diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6727 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.980443285528032,
+  "eval_steps": 500,
+  "global_step": 955,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005215123859191656,
+      "grad_norm": 8.640251838986332,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 1.3109,
+      "step": 1
+    },
+    {
+      "epoch": 0.010430247718383311,
+      "grad_norm": 8.50165895125568,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.2827,
+      "step": 2
+    },
+    {
+      "epoch": 0.01564537157757497,
+      "grad_norm": 8.535323811460232,
+      "learning_rate": 2.5e-06,
+      "loss": 1.2971,
+      "step": 3
+    },
+    {
+      "epoch": 0.020860495436766623,
+      "grad_norm": 8.005773755772996,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.2835,
+      "step": 4
+    },
+    {
+      "epoch": 0.02607561929595828,
+      "grad_norm": 6.542532290299446,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.2424,
+      "step": 5
+    },
+    {
+      "epoch": 0.03129074315514994,
+      "grad_norm": 3.513863098870053,
+      "learning_rate": 5e-06,
+      "loss": 1.1855,
+      "step": 6
+    },
+    {
+      "epoch": 0.03650586701434159,
+      "grad_norm": 2.7215053265085425,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.1771,
+      "step": 7
+    },
+    {
+      "epoch": 0.041720990873533245,
+      "grad_norm": 6.45835816206034,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.1818,
+      "step": 8
+    },
+    {
+      "epoch": 0.0469361147327249,
+      "grad_norm": 6.896643054762279,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.1927,
+      "step": 9
+    },
+    {
+      "epoch": 0.05215123859191656,
+      "grad_norm": 7.106049193713323,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.192,
+      "step": 10
+    },
+    {
+      "epoch": 0.05736636245110821,
+      "grad_norm": 5.8848429339749755,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.1358,
+      "step": 11
+    },
+    {
+      "epoch": 0.06258148631029987,
+      "grad_norm": 5.16279034442262,
+      "learning_rate": 1e-05,
+      "loss": 1.1187,
+      "step": 12
+    },
+    {
+      "epoch": 0.06779661016949153,
+      "grad_norm": 3.4057013372590914,
+      "learning_rate": 1.0833333333333334e-05,
+      "loss": 1.1044,
+      "step": 13
+    },
+    {
+      "epoch": 0.07301173402868318,
+      "grad_norm": 2.2087972161452516,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 1.0906,
+      "step": 14
+    },
+    {
+      "epoch": 0.07822685788787484,
+      "grad_norm": 2.1528478641437006,
+      "learning_rate": 1.25e-05,
+      "loss": 1.0749,
+      "step": 15
+    },
+    {
+      "epoch": 0.08344198174706649,
+      "grad_norm": 2.0837468590704984,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.041,
+      "step": 16
+    },
+    {
+      "epoch": 0.08865710560625815,
+      "grad_norm": 2.3435856245272064,
+      "learning_rate": 1.416666666666667e-05,
+      "loss": 1.0308,
+      "step": 17
+    },
+    {
+      "epoch": 0.0938722294654498,
+      "grad_norm": 1.7734881445436932,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.0104,
+      "step": 18
+    },
+    {
+      "epoch": 0.09908735332464146,
+      "grad_norm": 1.2852782680220982,
+      "learning_rate": 1.5833333333333333e-05,
+      "loss": 1.0159,
+      "step": 19
+    },
+    {
+      "epoch": 0.10430247718383312,
+      "grad_norm": 1.4664459559013807,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9997,
+      "step": 20
+    },
+    {
+      "epoch": 0.10951760104302477,
+      "grad_norm": 1.2747674832880032,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.9999,
+      "step": 21
+    },
+    {
+      "epoch": 0.11473272490221642,
+      "grad_norm": 1.1325966855930794,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9845,
+      "step": 22
+    },
+    {
+      "epoch": 0.11994784876140809,
+      "grad_norm": 1.178213874446251,
+      "learning_rate": 1.916666666666667e-05,
+      "loss": 0.9754,
+      "step": 23
+    },
+    {
+      "epoch": 0.12516297262059975,
+      "grad_norm": 1.0070300787437625,
+      "learning_rate": 2e-05,
+      "loss": 0.9916,
+      "step": 24
+    },
+    {
+      "epoch": 0.1303780964797914,
+      "grad_norm": 1.1301814564159125,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.9535,
+      "step": 25
+    },
+    {
+      "epoch": 0.13559322033898305,
+      "grad_norm": 1.1582102527170561,
+      "learning_rate": 2.1666666666666667e-05,
+      "loss": 0.9559,
+      "step": 26
+    },
+    {
+      "epoch": 0.1408083441981747,
+      "grad_norm": 1.28096007382199,
+      "learning_rate": 2.25e-05,
+      "loss": 0.953,
+      "step": 27
+    },
+    {
+      "epoch": 0.14602346805736635,
+      "grad_norm": 1.3963901299642703,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.9467,
+      "step": 28
+    },
+    {
+      "epoch": 0.15123859191655803,
+      "grad_norm": 1.2008437237669338,
+      "learning_rate": 2.4166666666666667e-05,
+      "loss": 0.9319,
+      "step": 29
+    },
+    {
+      "epoch": 0.15645371577574968,
+      "grad_norm": 1.4579402450445522,
+      "learning_rate": 2.5e-05,
+      "loss": 0.9583,
+      "step": 30
+    },
+    {
+      "epoch": 0.16166883963494133,
+      "grad_norm": 1.332654985022459,
+      "learning_rate": 2.5833333333333336e-05,
+      "loss": 0.9222,
+      "step": 31
+    },
+    {
+      "epoch": 0.16688396349413298,
+      "grad_norm": 1.0118249538528512,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.919,
+      "step": 32
+    },
+    {
+      "epoch": 0.17209908735332463,
+      "grad_norm": 1.9244594897055562,
+      "learning_rate": 2.75e-05,
+      "loss": 0.9432,
+      "step": 33
+    },
+    {
+      "epoch": 0.1773142112125163,
+      "grad_norm": 1.1988103983333642,
+      "learning_rate": 2.833333333333334e-05,
+      "loss": 0.9309,
+      "step": 34
+    },
+    {
+      "epoch": 0.18252933507170796,
+      "grad_norm": 1.375838275777245,
+      "learning_rate": 2.9166666666666666e-05,
+      "loss": 0.9164,
+      "step": 35
+    },
+    {
+      "epoch": 0.1877444589308996,
+      "grad_norm": 1.662174214128309,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 0.9116,
+      "step": 36
+    },
+    {
+      "epoch": 0.19295958279009126,
+      "grad_norm": 19.880826748141878,
+      "learning_rate": 3.0833333333333335e-05,
+      "loss": 0.9136,
+      "step": 37
+    },
+    {
+      "epoch": 0.1981747066492829,
+      "grad_norm": 2.1089494461311897,
+      "learning_rate": 3.1666666666666666e-05,
+      "loss": 0.9288,
+      "step": 38
+    },
+    {
+      "epoch": 0.2033898305084746,
+      "grad_norm": 1.0754107081344952,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.9173,
+      "step": 39
+    },
+    {
+      "epoch": 0.20860495436766624,
+      "grad_norm": 3.9611232218120964,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.9227,
+      "step": 40
+    },
+    {
+      "epoch": 0.2138200782268579,
+      "grad_norm": 3.0041123063486306,
+      "learning_rate": 3.4166666666666666e-05,
+      "loss": 0.9319,
+      "step": 41
+    },
+    {
+      "epoch": 0.21903520208604954,
+      "grad_norm": 2.7847797531368066,
+      "learning_rate": 3.5000000000000004e-05,
+      "loss": 0.9233,
+      "step": 42
+    },
+    {
+      "epoch": 0.2242503259452412,
+      "grad_norm": 2.2845773931740725,
+      "learning_rate": 3.5833333333333335e-05,
+      "loss": 0.9286,
+      "step": 43
+    },
+    {
+      "epoch": 0.22946544980443284,
+      "grad_norm": 2.9550586599454363,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.9194,
+      "step": 44
+    },
+    {
+      "epoch": 0.23468057366362452,
+      "grad_norm": 1.713180532466099,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.8909,
+      "step": 45
+    },
+    {
+      "epoch": 0.23989569752281617,
+      "grad_norm": 3.6245500815283984,
+      "learning_rate": 3.833333333333334e-05,
+      "loss": 0.9061,
+      "step": 46
+    },
+    {
+      "epoch": 0.24511082138200782,
+      "grad_norm": 2.7780460507158034,
+      "learning_rate": 3.9166666666666665e-05,
+      "loss": 0.9165,
+      "step": 47
+    },
+    {
+      "epoch": 0.2503259452411995,
+      "grad_norm": 3.3854708521313563,
+      "learning_rate": 4e-05,
+      "loss": 0.9076,
+      "step": 48
+    },
+    {
+      "epoch": 0.25554106910039115,
+      "grad_norm": 2.788766938646062,
+      "learning_rate": 4.0833333333333334e-05,
+      "loss": 0.9086,
+      "step": 49
+    },
+    {
+      "epoch": 0.2607561929595828,
+      "grad_norm": 3.769154968802035,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.8985,
+      "step": 50
+    },
+    {
+      "epoch": 0.26597131681877445,
+      "grad_norm": 3.6676032273552783,
+      "learning_rate": 4.25e-05,
+      "loss": 0.9186,
+      "step": 51
+    },
+    {
+      "epoch": 0.2711864406779661,
+      "grad_norm": 2.300855067051272,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.8785,
+      "step": 52
+    },
+    {
+      "epoch": 0.27640156453715775,
+      "grad_norm": 2.0855212942964156,
+      "learning_rate": 4.416666666666667e-05,
+      "loss": 0.9044,
+      "step": 53
+    },
+    {
+      "epoch": 0.2816166883963494,
+      "grad_norm": 3.1541260770808144,
+      "learning_rate": 4.5e-05,
+      "loss": 0.9134,
+      "step": 54
+    },
+    {
+      "epoch": 0.28683181225554105,
+      "grad_norm": 2.3439141139233635,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.8849,
+      "step": 55
+    },
+    {
+      "epoch": 0.2920469361147327,
+      "grad_norm": 3.622853136869387,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.8968,
+      "step": 56
+    },
+    {
+      "epoch": 0.29726205997392435,
+      "grad_norm": 3.3444739650293793,
+      "learning_rate": 4.75e-05,
+      "loss": 0.9119,
+      "step": 57
+    },
+    {
