312927,
+      "learning_rate": 0.0001302615664025216,
+      "loss": 0.6447,
+      "step": 1175
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.09226062148809433,
+      "learning_rate": 0.00013015819026675448,
+      "loss": 0.6655,
+      "step": 1176
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.09220758080482483,
+      "learning_rate": 0.00013005477866533203,
+      "loss": 0.6737,
+      "step": 1177
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.0914442241191864,
+      "learning_rate": 0.000129951331719865,
+      "loss": 0.7703,
+      "step": 1178
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.08964721113443375,
+      "learning_rate": 0.0001298478495520057,
+      "loss": 0.7136,
+      "step": 1179
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.08899814635515213,
+      "learning_rate": 0.00012974433228344788,
+      "loss": 0.6389,
+      "step": 1180
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.08671650290489197,
+      "learning_rate": 0.0001296407800359265,
+      "loss": 0.6658,
+      "step": 1181
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.09150044620037079,
+      "learning_rate": 0.00012953719293121776,
+      "loss": 0.7211,
+      "step": 1182
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.08706536889076233,
+      "learning_rate": 0.0001294335710911388,
+      "loss": 0.6188,
+      "step": 1183
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.09030577540397644,
+      "learning_rate": 0.00012932991463754758,
+      "loss": 0.5925,
+      "step": 1184
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.0844309851527214,
+      "learning_rate": 0.0001292262236923429,
+      "loss": 0.628,
+      "step": 1185
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.08728198707103729,
+      "learning_rate": 0.00012912249837746382,
+      "loss": 0.6787,
+      "step": 1186
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08774858713150024,
+      "learning_rate": 0.0001290187388148902,
+      "loss": 0.7292,
+      "step": 1187
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08736085146665573,
+      "learning_rate": 0.0001289149451266419,
+      "loss": 0.6155,
+      "step": 1188
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08831470459699631,
+      "learning_rate": 0.00012881111743477907,
+      "loss": 0.6692,
+      "step": 1189
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08998765051364899,
+      "learning_rate": 0.0001287072558614017,
+      "loss": 0.6844,
+      "step": 1190
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08641190081834793,
+      "learning_rate": 0.00012860336052864977,
+      "loss": 0.654,
+      "step": 1191
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08620987832546234,
+      "learning_rate": 0.00012849943155870283,
+      "loss": 0.5701,
+      "step": 1192
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.09002070128917694,
+      "learning_rate": 0.0001283954690737801,
+      "loss": 0.7221,
+      "step": 1193
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08585362881422043,
+      "learning_rate": 0.00012829147319614014,
+      "loss": 0.6972,
+      "step": 1194
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.09278030693531036,
+      "learning_rate": 0.00012818744404808081,
+      "loss": 0.7298,
+      "step": 1195
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.08575819432735443,
+      "learning_rate": 0.00012808338175193912,
+      "loss": 0.68,
+      "step": 1196
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.08484896272420883,
+      "learning_rate": 0.00012797928643009095,
+      "loss": 0.6956,
+      "step": 1197
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.1123301237821579,
+      "learning_rate": 0.00012787515820495123,
+      "loss": 0.653,
+      "step": 1198
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.08775759488344193,
+      "learning_rate": 0.00012777099719897333,
+      "loss": 0.6974,
+      "step": 1199
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.09750175476074219,
+      "learning_rate": 0.00012766680353464938,
+      "loss": 0.6307,
+      "step": 1200
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.08309242129325867,
+      "learning_rate": 0.00012756257733450972,
+      "loss": 0.6872,
+      "step": 1201
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.09836401790380478,
+      "learning_rate": 0.00012745831872112317,
+      "loss": 0.6586,
+      "step": 1202
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.09071850031614304,
+      "learning_rate": 0.00012735402781709654,
+      "loss": 0.5934,
+      "step": 1203
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.08511752635240555,
+      "learning_rate": 0.00012724970474507457,
+      "loss": 0.586,
+      "step": 1204
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.09099071472883224,
+      "learning_rate": 0.00012714534962773993,
+      "loss": 0.6747,
+      "step": 1205
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.08356339484453201,
+      "learning_rate": 0.00012704096258781296,
+      "loss": 0.6421,
+      "step": 1206
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.09262317419052124,
+      "learning_rate": 0.0001269365437480515,
+      "loss": 0.6963,
+      "step": 1207
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.08675730228424072,
+      "learning_rate": 0.00012683209323125074,
+      "loss": 0.7039,
+      "step": 1208
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.0890946015715599,
+      "learning_rate": 0.0001267276111602433,
+      "loss": 0.6936,
+      "step": 1209
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.09518737345933914,
+      "learning_rate": 0.0001266230976578987,
+      "loss": 0.6067,
+      "step": 1210
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.08667458593845367,
+      "learning_rate": 0.00012651855284712355,
+      "loss": 0.6358,
+      "step": 1211
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.08841356635093689,
+      "learning_rate": 0.00012641397685086124,
+      "loss": 0.638,
+      "step": 1212
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.08399655669927597,
+      "learning_rate": 0.0001263093697920918,
+      "loss": 0.6377,
+      "step": 1213
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.09324619174003601,
+      "learning_rate": 0.0001262047317938319,
+      "loss": 0.7131,
+      "step": 1214
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.08485792577266693,
+      "learning_rate": 0.00012610006297913444,
+      "loss": 0.701,
+      "step": 1215
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.8341993689537048,
+      "eval_runtime": 111.9397,
+      "eval_samples_per_second": 88.78,
+      "eval_steps_per_second": 22.199,
+      "step": 1215
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.0880061462521553,
+      "learning_rate": 0.0001259953634710887,
+      "loss": 0.7021,
+      "step": 1216
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.0873493030667305,
+      "learning_rate": 0.00012589063339281994,
+      "loss": 0.6999,
+      "step": 1217
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.08342842012643814,
+      "learning_rate": 0.00012578587286748953,
+      "loss": 0.6034,
+      "step": 1218
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.09012749046087265,
+      "learning_rate": 0.0001256810820182944,
+      "loss": 0.6671,
+      "step": 1219
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.09362991899251938,
+      "learning_rate": 0.0001255762609684674,
+      "loss": 0.6353,
+      "step": 1220
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.08616290241479874,
+      "learning_rate": 0.00012547140984127668,
+      "loss": 0.6331,
+      "step": 1221
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.09951253235340118,
+      "learning_rate": 0.0001253665287600259,
+      "loss": 0.6234,
+      "step": 1222
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.0860767513513565,
+      "learning_rate": 0.00012526161784805393,
+      "loss": 0.6465,
+      "step": 1223
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.0897819995880127,
+      "learning_rate": 0.00012515667722873464,
+      "loss": 0.6514,
+      "step": 1224
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.08401615172624588,
+      "learning_rate": 0.00012505170702547698,
+      "loss": 0.6334,
+      "step": 1225
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.09714555740356445,
+      "learning_rate": 0.00012494670736172444,
+      "loss": 0.7396,
+      "step": 1226
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.08669693768024445,
+      "learning_rate": 0.00012484167836095548,
+      "loss": 0.63,
+      "step": 1227
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.08610698580741882,
+      "learning_rate": 0.0001247366201466828,
+      "loss": 0.6939,
+      "step": 1228
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.08484649658203125,
+      "learning_rate": 0.00012463153284245358,
+      "loss": 0.5682,
+      "step": 1229
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.09587504714727402,
+      "learning_rate": 0.00012452641657184914,
+      "loss": 0.5824,
+      "step": 1230
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.0877947211265564,
+      "learning_rate": 0.00012442127145848494,
+      "loss": 0.6595,
+      "step": 1231
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.09148520231246948,
+      "learning_rate": 0.00012431609762601035,
+      "loss": 0.6649,
+      "step": 1232
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.08470945805311203,
+      "learning_rate": 0.0001242108951981084,
+      "loss": 0.6373,
+      "step": 1233
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.09003385901451111,
+      "learning_rate": 0.00012410566429849594,
+      "loss": 0.6727,
+      "step": 1234
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.0894625186920166,
+      "learning_rate": 0.00012400040505092306,
+      "loss": 0.6538,
+      "step": 1235
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.08713934570550919,
+      "learning_rate": 0.00012389511757917345,
+      "loss": 0.675,
+      "step": 1236
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.09119120985269547,
+      "learning_rate": 0.00012378980200706375,
+      "loss": 0.6835,
+      "step": 1237
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.1192459687590599,
+      "learning_rate": 0.0001236844584584439,
+      "loss": 0.6493,
+      "step": 1238
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.09404490888118744,
+      "learning_rate": 0.0001235790870571964,
+      "loss": 0.7077,
+      "step": 1239
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.08754786103963852,
+      "learning_rate": 0.00012347368792723693,
+      "loss": 0.7006,
+      "step": 1240
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.08642097562551498,
+      "learning_rate": 0.00012336826119251337,
+      "loss": 0.6489,
+      "step": 1241
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.08529900014400482,
+      "learning_rate": 0.0001232628069770063,
+      "loss": 0.624,
+      "step": 1242
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.09021829068660736,
+      "learning_rate": 0.00012315732540472865,
+      "loss": 0.6421,
+      "step": 1243
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.09515196084976196,
+      "learning_rate": 0.00012305181659972528,
+      "loss": 0.6257,
+      "step": 1244
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.09572126716375351,
+      "learning_rate": 0.00012294628068607337,
+      "loss": 0.648,
+      "step": 1245
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.09475268423557281,
+      "learning_rate": 0.0001228407177878817,
+      "loss": 0.5902,
+      "step": 1246
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.09514112025499344,
+      "learning_rate": 0.00012273512802929106,
+      "loss": 0.6196,
+      "step": 1247
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.09141046553850174,
+      "learning_rate": 0.00012262951153447355,
+      "loss": 0.6315,
+      "step": 1248
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.09266509115695953,
+      "learning_rate": 0.00012252386842763298,
+      "loss": 0.6672,
+      "step": 1249
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.08980372548103333,
+      "learning_rate": 0.00012241819883300424,
+      "loss": 0.6608,
+      "step": 1250
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.08771836012601852,
+      "learning_rate": 0.00012231250287485348,
+      "loss": 0.6598,
+      "step": 1251
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.08765268325805664,
+      "learning_rate": 0.00012220678067747785,
+      "loss": 0.664,
+      "step": 1252
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.08989165723323822,
+      "learning_rate": 0.00012210103236520524,
+      "loss": 0.6212,
+      "step": 1253
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.08498332649469376,
+      "learning_rate": 0.00012199525806239452,
+      "loss": 0.662,
+      "step": 1254
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.09044373035430908,
+      "learning_rate": 0.00012188945789343476,
+      "loss": 0.6522,
+      "step": 1255
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.08708745986223221,
+      "learning_rate": 0.00012178363198274578,
+      "loss": 0.6737,
+      "step": 1256
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.08722998946905136,
+      "learning_rate": 0.00012167778045477742,
+      "loss": 0.5989,
+      "step": 1257
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.0856277123093605,
+      "learning_rate": 0.00012157190343400989,
+      "loss": 0.573,
+      "step": 1258
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.08961885422468185,
+      "learning_rate": 0.0001214660010449531,
+      "loss": 0.6514,
+      "step": 1259
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.09075748920440674,
+      "learning_rate": 0.00012136007341214698,
+      "loss": 0.6519,
+      "step": 1260
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.0942176952958107,
+      "learning_rate": 0.0001212541206601612,
+      "loss": 0.6407,
+      "step": 1261
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.08544857800006866,
+      "learning_rate": 0.00012114814291359475,
+      "loss": 0.682,
+      "step": 1262
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.09113414585590363,
+      "learning_rate": 0.00012104214029707624,
+      "loss": 0.6324,
+      "step": 1263
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.086378313601017,
+      "learning_rate": 0.00012093611293526332,
+      "loss": 0.6566,
+      "step": 1264
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.08712925761938095,
+      "learning_rate": 0.00012083006095284295,
+      "loss": 0.6291,
+      "step": 1265
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.08669079840183258,
+      "learning_rate": 0.00012072398447453084,
+      "loss": 0.6849,
+      "step": 1266
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.08963727951049805,
+      "learning_rate": 0.00012061788362507167,
+      "loss": 0.691,
+      "step": 1267
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.08797255903482437,
+      "learning_rate": 0.00012051175852923866,
+      "loss": 0.6389,
+      "step": 1268
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.0952412411570549,
+      "learning_rate": 0.00012040560931183366,
+      "loss": 0.6361,
+      "step": 1269
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.08696448802947998,
+      "learning_rate": 0.00012029943609768679,
+      "loss": 0.6175,
+      "step": 1270
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.08767754584550858,
+      "learning_rate": 0.00012019323901165645,
+      "loss": 0.5771,
+      "step": 1271
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.08599507808685303,
+      "learning_rate": 0.00012008701817862906,
+      "loss": 0.6333,
+      "step": 1272
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.09141266345977783,
+      "learning_rate": 0.00011998077372351897,
+      "loss": 0.6368,
+      "step": 1273
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.0872783362865448,
+      "learning_rate": 0.00011987450577126842,
+      "loss": 0.57,
+      "step": 1274
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.0868949443101883,
+      "learning_rate": 0.00011976821444684709,
+      "loss": 0.6771,
+      "step": 1275
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.08809586614370346,
+      "learning_rate": 0.00011966189987525236,
+      "loss": 0.6872,
+      "step": 1276
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.0986771360039711,
+      "learning_rate": 0.00011955556218150872,
+      "loss": 0.5823,
+      "step": 1277
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.08900313824415207,
+      "learning_rate": 0.00011944920149066808,
+      "loss": 0.6385,
+      "step": 1278
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.09483526647090912,
+      "learning_rate": 0.00011934281792780919,
+      "loss": 0.6657,
+      "step": 1279
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.08786272257566452,
+      "learning_rate": 0.00011923641161803787,
+      "loss": 0.6542,
+      "step": 1280
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.08598706126213074,
+      "learning_rate": 0.00011912998268648659,
+      "loss": 0.6022,
+      "step": 1281
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.09137511253356934,
+      "learning_rate": 0.0001190235312583144,
+      "loss": 0.6297,
+      "step": 1282
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.08735531568527222,
+      "learning_rate": 0.00011891705745870694,
+      "loss": 0.675,
+      "step": 1283
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.09206798672676086,
+      "learning_rate": 0.00011881056141287597,
+      "loss": 0.7004,
+      "step": 1284
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.08716384321451187,
+      "learning_rate": 0.00011870404324605959,
+      "loss": 0.6388,
+      "step": 1285
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.08952728658914566,
+      "learning_rate": 0.00011859750308352176,
+      "loss": 0.6444,
+      "step": 1286
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.08607198297977448,
+      "learning_rate": 0.00011849094105055248,
+      "loss": 0.6152,
+      "step": 1287
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.09064583480358124,
+      "learning_rate": 0.00011838435727246726,
+      "loss": 0.63,
+      "step": 1288
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.0823214203119278,
+      "learning_rate": 0.00011827775187460737,
+      "loss": 0.6228,
+      "step": 1289
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.08422766625881195,
+      "learning_rate": 0.0001181711249823394,
+      "loss": 0.653,
+      "step": 1290
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.0890611931681633,
+      "learning_rate": 0.00011806447672105528,
+      "loss": 0.6319,
+      "step": 1291
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.0885709598660469,
+      "learning_rate": 0.00011795780721617198,
+      "loss": 0.6604,
+      "step": 1292
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.08804146200418472,
+      "learning_rate": 0.0001178511165931315,
+      "loss": 0.6348,
+      "step": 1293
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.08840359002351761,
+      "learning_rate": 0.00011774440497740078,
+      "loss": 0.6119,
+      "step": 1294
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.08792798966169357,
+      "learning_rate": 0.00011763767249447121,
+      "loss": 0.6241,
+      "step": 1295
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.09361138939857483,
+      "learning_rate": 0.00011753091926985897,
+      "loss": 0.6654,
+      "step": 1296
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.09208978712558746,
+      "learning_rate": 0.00011742414542910445,
+      "loss": 0.6393,
+      "step": 1297
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.08794102072715759,
+      "learning_rate": 0.00011731735109777237,
+      "loss": 0.688,
+      "step": 1298
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.09035321325063705,
+      "learning_rate": 0.00011721053640145156,
+      "loss": 0.6704,
+      "step": 1299
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.10449490696191788,
+      "learning_rate": 0.00011710370146575475,
+      "loss": 0.6889,
+      "step": 1300
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.08669156581163406,
+      "learning_rate": 0.00011699684641631849,
+      "loss": 0.5787,
+      "step": 1301
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.09480208158493042,
+      "learning_rate": 0.000116889971378803,
+      "loss": 0.6785,
+      "step": 1302
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.08715347200632095,
+      "learning_rate": 0.000116783076478892,
+      "loss": 0.6661,
+      "step": 1303
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.08632932603359222,
+      "learning_rate": 0.00011667616184229246,
+      "loss": 0.6046,
+      "step": 1304
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.09182171523571014,
+      "learning_rate": 0.00011656922759473481,
+      "loss": 0.6524,
+      "step": 1305
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.09081839770078659,
+      "learning_rate": 0.00011646227386197224,
+      "loss": 0.6229,
+      "step": 1306
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.09038412570953369,
+      "learning_rate": 0.00011635530076978114,
+      "loss": 0.7052,
+      "step": 1307
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.09159750491380692,
+      "learning_rate": 0.00011624830844396043,
+      "loss": 0.6417,
+      "step": 1308
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.09509099274873734,
+      "learning_rate": 0.00011614129701033179,
+      "loss": 0.6892,
+      "step": 1309
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.09086260944604874,
+      "learning_rate": 0.00011603426659473933,
+      "loss": 0.6754,
+      "step": 1310
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.08971850574016571,
+      "learning_rate": 0.00011592721732304949,
+      "loss": 0.591,
+      "step": 1311
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.09204638004302979,
+      "learning_rate": 0.00011582014932115089,
+      "loss": 0.6079,
+      "step": 1312
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09222222119569778,
+      "learning_rate": 0.00011571306271495413,
+      "loss": 0.637,
+      "step": 1313
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09187471121549606,
+      "learning_rate": 0.00011560595763039175,
+      "loss": 0.6292,
+      "step": 1314
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09650535136461258,
+      "learning_rate": 0.00011549883419341791,
+      "loss": 0.6317,
+      "step": 1315
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09161857515573502,
+      "learning_rate": 0.00011539169253000859,
+      "loss": 0.5598,
+      "step": 1316
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.0929383784532547,
+      "learning_rate": 0.00011528453276616089,
+      "loss": 0.678,
+      "step": 1317
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09310389310121536,
+      "learning_rate": 0.00011517735502789346,
+      "loss": 0.5582,
+      "step": 1318
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09871131181716919,
+      "learning_rate": 0.00011507015944124595,
+      "loss": 0.7079,
+      "step": 1319
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09423500299453735,
+      "learning_rate": 0.00011496294613227905,
+      "loss": 0.6574,
+      "step": 1320
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09618941694498062,
+      "learning_rate": 0.00011485571522707423,
+      "loss": 0.606,
+      "step": 1321
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.09104840457439423,
+      "learning_rate": 0.00011474846685173372,
+      "loss": 0.6633,
+      "step": 1322
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09393802285194397,
+      "learning_rate": 0.00011464120113238031,
+      "loss": 0.6461,
+      "step": 1323
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09360254555940628,
+      "learning_rate": 0.00011453391819515704,
+      "loss": 0.6616,
+      "step": 1324
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09270689636468887,
+      "learning_rate": 0.00011442661816622744,
+      "loss": 0.5967,
+      "step": 1325
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09051988273859024,
+      "learning_rate": 0.00011431930117177487,
+      "loss": 0.6577,
+      "step": 1326
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.0916256383061409,
+      "learning_rate": 0.00011421196733800291,
+      "loss": 0.6097,
+      "step": 1327
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09010876715183258,
+      "learning_rate": 0.0001141046167911347,
+      "loss": 0.7081,
+      "step": 1328
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09471412748098373,
+      "learning_rate": 0.0001139972496574132,
+      "loss": 0.6528,
+      "step": 1329
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.08696902543306351,
+      "learning_rate": 0.00011388986606310081,
+      "loss": 0.6168,
+      "step": 1330
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09542158991098404,
+      "learning_rate": 0.0001137824661344793,
+      "loss": 0.6189,
+      "step": 1331
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.09327239543199539,
+      "learning_rate": 0.00011367504999784964,
+      "loss": 0.6358,
+      "step": 1332
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.09684541821479797,
+      "learning_rate": 0.00011356761777953184,
+      "loss": 0.6912,
+      "step": 1333
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.10030949115753174,
+      "learning_rate": 0.00011346016960586489,
+      "loss": 0.661,
+      "step": 1334
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.09399015456438065,
+      "learning_rate": 0.00011335270560320647,
+      "loss": 0.6283,
+      "step": 1335
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.0928402990102768,
+      "learning_rate": 0.00011324522589793296,
+      "loss": 0.6763,
+      "step": 1336
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.0895305871963501,
+      "learning_rate": 0.00011313773061643904,
+      "loss": 0.6161,
+      "step": 1337
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.09813341498374939,
+      "learning_rate": 0.00011303021988513794,
+      "loss": 0.7036,
+      "step": 1338
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.09242972731590271,
+      "learning_rate": 0.00011292269383046089,
+      "loss": 0.6662,
+      "step": 1339
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.0893983542919159,
+      "learning_rate": 0.00011281515257885715,
+      "loss": 0.6323,
+      "step": 1340
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.09332004189491272,
+      "learning_rate": 0.00011270759625679397,
+      "loss": 0.6544,
+      "step": 1341
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.08976104855537415,
+      "learning_rate": 0.00011260002499075617,
+      "loss": 0.635,
+      "step": 1342
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09224797040224075,
+      "learning_rate": 0.00011249243890724627,
+      "loss": 0.6829,
+      "step": 1343
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09009308367967606,
+      "learning_rate": 0.00011238483813278408,
+      "loss": 0.7072,
+      "step": 1344
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09524769335985184,
+      "learning_rate": 0.00011227722279390691,
+      "loss": 0.6613,
+      "step": 1345
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09549695998430252,
+      "learning_rate": 0.00011216959301716889,
+      "loss": 0.6361,
+      "step": 1346
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09414596855640411,
+      "learning_rate": 0.00011206194892914141,
+      "loss": 0.6609,
+      "step": 1347
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09084893763065338,
+      "learning_rate": 0.00011195429065641252,
+      "loss": 0.6662,
+      "step": 1348
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09179358929395676,
+      "learning_rate": 0.00011184661832558705,
+      "loss": 0.7088,
+      "step": 1349
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09391826391220093,
+      "learning_rate": 0.00011173893206328627,
+      "loss": 0.668,
+      "step": 1350
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.09001519531011581,
+      "learning_rate": 0.00011163123199614791,
+      "loss": 0.6405,
+      "step": 1351
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09280743449926376,
+      "learning_rate": 0.00011152351825082588,
+      "loss": 0.6552,
+      "step": 1352
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09547816216945648,
+      "learning_rate": 0.00011141579095399021,
+      "loss": 0.6102,
+      "step": 1353
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09413521736860275,
+      "learning_rate": 0.00011130805023232691,
+      "loss": 0.636,
+      "step": 1354
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.0951075479388237,
+      "learning_rate": 0.00011120029621253763,
+      "loss": 0.6169,
+      "step": 1355
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09534990787506104,
+      "learning_rate": 0.0001110925290213398,
+      "loss": 0.6322,
+      "step": 1356
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09354977309703827,
+      "learning_rate": 0.0001109847487854663,
+      "loss": 0.6345,
+      "step": 1357
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09240519255399704,
+      "learning_rate": 0.00011087695563166534,
+      "loss": 0.6763,
+      "step": 1358
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09374047070741653,
+      "learning_rate": 0.00011076914968670029,
+      "loss": 0.6515,
+      "step": 1359
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09380830079317093,
+      "learning_rate": 0.00011066133107734963,
+      "loss": 0.7167,
+      "step": 1360
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.09072554856538773,
+      "learning_rate": 0.0001105534999304067,
+      "loss": 0.707,
+      "step": 1361
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.09862936288118362,
+      "learning_rate": 0.00011044565637267956,
+      "loss": 0.698,
+      "step": 1362
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.0963480994105339,
+      "learning_rate": 0.00011033780053099094,
+      "loss": 0.6728,
+      "step": 1363
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.08748151361942291,
+      "learning_rate": 0.00011022993253217787,
+      "loss": 0.6519,
+      "step": 1364
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.10116685181856155,
+      "learning_rate": 0.00011012205250309184,
+      "loss": 0.6559,
+      "step": 1365
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.08681276440620422,
+      "learning_rate": 0.00011001416057059841,
+      "loss": 0.6299,
+      "step": 1366
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.09113677591085434,
+      "learning_rate": 0.00010990625686157713,
+      "loss": 0.6868,
+      "step": 1367
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.08985843509435654,
+      "learning_rate": 0.00010979834150292144,
+      "loss": 0.6203,
+      "step": 1368
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.09041351079940796,
+      "learning_rate": 0.00010969041462153841,
+      "loss": 0.6151,
+      "step": 1369
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.09096385538578033,
+      "learning_rate": 0.00010958247634434875,
+      "loss": 0.7164,
+      "step": 1370
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.09355875849723816,
+      "learning_rate": 0.00010947452679828651,
+      "loss": 0.6537,
+      "step": 1371
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.09276916086673737,
+      "learning_rate": 0.000109366566110299,
+      "loss": 0.6733,
+      "step": 1372
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.09584525972604752,
+      "learning_rate": 0.00010925859440734667,
+      "loss": 0.671,
+      "step": 1373
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.0908685103058815,
+      "learning_rate": 0.00010915061181640289,
+      "loss": 0.6693,
+      "step": 1374
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.0912531241774559,
+      "learning_rate": 0.00010904261846445378,
+      "loss": 0.6169,
+      "step": 1375
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.0968153327703476,
+      "learning_rate": 0.00010893461447849825,
+      "loss": 0.658,
+      "step": 1376
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.09646685421466827,
+      "learning_rate": 0.00010882659998554758,
+      "loss": 0.6693,
+      "step": 1377
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.09004257619380951,
+      "learning_rate": 0.00010871857511262553,
+      "loss": 0.7048,
+      "step": 1378
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.08733712136745453,
+      "learning_rate": 0.00010861053998676794,
+      "loss": 0.6468,
+      "step": 1379
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.08630778640508652,
+      "learning_rate": 0.0001085024947350228,
+      "loss": 0.5791,
+      "step": 1380
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.098861463367939,
+      "learning_rate": 0.00010839443948444998,
+      "loss": 0.5865,
+      "step": 1381
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.08725135773420334,
+      "learning_rate": 0.0001082863743621211,
+      "loss": 0.6592,
+      "step": 1382
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.08900900930166245,
+      "learning_rate": 0.0001081782994951194,
+      "loss": 0.6866,
+      "step": 1383
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.09677062183618546,
+      "learning_rate": 0.00010807021501053954,
+      "loss": 0.6308,
+      "step": 1384
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.08833172917366028,
+      "learning_rate": 0.00010796212103548758,
+      "loss": 0.6261,
+      "step": 1385
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.0874096006155014,
+      "learning_rate": 0.00010785401769708066,
+      "loss": 0.6058,
+      "step": 1386
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.09676564484834671,
+      "learning_rate": 0.00010774590512244693,
+      "loss": 0.6907,
+      "step": 1387
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.09236234426498413,
+      "learning_rate": 0.00010763778343872549,
+      "loss": 0.5989,
+      "step": 1388
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.09421522915363312,
+      "learning_rate": 0.00010752965277306604,
+      "loss": 0.6793,
+      "step": 1389
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.09601663053035736,
+      "learning_rate": 0.0001074215132526289,
+      "loss": 0.6818,
+      "step": 1390
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.0974084734916687,
