diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4137 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9952,
+  "eval_steps": 500,
+  "global_step": 585,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00512,
+      "grad_norm": 7.427512523661066,
+      "learning_rate": 1.3559322033898307e-06,
+      "loss": 1.1833,
+      "step": 1
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 7.378805913304633,
+      "learning_rate": 2.7118644067796613e-06,
+      "loss": 1.1693,
+      "step": 2
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 7.387359456401948,
+      "learning_rate": 4.067796610169492e-06,
+      "loss": 1.1699,
+      "step": 3
+    },
+    {
+      "epoch": 0.02048,
+      "grad_norm": 5.424792613029199,
+      "learning_rate": 5.423728813559323e-06,
+      "loss": 1.112,
+      "step": 4
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 3.3531117149966567,
+      "learning_rate": 6.779661016949153e-06,
+      "loss": 1.0668,
+      "step": 5
+    },
+    {
+      "epoch": 0.03072,
+      "grad_norm": 2.9208579813945645,
+      "learning_rate": 8.135593220338983e-06,
+      "loss": 1.0554,
+      "step": 6
+    },
+    {
+      "epoch": 0.03584,
+      "grad_norm": 5.338271857324549,
+      "learning_rate": 9.491525423728815e-06,
+      "loss": 1.0192,
+      "step": 7
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 5.570953182353899,
+      "learning_rate": 1.0847457627118645e-05,
+      "loss": 1.0134,
+      "step": 8
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 5.770745007584102,
+      "learning_rate": 1.2203389830508477e-05,
+      "loss": 0.9676,
+      "step": 9
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 4.652978095614179,
+      "learning_rate": 1.3559322033898305e-05,
+      "loss": 0.9515,
+      "step": 10
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 2.8588327705676626,
+      "learning_rate": 1.4915254237288137e-05,
+      "loss": 0.9188,
+      "step": 11
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 3.6027109410465066,
+      "learning_rate": 1.6271186440677967e-05,
+      "loss": 0.8979,
+      "step": 12
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 2.9981487608394635,
+      "learning_rate": 1.76271186440678e-05,
+      "loss": 0.8804,
+      "step": 13
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 1.9815532395129927,
+      "learning_rate": 1.898305084745763e-05,
+      "loss": 0.85,
+      "step": 14
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 2.0993713191830765,
+      "learning_rate": 2.033898305084746e-05,
+      "loss": 0.8359,
+      "step": 15
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 1.8644702221939118,
+      "learning_rate": 2.169491525423729e-05,
+      "loss": 0.8325,
+      "step": 16
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 1.4001838150615264,
+      "learning_rate": 2.3050847457627122e-05,
+      "loss": 0.8145,
+      "step": 17
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 1.5105243610460197,
+      "learning_rate": 2.4406779661016954e-05,
+      "loss": 0.8061,
+      "step": 18
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 1.508229346305267,
+      "learning_rate": 2.576271186440678e-05,
+      "loss": 0.8019,
+      "step": 19
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 1.313209951669769,
+      "learning_rate": 2.711864406779661e-05,
+      "loss": 0.8,
+      "step": 20
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 1.173580022336048,
+      "learning_rate": 2.8474576271186442e-05,
+      "loss": 0.7764,
+      "step": 21
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 1.293902010777013,
+      "learning_rate": 2.9830508474576274e-05,
+      "loss": 0.7891,
+      "step": 22
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 0.9680297788054414,
+      "learning_rate": 3.1186440677966106e-05,
+      "loss": 0.7745,
+      "step": 23
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 0.9265872685281233,
+      "learning_rate": 3.2542372881355934e-05,
+      "loss": 0.7678,
+      "step": 24
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.9305803654580205,
+      "learning_rate": 3.389830508474576e-05,
+      "loss": 0.7559,
+      "step": 25
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 1.334014536013991,
+      "learning_rate": 3.52542372881356e-05,
+      "loss": 0.7697,
+      "step": 26
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 1.7668925981671184,
+      "learning_rate": 3.6610169491525426e-05,
+      "loss": 0.7506,
+      "step": 27
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 1.2041359597289543,
+      "learning_rate": 3.796610169491526e-05,
+      "loss": 0.7446,
+      "step": 28
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 1.9838038548617862,
+      "learning_rate": 3.932203389830509e-05,
+      "loss": 0.7511,
+      "step": 29
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 1.2091908618849432,
+      "learning_rate": 4.067796610169492e-05,
+      "loss": 0.743,
+      "step": 30
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 2.013739044410424,
+      "learning_rate": 4.203389830508475e-05,
+      "loss": 0.739,
+      "step": 31
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 1.7387033344897997,
+      "learning_rate": 4.338983050847458e-05,
+      "loss": 0.737,
+      "step": 32
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 1.7866106146689165,
+      "learning_rate": 4.474576271186441e-05,
+      "loss": 0.7392,
+      "step": 33
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 1.5407095248074152,
+      "learning_rate": 4.6101694915254244e-05,
+      "loss": 0.7286,
+      "step": 34
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 1.7355594859856742,
+      "learning_rate": 4.745762711864407e-05,
+      "loss": 0.7429,
+      "step": 35
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 1.4876420903172791,
+      "learning_rate": 4.881355932203391e-05,
+      "loss": 0.7229,
+      "step": 36
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 2.2843469216868098,
+      "learning_rate": 5.016949152542373e-05,
+      "loss": 0.7288,
+      "step": 37
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 1.6645490807582124,
+      "learning_rate": 5.152542372881356e-05,
+      "loss": 0.7189,
+      "step": 38
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 2.3532899761581096,
+      "learning_rate": 5.288135593220339e-05,
+      "loss": 0.7163,
+      "step": 39
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 2.497396548783959,
+      "learning_rate": 5.423728813559322e-05,
+      "loss": 0.7264,
+      "step": 40
+    },
+    {
+      "epoch": 0.20992,
+      "grad_norm": 1.4481682870152812,
+      "learning_rate": 5.5593220338983056e-05,
+      "loss": 0.7144,
+      "step": 41
+    },
+    {
+      "epoch": 0.21504,
+      "grad_norm": 3.1370809945974587,
+      "learning_rate": 5.6949152542372884e-05,
+      "loss": 0.7002,
+      "step": 42
+    },
+    {
+      "epoch": 0.22016,
+      "grad_norm": 2.6339360870235264,
+      "learning_rate": 5.830508474576271e-05,
+      "loss": 0.7403,
+      "step": 43
+    },
+    {
+      "epoch": 0.22528,
+      "grad_norm": 2.2084591884302647,
+      "learning_rate": 5.966101694915255e-05,
+      "loss": 0.7236,
+      "step": 44
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 2.7378970801102107,
+      "learning_rate": 6.1016949152542376e-05,
+      "loss": 0.6975,
+      "step": 45
+    },
+    {
+      "epoch": 0.23552,
+      "grad_norm": 1.8558926790043844,
+      "learning_rate": 6.237288135593221e-05,
+      "loss": 0.7187,
+      "step": 46
+    },
+    {
+      "epoch": 0.24064,
+      "grad_norm": 2.2724542031435866,
+      "learning_rate": 6.372881355932204e-05,
+      "loss": 0.7264,
+      "step": 47
+    },
+    {
+      "epoch": 0.24576,
+      "grad_norm": 2.8585186655333605,
+      "learning_rate": 6.508474576271187e-05,
+      "loss": 0.7122,
+      "step": 48
+    },
+    {
+      "epoch": 0.25088,
+      "grad_norm": 2.2069560445393246,
+      "learning_rate": 6.64406779661017e-05,
+      "loss": 0.727,
+      "step": 49
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 3.1071789158018173,
+      "learning_rate": 6.779661016949152e-05,
+      "loss": 0.6951,
+      "step": 50
+    },
+    {
+      "epoch": 0.26112,
+      "grad_norm": 2.0255083166280694,
+      "learning_rate": 6.915254237288137e-05,
+      "loss": 0.6928,
+      "step": 51
+    },
+    {
+      "epoch": 0.26624,
+      "grad_norm": 3.3982026768893427,
+      "learning_rate": 7.05084745762712e-05,
+      "loss": 0.6963,
+      "step": 52
+    },
+    {
+      "epoch": 0.27136,
+      "grad_norm": 1.7595270818326945,
+      "learning_rate": 7.186440677966102e-05,
+      "loss": 0.7136,
+      "step": 53
+    },
+    {
+      "epoch": 0.27648,
+      "grad_norm": 3.522255480730156,
+      "learning_rate": 7.322033898305085e-05,
+      "loss": 0.701,
+      "step": 54
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 2.0471574026263473,
+      "learning_rate": 7.457627118644068e-05,
+      "loss": 0.7015,
+      "step": 55
+    },
+    {
+      "epoch": 0.28672,
+      "grad_norm": 4.076903746052479,
+      "learning_rate": 7.593220338983052e-05,
+      "loss": 0.7213,
+      "step": 56
+    },
+    {
+      "epoch": 0.29184,
+      "grad_norm": 2.9540886546409,
+      "learning_rate": 7.728813559322035e-05,
+      "loss": 0.7072,
+      "step": 57
+    },
+    {
+      "epoch": 0.29696,
+      "grad_norm": 2.930300987052726,
+      "learning_rate": 7.864406779661018e-05,
+      "loss": 0.7075,
+      "step": 58
+    },
+    {
+      "epoch": 0.30208,
+      "grad_norm": 2.309566711188816,
+      "learning_rate": 8e-05,
+      "loss": 0.6845,
+      "step": 59
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 3.0240560677976482,
+      "learning_rate": 7.999928656081034e-05,
+      "loss": 0.7071,
+      "step": 60
+    },
+    {
+      "epoch": 0.31232,
+      "grad_norm": 2.1478418033816076,
+      "learning_rate": 7.999714626869112e-05,
+      "loss": 0.6906,
+      "step": 61
+    },
+    {
+      "epoch": 0.31744,
+      "grad_norm": 3.179230462128308,
+      "learning_rate": 7.999357919999074e-05,
+      "loss": 0.7038,
+      "step": 62
+    },
+    {
+      "epoch": 0.32256,
+      "grad_norm": 2.4903190109477693,
+      "learning_rate": 7.998858548195353e-05,
+      "loss": 0.7036,
+      "step": 63
+    },
+    {
+      "epoch": 0.32768,
+      "grad_norm": 2.5167791307656717,
+      "learning_rate": 7.998216529271523e-05,
+      "loss": 0.7022,
+      "step": 64
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 2.1782283151414723,
+      "learning_rate": 7.997431886129654e-05,
+      "loss": 0.7056,
+      "step": 65
+    },
+    {
+      "epoch": 0.33792,
+      "grad_norm": 2.4061240485846707,
+      "learning_rate": 7.996504646759507e-05,
+      "loss": 0.7033,
+      "step": 66
+    },
+    {
+      "epoch": 0.34304,
+      "grad_norm": 1.8094645315702094,
+      "learning_rate": 7.995434844237524e-05,
+      "loss": 0.703,
+      "step": 67
+    },
+    {
+      "epoch": 0.34816,
+      "grad_norm": 2.3445365799928384,
+      "learning_rate": 7.994222516725659e-05,
+      "loss": 0.6953,
+      "step": 68
+    },
+    {
+      "epoch": 0.35328,
+      "grad_norm": 1.99410664221794,
+      "learning_rate": 7.99286770747001e-05,
+      "loss": 0.6919,
+      "step": 69
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 2.230153791562052,
+      "learning_rate": 7.991370464799278e-05,
+      "loss": 0.6862,
+      "step": 70
+    },
+    {
+      "epoch": 0.36352,
+      "grad_norm": 1.9601599085514845,
+      "learning_rate": 7.989730842123042e-05,
+      "loss": 0.7128,
+      "step": 71
+    },
+    {
+      "epoch": 0.36864,
+      "grad_norm": 1.5761765132400967,
+      "learning_rate": 7.987948897929855e-05,
+      "loss": 0.6745,
+      "step": 72
+    },
+    {
+      "epoch": 0.37376,
+      "grad_norm": 1.679276038030434,
+      "learning_rate": 7.986024695785161e-05,
+      "loss": 0.6796,
+      "step": 73
+    },
+    {
+      "epoch": 0.37888,
+      "grad_norm": 1.6649601965609995,
+      "learning_rate": 7.98395830432902e-05,
+      "loss": 0.6745,
+      "step": 74
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 1.4057964596604813,
+      "learning_rate": 7.981749797273661e-05,
+      "loss": 0.6979,