+      "epoch": 0.30247718383311606,
+      "grad_norm": 2.4056743976018007,
+      "learning_rate": 4.8333333333333334e-05,
+      "loss": 0.8978,
+      "step": 58
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 2.0165924530855692,
+      "learning_rate": 4.916666666666667e-05,
+      "loss": 0.8842,
+      "step": 59
+    },
+    {
+      "epoch": 0.31290743155149936,
+      "grad_norm": 3.199274379857031,
+      "learning_rate": 5e-05,
+      "loss": 0.8926,
+      "step": 60
+    },
+    {
+      "epoch": 0.318122555410691,
+      "grad_norm": 2.145560466910737,
+      "learning_rate": 5.0833333333333333e-05,
+      "loss": 0.8875,
+      "step": 61
+    },
+    {
+      "epoch": 0.32333767926988266,
+      "grad_norm": 3.3364603622456923,
+      "learning_rate": 5.166666666666667e-05,
+      "loss": 0.8891,
+      "step": 62
+    },
+    {
+      "epoch": 0.3285528031290743,
+      "grad_norm": 3.088241155388676,
+      "learning_rate": 5.25e-05,
+      "loss": 0.8989,
+      "step": 63
+    },
+    {
+      "epoch": 0.33376792698826596,
+      "grad_norm": 2.2703016974586836,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.8848,
+      "step": 64
+    },
+    {
+      "epoch": 0.3389830508474576,
+      "grad_norm": 2.1056335353722875,
+      "learning_rate": 5.416666666666667e-05,
+      "loss": 0.8895,
+      "step": 65
+    },
+    {
+      "epoch": 0.34419817470664926,
+      "grad_norm": 2.195474617684906,
+      "learning_rate": 5.5e-05,
+      "loss": 0.8855,
+      "step": 66
+    },
+    {
+      "epoch": 0.3494132985658409,
+      "grad_norm": 1.8388111784516368,
+      "learning_rate": 5.583333333333333e-05,
+      "loss": 0.88,
+      "step": 67
+    },
+    {
+      "epoch": 0.3546284224250326,
+      "grad_norm": 2.205537218893776,
+      "learning_rate": 5.666666666666668e-05,
+      "loss": 0.8923,
+      "step": 68
+    },
+    {
+      "epoch": 0.35984354628422427,
+      "grad_norm": 1.43087259231813,
+      "learning_rate": 5.75e-05,
+      "loss": 0.8783,
+      "step": 69
+    },
+    {
+      "epoch": 0.3650586701434159,
+      "grad_norm": 3.170207545618331,
+      "learning_rate": 5.833333333333333e-05,
+      "loss": 0.8979,
+      "step": 70
+    },
+    {
+      "epoch": 0.37027379400260757,
+      "grad_norm": 2.4965183591219686,
+      "learning_rate": 5.916666666666668e-05,
+      "loss": 0.887,
+      "step": 71
+    },
+    {
+      "epoch": 0.3754889178617992,
+      "grad_norm": 2.18025601493941,
+      "learning_rate": 6.000000000000001e-05,
+      "loss": 0.8885,
+      "step": 72
+    },
+    {
+      "epoch": 0.38070404172099087,
+      "grad_norm": 2.320595439694094,
+      "learning_rate": 6.083333333333333e-05,
+      "loss": 0.8815,
+      "step": 73
+    },
+    {
+      "epoch": 0.3859191655801825,
+      "grad_norm": 2.0185818109580467,
+      "learning_rate": 6.166666666666667e-05,
+      "loss": 0.8831,
+      "step": 74
+    },
+    {
+      "epoch": 0.39113428943937417,
+      "grad_norm": 2.2105832623786745,
+      "learning_rate": 6.25e-05,
+      "loss": 0.8666,
+      "step": 75
+    },
+    {
+      "epoch": 0.3963494132985658,
+      "grad_norm": 1.9340187658703214,
+      "learning_rate": 6.333333333333333e-05,
+      "loss": 0.888,
+      "step": 76
+    },
+    {
+      "epoch": 0.4015645371577575,
+      "grad_norm": 1.811786813233452,
+      "learning_rate": 6.416666666666668e-05,
+      "loss": 0.889,
+      "step": 77
+    },
+    {
+      "epoch": 0.4067796610169492,
+      "grad_norm": 2.0985830559238563,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 0.8945,
+      "step": 78
+    },
+    {
+      "epoch": 0.41199478487614083,
+      "grad_norm": 2.7706210707741885,
+      "learning_rate": 6.583333333333334e-05,
+      "loss": 0.903,
+      "step": 79
+    },
+    {
+      "epoch": 0.4172099087353325,
+      "grad_norm": 2.328367128642788,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.886,
+      "step": 80
+    },
+    {
+      "epoch": 0.42242503259452413,
+      "grad_norm": 2.7310201710261857,
+      "learning_rate": 6.75e-05,
+      "loss": 0.8728,
+      "step": 81
+    },
+    {
+      "epoch": 0.4276401564537158,
+      "grad_norm": 2.012060559904938,
+      "learning_rate": 6.833333333333333e-05,
+      "loss": 0.8807,
+      "step": 82
+    },
+    {
+      "epoch": 0.43285528031290743,
+      "grad_norm": 1.8442023724268992,
+      "learning_rate": 6.916666666666668e-05,
+      "loss": 0.8832,
+      "step": 83
+    },
+    {
+      "epoch": 0.4380704041720991,
+      "grad_norm": 3.3879560119725225,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 0.8848,
+      "step": 84
+    },
+    {
+      "epoch": 0.44328552803129073,
+      "grad_norm": 2.0302673957424457,
+      "learning_rate": 7.083333333333334e-05,
+      "loss": 0.8716,
+      "step": 85
+    },
+    {
+      "epoch": 0.4485006518904824,
+      "grad_norm": 3.238439147300149,
+      "learning_rate": 7.166666666666667e-05,
+      "loss": 0.8896,
+      "step": 86
+    },
+    {
+      "epoch": 0.45371577574967403,
+      "grad_norm": 2.718682181237891,
+      "learning_rate": 7.25e-05,
+      "loss": 0.902,
+      "step": 87
+    },
+    {
+      "epoch": 0.4589308996088657,
+      "grad_norm": 2.4102500849494035,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 0.8865,
+      "step": 88
+    },
+    {
+      "epoch": 0.4641460234680574,
+      "grad_norm": 2.057159117257536,
+      "learning_rate": 7.416666666666668e-05,
+      "loss": 0.8714,
+      "step": 89
+    },
+    {
+      "epoch": 0.46936114732724904,
+      "grad_norm": 2.303058894217337,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.875,
+      "step": 90
+    },
+    {
+      "epoch": 0.4745762711864407,
+      "grad_norm": 1.853131768077206,
+      "learning_rate": 7.583333333333334e-05,
+      "loss": 0.8697,
+      "step": 91
+    },
+    {
+      "epoch": 0.47979139504563234,
+      "grad_norm": 1.5089115884426068,
+      "learning_rate": 7.666666666666668e-05,
+      "loss": 0.8672,
+      "step": 92
+    },
+    {
+      "epoch": 0.485006518904824,
+      "grad_norm": 3.288728672540363,
+      "learning_rate": 7.75e-05,
+      "loss": 0.8787,
+      "step": 93
+    },
+    {
+      "epoch": 0.49022164276401564,
+      "grad_norm": 2.8218607718403197,
+      "learning_rate": 7.833333333333333e-05,
+      "loss": 0.8886,
+      "step": 94
+    },
+    {
+      "epoch": 0.4954367666232073,
+      "grad_norm": 1.966995373674031,
+      "learning_rate": 7.916666666666668e-05,
+      "loss": 0.8738,
+      "step": 95
+    },
+    {
+      "epoch": 0.500651890482399,
+      "grad_norm": 4.193763943813693,
+      "learning_rate": 8e-05,
+      "loss": 0.8813,
+      "step": 96
+    },
+    {
+      "epoch": 0.5058670143415906,
+      "grad_norm": 2.2253144313432514,
+      "learning_rate": 7.99997324882088e-05,
+      "loss": 0.8778,
+      "step": 97
+    },
+    {
+      "epoch": 0.5110821382007823,
+      "grad_norm": 4.059040973559077,
+      "learning_rate": 7.999892995641334e-05,
+      "loss": 0.8844,
+      "step": 98
+    },
+    {
+      "epoch": 0.516297262059974,
+      "grad_norm": 2.814215374527903,
+      "learning_rate": 7.999759241534794e-05,
+      "loss": 0.8899,
+      "step": 99
+    },
+    {
+      "epoch": 0.5215123859191656,
+      "grad_norm": 3.1446586623159383,
+      "learning_rate": 7.9995719882903e-05,
+      "loss": 0.8933,
+      "step": 100
+    },
+    {
+      "epoch": 0.5267275097783573,
+      "grad_norm": 2.5831777088882335,
+      "learning_rate": 7.999331238412474e-05,
+      "loss": 0.885,
+      "step": 101
+    },
+    {
+      "epoch": 0.5319426336375489,
+      "grad_norm": 2.571515738465628,
+      "learning_rate": 7.99903699512149e-05,
+      "loss": 0.8773,
+      "step": 102
+    },
+    {
+      "epoch": 0.5371577574967406,
+      "grad_norm": 2.028582558881839,
+      "learning_rate": 7.998689262353024e-05,
+      "loss": 0.875,
+      "step": 103
+    },
+    {
+      "epoch": 0.5423728813559322,
+      "grad_norm": 1.9833279742128762,
+      "learning_rate": 7.998288044758206e-05,
+      "loss": 0.8864,
+      "step": 104
+    },
+    {
+      "epoch": 0.5475880052151239,
+      "grad_norm": 1.595344120443328,
+      "learning_rate": 7.99783334770356e-05,
+      "loss": 0.8772,
+      "step": 105
+    },
+    {
+      "epoch": 0.5528031290743155,
+      "grad_norm": 2.3698581665478025,
+      "learning_rate": 7.997325177270926e-05,
+      "loss": 0.8587,
+      "step": 106
+    },
+    {
+      "epoch": 0.5580182529335072,
+      "grad_norm": 1.8069354168584124,
+      "learning_rate": 7.996763540257382e-05,
+      "loss": 0.8786,
+      "step": 107
+    },
+    {
+      "epoch": 0.5632333767926988,
+      "grad_norm": 2.8802931859293968,
+      "learning_rate": 7.996148444175155e-05,
+      "loss": 0.8651,
+      "step": 108
+    },
+    {
+      "epoch": 0.5684485006518905,
+      "grad_norm": 2.163396645739927,
+      "learning_rate": 7.99547989725152e-05,
+      "loss": 0.8779,
+      "step": 109
+    },
+    {
+      "epoch": 0.5736636245110821,
+      "grad_norm": 1.7103752366008835,
+      "learning_rate": 7.994757908428683e-05,
+      "loss": 0.8627,
+      "step": 110
+    },
+    {
+      "epoch": 0.5788787483702738,
+      "grad_norm": 2.6236911355349366,
+      "learning_rate": 7.99398248736367e-05,
+      "loss": 0.8775,
+      "step": 111
+    },
+    {
+      "epoch": 0.5840938722294654,
+      "grad_norm": 1.7695200626233123,
+      "learning_rate": 7.993153644428198e-05,
+      "loss": 0.8675,
+      "step": 112
+    },
+    {
+      "epoch": 0.5893089960886571,
+      "grad_norm": 3.1189819236023597,
+      "learning_rate": 7.992271390708529e-05,
+      "loss": 0.864,
+      "step": 113
+    },
+    {
+      "epoch": 0.5945241199478487,
+      "grad_norm": 2.519671466442974,
+      "learning_rate": 7.991335738005325e-05,
+      "loss": 0.8753,
+      "step": 114