+      "learning_rate": 0.00010731336500458479,
+      "loss": 0.6542,
+      "step": 1391
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.09050207585096359,
+      "learning_rate": 0.00010720520815611474,
+      "loss": 0.6202,
+      "step": 1392
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.08748801052570343,
+      "learning_rate": 0.00010709704283440983,
+      "loss": 0.6336,
+      "step": 1393
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.09026212990283966,
+      "learning_rate": 0.00010698886916667113,
+      "loss": 0.6232,
+      "step": 1394
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.09674297273159027,
+      "learning_rate": 0.0001068806872801095,
+      "loss": 0.6088,
+      "step": 1395
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.09549888223409653,
+      "learning_rate": 0.00010677249730194551,
+      "loss": 0.6629,
+      "step": 1396
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.08924469351768494,
+      "learning_rate": 0.00010666429935940924,
+      "loss": 0.6783,
+      "step": 1397
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.09524770826101303,
+      "learning_rate": 0.0001065560935797401,
+      "loss": 0.6488,
+      "step": 1398
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.08813230693340302,
+      "learning_rate": 0.00010644788009018671,
+      "loss": 0.5909,
+      "step": 1399
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.09871229529380798,
+      "learning_rate": 0.00010633965901800685,
+      "loss": 0.6397,
+      "step": 1400
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.09156154096126556,
+      "learning_rate": 0.00010623143049046709,
+      "loss": 0.6349,
+      "step": 1401
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.08990387618541718,
+      "learning_rate": 0.00010612319463484286,
+      "loss": 0.6198,
+      "step": 1402
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.09997470676898956,
+      "learning_rate": 0.00010601495157841816,
+      "loss": 0.7134,
+      "step": 1403
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.0936041846871376,
+      "learning_rate": 0.00010590670144848546,
+      "loss": 0.6361,
+      "step": 1404
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.09657888859510422,
+      "learning_rate": 0.00010579844437234558,
+      "loss": 0.6524,
+      "step": 1405
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.09213721752166748,
+      "learning_rate": 0.00010569018047730752,
+      "loss": 0.6997,
+      "step": 1406
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.09132319688796997,
+      "learning_rate": 0.00010558190989068822,
+      "loss": 0.6547,
+      "step": 1407
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.0957031324505806,
+      "learning_rate": 0.00010547363273981257,
+      "loss": 0.6485,
+      "step": 1408
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.09112731367349625,
+      "learning_rate": 0.00010536534915201312,
+      "loss": 0.7171,
+      "step": 1409
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.10283418744802475,
+      "learning_rate": 0.00010525705925463004,
+      "loss": 0.695,
+      "step": 1410
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.08648620545864105,
+      "learning_rate": 0.0001051487631750109,
+      "loss": 0.6091,
+      "step": 1411
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.08666687458753586,
+      "learning_rate": 0.00010504046104051051,
+      "loss": 0.6033,
+      "step": 1412
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.09204310923814774,
+      "learning_rate": 0.00010493215297849085,
+      "loss": 0.6394,
+      "step": 1413
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.091493159532547,
+      "learning_rate": 0.00010482383911632086,
+      "loss": 0.6064,
+      "step": 1414
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.09357112646102905,
+      "learning_rate": 0.00010471551958137626,
+      "loss": 0.6272,
+      "step": 1415
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.0966016948223114,
+      "learning_rate": 0.00010460719450103951,
+      "loss": 0.6839,
+      "step": 1416
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.0946865901350975,
+      "learning_rate": 0.00010449886400269952,
+      "loss": 0.6571,
+      "step": 1417
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.08961882442235947,
+      "learning_rate": 0.00010439052821375161,
+      "loss": 0.5751,
+      "step": 1418
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.09724784642457962,
+      "learning_rate": 0.00010428218726159733,
+      "loss": 0.7059,
+      "step": 1419
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.09249065816402435,
+      "learning_rate": 0.00010417384127364427,
+      "loss": 0.5639,
+      "step": 1420
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.09107667952775955,
+      "learning_rate": 0.00010406549037730597,
+      "loss": 0.6601,
+      "step": 1421
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.09111786633729935,
+      "learning_rate": 0.00010395713470000173,
+      "loss": 0.6667,
+      "step": 1422
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.09122546762228012,
+      "learning_rate": 0.00010384877436915646,
+      "loss": 0.6716,
+      "step": 1423
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.09497132152318954,
+      "learning_rate": 0.00010374040951220059,
+      "loss": 0.7077,
+      "step": 1424
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.09761511534452438,
+      "learning_rate": 0.0001036320402565698,
+      "loss": 0.6634,
+      "step": 1425
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.09016017615795135,
+      "learning_rate": 0.00010352366672970499,
+      "loss": 0.6065,
+      "step": 1426
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.10208307951688766,
+      "learning_rate": 0.00010341528905905211,
+      "loss": 0.5917,
+      "step": 1427
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.0971704050898552,
+      "learning_rate": 0.00010330690737206192,
+      "loss": 0.6092,
+      "step": 1428
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.09399619698524475,
+      "learning_rate": 0.00010319852179618994,
+      "loss": 0.7323,
+      "step": 1429
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.09294161200523376,
+      "learning_rate": 0.00010309013245889623,
+      "loss": 0.6756,
+      "step": 1430
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.09187652170658112,
+      "learning_rate": 0.00010298173948764534,
+      "loss": 0.5659,
+      "step": 1431
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.09526228159666061,
+      "learning_rate": 0.000102873343009906,
+      "loss": 0.6207,
+      "step": 1432
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.0946209579706192,
+      "learning_rate": 0.00010276494315315117,
+      "loss": 0.6189,
+      "step": 1433
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.09191188961267471,
+      "learning_rate": 0.0001026565400448577,
+      "loss": 0.6445,
+      "step": 1434
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.0986436977982521,
+      "learning_rate": 0.00010254813381250628,
+      "loss": 0.6626,
+      "step": 1435
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.0932629257440567,
+      "learning_rate": 0.00010243972458358129,
+      "loss": 0.6448,
+      "step": 1436
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.09336506575345993,
+      "learning_rate": 0.00010233131248557066,
+      "loss": 0.6599,
+      "step": 1437
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.0949806496500969,
+      "learning_rate": 0.00010222289764596564,
+      "loss": 0.6024,
+      "step": 1438
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.087349072098732,
+      "learning_rate": 0.00010211448019226073,
+      "loss": 0.5655,
+      "step": 1439
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.09715444594621658,
+      "learning_rate": 0.00010200606025195352,
+      "loss": 0.6225,
+      "step": 1440
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.10039708018302917,
+      "learning_rate": 0.00010189763795254447,
+      "loss": 0.6756,
+      "step": 1441
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.10806684195995331,
+      "learning_rate": 0.00010178921342153688,
+      "loss": 0.6251,
+      "step": 1442
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.10099483281373978,
+      "learning_rate": 0.00010168078678643665,
+      "loss": 0.638,
+      "step": 1443
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.102535180747509,
+      "learning_rate": 0.00010157235817475213,
+      "loss": 0.6353,
+      "step": 1444
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.090107262134552,
+      "learning_rate": 0.00010146392771399402,
+      "loss": 0.5738,
+      "step": 1445
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.09371799975633621,
+      "learning_rate": 0.00010135549553167518,
+      "loss": 0.5767,
+      "step": 1446
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.09494274854660034,
+      "learning_rate": 0.00010124706175531053,
+      "loss": 0.7031,
+      "step": 1447
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.09525119513273239,
+      "learning_rate": 0.00010113862651241681,
+      "loss": 0.6251,
+      "step": 1448
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09051448851823807,
+      "learning_rate": 0.00010103018993051252,
+      "loss": 0.6045,
+      "step": 1449
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09272157400846481,
+      "learning_rate": 0.00010092175213711774,
+      "loss": 0.6781,
+      "step": 1450
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.0994546040892601,
+      "learning_rate": 0.00010081331325975396,
+      "loss": 0.5494,
+      "step": 1451
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09096287935972214,
+      "learning_rate": 0.00010070487342594392,
+      "loss": 0.6451,
+      "step": 1452
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09607745707035065,
+      "learning_rate": 0.00010059643276321153,
+      "loss": 0.6082,
+      "step": 1453
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09483859688043594,
+      "learning_rate": 0.00010048799139908165,
+      "loss": 0.6558,
+      "step": 1454
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09446833282709122,
+      "learning_rate": 0.00010037954946108,
+      "loss": 0.6186,
+      "step": 1455
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.0962926596403122,
+      "learning_rate": 0.0001002711070767329,
+      "loss": 0.6494,
+      "step": 1456
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09676545858383179,
+      "learning_rate": 0.00010016266437356728,
+      "loss": 0.6162,
+      "step": 1457
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.09890628606081009,
+      "learning_rate": 0.00010005422147911036,
+      "loss": 0.6826,
+      "step": 1458
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.8372865319252014,
+      "eval_runtime": 111.9226,
+      "eval_samples_per_second": 88.793,
+      "eval_steps_per_second": 22.203,
+      "step": 1458
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09064832329750061,
+      "learning_rate": 9.994577852088966e-05,
+      "loss": 0.6334,
+      "step": 1459
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09355921298265457,
+      "learning_rate": 9.983733562643273e-05,
+      "loss": 0.6412,
+      "step": 1460
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09292005747556686,
+      "learning_rate": 9.972889292326712e-05,
+      "loss": 0.609,
+      "step": 1461
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09495711326599121,
+      "learning_rate": 9.962045053892003e-05,
+      "loss": 0.6507,
+      "step": 1462
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09616797417402267,
+      "learning_rate": 9.951200860091836e-05,
+      "loss": 0.6625,
+      "step": 1463
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09481047093868256,
+      "learning_rate": 9.940356723678852e-05,
+      "loss": 0.7114,
+      "step": 1464
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.08921283483505249,
+      "learning_rate": 9.92951265740561e-05,
+      "loss": 0.6313,
+      "step": 1465
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09051833301782608,
+      "learning_rate": 9.918668674024607e-05,
+      "loss": 0.5885,
+      "step": 1466
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.09668543934822083,
+      "learning_rate": 9.907824786288225e-05,
+      "loss": 0.6398,
+      "step": 1467
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09285859018564224,
+      "learning_rate": 9.896981006948752e-05,
+      "loss": 0.7246,
+      "step": 1468
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09071426838636398,
+      "learning_rate": 9.886137348758321e-05,
+      "loss": 0.6584,
+      "step": 1469
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.0932983011007309,
+      "learning_rate": 9.87529382446895e-05,
+      "loss": 0.6761,
+      "step": 1470
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09152944386005402,
+      "learning_rate": 9.864450446832482e-05,
+      "loss": 0.6771,
+      "step": 1471
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09786435961723328,
+      "learning_rate": 9.8536072286006e-05,
+      "loss": 0.6662,
+      "step": 1472
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09742102026939392,
+      "learning_rate": 9.84276418252479e-05,
+      "loss": 0.6739,
+      "step": 1473
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09501685202121735,
+      "learning_rate": 9.831921321356337e-05,
+      "loss": 0.7016,
+      "step": 1474
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09399169683456421,
+      "learning_rate": 9.821078657846315e-05,
+      "loss": 0.6221,
+      "step": 1475
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09704238921403885,
+      "learning_rate": 9.810236204745554e-05,
+      "loss": 0.561,
+      "step": 1476
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.09105955064296722,
+      "learning_rate": 9.79939397480465e-05,
+      "loss": 0.6293,
+      "step": 1477
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.09872641414403915,
+      "learning_rate": 9.788551980773927e-05,
+      "loss": 0.6017,
+      "step": 1478
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.1039251834154129,
+      "learning_rate": 9.77771023540344e-05,
+      "loss": 0.7524,
+      "step": 1479
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.09600313752889633,
+      "learning_rate": 9.766868751442935e-05,
+      "loss": 0.5687,
+      "step": 1480
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.0960647240281105,
+      "learning_rate": 9.756027541641872e-05,
+      "loss": 0.5441,
+      "step": 1481
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.09576267004013062,
+      "learning_rate": 9.745186618749372e-05,
+      "loss": 0.676,
+      "step": 1482
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.09594549983739853,
+      "learning_rate": 9.734345995514234e-05,
+      "loss": 0.707,
+      "step": 1483
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.09687156975269318,
+      "learning_rate": 9.723505684684886e-05,
+      "loss": 0.6456,
+      "step": 1484
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.09854575991630554,
+      "learning_rate": 9.7126656990094e-05,
+      "loss": 0.6885,
+      "step": 1485
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.08853231370449066,
+      "learning_rate": 9.70182605123547e-05,
+      "loss": 0.618,
+      "step": 1486
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.09728317707777023,
+      "learning_rate": 9.690986754110378e-05,
+      "loss": 0.6684,
+      "step": 1487
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.09424198418855667,
+      "learning_rate": 9.68014782038101e-05,
+      "loss": 0.6228,
+      "step": 1488
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.10386296361684799,
+      "learning_rate": 9.669309262793808e-05,
+      "loss": 0.6592,
+      "step": 1489
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.09128458797931671,
+      "learning_rate": 9.658471094094791e-05,
+      "loss": 0.6857,
+      "step": 1490
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.09904751926660538,
+      "learning_rate": 9.647633327029502e-05,
+      "loss": 0.6183,
+      "step": 1491
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.09605219215154648,
+      "learning_rate": 9.636795974343023e-05,
+      "loss": 0.6413,
+      "step": 1492
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.10036271810531616,
+      "learning_rate": 9.625959048779942e-05,
+      "loss": 0.6515,
+      "step": 1493
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.10257143527269363,
+      "learning_rate": 9.615122563084355e-05,
+      "loss": 0.6684,
+      "step": 1494
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.0906180739402771,
+      "learning_rate": 9.604286529999829e-05,
+      "loss": 0.6137,
+      "step": 1495
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.09586507081985474,
+      "learning_rate": 9.593450962269404e-05,
+      "loss": 0.5627,
+      "step": 1496
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.09247685968875885,
+      "learning_rate": 9.582615872635578e-05,
+      "loss": 0.6114,
+      "step": 1497
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.09619525074958801,
+      "learning_rate": 9.571781273840269e-05,
+      "loss": 0.6915,
+      "step": 1498
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.09084618836641312,
+      "learning_rate": 9.56094717862484e-05,
+      "loss": 0.5918,
+      "step": 1499
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.09566932171583176,
+      "learning_rate": 9.550113599730049e-05,
+      "loss": 0.66,
+      "step": 1500
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.10314442962408066,
+      "learning_rate": 9.539280549896051e-05,
+      "loss": 0.665,
+      "step": 1501
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.10305890440940857,
+      "learning_rate": 9.528448041862375e-05,
+      "loss": 0.6872,
+      "step": 1502
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.09835629910230637,
+      "learning_rate": 9.517616088367915e-05,
+      "loss": 0.6226,
+      "step": 1503
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.10013748705387115,
+      "learning_rate": 9.506784702150918e-05,
+      "loss": 0.6728,
+      "step": 1504
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.0961749330163002,
+      "learning_rate": 9.495953895948952e-05,
+      "loss": 0.6466,
+      "step": 1505
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.08960247784852982,
+      "learning_rate": 9.485123682498913e-05,
+      "loss": 0.618,
+      "step": 1506
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.08642073720693588,
+      "learning_rate": 9.474294074536996e-05,
+      "loss": 0.5811,
+      "step": 1507
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.09075295925140381,
+      "learning_rate": 9.463465084798693e-05,
+      "loss": 0.6572,
+      "step": 1508
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.09943678230047226,
+      "learning_rate": 9.452636726018745e-05,
+      "loss": 0.7361,
+      "step": 1509
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.0944795086979866,
+      "learning_rate": 9.441809010931179e-05,
+      "loss": 0.7083,
+      "step": 1510
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.10334914922714233,
+      "learning_rate": 9.430981952269247e-05,
+      "loss": 0.7174,
+      "step": 1511
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.09473937004804611,
+      "learning_rate": 9.420155562765443e-05,
+      "loss": 0.6937,
+      "step": 1512
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.09412137418985367,
+      "learning_rate": 9.409329855151456e-05,
+      "loss": 0.6839,
+      "step": 1513
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.09418198466300964,
+      "learning_rate": 9.398504842158185e-05,
+      "loss": 0.6693,
+      "step": 1514
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.09520245343446732,
+      "learning_rate": 9.38768053651572e-05,
+      "loss": 0.6691,
+      "step": 1515
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.0990833193063736,
+      "learning_rate": 9.376856950953293e-05,
+      "loss": 0.6744,
+      "step": 1516
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.0982571691274643,
+      "learning_rate": 9.366034098199318e-05,
+      "loss": 0.6269,
+      "step": 1517
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.09549887478351593,
+      "learning_rate": 9.355211990981327e-05,
+      "loss": 0.6306,
+      "step": 1518
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.08750772476196289,
+      "learning_rate": 9.344390642025993e-05,
+      "loss": 0.6336,
+      "step": 1519
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.10237884521484375,
+      "learning_rate": 9.333570064059078e-05,
+      "loss": 0.6304,
+      "step": 1520
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.09283220767974854,
+      "learning_rate": 9.322750269805451e-05,
+      "loss": 0.5755,
+      "step": 1521
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.09829011559486389,
+      "learning_rate": 9.31193127198905e-05,
+      "loss": 0.6247,
+      "step": 1522
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.10395735502243042,
+      "learning_rate": 9.301113083332889e-05,
+      "loss": 0.6855,
+      "step": 1523
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.09860970079898834,
+      "learning_rate": 9.290295716559018e-05,
+      "loss": 0.6444,
+      "step": 1524
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.10212879627943039,
+      "learning_rate": 9.279479184388525e-05,
+      "loss": 0.6335,
+      "step": 1525
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.0931490957736969,
+      "learning_rate": 9.268663499541524e-05,
+      "loss": 0.6548,
+      "step": 1526
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.09232822060585022,
+      "learning_rate": 9.257848674737114e-05,
+      "loss": 0.6805,
+      "step": 1527
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.09733033925294876,
+      "learning_rate": 9.2470347226934e-05,
+      "loss": 0.5705,
+      "step": 1528
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.09804414212703705,
+      "learning_rate": 9.236221656127452e-05,
+      "loss": 0.6082,
+      "step": 1529
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.10374914854764938,
+      "learning_rate": 9.225409487755308e-05,
+      "loss": 0.6432,
+      "step": 1530
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.09214441478252411,
+      "learning_rate": 9.214598230291938e-05,
+      "loss": 0.5682,
+      "step": 1531
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.09109389781951904,
+      "learning_rate": 9.203787896451245e-05,
+      "loss": 0.7019,
+      "step": 1532
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.09263739734888077,
+      "learning_rate": 9.192978498946046e-05,
+      "loss": 0.6476,
+      "step": 1533
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.09582334011793137,
+      "learning_rate": 9.182170050488064e-05,
+      "loss": 0.6477,
+      "step": 1534
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.1023733913898468,
+      "learning_rate": 9.171362563787893e-05,
+      "loss": 0.7021,
+      "step": 1535
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09692607820034027,
+      "learning_rate": 9.160556051555003e-05,
+      "loss": 0.6299,
+      "step": 1536
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09542369842529297,
+      "learning_rate": 9.149750526497725e-05,
+      "loss": 0.6271,
+      "step": 1537
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09215032309293747,
+      "learning_rate": 9.138946001323208e-05,
+      "loss": 0.6061,
+      "step": 1538
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.10323350131511688,
+      "learning_rate": 9.12814248873745e-05,
+      "loss": 0.7484,
+      "step": 1539
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09550279378890991,
+      "learning_rate": 9.117340001445242e-05,
+      "loss": 0.6505,
+      "step": 1540
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09404686093330383,
+      "learning_rate": 9.106538552150178e-05,
+      "loss": 0.632,
+      "step": 1541
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.0932483971118927,
+      "learning_rate": 9.095738153554624e-05,
+      "loss": 0.6026,
+      "step": 1542
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09166806936264038,
+      "learning_rate": 9.084938818359713e-05,
+      "loss": 0.6579,
+      "step": 1543
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09238806366920471,
+      "learning_rate": 9.074140559265333e-05,
+      "loss": 0.6637,
+      "step": 1544
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.09482090920209885,
+      "learning_rate": 9.063343388970101e-05,
+      "loss": 0.6356,
+      "step": 1545
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.11874199658632278,
+      "learning_rate": 9.052547320171351e-05,
+      "loss": 0.6153,
+      "step": 1546
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.09669456630945206,
+      "learning_rate": 9.041752365565124e-05,
+      "loss": 0.6105,
+      "step": 1547
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.0983959510922432,
+      "learning_rate": 9.03095853784616e-05,
+      "loss": 0.5967,
+      "step": 1548
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.09383951127529144,
+      "learning_rate": 9.020165849707859e-05,
+      "loss": 0.6831,
+      "step": 1549
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.10585930943489075,
+      "learning_rate": 9.009374313842289e-05,
+      "loss": 0.6989,
+      "step": 1550
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.10016348212957382,
+      "learning_rate": 8.99858394294016e-05,
+      "loss": 0.6561,
+      "step": 1551
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.08909649401903152,
+      "learning_rate": 8.987794749690818e-05,
+      "loss": 0.6056,
+      "step": 1552
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.09735802561044693,
+      "learning_rate": 8.977006746782216e-05,
+      "loss": 0.6569,
+      "step": 1553
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.09465418010950089,
+      "learning_rate": 8.96621994690091e-05,
+      "loss": 0.6311,
+      "step": 1554
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.09621430933475494,
+      "learning_rate": 8.955434362732048e-05,
+      "loss": 0.6245,
+      "step": 1555
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.09657291322946548,
+      "learning_rate": 8.944650006959332e-05,
+      "loss": 0.692,
+      "step": 1556
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.09351187199354172,
+      "learning_rate": 8.933866892265039e-05,
+      "loss": 0.7121,
+      "step": 1557
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.09253359586000443,
+      "learning_rate": 8.923085031329971e-05,
+      "loss": 0.5976,
+      "step": 1558
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.11010093241930008,
+      "learning_rate": 8.91230443683347e-05,
+      "loss": 0.6664,
+      "step": 1559
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.10193350166082382,
+      "learning_rate": 8.901525121453373e-05,
+      "loss": 0.652,
+      "step": 1560
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.094691202044487,
+      "learning_rate": 8.890747097866023e-05,
+      "loss": 0.6666,
+      "step": 1561
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.09835696220397949,
+      "learning_rate": 8.879970378746238e-05,
+      "loss": 0.6804,
+      "step": 1562
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.09191767126321793,
+      "learning_rate": 8.869194976767313e-05,
+      "loss": 0.6396,
+      "step": 1563
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.09501973539590836,
+      "learning_rate": 8.85842090460098e-05,
+      "loss": 0.6044,
+      "step": 1564
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09784456342458725,
+      "learning_rate": 8.847648174917412e-05,
+      "loss": 0.637,
+      "step": 1565
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09819415956735611,
+      "learning_rate": 8.836876800385214e-05,
+      "loss": 0.6483,
+      "step": 1566
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09621243923902512,
+      "learning_rate": 8.826106793671376e-05,
+      "loss": 0.6621,
+      "step": 1567
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09610548615455627,
+      "learning_rate": 8.815338167441299e-05,
+      "loss": 0.6157,
+      "step": 1568
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.0955331027507782,
+      "learning_rate": 8.804570934358746e-05,
+      "loss": 0.5699,
+      "step": 1569
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09951046109199524,
+      "learning_rate": 8.793805107085861e-05,
+      "loss": 0.655,
+      "step": 1570
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09489110857248306,
+      "learning_rate": 8.783040698283112e-05,
+      "loss": 0.6441,
+      "step": 1571
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09634937345981598,
+      "learning_rate": 8.772277720609311e-05,
+      "loss": 0.6661,
+      "step": 1572
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.10194681584835052,
+      "learning_rate": 8.76151618672159e-05,
+      "loss": 0.7166,
+      "step": 1573
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.09606891125440598,
+      "learning_rate": 8.750756109275378e-05,
+      "loss": 0.6523,
+      "step": 1574
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.08936757594347,
+      "learning_rate": 8.739997500924386e-05,
+      "loss": 0.6364,
+      "step": 1575
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.08927170187234879,
+      "learning_rate": 8.729240374320604e-05,
+      "loss": 0.5381,
+      "step": 1576
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09218922257423401,
+      "learning_rate": 8.718484742114286e-05,
+      "loss": 0.6334,
+      "step": 1577
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09550817310810089,
+      "learning_rate": 8.707730616953915e-05,
+      "loss": 0.6351,
+      "step": 1578
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09181966632604599,
+      "learning_rate": 8.696978011486209e-05,
+      "loss": 0.6816,
+      "step": 1579
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09721545875072479,
+      "learning_rate": 8.686226938356095e-05,
+      "loss": 0.6784,
+      "step": 1580
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09176433086395264,
+      "learning_rate": 8.675477410206709e-05,
+      "loss": 0.5591,
+      "step": 1581
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09666182845830917,
+      "learning_rate": 8.664729439679355e-05,
+      "loss": 0.6025,
+      "step": 1582
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09324358403682709,
+      "learning_rate": 8.653983039413512e-05,
+      "loss": 0.6004,
+      "step": 1583
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.09725426137447357,
+      "learning_rate": 8.643238222046815e-05,
+      "loss": 0.67,
+      "step": 1584
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.09565319865942001,
+      "learning_rate": 8.63249500021504e-05,
+      "loss": 0.6668,
+      "step": 1585
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.10329779982566833,
+      "learning_rate": 8.621753386552072e-05,
+      "loss": 0.6298,
+      "step": 1586
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.0949530079960823,
+      "learning_rate": 8.611013393689918e-05,
+      "loss": 0.6593,
+      "step": 1587
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.10827862471342087,
+      "learning_rate": 8.600275034258682e-05,
+      "loss": 0.7111,
+      "step": 1588
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.09596540778875351,
+      "learning_rate": 8.589538320886532e-05,
+      "loss": 0.6414,
+      "step": 1589
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.09415283054113388,
+      "learning_rate": 8.578803266199712e-05,
+      "loss": 0.6095,
+      "step": 1590
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.09853770583868027,
+      "learning_rate": 8.568069882822511e-05,
+      "loss": 0.6357,
+      "step": 1591
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.09898513555526733,
+      "learning_rate": 8.55733818337726e-05,
+      "loss": 0.682,
+      "step": 1592
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.0918831080198288,
+      "learning_rate": 8.546608180484298e-05,
+      "loss": 0.5839,
+      "step": 1593
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.09902951121330261,
+      "learning_rate": 8.535879886761971e-05,
+      "loss": 0.6987,
+      "step": 1594
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.10585609823465347,
+      "learning_rate": 8.52515331482663e-05,
+      "loss": 0.6835,
+      "step": 1595
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.09927651286125183,
+      "learning_rate": 8.514428477292579e-05,
+      "loss": 0.6576,
+      "step": 1596
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.09774427115917206,
+      "learning_rate": 8.503705386772099e-05,
+      "loss": 0.658,
+      "step": 1597
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.0890786275267601,
+      "learning_rate": 8.492984055875404e-05,
+      "loss": 0.6024,
+      "step": 1598
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.09146375209093094,
+      "learning_rate": 8.482264497210655e-05,
+      "loss": 0.6192,
+      "step": 1599
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.09767889231443405,
+      "learning_rate": 8.471546723383912e-05,
+      "loss": 0.6899,
+      "step": 1600
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.08738665282726288,
+      "learning_rate": 8.460830746999143e-05,
+      "loss": 0.6845,
+      "step": 1601
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.09772787243127823,
+      "learning_rate": 8.450116580658207e-05,
+      "loss": 0.6081,
+      "step": 1602
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.0991789773106575,
+      "learning_rate": 8.43940423696083e-05,
+      "loss": 0.7227,
+      "step": 1603
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.09454263001680374,
+      "learning_rate": 8.428693728504591e-05,
+      "loss": 0.6521,
+      "step": 1604
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.09202510118484497,
+      "learning_rate": 8.417985067884912e-05,
+      "loss": 0.5679,
+      "step": 1605
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.08857171982526779,
+      "learning_rate": 8.407278267695052e-05,
+      "loss": 0.5613,
+      "step": 1606
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.0959198921918869,
+      "learning_rate": 8.396573340526068e-05,
+      "loss": 0.6193,
+      "step": 1607
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.09435534477233887,