+      "step": 75
+    },
+    {
+      "epoch": 0.38912,
+      "grad_norm": 2.014531329083667,
+      "learning_rate": 7.979399253400862e-05,
+      "loss": 0.7037,
+      "step": 76
+    },
+    {
+      "epoch": 0.39424,
+      "grad_norm": 1.7428321466528032,
+      "learning_rate": 7.976906756559127e-05,
+      "loss": 0.6832,
+      "step": 77
+    },
+    {
+      "epoch": 0.39936,
+      "grad_norm": 2.167027629573797,
+      "learning_rate": 7.974272395660703e-05,
+      "loss": 0.6879,
+      "step": 78
+    },
+    {
+      "epoch": 0.40448,
+      "grad_norm": 1.6144350155950205,
+      "learning_rate": 7.971496264678404e-05,
+      "loss": 0.6877,
+      "step": 79
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 1.5708068654881047,
+      "learning_rate": 7.968578462642265e-05,
+      "loss": 0.6989,
+      "step": 80
+    },
+    {
+      "epoch": 0.41472,
+      "grad_norm": 1.555445454516115,
+      "learning_rate": 7.965519093636e-05,
+      "loss": 0.6992,
+      "step": 81
+    },
+    {
+      "epoch": 0.41984,
+      "grad_norm": 1.8059552941628938,
+      "learning_rate": 7.962318266793294e-05,
+      "loss": 0.6912,
+      "step": 82
+    },
+    {
+      "epoch": 0.42496,
+      "grad_norm": 1.0063199261184264,
+      "learning_rate": 7.958976096293916e-05,
+      "loss": 0.6929,
+      "step": 83
+    },
+    {
+      "epoch": 0.43008,
+      "grad_norm": 1.7307326550028748,
+      "learning_rate": 7.955492701359636e-05,
+      "loss": 0.6902,
+      "step": 84
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 1.205147537826931,
+      "learning_rate": 7.951868206249975e-05,
+      "loss": 0.6943,
+      "step": 85
+    },
+    {
+      "epoch": 0.44032,
+      "grad_norm": 1.4160221907669925,
+      "learning_rate": 7.948102740257776e-05,
+      "loss": 0.6976,
+      "step": 86
+    },
+    {
+      "epoch": 0.44544,
+      "grad_norm": 2.001829250494979,
+      "learning_rate": 7.944196437704593e-05,
+      "loss": 0.7038,
+      "step": 87
+    },
+    {
+      "epoch": 0.45056,
+      "grad_norm": 1.1653512775535024,
+      "learning_rate": 7.940149437935887e-05,
+      "loss": 0.6849,
+      "step": 88
+    },
+    {
+      "epoch": 0.45568,
+      "grad_norm": 1.6823929061601692,
+      "learning_rate": 7.935961885316074e-05,
+      "loss": 0.6925,
+      "step": 89
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 1.533381264168228,
+      "learning_rate": 7.93163392922336e-05,
+      "loss": 0.7078,
+      "step": 90
+    },
+    {
+      "epoch": 0.46592,
+      "grad_norm": 1.3854602682735884,
+      "learning_rate": 7.927165724044418e-05,
+      "loss": 0.6905,
+      "step": 91
+    },
+    {
+      "epoch": 0.47104,
+      "grad_norm": 1.6672169257948595,
+      "learning_rate": 7.922557429168884e-05,
+      "loss": 0.6868,
+      "step": 92
+    },
+    {
+      "epoch": 0.47616,
+      "grad_norm": 1.3374855328007862,
+      "learning_rate": 7.917809208983667e-05,
+      "loss": 0.6872,
+      "step": 93
+    },
+    {
+      "epoch": 0.48128,
+      "grad_norm": 1.1062013903355805,
+      "learning_rate": 7.912921232867082e-05,
+      "loss": 0.6829,
+      "step": 94
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 1.2577822175567075,
+      "learning_rate": 7.907893675182816e-05,
+      "loss": 0.6851,
+      "step": 95
+    },
+    {
+      "epoch": 0.49152,
+      "grad_norm": 1.1436105339986973,
+      "learning_rate": 7.902726715273704e-05,
+      "loss": 0.6801,
+      "step": 96
+    },
+    {
+      "epoch": 0.49664,
+      "grad_norm": 1.3288774009958406,
+      "learning_rate": 7.89742053745533e-05,
+      "loss": 0.7013,
+      "step": 97
+    },
+    {
+      "epoch": 0.50176,
+      "grad_norm": 1.2793250945243777,
+      "learning_rate": 7.891975331009454e-05,
+      "loss": 0.6742,
+      "step": 98
+    },
+    {
+      "epoch": 0.50688,
+      "grad_norm": 1.3915794984714676,
+      "learning_rate": 7.886391290177259e-05,
+      "loss": 0.692,
+      "step": 99
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 1.683470719946437,
+      "learning_rate": 7.880668614152426e-05,
+      "loss": 0.6903,
+      "step": 100
+    },
+    {
+      "epoch": 0.51712,
+      "grad_norm": 1.1736596909100243,
+      "learning_rate": 7.874807507074019e-05,
+      "loss": 0.6769,
+      "step": 101
+    },
+    {
+      "epoch": 0.52224,
+      "grad_norm": 1.7187295695085814,
+      "learning_rate": 7.868808178019215e-05,
+      "loss": 0.6904,
+      "step": 102
+    },
+    {
+      "epoch": 0.52736,
+      "grad_norm": 0.9916974084040032,
+      "learning_rate": 7.862670840995836e-05,
+      "loss": 0.6788,
+      "step": 103
+    },
+    {
+      "epoch": 0.53248,
+      "grad_norm": 1.7935336303746816,
+      "learning_rate": 7.856395714934718e-05,
+      "loss": 0.6952,
+      "step": 104
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 1.4946062420453348,
+      "learning_rate": 7.849983023681905e-05,
+      "loss": 0.6853,
+      "step": 105
+    },
+    {
+      "epoch": 0.54272,
+      "grad_norm": 1.245185249022715,
+      "learning_rate": 7.84343299599066e-05,
+      "loss": 0.6729,
+      "step": 106
+    },
+    {
+      "epoch": 0.54784,
+      "grad_norm": 2.0094790105419738,
+      "learning_rate": 7.836745865513304e-05,
+      "loss": 0.682,
+      "step": 107
+    },
+    {
+      "epoch": 0.55296,
+      "grad_norm": 1.065814539722233,
+      "learning_rate": 7.829921870792886e-05,
+      "loss": 0.6764,
+      "step": 108
+    },
+    {
+      "epoch": 0.55808,
+      "grad_norm": 1.9579572631250968,
+      "learning_rate": 7.822961255254669e-05,
+      "loss": 0.6756,
+      "step": 109
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 1.52170370675673,
+      "learning_rate": 7.815864267197448e-05,
+      "loss": 0.6816,
+      "step": 110
+    },
+    {
+      "epoch": 0.56832,
+      "grad_norm": 1.704295073143993,
+      "learning_rate": 7.808631159784693e-05,
+      "loss": 0.6807,
+      "step": 111
+    },
+    {
+      "epoch": 0.57344,
+      "grad_norm": 1.0971860063652723,
+      "learning_rate": 7.801262191035518e-05,
+      "loss": 0.6775,
+      "step": 112
+    },
+    {
+      "epoch": 0.57856,
+      "grad_norm": 1.2658364692181785,
+      "learning_rate": 7.79375762381548e-05,
+      "loss": 0.6677,
+      "step": 113
+    },
+    {
+      "epoch": 0.58368,
+      "grad_norm": 1.1136078326808156,
+      "learning_rate": 7.786117725827195e-05,
+      "loss": 0.6706,
+      "step": 114
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 1.2432841023033496,
+      "learning_rate": 7.778342769600796e-05,
+      "loss": 0.6604,
+      "step": 115
+    },
+    {
+      "epoch": 0.59392,
+      "grad_norm": 0.9447874229879114,
+      "learning_rate": 7.770433032484204e-05,
+      "loss": 0.6674,
+      "step": 116
+    },
+    {
+      "epoch": 0.59904,
+      "grad_norm": 1.1728862828599786,
+      "learning_rate": 7.762388796633243e-05,
+      "loss": 0.6743,
+      "step": 117
+    },
+    {
+      "epoch": 0.60416,
+      "grad_norm": 1.0379645631823264,
+      "learning_rate": 7.754210349001568e-05,
+      "loss": 0.6746,
+      "step": 118
+    },
+    {
+      "epoch": 0.60928,
+      "grad_norm": 0.6784319178622393,
+      "learning_rate": 7.74589798133043e-05,
+      "loss": 0.6613,
+      "step": 119
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 1.2334508195450562,
+      "learning_rate": 7.737451990138275e-05,
+      "loss": 0.6832,
+      "step": 120
+    },
+    {
+      "epoch": 0.61952,
+      "grad_norm": 1.390023864901241,
+      "learning_rate": 7.728872676710155e-05,
+      "loss": 0.667,
+      "step": 121
+    },
+    {
+      "epoch": 0.62464,
+      "grad_norm": 1.0288050461695384,
+      "learning_rate": 7.720160347086994e-05,
+      "loss": 0.6823,
+      "step": 122
+    },
+    {
+      "epoch": 0.62976,
+      "grad_norm": 1.144163750515486,
+      "learning_rate": 7.711315312054659e-05,
+      "loss": 0.6692,
+      "step": 123
+    },
+    {
+      "epoch": 0.63488,
+      "grad_norm": 0.9460143556927034,
+      "learning_rate": 7.702337887132883e-05,
+      "loss": 0.6608,
+      "step": 124
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.617773004869984,
+      "learning_rate": 7.693228392564003e-05,
+      "loss": 0.6912,
+      "step": 125
+    },
+    {
+      "epoch": 0.64512,
+      "grad_norm": 0.8796030638057838,
+      "learning_rate": 7.68398715330154e-05,
+      "loss": 0.6545,
+      "step": 126
+    },
+    {
+      "epoch": 0.65024,
+      "grad_norm": 1.4530036523303451,
+      "learning_rate": 7.674614498998608e-05,
+      "loss": 0.667,
+      "step": 127
+    },
+    {
+      "epoch": 0.65536,
+      "grad_norm": 1.6001652553788648,
+      "learning_rate": 7.66511076399615e-05,
+      "loss": 0.686,
+      "step": 128
+    },
+    {
+      "epoch": 0.66048,
+      "grad_norm": 0.9295070052125337,
+      "learning_rate": 7.655476287311017e-05,
+      "loss": 0.6726,
+      "step": 129
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 1.981385007424261,
+      "learning_rate": 7.645711412623871e-05,
+      "loss": 0.6955,
+      "step": 130
+    },
+    {
+      "epoch": 0.67072,
+      "grad_norm": 1.1198337635879976,
+      "learning_rate": 7.635816488266925e-05,
+      "loss": 0.6747,
+      "step": 131
+    },
+    {
+      "epoch": 0.67584,
+      "grad_norm": 1.8403136092865524,
+      "learning_rate": 7.62579186721152e-05,
+      "loss": 0.6817,
+      "step": 132
+    },
+    {
+      "epoch": 0.68096,
+      "grad_norm": 1.211708500347191,
+      "learning_rate": 7.615637907055533e-05,
+      "loss": 0.6673,
+      "step": 133
+    },
+    {
+      "epoch": 0.68608,
+      "grad_norm": 1.619702658720517,
+      "learning_rate": 7.60535497001062e-05,
+      "loss": 0.6659,
+      "step": 134
+    },
+    {
+      "epoch": 0.6912,
+      "grad_norm": 1.3880946828869691,
+      "learning_rate": 7.594943422889293e-05,
+      "loss": 0.6876,
+      "step": 135
+    },
+    {
+      "epoch": 0.69632,
+      "grad_norm": 1.3898444786546138,
+      "learning_rate": 7.584403637091839e-05,
+      "loss": 0.6697,
+      "step": 136
+    },
+    {
+      "epoch": 0.70144,
+      "grad_norm": 1.1550848443588233,
+      "learning_rate": 7.57373598859307e-05,
+      "loss": 0.6708,
+      "step": 137
+    },
+    {
+      "epoch": 0.70656,
+      "grad_norm": 1.2208384976092193,
+      "learning_rate": 7.562940857928914e-05,
+      "loss": 0.6547,
+      "step": 138
+    },
+    {
+      "epoch": 0.71168,
+      "grad_norm": 0.7766104942706619,
+      "learning_rate": 7.552018630182829e-05,
+      "loss": 0.6548,
+      "step": 139
+    },
+    {
+      "epoch": 0.7168,
+      "grad_norm": 1.2980119993455839,
+      "learning_rate": 7.540969694972086e-05,
+      "loss": 0.6642,
+      "step": 140
+    },
+    {
+      "epoch": 0.72192,
+      "grad_norm": 1.1032988591515491,
+      "learning_rate": 7.529794446433853e-05,
+      "loss": 0.6567,
+      "step": 141
+    },
+    {
+      "epoch": 0.72704,
+      "grad_norm": 1.0687712643010066,
+      "learning_rate": 7.51849328321114e-05,
+      "loss": 0.659,
+      "step": 142
+    },
+    {
+      "epoch": 0.73216,
+      "grad_norm": 0.9993909705876705,
+      "learning_rate": 7.507066608438587e-05,
+      "loss": 0.6653,
+      "step": 143
+    },
+    {
+      "epoch": 0.73728,
+      "grad_norm": 1.4634026159707303,
+      "learning_rate": 7.495514829728072e-05,
+      "loss": 0.6633,
+      "step": 144
+    },
+    {
+      "epoch": 0.7424,
+      "grad_norm": 0.9096724763021085,
+      "learning_rate": 7.483838359154175e-05,
+      "loss": 0.6735,
+      "step": 145
+    },
+    {
+      "epoch": 0.74752,
+      "grad_norm": 1.359529554918389,
+      "learning_rate": 7.472037613239487e-05,
+      "loss": 0.6678,
+      "step": 146
+    },
+    {
+      "epoch": 0.75264,
+      "grad_norm": 0.989415493913379,
+      "learning_rate": 7.460113012939733e-05,
+      "loss": 0.6637,
+      "step": 147
+    },
+    {
+      "epoch": 0.75776,
+      "grad_norm": 1.3223608862498835,
+      "learning_rate": 7.448064983628775e-05,
+      "loss": 0.6645,
+      "step": 148
+    },
+    {
+      "epoch": 0.76288,
+      "grad_norm": 1.1620124411680315,
+      "learning_rate": 7.435893955083422e-05,
+      "loss": 0.6513,
+      "step": 149
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 1.0648774531623415,