+    },
+    {
+      "epoch": 0.5997392438070405,
+      "grad_norm": 2.445885184903824,
+      "learning_rate": 7.990346698833493e-05,
+      "loss": 0.8791,
+      "step": 115
+    },
+    {
+      "epoch": 0.6049543676662321,
+      "grad_norm": 2.6107239711114634,
+      "learning_rate": 7.989304286422016e-05,
+      "loss": 0.8688,
+      "step": 116
+    },
+    {
+      "epoch": 0.6101694915254238,
+      "grad_norm": 2.3280917526561926,
+      "learning_rate": 7.988208514713773e-05,
+      "loss": 0.8558,
+      "step": 117
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 2.01951998260186,
+      "learning_rate": 7.987059398365358e-05,
+      "loss": 0.8611,
+      "step": 118
+    },
+    {
+      "epoch": 0.6205997392438071,
+      "grad_norm": 2.67003693356188,
+      "learning_rate": 7.98585695274688e-05,
+      "loss": 0.8599,
+      "step": 119
+    },
+    {
+      "epoch": 0.6258148631029987,
+      "grad_norm": 1.3989184556669023,
+      "learning_rate": 7.984601193941757e-05,
+      "loss": 0.8619,
+      "step": 120
+    },
+    {
+      "epoch": 0.6310299869621904,
+      "grad_norm": 2.9657726548979007,
+      "learning_rate": 7.983292138746504e-05,
+      "loss": 0.8533,
+      "step": 121
+    },
+    {
+      "epoch": 0.636245110821382,
+      "grad_norm": 2.2111378936066717,
+      "learning_rate": 7.981929804670505e-05,
+      "loss": 0.8647,
+      "step": 122
+    },
+    {
+      "epoch": 0.6414602346805737,
+      "grad_norm": 1.7933021378995146,
+      "learning_rate": 7.980514209935783e-05,
+      "loss": 0.8572,
+      "step": 123
+    },
+    {
+      "epoch": 0.6466753585397653,
+      "grad_norm": 2.6146019104563036,
+      "learning_rate": 7.97904537347675e-05,
+      "loss": 0.8629,
+      "step": 124
+    },
+    {
+      "epoch": 0.651890482398957,
+      "grad_norm": 1.7226667254129786,
+      "learning_rate": 7.977523314939961e-05,
+      "loss": 0.8728,
+      "step": 125
+    },
+    {
+      "epoch": 0.6571056062581486,
+      "grad_norm": 2.4681433772246804,
+      "learning_rate": 7.975948054683847e-05,
+      "loss": 0.8772,
+      "step": 126
+    },
+    {
+      "epoch": 0.6623207301173403,
+      "grad_norm": 1.8531960262958524,
+      "learning_rate": 7.974319613778441e-05,
+      "loss": 0.8568,
+      "step": 127
+    },
+    {
+      "epoch": 0.6675358539765319,
+      "grad_norm": 3.3273564870681067,
+      "learning_rate": 7.972638014005102e-05,
+      "loss": 0.8674,
+      "step": 128
+    },
+    {
+      "epoch": 0.6727509778357236,
+      "grad_norm": 1.6565232049947463,
+      "learning_rate": 7.970903277856216e-05,
+      "loss": 0.8593,
+      "step": 129
+    },
+    {
+      "epoch": 0.6779661016949152,
+      "grad_norm": 3.2772231491550157,
+      "learning_rate": 7.969115428534904e-05,
+      "loss": 0.8689,
+      "step": 130
+    },
+    {
+      "epoch": 0.6831812255541069,
+      "grad_norm": 2.9366197326510175,
+      "learning_rate": 7.967274489954703e-05,
+      "loss": 0.8562,
+      "step": 131
+    },
+    {
+      "epoch": 0.6883963494132985,
+      "grad_norm": 1.579970961599737,
+      "learning_rate": 7.965380486739253e-05,
+      "loss": 0.8422,
+      "step": 132
+    },
+    {
+      "epoch": 0.6936114732724902,
+      "grad_norm": 3.1594495489774967,
+      "learning_rate": 7.963433444221964e-05,
+      "loss": 0.875,
+      "step": 133
+    },
+    {
+      "epoch": 0.6988265971316818,
+      "grad_norm": 2.3403703396896423,
+      "learning_rate": 7.961433388445676e-05,
+      "loss": 0.8591,
+      "step": 134
+    },
+    {
+      "epoch": 0.7040417209908736,
+      "grad_norm": 1.8764160357986022,
+      "learning_rate": 7.959380346162314e-05,
+      "loss": 0.8585,
+      "step": 135
+    },
+    {
+      "epoch": 0.7092568448500652,
+      "grad_norm": 2.954668141615744,
+      "learning_rate": 7.957274344832533e-05,
+      "loss": 0.8558,
+      "step": 136
+    },
+    {
+      "epoch": 0.7144719687092569,
+      "grad_norm": 2.1132431617150647,
+      "learning_rate": 7.955115412625337e-05,
+      "loss": 0.8597,
+      "step": 137
+    },
+    {
+      "epoch": 0.7196870925684485,
+      "grad_norm": 1.9269531529548758,
+      "learning_rate": 7.952903578417719e-05,
+      "loss": 0.8553,
+      "step": 138
+    },
+    {
+      "epoch": 0.7249022164276402,
+      "grad_norm": 1.9600506890917169,
+      "learning_rate": 7.950638871794268e-05,
+      "loss": 0.8498,
+      "step": 139
+    },
+    {
+      "epoch": 0.7301173402868318,
+      "grad_norm": 1.295131341774357,
+      "learning_rate": 7.948321323046766e-05,
+      "loss": 0.862,
+      "step": 140
+    },
+    {
+      "epoch": 0.7353324641460235,
+      "grad_norm": 2.598262766807541,
+      "learning_rate": 7.945950963173797e-05,
+      "loss": 0.8743,
+      "step": 141
+    },
+    {
+      "epoch": 0.7405475880052151,
+      "grad_norm": 1.7173227811522496,
+      "learning_rate": 7.943527823880321e-05,
+      "loss": 0.857,
+      "step": 142
+    },
+    {
+      "epoch": 0.7457627118644068,
+      "grad_norm": 2.778103098366881,
+      "learning_rate": 7.941051937577255e-05,
+      "loss": 0.8644,
+      "step": 143
+    },
+    {
+      "epoch": 0.7509778357235984,
+      "grad_norm": 1.9721804979121416,
+      "learning_rate": 7.938523337381036e-05,
+      "loss": 0.8641,
+      "step": 144
+    },
+    {
+      "epoch": 0.7561929595827901,
+      "grad_norm": 2.534467842230536,
+      "learning_rate": 7.935942057113185e-05,
+      "loss": 0.8607,
+      "step": 145
+    },
+    {
+      "epoch": 0.7614080834419817,
+      "grad_norm": 2.420990126278799,
+      "learning_rate": 7.933308131299846e-05,
+      "loss": 0.8532,
+      "step": 146
+    },
+    {
+      "epoch": 0.7666232073011734,
+      "grad_norm": 2.0250401094262775,
+      "learning_rate": 7.93062159517133e-05,
+      "loss": 0.8463,
+      "step": 147
+    },
+    {
+      "epoch": 0.771838331160365,
+      "grad_norm": 1.949074222131184,
+      "learning_rate": 7.92788248466164e-05,
+      "loss": 0.8505,
+      "step": 148
+    },
+    {
+      "epoch": 0.7770534550195567,
+      "grad_norm": 1.9136794648235018,
+      "learning_rate": 7.925090836407997e-05,
+      "loss": 0.8539,
+      "step": 149
+    },
+    {
+      "epoch": 0.7822685788787483,
+      "grad_norm": 1.3862902439859852,
+      "learning_rate": 7.922246687750341e-05,
+      "loss": 0.8511,
+      "step": 150
+    },
+    {
+      "epoch": 0.78748370273794,
+      "grad_norm": 2.238512517144782,
+      "learning_rate": 7.919350076730836e-05,
+      "loss": 0.846,
+      "step": 151
+    },
+    {
+      "epoch": 0.7926988265971316,
+      "grad_norm": 2.0835619443377134,
+      "learning_rate": 7.916401042093361e-05,
+      "loss": 0.8461,
+      "step": 152
+    },
+    {
+      "epoch": 0.7979139504563233,
+      "grad_norm": 1.8060011410999197,
+      "learning_rate": 7.913399623282997e-05,
+      "loss": 0.833,
+      "step": 153
+    },
+    {
+      "epoch": 0.803129074315515,
+      "grad_norm": 2.0703223704187357,
+      "learning_rate": 7.910345860445487e-05,
+      "loss": 0.8447,
+      "step": 154
+    },
+    {
+      "epoch": 0.8083441981747066,
+      "grad_norm": 2.299135311289539,
+      "learning_rate": 7.90723979442671e-05,
+      "loss": 0.8533,
+      "step": 155
+    },
+    {
+      "epoch": 0.8135593220338984,
+      "grad_norm": 1.8730518820522357,
+      "learning_rate": 7.90408146677213e-05,
+      "loss": 0.8372,
+      "step": 156
+    },
+    {
+      "epoch": 0.81877444589309,
+      "grad_norm": 1.5402936247797026,
+      "learning_rate": 7.900870919726244e-05,
+      "loss": 0.8408,
+      "step": 157
+    },
+    {
+      "epoch": 0.8239895697522817,
+      "grad_norm": 2.378224013864058,
+      "learning_rate": 7.897608196232007e-05,
+      "loss": 0.8492,
+      "step": 158
+    },
+    {
+      "epoch": 0.8292046936114733,
+      "grad_norm": 1.8241375846314414,
+      "learning_rate": 7.894293339930272e-05,
+      "loss": 0.8338,
+      "step": 159
+    },
+    {
+      "epoch": 0.834419817470665,
+      "grad_norm": 2.2711975231391826,
+      "learning_rate": 7.890926395159197e-05,
+      "loss": 0.8385,
+      "step": 160
+    },
+    {
+      "epoch": 0.8396349413298566,
+      "grad_norm": 1.6195285561704085,
+      "learning_rate": 7.887507406953651e-05,
+      "loss": 0.8489,
+      "step": 161
+    },
+    {
+      "epoch": 0.8448500651890483,
+      "grad_norm": 2.2453846392163768,
+      "learning_rate": 7.884036421044618e-05,
+      "loss": 0.8487,
+      "step": 162
+    },
+    {
+      "epoch": 0.8500651890482399,
+      "grad_norm": 1.940408392169657,
+      "learning_rate": 7.880513483858583e-05,
+      "loss": 0.8398,
+      "step": 163
+    },
+    {
+      "epoch": 0.8552803129074316,
+      "grad_norm": 2.2159286823222675,
+      "learning_rate": 7.876938642516905e-05,
+      "loss": 0.8492,
+      "step": 164
+    },
+    {
+      "epoch": 0.8604954367666232,
+      "grad_norm": 1.4842300453687591,
+      "learning_rate": 7.873311944835195e-05,
+      "loss": 0.8376,
+      "step": 165
+    },
+    {
+      "epoch": 0.8657105606258149,
+      "grad_norm": 2.491755340230041,
+      "learning_rate": 7.869633439322674e-05,
+      "loss": 0.8386,
+      "step": 166
+    },
+    {
+      "epoch": 0.8709256844850065,
+      "grad_norm": 1.8127014745278216,
+      "learning_rate": 7.865903175181521e-05,
+      "loss": 0.8318,
+      "step": 167
+    },
+    {
+      "epoch": 0.8761408083441982,
+      "grad_norm": 1.5105514574695558,
+      "learning_rate": 7.862121202306217e-05,
+      "loss": 0.8317,
+      "step": 168
+    },
+    {
+      "epoch": 0.8813559322033898,
+      "grad_norm": 2.485520131584514,
+      "learning_rate": 7.858287571282882e-05,
+      "loss": 0.8496,
+      "step": 169
+    },
+    {
+      "epoch": 0.8865710560625815,
+      "grad_norm": 1.7516963763143856,
+      "learning_rate": 7.854402333388587e-05,