+      "learning_rate": 8.385870298966824e-05,
+      "loss": 0.631,
+      "step": 1608
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.0935453400015831,
+      "learning_rate": 8.375169155603958e-05,
+      "loss": 0.6126,
+      "step": 1609
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.09247604757547379,
+      "learning_rate": 8.36446992302189e-05,
+      "loss": 0.7089,
+      "step": 1610
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.09178217500448227,
+      "learning_rate": 8.353772613802777e-05,
+      "loss": 0.5758,
+      "step": 1611
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.09289277344942093,
+      "learning_rate": 8.343077240526523e-05,
+      "loss": 0.5971,
+      "step": 1612
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.10544314235448837,
+      "learning_rate": 8.332383815770753e-05,
+      "loss": 0.6718,
+      "step": 1613
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.09784182161092758,
+      "learning_rate": 8.321692352110805e-05,
+      "loss": 0.6519,
+      "step": 1614
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.09386929869651794,
+      "learning_rate": 8.311002862119703e-05,
+      "loss": 0.5884,
+      "step": 1615
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.0988815575838089,
+      "learning_rate": 8.300315358368149e-05,
+      "loss": 0.6322,
+      "step": 1616
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.0945940837264061,
+      "learning_rate": 8.289629853424527e-05,
+      "loss": 0.6483,
+      "step": 1617
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.10307436436414719,
+      "learning_rate": 8.278946359854845e-05,
+      "loss": 0.6825,
+      "step": 1618
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.09305360913276672,
+      "learning_rate": 8.268264890222764e-05,
+      "loss": 0.6556,
+      "step": 1619
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.10257001966238022,
+      "learning_rate": 8.257585457089556e-05,
+      "loss": 0.7531,
+      "step": 1620
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.09230564534664154,
+      "learning_rate": 8.246908073014106e-05,
+      "loss": 0.6461,
+      "step": 1621
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.10027793794870377,
+      "learning_rate": 8.236232750552881e-05,
+      "loss": 0.5969,
+      "step": 1622
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09907641261816025,
+      "learning_rate": 8.225559502259926e-05,
+      "loss": 0.6862,
+      "step": 1623
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09151662141084671,
+      "learning_rate": 8.214888340686848e-05,
+      "loss": 0.5988,
+      "step": 1624
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.10143401473760605,
+      "learning_rate": 8.204219278382806e-05,
+      "loss": 0.6927,
+      "step": 1625
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09213609248399734,
+      "learning_rate": 8.193552327894476e-05,
+      "loss": 0.6694,
+      "step": 1626
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09753984957933426,
+      "learning_rate": 8.182887501766059e-05,
+      "loss": 0.6117,
+      "step": 1627
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09462133795022964,
+      "learning_rate": 8.172224812539264e-05,
+      "loss": 0.6449,
+      "step": 1628
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09718816727399826,
+      "learning_rate": 8.161564272753276e-05,
+      "loss": 0.6256,
+      "step": 1629
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09913357347249985,
+      "learning_rate": 8.150905894944753e-05,
+      "loss": 0.6192,
+      "step": 1630
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09483817219734192,
+      "learning_rate": 8.140249691647824e-05,
+      "loss": 0.6338,
+      "step": 1631
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.09601510316133499,
+      "learning_rate": 8.129595675394045e-05,
+      "loss": 0.6594,
+      "step": 1632
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09573660045862198,
+      "learning_rate": 8.118943858712405e-05,
+      "loss": 0.674,
+      "step": 1633
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09435724467039108,
+      "learning_rate": 8.108294254129308e-05,
+      "loss": 0.6413,
+      "step": 1634
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09103036671876907,
+      "learning_rate": 8.09764687416856e-05,
+      "loss": 0.6055,
+      "step": 1635
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09245546907186508,
+      "learning_rate": 8.087001731351345e-05,
+      "loss": 0.6626,
+      "step": 1636
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09607689082622528,
+      "learning_rate": 8.076358838196216e-05,
+      "loss": 0.6426,
+      "step": 1637
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09229609370231628,
+      "learning_rate": 8.06571820721908e-05,
+      "loss": 0.6256,
+      "step": 1638
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09131552278995514,
+      "learning_rate": 8.055079850933197e-05,
+      "loss": 0.5735,
+      "step": 1639
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.0979132354259491,
+      "learning_rate": 8.04444378184913e-05,
+      "loss": 0.6276,
+      "step": 1640
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.09681771695613861,
+      "learning_rate": 8.033810012474767e-05,
+      "loss": 0.5942,
+      "step": 1641
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.0946432426571846,
+      "learning_rate": 8.02317855531529e-05,
+      "loss": 0.6477,
+      "step": 1642
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.09734468162059784,
+      "learning_rate": 8.012549422873163e-05,
+      "loss": 0.6561,
+      "step": 1643
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.10328742116689682,
+      "learning_rate": 8.001922627648104e-05,
+      "loss": 0.6894,
+      "step": 1644
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.0990242213010788,
+      "learning_rate": 7.991298182137095e-05,
+      "loss": 0.6356,
+      "step": 1645
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.09674172103404999,
+      "learning_rate": 7.980676098834359e-05,
+      "loss": 0.6138,
+      "step": 1646
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.10231846570968628,
+      "learning_rate": 7.970056390231323e-05,
+      "loss": 0.6936,
+      "step": 1647
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.0951048955321312,
+      "learning_rate": 7.959439068816636e-05,
+      "loss": 0.5863,
+      "step": 1648
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.0990295261144638,
+      "learning_rate": 7.948824147076134e-05,
+      "loss": 0.641,
+      "step": 1649
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.10110025107860565,
+      "learning_rate": 7.938211637492838e-05,
+      "loss": 0.7049,
+      "step": 1650
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.09786627441644669,
+      "learning_rate": 7.92760155254692e-05,
+      "loss": 0.713,
+      "step": 1651
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.0944487527012825,
+      "learning_rate": 7.916993904715708e-05,
+      "loss": 0.6201,
+      "step": 1652
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.09228896349668503,
+      "learning_rate": 7.906388706473667e-05,
+      "loss": 0.6371,
+      "step": 1653
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.09492463618516922,
+      "learning_rate": 7.89578597029238e-05,
+      "loss": 0.657,
+      "step": 1654
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.10092806071043015,
+      "learning_rate": 7.885185708640526e-05,
+      "loss": 0.6896,
+      "step": 1655
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.09815175086259842,
+      "learning_rate": 7.874587933983881e-05,
+      "loss": 0.7098,
+      "step": 1656
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.09346696734428406,
+      "learning_rate": 7.863992658785303e-05,
+      "loss": 0.6755,
+      "step": 1657
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.09460650384426117,
+      "learning_rate": 7.853399895504694e-05,
+      "loss": 0.649,
+      "step": 1658
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.10080888867378235,
+      "learning_rate": 7.842809656599015e-05,
+      "loss": 0.6447,
+      "step": 1659
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.08859395235776901,
+      "learning_rate": 7.832221954522258e-05,
+      "loss": 0.6333,
+      "step": 1660
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.09535930305719376,
+      "learning_rate": 7.821636801725427e-05,
+      "loss": 0.6791,
+      "step": 1661
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.0975249856710434,
+      "learning_rate": 7.811054210656527e-05,
+      "loss": 0.699,
+      "step": 1662
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09512438625097275,
+      "learning_rate": 7.800474193760552e-05,
+      "loss": 0.6563,
+      "step": 1663
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09686731547117233,
+      "learning_rate": 7.789896763479474e-05,
+      "loss": 0.6331,
+      "step": 1664
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09903857856988907,
+      "learning_rate": 7.77932193225222e-05,
+      "loss": 0.6666,
+      "step": 1665
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09601344168186188,
+      "learning_rate": 7.768749712514655e-05,
+      "loss": 0.582,
+      "step": 1666
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09814467281103134,
+      "learning_rate": 7.758180116699577e-05,
+      "loss": 0.6777,
+      "step": 1667
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09757161140441895,
+      "learning_rate": 7.747613157236704e-05,
+      "loss": 0.6892,
+      "step": 1668
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09715452790260315,
+      "learning_rate": 7.737048846552646e-05,
+      "loss": 0.6232,
+      "step": 1669
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09860831499099731,
+      "learning_rate": 7.726487197070897e-05,
+      "loss": 0.6312,
+      "step": 1670
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.09930217266082764,
+      "learning_rate": 7.71592822121183e-05,
+      "loss": 0.6023,
+      "step": 1671
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09746716916561127,
+      "learning_rate": 7.705371931392668e-05,
+      "loss": 0.7145,
+      "step": 1672
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.0996137335896492,
+      "learning_rate": 7.694818340027474e-05,
+      "loss": 0.6294,
+      "step": 1673
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09783168882131577,
+      "learning_rate": 7.684267459527137e-05,
+      "loss": 0.6166,
+      "step": 1674
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09177543222904205,
+      "learning_rate": 7.673719302299369e-05,
+      "loss": 0.668,
+      "step": 1675
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09733571112155914,
+      "learning_rate": 7.663173880748667e-05,
+      "loss": 0.6356,
+      "step": 1676
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09575813263654709,
+      "learning_rate": 7.65263120727631e-05,
+      "loss": 0.5661,
+      "step": 1677
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09399565309286118,
+      "learning_rate": 7.642091294280358e-05,
+      "loss": 0.6213,
+      "step": 1678
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09807447344064713,
+      "learning_rate": 7.631554154155616e-05,
+      "loss": 0.618,
+      "step": 1679
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.0962565466761589,
+      "learning_rate": 7.621019799293627e-05,
+      "loss": 0.7541,
+      "step": 1680
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.09099316596984863,
+      "learning_rate": 7.610488242082658e-05,
+      "loss": 0.5963,
+      "step": 1681
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.0969841405749321,
+      "learning_rate": 7.599959494907694e-05,
+      "loss": 0.6668,
+      "step": 1682
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.09640409052371979,
+      "learning_rate": 7.589433570150413e-05,
+      "loss": 0.6573,
+      "step": 1683
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.09514302760362625,
+      "learning_rate": 7.578910480189161e-05,
+      "loss": 0.6937,
+      "step": 1684
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.0959470346570015,
+      "learning_rate": 7.568390237398967e-05,
+      "loss": 0.6114,
+      "step": 1685
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.10075272619724274,
+      "learning_rate": 7.557872854151508e-05,
+      "loss": 0.6408,
+      "step": 1686
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.0935841053724289,
+      "learning_rate": 7.547358342815088e-05,
+      "loss": 0.5783,
+      "step": 1687
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.1003410667181015,
+      "learning_rate": 7.536846715754644e-05,
+      "loss": 0.572,
+      "step": 1688
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.15400315821170807,
+      "learning_rate": 7.52633798533172e-05,
+      "loss": 0.6688,
+      "step": 1689
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.09458847343921661,
+      "learning_rate": 7.515832163904456e-05,
+      "loss": 0.623,
+      "step": 1690
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.09835183620452881,
+      "learning_rate": 7.505329263827557e-05,
+      "loss": 0.6585,
+      "step": 1691
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.12145767360925674,
+      "learning_rate": 7.494829297452306e-05,
+      "loss": 0.5001,
+      "step": 1692
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.09669528156518936,
+      "learning_rate": 7.484332277126535e-05,
+      "loss": 0.6896,
+      "step": 1693
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.09365645051002502,
+      "learning_rate": 7.473838215194609e-05,
+      "loss": 0.6897,
+      "step": 1694
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.09353401511907578,
+      "learning_rate": 7.463347123997411e-05,
+      "loss": 0.6342,
+      "step": 1695
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.0940760150551796,
+      "learning_rate": 7.452859015872334e-05,
+      "loss": 0.6477,
+      "step": 1696
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.09309391677379608,
+      "learning_rate": 7.442373903153266e-05,
+      "loss": 0.6275,
+      "step": 1697
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.09288842976093292,
+      "learning_rate": 7.431891798170563e-05,
+      "loss": 0.6041,
+      "step": 1698
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.09645617753267288,
+      "learning_rate": 7.42141271325105e-05,
+      "loss": 0.6165,
+      "step": 1699
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.0938829854130745,
+      "learning_rate": 7.410936660718004e-05,
+      "loss": 0.6547,
+      "step": 1700
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.09161178767681122,
+      "learning_rate": 7.400463652891133e-05,
+      "loss": 0.6102,
+      "step": 1701
+    },
+    {
+      "epoch": 1.74,
+      "eval_loss": 0.8371061682701111,
+      "eval_runtime": 111.8891,
+      "eval_samples_per_second": 88.82,
+      "eval_steps_per_second": 22.209,
+      "step": 1701
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.09533992409706116,
+      "learning_rate": 7.38999370208656e-05,
+      "loss": 0.6343,
+      "step": 1702
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.10295694321393967,
+      "learning_rate": 7.379526820616814e-05,
+      "loss": 0.6374,
+      "step": 1703
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.09574326872825623,
+      "learning_rate": 7.36906302079082e-05,
+      "loss": 0.6461,
+      "step": 1704
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.10001670569181442,
+      "learning_rate": 7.358602314913881e-05,
+      "loss": 0.6596,
+      "step": 1705
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.0978463739156723,
+      "learning_rate": 7.348144715287648e-05,
+      "loss": 0.6998,
+      "step": 1706
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.0960407629609108,
+      "learning_rate": 7.337690234210132e-05,
+      "loss": 0.5749,
+      "step": 1707
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.10341007262468338,
+      "learning_rate": 7.327238883975674e-05,
+      "loss": 0.6611,
+      "step": 1708
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.09410237520933151,
+      "learning_rate": 7.316790676874928e-05,
+      "loss": 0.7077,
+      "step": 1709
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.09386301040649414,
+      "learning_rate": 7.306345625194852e-05,
+      "loss": 0.5772,
+      "step": 1710
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.0902644544839859,
+      "learning_rate": 7.295903741218705e-05,
+      "loss": 0.591,
+      "step": 1711
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.09402024745941162,
+      "learning_rate": 7.285465037226009e-05,
+      "loss": 0.6489,
+      "step": 1712
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.09027791768312454,
+      "learning_rate": 7.275029525492547e-05,
+      "loss": 0.6092,
+      "step": 1713
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.09368973970413208,
+      "learning_rate": 7.264597218290348e-05,
+      "loss": 0.6044,
+      "step": 1714
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.09430774301290512,
+      "learning_rate": 7.254168127887684e-05,
+      "loss": 0.5929,
+      "step": 1715
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.09778784215450287,
+      "learning_rate": 7.24374226654903e-05,
+      "loss": 0.6677,
+      "step": 1716
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.10461032390594482,
+      "learning_rate": 7.233319646535066e-05,
+      "loss": 0.652,
+      "step": 1717
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.09024833887815475,
+      "learning_rate": 7.222900280102669e-05,
+      "loss": 0.6162,
+      "step": 1718
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.09676025062799454,
+      "learning_rate": 7.212484179504882e-05,
+      "loss": 0.6984,
+      "step": 1719
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.09309113770723343,
+      "learning_rate": 7.202071356990904e-05,
+      "loss": 0.5851,
+      "step": 1720
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.09722644835710526,
+      "learning_rate": 7.191661824806091e-05,
+      "loss": 0.6278,
+      "step": 1721
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.09644819051027298,
+      "learning_rate": 7.18125559519192e-05,
+      "loss": 0.6012,
+      "step": 1722
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.1003071740269661,
+      "learning_rate": 7.170852680385989e-05,
+      "loss": 0.6875,
+      "step": 1723
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.1022835522890091,
+      "learning_rate": 7.160453092621994e-05,
+      "loss": 0.6246,
+      "step": 1724
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.10007347911596298,
+      "learning_rate": 7.150056844129718e-05,
+      "loss": 0.6191,
+      "step": 1725
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.0963386818766594,
+      "learning_rate": 7.139663947135027e-05,
+      "loss": 0.6031,
+      "step": 1726
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.09222231060266495,
+      "learning_rate": 7.129274413859832e-05,
+      "loss": 0.6137,
+      "step": 1727
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.10093759745359421,
+      "learning_rate": 7.118888256522093e-05,
+      "loss": 0.7364,
+      "step": 1728
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.094202920794487,
+      "learning_rate": 7.108505487335808e-05,
+      "loss": 0.6533,
+      "step": 1729
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.09733964502811432,
+      "learning_rate": 7.098126118510981e-05,
+      "loss": 0.709,
+      "step": 1730
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.09362705796957016,
+      "learning_rate": 7.087750162253619e-05,
+      "loss": 0.6303,
+      "step": 1731
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.0985776036977768,
+      "learning_rate": 7.077377630765716e-05,
+      "loss": 0.5934,
+      "step": 1732
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.10437599569559097,
+      "learning_rate": 7.067008536245242e-05,
+      "loss": 0.6266,
+      "step": 1733
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.09321979433298111,
+      "learning_rate": 7.056642890886124e-05,
+      "loss": 0.5763,
+      "step": 1734
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.10222089290618896,
+      "learning_rate": 7.046280706878225e-05,
+      "loss": 0.6378,
+      "step": 1735
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.09610838443040848,
+      "learning_rate": 7.035921996407353e-05,
+      "loss": 0.6738,
+      "step": 1736
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.1020408347249031,
+      "learning_rate": 7.025566771655219e-05,
+      "loss": 0.5614,
+      "step": 1737
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.09553298354148865,
+      "learning_rate": 7.015215044799434e-05,
+      "loss": 0.621,
+      "step": 1738
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.0918021947145462,
+      "learning_rate": 7.004866828013501e-05,
+      "loss": 0.5604,
+      "step": 1739
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.10076640546321869,
+      "learning_rate": 6.994522133466797e-05,
+      "loss": 0.6355,
+      "step": 1740
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.09725600481033325,
+      "learning_rate": 6.984180973324554e-05,
+      "loss": 0.6387,
+      "step": 1741
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.0945560559630394,
+      "learning_rate": 6.973843359747845e-05,
+      "loss": 0.6378,
+      "step": 1742
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.09677950292825699,
+      "learning_rate": 6.96350930489357e-05,
+      "loss": 0.6667,
+      "step": 1743
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.09894482791423798,
+      "learning_rate": 6.953178820914457e-05,
+      "loss": 0.7401,
+      "step": 1744
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.09255923330783844,
+      "learning_rate": 6.942851919959024e-05,
+      "loss": 0.6147,
+      "step": 1745
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.09706816077232361,
+      "learning_rate": 6.932528614171571e-05,
+      "loss": 0.6334,
+      "step": 1746
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.0974464863538742,
+      "learning_rate": 6.922208915692186e-05,
+      "loss": 0.5988,
+      "step": 1747
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.10448437184095383,
+      "learning_rate": 6.911892836656704e-05,
+      "loss": 0.6654,
+      "step": 1748
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.09835972636938095,
+      "learning_rate": 6.901580389196698e-05,
+      "loss": 0.5493,
+      "step": 1749
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.10718105733394623,
+      "learning_rate": 6.891271585439481e-05,
+      "loss": 0.6815,
+      "step": 1750
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.09446258097887039,
+      "learning_rate": 6.880966437508081e-05,
+      "loss": 0.6518,
+      "step": 1751
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.0932326465845108,
+      "learning_rate": 6.870664957521224e-05,
+      "loss": 0.6398,
+      "step": 1752
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.0950569435954094,
+      "learning_rate": 6.860367157593318e-05,
+      "loss": 0.6705,
+      "step": 1753
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.09746532142162323,
+      "learning_rate": 6.850073049834443e-05,
+      "loss": 0.6843,
+      "step": 1754
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.10170366615056992,
+      "learning_rate": 6.83978264635035e-05,
+      "loss": 0.6565,
+      "step": 1755
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.10435983538627625,
+      "learning_rate": 6.829495959242425e-05,
+      "loss": 0.6343,
+      "step": 1756
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.09789245575666428,
+      "learning_rate": 6.819213000607673e-05,
+      "loss": 0.6852,
+      "step": 1757
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.09626414626836777,
+      "learning_rate": 6.808933782538735e-05,
+      "loss": 0.6438,
+      "step": 1758
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.09926639497280121,
+      "learning_rate": 6.798658317123844e-05,
+      "loss": 0.6113,
+      "step": 1759
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.09973691403865814,
+      "learning_rate": 6.788386616446813e-05,
+      "loss": 0.6834,
+      "step": 1760
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.09590507298707962,
+      "learning_rate": 6.778118692587032e-05,
+      "loss": 0.7063,
+      "step": 1761
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.09678776562213898,
+      "learning_rate": 6.76785455761946e-05,
+      "loss": 0.5959,
+      "step": 1762
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.10482712835073471,
+      "learning_rate": 6.757594223614587e-05,
+      "loss": 0.6836,
+      "step": 1763
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.10087873041629791,
+      "learning_rate": 6.747337702638435e-05,
+      "loss": 0.688,
+      "step": 1764
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.0967005044221878,
+      "learning_rate": 6.73708500675255e-05,
+      "loss": 0.692,
+      "step": 1765
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.10612227022647858,
+      "learning_rate": 6.72683614801397e-05,
+      "loss": 0.6701,
+      "step": 1766
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.09886319935321808,
+      "learning_rate": 6.71659113847523e-05,
+      "loss": 0.7092,
+      "step": 1767
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.09441614151000977,
+      "learning_rate": 6.706349990184327e-05,
+      "loss": 0.6237,
+      "step": 1768
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09582387655973434,
+      "learning_rate": 6.696112715184728e-05,
+      "loss": 0.5424,
+      "step": 1769
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09584696590900421,
+      "learning_rate": 6.685879325515341e-05,
+      "loss": 0.6346,
+      "step": 1770
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09787273406982422,
+      "learning_rate": 6.675649833210503e-05,
+      "loss": 0.6828,
+      "step": 1771
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09562217444181442,
+      "learning_rate": 6.665424250299963e-05,
+      "loss": 0.6426,
+      "step": 1772
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.10484904795885086,
+      "learning_rate": 6.65520258880889e-05,
+      "loss": 0.6905,
+      "step": 1773
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09628799557685852,
+      "learning_rate": 6.644984860757826e-05,
+      "loss": 0.6124,
+      "step": 1774
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09628019481897354,
+      "learning_rate": 6.634771078162688e-05,
+      "loss": 0.6609,
+      "step": 1775
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09470858424901962,
+      "learning_rate": 6.624561253034762e-05,
+      "loss": 0.6147,
+      "step": 1776
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.10551173239946365,
+      "learning_rate": 6.614355397380673e-05,
+      "loss": 0.6161,
+      "step": 1777
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.09330208599567413,
+      "learning_rate": 6.60415352320238e-05,
+      "loss": 0.674,
+      "step": 1778
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.09938740730285645,
+      "learning_rate": 6.593955642497156e-05,
+      "loss": 0.7023,
+      "step": 1779
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.10205026715993881,
+      "learning_rate": 6.583761767257587e-05,
+      "loss": 0.6612,
+      "step": 1780
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.09328316897153854,
+      "learning_rate": 6.573571909471543e-05,
+      "loss": 0.6346,
+      "step": 1781
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.0924779623746872,
+      "learning_rate": 6.563386081122166e-05,
+      "loss": 0.6468,
+      "step": 1782
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.09521466493606567,
+      "learning_rate": 6.553204294187862e-05,
+      "loss": 0.6373,
+      "step": 1783
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.09920044988393784,
+      "learning_rate": 6.543026560642288e-05,
+      "loss": 0.6348,
+      "step": 1784
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.09735903888940811,
+      "learning_rate": 6.532852892454335e-05,
+      "loss": 0.7126,
+      "step": 1785
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.09407635033130646,
+      "learning_rate": 6.522683301588103e-05,
+      "loss": 0.6486,
+      "step": 1786
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.1004701629281044,
+      "learning_rate": 6.512517800002909e-05,
+      "loss": 0.6421,
+      "step": 1787
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.09651912748813629,
+      "learning_rate": 6.502356399653255e-05,
+      "loss": 0.6397,
+      "step": 1788
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.1026163101196289,
+      "learning_rate": 6.492199112488819e-05,
+      "loss": 0.6833,
+      "step": 1789
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.09641033411026001,
+      "learning_rate": 6.48204595045444e-05,
+      "loss": 0.6028,
+      "step": 1790
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.10638896375894547,
+      "learning_rate": 6.47189692549012e-05,
+      "loss": 0.6609,
+      "step": 1791
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.09619806706905365,
+      "learning_rate": 6.461752049530982e-05,
+      "loss": 0.6464,
+      "step": 1792
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.0988159254193306,
+      "learning_rate": 6.451611334507268e-05,
+      "loss": 0.5935,
+      "step": 1793
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.09818794578313828,
+      "learning_rate": 6.441474792344339e-05,
+      "loss": 0.6369,
+      "step": 1794
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.0979076698422432,
+      "learning_rate": 6.431342434962638e-05,
+      "loss": 0.628,
+      "step": 1795
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.10532572865486145,
+      "learning_rate": 6.421214274277696e-05,
+      "loss": 0.6542,
+      "step": 1796
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.09725423902273178,
+      "learning_rate": 6.411090322200094e-05,
+      "loss": 0.6387,
+      "step": 1797
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.09431372582912445,
+      "learning_rate": 6.400970590635482e-05,
+      "loss": 0.6734,
+      "step": 1798
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.09110352396965027,
+      "learning_rate": 6.390855091484537e-05,
+      "loss": 0.6014,
+      "step": 1799
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.09633510559797287,
+      "learning_rate": 6.380743836642955e-05,
+      "loss": 0.6659,
+      "step": 1800
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.09658326208591461,
+      "learning_rate": 6.370636838001447e-05,
+      "loss": 0.6609,
+      "step": 1801
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.1059308722615242,
+      "learning_rate": 6.360534107445721e-05,
+      "loss": 0.6506,
+      "step": 1802
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.0991239994764328,
+      "learning_rate": 6.350435656856461e-05,
+      "loss": 0.6392,
+      "step": 1803
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.10205315053462982,
+      "learning_rate": 6.340341498109314e-05,
+      "loss": 0.6068,
+      "step": 1804
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.1025339886546135,
+      "learning_rate": 6.33025164307489e-05,
+      "loss": 0.6806,
+      "step": 1805
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.10113833099603653,
+      "learning_rate": 6.32016610361873e-05,
+      "loss": 0.7183,
+      "step": 1806
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.09995482861995697,
+      "learning_rate": 6.310084891601306e-05,
+      "loss": 0.7109,
+      "step": 1807
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.09897339344024658,
+      "learning_rate": 6.30000801887799e-05,
+      "loss": 0.6647,
+      "step": 1808
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.10555449873209,
+      "learning_rate": 6.289935497299067e-05,
+      "loss": 0.6565,
+      "step": 1809
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.09625938534736633,
+      "learning_rate": 6.279867338709693e-05,
+      "loss": 0.6205,
+      "step": 1810
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.09640774875879288,
+      "learning_rate": 6.269803554949895e-05,
+      "loss": 0.6778,
+      "step": 1811
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.10027953237295151,
+      "learning_rate": 6.259744157854559e-05,
+      "loss": 0.5868,
+      "step": 1812
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.09466378390789032,
+      "learning_rate": 6.249689159253412e-05,
+      "loss": 0.7303,
+      "step": 1813
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.10513494908809662,
+      "learning_rate": 6.239638570971005e-05,
+      "loss": 0.7197,
+      "step": 1814
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.09241978079080582,
+      "learning_rate": 6.229592404826703e-05,
+      "loss": 0.5639,
+      "step": 1815
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.09812828153371811,
+      "learning_rate": 6.219550672634672e-05,
+      "loss": 0.6529,
+      "step": 1816
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.10196724534034729,
+      "learning_rate": 6.209513386203872e-05,
+      "loss": 0.6363,
+      "step": 1817
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.09870493412017822,
+      "learning_rate": 6.199480557338015e-05,
+      "loss": 0.6249,
+      "step": 1818
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.10112203657627106,
+      "learning_rate": 6.189452197835586e-05,
+      "loss": 0.607,
+      "step": 1819
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.10369737446308136,
+      "learning_rate": 6.179428319489814e-05,
+      "loss": 0.6382,
+      "step": 1820
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.09579195082187653,
+      "learning_rate": 6.169408934088655e-05,
+      "loss": 0.5788,
+      "step": 1821
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.09685015678405762,
+      "learning_rate": 6.159394053414775e-05,
+      "loss": 0.6578,
+      "step": 1822
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.09669235348701477,
+      "learning_rate": 6.149383689245554e-05,
+      "loss": 0.6115,
+      "step": 1823
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.09187985956668854,
+      "learning_rate": 6.139377853353054e-05,
+      "loss": 0.6238,
+      "step": 1824
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.09645017981529236,
+      "learning_rate": 6.129376557504017e-05,