+      "learning_rate": 7.423600361468117e-05,
+      "loss": 0.662,
+      "step": 150
+    },
+    {
+      "epoch": 0.77312,
+      "grad_norm": 1.3007725912379944,
+      "learning_rate": 7.411184641319429e-05,
+      "loss": 0.6631,
+      "step": 151
+    },
+    {
+      "epoch": 0.77824,
+      "grad_norm": 0.8104819966143042,
+      "learning_rate": 7.398647237530426e-05,
+      "loss": 0.64,
+      "step": 152
+    },
+    {
+      "epoch": 0.78336,
+      "grad_norm": 1.0085673778184652,
+      "learning_rate": 7.385988597334867e-05,
+      "loss": 0.6574,
+      "step": 153
+    },
+    {
+      "epoch": 0.78848,
+      "grad_norm": 1.2945419726019807,
+      "learning_rate": 7.373209172291253e-05,
+      "loss": 0.6475,
+      "step": 154
+    },
+    {
+      "epoch": 0.7936,
+      "grad_norm": 1.0304441022051967,
+      "learning_rate": 7.360309418266715e-05,
+      "loss": 0.6517,
+      "step": 155
+    },
+    {
+      "epoch": 0.79872,
+      "grad_norm": 0.9586021230391207,
+      "learning_rate": 7.347289795420759e-05,
+      "loss": 0.6582,
+      "step": 156
+    },
+    {
+      "epoch": 0.80384,
+      "grad_norm": 1.1290408208864031,
+      "learning_rate": 7.334150768188839e-05,
+      "loss": 0.6609,
+      "step": 157
+    },
+    {
+      "epoch": 0.80896,
+      "grad_norm": 1.1884226665875801,
+      "learning_rate": 7.320892805265807e-05,
+      "loss": 0.6699,
+      "step": 158
+    },
+    {
+      "epoch": 0.81408,
+      "grad_norm": 0.8502814476612911,
+      "learning_rate": 7.307516379589177e-05,
+      "loss": 0.6493,
+      "step": 159
+    },
+    {
+      "epoch": 0.8192,
+      "grad_norm": 0.9952888696907723,
+      "learning_rate": 7.294021968322261e-05,
+      "loss": 0.6484,
+      "step": 160
+    },
+    {
+      "epoch": 0.82432,
+      "grad_norm": 1.2290779564393681,
+      "learning_rate": 7.280410052837156e-05,
+      "loss": 0.6582,
+      "step": 161
+    },
+    {
+      "epoch": 0.82944,
+      "grad_norm": 0.8035227492015825,
+      "learning_rate": 7.266681118697555e-05,
+      "loss": 0.6699,
+      "step": 162
+    },
+    {
+      "epoch": 0.83456,
+      "grad_norm": 1.1834030000315219,
+      "learning_rate": 7.252835655641445e-05,
+      "loss": 0.6625,
+      "step": 163
+    },
+    {
+      "epoch": 0.83968,
+      "grad_norm": 0.8159782777745277,
+      "learning_rate": 7.23887415756362e-05,
+      "loss": 0.6474,
+      "step": 164
+    },
+    {
+      "epoch": 0.8448,
+      "grad_norm": 1.5925994868655784,
+      "learning_rate": 7.224797122498074e-05,
+      "loss": 0.6568,
+      "step": 165
+    },
+    {
+      "epoch": 0.84992,
+      "grad_norm": 0.7327801301614796,
+      "learning_rate": 7.210605052600233e-05,
+      "loss": 0.6535,
+      "step": 166
+    },
+    {
+      "epoch": 0.85504,
+      "grad_norm": 1.3897449798624986,
+      "learning_rate": 7.196298454129037e-05,
+      "loss": 0.6586,
+      "step": 167
+    },
+    {
+      "epoch": 0.86016,
+      "grad_norm": 0.9296756247599517,
+      "learning_rate": 7.18187783742889e-05,
+      "loss": 0.6493,
+      "step": 168
+    },
+    {
+      "epoch": 0.86528,
+      "grad_norm": 1.4033864221213181,
+      "learning_rate": 7.167343716911446e-05,
+      "loss": 0.6535,
+      "step": 169
+    },
+    {
+      "epoch": 0.8704,
+      "grad_norm": 1.2953910254634462,
+      "learning_rate": 7.15269661103726e-05,
+      "loss": 0.6522,
+      "step": 170
+    },
+    {
+      "epoch": 0.87552,
+      "grad_norm": 0.7630866652770001,
+      "learning_rate": 7.137937042297303e-05,
+      "loss": 0.6533,
+      "step": 171
+    },
+    {
+      "epoch": 0.88064,
+      "grad_norm": 1.6581576394049042,
+      "learning_rate": 7.123065537194311e-05,
+      "loss": 0.6466,
+      "step": 172
+    },
+    {
+      "epoch": 0.88576,
+      "grad_norm": 0.894168267427839,
+      "learning_rate": 7.108082626224014e-05,
+      "loss": 0.6749,
+      "step": 173
+    },
+    {
+      "epoch": 0.89088,
+      "grad_norm": 1.2060587657486854,
+      "learning_rate": 7.092988843856202e-05,
+      "loss": 0.6644,
+      "step": 174
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 0.9959417990440771,
+      "learning_rate": 7.077784728515668e-05,
+      "loss": 0.6578,
+      "step": 175
+    },
+    {
+      "epoch": 0.90112,
+      "grad_norm": 1.0821074322876068,
+      "learning_rate": 7.062470822563003e-05,
+      "loss": 0.6543,
+      "step": 176
+    },
+    {
+      "epoch": 0.90624,
+      "grad_norm": 0.7740564812790837,
+      "learning_rate": 7.047047672275235e-05,
+      "loss": 0.6595,
+      "step": 177
+    },
+    {
+      "epoch": 0.91136,
+      "grad_norm": 0.7885588166965639,
+      "learning_rate": 7.031515827826356e-05,
+      "loss": 0.6677,
+      "step": 178
+    },
+    {
+      "epoch": 0.91648,
+      "grad_norm": 0.8964594506190875,
+      "learning_rate": 7.015875843267696e-05,
+      "loss": 0.668,
+      "step": 179
+    },
+    {
+      "epoch": 0.9216,
+      "grad_norm": 0.7922996726063177,
+      "learning_rate": 7.000128276508146e-05,
+      "loss": 0.6613,
+      "step": 180
+    },
+    {
+      "epoch": 0.92672,
+      "grad_norm": 0.7656175088558786,
+      "learning_rate": 6.984273689294272e-05,
+      "loss": 0.6474,
+      "step": 181
+    },
+    {
+      "epoch": 0.93184,
+      "grad_norm": 0.9214614344976504,
+      "learning_rate": 6.968312647190267e-05,
+      "loss": 0.6326,
+      "step": 182
+    },
+    {
+      "epoch": 0.93696,
+      "grad_norm": 1.2164859593883526,
+      "learning_rate": 6.952245719557777e-05,
+      "loss": 0.6465,
+      "step": 183
+    },
+    {
+      "epoch": 0.94208,
+      "grad_norm": 0.6778217095702366,
+      "learning_rate": 6.936073479535592e-05,
+      "loss": 0.657,
+      "step": 184
+    },
+    {
+      "epoch": 0.9472,
+      "grad_norm": 0.8101497456748098,
+      "learning_rate": 6.919796504019208e-05,
+      "loss": 0.6762,
+      "step": 185
+    },
+    {
+      "epoch": 0.95232,
+      "grad_norm": 0.8876944556782685,
+      "learning_rate": 6.903415373640231e-05,
+      "loss": 0.6475,
+      "step": 186
+    },
+    {
+      "epoch": 0.95744,
+      "grad_norm": 1.0000115870935409,
+      "learning_rate": 6.886930672745684e-05,
+      "loss": 0.6561,
+      "step": 187
+    },
+    {
+      "epoch": 0.96256,
+      "grad_norm": 1.0242604450962167,
+      "learning_rate": 6.870342989377146e-05,
+      "loss": 0.6434,
+      "step": 188
+    },
+    {
+      "epoch": 0.96768,
+      "grad_norm": 0.9780023684498255,
+      "learning_rate": 6.853652915249786e-05,
+      "loss": 0.6545,
+      "step": 189
+    },
+    {
+      "epoch": 0.9728,
+      "grad_norm": 0.9266467661951784,
+      "learning_rate": 6.836861045731254e-05,
+      "loss": 0.6501,
+      "step": 190
+    },
+    {
+      "epoch": 0.97792,
+      "grad_norm": 0.9727642879091534,
+      "learning_rate": 6.819967979820441e-05,
+      "loss": 0.6457,
+      "step": 191
+    },
+    {
+      "epoch": 0.98304,
+      "grad_norm": 0.6002892901190532,
+      "learning_rate": 6.802974320126105e-05,
+      "loss": 0.6535,
+      "step": 192
+    },
+    {
+      "epoch": 0.98816,
+      "grad_norm": 0.7042555907925779,
+      "learning_rate": 6.785880672845391e-05,
+      "loss": 0.6497,
+      "step": 193
+    },
+    {
+      "epoch": 0.99328,
+      "grad_norm": 1.0659679020014914,
+      "learning_rate": 6.768687647742188e-05,
+      "loss": 0.6594,
+      "step": 194
+    },
+    {
+      "epoch": 0.9984,
+      "grad_norm": 0.919967416151866,
+      "learning_rate": 6.751395858125393e-05,
+      "loss": 0.6454,
+      "step": 195
+    },
+    {
+      "epoch": 1.00352,
+      "grad_norm": 0.832989636341993,
+      "learning_rate": 6.734005920827023e-05,
+      "loss": 0.6229,
+      "step": 196
+    },
+    {
+      "epoch": 1.00864,
+      "grad_norm": 0.9268003420050892,
+      "learning_rate": 6.716518456180219e-05,
+      "loss": 0.6067,
+      "step": 197
+    },
+    {
+      "epoch": 1.01376,
+      "grad_norm": 0.9264505307476663,
+      "learning_rate": 6.69893408799711e-05,
+      "loss": 0.594,
+      "step": 198
+    },
+    {
+      "epoch": 1.01888,
+      "grad_norm": 0.992069998033644,
+      "learning_rate": 6.681253443546566e-05,
+      "loss": 0.6027,
+      "step": 199
+    },
+    {
+      "epoch": 1.024,
+      "grad_norm": 0.9820410181623107,
+      "learning_rate": 6.663477153531819e-05,
+      "loss": 0.6091,
+      "step": 200
+    },
+    {
+      "epoch": 1.02912,
+      "grad_norm": 0.7956602066339694,
+      "learning_rate": 6.645605852067966e-05,
+      "loss": 0.6016,
+      "step": 201
+    },
+    {
+      "epoch": 1.03424,
+      "grad_norm": 1.1030274948771646,
+      "learning_rate": 6.627640176659349e-05,
+      "loss": 0.6128,
+      "step": 202
+    },
+    {
+      "epoch": 1.03936,
+      "grad_norm": 0.8027746768903161,
+      "learning_rate": 6.609580768176813e-05,
+      "loss": 0.6202,
+      "step": 203
+    },
+    {
+      "epoch": 1.04448,
+      "grad_norm": 0.6182011917758258,
+      "learning_rate": 6.591428270834848e-05,
+      "loss": 0.6197,
+      "step": 204
+    },
+    {
+      "epoch": 1.0496,
+      "grad_norm": 0.6933362237087946,
+      "learning_rate": 6.573183332168599e-05,
+      "loss": 0.5973,
+      "step": 205
+    },
+    {
+      "epoch": 1.05472,
+      "grad_norm": 0.7261413202596659,
+      "learning_rate": 6.554846603010783e-05,
+      "loss": 0.6033,
+      "step": 206
+    },
+    {
+      "epoch": 1.05984,
+      "grad_norm": 0.8175867268516304,
+      "learning_rate": 6.536418737468458e-05,
+      "loss": 0.5962,
+      "step": 207
+    },
+    {
+      "epoch": 1.06496,
+      "grad_norm": 0.8967324447987707,
+      "learning_rate": 6.517900392899696e-05,
+      "loss": 0.6138,
+      "step": 208
+    },
+    {
+      "epoch": 1.07008,
+      "grad_norm": 0.8644025442427714,
+      "learning_rate": 6.499292229890136e-05,
+      "loss": 0.6143,
+      "step": 209
+    },
+    {
+      "epoch": 1.0752,
+      "grad_norm": 0.8340587352046295,
+      "learning_rate": 6.480594912229415e-05,
+      "loss": 0.5932,
+      "step": 210
+    },
+    {
+      "epoch": 1.08032,
+      "grad_norm": 0.7992015971845043,
+      "learning_rate": 6.46180910688749e-05,
+      "loss": 0.5946,
+      "step": 211
+    },
+    {
+      "epoch": 1.08544,
+      "grad_norm": 0.7883742168020988,
+      "learning_rate": 6.442935483990848e-05,
+      "loss": 0.5912,
+      "step": 212
+    },
+    {
+      "epoch": 1.09056,
+      "grad_norm": 0.7062624710725375,
+      "learning_rate": 6.4239747167986e-05,
+      "loss": 0.6006,
+      "step": 213
+    },
+    {
+      "epoch": 1.09568,
+      "grad_norm": 0.4879826112399788,
+      "learning_rate": 6.404927481678465e-05,
+      "loss": 0.5908,
+      "step": 214
+    },
+    {
+      "epoch": 1.1008,
+      "grad_norm": 0.46707410887787343,
+      "learning_rate": 6.385794458082645e-05,
+      "loss": 0.6016,
+      "step": 215
+    },
+    {
+      "epoch": 1.10592,
+      "grad_norm": 0.46215479735805276,
+      "learning_rate": 6.366576328523579e-05,
+      "loss": 0.6059,
+      "step": 216
+    },
+    {
+      "epoch": 1.11104,
+      "grad_norm": 0.4872892344668909,
+      "learning_rate": 6.347273778549609e-05,
+      "loss": 0.5979,
+      "step": 217
+    },
+    {
+      "epoch": 1.11616,
+      "grad_norm": 0.4117850505120705,
+      "learning_rate": 6.327887496720513e-05,
+      "loss": 0.6042,
+      "step": 218
+    },
+    {
+      "epoch": 1.12128,
+      "grad_norm": 0.6094192942803528,
+      "learning_rate": 6.308418174582952e-05,
+      "loss": 0.6098,
+      "step": 219
+    },
+    {
+      "epoch": 1.1264,
+      "grad_norm": 0.47370404685867984,
+      "learning_rate": 6.288866506645799e-05,
+      "loss": 0.5967,
+      "step": 220
+    },
+    {
+      "epoch": 1.13152,
+      "grad_norm": 0.5812822920441131,
+      "learning_rate": 6.269233190355355e-05,
+      "loss": 0.589,
+      "step": 221
+    },
+    {
+      "epoch": 1.13664,
+      "grad_norm": 0.581689390915006,
+      "learning_rate": 6.249518926070488e-05,
+      "loss": 0.6033,
+      "step": 222
+    },
+    {
+      "epoch": 1.14176,
+      "grad_norm": 0.7708083440150924,
+      "learning_rate": 6.229724417037632e-05,
+      "loss": 0.6008,
+      "step": 223
+    },
+    {
+      "epoch": 1.14688,