+      "loss": 0.8433,
+      "step": 170
+    },
+    {
+      "epoch": 0.8917861799217731,
+      "grad_norm": 2.462296826145561,
+      "learning_rate": 7.850465540590684e-05,
+      "loss": 0.8537,
+      "step": 171
+    },
+    {
+      "epoch": 0.8970013037809648,
+      "grad_norm": 1.1958019189200906,
+      "learning_rate": 7.846477245546094e-05,
+      "loss": 0.8323,
+      "step": 172
+    },
+    {
+      "epoch": 0.9022164276401564,
+      "grad_norm": 2.6026749193200693,
+      "learning_rate": 7.842437501600616e-05,
+      "loss": 0.8516,
+      "step": 173
+    },
+    {
+      "epoch": 0.9074315514993481,
+      "grad_norm": 1.7619131106705077,
+      "learning_rate": 7.838346362788206e-05,
+      "loss": 0.838,
+      "step": 174
+    },
+    {
+      "epoch": 0.9126466753585397,
+      "grad_norm": 2.4386182828768312,
+      "learning_rate": 7.834203883830259e-05,
+      "loss": 0.8444,
+      "step": 175
+    },
+    {
+      "epoch": 0.9178617992177314,
+      "grad_norm": 2.0969277412774274,
+      "learning_rate": 7.830010120134873e-05,
+      "loss": 0.8622,
+      "step": 176
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 1.4692953467652599,
+      "learning_rate": 7.825765127796108e-05,
+      "loss": 0.8338,
+      "step": 177
+    },
+    {
+      "epoch": 0.9282920469361148,
+      "grad_norm": 1.5923417027165594,
+      "learning_rate": 7.821468963593242e-05,
+      "loss": 0.8391,
+      "step": 178
+    },
+    {
+      "epoch": 0.9335071707953064,
+      "grad_norm": 2.2009607061094942,
+      "learning_rate": 7.817121684990004e-05,
+      "loss": 0.8589,
+      "step": 179
+    },
+    {
+      "epoch": 0.9387222946544981,
+      "grad_norm": 1.5116758680331852,
+      "learning_rate": 7.812723350133805e-05,
+      "loss": 0.8341,
+      "step": 180
+    },
+    {
+      "epoch": 0.9439374185136897,
+      "grad_norm": 2.038874732993399,
+      "learning_rate": 7.80827401785497e-05,
+      "loss": 0.8547,
+      "step": 181
+    },
+    {
+      "epoch": 0.9491525423728814,
+      "grad_norm": 2.2985530062759514,
+      "learning_rate": 7.80377374766594e-05,
+      "loss": 0.8391,
+      "step": 182
+    },
+    {
+      "epoch": 0.954367666232073,
+      "grad_norm": 1.257017121878146,
+      "learning_rate": 7.799222599760481e-05,
+      "loss": 0.8325,
+      "step": 183
+    },
+    {
+      "epoch": 0.9595827900912647,
+      "grad_norm": 2.873446007386103,
+      "learning_rate": 7.794620635012883e-05,
+      "loss": 0.853,
+      "step": 184
+    },
+    {
+      "epoch": 0.9647979139504563,
+      "grad_norm": 2.0836694911738207,
+      "learning_rate": 7.789967914977134e-05,
+      "loss": 0.8554,
+      "step": 185
+    },
+    {
+      "epoch": 0.970013037809648,
+      "grad_norm": 2.095712201456811,
+      "learning_rate": 7.785264501886108e-05,
+      "loss": 0.8439,
+      "step": 186
+    },
+    {
+      "epoch": 0.9752281616688396,
+      "grad_norm": 1.668972201159559,
+      "learning_rate": 7.78051045865073e-05,
+      "loss": 0.8442,
+      "step": 187
+    },
+    {
+      "epoch": 0.9804432855280313,
+      "grad_norm": 2.6548529934832787,
+      "learning_rate": 7.77570584885913e-05,
+      "loss": 0.8509,
+      "step": 188
+    },
+    {
+      "epoch": 0.9856584093872229,
+      "grad_norm": 1.6440679506158635,
+      "learning_rate": 7.770850736775796e-05,
+      "loss": 0.8367,
+      "step": 189
+    },
+    {
+      "epoch": 0.9908735332464146,
+      "grad_norm": 2.895095407046929,
+      "learning_rate": 7.765945187340715e-05,
+      "loss": 0.8476,
+      "step": 190
+    },
+    {
+      "epoch": 0.9960886571056062,
+      "grad_norm": 2.1880375444766655,
+      "learning_rate": 7.760989266168503e-05,
+      "loss": 0.8466,
+      "step": 191
+    },
+    {
+      "epoch": 1.001303780964798,
+      "grad_norm": 2.491373495658068,
+      "learning_rate": 7.755983039547528e-05,
+      "loss": 1.0483,
+      "step": 192
+    },
+    {
+      "epoch": 1.0065189048239895,
+      "grad_norm": 2.1065510338703195,
+      "learning_rate": 7.750926574439019e-05,
+      "loss": 0.8317,
+      "step": 193
+    },
+    {
+      "epoch": 1.0117340286831813,
+      "grad_norm": 1.6048527323347803,
+      "learning_rate": 7.745819938476184e-05,
+      "loss": 0.8309,
+      "step": 194
+    },
+    {
+      "epoch": 1.0169491525423728,
+      "grad_norm": 1.3686572978259008,
+      "learning_rate": 7.740663199963284e-05,
+      "loss": 0.8207,
+      "step": 195
+    },
+    {
+      "epoch": 1.0221642764015646,
+      "grad_norm": 1.6395534429827354,
+      "learning_rate": 7.73545642787474e-05,
+      "loss": 0.824,
+      "step": 196
+    },
+    {
+      "epoch": 1.0273794002607561,
+      "grad_norm": 1.829519017394267,
+      "learning_rate": 7.730199691854198e-05,
+      "loss": 0.8346,
+      "step": 197
+    },
+    {
+      "epoch": 1.032594524119948,
+      "grad_norm": 2.086222113236519,
+      "learning_rate": 7.724893062213602e-05,
+      "loss": 0.8204,
+      "step": 198
+    },
+    {
+      "epoch": 1.0378096479791394,
+      "grad_norm": 1.6543925285010792,
+      "learning_rate": 7.71953660993225e-05,
+      "loss": 0.8238,
+      "step": 199
+    },
+    {
+      "epoch": 1.0430247718383312,
+      "grad_norm": 1.6861891172176104,
+      "learning_rate": 7.71413040665585e-05,
+      "loss": 0.8215,
+      "step": 200
+    },
+    {
+      "epoch": 1.0482398956975227,
+      "grad_norm": 2.313274305017442,
+      "learning_rate": 7.708674524695559e-05,
+      "loss": 0.8082,
+      "step": 201
+    },
+    {
+      "epoch": 1.0534550195567145,
+      "grad_norm": 1.8334252249650502,
+      "learning_rate": 7.703169037027014e-05,
+      "loss": 0.824,
+      "step": 202
+    },
+    {
+      "epoch": 1.058670143415906,
+      "grad_norm": 1.7126012022750563,
+      "learning_rate": 7.697614017289357e-05,
+      "loss": 0.8105,
+      "step": 203
+    },
+    {
+      "epoch": 1.0638852672750978,
+      "grad_norm": 1.3173869406989056,
+      "learning_rate": 7.692009539784255e-05,
+      "loss": 0.8303,
+      "step": 204
+    },
+    {
+      "epoch": 1.0691003911342893,
+      "grad_norm": 2.0262932335464012,
+      "learning_rate": 7.686355679474898e-05,
+      "loss": 0.8197,
+      "step": 205
+    },
+    {
+      "epoch": 1.074315514993481,
+      "grad_norm": 2.2714312463405397,
+      "learning_rate": 7.680652511985e-05,
+      "loss": 0.8196,
+      "step": 206
+    },
+    {
+      "epoch": 1.0795306388526726,
+      "grad_norm": 1.5002234528402432,
+      "learning_rate": 7.674900113597787e-05,
+      "loss": 0.8141,
+      "step": 207
+    },
+    {
+      "epoch": 1.0847457627118644,
+      "grad_norm": 2.059909369219652,
+      "learning_rate": 7.669098561254983e-05,
+      "loss": 0.826,
+      "step": 208
+    },
+    {
+      "epoch": 1.0899608865710562,
+      "grad_norm": 1.6415840594311673,
+      "learning_rate": 7.663247932555767e-05,
+      "loss": 0.8314,
+      "step": 209
+    },
+    {
+      "epoch": 1.0951760104302477,
+      "grad_norm": 2.094580841761285,
+      "learning_rate": 7.65734830575575e-05,
+      "loss": 0.8405,
+      "step": 210
+    },
+    {
+      "epoch": 1.1003911342894395,
+      "grad_norm": 1.4276266650819935,
+      "learning_rate": 7.651399759765915e-05,
+      "loss": 0.8172,
+      "step": 211
+    },
+    {
+      "epoch": 1.105606258148631,
+      "grad_norm": 2.406763290618582,
+      "learning_rate": 7.645402374151575e-05,
+      "loss": 0.8166,
+      "step": 212
+    },
+    {
+      "epoch": 1.1108213820078228,
+      "grad_norm": 1.4071963692309266,
+      "learning_rate": 7.639356229131298e-05,
+      "loss": 0.8252,
+      "step": 213
+    },
+    {
+      "epoch": 1.1160365058670143,
+      "grad_norm": 1.434300709398784,
+      "learning_rate": 7.633261405575838e-05,
+      "loss": 0.8342,
+      "step": 214
+    },
+    {
+      "epoch": 1.121251629726206,
+      "grad_norm": 2.6026163076276223,
+      "learning_rate": 7.627117985007052e-05,
+      "loss": 0.8277,
+      "step": 215
+    },
+    {
+      "epoch": 1.1264667535853976,
+      "grad_norm": 1.545445917096449,
+      "learning_rate": 7.620926049596814e-05,
+      "loss": 0.8224,
+      "step": 216
+    },
+    {
+      "epoch": 1.1316818774445894,
+      "grad_norm": 2.349483330688553,
+      "learning_rate": 7.61468568216591e-05,
+      "loss": 0.8317,
+      "step": 217
+    },
+    {
+      "epoch": 1.136897001303781,
+      "grad_norm": 1.6493926282653744,
+      "learning_rate": 7.60839696618293e-05,
+      "loss": 0.8235,
+      "step": 218
+    },
+    {
+      "epoch": 1.1421121251629727,
+      "grad_norm": 2.523447089439544,
+      "learning_rate": 7.602059985763165e-05,
+      "loss": 0.8259,
+      "step": 219
+    },
+    {
+      "epoch": 1.1473272490221642,
+      "grad_norm": 1.910781874766756,
+      "learning_rate": 7.595674825667457e-05,
+      "loss": 0.826,
+      "step": 220
+    },
+    {
+      "epoch": 1.152542372881356,
+      "grad_norm": 2.2270989973834032,
+      "learning_rate": 7.589241571301091e-05,
+      "loss": 0.8309,
+      "step": 221
+    },
+    {
+      "epoch": 1.1577574967405475,
+      "grad_norm": 1.7934683365685906,
+      "learning_rate": 7.582760308712634e-05,
+      "loss": 0.8302,
+      "step": 222
+    },
+    {
+      "epoch": 1.1629726205997393,
+      "grad_norm": 2.0662257630292062,
+      "learning_rate": 7.5762311245928e-05,
+      "loss": 0.8277,
+      "step": 223
+    },
+    {
+      "epoch": 1.1681877444589308,
+      "grad_norm": 1.6790104250183842,
+      "learning_rate": 7.569654106273268e-05,
+      "loss": 0.8254,
+      "step": 224
+    },
+    {
+      "epoch": 1.1734028683181226,
+      "grad_norm": 2.11893755909352,
+      "learning_rate": 7.563029341725541e-05,
+      "loss": 0.8183,
+      "step": 225
+    },
+    {
+      "epoch": 1.1786179921773141,
+      "grad_norm": 1.5877658958807297,