+      "loss": 0.6104,
+      "step": 1825
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.09995485842227936,
+      "learning_rate": 6.119379813459832e-05,
+      "loss": 0.7035,
+      "step": 1826
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.09640282392501831,
+      "learning_rate": 6.109387632976555e-05,
+      "loss": 0.5992,
+      "step": 1827
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.09824865311384201,
+      "learning_rate": 6.099400027804863e-05,
+      "loss": 0.6019,
+      "step": 1828
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.09777974337339401,
+      "learning_rate": 6.0894170096900524e-05,
+      "loss": 0.6683,
+      "step": 1829
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.10040297359228134,
+      "learning_rate": 6.0794385903720276e-05,
+      "loss": 0.6419,
+      "step": 1830
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.09963066130876541,
+      "learning_rate": 6.069464781585288e-05,
+      "loss": 0.6706,
+      "step": 1831
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.09737717360258102,
+      "learning_rate": 6.0594955950589104e-05,
+      "loss": 0.6959,
+      "step": 1832
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.09666707366704941,
+      "learning_rate": 6.049531042516527e-05,
+      "loss": 0.6304,
+      "step": 1833
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.10755129903554916,
+      "learning_rate": 6.039571135676332e-05,
+      "loss": 0.6474,
+      "step": 1834
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.10874524712562561,
+      "learning_rate": 6.029615886251051e-05,
+      "loss": 0.6356,
+      "step": 1835
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.09853578358888626,
+      "learning_rate": 6.019665305947935e-05,
+      "loss": 0.6847,
+      "step": 1836
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.10183116048574448,
+      "learning_rate": 6.0097194064687345e-05,
+      "loss": 0.6525,
+      "step": 1837
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.10062870383262634,
+      "learning_rate": 5.99977819950971e-05,
+      "loss": 0.5876,
+      "step": 1838
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.09886735677719116,
+      "learning_rate": 5.989841696761598e-05,
+      "loss": 0.5899,
+      "step": 1839
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.09872845560312271,
+      "learning_rate": 5.9799099099095955e-05,
+      "loss": 0.6432,
+      "step": 1840
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.09901401400566101,
+      "learning_rate": 5.969982850633365e-05,
+      "loss": 0.6734,
+      "step": 1841
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.09907763451337814,
+      "learning_rate": 5.960060530607001e-05,
+      "loss": 0.6259,
+      "step": 1842
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.09919259697198868,
+      "learning_rate": 5.950142961499032e-05,
+      "loss": 0.6271,
+      "step": 1843
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.09853816032409668,
+      "learning_rate": 5.94023015497239e-05,
+      "loss": 0.6366,
+      "step": 1844
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.0947277694940567,
+      "learning_rate": 5.9303221226844174e-05,
+      "loss": 0.5816,
+      "step": 1845
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.09792958945035934,
+      "learning_rate": 5.920418876286831e-05,
+      "loss": 0.6805,
+      "step": 1846
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.10146981477737427,
+      "learning_rate": 5.9105204274257344e-05,
+      "loss": 0.6384,
+      "step": 1847
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.09037366509437561,
+      "learning_rate": 5.900626787741567e-05,
+      "loss": 0.6045,
+      "step": 1848
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.0977599248290062,
+      "learning_rate": 5.890737968869133e-05,
+      "loss": 0.6032,
+      "step": 1849
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.0970187783241272,
+      "learning_rate": 5.880853982437562e-05,
+      "loss": 0.7293,
+      "step": 1850
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.09835325181484222,
+      "learning_rate": 5.870974840070291e-05,
+      "loss": 0.6096,
+      "step": 1851
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.09433458745479584,
+      "learning_rate": 5.861100553385076e-05,
+      "loss": 0.617,
+      "step": 1852
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.1016274243593216,
+      "learning_rate": 5.8512311339939505e-05,
+      "loss": 0.6759,
+      "step": 1853
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.09398248046636581,
+      "learning_rate": 5.841366593503232e-05,
+      "loss": 0.6219,
+      "step": 1854
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.10223573446273804,
+      "learning_rate": 5.8315069435134915e-05,
+      "loss": 0.5961,
+      "step": 1855
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.0986015796661377,
+      "learning_rate": 5.821652195619558e-05,
+      "loss": 0.667,
+      "step": 1856
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.09686239808797836,
+      "learning_rate": 5.811802361410492e-05,
+      "loss": 0.5717,
+      "step": 1857
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.09741710126399994,
+      "learning_rate": 5.8019574524695763e-05,
+      "loss": 0.6352,
+      "step": 1858
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.09348324686288834,
+      "learning_rate": 5.7921174803743e-05,
+      "loss": 0.628,
+      "step": 1859
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.15314215421676636,
+      "learning_rate": 5.782282456696344e-05,
+      "loss": 0.5806,
+      "step": 1860
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.0961272343993187,
+      "learning_rate": 5.772452393001579e-05,
+      "loss": 0.6204,
+      "step": 1861
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.0999687910079956,
+      "learning_rate": 5.762627300850034e-05,
+      "loss": 0.6213,
+      "step": 1862
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.09896641969680786,
+      "learning_rate": 5.752807191795897e-05,
+      "loss": 0.6439,
+      "step": 1863
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.09776857495307922,
+      "learning_rate": 5.7429920773874913e-05,
+      "loss": 0.6517,
+      "step": 1864
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.09646885842084885,
+      "learning_rate": 5.7331819691672696e-05,
+      "loss": 0.5246,
+      "step": 1865
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.09869589656591415,
+      "learning_rate": 5.723376878671798e-05,
+      "loss": 0.6551,
+      "step": 1866
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.10511456429958344,
+      "learning_rate": 5.7135768174317383e-05,
+      "loss": 0.6876,
+      "step": 1867
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.10038403421640396,
+      "learning_rate": 5.703781796971841e-05,
+      "loss": 0.6131,
+      "step": 1868
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.09927596896886826,
+      "learning_rate": 5.6939918288109295e-05,
+      "loss": 0.5928,
+      "step": 1869
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.09631170332431793,
+      "learning_rate": 5.684206924461884e-05,
+      "loss": 0.5922,
+      "step": 1870
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.10013575106859207,
+      "learning_rate": 5.674427095431627e-05,
+      "loss": 0.6924,
+      "step": 1871
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.09850646555423737,
+      "learning_rate": 5.664652353221118e-05,
+      "loss": 0.614,
+      "step": 1872
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.10419221222400665,
+      "learning_rate": 5.654882709325331e-05,
+      "loss": 0.6171,
+      "step": 1873
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.10470752418041229,
+      "learning_rate": 5.645118175233245e-05,
+      "loss": 0.6852,
+      "step": 1874
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.09644245356321335,
+      "learning_rate": 5.635358762427831e-05,
+      "loss": 0.6247,
+      "step": 1875
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.09776662290096283,
+      "learning_rate": 5.625604482386037e-05,
+      "loss": 0.6186,
+      "step": 1876
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.10003117471933365,
+      "learning_rate": 5.6158553465787734e-05,
+      "loss": 0.591,
+      "step": 1877
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.09640898555517197,
+      "learning_rate": 5.606111366470902e-05,
+      "loss": 0.6602,
+      "step": 1878
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.09401225298643112,
+      "learning_rate": 5.596372553521223e-05,
+      "loss": 0.5959,
+      "step": 1879
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.10302352905273438,
+      "learning_rate": 5.5866389191824584e-05,
+      "loss": 0.6692,
+      "step": 1880
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.09677159786224365,
+      "learning_rate": 5.57691047490124e-05,
+      "loss": 0.6012,
+      "step": 1881
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.10318101942539215,
+      "learning_rate": 5.5671872321180983e-05,
+      "loss": 0.6648,
+      "step": 1882
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.0969577431678772,
+      "learning_rate": 5.557469202267445e-05,
+      "loss": 0.6417,
+      "step": 1883
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.09170772135257721,
+      "learning_rate": 5.5477563967775615e-05,
+      "loss": 0.6058,
+      "step": 1884
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.11065363138914108,
+      "learning_rate": 5.538048827070585e-05,
+      "loss": 0.6407,
+      "step": 1885
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.10233711451292038,
+      "learning_rate": 5.5283465045624973e-05,
+      "loss": 0.6452,
+      "step": 1886
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.09645158052444458,
+      "learning_rate": 5.518649440663108e-05,
+      "loss": 0.6215,
+      "step": 1887
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.09994348138570786,
+      "learning_rate": 5.508957646776045e-05,
+      "loss": 0.5915,
+      "step": 1888
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.09882402420043945,
+      "learning_rate": 5.499271134298736e-05,
+      "loss": 0.599,
+      "step": 1889
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.10297371447086334,
+      "learning_rate": 5.489589914622397e-05,
+      "loss": 0.6088,
+      "step": 1890
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.10148060321807861,
+      "learning_rate": 5.4799139991320234e-05,
+      "loss": 0.6861,
+      "step": 1891
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.10107173770666122,
+      "learning_rate": 5.470243399206371e-05,
+      "loss": 0.6522,
+      "step": 1892
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.10026003420352936,
+      "learning_rate": 5.460578126217943e-05,
+      "loss": 0.519,
+      "step": 1893
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.10234738886356354,
+      "learning_rate": 5.450918191532982e-05,
+      "loss": 0.6261,
+      "step": 1894
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.10654883086681366,
+      "learning_rate": 5.44126360651145e-05,
+      "loss": 0.6174,
+      "step": 1895
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.10551703721284866,
+      "learning_rate": 5.431614382507017e-05,
+      "loss": 0.6327,
+      "step": 1896
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.09744749963283539,
+      "learning_rate": 5.421970530867051e-05,
+      "loss": 0.5979,
+      "step": 1897
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.10634944587945938,
+      "learning_rate": 5.4123320629326024e-05,
+      "loss": 0.6422,
+      "step": 1898
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.09777097404003143,
+      "learning_rate": 5.402698990038387e-05,
+      "loss": 0.5914,
+      "step": 1899
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.09305303543806076,
+      "learning_rate": 5.393071323512779e-05,
+      "loss": 0.5919,
+      "step": 1900
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.09990797936916351,
+      "learning_rate": 5.383449074677795e-05,
+      "loss": 0.6787,
+      "step": 1901
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.09466062486171722,
+      "learning_rate": 5.3738322548490805e-05,
+      "loss": 0.6273,
+      "step": 1902
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.09690329432487488,
+      "learning_rate": 5.3642208753358924e-05,
+      "loss": 0.6153,
+      "step": 1903
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.09096551686525345,
+      "learning_rate": 5.354614947441094e-05,
+      "loss": 0.5859,
+      "step": 1904
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.09688370674848557,
+      "learning_rate": 5.3450144824611394e-05,
+      "loss": 0.6186,
+      "step": 1905
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.10283687710762024,
+      "learning_rate": 5.335419491686051e-05,
+      "loss": 0.6789,
+      "step": 1906
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.09568949788808823,
+      "learning_rate": 5.325829986399422e-05,
+      "loss": 0.6123,
+      "step": 1907
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.09774908423423767,
+      "learning_rate": 5.316245977878389e-05,
+      "loss": 0.6894,
+      "step": 1908
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.0992112085223198,
+      "learning_rate": 5.3066674773936274e-05,
+      "loss": 0.6634,
+      "step": 1909
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.09688520431518555,
+      "learning_rate": 5.297094496209334e-05,
+      "loss": 0.6306,
+      "step": 1910
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.09815088659524918,
+      "learning_rate": 5.287527045583216e-05,
+      "loss": 0.6227,
+      "step": 1911
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.10484449565410614,
+      "learning_rate": 5.277965136766473e-05,
+      "loss": 0.6358,
+      "step": 1912
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.09204939007759094,
+      "learning_rate": 5.2684087810037926e-05,
+      "loss": 0.5693,
+      "step": 1913
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.09460483491420746,
+      "learning_rate": 5.25885798953333e-05,
+      "loss": 0.6268,
+      "step": 1914
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.09923145920038223,
+      "learning_rate": 5.249312773586696e-05,
+      "loss": 0.6568,
+      "step": 1915
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.09629133343696594,
+      "learning_rate": 5.2397731443889466e-05,
+      "loss": 0.666,
+      "step": 1916
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.1004701554775238,
+      "learning_rate": 5.230239113158566e-05,
+      "loss": 0.6536,
+      "step": 1917
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.1015523374080658,
+      "learning_rate": 5.2207106911074554e-05,
+      "loss": 0.638,
+      "step": 1918
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.10012494027614594,
+      "learning_rate": 5.2111878894409206e-05,
+      "loss": 0.6571,
+      "step": 1919
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.10520032048225403,
+      "learning_rate": 5.201670719357657e-05,
+      "loss": 0.7547,
+      "step": 1920
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.09958270192146301,
+      "learning_rate": 5.192159192049739e-05,
+      "loss": 0.6797,
+      "step": 1921
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.09816860407590866,
+      "learning_rate": 5.182653318702604e-05,
+      "loss": 0.6759,
+      "step": 1922
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.10633301734924316,
+      "learning_rate": 5.173153110495039e-05,
+      "loss": 0.6648,
+      "step": 1923
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.09729764610528946,
+      "learning_rate": 5.1636585785991706e-05,
+      "loss": 0.6941,
+      "step": 1924
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.09836485981941223,
+      "learning_rate": 5.154169734180452e-05,
+      "loss": 0.6101,
+      "step": 1925
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.0946260467171669,
+      "learning_rate": 5.1446865883976426e-05,
+      "loss": 0.5646,
+      "step": 1926
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.10132639110088348,
+      "learning_rate": 5.135209152402805e-05,
+      "loss": 0.6526,
+      "step": 1927
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.09764391928911209,
+      "learning_rate": 5.1257374373412846e-05,
+      "loss": 0.6224,
+      "step": 1928
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.09751232713460922,
+      "learning_rate": 5.116271454351702e-05,
+      "loss": 0.6649,
+      "step": 1929
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.1026969626545906,
+      "learning_rate": 5.106811214565934e-05,
+      "loss": 0.6471,
+      "step": 1930
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.10073301196098328,
+      "learning_rate": 5.0973567291091064e-05,
+      "loss": 0.6088,
+      "step": 1931
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.09634771198034286,
+      "learning_rate": 5.0879080090995724e-05,
+      "loss": 0.6002,
+      "step": 1932
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.09814572334289551,
+      "learning_rate": 5.0784650656489117e-05,
+      "loss": 0.6141,
+      "step": 1933
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.10008418560028076,
+      "learning_rate": 5.069027909861905e-05,
+      "loss": 0.6476,
+      "step": 1934
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.09970271587371826,
+      "learning_rate": 5.059596552836534e-05,
+      "loss": 0.686,
+      "step": 1935
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.10272549092769623,
+      "learning_rate": 5.050171005663954e-05,
+      "loss": 0.671,
+      "step": 1936
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.10756421834230423,
+      "learning_rate": 5.0407512794284907e-05,
+      "loss": 0.5907,
+      "step": 1937
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.09952995181083679,
+      "learning_rate": 5.031337385207626e-05,
+      "loss": 0.6834,
+      "step": 1938
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.09724836051464081,
+      "learning_rate": 5.02192933407198e-05,
+      "loss": 0.6516,
+      "step": 1939
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.09979628026485443,
+      "learning_rate": 5.0125271370853054e-05,
+      "loss": 0.7203,
+      "step": 1940
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.09599156677722931,
+      "learning_rate": 5.003130805304468e-05,
+      "loss": 0.6138,
+      "step": 1941
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.09719957411289215,
+      "learning_rate": 4.993740349779437e-05,
+      "loss": 0.6374,
+      "step": 1942
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.09641937911510468,
+      "learning_rate": 4.9843557815532694e-05,
+      "loss": 0.5733,
+      "step": 1943
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.09918541461229324,
+      "learning_rate": 4.974977111662101e-05,
+      "loss": 0.6007,
+      "step": 1944
+    },
+    {
+      "epoch": 1.99,
+      "eval_loss": 0.8391451835632324,
+      "eval_runtime": 111.6994,
+      "eval_samples_per_second": 88.971,
+      "eval_steps_per_second": 22.247,
+      "step": 1944
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.10276373475790024,
+      "learning_rate": 4.965604351135132e-05,
+      "loss": 0.7218,
+      "step": 1945
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.09531883895397186,
+      "learning_rate": 4.95623751099461e-05,
+      "loss": 0.6271,
+      "step": 1946
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.09618810564279556,
+      "learning_rate": 4.946876602255822e-05,
+      "loss": 0.672,
+      "step": 1947
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.09534689784049988,
+      "learning_rate": 4.937521635927081e-05,
+      "loss": 0.6468,
+      "step": 1948
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.10133272409439087,
+      "learning_rate": 4.9281726230097114e-05,
+      "loss": 0.7151,
+      "step": 1949
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.10226277261972427,
+      "learning_rate": 4.918829574498033e-05,
+      "loss": 0.6447,
+      "step": 1950
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.10175823420286179,
+      "learning_rate": 4.909492501379356e-05,
+      "loss": 0.5616,
+      "step": 1951
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.09981898963451385,
+      "learning_rate": 4.900161414633961e-05,
+      "loss": 0.6593,
+      "step": 1952
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.10000532865524292,
+      "learning_rate": 4.890836325235092e-05,
+      "loss": 0.6415,
+      "step": 1953
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09873597323894501,
+      "learning_rate": 4.8815172441489344e-05,
+      "loss": 0.6578,
+      "step": 1954
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.11193492263555527,
+      "learning_rate": 4.872204182334613e-05,
+      "loss": 0.6552,
+      "step": 1955
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09587957710027695,
+      "learning_rate": 4.862897150744171e-05,
+      "loss": 0.5441,
+      "step": 1956
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09471358358860016,
+      "learning_rate": 4.8535961603225644e-05,
+      "loss": 0.5831,
+      "step": 1957
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09220650047063828,
+      "learning_rate": 4.844301222007638e-05,
+      "loss": 0.6205,
+      "step": 1958
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.10302426666021347,
+      "learning_rate": 4.835012346730125e-05,
+      "loss": 0.6589,
+      "step": 1959
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09965970367193222,
+      "learning_rate": 4.8257295454136284e-05,
+      "loss": 0.6809,
+      "step": 1960
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09576713293790817,
+      "learning_rate": 4.816452828974604e-05,
+      "loss": 0.6379,
+      "step": 1961
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.10165003687143326,
+      "learning_rate": 4.807182208322356e-05,
+      "loss": 0.6725,
+      "step": 1962
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10322687774896622,
+      "learning_rate": 4.7979176943590174e-05,
+      "loss": 0.5905,
+      "step": 1963
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09559673815965652,
+      "learning_rate": 4.788659297979543e-05,
+      "loss": 0.5876,
+      "step": 1964
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09789896011352539,
+      "learning_rate": 4.77940703007169e-05,
+      "loss": 0.6243,
+      "step": 1965
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10369144380092621,
+      "learning_rate": 4.7701609015160096e-05,
+      "loss": 0.6923,
+      "step": 1966
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.1011209785938263,
+      "learning_rate": 4.7609209231858345e-05,
+      "loss": 0.6342,
+      "step": 1967
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10288915038108826,
+      "learning_rate": 4.7516871059472636e-05,
+      "loss": 0.6523,
+      "step": 1968
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09447886794805527,
+      "learning_rate": 4.74245946065915e-05,
+      "loss": 0.5762,
+      "step": 1969
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.0987362340092659,
+      "learning_rate": 4.7332379981730915e-05,
+      "loss": 0.6135,
+      "step": 1970
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09803710132837296,
+      "learning_rate": 4.7240227293334106e-05,
+      "loss": 0.641,
+      "step": 1971
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09446247667074203,
+      "learning_rate": 4.714813664977149e-05,
+      "loss": 0.622,
+      "step": 1972
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.09788127988576889,
+      "learning_rate": 4.705610815934052e-05,
+      "loss": 0.6332,
+      "step": 1973
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10170095413923264,
+      "learning_rate": 4.696414193026555e-05,
+      "loss": 0.6383,
+      "step": 1974
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10501042008399963,
+      "learning_rate": 4.687223807069772e-05,
+      "loss": 0.7155,
+      "step": 1975
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09749948233366013,
+      "learning_rate": 4.678039668871481e-05,
+      "loss": 0.6272,
+      "step": 1976
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09135567396879196,
+      "learning_rate": 4.668861789232115e-05,
+      "loss": 0.5994,
+      "step": 1977
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.09460758417844772,
+      "learning_rate": 4.6596901789447464e-05,
+      "loss": 0.5524,
+      "step": 1978
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.0967264249920845,
+      "learning_rate": 4.650524848795074e-05,
+      "loss": 0.6697,
+      "step": 1979
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09406213462352753,
+      "learning_rate": 4.64136580956141e-05,
+      "loss": 0.5412,
+      "step": 1980
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09364727139472961,
+      "learning_rate": 4.632213072014673e-05,
+      "loss": 0.6109,
+      "step": 1981
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.09998447448015213,
+      "learning_rate": 4.623066646918366e-05,
+      "loss": 0.6546,
+      "step": 1982
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10004132986068726,
+      "learning_rate": 4.613926545028571e-05,
+      "loss": 0.6286,
+      "step": 1983
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10160532593727112,
+      "learning_rate": 4.6047927770939347e-05,
+      "loss": 0.6424,
+      "step": 1984
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10370038449764252,
+      "learning_rate": 4.5956653538556536e-05,
+      "loss": 0.6075,
+      "step": 1985
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10678977519273758,
+      "learning_rate": 4.586544286047464e-05,
+      "loss": 0.5969,
+      "step": 1986
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10616578161716461,
+      "learning_rate": 4.577429584395626e-05,
+      "loss": 0.5871,
+      "step": 1987
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10694827139377594,
+      "learning_rate": 4.5683212596189184e-05,
+      "loss": 0.5823,
+      "step": 1988
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.10135646909475327,
+      "learning_rate": 4.559219322428614e-05,
+      "loss": 0.5714,
+      "step": 1989
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10127806663513184,
+      "learning_rate": 4.5501237835284795e-05,
+      "loss": 0.5574,
+      "step": 1990
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10665806382894516,
+      "learning_rate": 4.541034653614754e-05,
+      "loss": 0.5946,
+      "step": 1991
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10035495460033417,
+      "learning_rate": 4.5319519433761416e-05,
+      "loss": 0.5491,
+      "step": 1992
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.09943161904811859,
+      "learning_rate": 4.522875663493795e-05,
+      "loss": 0.6344,
+      "step": 1993
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.09872744232416153,
+      "learning_rate": 4.513805824641307e-05,
+      "loss": 0.5458,
+      "step": 1994
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10467476397752762,
+      "learning_rate": 4.504742437484696e-05,
+      "loss": 0.6104,
+      "step": 1995
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.09266231954097748,
+      "learning_rate": 4.495685512682389e-05,
+      "loss": 0.5422,
+      "step": 1996
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.09579336643218994,
+      "learning_rate": 4.4866350608852207e-05,
+      "loss": 0.5115,
+      "step": 1997
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10147663950920105,
+      "learning_rate": 4.477591092736405e-05,
+      "loss": 0.5676,
+      "step": 1998
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.10405425727367401,
+      "learning_rate": 4.468553618871539e-05,
+      "loss": 0.5667,
+      "step": 1999
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.10756871849298477,
+      "learning_rate": 4.459522649918578e-05,
+      "loss": 0.6366,
+      "step": 2000
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.09812985360622406,
+      "learning_rate": 4.450498196497828e-05,
+      "loss": 0.5416,
+      "step": 2001
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.09972529858350754,
+      "learning_rate": 4.4414802692219346e-05,
+      "loss": 0.5876,
+      "step": 2002
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.10081417858600616,
+      "learning_rate": 4.432468878695868e-05,
+      "loss": 0.5851,
+      "step": 2003
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.11247777938842773,
+      "learning_rate": 4.42346403551691e-05,
+      "loss": 0.6644,
+      "step": 2004
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.10756223648786545,
+      "learning_rate": 4.414465750274643e-05,
+      "loss": 0.6102,
+      "step": 2005
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.11053269356489182,
+      "learning_rate": 4.4054740335509404e-05,
+      "loss": 0.6136,
+      "step": 2006
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.10148683935403824,
+      "learning_rate": 4.396488895919948e-05,
+      "loss": 0.6576,
+      "step": 2007
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.11205828934907913,
+      "learning_rate": 4.387510347948075e-05,
+      "loss": 0.5965,
+      "step": 2008
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.10538581758737564,
+      "learning_rate": 4.378538400193982e-05,
+      "loss": 0.5825,
+      "step": 2009
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.10357122123241425,
+      "learning_rate": 4.369573063208568e-05,
+      "loss": 0.6026,
+      "step": 2010
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.10646289587020874,
+      "learning_rate": 4.360614347534956e-05,
+      "loss": 0.6271,
+      "step": 2011
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.10473828762769699,
+      "learning_rate": 4.351662263708486e-05,
+      "loss": 0.5845,
+      "step": 2012
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.1110227182507515,
+      "learning_rate": 4.342716822256695e-05,
+      "loss": 0.635,
+      "step": 2013
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.1004132404923439,
+      "learning_rate": 4.3337780336993104e-05,
+      "loss": 0.5776,
+      "step": 2014
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.09966661036014557,
+      "learning_rate": 4.3248459085482374e-05,
+      "loss": 0.5181,
+      "step": 2015
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.1144009679555893,
+      "learning_rate": 4.315920457307543e-05,
+      "loss": 0.6382,
+      "step": 2016
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.1163625493645668,
+      "learning_rate": 4.307001690473447e-05,
+      "loss": 0.5464,
+      "step": 2017
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.10942616313695908,
+      "learning_rate": 4.298089618534307e-05,
+      "loss": 0.6113,
+      "step": 2018
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10717497020959854,
+      "learning_rate": 4.289184251970608e-05,
+      "loss": 0.609,
+      "step": 2019
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10525099188089371,
+      "learning_rate": 4.280285601254952e-05,
+      "loss": 0.6324,
+      "step": 2020
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.1033729687333107,
+      "learning_rate": 4.271393676852039e-05,
+      "loss": 0.5736,
+      "step": 2021
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10033535212278366,
+      "learning_rate": 4.2625084892186616e-05,
+      "loss": 0.5992,
+      "step": 2022
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10182648152112961,
+      "learning_rate": 4.2536300488036906e-05,
+      "loss": 0.5816,
+      "step": 2023
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10433407127857208,
+      "learning_rate": 4.2447583660480605e-05,
+      "loss": 0.5528,
+      "step": 2024
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10535496473312378,
+      "learning_rate": 4.235893451384759e-05,
+      "loss": 0.5764,
+      "step": 2025
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10799122601747513,
+      "learning_rate": 4.227035315238815e-05,
+      "loss": 0.6081,
+      "step": 2026
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10689231008291245,
+      "learning_rate": 4.218183968027287e-05,
+      "loss": 0.5704,
+      "step": 2027
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.10079720616340637,
+      "learning_rate": 4.2093394201592496e-05,
+      "loss": 0.5725,
+      "step": 2028
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10539403557777405,
+      "learning_rate": 4.200501682035778e-05,
+      "loss": 0.5649,
+      "step": 2029
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.1011606827378273,
+      "learning_rate": 4.1916707640499444e-05,
+      "loss": 0.626,
+      "step": 2030
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10625197738409042,
+      "learning_rate": 4.1828466765867955e-05,
+      "loss": 0.6025,
+      "step": 2031
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10916422307491302,
+      "learning_rate": 4.174029430023352e-05,
+      "loss": 0.6075,
+      "step": 2032
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10839617252349854,
+      "learning_rate": 4.165219034728581e-05,
+      "loss": 0.5931,
+      "step": 2033
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10174981504678726,
+      "learning_rate": 4.156415501063402e-05,
+      "loss": 0.6112,
+      "step": 2034
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.1019379049539566,
+      "learning_rate": 4.147618839380658e-05,
+      "loss": 0.6291,
+      "step": 2035
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10948579013347626,
+      "learning_rate": 4.1388290600251136e-05,
+      "loss": 0.5754,
+      "step": 2036
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10134655982255936,
+      "learning_rate": 4.1300461733334395e-05,
+      "loss": 0.5451,
+      "step": 2037
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.10328393429517746,
+      "learning_rate": 4.1212701896342024e-05,
+      "loss": 0.5622,
+      "step": 2038
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.10247175395488739,
+      "learning_rate": 4.112501119247848e-05,
+      "loss": 0.6016,
+      "step": 2039
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.09854919463396072,
+      "learning_rate": 4.1037389724866935e-05,
+      "loss": 0.5438,
+      "step": 2040
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.10441790521144867,
+      "learning_rate": 4.094983759654916e-05,
+      "loss": 0.5923,
+      "step": 2041
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.10846678912639618,