+      "grad_norm": 0.9437745661416408,
+      "learning_rate": 6.209850369365714e-05,
+      "loss": 0.5872,
+      "step": 224
+    },
+    {
+      "epoch": 1.152,
+      "grad_norm": 1.0285174187897348,
+      "learning_rate": 6.189897492000953e-05,
+      "loss": 0.6034,
+      "step": 225
+    },
+    {
+      "epoch": 1.15712,
+      "grad_norm": 0.9256911501210775,
+      "learning_rate": 6.169866496701586e-05,
+      "loss": 0.6065,
+      "step": 226
+    },
+    {
+      "epoch": 1.16224,
+      "grad_norm": 0.8931346889464913,
+      "learning_rate": 6.149758098012464e-05,
+      "loss": 0.6162,
+      "step": 227
+    },
+    {
+      "epoch": 1.16736,
+      "grad_norm": 0.906249956782627,
+      "learning_rate": 6.129573013239571e-05,
+      "loss": 0.609,
+      "step": 228
+    },
+    {
+      "epoch": 1.17248,
+      "grad_norm": 0.9031799418470178,
+      "learning_rate": 6.109311962424433e-05,
+      "loss": 0.6043,
+      "step": 229
+    },
+    {
+      "epoch": 1.1776,
+      "grad_norm": 0.7749991050050256,
+      "learning_rate": 6.0889756683184324e-05,
+      "loss": 0.609,
+      "step": 230
+    },
+    {
+      "epoch": 1.18272,
+      "grad_norm": 0.6113291239741755,
+      "learning_rate": 6.068564856357031e-05,
+      "loss": 0.6041,
+      "step": 231
+    },
+    {
+      "epoch": 1.18784,
+      "grad_norm": 0.4610907845132779,
+      "learning_rate": 6.048080254633885e-05,
+      "loss": 0.6032,
+      "step": 232
+    },
+    {
+      "epoch": 1.19296,
+      "grad_norm": 0.7535273976467541,
+      "learning_rate": 6.0275225938748764e-05,
+      "loss": 0.5856,
+      "step": 233
+    },
+    {
+      "epoch": 1.19808,
+      "grad_norm": 0.7357293529763054,
+      "learning_rate": 6.006892607412047e-05,
+      "loss": 0.5871,
+      "step": 234
+    },
+    {
+      "epoch": 1.2032,
+      "grad_norm": 0.7850706730871837,
+      "learning_rate": 5.98619103115744e-05,
+      "loss": 0.6022,
+      "step": 235
+    },
+    {
+      "epoch": 1.20832,
+      "grad_norm": 0.9520874763788266,
+      "learning_rate": 5.965418603576842e-05,
+      "loss": 0.6029,
+      "step": 236
+    },
+    {
+      "epoch": 1.21344,
+      "grad_norm": 0.6234215842388057,
+      "learning_rate": 5.944576065663449e-05,
+      "loss": 0.5989,
+      "step": 237
+    },
+    {
+      "epoch": 1.21856,
+      "grad_norm": 0.5581618761087828,
+      "learning_rate": 5.9236641609114295e-05,
+      "loss": 0.5852,
+      "step": 238
+    },
+    {
+      "epoch": 1.2236799999999999,
+      "grad_norm": 0.8280934815331484,
+      "learning_rate": 5.902683635289403e-05,
+      "loss": 0.6065,
+      "step": 239
+    },
+    {
+      "epoch": 1.2288000000000001,
+      "grad_norm": 0.8871893394061044,
+      "learning_rate": 5.881635237213829e-05,
+      "loss": 0.5901,
+      "step": 240
+    },
+    {
+      "epoch": 1.23392,
+      "grad_norm": 0.6767238651749208,
+      "learning_rate": 5.8605197175223095e-05,
+      "loss": 0.6073,
+      "step": 241
+    },
+    {
+      "epoch": 1.23904,
+      "grad_norm": 0.688896399179518,
+      "learning_rate": 5.839337829446809e-05,
+      "loss": 0.6049,
+      "step": 242
+    },
+    {
+      "epoch": 1.24416,
+      "grad_norm": 0.5804470630643502,
+      "learning_rate": 5.8180903285867804e-05,
+      "loss": 0.6036,
+      "step": 243
+    },
+    {
+      "epoch": 1.24928,
+      "grad_norm": 0.6457016614930509,
+      "learning_rate": 5.7967779728822134e-05,
+      "loss": 0.5981,
+      "step": 244
+    },
+    {
+      "epoch": 1.2544,
+      "grad_norm": 0.8414381704157707,
+      "learning_rate": 5.775401522586596e-05,
+      "loss": 0.6109,
+      "step": 245
+    },
+    {
+      "epoch": 1.25952,
+      "grad_norm": 0.8374711125166566,
+      "learning_rate": 5.753961740239799e-05,
+      "loss": 0.6005,
+      "step": 246
+    },
+    {
+      "epoch": 1.26464,
+      "grad_norm": 0.5542646630500834,
+      "learning_rate": 5.732459390640868e-05,
+      "loss": 0.6103,
+      "step": 247
+    },
+    {
+      "epoch": 1.26976,
+      "grad_norm": 0.5875945873220082,
+      "learning_rate": 5.7108952408207485e-05,
+      "loss": 0.6128,
+      "step": 248
+    },
+    {
+      "epoch": 1.27488,
+      "grad_norm": 0.5087534889785712,
+      "learning_rate": 5.689270060014919e-05,
+      "loss": 0.5958,
+      "step": 249
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.4031024078366132,
+      "learning_rate": 5.66758461963595e-05,
+      "loss": 0.589,
+      "step": 250
+    },
+    {
+      "epoch": 1.28512,
+      "grad_norm": 0.4918002116220245,
+      "learning_rate": 5.645839693245995e-05,
+      "loss": 0.5982,
+      "step": 251
+    },
+    {
+      "epoch": 1.29024,
+      "grad_norm": 0.48513251035233335,
+      "learning_rate": 5.6240360565291874e-05,
+      "loss": 0.5981,
+      "step": 252
+    },
+    {
+      "epoch": 1.29536,
+      "grad_norm": 0.41091372229331263,
+      "learning_rate": 5.602174487263971e-05,
+      "loss": 0.5902,
+      "step": 253
+    },
+    {
+      "epoch": 1.30048,
+      "grad_norm": 0.5989492893332392,
+      "learning_rate": 5.580255765295358e-05,
+      "loss": 0.6129,
+      "step": 254
+    },
+    {
+      "epoch": 1.3056,
+      "grad_norm": 0.5776722852203174,
+      "learning_rate": 5.5582806725071135e-05,
+      "loss": 0.606,
+      "step": 255
+    },
+    {
+      "epoch": 1.3107199999999999,
+      "grad_norm": 0.4466289358616809,
+      "learning_rate": 5.536249992793856e-05,
+      "loss": 0.5992,
+      "step": 256
+    },
+    {
+      "epoch": 1.3158400000000001,
+      "grad_norm": 0.4281414323403905,
+      "learning_rate": 5.514164512033099e-05,
+      "loss": 0.5865,
+      "step": 257
+    },
+    {
+      "epoch": 1.32096,
+      "grad_norm": 0.3767339434891749,
+      "learning_rate": 5.4920250180572164e-05,
+      "loss": 0.6049,
+      "step": 258
+    },
+    {
+      "epoch": 1.32608,
+      "grad_norm": 0.4541584747628375,
+      "learning_rate": 5.469832300625343e-05,
+      "loss": 0.5891,
+      "step": 259
+    },
+    {
+      "epoch": 1.3312,
+      "grad_norm": 0.5852329624714224,
+      "learning_rate": 5.447587151395193e-05,
+      "loss": 0.6013,
+      "step": 260
+    },
+    {
+      "epoch": 1.33632,
+      "grad_norm": 0.5470836134042647,
+      "learning_rate": 5.4252903638948304e-05,
+      "loss": 0.5871,
+      "step": 261
+    },
+    {
+      "epoch": 1.34144,
+      "grad_norm": 0.3981165895490368,
+      "learning_rate": 5.402942733494355e-05,
+      "loss": 0.5988,
+      "step": 262
+    },
+    {
+      "epoch": 1.34656,
+      "grad_norm": 0.8090553651844287,
+      "learning_rate": 5.380545057377533e-05,
+      "loss": 0.5919,
+      "step": 263
+    },
+    {
+      "epoch": 1.35168,
+      "grad_norm": 0.6973495560228751,
+      "learning_rate": 5.358098134513359e-05,
+      "loss": 0.6099,
+      "step": 264
+    },
+    {
+      "epoch": 1.3568,
+      "grad_norm": 0.4439863252173243,
+      "learning_rate": 5.335602765627556e-05,
+      "loss": 0.6112,
+      "step": 265
+    },
+    {
+      "epoch": 1.36192,
+      "grad_norm": 0.45728264462579415,
+      "learning_rate": 5.313059753174011e-05,
+      "loss": 0.5999,
+      "step": 266
+    },
+    {
+      "epoch": 1.36704,
+      "grad_norm": 0.5814907075742051,
+      "learning_rate": 5.2904699013061525e-05,
+      "loss": 0.5996,
+      "step": 267
+    },
+    {
+      "epoch": 1.37216,
+      "grad_norm": 0.5358642327581291,
+      "learning_rate": 5.2678340158482605e-05,
+      "loss": 0.6085,
+      "step": 268
+    },
+    {
+      "epoch": 1.37728,
+      "grad_norm": 0.5406775247232574,
+      "learning_rate": 5.245152904266722e-05,
+      "loss": 0.6234,
+      "step": 269
+    },
+    {
+      "epoch": 1.3824,
+      "grad_norm": 0.7178735700752464,
+      "learning_rate": 5.2224273756412335e-05,
+      "loss": 0.6273,
+      "step": 270
+    },
+    {
+      "epoch": 1.3875199999999999,
+      "grad_norm": 0.7039970725546744,
+      "learning_rate": 5.1996582406359286e-05,
+      "loss": 0.5899,
+      "step": 271
+    },
+    {
+      "epoch": 1.39264,
+      "grad_norm": 0.5122004385795155,
+      "learning_rate": 5.17684631147047e-05,
+      "loss": 0.5772,
+      "step": 272
+    },
+    {
+      "epoch": 1.39776,
+      "grad_norm": 0.5292102951542996,
+      "learning_rate": 5.1539924018910705e-05,
+      "loss": 0.5963,
+      "step": 273
+    },
+    {
+      "epoch": 1.4028800000000001,
+      "grad_norm": 0.5393146339840262,
+      "learning_rate": 5.131097327141467e-05,
+      "loss": 0.6078,
+      "step": 274
+    },
+    {
+      "epoch": 1.408,
+      "grad_norm": 0.708939654981038,
+      "learning_rate": 5.108161903933837e-05,
+      "loss": 0.6005,
+      "step": 275
+    },
+    {
+      "epoch": 1.41312,
+      "grad_norm": 0.7037903511384231,
+      "learning_rate": 5.085186950419671e-05,
+      "loss": 0.5942,
+      "step": 276
+    },
+    {
+      "epoch": 1.41824,
+      "grad_norm": 0.4165090417742424,
+      "learning_rate": 5.0621732861605754e-05,
+      "loss": 0.5906,
+      "step": 277
+    },
+    {
+      "epoch": 1.42336,
+      "grad_norm": 0.6642875595643233,
+      "learning_rate": 5.039121732099052e-05,
+      "loss": 0.5891,
+      "step": 278
+    },
+    {
+      "epoch": 1.42848,
+      "grad_norm": 0.7547032791125439,
+      "learning_rate": 5.0160331105292036e-05,
+      "loss": 0.6008,
+      "step": 279
+    },
+    {
+      "epoch": 1.4336,
+      "grad_norm": 0.733986554958704,
+      "learning_rate": 4.992908245067403e-05,
+      "loss": 0.5915,
+      "step": 280
+    },
+    {
+      "epoch": 1.43872,
+      "grad_norm": 0.47614304475036945,
+      "learning_rate": 4.969747960622912e-05,
+      "loss": 0.6011,
+      "step": 281
+    },
+    {
+      "epoch": 1.44384,
+      "grad_norm": 0.585876239848498,
+      "learning_rate": 4.9465530833684614e-05,
+      "loss": 0.5912,
+      "step": 282
+    },
+    {
+      "epoch": 1.44896,
+      "grad_norm": 0.6945418578556748,
+      "learning_rate": 4.923324440710773e-05,
+      "loss": 0.6106,
+      "step": 283
+    },
+    {
+      "epoch": 1.45408,
+      "grad_norm": 0.6261522164548919,
+      "learning_rate": 4.900062861261044e-05,
+      "loss": 0.5941,
+      "step": 284
+    },
+    {
+      "epoch": 1.4592,
+      "grad_norm": 0.5342894182102067,
+      "learning_rate": 4.8767691748053955e-05,
+      "loss": 0.5922,
+      "step": 285
+    },
+    {
+      "epoch": 1.46432,
+      "grad_norm": 0.5838783944360918,
+      "learning_rate": 4.8534442122752674e-05,
+      "loss": 0.6075,
+      "step": 286
+    },
+    {
+      "epoch": 1.46944,
+      "grad_norm": 0.6486938746583781,
+      "learning_rate": 4.830088805717779e-05,
+      "loss": 0.5954,
+      "step": 287
+    },
+    {
+      "epoch": 1.4745599999999999,
+      "grad_norm": 0.5719218459201613,
+      "learning_rate": 4.806703788266045e-05,
+      "loss": 0.6018,
+      "step": 288
+    },
+    {
+      "epoch": 1.47968,
+      "grad_norm": 0.5120146948133751,
+      "learning_rate": 4.783289994109459e-05,
+      "loss": 0.5912,
+      "step": 289
+    },
+    {
+      "epoch": 1.4848,
+      "grad_norm": 0.6804527806275119,
+      "learning_rate": 4.759848258463942e-05,
+      "loss": 0.5896,
+      "step": 290
+    },
+    {
+      "epoch": 1.4899200000000001,
+      "grad_norm": 0.5487294767485589,
+      "learning_rate": 4.736379417542134e-05,
+      "loss": 0.5882,
+      "step": 291
+    },
+    {
+      "epoch": 1.49504,
+      "grad_norm": 0.4853965765407641,
+      "learning_rate": 4.712884308523579e-05,
+      "loss": 0.5888,
+      "step": 292
+    },
+    {
+      "epoch": 1.5001600000000002,
+      "grad_norm": 0.586848411445172,
+      "learning_rate": 4.689363769524854e-05,
+      "loss": 0.5959,
+      "step": 293
+    },
+    {
+      "epoch": 1.50528,
+      "grad_norm": 0.49095884598520845,
+      "learning_rate": 4.6658186395696735e-05,
+      "loss": 0.6054,
+      "step": 294
+    },
+    {
+      "epoch": 1.5104,
+      "grad_norm": 0.5564665878511703,
+      "learning_rate": 4.64224975855896e-05,
+      "loss": 0.6059,
+      "step": 295
+    },
+    {
+      "epoch": 1.51552,
+      "grad_norm": 0.46245666700892735,
+      "learning_rate": 4.618657967240879e-05,
+      "loss": 0.5891,
+      "step": 296
+    },
+    {
+      "epoch": 1.52064,