+      "learning_rate": 7.55635691955975e-05,
+      "loss": 0.8232,
+      "step": 226
+    },
+    {
+      "epoch": 1.1838331160365059,
+      "grad_norm": 2.147915736111546,
+      "learning_rate": 7.549636929023471e-05,
+      "loss": 0.8185,
+      "step": 227
+    },
+    {
+      "epoch": 1.1890482398956976,
+      "grad_norm": 1.7400398545071791,
+      "learning_rate": 7.542869460000544e-05,
+      "loss": 0.8219,
+      "step": 228
+    },
+    {
+      "epoch": 1.1942633637548892,
+      "grad_norm": 1.8017174411766266,
+      "learning_rate": 7.536054603009856e-05,
+      "loss": 0.8099,
+      "step": 229
+    },
+    {
+      "epoch": 1.1994784876140807,
+      "grad_norm": 1.7295101992953383,
+      "learning_rate": 7.529192449204137e-05,
+      "loss": 0.8199,
+      "step": 230
+    },
+    {
+      "epoch": 1.2046936114732725,
+      "grad_norm": 1.81480851347367,
+      "learning_rate": 7.522283090368739e-05,
+      "loss": 0.8163,
+      "step": 231
+    },
+    {
+      "epoch": 1.2099087353324642,
+      "grad_norm": 1.4532183826025098,
+      "learning_rate": 7.515326618920409e-05,
+      "loss": 0.8243,
+      "step": 232
+    },
+    {
+      "epoch": 1.2151238591916558,
+      "grad_norm": 2.025221417951849,
+      "learning_rate": 7.508323127906055e-05,
+      "loss": 0.8104,
+      "step": 233
+    },
+    {
+      "epoch": 1.2203389830508475,
+      "grad_norm": 2.839580655118928,
+      "learning_rate": 7.5012727110015e-05,
+      "loss": 0.8156,
+      "step": 234
+    },
+    {
+      "epoch": 1.225554106910039,
+      "grad_norm": 0.9612813608575648,
+      "learning_rate": 7.494175462510225e-05,
+      "loss": 0.8136,
+      "step": 235
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 4.580627208479269,
+      "learning_rate": 7.487031477362112e-05,
+      "loss": 0.8363,
+      "step": 236
+    },
+    {
+      "epoch": 1.2359843546284224,
+      "grad_norm": 3.3062212088133585,
+      "learning_rate": 7.479840851112175e-05,
+      "loss": 0.8457,
+      "step": 237
+    },
+    {
+      "epoch": 1.2411994784876141,
+      "grad_norm": 3.9085448168352577,
+      "learning_rate": 7.47260367993928e-05,
+      "loss": 0.8369,
+      "step": 238
+    },
+    {
+      "epoch": 1.2464146023468057,
+      "grad_norm": 3.2990078411550416,
+      "learning_rate": 7.465320060644857e-05,
+      "loss": 0.83,
+      "step": 239
+    },
+    {
+      "epoch": 1.2516297262059974,
+      "grad_norm": 3.171914889017843,
+      "learning_rate": 7.45799009065161e-05,
+      "loss": 0.8245,
+      "step": 240
+    },
+    {
+      "epoch": 1.256844850065189,
+      "grad_norm": 2.6816905524834436,
+      "learning_rate": 7.450613868002208e-05,
+      "loss": 0.8363,
+      "step": 241
+    },
+    {
+      "epoch": 1.2620599739243807,
+      "grad_norm": 3.0890905030641256,
+      "learning_rate": 7.443191491357976e-05,
+      "loss": 0.823,
+      "step": 242
+    },
+    {
+      "epoch": 1.2672750977835723,
+      "grad_norm": 2.3490265350478956,
+      "learning_rate": 7.435723059997581e-05,
+      "loss": 0.8276,
+      "step": 243
+    },
+    {
+      "epoch": 1.272490221642764,
+      "grad_norm": 3.5496378180168957,
+      "learning_rate": 7.428208673815693e-05,
+      "loss": 0.8247,
+      "step": 244
+    },
+    {
+      "epoch": 1.2777053455019556,
+      "grad_norm": 3.053170271456407,
+      "learning_rate": 7.420648433321659e-05,
+      "loss": 0.8306,
+      "step": 245
+    },
+    {
+      "epoch": 1.2829204693611473,
+      "grad_norm": 2.387063335590222,
+      "learning_rate": 7.41304243963815e-05,
+      "loss": 0.831,
+      "step": 246
+    },
+    {
+      "epoch": 1.288135593220339,
+      "grad_norm": 1.8497576227055157,
+      "learning_rate": 7.405390794499819e-05,
+      "loss": 0.8066,
+      "step": 247
+    },
+    {
+      "epoch": 1.2933507170795306,
+      "grad_norm": 2.9897041114580727,
+      "learning_rate": 7.397693600251929e-05,
+      "loss": 0.8134,
+      "step": 248
+    },
+    {
+      "epoch": 1.2985658409387222,
+      "grad_norm": 1.980452439525295,
+      "learning_rate": 7.389950959848992e-05,
+      "loss": 0.8252,
+      "step": 249
+    },
+    {
+      "epoch": 1.303780964797914,
+      "grad_norm": 3.4714237886205748,
+      "learning_rate": 7.382162976853387e-05,
+      "loss": 0.8294,
+      "step": 250
+    },
+    {
+      "epoch": 1.3089960886571057,
+      "grad_norm": 3.2963021601322535,
+      "learning_rate": 7.37432975543398e-05,
+      "loss": 0.8232,
+      "step": 251
+    },
+    {
+      "epoch": 1.3142112125162972,
+      "grad_norm": 2.1372765339871806,
+      "learning_rate": 7.366451400364723e-05,
+      "loss": 0.8228,
+      "step": 252
+    },
+    {
+      "epoch": 1.3194263363754888,
+      "grad_norm": 1.61329278641058,
+      "learning_rate": 7.358528017023262e-05,
+      "loss": 0.8149,
+      "step": 253
+    },
+    {
+      "epoch": 1.3246414602346805,
+      "grad_norm": 2.149446564090082,
+      "learning_rate": 7.350559711389518e-05,
+      "loss": 0.803,
+      "step": 254
+    },
+    {
+      "epoch": 1.3298565840938723,
+      "grad_norm": 1.2080706850732208,
+      "learning_rate": 7.342546590044279e-05,
+      "loss": 0.8309,
+      "step": 255
+    },
+    {
+      "epoch": 1.3350717079530638,
+      "grad_norm": 2.060776426075165,
+      "learning_rate": 7.334488760167768e-05,
+      "loss": 0.8218,
+      "step": 256
+    },
+    {
+      "epoch": 1.3402868318122556,
+      "grad_norm": 1.4270579486920896,
+      "learning_rate": 7.326386329538207e-05,
+      "loss": 0.8482,
+      "step": 257
+    },
+    {
+      "epoch": 1.3455019556714471,
+      "grad_norm": 2.320654966360905,
+      "learning_rate": 7.318239406530386e-05,
+      "loss": 0.8284,
+      "step": 258
+    },
+    {
+      "epoch": 1.350717079530639,
+      "grad_norm": 2.040787679253322,
+      "learning_rate": 7.3100481001142e-05,
+      "loss": 0.8436,
+      "step": 259
+    },
+    {
+      "epoch": 1.3559322033898304,
+      "grad_norm": 1.450733455499567,
+      "learning_rate": 7.301812519853203e-05,
+      "loss": 0.8067,
+      "step": 260
+    },
+    {
+      "epoch": 1.3611473272490222,
+      "grad_norm": 1.4473957425879223,
+      "learning_rate": 7.293532775903137e-05,
+      "loss": 0.8172,
+      "step": 261
+    },
+    {
+      "epoch": 1.3663624511082137,
+      "grad_norm": 1.4567295321246068,
+      "learning_rate": 7.285208979010458e-05,
+      "loss": 0.833,
+      "step": 262
+    },
+    {
+      "epoch": 1.3715775749674055,
+      "grad_norm": 2.2303402593920234,
+      "learning_rate": 7.276841240510858e-05,
+      "loss": 0.8241,
+      "step": 263
+    },
+    {
+      "epoch": 1.376792698826597,
+      "grad_norm": 1.3757747352916987,
+      "learning_rate": 7.26842967232777e-05,
+      "loss": 0.8106,
+      "step": 264
+    },
+    {
+      "epoch": 1.3820078226857888,
+      "grad_norm": 1.960651210566042,
+      "learning_rate": 7.25997438697088e-05,
+      "loss": 0.8258,
+      "step": 265
+    },
+    {
+      "epoch": 1.3872229465449806,
+      "grad_norm": 1.6973151493940888,
+      "learning_rate": 7.251475497534615e-05,
+      "loss": 0.8421,
+      "step": 266
+    },
+    {
+      "epoch": 1.3924380704041721,
+      "grad_norm": 1.4278961151807503,
+      "learning_rate": 7.242933117696628e-05,
+      "loss": 0.8107,
+      "step": 267
+    },
+    {
+      "epoch": 1.3976531942633637,
+      "grad_norm": 1.6765492543318952,
+      "learning_rate": 7.234347361716291e-05,
+      "loss": 0.8126,
+      "step": 268
+    },
+    {
+      "epoch": 1.4028683181225554,
+      "grad_norm": 1.7901534079828416,
+      "learning_rate": 7.225718344433149e-05,
+      "loss": 0.8209,
+      "step": 269
+    },
+    {
+      "epoch": 1.4080834419817472,
+      "grad_norm": 1.1879344799911462,
+      "learning_rate": 7.217046181265394e-05,
+      "loss": 0.8241,
+      "step": 270
+    },
+    {
+      "epoch": 1.4132985658409387,
+      "grad_norm": 2.316592389288362,
+      "learning_rate": 7.208330988208324e-05,
+      "loss": 0.8233,
+      "step": 271
+    },
+    {
+      "epoch": 1.4185136897001303,
+      "grad_norm": 1.3009721963229537,
+      "learning_rate": 7.199572881832784e-05,
+      "loss": 0.8094,
+      "step": 272
+    },
+    {
+      "epoch": 1.423728813559322,
+      "grad_norm": 2.068602624965321,
+      "learning_rate": 7.190771979283608e-05,
+      "loss": 0.8221,
+      "step": 273
+    },
+    {
+      "epoch": 1.4289439374185138,
+      "grad_norm": 1.9940233989402643,
+      "learning_rate": 7.181928398278058e-05,
+      "loss": 0.8225,
+      "step": 274
+    },
+    {
+      "epoch": 1.4341590612777053,
+      "grad_norm": 1.7328843255622977,
+      "learning_rate": 7.173042257104243e-05,
+      "loss": 0.8142,
+      "step": 275
+    },
+    {
+      "epoch": 1.439374185136897,
+      "grad_norm": 1.5984687241828228,
+      "learning_rate": 7.164113674619542e-05,
+      "loss": 0.8062,
+      "step": 276
+    },
+    {
+      "epoch": 1.4445893089960886,
+      "grad_norm": 1.9941450850887774,
+      "learning_rate": 7.155142770249008e-05,
+      "loss": 0.8156,
+      "step": 277
+    },
+    {
+      "epoch": 1.4498044328552804,
+      "grad_norm": 1.2629214490257985,
+      "learning_rate": 7.146129663983775e-05,
+      "loss": 0.8029,
+      "step": 278
+    },
+    {
+      "epoch": 1.455019556714472,
+      "grad_norm": 1.5341743253202347,
+      "learning_rate": 7.137074476379454e-05,
+      "loss": 0.8184,
+      "step": 279
+    },
+    {
+      "epoch": 1.4602346805736637,
+      "grad_norm": 1.205052243506182,
+      "learning_rate": 7.127977328554518e-05,
+      "loss": 0.8297,
+      "step": 280
+    },
+    {
+      "epoch": 1.4654498044328552,
+      "grad_norm": 1.7584530348059961,
+      "learning_rate": 7.118838342188683e-05,
+      "loss": 0.8183,
+      "step": 281
+    },
+    {
+      "epoch": 1.470664928292047,