+      "learning_rate": 4.086235491048535e-05,
+      "loss": 0.6536,
+      "step": 2042
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.11094819009304047,
+      "learning_rate": 4.0774941769554044e-05,
+      "loss": 0.6495,
+      "step": 2043
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.10526743531227112,
+      "learning_rate": 4.0687598276552016e-05,
+      "loss": 0.574,
+      "step": 2044
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.10770121961832047,
+      "learning_rate": 4.060032453419413e-05,
+      "loss": 0.5571,
+      "step": 2045
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.10666481405496597,
+      "learning_rate": 4.051312064511321e-05,
+      "loss": 0.5927,
+      "step": 2046
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.10898322612047195,
+      "learning_rate": 4.0425986711859944e-05,
+      "loss": 0.6415,
+      "step": 2047
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10477761179208755,
+      "learning_rate": 4.033892283690274e-05,
+      "loss": 0.5971,
+      "step": 2048
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10394734144210815,
+      "learning_rate": 4.025192912262764e-05,
+      "loss": 0.5543,
+      "step": 2049
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10821384936571121,
+      "learning_rate": 4.016500567133816e-05,
+      "loss": 0.6442,
+      "step": 2050
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10183420032262802,
+      "learning_rate": 4.00781525852552e-05,
+      "loss": 0.5757,
+      "step": 2051
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10447313636541367,
+      "learning_rate": 3.99913699665169e-05,
+      "loss": 0.6294,
+      "step": 2052
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10402779281139374,
+      "learning_rate": 3.990465791717853e-05,
+      "loss": 0.5567,
+      "step": 2053
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.1075742244720459,
+      "learning_rate": 3.98180165392124e-05,
+      "loss": 0.621,
+      "step": 2054
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10536019504070282,
+      "learning_rate": 3.973144593450766e-05,
+      "loss": 0.5917,
+      "step": 2055
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.10300448536872864,
+      "learning_rate": 3.964494620487028e-05,
+      "loss": 0.606,
+      "step": 2056
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.11228964477777481,
+      "learning_rate": 3.9558517452022857e-05,
+      "loss": 0.5439,
+      "step": 2057
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.10543844848871231,
+      "learning_rate": 3.9472159777604536e-05,
+      "loss": 0.6111,
+      "step": 2058
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.10760674625635147,
+      "learning_rate": 3.938587328317085e-05,
+      "loss": 0.6008,
+      "step": 2059
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.11331881582736969,
+      "learning_rate": 3.929965807019363e-05,
+      "loss": 0.5594,
+      "step": 2060
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.10293702036142349,
+      "learning_rate": 3.921351424006092e-05,
+      "loss": 0.5317,
+      "step": 2061
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.10818410664796829,
+      "learning_rate": 3.9127441894076764e-05,
+      "loss": 0.607,
+      "step": 2062
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.1045442670583725,
+      "learning_rate": 3.9041441133461175e-05,
+      "loss": 0.5772,
+      "step": 2063
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.10146302729845047,
+      "learning_rate": 3.895551205934999e-05,
+      "loss": 0.5235,
+      "step": 2064
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.11018965393304825,
+      "learning_rate": 3.8869654772794696e-05,
+      "loss": 0.6176,
+      "step": 2065
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.10732845962047577,
+      "learning_rate": 3.878386937476242e-05,
+      "loss": 0.5441,
+      "step": 2066
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.106468066573143,
+      "learning_rate": 3.86981559661357e-05,
+      "loss": 0.5482,
+      "step": 2067
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.10570799559354782,
+      "learning_rate": 3.8612514647712395e-05,
+      "loss": 0.552,
+      "step": 2068
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.11474299430847168,
+      "learning_rate": 3.852694552020576e-05,
+      "loss": 0.623,
+      "step": 2069
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.0995294377207756,
+      "learning_rate": 3.8441448684243886e-05,
+      "loss": 0.5944,
+      "step": 2070
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.10456088185310364,
+      "learning_rate": 3.835602424037005e-05,
+      "loss": 0.5011,
+      "step": 2071
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.10643794387578964,
+      "learning_rate": 3.82706722890423e-05,
+      "loss": 0.632,
+      "step": 2072
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.10490655899047852,
+      "learning_rate": 3.818539293063348e-05,
+      "loss": 0.5561,
+      "step": 2073
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.10175126045942307,
+      "learning_rate": 3.810018626543105e-05,
+      "loss": 0.5671,
+      "step": 2074
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.11086132377386093,
+      "learning_rate": 3.8015052393637e-05,
+      "loss": 0.542,
+      "step": 2075
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.10274316370487213,
+      "learning_rate": 3.792999141536767e-05,
+      "loss": 0.6362,
+      "step": 2076
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10360395908355713,
+      "learning_rate": 3.7845003430653725e-05,
+      "loss": 0.5712,
+      "step": 2077
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.09942255914211273,
+      "learning_rate": 3.7760088539439985e-05,
+      "loss": 0.5589,
+      "step": 2078
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10809047520160675,
+      "learning_rate": 3.767524684158525e-05,
+      "loss": 0.6207,
+      "step": 2079
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10145888477563858,
+      "learning_rate": 3.759047843686241e-05,
+      "loss": 0.5664,
+      "step": 2080
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10236692428588867,
+      "learning_rate": 3.750578342495795e-05,
+      "loss": 0.5962,
+      "step": 2081
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10957232862710953,
+      "learning_rate": 3.7421161905472176e-05,
+      "loss": 0.5551,
+      "step": 2082
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.11291569471359253,
+      "learning_rate": 3.733661397791893e-05,
+      "loss": 0.4689,
+      "step": 2083
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10823626816272736,
+      "learning_rate": 3.7252139741725544e-05,
+      "loss": 0.5461,
+      "step": 2084
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10669641196727753,
+      "learning_rate": 3.716773929623265e-05,
+      "loss": 0.5502,
+      "step": 2085
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.10702969133853912,
+      "learning_rate": 3.708341274069411e-05,
+      "loss": 0.617,
+      "step": 2086
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.1105843037366867,
+      "learning_rate": 3.6999160174276914e-05,
+      "loss": 0.5353,
+      "step": 2087
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.10661062598228455,
+      "learning_rate": 3.691498169606102e-05,
+      "loss": 0.5326,
+      "step": 2088
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.10327868163585663,
+      "learning_rate": 3.683087740503929e-05,
+      "loss": 0.5795,
+      "step": 2089
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.1088809072971344,
+      "learning_rate": 3.674684740011729e-05,
+      "loss": 0.6314,
+      "step": 2090
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.12325094640254974,
+      "learning_rate": 3.6662891780113265e-05,
+      "loss": 0.6059,
+      "step": 2091
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.10361084342002869,
+      "learning_rate": 3.657901064375798e-05,
+      "loss": 0.4702,
+      "step": 2092
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.10436633974313736,
+      "learning_rate": 3.649520408969462e-05,
+      "loss": 0.6298,
+      "step": 2093
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.10383755713701248,
+      "learning_rate": 3.641147221647863e-05,
+      "loss": 0.5774,
+      "step": 2094
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.10497051477432251,
+      "learning_rate": 3.6327815122577644e-05,
+      "loss": 0.6152,
+      "step": 2095
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.10531959682703018,
+      "learning_rate": 3.624423290637137e-05,
+      "loss": 0.593,
+      "step": 2096
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.11289358139038086,
+      "learning_rate": 3.616072566615141e-05,
+      "loss": 0.5328,
+      "step": 2097
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.10298408567905426,
+      "learning_rate": 3.6077293500121336e-05,
+      "loss": 0.5988,
+      "step": 2098
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.10554603487253189,
+      "learning_rate": 3.599393650639624e-05,
+      "loss": 0.6238,
+      "step": 2099
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.11433698236942291,
+      "learning_rate": 3.591065478300295e-05,
+      "loss": 0.5166,
+      "step": 2100
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.10721001774072647,
+      "learning_rate": 3.5827448427879716e-05,
+      "loss": 0.5233,
+      "step": 2101
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.11467213183641434,
+      "learning_rate": 3.574431753887617e-05,
+      "loss": 0.5611,
+      "step": 2102
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.11061830073595047,
+      "learning_rate": 3.566126221375321e-05,
+      "loss": 0.5957,
+      "step": 2103
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.11240416020154953,
+      "learning_rate": 3.557828255018286e-05,
+      "loss": 0.5553,
+      "step": 2104
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.10826044529676437,
+      "learning_rate": 3.549537864574818e-05,
+      "loss": 0.5569,
+      "step": 2105
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.1139349713921547,
+      "learning_rate": 3.5412550597943106e-05,
+      "loss": 0.6552,
+      "step": 2106
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.10119649767875671,
+      "learning_rate": 3.53297985041724e-05,
+      "loss": 0.5699,
+      "step": 2107
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.10227086395025253,
+      "learning_rate": 3.524712246175145e-05,
+      "loss": 0.5524,
+      "step": 2108
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.11410786211490631,
+      "learning_rate": 3.5164522567906364e-05,
+      "loss": 0.6543,
+      "step": 2109
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.11783404648303986,
+      "learning_rate": 3.5081998919773475e-05,
+      "loss": 0.5835,
+      "step": 2110
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.10686619579792023,
+      "learning_rate": 3.499955161439959e-05,
+      "loss": 0.5826,
+      "step": 2111
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.10940331965684891,
+      "learning_rate": 3.4917180748741743e-05,
+      "loss": 0.6086,
+      "step": 2112
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.10842977464199066,
+      "learning_rate": 3.4834886419667e-05,
+      "loss": 0.5225,
+      "step": 2113
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.10937518626451492,
+      "learning_rate": 3.4752668723952496e-05,
+      "loss": 0.6266,
+      "step": 2114
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.1140761747956276,
+      "learning_rate": 3.46705277582852e-05,
+      "loss": 0.6188,
+      "step": 2115
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.10321415215730667,
+      "learning_rate": 3.4588463619261866e-05,
+      "loss": 0.6214,
+      "step": 2116
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.10466670989990234,
+      "learning_rate": 3.450647640338891e-05,
+      "loss": 0.5681,
+      "step": 2117
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.10992269217967987,
+      "learning_rate": 3.442456620708225e-05,
+      "loss": 0.6083,
+      "step": 2118
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.10328445583581924,
+      "learning_rate": 3.434273312666725e-05,
+      "loss": 0.586,
+      "step": 2119
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.10694943368434906,
+      "learning_rate": 3.4260977258378666e-05,
+      "loss": 0.6369,
+      "step": 2120
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.1104106679558754,
+      "learning_rate": 3.4179298698360296e-05,
+      "loss": 0.6094,
+      "step": 2121
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.1093631237745285,
+      "learning_rate": 3.409769754266513e-05,
+      "loss": 0.5775,
+      "step": 2122
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.10708426684141159,
+      "learning_rate": 3.401617388725511e-05,
+      "loss": 0.597,
+      "step": 2123
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.11512976884841919,
+      "learning_rate": 3.393472782800103e-05,
+      "loss": 0.5716,
+      "step": 2124
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.10788550227880478,
+      "learning_rate": 3.385335946068247e-05,
+      "loss": 0.6065,
+      "step": 2125
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.10368785262107849,
+      "learning_rate": 3.377206888098754e-05,
+      "loss": 0.5412,
+      "step": 2126
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.10234733670949936,
+      "learning_rate": 3.3690856184513076e-05,
+      "loss": 0.5808,
+      "step": 2127
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.09982272237539291,
+      "learning_rate": 3.3609721466764074e-05,
+      "loss": 0.5127,
+      "step": 2128
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.10406281054019928,
+      "learning_rate": 3.352866482315401e-05,
+      "loss": 0.516,
+      "step": 2129
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.10658598691225052,
+      "learning_rate": 3.344768634900446e-05,
+      "loss": 0.5896,
+      "step": 2130
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.107061006128788,
+      "learning_rate": 3.3366786139545116e-05,
+      "loss": 0.5577,
+      "step": 2131
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.10693950206041336,
+      "learning_rate": 3.3285964289913595e-05,
+      "loss": 0.6025,
+      "step": 2132
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.10833391547203064,
+      "learning_rate": 3.320522089515541e-05,
+      "loss": 0.5991,
+      "step": 2133
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.10807321965694427,
+      "learning_rate": 3.3124556050223766e-05,
+      "loss": 0.6187,
+      "step": 2134
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.103877954185009,
+      "learning_rate": 3.30439698499795e-05,
+      "loss": 0.5499,
+      "step": 2135
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10636716336011887,
+      "learning_rate": 3.2963462389191026e-05,
+      "loss": 0.5709,
+      "step": 2136
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10491225868463516,
+      "learning_rate": 3.288303376253403e-05,
+      "loss": 0.5575,
+      "step": 2137
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10422631353139877,
+      "learning_rate": 3.28026840645917e-05,
+      "loss": 0.5745,
+      "step": 2138
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10408491641283035,
+      "learning_rate": 3.272241338985418e-05,
+      "loss": 0.529,
+      "step": 2139
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10744007676839828,
+      "learning_rate": 3.26422218327188e-05,
+      "loss": 0.6624,
+      "step": 2140
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10911328345537186,
+      "learning_rate": 3.256210948748983e-05,
+      "loss": 0.6219,
+      "step": 2141
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.09976132959127426,
+      "learning_rate": 3.2482076448378395e-05,
+      "loss": 0.5019,
+      "step": 2142
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10440421849489212,
+      "learning_rate": 3.2402122809502353e-05,
+      "loss": 0.5522,
+      "step": 2143
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.10908193141222,
+      "learning_rate": 3.232224866488619e-05,
+      "loss": 0.5584,
+      "step": 2144
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.11172449588775635,
+      "learning_rate": 3.2242454108460884e-05,
+      "loss": 0.6643,
+      "step": 2145
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.10436230897903442,
+      "learning_rate": 3.2162739234063856e-05,
+      "loss": 0.5435,
+      "step": 2146
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.10893420875072479,
+      "learning_rate": 3.208310413543879e-05,
+      "loss": 0.5507,
+      "step": 2147
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.10623294860124588,
+      "learning_rate": 3.200354890623555e-05,
+      "loss": 0.5968,
+      "step": 2148
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.1058548167347908,
+      "learning_rate": 3.192407364001017e-05,
+      "loss": 0.5894,
+      "step": 2149
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.10474307835102081,
+      "learning_rate": 3.18446784302245e-05,
+      "loss": 0.5165,
+      "step": 2150
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.11115308105945587,
+      "learning_rate": 3.176536337024631e-05,
+      "loss": 0.6179,
+      "step": 2151
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.11214645206928253,
+      "learning_rate": 3.168612855334916e-05,
+      "loss": 0.6108,
+      "step": 2152
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.10299092531204224,
+      "learning_rate": 3.1606974072712195e-05,
+      "loss": 0.5712,
+      "step": 2153
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.10900414735078812,
+      "learning_rate": 3.152790002142009e-05,
+      "loss": 0.5729,
+      "step": 2154
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.11017811298370361,
+      "learning_rate": 3.1448906492462895e-05,
+      "loss": 0.6124,
+      "step": 2155
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.10492198914289474,
+      "learning_rate": 3.136999357873614e-05,
+      "loss": 0.5869,
+      "step": 2156
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.10835305601358414,
+      "learning_rate": 3.129116137304031e-05,
+      "loss": 0.5703,
+      "step": 2157
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.10911907255649567,
+      "learning_rate": 3.121240996808115e-05,
+      "loss": 0.635,
+      "step": 2158
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.11002764105796814,
+      "learning_rate": 3.113373945646927e-05,
+      "loss": 0.5868,
+      "step": 2159
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.10397159308195114,
+      "learning_rate": 3.1055149930720305e-05,
+      "loss": 0.5284,
+      "step": 2160
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.10380441695451736,
+      "learning_rate": 3.0976641483254485e-05,
+      "loss": 0.576,
+      "step": 2161
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.1092744991183281,
+      "learning_rate": 3.089821420639679e-05,
+      "loss": 0.5491,
+      "step": 2162
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.10718989372253418,
+      "learning_rate": 3.081986819237671e-05,
+      "loss": 0.6361,
+      "step": 2163
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.11400022357702255,
+      "learning_rate": 3.0741603533328175e-05,
+      "loss": 0.6546,
+      "step": 2164
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.10823620855808258,
+      "learning_rate": 3.066342032128946e-05,
+      "loss": 0.5381,
+      "step": 2165
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.10693760961294174,
+      "learning_rate": 3.058531864820301e-05,
+      "loss": 0.5581,
+      "step": 2166
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.10529565066099167,
+      "learning_rate": 3.050729860591551e-05,
+      "loss": 0.56,
+      "step": 2167
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.11990929394960403,
+      "learning_rate": 3.0429360286177465e-05,
+      "loss": 0.6193,
+      "step": 2168
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.10708729922771454,
+      "learning_rate": 3.0351503780643386e-05,
+      "loss": 0.6205,
+      "step": 2169
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.10364414751529694,
+      "learning_rate": 3.0273729180871547e-05,
+      "loss": 0.5872,
+      "step": 2170
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.12280042469501495,
+      "learning_rate": 3.019603657832397e-05,
+      "loss": 0.6049,
+      "step": 2171
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.10781248658895493,
+      "learning_rate": 3.0118426064366112e-05,
+      "loss": 0.5479,
+      "step": 2172
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.11197660863399506,
+      "learning_rate": 3.0040897730266993e-05,
+      "loss": 0.6578,
+      "step": 2173
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.11088935285806656,
+      "learning_rate": 2.996345166719897e-05,
+      "loss": 0.5763,
+      "step": 2174
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.10757026076316833,
+      "learning_rate": 2.988608796623764e-05,
+      "loss": 0.5887,
+      "step": 2175
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.1040201336145401,
+      "learning_rate": 2.980880671836176e-05,
+      "loss": 0.5237,
+      "step": 2176
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.11220099031925201,
+      "learning_rate": 2.9731608014453073e-05,
+      "loss": 0.5879,
+      "step": 2177
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.10947141796350479,
+      "learning_rate": 2.965449194529638e-05,
+      "loss": 0.6328,
+      "step": 2178
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.10400321334600449,
+      "learning_rate": 2.9577458601579134e-05,
+      "loss": 0.4626,
+      "step": 2179
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.10130538046360016,
+      "learning_rate": 2.950050807389161e-05,
+      "loss": 0.4944,
+      "step": 2180
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.11106161028146744,
+      "learning_rate": 2.9423640452726676e-05,
+      "loss": 0.6256,
+      "step": 2181
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.11965084075927734,
+      "learning_rate": 2.934685582847968e-05,
+      "loss": 0.5942,
+      "step": 2182
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.1107841357588768,
+      "learning_rate": 2.9270154291448405e-05,
+      "loss": 0.6392,
+      "step": 2183
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.11000393331050873,
+      "learning_rate": 2.9193535931832848e-05,
+      "loss": 0.5245,
+      "step": 2184
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.11273583024740219,
+      "learning_rate": 2.911700083973534e-05,
+      "loss": 0.6261,
+      "step": 2185
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.11273231357336044,
+      "learning_rate": 2.904054910516012e-05,
+      "loss": 0.5406,
+      "step": 2186
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.11052940785884857,
+      "learning_rate": 2.8964180818013475e-05,
+      "loss": 0.6088,
+      "step": 2187
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 0.8577739596366882,
+      "eval_runtime": 111.8942,
+      "eval_samples_per_second": 88.816,
+      "eval_steps_per_second": 22.208,
+      "step": 2187
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.10331575572490692,
+      "learning_rate": 2.8887896068103547e-05,
+      "loss": 0.4955,
+      "step": 2188
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.11189525574445724,
+      "learning_rate": 2.8811694945140343e-05,
+      "loss": 0.6384,
+      "step": 2189
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1130823940038681,
+      "learning_rate": 2.8735577538735337e-05,
+      "loss": 0.528,
+      "step": 2190
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1075650006532669,
+      "learning_rate": 2.8659543938401666e-05,
+      "loss": 0.6076,
+      "step": 2191
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.1098262369632721,
+      "learning_rate": 2.858359423355392e-05,
+      "loss": 0.5885,
+      "step": 2192
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.10681478679180145,
+      "learning_rate": 2.850772851350798e-05,
+      "loss": 0.5966,
+      "step": 2193
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.11558255553245544,
+      "learning_rate": 2.8431946867481007e-05,
+      "loss": 0.6048,
+      "step": 2194
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.11117714643478394,
+      "learning_rate": 2.8356249384591216e-05,
+      "loss": 0.6637,
+      "step": 2195
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.10714937746524811,
+      "learning_rate": 2.828063615385802e-05,
+      "loss": 0.5821,
+      "step": 2196
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.10461150854825974,
+      "learning_rate": 2.8205107264201524e-05,
+      "loss": 0.5846,
+      "step": 2197
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.11354757100343704,
+      "learning_rate": 2.812966280444279e-05,
+      "loss": 0.5649,
+      "step": 2198
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.10682202875614166,
+      "learning_rate": 2.805430286330355e-05,
+      "loss": 0.6166,
+      "step": 2199
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.1067909449338913,
+      "learning_rate": 2.7979027529406232e-05,
+      "loss": 0.5679,
+      "step": 2200
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.10288982093334198,
+      "learning_rate": 2.7903836891273584e-05,
+      "loss": 0.6029,
+      "step": 2201
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.10488040745258331,
+      "learning_rate": 2.7828731037328913e-05,
+      "loss": 0.5403,
+      "step": 2202
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.10360350459814072,
+      "learning_rate": 2.7753710055895754e-05,
+      "loss": 0.5459,
+      "step": 2203
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.10860137641429901,
+      "learning_rate": 2.7678774035197853e-05,
+      "loss": 0.6116,
+      "step": 2204
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.10692019760608673,
+      "learning_rate": 2.7603923063359018e-05,
+      "loss": 0.5849,
+      "step": 2205
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.1102878749370575,
+      "learning_rate": 2.7529157228403047e-05,
+      "loss": 0.631,
+      "step": 2206
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.11322438716888428,
+      "learning_rate": 2.74544766182537e-05,
+      "loss": 0.6588,
+      "step": 2207
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.10744564235210419,
+      "learning_rate": 2.737988132073437e-05,
+      "loss": 0.6803,
+      "step": 2208
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.10900402814149857,
+      "learning_rate": 2.7305371423568216e-05,
+      "loss": 0.5589,
+      "step": 2209
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.10531095415353775,
+      "learning_rate": 2.7230947014377916e-05,
+      "loss": 0.5817,
+      "step": 2210
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.10973135381937027,
+      "learning_rate": 2.715660818068574e-05,
+      "loss": 0.5588,
+      "step": 2211
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.1086968332529068,
+      "learning_rate": 2.7082355009913142e-05,
+      "loss": 0.5785,
+      "step": 2212
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.11053763329982758,
+      "learning_rate": 2.7008187589380918e-05,
+      "loss": 0.6177,
+      "step": 2213
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.11204681545495987,
+      "learning_rate": 2.6934106006309125e-05,
+      "loss": 0.5543,
+      "step": 2214
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.11255240440368652,
+      "learning_rate": 2.686011034781668e-05,
+      "loss": 0.6466,
+      "step": 2215
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.11382347345352173,
+      "learning_rate": 2.678620070092158e-05,
+      "loss": 0.6611,
+      "step": 2216
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.11264553666114807,
+      "learning_rate": 2.6712377152540648e-05,
+      "loss": 0.6416,
+      "step": 2217
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.12115882337093353,
+      "learning_rate": 2.6638639789489516e-05,
+      "loss": 0.6289,
+      "step": 2218
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.10413646697998047,
+      "learning_rate": 2.656498869848232e-05,
+      "loss": 0.566,
+      "step": 2219
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.10919666290283203,
+      "learning_rate": 2.6491423966131856e-05,
+      "loss": 0.5854,
+      "step": 2220
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.11465396732091904,
+      "learning_rate": 2.6417945678949328e-05,
+      "loss": 0.5798,
+      "step": 2221
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.10808613896369934,
+      "learning_rate": 2.6344553923344283e-05,
+      "loss": 0.5593,
+      "step": 2222
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.10671836882829666,
+      "learning_rate": 2.6271248785624524e-05,
+      "loss": 0.6108,
+      "step": 2223
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.11143945157527924,
+      "learning_rate": 2.619803035199593e-05,
+      "loss": 0.5984,
+      "step": 2224
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.10562798380851746,
+      "learning_rate": 2.612489870856256e-05,
+      "loss": 0.4939,
+      "step": 2225
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.11124252527952194,
+      "learning_rate": 2.6051853941326232e-05,
+      "loss": 0.571,
+      "step": 2226
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.10681876540184021,
+      "learning_rate": 2.5978896136186705e-05,
+      "loss": 0.5418,
+      "step": 2227
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.11089930683374405,
+      "learning_rate": 2.5906025378941423e-05,
+      "loss": 0.6097,
+      "step": 2228
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.11197752505540848,
+      "learning_rate": 2.5833241755285565e-05,
+      "loss": 0.6439,
+      "step": 2229
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.11759854108095169,
+      "learning_rate": 2.576054535081167e-05,
+      "loss": 0.5491,
+      "step": 2230
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.11571485549211502,
+      "learning_rate": 2.5687936251009825e-05,
+      "loss": 0.5897,
+      "step": 2231
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.10776043683290482,
+      "learning_rate": 2.56154145412675e-05,
+      "loss": 0.5538,
+      "step": 2232
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.10648128390312195,
+      "learning_rate": 2.5542980306869223e-05,
+      "loss": 0.5294,
+      "step": 2233
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.11344112455844879,
+      "learning_rate": 2.5470633632996798e-05,
+      "loss": 0.6019,
+      "step": 2234
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.10959113389253616,
+      "learning_rate": 2.539837460472898e-05,
+      "loss": 0.5979,
+      "step": 2235
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.10097500681877136,
+      "learning_rate": 2.5326203307041552e-05,
+      "loss": 0.5485,
+      "step": 2236
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.10733914375305176,
+      "learning_rate": 2.5254119824806997e-05,
+      "loss": 0.5888,
+      "step": 2237
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.12054766714572906,
+      "learning_rate": 2.51821242427946e-05,
+      "loss": 0.5211,
+      "step": 2238
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.10666290670633316,
+      "learning_rate": 2.5110216645670247e-05,
+      "loss": 0.5903,
+      "step": 2239
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.11026865988969803,
+      "learning_rate": 2.503839711799646e-05,
+      "loss": 0.5783,
+      "step": 2240
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.10749930888414383,
+      "learning_rate": 2.496666574423201e-05,
+      "loss": 0.5669,
+      "step": 2241
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11575426906347275,
+      "learning_rate": 2.489502260873212e-05,
+      "loss": 0.6043,
+      "step": 2242
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.10946560651063919,
+      "learning_rate": 2.482346779574829e-05,
+      "loss": 0.5798,
+      "step": 2243
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11224581301212311,
+      "learning_rate": 2.4752001389428014e-05,
+      "loss": 0.5796,
+      "step": 2244
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11278235167264938,
+      "learning_rate": 2.468062347381491e-05,
+      "loss": 0.5726,
+      "step": 2245
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11235585808753967,
+      "learning_rate": 2.4609334132848493e-05,
+      "loss": 0.5869,
+      "step": 2246
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11066693812608719,
+      "learning_rate": 2.453813345036423e-05,
+      "loss": 0.5835,
+      "step": 2247
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11173537373542786,
+      "learning_rate": 2.4467021510093136e-05,
+      "loss": 0.597,
+      "step": 2248
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11050070822238922,
+      "learning_rate": 2.439599839566201e-05,
+      "loss": 0.5608,
+      "step": 2249
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.11474709212779999,
+      "learning_rate": 2.4325064190593106e-05,
+      "loss": 0.6124,
+      "step": 2250
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.12037473171949387,
+      "learning_rate": 2.4254218978304255e-05,
+      "loss": 0.647,
+      "step": 2251
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.11300857365131378,
+      "learning_rate": 2.418346284210845e-05,
+      "loss": 0.6662,
+      "step": 2252
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.11486240476369858,
+      "learning_rate": 2.4112795865214046e-05,
+      "loss": 0.5858,
+      "step": 2253
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.1133425161242485,
+      "learning_rate": 2.404221813072458e-05,
+      "loss": 0.6136,
+      "step": 2254
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.11130309849977493,
+      "learning_rate": 2.3971729721638524e-05,
+      "loss": 0.6833,
+      "step": 2255
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.10779284685850143,
+      "learning_rate": 2.3901330720849392e-05,
+      "loss": 0.5987,
+      "step": 2256
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.10761451721191406,
+      "learning_rate": 2.3831021211145487e-05,
+      "loss": 0.5504,
+      "step": 2257
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.1140928566455841,
+      "learning_rate": 2.376080127521002e-05,