+      "grad_norm": 0.533299845495822,
+      "learning_rate": 4.595044107180857e-05,
+      "loss": 0.6044,
+      "step": 297
+    },
+    {
+      "epoch": 1.52576,
+      "grad_norm": 0.42525913312635544,
+      "learning_rate": 4.571409020731554e-05,
+      "loss": 0.5869,
+      "step": 298
+    },
+    {
+      "epoch": 1.53088,
+      "grad_norm": 0.5215704284954067,
+      "learning_rate": 4.547753551002814e-05,
+      "loss": 0.5923,
+      "step": 299
+    },
+    {
+      "epoch": 1.536,
+      "grad_norm": 0.47973640340231194,
+      "learning_rate": 4.524078541831597e-05,
+      "loss": 0.5891,
+      "step": 300
+    },
+    {
+      "epoch": 1.54112,
+      "grad_norm": 0.4327598342601823,
+      "learning_rate": 4.500384837751868e-05,
+      "loss": 0.5958,
+      "step": 301
+    },
+    {
+      "epoch": 1.54624,
+      "grad_norm": 0.5990121963555108,
+      "learning_rate": 4.476673283964481e-05,
+      "loss": 0.5856,
+      "step": 302
+    },
+    {
+      "epoch": 1.5513599999999999,
+      "grad_norm": 0.7131944725290759,
+      "learning_rate": 4.4529447263070215e-05,
+      "loss": 0.5942,
+      "step": 303
+    },
+    {
+      "epoch": 1.55648,
+      "grad_norm": 0.5329359342241601,
+      "learning_rate": 4.429200011223638e-05,
+      "loss": 0.597,
+      "step": 304
+    },
+    {
+      "epoch": 1.5615999999999999,
+      "grad_norm": 0.4477137886623582,
+      "learning_rate": 4.4054399857348426e-05,
+      "loss": 0.5905,
+      "step": 305
+    },
+    {
+      "epoch": 1.5667200000000001,
+      "grad_norm": 0.7235716501912173,
+      "learning_rate": 4.381665497407304e-05,
+      "loss": 0.6044,
+      "step": 306
+    },
+    {
+      "epoch": 1.57184,
+      "grad_norm": 0.5656431568517662,
+      "learning_rate": 4.3578773943236066e-05,
+      "loss": 0.5935,
+      "step": 307
+    },
+    {
+      "epoch": 1.5769600000000001,
+      "grad_norm": 0.5443076040573943,
+      "learning_rate": 4.334076525051998e-05,
+      "loss": 0.6085,
+      "step": 308
+    },
+    {
+      "epoch": 1.58208,
+      "grad_norm": 0.44142667616814085,
+      "learning_rate": 4.3102637386161214e-05,
+      "loss": 0.5974,
+      "step": 309
+    },
+    {
+      "epoch": 1.5872000000000002,
+      "grad_norm": 0.4881724554496324,
+      "learning_rate": 4.2864398844647333e-05,
+      "loss": 0.5982,
+      "step": 310
+    },
+    {
+      "epoch": 1.59232,
+      "grad_norm": 0.48224923855420715,
+      "learning_rate": 4.2626058124413934e-05,
+      "loss": 0.5957,
+      "step": 311
+    },
+    {
+      "epoch": 1.59744,
+      "grad_norm": 0.5087339085522424,
+      "learning_rate": 4.23876237275415e-05,
+      "loss": 0.5919,
+      "step": 312
+    },
+    {
+      "epoch": 1.60256,
+      "grad_norm": 0.4533960557571388,
+      "learning_rate": 4.2149104159452194e-05,
+      "loss": 0.6004,
+      "step": 313
+    },
+    {
+      "epoch": 1.60768,
+      "grad_norm": 0.511203081310295,
+      "learning_rate": 4.191050792860641e-05,
+      "loss": 0.5994,
+      "step": 314
+    },
+    {
+      "epoch": 1.6128,
+      "grad_norm": 0.4223027905990951,
+      "learning_rate": 4.1671843546199195e-05,
+      "loss": 0.5887,
+      "step": 315
+    },
+    {
+      "epoch": 1.61792,
+      "grad_norm": 0.4494541281616636,
+      "learning_rate": 4.143311952585673e-05,
+      "loss": 0.5775,
+      "step": 316
+    },
+    {
+      "epoch": 1.62304,
+      "grad_norm": 0.49316734261443734,
+      "learning_rate": 4.119434438333261e-05,
+      "loss": 0.5937,
+      "step": 317
+    },
+    {
+      "epoch": 1.62816,
+      "grad_norm": 0.36392011844676875,
+      "learning_rate": 4.095552663620403e-05,
+      "loss": 0.5962,
+      "step": 318
+    },
+    {
+      "epoch": 1.63328,
+      "grad_norm": 0.4968107231897248,
+      "learning_rate": 4.0716674803568005e-05,
+      "loss": 0.5919,
+      "step": 319
+    },
+    {
+      "epoch": 1.6383999999999999,
+      "grad_norm": 0.37525246925036004,
+      "learning_rate": 4.0477797405737414e-05,
+      "loss": 0.5969,
+      "step": 320
+    },
+    {
+      "epoch": 1.64352,
+      "grad_norm": 0.5908101591078128,
+      "learning_rate": 4.023890296393714e-05,
+      "loss": 0.6099,
+      "step": 321
+    },
+    {
+      "epoch": 1.6486399999999999,
+      "grad_norm": 0.3984607947668425,
+      "learning_rate": 4e-05,
+      "loss": 0.6004,
+      "step": 322
+    },
+    {
+      "epoch": 1.6537600000000001,
+      "grad_norm": 0.5149940417738129,
+      "learning_rate": 3.976109703606288e-05,
+      "loss": 0.6006,
+      "step": 323
+    },
+    {
+      "epoch": 1.65888,
+      "grad_norm": 0.4005964486620091,
+      "learning_rate": 3.952220259426259e-05,
+      "loss": 0.5803,
+      "step": 324
+    },
+    {
+      "epoch": 1.6640000000000001,
+      "grad_norm": 0.35086767512207107,
+      "learning_rate": 3.928332519643201e-05,
+      "loss": 0.585,
+      "step": 325
+    },
+    {
+      "epoch": 1.66912,
+      "grad_norm": 0.42885330848275394,
+      "learning_rate": 3.9044473363796e-05,
+      "loss": 0.5958,
+      "step": 326
+    },
+    {
+      "epoch": 1.67424,
+      "grad_norm": 0.38751578060305375,
+      "learning_rate": 3.8805655616667406e-05,
+      "loss": 0.6014,
+      "step": 327
+    },
+    {
+      "epoch": 1.67936,
+      "grad_norm": 0.3603759776041056,
+      "learning_rate": 3.8566880474143284e-05,
+      "loss": 0.6001,
+      "step": 328
+    },
+    {
+      "epoch": 1.68448,
+      "grad_norm": 0.4283886804732866,
+      "learning_rate": 3.832815645380083e-05,
+      "loss": 0.5999,
+      "step": 329
+    },
+    {
+      "epoch": 1.6896,
+      "grad_norm": 0.3337168354335494,
+      "learning_rate": 3.80894920713936e-05,
+      "loss": 0.5918,
+      "step": 330
+    },
+    {
+      "epoch": 1.69472,
+      "grad_norm": 0.383147123590219,
+      "learning_rate": 3.785089584054781e-05,
+      "loss": 0.604,
+      "step": 331
+    },
+    {
+      "epoch": 1.69984,
+      "grad_norm": 0.39090517742398523,
+      "learning_rate": 3.761237627245851e-05,
+      "loss": 0.5784,
+      "step": 332
+    },
+    {
+      "epoch": 1.70496,
+      "grad_norm": 0.41160894437025963,
+      "learning_rate": 3.737394187558608e-05,
+      "loss": 0.5899,
+      "step": 333
+    },
+    {
+      "epoch": 1.71008,
+      "grad_norm": 0.4348994266327126,
+      "learning_rate": 3.713560115535268e-05,
+      "loss": 0.584,
+      "step": 334
+    },
+    {
+      "epoch": 1.7151999999999998,
+      "grad_norm": 0.44033711538569453,
+      "learning_rate": 3.689736261383879e-05,
+      "loss": 0.6019,
+      "step": 335
+    },
+    {
+      "epoch": 1.72032,
+      "grad_norm": 0.46185741930953395,
+      "learning_rate": 3.6659234749480043e-05,
+      "loss": 0.5979,
+      "step": 336
+    },
+    {
+      "epoch": 1.7254399999999999,
+      "grad_norm": 0.4557337264056941,
+      "learning_rate": 3.642122605676396e-05,
+      "loss": 0.6102,
+      "step": 337
+    },
+    {
+      "epoch": 1.73056,
+      "grad_norm": 0.6447579696011232,
+      "learning_rate": 3.6183345025926966e-05,
+      "loss": 0.6094,
+      "step": 338
+    },
+    {
+      "epoch": 1.73568,
+      "grad_norm": 0.35483221584671387,
+      "learning_rate": 3.594560014265159e-05,
+      "loss": 0.5859,
+      "step": 339
+    },
+    {
+      "epoch": 1.7408000000000001,
+      "grad_norm": 0.5619949022675687,
+      "learning_rate": 3.570799988776364e-05,
+      "loss": 0.602,
+      "step": 340
+    },
+    {
+      "epoch": 1.74592,
+      "grad_norm": 0.4808833851628103,
+      "learning_rate": 3.5470552736929785e-05,
+      "loss": 0.59,
+      "step": 341
+    },
+    {
+      "epoch": 1.7510400000000002,
+      "grad_norm": 0.4280664310256486,
+      "learning_rate": 3.5233267160355206e-05,
+      "loss": 0.5973,
+      "step": 342
+    },
+    {
+      "epoch": 1.75616,
+      "grad_norm": 0.34606786385637417,
+      "learning_rate": 3.4996151622481336e-05,
+      "loss": 0.5987,
+      "step": 343
+    },
+    {
+      "epoch": 1.76128,
+      "grad_norm": 0.47886946752668064,
+      "learning_rate": 3.475921458168404e-05,
+      "loss": 0.5974,
+      "step": 344
+    },
+    {
+      "epoch": 1.7664,
+      "grad_norm": 0.35895795443414463,
+      "learning_rate": 3.452246448997187e-05,
+      "loss": 0.5825,
+      "step": 345
+    },
+    {
+      "epoch": 1.77152,
+      "grad_norm": 0.6237524007460633,
+      "learning_rate": 3.428590979268448e-05,
+      "loss": 0.593,
+      "step": 346
+    },
+    {
+      "epoch": 1.77664,
+      "grad_norm": 0.5291377288153889,
+      "learning_rate": 3.4049558928191435e-05,
+      "loss": 0.5885,
+      "step": 347
+    },
+    {
+      "epoch": 1.78176,
+      "grad_norm": 0.5075647920106248,
+      "learning_rate": 3.381342032759122e-05,
+      "loss": 0.5982,
+      "step": 348
+    },
+    {
+      "epoch": 1.78688,
+      "grad_norm": 0.49287865419478344,
+      "learning_rate": 3.357750241441041e-05,
+      "loss": 0.5802,
+      "step": 349
+    },
+    {
+      "epoch": 1.792,
+      "grad_norm": 0.543103939185933,
+      "learning_rate": 3.334181360430327e-05,
+      "loss": 0.5936,
+      "step": 350
+    },
+    {
+      "epoch": 1.79712,
+      "grad_norm": 0.3894840623083752,
+      "learning_rate": 3.310636230475148e-05,
+      "loss": 0.588,
+      "step": 351
+    },
+    {
+      "epoch": 1.8022399999999998,
+      "grad_norm": 0.43995147863147344,
+      "learning_rate": 3.287115691476422e-05,
+      "loss": 0.6006,
+      "step": 352
+    },
+    {
+      "epoch": 1.80736,
+      "grad_norm": 0.3921709111241505,
+      "learning_rate": 3.263620582457868e-05,
+      "loss": 0.5939,
+      "step": 353
+    },
+    {
+      "epoch": 1.8124799999999999,
+      "grad_norm": 0.31433310266314246,
+      "learning_rate": 3.24015174153606e-05,
+      "loss": 0.5893,
+      "step": 354
+    },
+    {
+      "epoch": 1.8176,
+      "grad_norm": 0.43895209082760966,
+      "learning_rate": 3.2167100058905415e-05,
+      "loss": 0.5899,
+      "step": 355
+    },
+    {
+      "epoch": 1.82272,
+      "grad_norm": 0.35289039516589493,
+      "learning_rate": 3.1932962117339566e-05,
+      "loss": 0.6034,
+      "step": 356
+    },
+    {
+      "epoch": 1.8278400000000001,
+      "grad_norm": 0.3361411963291542,
+      "learning_rate": 3.169911194282222e-05,
+      "loss": 0.5794,
+      "step": 357
+    },
+    {
+      "epoch": 1.83296,
+      "grad_norm": 0.35346529955825,
+      "learning_rate": 3.1465557877247325e-05,
+      "loss": 0.5872,
+      "step": 358
+    },
+    {
+      "epoch": 1.8380800000000002,
+      "grad_norm": 0.3031581900611866,
+      "learning_rate": 3.123230825194606e-05,
+      "loss": 0.5771,
+      "step": 359
+    },
+    {
+      "epoch": 1.8432,
+      "grad_norm": 0.4329864321858136,
+      "learning_rate": 3.099937138738958e-05,
+      "loss": 0.5927,
+      "step": 360
+    },
+    {
+      "epoch": 1.84832,
+      "grad_norm": 0.35326273185796286,
+      "learning_rate": 3.076675559289228e-05,
+      "loss": 0.5996,
+      "step": 361
+    },
+    {
+      "epoch": 1.85344,
+      "grad_norm": 0.35925676205986545,
+      "learning_rate": 3.05344691663154e-05,
+      "loss": 0.5781,
+      "step": 362
+    },
+    {
+      "epoch": 1.85856,
+      "grad_norm": 0.3536221477128384,
+      "learning_rate": 3.0302520393770894e-05,
+      "loss": 0.5934,
+      "step": 363
+    },
+    {
+      "epoch": 1.86368,
+      "grad_norm": 0.42419086729757743,
+      "learning_rate": 3.0070917549325983e-05,
+      "loss": 0.592,
+      "step": 364
+    },
+    {
+      "epoch": 1.8688,
+      "grad_norm": 0.3833286680510395,
+      "learning_rate": 2.9839668894707974e-05,
+      "loss": 0.5858,
+      "step": 365
+    },
+    {
+      "epoch": 1.87392,
+      "grad_norm": 0.3817012337576983,
+      "learning_rate": 2.960878267900948e-05,
+      "loss": 0.5816,
+      "step": 366
+    },
+    {
+      "epoch": 1.87904,
+      "grad_norm": 0.400413399573491,
+      "learning_rate": 2.937826713839426e-05,
+      "loss": 0.5839,
+      "step": 367
+    },
+    {
+      "epoch": 1.88416,
+      "grad_norm": 0.36874490386425984,
+      "learning_rate": 2.9148130495803307e-05,
+      "loss": 0.5854,
+      "step": 368
+    },
+    {
+      "epoch": 1.8892799999999998,
+      "grad_norm": 0.39202924423336344,
+      "learning_rate": 2.8918380960661624e-05,