+      "grad_norm": 1.8223656535526023,
+      "learning_rate": 7.10965763952128e-05,
+      "loss": 0.8154,
+      "step": 282
+    },
+    {
+      "epoch": 1.4758800521512385,
+      "grad_norm": 1.3665387213451412,
+      "learning_rate": 7.100435343349617e-05,
+      "loss": 0.8114,
+      "step": 283
+    },
+    {
+      "epoch": 1.4810951760104303,
+      "grad_norm": 1.499553247270366,
+      "learning_rate": 7.091171577027344e-05,
+      "loss": 0.813,
+      "step": 284
+    },
+    {
+      "epoch": 1.4863102998696218,
+      "grad_norm": 2.1007419745906866,
+      "learning_rate": 7.081866464462798e-05,
+      "loss": 0.8216,
+      "step": 285
+    },
+    {
+      "epoch": 1.4915254237288136,
+      "grad_norm": 1.512999754469945,
+      "learning_rate": 7.072520130117344e-05,
+      "loss": 0.8182,
+      "step": 286
+    },
+    {
+      "epoch": 1.4967405475880051,
+      "grad_norm": 1.6975647047303697,
+      "learning_rate": 7.063132699003716e-05,
+      "loss": 0.8244,
+      "step": 287
+    },
+    {
+      "epoch": 1.5019556714471969,
+      "grad_norm": 1.7957380356125672,
+      "learning_rate": 7.053704296684337e-05,
+      "loss": 0.8162,
+      "step": 288
+    },
+    {
+      "epoch": 1.5071707953063886,
+      "grad_norm": 1.2813149409377043,
+      "learning_rate": 7.044235049269649e-05,
+      "loss": 0.8095,
+      "step": 289
+    },
+    {
+      "epoch": 1.5123859191655802,
+      "grad_norm": 1.56534754361993,
+      "learning_rate": 7.034725083416419e-05,
+      "loss": 0.8258,
+      "step": 290
+    },
+    {
+      "epoch": 1.5176010430247717,
+      "grad_norm": 1.7975943131313281,
+      "learning_rate": 7.025174526326045e-05,
+      "loss": 0.8025,
+      "step": 291
+    },
+    {
+      "epoch": 1.5228161668839635,
+      "grad_norm": 1.1853911402078072,
+      "learning_rate": 7.015583505742857e-05,
+      "loss": 0.8249,
+      "step": 292
+    },
+    {
+      "epoch": 1.5280312907431552,
+      "grad_norm": 1.8494955443278849,
+      "learning_rate": 7.005952149952416e-05,
+      "loss": 0.8378,
+      "step": 293
+    },
+    {
+      "epoch": 1.5332464146023468,
+      "grad_norm": 1.3622849536120247,
+      "learning_rate": 6.996280587779778e-05,
+      "loss": 0.8354,
+      "step": 294
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 2.0112831346634708,
+      "learning_rate": 6.986568948587792e-05,
+      "loss": 0.83,
+      "step": 295
+    },
+    {
+      "epoch": 1.54367666232073,
+      "grad_norm": 1.5420042484474124,
+      "learning_rate": 6.976817362275357e-05,
+      "loss": 0.8109,
+      "step": 296
+    },
+    {
+      "epoch": 1.5488917861799218,
+      "grad_norm": 2.0929864944342804,
+      "learning_rate": 6.96702595927569e-05,
+      "loss": 0.8334,
+      "step": 297
+    },
+    {
+      "epoch": 1.5541069100391134,
+      "grad_norm": 1.8922406476791735,
+      "learning_rate": 6.957194870554578e-05,
+      "loss": 0.806,
+      "step": 298
+    },
+    {
+      "epoch": 1.559322033898305,
+      "grad_norm": 1.6880868554685944,
+      "learning_rate": 6.947324227608628e-05,
+      "loss": 0.8212,
+      "step": 299
+    },
+    {
+      "epoch": 1.5645371577574967,
+      "grad_norm": 1.6346451668072235,
+      "learning_rate": 6.937414162463509e-05,
+      "loss": 0.8014,
+      "step": 300
+    },
+    {
+      "epoch": 1.5697522816166884,
+      "grad_norm": 1.577939054921778,
+      "learning_rate": 6.927464807672186e-05,
+      "loss": 0.8187,
+      "step": 301
+    },
+    {
+      "epoch": 1.57496740547588,
+      "grad_norm": 1.4316029714279948,
+      "learning_rate": 6.917476296313145e-05,
+      "loss": 0.8046,
+      "step": 302
+    },
+    {
+      "epoch": 1.5801825293350718,
+      "grad_norm": 1.5381955264209675,
+      "learning_rate": 6.907448761988612e-05,
+      "loss": 0.8077,
+      "step": 303
+    },
+    {
+      "epoch": 1.5853976531942635,
+      "grad_norm": 1.0983288794814963,
+      "learning_rate": 6.897382338822772e-05,
+      "loss": 0.804,
+      "step": 304
+    },
+    {
+      "epoch": 1.590612777053455,
+      "grad_norm": 1.2952124244327796,
+      "learning_rate": 6.88727716145997e-05,
+      "loss": 0.8014,
+      "step": 305
+    },
+    {
+      "epoch": 1.5958279009126466,
+      "grad_norm": 1.1248306945768556,
+      "learning_rate": 6.877133365062911e-05,
+      "loss": 0.8086,
+      "step": 306
+    },
+    {
+      "epoch": 1.6010430247718384,
+      "grad_norm": 1.7722044591752057,
+      "learning_rate": 6.86695108531085e-05,
+      "loss": 0.8061,
+      "step": 307
+    },
+    {
+      "epoch": 1.6062581486310301,
+      "grad_norm": 1.8332828396584284,
+      "learning_rate": 6.856730458397787e-05,
+      "loss": 0.8177,
+      "step": 308
+    },
+    {
+      "epoch": 1.6114732724902217,
+      "grad_norm": 1.3222660990882034,
+      "learning_rate": 6.846471621030626e-05,
+      "loss": 0.8012,
+      "step": 309
+    },
+    {
+      "epoch": 1.6166883963494132,
+      "grad_norm": 1.3548943859194524,
+      "learning_rate": 6.836174710427369e-05,
+      "loss": 0.8171,
+      "step": 310
+    },
+    {
+      "epoch": 1.621903520208605,
+      "grad_norm": 1.2347347259073755,
+      "learning_rate": 6.825839864315264e-05,
+      "loss": 0.7839,
+      "step": 311
+    },
+    {
+      "epoch": 1.6271186440677967,
+      "grad_norm": 1.4912678731047493,
+      "learning_rate": 6.815467220928972e-05,
+      "loss": 0.8004,
+      "step": 312
+    },
+    {
+      "epoch": 1.6323337679269883,
+      "grad_norm": 1.0089833155389005,
+      "learning_rate": 6.805056919008714e-05,
+      "loss": 0.806,
+      "step": 313
+    },
+    {
+      "epoch": 1.6375488917861798,
+      "grad_norm": 1.8923057463186084,
+      "learning_rate": 6.794609097798414e-05,
+      "loss": 0.8149,
+      "step": 314
+    },
+    {
+      "epoch": 1.6427640156453716,
+      "grad_norm": 1.3911432602736478,
+      "learning_rate": 6.784123897043841e-05,
+      "loss": 0.8261,
+      "step": 315
+    },
+    {
+      "epoch": 1.6479791395045633,
+      "grad_norm": 1.2299562475949994,
+      "learning_rate": 6.773601456990739e-05,
+      "loss": 0.8025,
+      "step": 316
+    },
+    {
+      "epoch": 1.6531942633637549,
+      "grad_norm": 1.749738991866774,
+      "learning_rate": 6.763041918382945e-05,
+      "loss": 0.8087,
+      "step": 317
+    },
+    {
+      "epoch": 1.6584093872229464,
+      "grad_norm": 1.2129819476495074,
+      "learning_rate": 6.752445422460513e-05,
+      "loss": 0.8058,
+      "step": 318
+    },
+    {
+      "epoch": 1.6636245110821382,
+      "grad_norm": 1.9218816457335506,
+      "learning_rate": 6.741812110957823e-05,
+      "loss": 0.8199,
+      "step": 319
+    },
+    {
+      "epoch": 1.66883963494133,
+      "grad_norm": 1.5263095695339857,
+      "learning_rate": 6.731142126101688e-05,
+      "loss": 0.8098,
+      "step": 320
+    },
+    {
+      "epoch": 1.6740547588005215,
+      "grad_norm": 1.7388018103606915,
+      "learning_rate": 6.720435610609443e-05,
+      "loss": 0.7964,
+      "step": 321
+    },
+    {
+      "epoch": 1.6792698826597132,
+      "grad_norm": 1.1143428753117783,
+      "learning_rate": 6.709692707687047e-05,
+      "loss": 0.8026,
+      "step": 322
+    },
+    {
+      "epoch": 1.684485006518905,
+      "grad_norm": 1.4518152074151873,
+      "learning_rate": 6.69891356102716e-05,
+      "loss": 0.8199,
+      "step": 323
+    },
+    {
+      "epoch": 1.6897001303780965,
+      "grad_norm": 1.2302866722544838,
+      "learning_rate": 6.688098314807221e-05,
+      "loss": 0.8116,
+      "step": 324
+    },
+    {
+      "epoch": 1.694915254237288,
+      "grad_norm": 1.355915097493558,
+      "learning_rate": 6.677247113687527e-05,
+      "loss": 0.8184,
+      "step": 325
+    },
+    {
+      "epoch": 1.7001303780964798,
+      "grad_norm": 1.0973851180923702,
+      "learning_rate": 6.666360102809289e-05,
+      "loss": 0.8066,
+      "step": 326
+    },
+    {
+      "epoch": 1.7053455019556716,
+      "grad_norm": 2.034989183783714,
+      "learning_rate": 6.655437427792698e-05,
+      "loss": 0.8068,
+      "step": 327
+    },
+    {
+      "epoch": 1.7105606258148631,
+      "grad_norm": 1.352963702330486,
+      "learning_rate": 6.644479234734971e-05,
+      "loss": 0.8337,
+      "step": 328
+    },
+    {
+      "epoch": 1.7157757496740547,
+      "grad_norm": 1.486986240165971,
+      "learning_rate": 6.6334856702084e-05,
+      "loss": 0.8142,
+      "step": 329
+    },
+    {
+      "epoch": 1.7209908735332464,
+      "grad_norm": 1.2755542839313279,
+      "learning_rate": 6.622456881258392e-05,
+      "loss": 0.8224,
+      "step": 330
+    },
+    {
+      "epoch": 1.7262059973924382,
+      "grad_norm": 1.4619956893213686,
+      "learning_rate": 6.6113930154015e-05,
+      "loss": 0.8077,
+      "step": 331
+    },
+    {
+      "epoch": 1.7314211212516297,
+      "grad_norm": 1.7136751212219052,
+      "learning_rate": 6.600294220623457e-05,
+      "loss": 0.8089,
+      "step": 332
+    },
+    {
+      "epoch": 1.7366362451108213,
+      "grad_norm": 1.1691508723533794,
+      "learning_rate": 6.589160645377181e-05,
+      "loss": 0.8192,
+      "step": 333
+    },
+    {
+      "epoch": 1.741851368970013,
+      "grad_norm": 2.2934106958814544,
+      "learning_rate": 6.57799243858081e-05,
+      "loss": 0.8123,
+      "step": 334
+    },
+    {
+      "epoch": 1.7470664928292048,
+      "grad_norm": 1.3479467101781495,
+      "learning_rate": 6.566789749615691e-05,
+      "loss": 0.8016,
+      "step": 335
+    },
+    {
+      "epoch": 1.7522816166883963,
+      "grad_norm": 1.4715014206564454,
+      "learning_rate": 6.555552728324394e-05,
+      "loss": 0.8135,
+      "step": 336
+    },
+    {
+      "epoch": 1.7574967405475879,
+      "grad_norm": 2.4679525667464155,
+      "learning_rate": 6.544281525008703e-05,
+      "loss": 0.8047,