+      "loss": 0.5765,
+      "step": 2258
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.10785611718893051,
+      "learning_rate": 2.3690670995620668e-05,
+      "loss": 0.6142,
+      "step": 2259
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.11109215021133423,
+      "learning_rate": 2.3620630454849758e-05,
+      "loss": 0.5675,
+      "step": 2260
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.11377670615911484,
+      "learning_rate": 2.355067973526418e-05,
+      "loss": 0.6017,
+      "step": 2261
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.10863988101482391,
+      "learning_rate": 2.3480818919125026e-05,
+      "loss": 0.565,
+      "step": 2262
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.10676811635494232,
+      "learning_rate": 2.3411048088587762e-05,
+      "loss": 0.552,
+      "step": 2263
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.1048518717288971,
+      "learning_rate": 2.3341367325702003e-05,
+      "loss": 0.5849,
+      "step": 2264
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.1173892617225647,
+      "learning_rate": 2.3271776712411518e-05,
+      "loss": 0.5725,
+      "step": 2265
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.10788566619157791,
+      "learning_rate": 2.320227633055394e-05,
+      "loss": 0.604,
+      "step": 2266
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.11027087271213531,
+      "learning_rate": 2.3132866261860863e-05,
+      "loss": 0.602,
+      "step": 2267
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.10224713385105133,
+      "learning_rate": 2.306354658795764e-05,
+      "loss": 0.5785,
+      "step": 2268
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.11403778940439224,
+      "learning_rate": 2.299431739036342e-05,
+      "loss": 0.5869,
+      "step": 2269
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.11354638636112213,
+      "learning_rate": 2.2925178750490793e-05,
+      "loss": 0.6146,
+      "step": 2270
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.09810806810855865,
+      "learning_rate": 2.2856130749645942e-05,
+      "loss": 0.5828,
+      "step": 2271
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.10592757910490036,
+      "learning_rate": 2.2787173469028533e-05,
+      "loss": 0.6504,
+      "step": 2272
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.11044619977474213,
+      "learning_rate": 2.2718306989731387e-05,
+      "loss": 0.5564,
+      "step": 2273
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.1082218736410141,
+      "learning_rate": 2.264953139274065e-05,
+      "loss": 0.5176,
+      "step": 2274
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.11557526141405106,
+      "learning_rate": 2.2580846758935535e-05,
+      "loss": 0.6407,
+      "step": 2275
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.11399544775485992,
+      "learning_rate": 2.2512253169088403e-05,
+      "loss": 0.5726,
+      "step": 2276
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.10828489065170288,
+      "learning_rate": 2.2443750703864376e-05,
+      "loss": 0.5454,
+      "step": 2277
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.11010563373565674,
+      "learning_rate": 2.237533944382153e-05,
+      "loss": 0.587,
+      "step": 2278
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.11093679815530777,
+      "learning_rate": 2.2307019469410606e-05,
+      "loss": 0.5828,
+      "step": 2279
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.11040529608726501,
+      "learning_rate": 2.2238790860975144e-05,
+      "loss": 0.6292,
+      "step": 2280
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.10993993282318115,
+      "learning_rate": 2.2170653698751044e-05,
+      "loss": 0.6069,
+      "step": 2281
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.1121375635266304,
+      "learning_rate": 2.2102608062866758e-05,
+      "loss": 0.6031,
+      "step": 2282
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.10904592275619507,
+      "learning_rate": 2.2034654033343183e-05,
+      "loss": 0.6174,
+      "step": 2283
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.11172360181808472,
+      "learning_rate": 2.1966791690093348e-05,
+      "loss": 0.5759,
+      "step": 2284
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.10993849486112595,
+      "learning_rate": 2.1899021112922525e-05,
+      "loss": 0.5572,
+      "step": 2285
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.1052040308713913,
+      "learning_rate": 2.183134238152805e-05,
+      "loss": 0.5826,
+      "step": 2286
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.10786193609237671,
+      "learning_rate": 2.1763755575499355e-05,
+      "loss": 0.5346,
+      "step": 2287
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.11253299564123154,
+      "learning_rate": 2.169626077431758e-05,
+      "loss": 0.5709,
+      "step": 2288
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.11083231121301651,
+      "learning_rate": 2.1628858057355783e-05,
+      "loss": 0.5731,
+      "step": 2289
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.11575774103403091,
+      "learning_rate": 2.1561547503878777e-05,
+      "loss": 0.6008,
+      "step": 2290
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.10586890578269958,
+      "learning_rate": 2.1494329193042916e-05,
+      "loss": 0.5092,
+      "step": 2291
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.10483654588460922,
+      "learning_rate": 2.1427203203896052e-05,
+      "loss": 0.5117,
+      "step": 2292
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.10706888139247894,
+      "learning_rate": 2.136016961537751e-05,
+      "loss": 0.5994,
+      "step": 2293
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.12088432908058167,
+      "learning_rate": 2.129322850631803e-05,
+      "loss": 0.6985,
+      "step": 2294
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.10278206318616867,
+      "learning_rate": 2.1226379955439446e-05,
+      "loss": 0.5754,
+      "step": 2295
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.1108093112707138,
+      "learning_rate": 2.115962404135484e-05,
+      "loss": 0.6059,
+      "step": 2296
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.1103772297501564,
+      "learning_rate": 2.109296084256831e-05,
+      "loss": 0.5146,
+      "step": 2297
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.10462667793035507,
+      "learning_rate": 2.102639043747502e-05,
+      "loss": 0.5426,
+      "step": 2298
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.10616558790206909,
+      "learning_rate": 2.0959912904360867e-05,
+      "loss": 0.6338,
+      "step": 2299
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.11522235721349716,
+      "learning_rate": 2.089352832140259e-05,
+      "loss": 0.6269,
+      "step": 2300
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.10954602807760239,
+      "learning_rate": 2.082723676666769e-05,
+      "loss": 0.5534,
+      "step": 2301
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.10834358632564545,
+      "learning_rate": 2.07610383181142e-05,
+      "loss": 0.5575,
+      "step": 2302
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.11558464914560318,
+      "learning_rate": 2.0694933053590627e-05,
+      "loss": 0.6484,
+      "step": 2303
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.11200366914272308,
+      "learning_rate": 2.0628921050835925e-05,
+      "loss": 0.6204,
+      "step": 2304
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.10741502791643143,
+      "learning_rate": 2.056300238747948e-05,
+      "loss": 0.4585,
+      "step": 2305
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.11143650859594345,
+      "learning_rate": 2.0497177141040725e-05,
+      "loss": 0.607,
+      "step": 2306
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.10890121012926102,
+      "learning_rate": 2.043144538892936e-05,
+      "loss": 0.6059,
+      "step": 2307
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.11081679165363312,
+      "learning_rate": 2.0365807208445074e-05,
+      "loss": 0.5899,
+      "step": 2308
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.11185185611248016,
+      "learning_rate": 2.0300262676777626e-05,
+      "loss": 0.6289,
+      "step": 2309
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.10670889168977737,
+      "learning_rate": 2.02348118710065e-05,
+      "loss": 0.5021,
+      "step": 2310
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.11716235429048538,
+      "learning_rate": 2.0169454868101023e-05,
+      "loss": 0.6286,
+      "step": 2311
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.12103371322154999,
+      "learning_rate": 2.0104191744920288e-05,
+      "loss": 0.5999,
+      "step": 2312
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.10644591599702835,
+      "learning_rate": 2.003902257821283e-05,
+      "loss": 0.6153,
+      "step": 2313
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.11589517444372177,
+      "learning_rate": 1.9973947444616814e-05,
+      "loss": 0.5952,
+      "step": 2314
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.11601000279188156,
+      "learning_rate": 1.9908966420659746e-05,
+      "loss": 0.5918,
+      "step": 2315
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.11709819734096527,
+      "learning_rate": 1.984407958275859e-05,
+      "loss": 0.4939,
+      "step": 2316
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.1103871688246727,
+      "learning_rate": 1.977928700721937e-05,
+      "loss": 0.5342,
+      "step": 2317
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.10299322009086609,
+      "learning_rate": 1.9714588770237352e-05,
+      "loss": 0.5938,
+      "step": 2318
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.11013869941234589,
+      "learning_rate": 1.9649984947896915e-05,
+      "loss": 0.5979,
+      "step": 2319
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.11123698204755783,
+      "learning_rate": 1.958547561617132e-05,
+      "loss": 0.61,
+      "step": 2320
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.10424362123012543,
+      "learning_rate": 1.952106085092271e-05,
+      "loss": 0.5745,
+      "step": 2321
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.10567298531532288,
+      "learning_rate": 1.9456740727902045e-05,
+      "loss": 0.5223,
+      "step": 2322
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.12392066419124603,
+      "learning_rate": 1.939251532274905e-05,
+      "loss": 0.4739,
+      "step": 2323
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.11293203383684158,
+      "learning_rate": 1.932838471099193e-05,
+      "loss": 0.6085,
+      "step": 2324
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.11288765072822571,
+      "learning_rate": 1.9264348968047507e-05,
+      "loss": 0.5985,
+      "step": 2325
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.10648974031209946,
+      "learning_rate": 1.9200408169220986e-05,
+      "loss": 0.5726,
+      "step": 2326
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.11502914875745773,
+      "learning_rate": 1.9136562389706037e-05,
+      "loss": 0.6022,
+      "step": 2327
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.11790643632411957,
+      "learning_rate": 1.9072811704584416e-05,
+      "loss": 0.5842,
+      "step": 2328
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.12253357470035553,
+      "learning_rate": 1.9009156188826128e-05,
+      "loss": 0.6484,
+      "step": 2329
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.11219343543052673,
+      "learning_rate": 1.8945595917289326e-05,
+      "loss": 0.5981,
+      "step": 2330
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.11103370040655136,
+      "learning_rate": 1.88821309647201e-05,
+      "loss": 0.5743,
+      "step": 2331
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.1079200804233551,
+      "learning_rate": 1.8818761405752373e-05,
+      "loss": 0.5569,
+      "step": 2332
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.10873348265886307,
+      "learning_rate": 1.875548731490796e-05,
+      "loss": 0.5897,
+      "step": 2333
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.1107969582080841,
+      "learning_rate": 1.86923087665965e-05,
+      "loss": 0.6159,
+      "step": 2334
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.11332055926322937,
+      "learning_rate": 1.862922583511507e-05,
+      "loss": 0.6554,
+      "step": 2335
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.1070738434791565,
+      "learning_rate": 1.8566238594648455e-05,
+      "loss": 0.5856,
+      "step": 2336
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.11095769703388214,
+      "learning_rate": 1.8503347119268834e-05,
+      "loss": 0.6234,
+      "step": 2337
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.11429719626903534,
+      "learning_rate": 1.8440551482935876e-05,
+      "loss": 0.658,
+      "step": 2338
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.11155037581920624,
+      "learning_rate": 1.837785175949638e-05,
+      "loss": 0.6122,
+      "step": 2339
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.10984606295824051,
+      "learning_rate": 1.831524802268445e-05,
+      "loss": 0.5952,
+      "step": 2340
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.11355830729007721,
+      "learning_rate": 1.8252740346121344e-05,
+      "loss": 0.5783,
+      "step": 2341
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.11064739525318146,
+      "learning_rate": 1.8190328803315304e-05,
+      "loss": 0.5594,
+      "step": 2342
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.11037656664848328,
+      "learning_rate": 1.8128013467661485e-05,
+      "loss": 0.6119,
+      "step": 2343
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.10697721689939499,
+      "learning_rate": 1.8065794412441937e-05,
+      "loss": 0.5794,
+      "step": 2344
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.11225653439760208,
+      "learning_rate": 1.8003671710825576e-05,
+      "loss": 0.636,
+      "step": 2345
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.11298608034849167,
+      "learning_rate": 1.7941645435867838e-05,
+      "loss": 0.6186,
+      "step": 2346
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.1135297417640686,
+      "learning_rate": 1.7879715660510843e-05,
+      "loss": 0.6218,
+      "step": 2347
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.11299118399620056,
+      "learning_rate": 1.78178824575833e-05,
+      "loss": 0.6276,
+      "step": 2348
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.11816900968551636,
+      "learning_rate": 1.7756145899800292e-05,
+      "loss": 0.5468,
+      "step": 2349
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.1023438572883606,
+      "learning_rate": 1.769450605976316e-05,
+      "loss": 0.5529,
+      "step": 2350
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.11418010294437408,
+      "learning_rate": 1.76329630099596e-05,
+      "loss": 0.6445,
+      "step": 2351
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.11138451844453812,
+      "learning_rate": 1.7571516822763566e-05,
+      "loss": 0.5286,
+      "step": 2352
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.10600920021533966,
+      "learning_rate": 1.7510167570434898e-05,
+      "loss": 0.5668,
+      "step": 2353
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.11446873098611832,
+      "learning_rate": 1.7448915325119596e-05,
+      "loss": 0.603,
+      "step": 2354
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.10648351907730103,
+      "learning_rate": 1.738776015884951e-05,
+      "loss": 0.5847,
+      "step": 2355
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.10541409999132156,
+      "learning_rate": 1.732670214354243e-05,
+      "loss": 0.5322,
+      "step": 2356
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.107686847448349,
+      "learning_rate": 1.726574135100173e-05,
+      "loss": 0.5418,
+      "step": 2357
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.10671236366033554,
+      "learning_rate": 1.7204877852916545e-05,
+      "loss": 0.5728,
+      "step": 2358
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.10845701396465302,
+      "learning_rate": 1.7144111720861654e-05,
+      "loss": 0.6356,
+      "step": 2359
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.11120231449604034,
+      "learning_rate": 1.7083443026297263e-05,
+      "loss": 0.5314,
+      "step": 2360
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.11589409410953522,
+      "learning_rate": 1.7022871840568944e-05,
+      "loss": 0.5284,
+      "step": 2361
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.11374478787183762,
+      "learning_rate": 1.696239823490766e-05,
+      "loss": 0.5841,
+      "step": 2362
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.11041732877492905,
+      "learning_rate": 1.6902022280429687e-05,
+      "loss": 0.5457,
+      "step": 2363
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.10890837013721466,
+      "learning_rate": 1.6841744048136322e-05,
+      "loss": 0.6171,
+      "step": 2364
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.11266002804040909,
+      "learning_rate": 1.6781563608914043e-05,
+      "loss": 0.5184,
+      "step": 2365
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.11089814454317093,
+      "learning_rate": 1.672148103353426e-05,
+      "loss": 0.6043,
+      "step": 2366
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.10790608078241348,
+      "learning_rate": 1.6661496392653407e-05,
+      "loss": 0.5531,
+      "step": 2367
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.11126933991909027,
+      "learning_rate": 1.660160975681262e-05,
+      "loss": 0.6515,
+      "step": 2368
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.11195553094148636,
+      "learning_rate": 1.6541821196437812e-05,
+      "loss": 0.5477,
+      "step": 2369
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.1129053458571434,
+      "learning_rate": 1.648213078183963e-05,
+      "loss": 0.619,
+      "step": 2370
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.11396722495555878,
+      "learning_rate": 1.642253858321329e-05,
+      "loss": 0.5854,
+      "step": 2371
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.11289696395397186,
+      "learning_rate": 1.6363044670638383e-05,
+      "loss": 0.5908,
+      "step": 2372
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.1255417913198471,
+      "learning_rate": 1.6303649114079033e-05,
+      "loss": 0.5439,
+      "step": 2373
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.10820462554693222,
+      "learning_rate": 1.6244351983383743e-05,
+      "loss": 0.5648,
+      "step": 2374
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.11893779039382935,
+      "learning_rate": 1.6185153348285108e-05,
+      "loss": 0.5734,
+      "step": 2375
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.1113956943154335,
+      "learning_rate": 1.612605327840001e-05,
+      "loss": 0.5823,
+      "step": 2376
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.1090039387345314,
+      "learning_rate": 1.606705184322943e-05,
+      "loss": 0.5783,
+      "step": 2377
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.10794378072023392,
+      "learning_rate": 1.600814911215832e-05,
+      "loss": 0.5986,
+      "step": 2378
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.11214049160480499,
+      "learning_rate": 1.594934515445551e-05,
+      "loss": 0.5425,
+      "step": 2379
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.11800431460142136,
+      "learning_rate": 1.589064003927372e-05,
+      "loss": 0.7081,
+      "step": 2380
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.11361788958311081,
+      "learning_rate": 1.5832033835649472e-05,
+      "loss": 0.5862,
+      "step": 2381
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.11899910867214203,
+      "learning_rate": 1.5773526612502943e-05,
+      "loss": 0.585,
+      "step": 2382
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.11017204821109772,
+      "learning_rate": 1.5715118438637832e-05,
+      "loss": 0.612,
+      "step": 2383
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.10658569633960724,
+      "learning_rate": 1.5656809382741443e-05,
+      "loss": 0.5309,
+      "step": 2384
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.1027158573269844,
+      "learning_rate": 1.559859951338456e-05,
+      "loss": 0.5402,
+      "step": 2385
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.11072217673063278,
+      "learning_rate": 1.554048889902119e-05,
+      "loss": 0.5814,
+      "step": 2386
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.11142212897539139,
+      "learning_rate": 1.54824776079887e-05,
+      "loss": 0.6465,
+      "step": 2387
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.10783372819423676,
+      "learning_rate": 1.5424565708507667e-05,
+      "loss": 0.5934,
+      "step": 2388
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.11221033334732056,
+      "learning_rate": 1.536675326868179e-05,
+      "loss": 0.603,
+      "step": 2389
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.1094745472073555,
+      "learning_rate": 1.5309040356497717e-05,
+      "loss": 0.5593,
+      "step": 2390
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.10428839176893234,
+      "learning_rate": 1.5251427039825117e-05,
+      "loss": 0.6142,
+      "step": 2391
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.11285248398780823,
+      "learning_rate": 1.5193913386416569e-05,
+      "loss": 0.6565,
+      "step": 2392
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.10441061854362488,
+      "learning_rate": 1.5136499463907416e-05,
+      "loss": 0.6393,
+      "step": 2393
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.11005312204360962,
+      "learning_rate": 1.507918533981565e-05,
+      "loss": 0.5398,
+      "step": 2394
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.1100429818034172,
+      "learning_rate": 1.502197108154203e-05,
+      "loss": 0.5563,
+      "step": 2395
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.11301954090595245,
+      "learning_rate": 1.4964856756369804e-05,
+      "loss": 0.5971,
+      "step": 2396
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.11425873637199402,
+      "learning_rate": 1.490784243146468e-05,
+      "loss": 0.5392,
+      "step": 2397
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.11055084317922592,
+      "learning_rate": 1.4850928173874768e-05,
+      "loss": 0.5816,
+      "step": 2398
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.10878891497850418,
+      "learning_rate": 1.4794114050530583e-05,
+      "loss": 0.5843,
+      "step": 2399
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.11173763871192932,
+      "learning_rate": 1.4737400128244827e-05,
+      "loss": 0.6309,
+      "step": 2400
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.12217356264591217,
+      "learning_rate": 1.4680786473712305e-05,
+      "loss": 0.5918,
+      "step": 2401
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.11374884843826294,
+      "learning_rate": 1.4624273153509983e-05,
+      "loss": 0.6303,
+      "step": 2402
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.10755521059036255,
+      "learning_rate": 1.4567860234096886e-05,
+      "loss": 0.5371,
+      "step": 2403
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.10902739316225052,
+      "learning_rate": 1.4511547781813828e-05,
+      "loss": 0.594,
+      "step": 2404
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.11338461190462112,
+      "learning_rate": 1.4455335862883557e-05,
+      "loss": 0.6449,
+      "step": 2405
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.10747576504945755,
+      "learning_rate": 1.4399224543410638e-05,
+      "loss": 0.5825,
+      "step": 2406
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.12629015743732452,
+      "learning_rate": 1.4343213889381268e-05,
+      "loss": 0.6265,
+      "step": 2407
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.11493159085512161,
+      "learning_rate": 1.4287303966663235e-05,
+      "loss": 0.5475,
+      "step": 2408
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.1143827885389328,
+      "learning_rate": 1.4231494841005921e-05,
+      "loss": 0.5692,
+      "step": 2409
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.11103402078151703,
+      "learning_rate": 1.4175786578040195e-05,
+      "loss": 0.6013,
+      "step": 2410
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.11517591029405594,
+      "learning_rate": 1.4120179243278286e-05,
+      "loss": 0.6304,
+      "step": 2411
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.11341342329978943,
+      "learning_rate": 1.4064672902113685e-05,
+      "loss": 0.614,
+      "step": 2412
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.11039137095212936,
+      "learning_rate": 1.4009267619821143e-05,
+      "loss": 0.6061,
+      "step": 2413
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.11438779532909393,
+      "learning_rate": 1.3953963461556651e-05,
+      "loss": 0.5409,
+      "step": 2414
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.11559653282165527,
+      "learning_rate": 1.3898760492357144e-05,
+      "loss": 0.6331,
+      "step": 2415
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.1088695079088211,
+      "learning_rate": 1.3843658777140633e-05,
+      "loss": 0.5829,
+      "step": 2416
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.11227795481681824,
+      "learning_rate": 1.3788658380706088e-05,
+      "loss": 0.5463,
+      "step": 2417
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.107810840010643,
+      "learning_rate": 1.373375936773328e-05,
+      "loss": 0.5512,
+      "step": 2418
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.11541516333818436,
+      "learning_rate": 1.3678961802782719e-05,
+      "loss": 0.5623,
+      "step": 2419
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.10694299638271332,
+      "learning_rate": 1.3624265750295673e-05,
+      "loss": 0.622,
+      "step": 2420
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.11387132853269577,
+      "learning_rate": 1.3569671274594042e-05,
+      "loss": 0.6247,
+      "step": 2421
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.11427866667509079,
+      "learning_rate": 1.3515178439880261e-05,
+      "loss": 0.58,
+      "step": 2422
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.10684357583522797,
+      "learning_rate": 1.3460787310237155e-05,
+      "loss": 0.5802,
+      "step": 2423
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.10446320474147797,
+      "learning_rate": 1.3406497949628071e-05,
+      "loss": 0.5585,
+      "step": 2424
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.11435571312904358,
+      "learning_rate": 1.335231042189663e-05,
+      "loss": 0.6226,
+      "step": 2425
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.1081709936261177,
+      "learning_rate": 1.3298224790766644e-05,
+      "loss": 0.63,
+      "step": 2426
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.10940678417682648,
+      "learning_rate": 1.3244241119842138e-05,
+      "loss": 0.5814,
+      "step": 2427
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.10868065804243088,
+      "learning_rate": 1.3190359472607272e-05,
+      "loss": 0.5354,
+      "step": 2428
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.12045648694038391,
+      "learning_rate": 1.3136579912426206e-05,
+      "loss": 0.5788,
+      "step": 2429
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.10885733366012573,
+      "learning_rate": 1.308290250254297e-05,
+      "loss": 0.6148,
+      "step": 2430
+    },
+    {
+      "epoch": 2.47,
+      "eval_loss": 0.8587498664855957,
+      "eval_runtime": 111.8567,
+      "eval_samples_per_second": 88.846,
+      "eval_steps_per_second": 22.216,
+      "step": 2430
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.10911186784505844,
+      "learning_rate": 1.3029327306081528e-05,
+      "loss": 0.586,
+      "step": 2431
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.10758145153522491,
+      "learning_rate": 1.2975854386045716e-05,
+      "loss": 0.5322,
+      "step": 2432
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.11344973742961884,
+      "learning_rate": 1.2922483805318986e-05,
+      "loss": 0.6419,
+      "step": 2433
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.11634604632854462,
+      "learning_rate": 1.2869215626664422e-05,
+      "loss": 0.5715,
+      "step": 2434
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.11084653437137604,
+      "learning_rate": 1.2816049912724804e-05,
+      "loss": 0.567,
+      "step": 2435
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.1127915009856224,
+      "learning_rate": 1.2762986726022352e-05,
+      "loss": 0.5653,
+      "step": 2436
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11595485359430313,
+      "learning_rate": 1.2710026128958652e-05,
+      "loss": 0.5807,
+      "step": 2437
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11204774677753448,
+      "learning_rate": 1.2657168183814727e-05,
+      "loss": 0.5677,
+      "step": 2438
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11380517482757568,
+      "learning_rate": 1.2604412952750887e-05,
+      "loss": 0.6064,
+      "step": 2439
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11020273715257645,
+      "learning_rate": 1.2551760497806641e-05,
+      "loss": 0.6101,
+      "step": 2440
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.1060190200805664,
+      "learning_rate": 1.2499210880900558e-05,
+      "loss": 0.5852,
+      "step": 2441
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11431319266557693,
+      "learning_rate": 1.2446764163830338e-05,
+      "loss": 0.6565,
+      "step": 2442
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11377350986003876,
+      "learning_rate": 1.2394420408272733e-05,
+      "loss": 0.6057,
+      "step": 2443
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11644969880580902,
+      "learning_rate": 1.2342179675783283e-05,
+      "loss": 0.5769,
+      "step": 2444
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.11489622294902802,
+      "learning_rate": 1.2290042027796434e-05,
+      "loss": 0.6265,
+      "step": 2445
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.10850931704044342,
+      "learning_rate": 1.223800752562545e-05,
+      "loss": 0.5979,
+      "step": 2446
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.11490161716938019,
+      "learning_rate": 1.2186076230462262e-05,
+      "loss": 0.5691,
+      "step": 2447
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.11626631021499634,
+      "learning_rate": 1.2134248203377386e-05,
+      "loss": 0.5532,
+      "step": 2448
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.11156939715147018,
+      "learning_rate": 1.208252350531992e-05,
+      "loss": 0.561,
+      "step": 2449
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.11055629700422287,
+      "learning_rate": 1.2030902197117522e-05,
+      "loss": 0.5947,
+      "step": 2450
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.1139637902379036,
+      "learning_rate": 1.1979384339476186e-05,
+      "loss": 0.6587,
+      "step": 2451
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.10098773241043091,
+      "learning_rate": 1.1927969992980226e-05,
+      "loss": 0.5131,
+      "step": 2452
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.11296376585960388,
+      "learning_rate": 1.1876659218092323e-05,
+      "loss": 0.5932,
+      "step": 2453
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.1126590147614479,
+      "learning_rate": 1.1825452075153321e-05,
+      "loss": 0.5879,
+      "step": 2454
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.11482881009578705,
+      "learning_rate": 1.1774348624382126e-05,
+      "loss": 0.5846,
+      "step": 2455
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.10599429160356522,
+      "learning_rate": 1.1723348925875766e-05,
+      "loss": 0.5799,
+      "step": 2456
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.1136750653386116,
+      "learning_rate": 1.1672453039609289e-05,
+      "loss": 0.5928,
+      "step": 2457
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.11169691383838654,
+      "learning_rate": 1.1621661025435614e-05,
+      "loss": 0.546,
+      "step": 2458
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.1178235411643982,
+      "learning_rate": 1.1570972943085479e-05,
+      "loss": 0.5899,
+      "step": 2459
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.10910765081644058,
+      "learning_rate": 1.1520388852167452e-05,
+      "loss": 0.5601,
+      "step": 2460
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.11287535727024078,
+      "learning_rate": 1.1469908812167806e-05,
+      "loss": 0.6342,
+      "step": 2461
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.11627475917339325,
+      "learning_rate": 1.1419532882450468e-05,
+      "loss": 0.5621,
+      "step": 2462
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.11356250196695328,
+      "learning_rate": 1.1369261122256814e-05,
+      "loss": 0.5674,
+      "step": 2463
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.1633225381374359,
+      "learning_rate": 1.1319093590705887e-05,
+      "loss": 0.5584,
+      "step": 2464
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.1137564405798912,
+      "learning_rate": 1.1269030346794063e-05,
+      "loss": 0.667,
+      "step": 2465
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.11126105487346649,
+      "learning_rate": 1.1219071449395068e-05,
+      "loss": 0.5775,
+      "step": 2466
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.10923659056425095,
+      "learning_rate": 1.116921695725992e-05,
+      "loss": 0.6233,
+      "step": 2467
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.10701790452003479,
+      "learning_rate": 1.1119466929016942e-05,
+      "loss": 0.5704,
+      "step": 2468
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.126361683011055,
+      "learning_rate": 1.1069821423171544e-05,
+      "loss": 0.6708,
+      "step": 2469
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.109717957675457,
+      "learning_rate": 1.1020280498106162e-05,
+      "loss": 0.5584,
+      "step": 2470
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.10662524402141571,
+      "learning_rate": 1.0970844212080322e-05,
+      "loss": 0.6061,
+      "step": 2471
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.11310931295156479,
+      "learning_rate": 1.0921512623230512e-05,
+      "loss": 0.6731,
+      "step": 2472
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.11701517552137375,
+      "learning_rate": 1.087228578957007e-05,
+      "loss": 0.5944,
+      "step": 2473
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.11389590054750443,
+      "learning_rate": 1.0823163768989075e-05,