+      "loss": 0.5941,
+      "step": 369
+    },
+    {
+      "epoch": 1.8944,
+      "grad_norm": 0.35104738229454596,
+      "learning_rate": 2.8689026728585338e-05,
+      "loss": 0.5845,
+      "step": 370
+    },
+    {
+      "epoch": 1.8995199999999999,
+      "grad_norm": 0.42429376258538937,
+      "learning_rate": 2.8460075981089305e-05,
+      "loss": 0.5827,
+      "step": 371
+    },
+    {
+      "epoch": 1.90464,
+      "grad_norm": 0.36436129019616925,
+      "learning_rate": 2.82315368852953e-05,
+      "loss": 0.5754,
+      "step": 372
+    },
+    {
+      "epoch": 1.90976,
+      "grad_norm": 0.42556325640476184,
+      "learning_rate": 2.8003417593640727e-05,
+      "loss": 0.599,
+      "step": 373
+    },
+    {
+      "epoch": 1.9148800000000001,
+      "grad_norm": 0.3851458476612929,
+      "learning_rate": 2.7775726243587685e-05,
+      "loss": 0.5931,
+      "step": 374
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3737224724659946,
+      "learning_rate": 2.754847095733278e-05,
+      "loss": 0.6015,
+      "step": 375
+    },
+    {
+      "epoch": 1.9251200000000002,
+      "grad_norm": 0.36520508052939255,
+      "learning_rate": 2.732165984151741e-05,
+      "loss": 0.5961,
+      "step": 376
+    },
+    {
+      "epoch": 1.93024,
+      "grad_norm": 0.3088144098356228,
+      "learning_rate": 2.709530098693849e-05,
+      "loss": 0.5898,
+      "step": 377
+    },
+    {
+      "epoch": 1.93536,
+      "grad_norm": 0.4046047102015038,
+      "learning_rate": 2.686940246825989e-05,
+      "loss": 0.6059,
+      "step": 378
+    },
+    {
+      "epoch": 1.94048,
+      "grad_norm": 0.3815617952147954,
+      "learning_rate": 2.6643972343724453e-05,
+      "loss": 0.5785,
+      "step": 379
+    },
+    {
+      "epoch": 1.9456,
+      "grad_norm": 0.3351980518182547,
+      "learning_rate": 2.6419018654866415e-05,
+      "loss": 0.585,
+      "step": 380
+    },
+    {
+      "epoch": 1.95072,
+      "grad_norm": 0.4441803690810445,
+      "learning_rate": 2.6194549426224684e-05,
+      "loss": 0.5955,
+      "step": 381
+    },
+    {
+      "epoch": 1.95584,
+      "grad_norm": 0.2725524980139723,
+      "learning_rate": 2.5970572665056465e-05,
+      "loss": 0.5871,
+      "step": 382
+    },
+    {
+      "epoch": 1.96096,
+      "grad_norm": 0.4789800834824601,
+      "learning_rate": 2.57470963610517e-05,
+      "loss": 0.597,
+      "step": 383
+    },
+    {
+      "epoch": 1.96608,
+      "grad_norm": 0.3196138403431241,
+      "learning_rate": 2.5524128486048073e-05,
+      "loss": 0.5948,
+      "step": 384
+    },
+    {
+      "epoch": 1.9712,
+      "grad_norm": 0.4883702808240898,
+      "learning_rate": 2.5301676993746592e-05,
+      "loss": 0.588,
+      "step": 385
+    },
+    {
+      "epoch": 1.9763199999999999,
+      "grad_norm": 0.4304710776740572,
+      "learning_rate": 2.5079749819427842e-05,
+      "loss": 0.5884,
+      "step": 386
+    },
+    {
+      "epoch": 1.98144,
+      "grad_norm": 0.364646635973235,
+      "learning_rate": 2.4858354879669025e-05,
+      "loss": 0.5776,
+      "step": 387
+    },
+    {
+      "epoch": 1.9865599999999999,
+      "grad_norm": 0.385522115127419,
+      "learning_rate": 2.463750007206146e-05,
+      "loss": 0.5842,
+      "step": 388
+    },
+    {
+      "epoch": 1.9916800000000001,
+      "grad_norm": 0.3482195171706265,
+      "learning_rate": 2.441719327492887e-05,
+      "loss": 0.5886,
+      "step": 389
+    },
+    {
+      "epoch": 1.9968,
+      "grad_norm": 0.4205224902290012,
+      "learning_rate": 2.4197442347046432e-05,
+      "loss": 0.5969,
+      "step": 390
+    },
+    {
+      "epoch": 2.00192,
+      "grad_norm": 0.40631741625288587,
+      "learning_rate": 2.397825512736032e-05,
+      "loss": 0.5662,
+      "step": 391
+    },
+    {
+      "epoch": 2.00704,
+      "grad_norm": 0.4013123221753668,
+      "learning_rate": 2.3759639434708142e-05,
+      "loss": 0.5248,
+      "step": 392
+    },
+    {
+      "epoch": 2.01216,
+      "grad_norm": 0.4323654857095571,
+      "learning_rate": 2.3541603067540063e-05,
+      "loss": 0.5389,
+      "step": 393
+    },
+    {
+      "epoch": 2.01728,
+      "grad_norm": 0.4365874888455653,
+      "learning_rate": 2.332415380364052e-05,
+      "loss": 0.5172,
+      "step": 394
+    },
+    {
+      "epoch": 2.0224,
+      "grad_norm": 0.38461821215764164,
+      "learning_rate": 2.3107299399850822e-05,
+      "loss": 0.5262,
+      "step": 395
+    },
+    {
+      "epoch": 2.02752,
+      "grad_norm": 0.47903986310634056,
+      "learning_rate": 2.2891047591792532e-05,
+      "loss": 0.5212,
+      "step": 396
+    },
+    {
+      "epoch": 2.03264,
+      "grad_norm": 0.35779675128884125,
+      "learning_rate": 2.2675406093591316e-05,
+      "loss": 0.5293,
+      "step": 397
+    },
+    {
+      "epoch": 2.03776,
+      "grad_norm": 0.4808543092900177,
+      "learning_rate": 2.2460382597602016e-05,
+      "loss": 0.5352,
+      "step": 398
+    },
+    {
+      "epoch": 2.04288,
+      "grad_norm": 0.3753156744242789,
+      "learning_rate": 2.2245984774134063e-05,
+      "loss": 0.5214,
+      "step": 399
+    },
+    {
+      "epoch": 2.048,
+      "grad_norm": 0.39281547945021766,
+      "learning_rate": 2.2032220271177876e-05,
+      "loss": 0.5287,
+      "step": 400
+    },
+    {
+      "epoch": 2.05312,
+      "grad_norm": 0.40725978791424594,
+      "learning_rate": 2.1819096714132206e-05,
+      "loss": 0.526,
+      "step": 401
+    },
+    {
+      "epoch": 2.05824,
+      "grad_norm": 0.36597234139560436,
+      "learning_rate": 2.1606621705531934e-05,
+      "loss": 0.5073,
+      "step": 402
+    },
+    {
+      "epoch": 2.06336,
+      "grad_norm": 0.4434308159144889,
+      "learning_rate": 2.139480282477691e-05,
+      "loss": 0.5259,
+      "step": 403
+    },
+    {
+      "epoch": 2.06848,
+      "grad_norm": 0.36571996113644073,
+      "learning_rate": 2.1183647627861736e-05,
+      "loss": 0.5375,
+      "step": 404
+    },
+    {
+      "epoch": 2.0736,
+      "grad_norm": 0.5190659641286463,
+      "learning_rate": 2.097316364710599e-05,
+      "loss": 0.5204,
+      "step": 405
+    },
+    {
+      "epoch": 2.07872,
+      "grad_norm": 0.34838410996336344,
+      "learning_rate": 2.0763358390885705e-05,
+      "loss": 0.5183,
+      "step": 406
+    },
+    {
+      "epoch": 2.08384,
+      "grad_norm": 0.464065008178319,
+      "learning_rate": 2.0554239343365524e-05,
+      "loss": 0.5181,
+      "step": 407
+    },
+    {
+      "epoch": 2.08896,
+      "grad_norm": 0.3876057469692389,
+      "learning_rate": 2.0345813964231604e-05,
+      "loss": 0.5283,
+      "step": 408
+    },
+    {
+      "epoch": 2.09408,
+      "grad_norm": 0.38233945304862516,
+      "learning_rate": 2.013808968842561e-05,
+      "loss": 0.5253,
+      "step": 409
+    },
+    {
+      "epoch": 2.0992,
+      "grad_norm": 0.38531045185458873,
+      "learning_rate": 1.9931073925879544e-05,
+      "loss": 0.5286,
+      "step": 410
+    },
+    {
+      "epoch": 2.10432,
+      "grad_norm": 0.3441646148982092,
+      "learning_rate": 1.9724774061251253e-05,
+      "loss": 0.5289,
+      "step": 411
+    },
+    {
+      "epoch": 2.10944,
+      "grad_norm": 0.3059332693194638,
+      "learning_rate": 1.9519197453661166e-05,
+      "loss": 0.5268,
+      "step": 412
+    },
+    {
+      "epoch": 2.11456,
+      "grad_norm": 0.34251732288777326,
+      "learning_rate": 1.9314351436429703e-05,
+      "loss": 0.535,
+      "step": 413
+    },
+    {
+      "epoch": 2.11968,
+      "grad_norm": 0.27043155273405406,
+      "learning_rate": 1.9110243316815672e-05,
+      "loss": 0.519,
+      "step": 414
+    },
+    {
+      "epoch": 2.1248,
+      "grad_norm": 0.3287810309710835,
+      "learning_rate": 1.8906880375755686e-05,
+      "loss": 0.5123,
+      "step": 415
+    },
+    {
+      "epoch": 2.12992,
+      "grad_norm": 0.31017397419120324,
+      "learning_rate": 1.87042698676043e-05,
+      "loss": 0.5271,
+      "step": 416
+    },
+    {
+      "epoch": 2.13504,
+      "grad_norm": 0.27534025617214347,
+      "learning_rate": 1.8502419019875357e-05,
+      "loss": 0.5148,
+      "step": 417
+    },
+    {
+      "epoch": 2.14016,
+      "grad_norm": 0.34713006182861084,
+      "learning_rate": 1.8301335032984157e-05,
+      "loss": 0.5263,
+      "step": 418
+    },
+    {
+      "epoch": 2.14528,
+      "grad_norm": 0.2944738685945991,
+      "learning_rate": 1.8101025079990485e-05,
+      "loss": 0.5082,
+      "step": 419
+    },
+    {
+      "epoch": 2.1504,
+      "grad_norm": 0.2795225009877447,
+      "learning_rate": 1.7901496306342886e-05,
+      "loss": 0.5162,
+      "step": 420
+    },
+    {
+      "epoch": 2.15552,
+      "grad_norm": 0.2792843641392462,
+      "learning_rate": 1.7702755829623696e-05,
+      "loss": 0.516,
+      "step": 421
+    },
+    {
+      "epoch": 2.16064,
+      "grad_norm": 0.31863845049393985,
+      "learning_rate": 1.750481073929514e-05,
+      "loss": 0.5294,
+      "step": 422
+    },
+    {
+      "epoch": 2.16576,
+      "grad_norm": 0.2926787760079869,
+      "learning_rate": 1.730766809644646e-05,
+      "loss": 0.5436,
+      "step": 423
+    },
+    {
+      "epoch": 2.17088,
+      "grad_norm": 0.26756179726965906,
+      "learning_rate": 1.711133493354203e-05,
+      "loss": 0.5391,
+      "step": 424
+    },
+    {
+      "epoch": 2.176,
+      "grad_norm": 0.26829916228856365,
+      "learning_rate": 1.6915818254170485e-05,
+      "loss": 0.5293,
+      "step": 425
+    },
+    {
+      "epoch": 2.18112,
+      "grad_norm": 0.2916809848264421,
+      "learning_rate": 1.672112503279488e-05,
+      "loss": 0.5239,
+      "step": 426
+    },
+    {
+      "epoch": 2.18624,
+      "grad_norm": 0.2847780874837289,
+      "learning_rate": 1.6527262214503917e-05,
+      "loss": 0.5206,
+      "step": 427
+    },
+    {
+      "epoch": 2.19136,
+      "grad_norm": 0.31205176032409004,
+      "learning_rate": 1.6334236714764215e-05,
+      "loss": 0.5219,
+      "step": 428
+    },
+    {
+      "epoch": 2.19648,
+      "grad_norm": 0.28653984101417374,
+      "learning_rate": 1.6142055419173556e-05,
+      "loss": 0.5396,
+      "step": 429
+    },
+    {
+      "epoch": 2.2016,
+      "grad_norm": 0.29293525576131924,
+      "learning_rate": 1.595072518321535e-05,
+      "loss": 0.5186,
+      "step": 430
+    },
+    {
+      "epoch": 2.20672,
+      "grad_norm": 0.32036644737076314,
+      "learning_rate": 1.576025283201401e-05,
+      "loss": 0.5293,
+      "step": 431
+    },
+    {
+      "epoch": 2.21184,
+      "grad_norm": 0.2746440977694007,
+      "learning_rate": 1.5570645160091534e-05,
+      "loss": 0.5238,
+      "step": 432
+    },
+    {
+      "epoch": 2.21696,
+      "grad_norm": 0.2763614067422842,
+      "learning_rate": 1.5381908931125112e-05,
+      "loss": 0.5315,
+      "step": 433
+    },
+    {
+      "epoch": 2.22208,
+      "grad_norm": 0.3206600626128733,
+      "learning_rate": 1.519405087770586e-05,
+      "loss": 0.5155,
+      "step": 434
+    },
+    {
+      "epoch": 2.2272,
+      "grad_norm": 0.28546354885077907,
+      "learning_rate": 1.5007077701098643e-05,
+      "loss": 0.5153,
+      "step": 435
+    },
+    {
+      "epoch": 2.23232,
+      "grad_norm": 0.32183823115025384,
+      "learning_rate": 1.4820996071003042e-05,
+      "loss": 0.5244,
+      "step": 436
+    },
+    {
+      "epoch": 2.23744,
+      "grad_norm": 0.2833366198667468,
+      "learning_rate": 1.4635812625315428e-05,
+      "loss": 0.5291,
+      "step": 437
+    },
+    {
+      "epoch": 2.24256,
+      "grad_norm": 0.35259963599910704,
+      "learning_rate": 1.445153396989218e-05,
+      "loss": 0.5334,
+      "step": 438
+    },
+    {
+      "epoch": 2.24768,
+      "grad_norm": 0.31168802605945267,
+      "learning_rate": 1.4268166678314029e-05,
+      "loss": 0.522,
+      "step": 439
+    },
+    {
+      "epoch": 2.2528,
+      "grad_norm": 0.32274340867941653,
+      "learning_rate": 1.4085717291651534e-05,
+      "loss": 0.5291,
+      "step": 440
+    },
+    {
+      "epoch": 2.25792,
+      "grad_norm": 0.2948948221905903,
+      "learning_rate": 1.390419231823187e-05,
+      "loss": 0.5394,
+      "step": 441
+    },
+    {
+      "epoch": 2.26304,
+      "grad_norm": 0.2743662929170965,
+      "learning_rate": 1.3723598233406525e-05,