+      "step": 337
+    },
+    {
+      "epoch": 1.7627118644067796,
+      "grad_norm": 1.2701148601075805,
+      "learning_rate": 6.532976290427611e-05,
+      "loss": 0.8155,
+      "step": 338
+    },
+    {
+      "epoch": 1.7679269882659714,
+      "grad_norm": 2.7926000172133967,
+      "learning_rate": 6.521637175795292e-05,
+      "loss": 0.8114,
+      "step": 339
+    },
+    {
+      "epoch": 1.773142112125163,
+      "grad_norm": 1.8634719780362607,
+      "learning_rate": 6.51026433277909e-05,
+      "loss": 0.8279,
+      "step": 340
+    },
+    {
+      "epoch": 1.7783572359843545,
+      "grad_norm": 2.527061321023331,
+      "learning_rate": 6.498857913497485e-05,
+      "loss": 0.8308,
+      "step": 341
+    },
+    {
+      "epoch": 1.7835723598435462,
+      "grad_norm": 1.7861432214936908,
+      "learning_rate": 6.487418070518063e-05,
+      "loss": 0.8237,
+      "step": 342
+    },
+    {
+      "epoch": 1.788787483702738,
+      "grad_norm": 2.4741757997509812,
+      "learning_rate": 6.475944956855463e-05,
+      "loss": 0.8098,
+      "step": 343
+    },
+    {
+      "epoch": 1.7940026075619295,
+      "grad_norm": 2.5470632776776583,
+      "learning_rate": 6.464438725969348e-05,
+      "loss": 0.8153,
+      "step": 344
+    },
+    {
+      "epoch": 1.7992177314211213,
+      "grad_norm": 1.4497598146752289,
+      "learning_rate": 6.452899531762338e-05,
+      "loss": 0.809,
+      "step": 345
+    },
+    {
+      "epoch": 1.804432855280313,
+      "grad_norm": 1.6671814059233867,
+      "learning_rate": 6.44132752857796e-05,
+      "loss": 0.8206,
+      "step": 346
+    },
+    {
+      "epoch": 1.8096479791395046,
+      "grad_norm": 1.4862285893517284,
+      "learning_rate": 6.429722871198579e-05,
+      "loss": 0.8125,
+      "step": 347
+    },
+    {
+      "epoch": 1.8148631029986961,
+      "grad_norm": 1.4717475694489375,
+      "learning_rate": 6.418085714843328e-05,
+      "loss": 0.8028,
+      "step": 348
+    },
+    {
+      "epoch": 1.820078226857888,
+      "grad_norm": 1.2961734030931862,
+      "learning_rate": 6.406416215166035e-05,
+      "loss": 0.814,
+      "step": 349
+    },
+    {
+      "epoch": 1.8252933507170797,
+      "grad_norm": 1.375416590864907,
+      "learning_rate": 6.394714528253137e-05,
+      "loss": 0.8084,
+      "step": 350
+    },
+    {
+      "epoch": 1.8305084745762712,
+      "grad_norm": 1.1939033449159728,
+      "learning_rate": 6.382980810621595e-05,
+      "loss": 0.8054,
+      "step": 351
+    },
+    {
+      "epoch": 1.8357235984354627,
+      "grad_norm": 1.905725265353936,
+      "learning_rate": 6.371215219216801e-05,
+      "loss": 0.799,
+      "step": 352
+    },
+    {
+      "epoch": 1.8409387222946545,
+      "grad_norm": 1.5504247264478312,
+      "learning_rate": 6.359417911410477e-05,
+      "loss": 0.8184,
+      "step": 353
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 1.034848478726046,
+      "learning_rate": 6.347589044998568e-05,
+      "loss": 0.8058,
+      "step": 354
+    },
+    {
+      "epoch": 1.8513689700130378,
+      "grad_norm": 2.3826958255372217,
+      "learning_rate": 6.335728778199139e-05,
+      "loss": 0.8066,
+      "step": 355
+    },
+    {
+      "epoch": 1.8565840938722293,
+      "grad_norm": 1.3959972285328377,
+      "learning_rate": 6.323837269650249e-05,
+      "loss": 0.8217,
+      "step": 356
+    },
+    {
+      "epoch": 1.861799217731421,
+      "grad_norm": 2.506085020659937,
+      "learning_rate": 6.311914678407837e-05,
+      "loss": 0.8216,
+      "step": 357
+    },
+    {
+      "epoch": 1.8670143415906129,
+      "grad_norm": 1.8279565139453056,
+      "learning_rate": 6.299961163943587e-05,
+      "loss": 0.8213,
+      "step": 358
+    },
+    {
+      "epoch": 1.8722294654498044,
+      "grad_norm": 2.0962030329324204,
+      "learning_rate": 6.287976886142806e-05,
+      "loss": 0.8206,
+      "step": 359
+    },
+    {
+      "epoch": 1.877444589308996,
+      "grad_norm": 1.70186031477118,
+      "learning_rate": 6.275962005302273e-05,
+      "loss": 0.8301,
+      "step": 360
+    },
+    {
+      "epoch": 1.8826597131681877,
+      "grad_norm": 2.1788285828805365,
+      "learning_rate": 6.263916682128104e-05,
+      "loss": 0.817,
+      "step": 361
+    },
+    {
+      "epoch": 1.8878748370273795,
+      "grad_norm": 1.805067718405396,
+      "learning_rate": 6.251841077733595e-05,
+      "loss": 0.8158,
+      "step": 362
+    },
+    {
+      "epoch": 1.893089960886571,
+      "grad_norm": 2.1305468630587066,
+      "learning_rate": 6.239735353637076e-05,
+      "loss": 0.8086,
+      "step": 363
+    },
+    {
+      "epoch": 1.8983050847457628,
+      "grad_norm": 1.8535876445861497,
+      "learning_rate": 6.227599671759745e-05,
+      "loss": 0.8088,
+      "step": 364
+    },
+    {
+      "epoch": 1.9035202086049545,
+      "grad_norm": 2.0851098014276914,
+      "learning_rate": 6.215434194423499e-05,
+      "loss": 0.8053,
+      "step": 365
+    },
+    {
+      "epoch": 1.908735332464146,
+      "grad_norm": 1.6617436320103747,
+      "learning_rate": 6.203239084348772e-05,
+      "loss": 0.8272,
+      "step": 366
+    },
+    {
+      "epoch": 1.9139504563233376,
+      "grad_norm": 1.9519893634609065,
+      "learning_rate": 6.191014504652352e-05,
+      "loss": 0.8052,
+      "step": 367
+    },
+    {
+      "epoch": 1.9191655801825294,
+      "grad_norm": 1.8496089676435563,
+      "learning_rate": 6.178760618845194e-05,
+      "loss": 0.8152,
+      "step": 368
+    },
+    {
+      "epoch": 1.9243807040417211,
+      "grad_norm": 1.5391906529413488,
+      "learning_rate": 6.166477590830252e-05,
+      "loss": 0.8129,
+      "step": 369
+    },
+    {
+      "epoch": 1.9295958279009127,
+      "grad_norm": 1.5353016752039255,
+      "learning_rate": 6.154165584900263e-05,
+      "loss": 0.7994,
+      "step": 370
+    },
+    {
+      "epoch": 1.9348109517601042,
+      "grad_norm": 1.4284130690464092,
+      "learning_rate": 6.141824765735567e-05,
+      "loss": 0.8169,
+      "step": 371
+    },
+    {
+      "epoch": 1.940026075619296,
+      "grad_norm": 1.3150434684077639,
+      "learning_rate": 6.129455298401894e-05,
+      "loss": 0.7936,
+      "step": 372
+    },
+    {
+      "epoch": 1.9452411994784877,
+      "grad_norm": 1.3747947345260623,
+      "learning_rate": 6.117057348348164e-05,
+      "loss": 0.8007,
+      "step": 373
+    },
+    {
+      "epoch": 1.9504563233376793,
+      "grad_norm": 1.0176646734905497,
+      "learning_rate": 6.104631081404269e-05,
+      "loss": 0.8085,
+      "step": 374
+    },
+    {
+      "epoch": 1.9556714471968708,
+      "grad_norm": 1.773293381624066,
+      "learning_rate": 6.092176663778851e-05,
+      "loss": 0.8075,
+      "step": 375
+    },
+    {
+      "epoch": 1.9608865710560626,
+      "grad_norm": 1.4739056885963973,
+      "learning_rate": 6.079694262057094e-05,
+      "loss": 0.8125,
+      "step": 376
+    },
+    {
+      "epoch": 1.9661016949152543,
+      "grad_norm": 1.6264911494594487,
+      "learning_rate": 6.067184043198476e-05,
+      "loss": 0.8093,
+      "step": 377
+    },
+    {
+      "epoch": 1.9713168187744459,
+      "grad_norm": 1.5269696160976443,
+      "learning_rate": 6.054646174534552e-05,
+      "loss": 0.8073,
+      "step": 378
+    },
+    {
+      "epoch": 1.9765319426336374,
+      "grad_norm": 1.4672155013368013,
+      "learning_rate": 6.0420808237667055e-05,
+      "loss": 0.7873,
+      "step": 379
+    },
+    {
+      "epoch": 1.9817470664928292,
+      "grad_norm": 1.6071190706840761,
+      "learning_rate": 6.029488158963912e-05,
+      "loss": 0.8185,
+      "step": 380
+    },
+    {
+      "epoch": 1.986962190352021,
+      "grad_norm": 1.1200471714942668,
+      "learning_rate": 6.016868348560488e-05,
+      "loss": 0.7945,
+      "step": 381
+    },
+    {
+      "epoch": 1.9921773142112125,
+      "grad_norm": 1.2021517615584545,
+      "learning_rate": 6.004221561353838e-05,
+      "loss": 0.8019,
+      "step": 382
+    },
+    {
+      "epoch": 1.997392438070404,
+      "grad_norm": 1.1953523539998745,
+      "learning_rate": 5.991547966502195e-05,
+      "loss": 0.8847,
+      "step": 383
+    },
+    {
+      "epoch": 2.002607561929596,
+      "grad_norm": 1.2139914038103075,
+      "learning_rate": 5.978847733522363e-05,
+      "loss": 0.9192,
+      "step": 384
+    },
+    {
+      "epoch": 2.0078226857887875,
+      "grad_norm": 1.2663014964519022,
+      "learning_rate": 5.9661210322874456e-05,
+      "loss": 0.7934,
+      "step": 385
+    },
+    {
+      "epoch": 2.013037809647979,
+      "grad_norm": 1.7536553812874152,
+      "learning_rate": 5.953368033024576e-05,
+      "loss": 0.7716,
+      "step": 386
+    },
+    {
+      "epoch": 2.0182529335071706,
+      "grad_norm": 1.3505679779762105,
+      "learning_rate": 5.940588906312636e-05,
+      "loss": 0.79,
+      "step": 387
+    },
+    {
+      "epoch": 2.0234680573663626,
+      "grad_norm": 1.100812193500458,
+      "learning_rate": 5.9277838230799816e-05,
+      "loss": 0.7831,
+      "step": 388
+    },
+    {
+      "epoch": 2.028683181225554,
+      "grad_norm": 1.312836314335182,
+      "learning_rate": 5.9149529546021486e-05,
+      "loss": 0.7833,
+      "step": 389
+    },
+    {
+      "epoch": 2.0338983050847457,
+      "grad_norm": 1.9463142850045887,
+      "learning_rate": 5.902096472499569e-05,
+      "loss": 0.7829,
+      "step": 390
+    },
+    {
+      "epoch": 2.039113428943937,
+      "grad_norm": 0.7033565907846064,
+      "learning_rate": 5.889214548735269e-05,
+      "loss": 0.7842,
+      "step": 391
+    },
+    {
+      "epoch": 2.044328552803129,
+      "grad_norm": 1.743510351487419,
+      "learning_rate": 5.876307355612575e-05,
+      "loss": 0.7679,
+      "step": 392
+    },
+    {
+      "epoch": 2.0495436766623207,
+      "grad_norm": 1.2465470038883548,