+      "loss": 0.581,
+      "step": 2474
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.11175854504108429,
+      "learning_rate": 1.0774146619254466e-05,
+      "loss": 0.6076,
+      "step": 2475
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.11090891063213348,
+      "learning_rate": 1.072523439800981e-05,
+      "loss": 0.5624,
+      "step": 2476
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.11080201715230942,
+      "learning_rate": 1.0676427162775215e-05,
+      "loss": 0.6075,
+      "step": 2477
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.11529351770877838,
+      "learning_rate": 1.0627724970947395e-05,
+      "loss": 0.5912,
+      "step": 2478
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.1041053980588913,
+      "learning_rate": 1.0579127879799555e-05,
+      "loss": 0.4984,
+      "step": 2479
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.10762374848127365,
+      "learning_rate": 1.053063594648126e-05,
+      "loss": 0.5384,
+      "step": 2480
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.1082184761762619,
+      "learning_rate": 1.0482249228018371e-05,
+      "loss": 0.5349,
+      "step": 2481
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.11111825704574585,
+      "learning_rate": 1.0433967781313114e-05,
+      "loss": 0.5997,
+      "step": 2482
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.10750050097703934,
+      "learning_rate": 1.038579166314384e-05,
+      "loss": 0.5837,
+      "step": 2483
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.11059720814228058,
+      "learning_rate": 1.0337720930165107e-05,
+      "loss": 0.5842,
+      "step": 2484
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.11312595009803772,
+      "learning_rate": 1.0289755638907428e-05,
+      "loss": 0.59,
+      "step": 2485
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.11058922857046127,
+      "learning_rate": 1.0241895845777438e-05,
+      "loss": 0.6491,
+      "step": 2486
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.10659688711166382,
+      "learning_rate": 1.019414160705765e-05,
+      "loss": 0.5263,
+      "step": 2487
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.1102498471736908,
+      "learning_rate": 1.0146492978906441e-05,
+      "loss": 0.534,
+      "step": 2488
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.11419513821601868,
+      "learning_rate": 1.0098950017357978e-05,
+      "loss": 0.5067,
+      "step": 2489
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.11710280179977417,
+      "learning_rate": 1.0051512778322236e-05,
+      "loss": 0.6487,
+      "step": 2490
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.11812123656272888,
+      "learning_rate": 1.0004181317584827e-05,
+      "loss": 0.5423,
+      "step": 2491
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.18139739334583282,
+      "learning_rate": 9.95695569080688e-06,
+      "loss": 0.5792,
+      "step": 2492
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.11112868040800095,
+      "learning_rate": 9.909835953525226e-06,
+      "loss": 0.5472,
+      "step": 2493
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.10721415281295776,
+      "learning_rate": 9.862822161152064e-06,
+      "loss": 0.5176,
+      "step": 2494
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.1164349690079689,
+      "learning_rate": 9.815914368975e-06,
+      "loss": 0.5814,
+      "step": 2495
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.11200873553752899,
+      "learning_rate": 9.769112632157006e-06,
+      "loss": 0.6139,
+      "step": 2496
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.10787878930568695,
+      "learning_rate": 9.722417005736383e-06,
+      "loss": 0.6545,
+      "step": 2497
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.10708435624837875,
+      "learning_rate": 9.675827544626603e-06,
+      "loss": 0.5666,
+      "step": 2498
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.10606389492750168,
+      "learning_rate": 9.629344303616261e-06,
+      "loss": 0.549,
+      "step": 2499
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.11322688311338425,
+      "learning_rate": 9.582967337369065e-06,
+      "loss": 0.5731,
+      "step": 2500
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.11601674556732178,
+      "learning_rate": 9.536696700423776e-06,
+      "loss": 0.6011,
+      "step": 2501
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.11646999418735504,
+      "learning_rate": 9.49053244719409e-06,
+      "loss": 0.5325,
+      "step": 2502
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.11576241999864578,
+      "learning_rate": 9.444474631968547e-06,
+      "loss": 0.5892,
+      "step": 2503
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11095581203699112,
+      "learning_rate": 9.398523308910601e-06,
+      "loss": 0.5383,
+      "step": 2504
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11017283797264099,
+      "learning_rate": 9.352678532058446e-06,
+      "loss": 0.5641,
+      "step": 2505
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11411051452159882,
+      "learning_rate": 9.306940355324934e-06,
+      "loss": 0.5765,
+      "step": 2506
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11091768741607666,
+      "learning_rate": 9.261308832497583e-06,
+      "loss": 0.5837,
+      "step": 2507
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11221011728048325,
+      "learning_rate": 9.215784017238527e-06,
+      "loss": 0.5958,
+      "step": 2508
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.10979009419679642,
+      "learning_rate": 9.170365963084393e-06,
+      "loss": 0.5806,
+      "step": 2509
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11538652330636978,
+      "learning_rate": 9.12505472344618e-06,
+      "loss": 0.5749,
+      "step": 2510
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11941476911306381,
+      "learning_rate": 9.079850351609398e-06,
+      "loss": 0.5649,
+      "step": 2511
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11119315028190613,
+      "learning_rate": 9.034752900733812e-06,
+      "loss": 0.6115,
+      "step": 2512
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.11279363930225372,
+      "learning_rate": 8.989762423853477e-06,
+      "loss": 0.6451,
+      "step": 2513
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.10634981095790863,
+      "learning_rate": 8.944878973876558e-06,
+      "loss": 0.5799,
+      "step": 2514
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.10545600205659866,
+      "learning_rate": 8.900102603585526e-06,
+      "loss": 0.5365,
+      "step": 2515
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.11213669180870056,
+      "learning_rate": 8.85543336563679e-06,
+      "loss": 0.6361,
+      "step": 2516
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.11364196240901947,
+      "learning_rate": 8.81087131256082e-06,
+      "loss": 0.6063,
+      "step": 2517
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.12257219851016998,
+      "learning_rate": 8.766416496761998e-06,
+      "loss": 0.5199,
+      "step": 2518
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.11508413404226303,
+      "learning_rate": 8.722068970518692e-06,
+      "loss": 0.5844,
+      "step": 2519
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.10994384437799454,
+      "learning_rate": 8.677828785983044e-06,
+      "loss": 0.5633,
+      "step": 2520
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.112838014960289,
+      "learning_rate": 8.63369599518088e-06,
+      "loss": 0.6169,
+      "step": 2521
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.1092621237039566,
+      "learning_rate": 8.589670650011894e-06,
+      "loss": 0.6375,
+      "step": 2522
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.12111138552427292,
+      "learning_rate": 8.545752802249296e-06,
+      "loss": 0.5994,
+      "step": 2523
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.11666639149188995,
+      "learning_rate": 8.501942503539961e-06,
+      "loss": 0.5762,
+      "step": 2524
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.11110684275627136,
+      "learning_rate": 8.4582398054042e-06,
+      "loss": 0.5447,
+      "step": 2525
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.10998429358005524,
+      "learning_rate": 8.414644759235901e-06,
+      "loss": 0.5794,
+      "step": 2526
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.11470309644937515,
+      "learning_rate": 8.371157416302266e-06,
+      "loss": 0.5594,
+      "step": 2527
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.11609251797199249,
+      "learning_rate": 8.327777827743843e-06,
+      "loss": 0.6511,
+      "step": 2528
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.10859417915344238,
+      "learning_rate": 8.284506044574536e-06,
+      "loss": 0.608,
+      "step": 2529
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.10628614574670792,
+      "learning_rate": 8.241342117681382e-06,
+      "loss": 0.5957,
+      "step": 2530
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.11116372048854828,
+      "learning_rate": 8.19828609782467e-06,
+      "loss": 0.5539,
+      "step": 2531
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.11102517694234848,
+      "learning_rate": 8.155338035637661e-06,
+      "loss": 0.5982,
+      "step": 2532
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.11213390529155731,
+      "learning_rate": 8.11249798162681e-06,
+      "loss": 0.6169,
+      "step": 2533
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.11388233304023743,
+      "learning_rate": 8.069765986171485e-06,
+      "loss": 0.6001,
+      "step": 2534
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.11528266221284866,
+      "learning_rate": 8.027142099523932e-06,
+      "loss": 0.5762,
+      "step": 2535
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.11387170106172562,
+      "learning_rate": 7.984626371809323e-06,
+      "loss": 0.6159,
+      "step": 2536
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.109981469810009,
+      "learning_rate": 7.942218853025652e-06,
+      "loss": 0.5774,
+      "step": 2537
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.11311296373605728,
+      "learning_rate": 7.899919593043637e-06,
+      "loss": 0.5422,
+      "step": 2538
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.11253662407398224,
+      "learning_rate": 7.85772864160661e-06,
+      "loss": 0.5821,
+      "step": 2539
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.1171010285615921,
+      "learning_rate": 7.815646048330683e-06,
+      "loss": 0.6443,
+      "step": 2540
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.1133623942732811,
+      "learning_rate": 7.773671862704413e-06,
+      "loss": 0.6202,
+      "step": 2541
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.1137334555387497,
+      "learning_rate": 7.731806134088959e-06,
+      "loss": 0.6209,
+      "step": 2542
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.11073236912488937,
+      "learning_rate": 7.69004891171784e-06,
+      "loss": 0.5792,
+      "step": 2543
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.10371290892362595,
+      "learning_rate": 7.648400244697074e-06,
+      "loss": 0.4989,
+      "step": 2544
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.10923890024423599,
+      "learning_rate": 7.606860182004971e-06,
+      "loss": 0.557,
+      "step": 2545
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.1077064573764801,
+      "learning_rate": 7.5654287724921e-06,
+      "loss": 0.5729,
+      "step": 2546
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.10521820187568665,
+      "learning_rate": 7.524106064881276e-06,
+      "loss": 0.5337,
+      "step": 2547
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.11200764030218124,
+      "learning_rate": 7.482892107767525e-06,
+      "loss": 0.6299,
+      "step": 2548
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.1104995608329773,
+      "learning_rate": 7.441786949617946e-06,
+      "loss": 0.6105,
+      "step": 2549
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.11246494948863983,
+      "learning_rate": 7.400790638771637e-06,
+      "loss": 0.5667,
+      "step": 2550
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.11189506202936172,
+      "learning_rate": 7.3599032234398015e-06,
+      "loss": 0.5911,
+      "step": 2551
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.11209063977003098,
+      "learning_rate": 7.31912475170552e-06,
+      "loss": 0.5577,
+      "step": 2552
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.11177618056535721,
+      "learning_rate": 7.278455271523765e-06,
+      "loss": 0.6044,
+      "step": 2553
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.11701975017786026,
+      "learning_rate": 7.2378948307213035e-06,
+      "loss": 0.6058,
+      "step": 2554
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.1151755303144455,
+      "learning_rate": 7.197443476996746e-06,
+      "loss": 0.6184,
+      "step": 2555
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.11058969795703888,
+      "learning_rate": 7.157101257920396e-06,
+      "loss": 0.6016,
+      "step": 2556
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.11069056391716003,
+      "learning_rate": 7.116868220934126e-06,
+      "loss": 0.5728,
+      "step": 2557
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.11904621124267578,
+      "learning_rate": 7.076744413351555e-06,
+      "loss": 0.7044,
+      "step": 2558
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.1174590215086937,
+      "learning_rate": 7.036729882357762e-06,
+      "loss": 0.6094,
+      "step": 2559
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.11238753795623779,
+      "learning_rate": 6.996824675009339e-06,
+      "loss": 0.5916,
+      "step": 2560
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.11227459460496902,
+      "learning_rate": 6.957028838234281e-06,
+      "loss": 0.5859,
+      "step": 2561
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11649691313505173,
+      "learning_rate": 6.917342418832029e-06,
+      "loss": 0.6621,
+      "step": 2562
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.1114664152264595,
+      "learning_rate": 6.877765463473307e-06,
+      "loss": 0.6558,
+      "step": 2563
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11019684374332428,
+      "learning_rate": 6.838298018700129e-06,
+      "loss": 0.5782,
+      "step": 2564
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11766045540571213,
+      "learning_rate": 6.79894013092568e-06,
+      "loss": 0.6245,
+      "step": 2565
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11511044949293137,
+      "learning_rate": 6.759691846434391e-06,
+      "loss": 0.6836,
+      "step": 2566
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11136568337678909,
+      "learning_rate": 6.720553211381753e-06,
+      "loss": 0.5568,
+      "step": 2567
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.10685116052627563,
+      "learning_rate": 6.681524271794259e-06,
+      "loss": 0.5213,
+      "step": 2568
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11732318252325058,
+      "learning_rate": 6.6426050735695185e-06,
+      "loss": 0.5607,
+      "step": 2569
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11522579193115234,
+      "learning_rate": 6.603795662475987e-06,
+      "loss": 0.6233,
+      "step": 2570
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.11714325845241547,
+      "learning_rate": 6.565096084153066e-06,
+      "loss": 0.6323,
+      "step": 2571
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.11102015525102615,
+      "learning_rate": 6.526506384110953e-06,
+      "loss": 0.5952,
+      "step": 2572
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.11005159467458725,
+      "learning_rate": 6.4880266077306684e-06,
+      "loss": 0.5619,
+      "step": 2573
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.11469392478466034,
+      "learning_rate": 6.4496568002639705e-06,
+      "loss": 0.6466,
+      "step": 2574
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.11261619627475739,
+      "learning_rate": 6.411397006833231e-06,
+      "loss": 0.6101,
+      "step": 2575
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.11512603610754013,
+      "learning_rate": 6.373247272431504e-06,
+      "loss": 0.6057,
+      "step": 2576
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.1119186133146286,
+      "learning_rate": 6.335207641922414e-06,
+      "loss": 0.6121,
+      "step": 2577
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.1133420392870903,
+      "learning_rate": 6.297278160040099e-06,
+      "loss": 0.5755,
+      "step": 2578
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.1077301874756813,
+      "learning_rate": 6.259458871389123e-06,
+      "loss": 0.5307,
+      "step": 2579
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.10843468457460403,
+      "learning_rate": 6.221749820444522e-06,
+      "loss": 0.5665,
+      "step": 2580
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.12884122133255005,
+      "learning_rate": 6.184151051551668e-06,
+      "loss": 0.6238,
+      "step": 2581
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.1070568785071373,
+      "learning_rate": 6.146662608926246e-06,
+      "loss": 0.5822,
+      "step": 2582
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.10766773670911789,
+      "learning_rate": 6.109284536654169e-06,
+      "loss": 0.513,
+      "step": 2583
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.1223384216427803,
+      "learning_rate": 6.072016878691622e-06,
+      "loss": 0.6619,
+      "step": 2584
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.11201731860637665,
+      "learning_rate": 6.034859678864901e-06,
+      "loss": 0.5906,
+      "step": 2585
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.11305366456508636,
+      "learning_rate": 5.997812980870365e-06,
+      "loss": 0.6322,
+      "step": 2586
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.11629898101091385,
+      "learning_rate": 5.960876828274508e-06,
+      "loss": 0.627,
+      "step": 2587
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.12038525193929672,
+      "learning_rate": 5.924051264513786e-06,
+      "loss": 0.6094,
+      "step": 2588
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.11716024577617645,
+      "learning_rate": 5.88733633289461e-06,
+      "loss": 0.6073,
+      "step": 2589
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.11274966597557068,
+      "learning_rate": 5.850732076593224e-06,
+      "loss": 0.5709,
+      "step": 2590
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.11597256362438202,
+      "learning_rate": 5.8142385386558364e-06,
+      "loss": 0.6279,
+      "step": 2591
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.11708580702543259,
+      "learning_rate": 5.777855761998374e-06,
+      "loss": 0.6074,
+      "step": 2592
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.11434788256883621,
+      "learning_rate": 5.741583789406546e-06,
+      "loss": 0.6123,
+      "step": 2593
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.11441222578287125,
+      "learning_rate": 5.705422663535687e-06,
+      "loss": 0.6432,
+      "step": 2594
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.11490438133478165,
+      "learning_rate": 5.66937242691088e-06,
+      "loss": 0.6401,
+      "step": 2595
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.10856378823518753,
+      "learning_rate": 5.633433121926746e-06,
+      "loss": 0.5954,
+      "step": 2596
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.11231663823127747,
+      "learning_rate": 5.597604790847422e-06,
+      "loss": 0.6223,
+      "step": 2597
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.12034507840871811,
+      "learning_rate": 5.5618874758066266e-06,
+      "loss": 0.5597,
+      "step": 2598
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.1154862716794014,
+      "learning_rate": 5.52628121880745e-06,
+      "loss": 0.5135,
+      "step": 2599
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.11328060925006866,
+      "learning_rate": 5.490786061722419e-06,
+      "loss": 0.6006,
+      "step": 2600
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.11254023015499115,
+      "learning_rate": 5.455402046293368e-06,
+      "loss": 0.5919,
+      "step": 2601
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.11789064854383469,
+      "learning_rate": 5.420129214131486e-06,
+      "loss": 0.6148,
+      "step": 2602
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.10514586418867111,
+      "learning_rate": 5.384967606717173e-06,
+      "loss": 0.5447,
+      "step": 2603
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.11419537663459778,
+      "learning_rate": 5.349917265400062e-06,
+      "loss": 0.6156,
+      "step": 2604
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.11480317264795303,
+      "learning_rate": 5.314978231398871e-06,
+      "loss": 0.5601,
+      "step": 2605
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.10798463970422745,
+      "learning_rate": 5.2801505458015095e-06,
+      "loss": 0.5791,
+      "step": 2606
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.10622647404670715,
+      "learning_rate": 5.245434249564884e-06,
+      "loss": 0.5343,
+      "step": 2607
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.11073992401361465,
+      "learning_rate": 5.210829383514914e-06,
+      "loss": 0.5829,
+      "step": 2608
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.1222434788942337,
+      "learning_rate": 5.176335988346515e-06,
+      "loss": 0.5863,
+      "step": 2609
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.11256115138530731,
+      "learning_rate": 5.14195410462347e-06,
+      "loss": 0.5155,
+      "step": 2610
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.1130778044462204,
+      "learning_rate": 5.107683772778471e-06,
+      "loss": 0.6063,
+      "step": 2611
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.11926062405109406,
+      "learning_rate": 5.073525033112958e-06,
+      "loss": 0.6576,
+      "step": 2612
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.11267580837011337,
+      "learning_rate": 5.03947792579722e-06,
+      "loss": 0.6242,
+      "step": 2613
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.10830764472484589,
+      "learning_rate": 5.005542490870207e-06,
+      "loss": 0.6344,
+      "step": 2614
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.10857363790273666,
+      "learning_rate": 4.971718768239586e-06,
+      "loss": 0.5837,
+      "step": 2615
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.10756043344736099,
+      "learning_rate": 4.938006797681616e-06,
+      "loss": 0.5239,
+      "step": 2616
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.12172286212444305,
+      "learning_rate": 4.904406618841173e-06,
+      "loss": 0.5487,
+      "step": 2617
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.11061225086450577,
+      "learning_rate": 4.870918271231639e-06,
+      "loss": 0.6108,
+      "step": 2618
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.11104428023099899,
+      "learning_rate": 4.8375417942348675e-06,
+      "loss": 0.6531,
+      "step": 2619
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.11502431333065033,
+      "learning_rate": 4.804277227101206e-06,
+      "loss": 0.6029,
+      "step": 2620
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.11490488052368164,
+      "learning_rate": 4.7711246089493734e-06,
+      "loss": 0.5772,
+      "step": 2621
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.111418217420578,
+      "learning_rate": 4.738083978766439e-06,
+      "loss": 0.5912,
+      "step": 2622
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.10899946093559265,
+      "learning_rate": 4.705155375407733e-06,
+      "loss": 0.5964,
+      "step": 2623
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.11127331107854843,
+      "learning_rate": 4.672338837596923e-06,
+      "loss": 0.6143,
+      "step": 2624
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.11228065937757492,
+      "learning_rate": 4.639634403925841e-06,
+      "loss": 0.5971,
+      "step": 2625
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.11837125569581985,
+      "learning_rate": 4.607042112854465e-06,
+      "loss": 0.6638,
+      "step": 2626
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.10957052558660507,
+      "learning_rate": 4.574562002710947e-06,
+      "loss": 0.6007,
+      "step": 2627
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.11680156737565994,
+      "learning_rate": 4.542194111691489e-06,
+      "loss": 0.6363,
+      "step": 2628
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.1148158386349678,
+      "learning_rate": 4.509938477860343e-06,
+      "loss": 0.6419,
+      "step": 2629
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.11126531660556793,
+      "learning_rate": 4.477795139149688e-06,
+      "loss": 0.584,
+      "step": 2630
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.11134292185306549,
+      "learning_rate": 4.44576413335972e-06,
+      "loss": 0.6004,
+      "step": 2631
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.10785213112831116,
+      "learning_rate": 4.413845498158497e-06,
+      "loss": 0.5564,
+      "step": 2632
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.11112633347511292,
+      "learning_rate": 4.38203927108195e-06,
+      "loss": 0.5846,
+      "step": 2633
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.11770457029342651,
+      "learning_rate": 4.3503454895337695e-06,
+      "loss": 0.6012,
+      "step": 2634
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.12440262734889984,
+      "learning_rate": 4.318764190785474e-06,
+      "loss": 0.579,
+      "step": 2635
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.11246594041585922,
+      "learning_rate": 4.2872954119762775e-06,
+      "loss": 0.5514,
+      "step": 2636
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.11038460582494736,
+      "learning_rate": 4.255939190113045e-06,
+      "loss": 0.5646,
+      "step": 2637
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.11352162063121796,
+      "learning_rate": 4.224695562070335e-06,
+      "loss": 0.598,
+      "step": 2638
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.10908433794975281,
+      "learning_rate": 4.193564564590246e-06,
+      "loss": 0.593,
+      "step": 2639
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.10671448707580566,
+      "learning_rate": 4.162546234282472e-06,
+      "loss": 0.5193,
+      "step": 2640
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.10968731343746185,
+      "learning_rate": 4.131640607624121e-06,
+      "loss": 0.5699,
+      "step": 2641
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.11356443166732788,
+      "learning_rate": 4.100847720959866e-06,
+      "loss": 0.6135,
+      "step": 2642
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.1150333434343338,
+      "learning_rate": 4.0701676105017625e-06,
+      "loss": 0.593,
+      "step": 2643
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.1116378977894783,
+      "learning_rate": 4.039600312329217e-06,
+      "loss": 0.5868,
+      "step": 2644
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.11307267844676971,
+      "learning_rate": 4.0091458623890075e-06,
+      "loss": 0.5428,
+      "step": 2645
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.10892671346664429,
+      "learning_rate": 3.978804296495176e-06,
+      "loss": 0.5858,
+      "step": 2646
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.10631313174962997,
+      "learning_rate": 3.948575650329045e-06,
+      "loss": 0.5548,
+      "step": 2647
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.1064402386546135,
+      "learning_rate": 3.918459959439081e-06,
+      "loss": 0.5494,
+      "step": 2648
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.11454974114894867,
+      "learning_rate": 3.888457259240996e-06,
+      "loss": 0.5832,
+      "step": 2649
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.10831843316555023,
+      "learning_rate": 3.8585675850175914e-06,
+      "loss": 0.5666,
+      "step": 2650
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.11305207014083862,
+      "learning_rate": 3.82879097191875e-06,
+      "loss": 0.5954,
+      "step": 2651
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.11256063729524612,
+      "learning_rate": 3.799127454961371e-06,
+      "loss": 0.623,
+      "step": 2652
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.12058280408382416,
+      "learning_rate": 3.7695770690294065e-06,
+      "loss": 0.6353,
+      "step": 2653
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.11177097260951996,
+      "learning_rate": 3.740139848873736e-06,
+      "loss": 0.6177,
+      "step": 2654
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.10995835065841675,
+      "learning_rate": 3.7108158291121686e-06,
+      "loss": 0.5577,
+      "step": 2655
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.11444782465696335,
+      "learning_rate": 3.6816050442293635e-06,
+      "loss": 0.6436,
+      "step": 2656
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.11205881088972092,
+      "learning_rate": 3.652507528576865e-06,
+      "loss": 0.607,
+      "step": 2657
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.10996409505605698,
+      "learning_rate": 3.6235233163729897e-06,
+      "loss": 0.5551,
+      "step": 2658
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.10964904725551605,
+      "learning_rate": 3.5946524417027615e-06,
+      "loss": 0.564,
+      "step": 2659
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.11075043678283691,
+      "learning_rate": 3.5658949385180327e-06,
+      "loss": 0.5318,
+      "step": 2660
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.10882321000099182,
+      "learning_rate": 3.5372508406372296e-06,
+      "loss": 0.5635,
+      "step": 2661
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.11170359700918198,
+      "learning_rate": 3.508720181745473e-06,
+      "loss": 0.5894,
+      "step": 2662
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.11337607353925705,
+      "learning_rate": 3.4803029953944356e-06,
+      "loss": 0.5416,
+      "step": 2663
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.11241286993026733,
+      "learning_rate": 3.4519993150023966e-06,
+      "loss": 0.5693,
+      "step": 2664
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.1155572235584259,
+      "learning_rate": 3.4238091738541424e-06,
+      "loss": 0.5714,
+      "step": 2665
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.12784455716609955,
+      "learning_rate": 3.3957326051008765e-06,
+      "loss": 0.7062,
+      "step": 2666
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.11534856259822845,
+      "learning_rate": 3.3677696417603434e-06,
+      "loss": 0.5479,
+      "step": 2667
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.1193825826048851,
+      "learning_rate": 3.339920316716627e-06,
+      "loss": 0.5907,
+      "step": 2668
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.1079108864068985,
+      "learning_rate": 3.312184662720186e-06,
+      "loss": 0.6232,
+      "step": 2669
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.10980544984340668,
+      "learning_rate": 3.284562712387784e-06,
+      "loss": 0.5501,
+      "step": 2670
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.10935201495885849,
+      "learning_rate": 3.2570544982025053e-06,
+      "loss": 0.5814,
+      "step": 2671
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.10963080823421478,
+      "learning_rate": 3.2296600525136722e-06,
+      "loss": 0.6194,
+      "step": 2672
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.12110233306884766,
+      "learning_rate": 3.202379407536804e-06,
+      "loss": 0.5864,
+      "step": 2673
+    },
+    {
+      "epoch": 2.72,
+      "eval_loss": 0.8584994673728943,
+      "eval_runtime": 111.5605,
+      "eval_samples_per_second": 89.082,
+      "eval_steps_per_second": 22.275,
+      "step": 2673
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.11464009433984756,
+      "learning_rate": 3.1752125953535936e-06,
+      "loss": 0.6142,
+      "step": 2674
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.10756965726613998,
+      "learning_rate": 3.1481596479118856e-06,
+      "loss": 0.6132,
+      "step": 2675
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.11245951056480408,
+      "learning_rate": 3.121220597025609e-06,
+      "loss": 0.5407,
+      "step": 2676
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.11353651434183121,
+      "learning_rate": 3.0943954743747116e-06,
+      "loss": 0.6101,
+      "step": 2677
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.11170332878828049,
+      "learning_rate": 3.0676843115052588e-06,
+      "loss": 0.5426,
+      "step": 2678
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.11404453963041306,
+      "learning_rate": 3.041087139829213e-06,
+      "loss": 0.6005,
+      "step": 2679
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.10731244832277298,
+      "learning_rate": 3.0146039906245426e-06,
+      "loss": 0.593,
+      "step": 2680
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.11844407021999359,
+      "learning_rate": 2.988234895035058e-06,
+      "loss": 0.5947,
+      "step": 2681
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.115105539560318,
+      "learning_rate": 2.961979884070509e-06,
+      "loss": 0.5784,
+      "step": 2682
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.1156916469335556,
+      "learning_rate": 2.9358389886064763e-06,
+      "loss": 0.5635,
+      "step": 2683
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.10522948205471039,
+      "learning_rate": 2.909812239384313e-06,
+      "loss": 0.5477,
+      "step": 2684
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.11422282457351685,
+      "learning_rate": 2.883899667011136e-06,
+      "loss": 0.6244,
+      "step": 2685
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.11447778344154358,
+      "learning_rate": 2.8581013019598367e-06,
+      "loss": 0.6572,
+      "step": 2686
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.13662485778331757,
+      "learning_rate": 2.8324171745689576e-06,
+      "loss": 0.5307,
+      "step": 2687
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.1115885004401207,
+      "learning_rate": 2.806847315042693e-06,
+      "loss": 0.5911,
+      "step": 2688
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.111208975315094,
+      "learning_rate": 2.7813917534509017e-06,
+      "loss": 0.5648,
+      "step": 2689
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.11762652546167374,
+      "learning_rate": 2.756050519729003e-06,
+      "loss": 0.6054,
+      "step": 2690
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.11107826977968216,