+      "loss": 0.5271,
+      "step": 442
+    },
+    {
+      "epoch": 2.26816,
+      "grad_norm": 0.2556644467949261,
+      "learning_rate": 1.3543941479320344e-05,
+      "loss": 0.5253,
+      "step": 443
+    },
+    {
+      "epoch": 2.27328,
+      "grad_norm": 0.2611595534203661,
+      "learning_rate": 1.3365228464681814e-05,
+      "loss": 0.5262,
+      "step": 444
+    },
+    {
+      "epoch": 2.2784,
+      "grad_norm": 0.26247038877870327,
+      "learning_rate": 1.3187465564534359e-05,
+      "loss": 0.5257,
+      "step": 445
+    },
+    {
+      "epoch": 2.28352,
+      "grad_norm": 0.2536877324922943,
+      "learning_rate": 1.30106591200289e-05,
+      "loss": 0.5174,
+      "step": 446
+    },
+    {
+      "epoch": 2.28864,
+      "grad_norm": 0.2691257976797504,
+      "learning_rate": 1.2834815438197827e-05,
+      "loss": 0.5421,
+      "step": 447
+    },
+    {
+      "epoch": 2.29376,
+      "grad_norm": 0.21048332739110232,
+      "learning_rate": 1.2659940791729773e-05,
+      "loss": 0.5183,
+      "step": 448
+    },
+    {
+      "epoch": 2.29888,
+      "grad_norm": 0.26799152242066077,
+      "learning_rate": 1.248604141874608e-05,
+      "loss": 0.5347,
+      "step": 449
+    },
+    {
+      "epoch": 2.304,
+      "grad_norm": 0.23183888170912065,
+      "learning_rate": 1.2313123522578141e-05,
+      "loss": 0.528,
+      "step": 450
+    },
+    {
+      "epoch": 2.30912,
+      "grad_norm": 0.23153879138735126,
+      "learning_rate": 1.2141193271546104e-05,
+      "loss": 0.5402,
+      "step": 451
+    },
+    {
+      "epoch": 2.31424,
+      "grad_norm": 0.2365803046413853,
+      "learning_rate": 1.1970256798738946e-05,
+      "loss": 0.524,
+      "step": 452
+    },
+    {
+      "epoch": 2.31936,
+      "grad_norm": 0.23790371050570536,
+      "learning_rate": 1.1800320201795609e-05,
+      "loss": 0.5236,
+      "step": 453
+    },
+    {
+      "epoch": 2.32448,
+      "grad_norm": 0.21840868912845707,
+      "learning_rate": 1.1631389542687455e-05,
+      "loss": 0.5158,
+      "step": 454
+    },
+    {
+      "epoch": 2.3296,
+      "grad_norm": 0.22811000289018,
+      "learning_rate": 1.1463470847502154e-05,
+      "loss": 0.5491,
+      "step": 455
+    },
+    {
+      "epoch": 2.33472,
+      "grad_norm": 0.2441771477812667,
+      "learning_rate": 1.1296570106228568e-05,
+      "loss": 0.5119,
+      "step": 456
+    },
+    {
+      "epoch": 2.33984,
+      "grad_norm": 0.2504770922206427,
+      "learning_rate": 1.1130693272543174e-05,
+      "loss": 0.5259,
+      "step": 457
+    },
+    {
+      "epoch": 2.34496,
+      "grad_norm": 0.22307013304308804,
+      "learning_rate": 1.0965846263597704e-05,
+      "loss": 0.5229,
+      "step": 458
+    },
+    {
+      "epoch": 2.35008,
+      "grad_norm": 0.20777825026632868,
+      "learning_rate": 1.0802034959807934e-05,
+      "loss": 0.5327,
+      "step": 459
+    },
+    {
+      "epoch": 2.3552,
+      "grad_norm": 0.2496892379371564,
+      "learning_rate": 1.063926520464407e-05,
+      "loss": 0.5173,
+      "step": 460
+    },
+    {
+      "epoch": 2.3603199999999998,
+      "grad_norm": 0.2710443329473512,
+      "learning_rate": 1.047754280442225e-05,
+      "loss": 0.5093,
+      "step": 461
+    },
+    {
+      "epoch": 2.36544,
+      "grad_norm": 0.2221839293934009,
+      "learning_rate": 1.0316873528097333e-05,
+      "loss": 0.5339,
+      "step": 462
+    },
+    {
+      "epoch": 2.3705600000000002,
+      "grad_norm": 0.21317614214755715,
+      "learning_rate": 1.0157263107057291e-05,
+      "loss": 0.5284,
+      "step": 463
+    },
+    {
+      "epoch": 2.37568,
+      "grad_norm": 0.22721711775906023,
+      "learning_rate": 9.998717234918557e-06,
+      "loss": 0.5351,
+      "step": 464
+    },
+    {
+      "epoch": 2.3808,
+      "grad_norm": 0.22893893378395325,
+      "learning_rate": 9.84124156732305e-06,
+      "loss": 0.5173,
+      "step": 465
+    },
+    {
+      "epoch": 2.38592,
+      "grad_norm": 0.2312282069514675,
+      "learning_rate": 9.684841721736449e-06,
+      "loss": 0.5099,
+      "step": 466
+    },
+    {
+      "epoch": 2.39104,
+      "grad_norm": 0.23235918033484193,
+      "learning_rate": 9.529523277247672e-06,
+      "loss": 0.5353,
+      "step": 467
+    },
+    {
+      "epoch": 2.39616,
+      "grad_norm": 0.2122801075559491,
+      "learning_rate": 9.37529177436998e-06,
+      "loss": 0.524,
+      "step": 468
+    },
+    {
+      "epoch": 2.40128,
+      "grad_norm": 0.2281610275331807,
+      "learning_rate": 9.222152714843324e-06,
+      "loss": 0.5198,
+      "step": 469
+    },
+    {
+      "epoch": 2.4064,
+      "grad_norm": 0.2276209528821846,
+      "learning_rate": 9.070111561437994e-06,
+      "loss": 0.5125,
+      "step": 470
+    },
+    {
+      "epoch": 2.41152,
+      "grad_norm": 0.25863996249108634,
+      "learning_rate": 8.919173737759878e-06,
+      "loss": 0.5278,
+      "step": 471
+    },
+    {
+      "epoch": 2.41664,
+      "grad_norm": 0.24519777588872835,
+      "learning_rate": 8.769344628056893e-06,
+      "loss": 0.548,
+      "step": 472
+    },
+    {
+      "epoch": 2.42176,
+      "grad_norm": 0.22910426740880335,
+      "learning_rate": 8.62062957702698e-06,
+      "loss": 0.5186,
+      "step": 473
+    },
+    {
+      "epoch": 2.42688,
+      "grad_norm": 0.2172444686293662,
+      "learning_rate": 8.473033889627404e-06,
+      "loss": 0.5316,
+      "step": 474
+    },
+    {
+      "epoch": 2.432,
+      "grad_norm": 0.21789541593913744,
+      "learning_rate": 8.326562830885559e-06,
+      "loss": 0.4964,
+      "step": 475
+    },
+    {
+      "epoch": 2.43712,
+      "grad_norm": 0.26419477579272155,
+      "learning_rate": 8.181221625711102e-06,
+      "loss": 0.5285,
+      "step": 476
+    },
+    {
+      "epoch": 2.44224,
+      "grad_norm": 0.22975957761205745,
+      "learning_rate": 8.03701545870963e-06,
+      "loss": 0.5323,
+      "step": 477
+    },
+    {
+      "epoch": 2.4473599999999998,
+      "grad_norm": 0.2192599105025437,
+      "learning_rate": 7.893949473997682e-06,
+      "loss": 0.513,
+      "step": 478
+    },
+    {
+      "epoch": 2.45248,
+      "grad_norm": 0.21383791397200727,
+      "learning_rate": 7.752028775019264e-06,
+      "loss": 0.5287,
+      "step": 479
+    },
+    {
+      "epoch": 2.4576000000000002,
+      "grad_norm": 0.23011437791676428,
+      "learning_rate": 7.61125842436381e-06,
+      "loss": 0.5158,
+      "step": 480
+    },
+    {
+      "epoch": 2.46272,
+      "grad_norm": 0.23089368890405654,
+      "learning_rate": 7.471643443585561e-06,
+      "loss": 0.5286,
+      "step": 481
+    },
+    {
+      "epoch": 2.46784,
+      "grad_norm": 0.21550660720139894,
+      "learning_rate": 7.33318881302445e-06,
+      "loss": 0.5315,
+      "step": 482
+    },
+    {
+      "epoch": 2.47296,
+      "grad_norm": 0.21786834523919976,
+      "learning_rate": 7.1958994716284556e-06,
+      "loss": 0.5344,
+      "step": 483
+    },
+    {
+      "epoch": 2.47808,
+      "grad_norm": 0.23493881832666266,
+      "learning_rate": 7.059780316777396e-06,
+      "loss": 0.5298,
+      "step": 484
+    },
+    {
+      "epoch": 2.4832,
+      "grad_norm": 0.23179568273154502,
+      "learning_rate": 6.9248362041082514e-06,
+      "loss": 0.5395,
+      "step": 485
+    },
+    {
+      "epoch": 2.48832,
+      "grad_norm": 0.1997326200659906,
+      "learning_rate": 6.791071947341939e-06,
+      "loss": 0.5344,
+      "step": 486
+    },
+    {
+      "epoch": 2.49344,
+      "grad_norm": 0.18863063393495233,
+      "learning_rate": 6.658492318111611e-06,
+      "loss": 0.528,
+      "step": 487
+    },
+    {
+      "epoch": 2.49856,
+      "grad_norm": 0.2042454987848812,
+      "learning_rate": 6.527102045792424e-06,
+      "loss": 0.5057,
+      "step": 488
+    },
+    {
+      "epoch": 2.50368,
+      "grad_norm": 0.21633481485073822,
+      "learning_rate": 6.39690581733285e-06,
+      "loss": 0.5231,
+      "step": 489
+    },
+    {
+      "epoch": 2.5088,
+      "grad_norm": 0.20671940715964057,
+      "learning_rate": 6.267908277087489e-06,
+      "loss": 0.519,
+      "step": 490
+    },
+    {
+      "epoch": 2.51392,
+      "grad_norm": 0.20567854945985983,
+      "learning_rate": 6.140114026651338e-06,
+      "loss": 0.5246,
+      "step": 491
+    },
+    {
+      "epoch": 2.51904,
+      "grad_norm": 0.21668723148852223,
+      "learning_rate": 6.01352762469575e-06,
+      "loss": 0.5234,
+      "step": 492
+    },
+    {
+      "epoch": 2.52416,
+      "grad_norm": 0.1955709273332618,
+      "learning_rate": 5.888153586805723e-06,
+      "loss": 0.4833,
+      "step": 493
+    },
+    {
+      "epoch": 2.52928,
+      "grad_norm": 0.19018764122168655,
+      "learning_rate": 5.763996385318838e-06,
+      "loss": 0.5241,
+      "step": 494
+    },
+    {
+      "epoch": 2.5343999999999998,
+      "grad_norm": 0.21895623789510293,
+      "learning_rate": 5.641060449165774e-06,
+      "loss": 0.5022,
+      "step": 495
+    },
+    {
+      "epoch": 2.53952,
+      "grad_norm": 0.23471379305483125,
+      "learning_rate": 5.5193501637122605e-06,
+      "loss": 0.527,
+      "step": 496
+    },
+    {
+      "epoch": 2.5446400000000002,
+      "grad_norm": 0.2119274689582658,
+      "learning_rate": 5.39886987060267e-06,
+      "loss": 0.5376,
+      "step": 497
+    },
+    {
+      "epoch": 2.54976,
+      "grad_norm": 0.19216483721783073,
+      "learning_rate": 5.279623867605144e-06,
+      "loss": 0.5221,
+      "step": 498
+    },
+    {
+      "epoch": 2.55488,
+      "grad_norm": 0.2154169515261133,
+      "learning_rate": 5.161616408458239e-06,
+      "loss": 0.5188,
+      "step": 499
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.20441281420775267,
+      "learning_rate": 5.044851702719289e-06,
+      "loss": 0.5303,
+      "step": 500
+    },
+    {
+      "epoch": 2.56512,
+      "grad_norm": 0.2107906688631621,
+      "learning_rate": 4.929333915614147e-06,
+      "loss": 0.511,
+      "step": 501
+    },
+    {
+      "epoch": 2.57024,
+      "grad_norm": 0.21514247844377765,
+      "learning_rate": 4.815067167888603e-06,
+      "loss": 0.5245,
+      "step": 502
+    },
+    {
+      "epoch": 2.57536,
+      "grad_norm": 0.2061450148487434,
+      "learning_rate": 4.702055535661481e-06,
+      "loss": 0.5079,
+      "step": 503
+    },
+    {
+      "epoch": 2.58048,
+      "grad_norm": 0.19909839216020486,
+      "learning_rate": 4.590303050279144e-06,
+      "loss": 0.5192,
+      "step": 504
+    },
+    {
+      "epoch": 2.5856,
+      "grad_norm": 0.20208792009677243,
+      "learning_rate": 4.479813698171702e-06,
+      "loss": 0.5102,
+      "step": 505
+    },
+    {
+      "epoch": 2.59072,
+      "grad_norm": 0.2120494653558063,
+      "learning_rate": 4.370591420710879e-06,
+      "loss": 0.5276,
+      "step": 506
+    },
+    {
+      "epoch": 2.59584,
+      "grad_norm": 0.19182746478809978,
+      "learning_rate": 4.262640114069303e-06,
+      "loss": 0.5297,
+      "step": 507
+    },
+    {
+      "epoch": 2.60096,
+      "grad_norm": 0.2049595817249684,
+      "learning_rate": 4.1559636290816165e-06,
+      "loss": 0.5089,
+      "step": 508
+    },
+    {
+      "epoch": 2.60608,
+      "grad_norm": 0.20334632758136215,
+      "learning_rate": 4.050565771107086e-06,
+      "loss": 0.5341,
+      "step": 509
+    },
+    {
+      "epoch": 2.6112,
+      "grad_norm": 0.21626882759385432,
+      "learning_rate": 3.946450299893813e-06,
+      "loss": 0.5356,
+      "step": 510
+    },
+    {
+      "epoch": 2.61632,
+      "grad_norm": 0.22272362239503654,
+      "learning_rate": 3.843620929444667e-06,
+      "loss": 0.5264,
+      "step": 511
+    },
+    {
+      "epoch": 2.6214399999999998,
+      "grad_norm": 0.20142676114788888,
+      "learning_rate": 3.74208132788481e-06,
+      "loss": 0.5277,
+      "step": 512
+    },
+    {
+      "epoch": 2.62656,
+      "grad_norm": 0.19564873650299325,
+      "learning_rate": 3.641835117330761e-06,
+      "loss": 0.532,
+      "step": 513
+    },
+    {
+      "epoch": 2.6316800000000002,
+      "grad_norm": 0.21279829206795126,
+      "learning_rate": 3.542885873761308e-06,
+      "loss": 0.5163,
+      "step": 514
+    },
+    {
+      "epoch": 2.6368,
+      "grad_norm": 0.20724605046419126,
+      "learning_rate": 3.4452371268898444e-06,
+      "loss": 0.5074,
+      "step": 515
+    },
+    {
+      "epoch": 2.64192,
+      "grad_norm": 0.2140560895573128,
+      "learning_rate": 3.3488923600385028e-06,
+      "loss": 0.5331,
+      "step": 516
+    },
+    {
+      "epoch": 2.64704,
+      "grad_norm": 0.19348656997795682,
+      "learning_rate": 3.2538550100139353e-06,
+      "loss": 0.5273,
+      "step": 517
+    },
+    {
+      "epoch": 2.65216,
+      "grad_norm": 0.1959351210860576,
+      "learning_rate": 3.160128466984609e-06,
+      "loss": 0.5189,
+      "step": 518
+    },
+    {
+      "epoch": 2.65728,
+      "grad_norm": 0.2043908489285227,
+      "learning_rate": 3.0677160743599788e-06,
+      "loss": 0.5349,
+      "step": 519
+    },
+    {
+      "epoch": 2.6624,
+      "grad_norm": 0.20288115925706235,
+      "learning_rate": 2.9766211286711868e-06,
+      "loss": 0.5248,
+      "step": 520
+    },
+    {
+      "epoch": 2.66752,
+      "grad_norm": 0.20731847438641798,
+      "learning_rate": 2.8868468794534243e-06,
+      "loss": 0.5181,
+      "step": 521
+    },
+    {
+      "epoch": 2.67264,
+      "grad_norm": 0.19303823605826137,
+      "learning_rate": 2.7983965291300765e-06,
+      "loss": 0.5309,
+      "step": 522
+    },
+    {
+      "epoch": 2.67776,
+      "grad_norm": 0.2049514957880367,
+      "learning_rate": 2.7112732328984594e-06,
+      "loss": 0.5301,
+      "step": 523
+    },
+    {
+      "epoch": 2.68288,
+      "grad_norm": 0.19908648001763588,
+      "learning_rate": 2.6254800986172635e-06,
+      "loss": 0.5278,
+      "step": 524
+    },
+    {
+      "epoch": 2.6879999999999997,
+      "grad_norm": 0.20892226869833805,
+      "learning_rate": 2.5410201866957042e-06,
+      "loss": 0.5236,
+      "step": 525
+    },
+    {
+      "epoch": 2.69312,
+      "grad_norm": 0.18255526452984,
+      "learning_rate": 2.457896509984332e-06,
+      "loss": 0.525,
+      "step": 526
+    },
+    {
+      "epoch": 2.69824,
+      "grad_norm": 0.19075543504463546,
+      "learning_rate": 2.376112033667579e-06,
+      "loss": 0.5139,
+      "step": 527
+    },
+    {
+      "epoch": 2.70336,
+      "grad_norm": 0.19978943687563103,
+      "learning_rate": 2.29566967515797e-06,
+      "loss": 0.523,
+      "step": 528
+    },
+    {
+      "epoch": 2.7084799999999998,
+      "grad_norm": 0.17850021044317485,
+      "learning_rate": 2.216572303992055e-06,
+      "loss": 0.5327,
+      "step": 529
+    },
+    {
+      "epoch": 2.7136,
+      "grad_norm": 0.1854277104912946,
+      "learning_rate": 2.1388227417280527e-06,
+      "loss": 0.5233,
+      "step": 530
+    },
+    {
+      "epoch": 2.7187200000000002,
+      "grad_norm": 0.18348286998112376,
+      "learning_rate": 2.0624237618452047e-06,
+      "loss": 0.5396,
+      "step": 531
+    },
+    {
+      "epoch": 2.72384,
+      "grad_norm": 0.1883434076908113,
+      "learning_rate": 1.98737808964482e-06,
+      "loss": 0.5197,
+      "step": 532
+    },
+    {
+      "epoch": 2.72896,
+      "grad_norm": 0.1841837424394637,
+      "learning_rate": 1.913688402153082e-06,
+      "loss": 0.5297,
+      "step": 533
+    },
+    {
+      "epoch": 2.73408,
+      "grad_norm": 0.17622787049259134,
+      "learning_rate": 1.8413573280255326e-06,
+      "loss": 0.5285,
+      "step": 534
+    },
+    {
+      "epoch": 2.7392,
+      "grad_norm": 0.17806753215393287,
+      "learning_rate": 1.7703874474533167e-06,
+      "loss": 0.5278,
+      "step": 535
+    },
+    {
+      "epoch": 2.74432,
+      "grad_norm": 0.1958676559056442,
+      "learning_rate": 1.7007812920711408e-06,
+      "loss": 0.5044,
+      "step": 536
+    },
+    {
+      "epoch": 2.74944,
+      "grad_norm": 0.16605984676027447,
+      "learning_rate": 1.6325413448669625e-06,
+      "loss": 0.514,
+      "step": 537
+    },
+    {
+      "epoch": 2.75456,
+      "grad_norm": 0.18195390968020672,
+      "learning_rate": 1.5656700400934121e-06,
+      "loss": 0.5445,
+      "step": 538
+    },
+    {
+      "epoch": 2.75968,
+      "grad_norm": 0.16824479934145423,
+      "learning_rate": 1.500169763180961e-06,
+      "loss": 0.5212,
+      "step": 539
+    },
+    {
+      "epoch": 2.7648,
+      "grad_norm": 0.1819730086116406,
+      "learning_rate": 1.4360428506528323e-06,
+      "loss": 0.5232,
+      "step": 540
+    },
+    {
+      "epoch": 2.76992,
+      "grad_norm": 0.17926267467820442,
+      "learning_rate": 1.373291590041661e-06,
+      "loss": 0.5273,
+      "step": 541
+    },
+    {
+      "epoch": 2.7750399999999997,
+      "grad_norm": 0.21059680680377105,
+      "learning_rate": 1.3119182198078596e-06,
+      "loss": 0.5131,
+      "step": 542
+    },
+    {
+      "epoch": 2.78016,
+      "grad_norm": 0.18407921454967086,
+      "learning_rate": 1.2519249292598112e-06,
+      "loss": 0.4969,
+      "step": 543
+    },
+    {
+      "epoch": 2.78528,
+      "grad_norm": 0.18395693454820505,
+      "learning_rate": 1.1933138584757508e-06,
+      "loss": 0.5188,
+      "step": 544
+    },
+    {
+      "epoch": 2.7904,
+      "grad_norm": 0.19315285688390138,
+      "learning_rate": 1.1360870982274118e-06,
+      "loss": 0.5186,
+      "step": 545
+    },
+    {
+      "epoch": 2.79552,
+      "grad_norm": 0.1887790134903703,
+      "learning_rate": 1.080246689905473e-06,
+      "loss": 0.5271,
+      "step": 546
+    },
+    {
+      "epoch": 2.80064,
+      "grad_norm": 0.17959548964956856,
+      "learning_rate": 1.0257946254467145e-06,
+      "loss": 0.5243,
+      "step": 547
+    },
+    {
+      "epoch": 2.8057600000000003,
+      "grad_norm": 0.1817474574269339,
+      "learning_rate": 9.727328472629716e-07,
+      "loss": 0.5287,
+      "step": 548
+    },
+    {
+      "epoch": 2.81088,
+      "grad_norm": 0.17642219828487612,
+      "learning_rate": 9.21063248171854e-07,
+      "loss": 0.526,
+      "step": 549
+    },
+    {
+      "epoch": 2.816,
+      "grad_norm": 0.1904246025654594,
+      "learning_rate": 8.707876713291941e-07,
+      "loss": 0.5135,
+      "step": 550
+    },
+    {
+      "epoch": 2.82112,
+      "grad_norm": 0.17957762634349514,
+      "learning_rate": 8.219079101633443e-07,
+      "loss": 0.5215,
+      "step": 551
+    },
+    {
+      "epoch": 2.82624,
+      "grad_norm": 0.16263065954048853,
+      "learning_rate": 7.744257083111662e-07,
+      "loss": 0.5193,
+      "step": 552
+    },
+    {
+      "epoch": 2.83136,
+      "grad_norm": 0.17484720518211333,
+      "learning_rate": 7.283427595558224e-07,
+      "loss": 0.5296,
+      "step": 553
+    },
+    {
+      "epoch": 2.83648,
+      "grad_norm": 0.1867914151335944,
+      "learning_rate": 6.836607077664115e-07,
+      "loss": 0.5383,
+      "step": 554
+    },
+    {
+      "epoch": 2.8416,
+      "grad_norm": 0.18028174989209536,
+      "learning_rate": 6.403811468392685e-07,
+      "loss": 0.5453,
+      "step": 555
+    },
+    {
+      "epoch": 2.84672,
+      "grad_norm": 0.1730198700904891,
+      "learning_rate": 5.9850562064113e-07,
+      "loss": 0.5272,
+      "step": 556
+    },
+    {
+      "epoch": 2.85184,
+      "grad_norm": 0.17377782938839437,
+      "learning_rate": 5.580356229540851e-07,
+      "loss": 0.5298,
+      "step": 557
+    },
+    {
+      "epoch": 2.85696,
+      "grad_norm": 0.16971911427369651,
+      "learning_rate": 5.189725974222448e-07,
+      "loss": 0.5239,
+      "step": 558
+    },
+    {
+      "epoch": 2.8620799999999997,
+      "grad_norm": 0.1853393873016306,
+      "learning_rate": 4.81317937500263e-07,
+      "loss": 0.5358,
+      "step": 559
+    },
+    {
+      "epoch": 2.8672,
+      "grad_norm": 0.16321076208370755,
+      "learning_rate": 4.450729864036607e-07,
+      "loss": 0.511,
+      "step": 560
+    },
+    {
+      "epoch": 2.87232,
+      "grad_norm": 0.17617659485628395,
+      "learning_rate": 4.1023903706084666e-07,
+      "loss": 0.5251,
+      "step": 561
+    },
+    {
+      "epoch": 2.87744,
+      "grad_norm": 0.17229617450371348,
+      "learning_rate": 3.768173320670654e-07,
+      "loss": 0.516,
+      "step": 562
+    },
+    {
+      "epoch": 2.88256,
+      "grad_norm": 0.18627902011447772,
+      "learning_rate": 3.448090636400192e-07,
+      "loss": 0.527,
+      "step": 563
+    },
+    {
+      "epoch": 2.88768,
+      "grad_norm": 0.1692917592955478,
+      "learning_rate": 3.1421537357735567e-07,
+      "loss": 0.5203,
+      "step": 564
+    },
+    {
+      "epoch": 2.8928000000000003,
+      "grad_norm": 0.1742890382438682,
+      "learning_rate": 2.850373532159578e-07,
+      "loss": 0.5099,
+      "step": 565
+    },
+    {
+      "epoch": 2.89792,
+      "grad_norm": 0.18401122824627672,
+      "learning_rate": 2.5727604339297996e-07,
+      "loss": 0.5303,
+      "step": 566
+    },
+    {
+      "epoch": 2.90304,
+      "grad_norm": 0.17215703579697378,
+      "learning_rate": 2.309324344087349e-07,
+      "loss": 0.5248,
+      "step": 567
+    },
+    {
+      "epoch": 2.90816,
+      "grad_norm": 0.1744660017717841,
+      "learning_rate": 2.060074659913891e-07,
+      "loss": 0.5067,
+      "step": 568
+    },
+    {
+      "epoch": 2.91328,
+      "grad_norm": 0.1753887233423339,
+      "learning_rate": 1.8250202726339815e-07,
+      "loss": 0.5142,
+      "step": 569
+    },
+    {
+      "epoch": 2.9184,
+      "grad_norm": 0.17222558777930064,
+      "learning_rate": 1.6041695670981684e-07,
+      "loss": 0.5271,
+      "step": 570
+    },
+    {
+      "epoch": 2.92352,
+      "grad_norm": 0.1749876224664464,
+      "learning_rate": 1.397530421483939e-07,
+      "loss": 0.5215,
+      "step": 571
+    },
+    {
+      "epoch": 2.92864,
+      "grad_norm": 0.17369587300943137,
+      "learning_rate": 1.2051102070144816e-07,
+      "loss": 0.5248,
+      "step": 572
+    },
+    {
+      "epoch": 2.93376,
+      "grad_norm": 0.1720447918581037,
+      "learning_rate": 1.0269157876959146e-07,
+      "loss": 0.5198,
+      "step": 573
+    },
+    {
+      "epoch": 2.93888,
+      "grad_norm": 0.18006932981535365,
+      "learning_rate": 8.62953520072285e-08,
+      "loss": 0.5081,
+      "step": 574
+    },
+    {
+      "epoch": 2.944,
+      "grad_norm": 0.16965260368509588,
+      "learning_rate": 7.132292529990814e-08,
+      "loss": 0.5227,
+      "step": 575
+    },
+    {
+      "epoch": 2.9491199999999997,
+      "grad_norm": 0.17833147445276212,
+      "learning_rate": 5.7774832743415776e-08,
+      "loss": 0.5269,
+      "step": 576
+    },
+    {
+      "epoch": 2.95424,
+      "grad_norm": 0.17095515499148287,
+      "learning_rate": 4.565155762477069e-08,
+      "loss": 0.5416,
+      "step": 577
+    },
+    {
+      "epoch": 2.95936,
+      "grad_norm": 0.17471514998398122,
+      "learning_rate": 3.4953532404942146e-08,
+      "loss": 0.5026,
+      "step": 578
+    },
+    {
+      "epoch": 2.96448,
+      "grad_norm": 0.1721020510585504,
+      "learning_rate": 2.568113870346167e-08,
+      "loss": 0.5092,
+      "step": 579
+    },
+    {
+      "epoch": 2.9696,
+      "grad_norm": 0.18037480336003978,
+      "learning_rate": 1.783470728477621e-08,
+      "loss": 0.514,
+      "step": 580
+    },
+    {
+      "epoch": 2.97472,
+      "grad_norm": 0.17100517198442003,
+      "learning_rate": 1.1414518046470868e-08,
+      "loss": 0.5405,
+      "step": 581
+    },
+    {
+      "epoch": 2.9798400000000003,
+      "grad_norm": 0.1702692183309079,
+      "learning_rate": 6.4208000092769174e-09,
+      "loss": 0.5228,
+      "step": 582
+    },
+    {
+      "epoch": 2.98496,
+      "grad_norm": 0.17252007004331688,
+      "learning_rate": 2.8537313088961015e-09,
+      "loss": 0.5295,
+      "step": 583
+    },
+    {
+      "epoch": 2.99008,
+      "grad_norm": 0.18840800940002203,
+      "learning_rate": 7.134391896679305e-10,
+      "loss": 0.5159,
+      "step": 584
+    },
+    {
+      "epoch": 2.9952,
+      "grad_norm": 0.1870617408251541,
+      "learning_rate": 0.0,
+      "loss": 0.4924,
+      "step": 585
+    },
+    {
+      "epoch": 2.9952,
+      "step": 585,
+      "total_flos": 4.2089030343416873e+18,
+      "train_loss": 0.0,
+      "train_runtime": 5.9478,
+      "train_samples_per_second": 50438.911,
+      "train_steps_per_second": 98.356
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 585,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.2089030343416873e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}