+      "learning_rate": 2.730823643677971e-06,
+      "loss": 0.6203,
+      "step": 2691
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.11136118322610855,
+      "learning_rate": 2.7057111549642964e-06,
+      "loss": 0.5723,
+      "step": 2692
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.10628387331962585,
+      "learning_rate": 2.6807130831199455e-06,
+      "loss": 0.5411,
+      "step": 2693
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.11628863960504532,
+      "learning_rate": 2.6558294575423472e-06,
+      "loss": 0.6226,
+      "step": 2694
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.11020716279745102,
+      "learning_rate": 2.631060307494326e-06,
+      "loss": 0.5453,
+      "step": 2695
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.11366419494152069,
+      "learning_rate": 2.6064056621040946e-06,
+      "loss": 0.5846,
+      "step": 2696
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.11488842964172363,
+      "learning_rate": 2.5818655503652258e-06,
+      "loss": 0.6245,
+      "step": 2697
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.11542273312807083,
+      "learning_rate": 2.5574400011365684e-06,
+      "loss": 0.6334,
+      "step": 2698
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.12216752767562866,
+      "learning_rate": 2.533129043142257e-06,
+      "loss": 0.5436,
+      "step": 2699
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.1085500568151474,
+      "learning_rate": 2.5089327049716894e-06,
+      "loss": 0.6075,
+      "step": 2700
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.11838563531637192,
+      "learning_rate": 2.484851015079448e-06,
+      "loss": 0.5929,
+      "step": 2701
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.12044553458690643,
+      "learning_rate": 2.460884001785313e-06,
+      "loss": 0.623,
+      "step": 2702
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.1102093905210495,
+      "learning_rate": 2.437031693274172e-06,
+      "loss": 0.5605,
+      "step": 2703
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.11271727085113525,
+      "learning_rate": 2.4132941175960655e-06,
+      "loss": 0.6069,
+      "step": 2704
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.10518582910299301,
+      "learning_rate": 2.389671302666063e-06,
+      "loss": 0.5703,
+      "step": 2705
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.10748353600502014,
+      "learning_rate": 2.3661632762643214e-06,
+      "loss": 0.6061,
+      "step": 2706
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.11418744921684265,
+      "learning_rate": 2.342770066035971e-06,
+      "loss": 0.5951,
+      "step": 2707
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.10873697698116302,
+      "learning_rate": 2.3194916994911386e-06,
+      "loss": 0.5733,
+      "step": 2708
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.11202026158571243,
+      "learning_rate": 2.296328204004905e-06,
+      "loss": 0.6471,
+      "step": 2709
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.11444363743066788,
+      "learning_rate": 2.273279606817225e-06,
+      "loss": 0.5909,
+      "step": 2710
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.10695597529411316,
+      "learning_rate": 2.2503459350329824e-06,
+      "loss": 0.533,
+      "step": 2711
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.11246615648269653,
+      "learning_rate": 2.227527215621883e-06,
+      "loss": 0.532,
+      "step": 2712
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.11318053305149078,
+      "learning_rate": 2.2048234754184603e-06,
+      "loss": 0.6185,
+      "step": 2713
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.10494859516620636,
+      "learning_rate": 2.1822347411220246e-06,
+      "loss": 0.578,
+      "step": 2714
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.1117667704820633,
+      "learning_rate": 2.159761039296648e-06,
+      "loss": 0.6141,
+      "step": 2715
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.10902206599712372,
+      "learning_rate": 2.1374023963711244e-06,
+      "loss": 0.5295,
+      "step": 2716
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.10915222018957138,
+      "learning_rate": 2.115158838638931e-06,
+      "loss": 0.5343,
+      "step": 2717
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.11486534774303436,
+      "learning_rate": 2.0930303922582216e-06,
+      "loss": 0.599,
+      "step": 2718
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.11193366348743439,
+      "learning_rate": 2.0710170832517695e-06,
+      "loss": 0.569,
+      "step": 2719
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.10730282217264175,
+      "learning_rate": 2.049118937506955e-06,
+      "loss": 0.5744,
+      "step": 2720
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.10701372474431992,
+      "learning_rate": 2.0273359807757243e-06,
+      "loss": 0.5462,
+      "step": 2721
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.10689830780029297,
+      "learning_rate": 2.005668238674552e-06,
+      "loss": 0.5613,
+      "step": 2722
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.10967187583446503,
+      "learning_rate": 1.9841157366844442e-06,
+      "loss": 0.53,
+      "step": 2723
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.1149376854300499,
+      "learning_rate": 1.962678500150883e-06,
+      "loss": 0.574,
+      "step": 2724
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.11443246155977249,
+      "learning_rate": 1.9413565542837686e-06,
+      "loss": 0.6061,
+      "step": 2725
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.10937022417783737,
+      "learning_rate": 1.920149924157455e-06,
+      "loss": 0.5687,
+      "step": 2726
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11562293022871017,
+      "learning_rate": 1.8990586347106708e-06,
+      "loss": 0.5484,
+      "step": 2727
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.12192241102457047,
+      "learning_rate": 1.8780827107465092e-06,
+      "loss": 0.62,
+      "step": 2728
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.1137763112783432,
+      "learning_rate": 1.857222176932405e-06,
+      "loss": 0.5311,
+      "step": 2729
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11711198091506958,
+      "learning_rate": 1.836477057800068e-06,
+      "loss": 0.5654,
+      "step": 2730
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11398762464523315,
+      "learning_rate": 1.8158473777455165e-06,
+      "loss": 0.5958,
+      "step": 2731
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11018142104148865,
+      "learning_rate": 1.7953331610289892e-06,
+      "loss": 0.5933,
+      "step": 2732
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11277370899915695,
+      "learning_rate": 1.7749344317749439e-06,
+      "loss": 0.648,
+      "step": 2733
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11894101649522781,
+      "learning_rate": 1.7546512139720362e-06,
+      "loss": 0.5985,
+      "step": 2734
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11601237952709198,
+      "learning_rate": 1.7344835314730856e-06,
+      "loss": 0.5747,
+      "step": 2735
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.11657615005970001,
+      "learning_rate": 1.7144314079950208e-06,
+      "loss": 0.6128,
+      "step": 2736
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.10841073095798492,
+      "learning_rate": 1.694494867118901e-06,
+      "loss": 0.5359,
+      "step": 2737
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.1163870096206665,
+      "learning_rate": 1.6746739322898385e-06,
+      "loss": 0.5529,
+      "step": 2738
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.11136572062969208,
+      "learning_rate": 1.6549686268170107e-06,
+      "loss": 0.5948,
+      "step": 2739
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.10908970981836319,
+      "learning_rate": 1.6353789738736026e-06,
+      "loss": 0.5998,
+      "step": 2740
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.11252446472644806,
+      "learning_rate": 1.6159049964968197e-06,
+      "loss": 0.6152,
+      "step": 2741
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.11921734362840652,
+      "learning_rate": 1.5965467175878102e-06,
+      "loss": 0.6083,
+      "step": 2742
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.11046607792377472,
+      "learning_rate": 1.5773041599116523e-06,
+      "loss": 0.5949,
+      "step": 2743
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.11592896282672882,
+      "learning_rate": 1.558177346097367e-06,
+      "loss": 0.544,
+      "step": 2744
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.1106400117278099,
+      "learning_rate": 1.5391662986378507e-06,
+      "loss": 0.6443,
+      "step": 2745
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.10922834277153015,
+      "learning_rate": 1.5202710398898646e-06,
+      "loss": 0.5723,
+      "step": 2746
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.11416708678007126,
+      "learning_rate": 1.501491592073978e-06,
+      "loss": 0.5944,
+      "step": 2747
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.11159468442201614,
+      "learning_rate": 1.4828279772745924e-06,
+      "loss": 0.6148,
+      "step": 2748
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.1149953305721283,
+      "learning_rate": 1.4642802174398952e-06,
+      "loss": 0.5913,
+      "step": 2749
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.10927271097898483,
+      "learning_rate": 1.4458483343818052e-06,
+      "loss": 0.5284,
+      "step": 2750
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.11257104575634003,
+      "learning_rate": 1.4275323497760063e-06,
+      "loss": 0.5917,
+      "step": 2751
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.11167798191308975,
+      "learning_rate": 1.4093322851618573e-06,
+      "loss": 0.5838,
+      "step": 2752
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.114674873650074,
+      "learning_rate": 1.3912481619424044e-06,
+      "loss": 0.6022,
+      "step": 2753
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.11159475892782211,
+      "learning_rate": 1.3732800013843584e-06,
+      "loss": 0.6218,
+      "step": 2754
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.11362778395414352,
+      "learning_rate": 1.3554278246180385e-06,
+      "loss": 0.6123,
+      "step": 2755
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.11442891508340836,
+      "learning_rate": 1.3376916526374072e-06,
+      "loss": 0.5895,
+      "step": 2756
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.11061660200357437,
+      "learning_rate": 1.320071506299947e-06,
+      "loss": 0.574,
+      "step": 2757
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.110038161277771,
+      "learning_rate": 1.302567406326749e-06,
+      "loss": 0.578,
+      "step": 2758
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.11182013154029846,
+      "learning_rate": 1.2851793733024253e-06,
+      "loss": 0.6194,
+      "step": 2759
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.11634350568056107,
+      "learning_rate": 1.2679074276750746e-06,
+      "loss": 0.5638,
+      "step": 2760
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.12204045057296753,
+      "learning_rate": 1.2507515897562827e-06,
+      "loss": 0.5914,
+      "step": 2761
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.12485633045434952,
+      "learning_rate": 1.2337118797211223e-06,
+      "loss": 0.491,
+      "step": 2762
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.1096053496003151,
+      "learning_rate": 1.216788317608053e-06,
+      "loss": 0.6002,
+      "step": 2763
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.10932616144418716,
+      "learning_rate": 1.1999809233189885e-06,
+      "loss": 0.6111,
+      "step": 2764
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.11098449677228928,
+      "learning_rate": 1.183289716619218e-06,
+      "loss": 0.6271,
+      "step": 2765
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.1182386726140976,
+      "learning_rate": 1.1667147171373848e-06,
+      "loss": 0.5733,
+      "step": 2766
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.12654823064804077,
+      "learning_rate": 1.150255944365486e-06,
+      "loss": 0.5121,
+      "step": 2767
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.10396461188793182,
+      "learning_rate": 1.1339134176588162e-06,
+      "loss": 0.4755,
+      "step": 2768
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.1107998713850975,
+      "learning_rate": 1.1176871562360025e-06,
+      "loss": 0.5951,
+      "step": 2769
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.11483904719352722,
+      "learning_rate": 1.1015771791789254e-06,
+      "loss": 0.6418,
+      "step": 2770
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.11298028379678726,
+      "learning_rate": 1.0855835054327079e-06,
+      "loss": 0.5731,
+      "step": 2771
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.11473549157381058,
+      "learning_rate": 1.0697061538057163e-06,
+      "loss": 0.6362,
+      "step": 2772
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.11135080456733704,
+      "learning_rate": 1.0539451429695036e-06,
+      "loss": 0.5851,
+      "step": 2773
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.12176463007926941,
+      "learning_rate": 1.0383004914588435e-06,
+      "loss": 0.5927,
+      "step": 2774
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.11966660618782043,
+      "learning_rate": 1.0227722176716415e-06,
+      "loss": 0.5216,
+      "step": 2775
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.1088603138923645,
+      "learning_rate": 1.0073603398689568e-06,
+      "loss": 0.5488,
+      "step": 2776
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.13011470437049866,
+      "learning_rate": 9.920648761749585e-07,
+      "loss": 0.5306,
+      "step": 2777
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.11066142469644547,
+      "learning_rate": 9.768858445769247e-07,
+      "loss": 0.602,
+      "step": 2778
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.11068571358919144,
+      "learning_rate": 9.6182326292521e-07,
+      "loss": 0.6626,
+      "step": 2779
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.11232766509056091,
+      "learning_rate": 9.468771489332118e-07,
+      "loss": 0.6698,
+      "step": 2780
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.1098511815071106,
+      "learning_rate": 9.320475201773816e-07,
+      "loss": 0.6321,
+      "step": 2781
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.11290988326072693,
+      "learning_rate": 9.173343940971801e-07,
+      "loss": 0.6224,
+      "step": 2782
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.10911401361227036,
+      "learning_rate": 9.027377879950671e-07,
+      "loss": 0.6291,
+      "step": 2783
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.1166314110159874,
+      "learning_rate": 8.882577190364449e-07,
+      "loss": 0.5967,
+      "step": 2784
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11711157113313675,
+      "learning_rate": 8.738942042497145e-07,
+      "loss": 0.5577,
+      "step": 2785
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.12066779285669327,
+      "learning_rate": 8.596472605261863e-07,
+      "loss": 0.6397,
+      "step": 2786
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11296729743480682,
+      "learning_rate": 8.455169046200806e-07,
+      "loss": 0.6414,
+      "step": 2787
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11596448719501495,
+      "learning_rate": 8.315031531485273e-07,
+      "loss": 0.6041,
+      "step": 2788
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11018485575914383,
+      "learning_rate": 8.176060225915216e-07,
+      "loss": 0.5743,
+      "step": 2789
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11164645850658417,
+      "learning_rate": 8.038255292919128e-07,
+      "loss": 0.5798,
+      "step": 2790
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11427861452102661,
+      "learning_rate": 7.901616894553821e-07,
+      "loss": 0.6275,
+      "step": 2791
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11207189410924911,
+      "learning_rate": 7.766145191504536e-07,
+      "loss": 0.5658,
+      "step": 2792
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.10966978222131729,
+      "learning_rate": 7.631840343084174e-07,
+      "loss": 0.5871,
+      "step": 2793
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.11263763159513474,
+      "learning_rate": 7.498702507233613e-07,
+      "loss": 0.5824,
+      "step": 2794
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.11337420344352722,
+      "learning_rate": 7.366731840521279e-07,
+      "loss": 0.5991,
+      "step": 2795
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.11061732470989227,
+      "learning_rate": 7.23592849814303e-07,
+      "loss": 0.5515,
+      "step": 2796
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.10715221613645554,
+      "learning_rate": 7.106292633921818e-07,
+      "loss": 0.5595,
+      "step": 2797
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.10988996922969818,
+      "learning_rate": 6.97782440030803e-07,
+      "loss": 0.5419,
+      "step": 2798
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.11286706477403641,
+      "learning_rate": 6.850523948378596e-07,
+      "loss": 0.5829,
+      "step": 2799
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.11395177990198135,
+      "learning_rate": 6.724391427837206e-07,
+      "loss": 0.6399,
+      "step": 2800
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.12327868491411209,
+      "learning_rate": 6.599426987014323e-07,
+      "loss": 0.6341,
+      "step": 2801
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.11015816032886505,
+      "learning_rate": 6.475630772866282e-07,
+      "loss": 0.5587,
+      "step": 2802
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.11080652475357056,
+      "learning_rate": 6.353002930976182e-07,
+      "loss": 0.6178,
+      "step": 2803
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.12053908407688141,
+      "learning_rate": 6.231543605552559e-07,
+      "loss": 0.5465,
+      "step": 2804
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.11483890563249588,
+      "learning_rate": 6.111252939430268e-07,
+      "loss": 0.5825,
+      "step": 2805
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.11099923402070999,
+      "learning_rate": 5.992131074069595e-07,
+      "loss": 0.6066,
+      "step": 2806
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.10981005430221558,
+      "learning_rate": 5.874178149556264e-07,
+      "loss": 0.6131,
+      "step": 2807
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.11368913948535919,
+      "learning_rate": 5.757394304601538e-07,
+      "loss": 0.5983,
+      "step": 2808
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.11302459239959717,
+      "learning_rate": 5.641779676541781e-07,
+      "loss": 0.5453,
+      "step": 2809
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.11117744445800781,
+      "learning_rate": 5.527334401338236e-07,
+      "loss": 0.611,
+      "step": 2810
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.11395902186632156,
+      "learning_rate": 5.414058613577245e-07,
+      "loss": 0.7409,
+      "step": 2811
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.11279110610485077,
+      "learning_rate": 5.301952446469583e-07,
+      "loss": 0.5562,
+      "step": 2812
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.1126856654882431,
+      "learning_rate": 5.191016031850904e-07,
+      "loss": 0.5564,
+      "step": 2813
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11236195266246796,
+      "learning_rate": 5.081249500180851e-07,
+      "loss": 0.5958,
+      "step": 2814
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11353650689125061,
+      "learning_rate": 4.972652980543613e-07,
+      "loss": 0.5868,
+      "step": 2815
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.10875215381383896,
+      "learning_rate": 4.865226600647254e-07,
+      "loss": 0.5932,
+      "step": 2816
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11873544007539749,
+      "learning_rate": 4.7589704868239436e-07,
+      "loss": 0.5258,
+      "step": 2817
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11428482830524445,
+      "learning_rate": 4.653884764029393e-07,
+      "loss": 0.633,
+      "step": 2818
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11285462975502014,
+      "learning_rate": 4.549969555843303e-07,
+      "loss": 0.6319,
+      "step": 2819
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11439676582813263,
+      "learning_rate": 4.44722498446859e-07,
+      "loss": 0.5851,
+      "step": 2820
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11818290501832962,
+      "learning_rate": 4.3456511707316017e-07,
+      "loss": 0.577,
+      "step": 2821
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11272864788770676,
+      "learning_rate": 4.245248234081789e-07,
+      "loss": 0.5934,
+      "step": 2822
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.11173795163631439,
+      "learning_rate": 4.146016292591703e-07,
+      "loss": 0.605,
+      "step": 2823
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.10947389900684357,
+      "learning_rate": 4.047955462957109e-07,
+      "loss": 0.5318,
+      "step": 2824
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.10828577727079391,
+      "learning_rate": 3.951065860496095e-07,
+      "loss": 0.556,
+      "step": 2825
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.11082903295755386,
+      "learning_rate": 3.855347599149739e-07,
+      "loss": 0.5723,
+      "step": 2826
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.11096781492233276,
+      "learning_rate": 3.760800791481556e-07,
+      "loss": 0.6666,
+      "step": 2827
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.10990378260612488,
+      "learning_rate": 3.6674255486772725e-07,
+      "loss": 0.5994,
+      "step": 2828
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.10654883831739426,
+      "learning_rate": 3.5752219805450516e-07,
+      "loss": 0.5981,
+      "step": 2829
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.11795970052480698,
+      "learning_rate": 3.484190195515269e-07,
+      "loss": 0.5836,
+      "step": 2830
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.1102161705493927,
+      "learning_rate": 3.3943303006400694e-07,
+      "loss": 0.59,
+      "step": 2831
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.11054965108633041,
+      "learning_rate": 3.3056424015934785e-07,
+      "loss": 0.5724,
+      "step": 2832
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.11192359030246735,
+      "learning_rate": 3.218126602671512e-07,
+      "loss": 0.5748,
+      "step": 2833
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.11256634443998337,
+      "learning_rate": 3.131783006791511e-07,
+      "loss": 0.5824,
+      "step": 2834
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.11232349276542664,
+      "learning_rate": 3.0466117154925866e-07,
+      "loss": 0.5528,
+      "step": 2835
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.11651121079921722,
+      "learning_rate": 2.962612828935174e-07,
+      "loss": 0.5844,
+      "step": 2836
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.10736921429634094,
+      "learning_rate": 2.8797864459007006e-07,
+      "loss": 0.5686,
+      "step": 2837
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.11939041316509247,
+      "learning_rate": 2.798132663792141e-07,
+      "loss": 0.5991,
+      "step": 2838
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.11100921034812927,
+      "learning_rate": 2.71765157863324e-07,
+      "loss": 0.602,
+      "step": 2839
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.10834819078445435,
+      "learning_rate": 2.638343285068734e-07,
+      "loss": 0.5816,
+      "step": 2840
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.11037305742502213,
+      "learning_rate": 2.56020787636424e-07,
+      "loss": 0.5692,
+      "step": 2841
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.11000951379537582,
+      "learning_rate": 2.4832454444060347e-07,
+      "loss": 0.6585,
+      "step": 2842
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.11251745373010635,
+      "learning_rate": 2.4074560797008315e-07,
+      "loss": 0.6148,
+      "step": 2843
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.10755830258131027,
+      "learning_rate": 2.3328398713762244e-07,
+      "loss": 0.5234,
+      "step": 2844
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.11007846891880035,
+      "learning_rate": 2.2593969071798005e-07,
+      "loss": 0.5583,
+      "step": 2845
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.11691147089004517,
+      "learning_rate": 2.1871272734794723e-07,
+      "loss": 0.5951,
+      "step": 2846
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.11737576127052307,
+      "learning_rate": 2.1160310552635898e-07,
+      "loss": 0.6177,
+      "step": 2847
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.1089283749461174,
+      "learning_rate": 2.0461083361403844e-07,
+      "loss": 0.6085,
+      "step": 2848
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.1154700219631195,
+      "learning_rate": 1.9773591983380803e-07,
+      "loss": 0.6864,
+      "step": 2849
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.12094154953956604,
+      "learning_rate": 1.9097837227047832e-07,
+      "loss": 0.6226,
+      "step": 2850
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.1133342981338501,
+      "learning_rate": 1.8433819887084812e-07,
+      "loss": 0.5913,
+      "step": 2851
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.11688636988401413,
+      "learning_rate": 1.778154074436711e-07,
+      "loss": 0.6063,
+      "step": 2852
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.1098800078034401,
+      "learning_rate": 1.7141000565966681e-07,
+      "loss": 0.591,
+      "step": 2853
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.11573871225118637,
+      "learning_rate": 1.6512200105150976e-07,
+      "loss": 0.5913,
+      "step": 2854
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.11156423389911652,
+      "learning_rate": 1.5895140101382933e-07,
+      "loss": 0.5526,
+      "step": 2855
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.11374085396528244,
+      "learning_rate": 1.528982128031542e-07,
+      "loss": 0.5696,
+      "step": 2856
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.11236458271741867,
+      "learning_rate": 1.4696244353795685e-07,
+      "loss": 0.6051,
+      "step": 2857
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.11232428252696991,
+      "learning_rate": 1.4114410019865353e-07,
+      "loss": 0.5529,
+      "step": 2858
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.10971636325120926,
+      "learning_rate": 1.3544318962751547e-07,
+      "loss": 0.632,
+      "step": 2859
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.11082550883293152,
+      "learning_rate": 1.2985971852875755e-07,
+      "loss": 0.6308,
+      "step": 2860
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.10448183119297028,
+      "learning_rate": 1.2439369346846086e-07,
+      "loss": 0.5714,
+      "step": 2861
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.11496017873287201,
+      "learning_rate": 1.1904512087461683e-07,
+      "loss": 0.6109,
+      "step": 2862
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.11100126802921295,
+      "learning_rate": 1.1381400703707191e-07,
+      "loss": 0.5551,
+      "step": 2863
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.11645042896270752,
+      "learning_rate": 1.0870035810754964e-07,
+      "loss": 0.5797,
+      "step": 2864
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.11764471232891083,
+      "learning_rate": 1.0370418009963967e-07,
+      "loss": 0.6625,
+      "step": 2865
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.12045713514089584,
+      "learning_rate": 9.882547888878658e-08,
+      "loss": 0.5573,
+      "step": 2866
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.11804057657718658,
+      "learning_rate": 9.406426021228987e-08,
+      "loss": 0.6204,
+      "step": 2867
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.12836426496505737,
+      "learning_rate": 8.942052966925962e-08,
+      "loss": 0.5685,
+      "step": 2868
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.11208245903253555,
+      "learning_rate": 8.489429272068306e-08,
+      "loss": 0.499,
+      "step": 2869
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.11900430172681808,
+      "learning_rate": 8.048555468935792e-08,
+      "loss": 0.5828,
+      "step": 2870
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.11285870522260666,
+      "learning_rate": 7.619432075989252e-08,
+      "loss": 0.5847,
+      "step": 2871
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.10991118103265762,
+      "learning_rate": 7.202059597873901e-08,
+      "loss": 0.5613,
+      "step": 2872
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.1170453429222107,
+      "learning_rate": 6.796438525414895e-08,
+      "loss": 0.5353,
+      "step": 2873
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.12400245666503906,
+      "learning_rate": 6.402569335615117e-08,
+      "loss": 0.5933,
+      "step": 2874
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.11195569485425949,
+      "learning_rate": 6.020452491662942e-08,
+      "loss": 0.5527,
+      "step": 2875
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.10715174674987793,
+      "learning_rate": 5.6500884429200316e-08,
+      "loss": 0.5702,
+      "step": 2876
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.11766745895147324,
+      "learning_rate": 5.291477624931318e-08,
+      "loss": 0.604,
+      "step": 2877
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.11058572679758072,
+      "learning_rate": 4.9446204594194576e-08,
+      "loss": 0.5736,
+      "step": 2878
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.11179894953966141,
+      "learning_rate": 4.609517354282611e-08,
+      "loss": 0.6787,
+      "step": 2879
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.11611034721136093,
+      "learning_rate": 4.2861687035999906e-08,
+      "loss": 0.5486,
+      "step": 2880
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.11076311767101288,
+      "learning_rate": 3.974574887622984e-08,
+      "loss": 0.5876,
+      "step": 2881
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.11752700060606003,
+      "learning_rate": 3.6747362727829194e-08,
+      "loss": 0.6271,
+      "step": 2882
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.11534613370895386,
+      "learning_rate": 3.38665321168663e-08,
+      "loss": 0.6206,
+      "step": 2883
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.11212414503097534,
+      "learning_rate": 3.1103260431164514e-08,
+      "loss": 0.623,
+      "step": 2884
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.10658472776412964,
+      "learning_rate": 2.845755092029112e-08,
+      "loss": 0.5904,
+      "step": 2885
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.11601762473583221,
+      "learning_rate": 2.5929406695568425e-08,
+      "loss": 0.5631,
+      "step": 2886
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.10685524344444275,
+      "learning_rate": 2.351883073006267e-08,
+      "loss": 0.6448,
+      "step": 2887
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.10688942670822144,
+      "learning_rate": 2.1225825858584013e-08,
+      "loss": 0.5398,
+      "step": 2888
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.10801198333501816,
+      "learning_rate": 1.905039477767545e-08,
+      "loss": 0.5871,
+      "step": 2889
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.10796801000833511,
+      "learning_rate": 1.6992540045623895e-08,
+      "loss": 0.5745,
+      "step": 2890
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.10827478021383286,
+      "learning_rate": 1.505226408242688e-08,
+      "loss": 0.5726,
+      "step": 2891
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.11067235469818115,
+      "learning_rate": 1.3229569169836975e-08,
+      "loss": 0.5197,
+      "step": 2892
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.10949043929576874,
+      "learning_rate": 1.152445745130626e-08,
+      "loss": 0.5673,
+      "step": 2893
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.10996787995100021,
+      "learning_rate": 9.936930932041843e-09,
+      "loss": 0.5671,
+      "step": 2894
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.11858395487070084,
+      "learning_rate": 8.466991478939257e-09,
+      "loss": 0.6292,
+      "step": 2895
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.11117985099554062,
+      "learning_rate": 7.114640820637952e-09,
+      "loss": 0.5808,
+      "step": 2896
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.11092160642147064,
+      "learning_rate": 5.879880547488003e-09,
+      "loss": 0.5796,
+      "step": 2897
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.10951466858386993,
+      "learning_rate": 4.762712111550105e-09,
+      "loss": 0.6009,
+      "step": 2898
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.11280664801597595,
+      "learning_rate": 3.76313682659557e-09,
+      "loss": 0.5715,
+      "step": 2899
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.1151258572936058,
+      "learning_rate": 2.8811558681174354e-09,
+      "loss": 0.6435,
+      "step": 2900
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.10584615916013718,
+      "learning_rate": 2.116770273319357e-09,
+      "loss": 0.5999,
+      "step": 2901
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.11521152406930923,
+      "learning_rate": 1.4699809410934074e-09,
+      "loss": 0.6089,
+      "step": 2902
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.11263085901737213,
+      "learning_rate": 9.40788632064482e-10,
+      "loss": 0.5642,
+      "step": 2903
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.10475903749465942,
+      "learning_rate": 5.291939685569957e-10,
+      "loss": 0.5743,
+      "step": 2904
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.11360141634941101,
+      "learning_rate": 2.3519743460598267e-10,
+      "loss": 0.6425,
+      "step": 2905
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.11684887856245041,
+      "learning_rate": 5.87993759348926e-11,
+      "loss": 0.6022,
+      "step": 2906
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.118344247341156,
+      "learning_rate": 0.0,
+      "loss": 0.5936,
+      "step": 2907
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 2907,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 969,
+  "total_flos": 8.674609537770062e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}