+      "learning_rate": 5.8633750657728033e-05,
+      "loss": 0.7712,
+      "step": 393
+    },
+    {
+      "epoch": 2.0547588005215123,
+      "grad_norm": 2.068655430465104,
+      "learning_rate": 5.850417852192956e-05,
+      "loss": 0.7853,
+      "step": 394
+    },
+    {
+      "epoch": 2.0599739243807043,
+      "grad_norm": 0.9560005505258112,
+      "learning_rate": 5.837435888183403e-05,
+      "loss": 0.785,
+      "step": 395
+    },
+    {
+      "epoch": 2.065189048239896,
+      "grad_norm": 1.7062918904685223,
+      "learning_rate": 5.8244293473855664e-05,
+      "loss": 0.7849,
+      "step": 396
+    },
+    {
+      "epoch": 2.0704041720990873,
+      "grad_norm": 1.623721715677207,
+      "learning_rate": 5.8113984037695984e-05,
+      "loss": 0.7857,
+      "step": 397
+    },
+    {
+      "epoch": 2.075619295958279,
+      "grad_norm": 1.484064136635307,
+      "learning_rate": 5.798343231632053e-05,
+      "loss": 0.7779,
+      "step": 398
+    },
+    {
+      "epoch": 2.080834419817471,
+      "grad_norm": 1.34676834959028,
+      "learning_rate": 5.785264005593553e-05,
+      "loss": 0.7934,
+      "step": 399
+    },
+    {
+      "epoch": 2.0860495436766624,
+      "grad_norm": 1.6199775628522408,
+      "learning_rate": 5.772160900596456e-05,
+      "loss": 0.7735,
+      "step": 400
+    },
+    {
+      "epoch": 2.091264667535854,
+      "grad_norm": 1.5660844096047937,
+      "learning_rate": 5.7590340919025204e-05,
+      "loss": 0.7833,
+      "step": 401
+    },
+    {
+      "epoch": 2.0964797913950455,
+      "grad_norm": 1.0421277103166564,
+      "learning_rate": 5.7458837550905486e-05,
+      "loss": 0.7837,
+      "step": 402
+    },
+    {
+      "epoch": 2.1016949152542375,
+      "grad_norm": 1.7458400592594279,
+      "learning_rate": 5.73271006605405e-05,
+      "loss": 0.7855,
+      "step": 403
+    },
+    {
+      "epoch": 2.106910039113429,
+      "grad_norm": 1.5186591541100996,
+      "learning_rate": 5.7195132009988814e-05,
+      "loss": 0.7853,
+      "step": 404
+    },
+    {
+      "epoch": 2.1121251629726205,
+      "grad_norm": 0.8719695266664192,
+      "learning_rate": 5.706293336440894e-05,
+      "loss": 0.7896,
+      "step": 405
+    },
+    {
+      "epoch": 2.117340286831812,
+      "grad_norm": 1.0314918237610309,
+      "learning_rate": 5.693050649203568e-05,
+      "loss": 0.7807,
+      "step": 406
+    },
+    {
+      "epoch": 2.122555410691004,
+      "grad_norm": 0.7603723547082539,
+      "learning_rate": 5.679785316415654e-05,
+      "loss": 0.7763,
+      "step": 407
+    },
+    {
+      "epoch": 2.1277705345501956,
+      "grad_norm": 0.9173514911262751,
+      "learning_rate": 5.6664975155088004e-05,
+      "loss": 0.7919,
+      "step": 408
+    },
+    {
+      "epoch": 2.132985658409387,
+      "grad_norm": 1.2215202346945515,
+      "learning_rate": 5.6531874242151746e-05,
+      "loss": 0.772,
+      "step": 409
+    },
+    {
+      "epoch": 2.1382007822685787,
+      "grad_norm": 2.368630198683683,
+      "learning_rate": 5.639855220565098e-05,
+      "loss": 0.801,
+      "step": 410
+    },
+    {
+      "epoch": 2.1434159061277707,
+      "grad_norm": 1.2419888378957813,
+      "learning_rate": 5.626501082884654e-05,
+      "loss": 0.7948,
+      "step": 411
+    },
+    {
+      "epoch": 2.148631029986962,
+      "grad_norm": 3.0547619405266446,
+      "learning_rate": 5.613125189793305e-05,
+      "loss": 0.7761,
+      "step": 412
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 2.1394143129005987,
+      "learning_rate": 5.5997277202015085e-05,
+      "loss": 0.7815,
+      "step": 413
+    },
+    {
+      "epoch": 2.1590612777053453,
+      "grad_norm": 3.041921943681318,
+      "learning_rate": 5.586308853308319e-05,
+      "loss": 0.7889,
+      "step": 414
+    },
+    {
+      "epoch": 2.1642764015645373,
+      "grad_norm": 2.4712686274458853,
+      "learning_rate": 5.572868768598993e-05,
+      "loss": 0.7764,
+      "step": 415
+    },
+    {
+      "epoch": 2.169491525423729,
+      "grad_norm": 2.5501423533072605,
+      "learning_rate": 5.5594076458425864e-05,
+      "loss": 0.7995,
+      "step": 416
+    },
+    {
+      "epoch": 2.1747066492829203,
+      "grad_norm": 2.2877612282195416,
+      "learning_rate": 5.545925665089552e-05,
+      "loss": 0.7882,
+      "step": 417
+    },
+    {
+      "epoch": 2.1799217731421123,
+      "grad_norm": 2.1483606998937144,
+      "learning_rate": 5.532423006669332e-05,
+      "loss": 0.7873,
+      "step": 418
+    },
+    {
+      "epoch": 2.185136897001304,
+      "grad_norm": 1.7462471775177655,
+      "learning_rate": 5.518899851187942e-05,
+      "loss": 0.7964,
+      "step": 419
+    },
+    {
+      "epoch": 2.1903520208604954,
+      "grad_norm": 2.756935578457596,
+      "learning_rate": 5.505356379525559e-05,
+      "loss": 0.7891,
+      "step": 420
+    },
+    {
+      "epoch": 2.195567144719687,
+      "grad_norm": 2.4106523309041314,
+      "learning_rate": 5.491792772834103e-05,
+      "loss": 0.7804,
+      "step": 421
+    },
+    {
+      "epoch": 2.200782268578879,
+      "grad_norm": 1.8981144967918866,
+      "learning_rate": 5.478209212534809e-05,
+      "loss": 0.7796,
+      "step": 422
+    },
+    {
+      "epoch": 2.2059973924380705,
+      "grad_norm": 1.487529019552184,
+      "learning_rate": 5.464605880315803e-05,
+      "loss": 0.7773,
+      "step": 423
+    },
+    {
+      "epoch": 2.211212516297262,
+      "grad_norm": 2.44109555313169,
+      "learning_rate": 5.4509829581296774e-05,
+      "loss": 0.777,
+      "step": 424
+    },
+    {
+      "epoch": 2.2164276401564535,
+      "grad_norm": 1.9157281312835157,
+      "learning_rate": 5.4373406281910434e-05,
+      "loss": 0.7835,
+      "step": 425
+    },
+    {
+      "epoch": 2.2216427640156455,
+      "grad_norm": 2.6110440355014477,
+      "learning_rate": 5.423679072974109e-05,
+      "loss": 0.7792,
+      "step": 426
+    },
+    {
+      "epoch": 2.226857887874837,
+      "grad_norm": 2.3658010170065853,
+      "learning_rate": 5.4099984752102295e-05,
+      "loss": 0.7914,
+      "step": 427
+    },
+    {
+      "epoch": 2.2320730117340286,
+      "grad_norm": 1.768060696859615,
+      "learning_rate": 5.396299017885465e-05,
+      "loss": 0.7789,
+      "step": 428
+    },
+    {
+      "epoch": 2.23728813559322,
+      "grad_norm": 1.661248122612779,
+      "learning_rate": 5.3825808842381345e-05,
+      "loss": 0.7735,
+      "step": 429
+    },
+    {
+      "epoch": 2.242503259452412,
+      "grad_norm": 2.362109109009534,
+      "learning_rate": 5.3688442577563646e-05,
+      "loss": 0.7886,
+      "step": 430
+    },
+    {
+      "epoch": 2.2477183833116037,
+      "grad_norm": 2.0017993267041905,
+      "learning_rate": 5.355089322175629e-05,
+      "loss": 0.7735,
+      "step": 431
+    },
+    {
+      "epoch": 2.252933507170795,
+      "grad_norm": 1.985452281861886,
+      "learning_rate": 5.3413162614763043e-05,
+      "loss": 0.7904,
+      "step": 432
+    },
+    {
+      "epoch": 2.2581486310299868,
+      "grad_norm": 1.8249568105371123,
+      "learning_rate": 5.327525259881196e-05,
+      "loss": 0.7792,
+      "step": 433
+    },
+    {
+      "epoch": 2.2633637548891787,
+      "grad_norm": 1.987099316108934,
+      "learning_rate": 5.3137165018530805e-05,
+      "loss": 0.7718,
+      "step": 434
+    },
+    {
+      "epoch": 2.2685788787483703,
+      "grad_norm": 1.7309972017145459,
+      "learning_rate": 5.299890172092238e-05,
+      "loss": 0.7737,
+      "step": 435
+    },
+    {
+      "epoch": 2.273794002607562,
+      "grad_norm": 2.083040780695706,
+      "learning_rate": 5.286046455533981e-05,
+      "loss": 0.7861,
+      "step": 436
+    },
+    {
+      "epoch": 2.279009126466754,
+      "grad_norm": 1.7833638732253791,
+      "learning_rate": 5.27218553734618e-05,
+      "loss": 0.7688,
+      "step": 437
+    },
+    {
+      "epoch": 2.2842242503259453,
+      "grad_norm": 2.2540602964270997,
+      "learning_rate": 5.2583076029267864e-05,
+      "loss": 0.7868,
+      "step": 438
+    },
+    {
+      "epoch": 2.289439374185137,
+      "grad_norm": 2.1042743671125046,
+      "learning_rate": 5.2444128379013564e-05,
+      "loss": 0.7806,
+      "step": 439
+    },
+    {
+      "epoch": 2.2946544980443284,
+      "grad_norm": 1.638491731440727,
+      "learning_rate": 5.2305014281205634e-05,
+      "loss": 0.7778,
+      "step": 440
+    },
+    {
+      "epoch": 2.2998696219035204,
+      "grad_norm": 1.4490203056402047,
+      "learning_rate": 5.2165735596577146e-05,
+      "loss": 0.7776,
+      "step": 441
+    },
+    {
+      "epoch": 2.305084745762712,
+      "grad_norm": 2.285181885567823,
+      "learning_rate": 5.202629418806263e-05,
+      "loss": 0.7767,
+      "step": 442
+    },
+    {
+      "epoch": 2.3102998696219035,
+      "grad_norm": 1.9464604198183288,
+      "learning_rate": 5.1886691920773116e-05,
+      "loss": 0.7651,
+      "step": 443
+    },
+    {
+      "epoch": 2.315514993481095,
+      "grad_norm": 1.7631237894703213,
+      "learning_rate": 5.174693066197125e-05,
+      "loss": 0.7726,
+      "step": 444
+    },
+    {
+      "epoch": 2.320730117340287,
+      "grad_norm": 1.56838244742276,
+      "learning_rate": 5.160701228104626e-05,
+      "loss": 0.7819,
+      "step": 445
+    },
+    {
+      "epoch": 2.3259452411994785,
+      "grad_norm": 2.202901692990395,
+      "learning_rate": 5.146693864948898e-05,
+      "loss": 0.7856,
+      "step": 446
+    },
+    {
+      "epoch": 2.33116036505867,
+      "grad_norm": 1.92108099189494,
+      "learning_rate": 5.132671164086682e-05,
+      "loss": 0.775,
+      "step": 447
+    },
+    {
+      "epoch": 2.3363754889178616,
+      "grad_norm": 1.760952455824346,
+      "learning_rate": 5.118633313079869e-05,
+      "loss": 0.7807,
+      "step": 448
+    },
+    {
+      "epoch": 2.3415906127770536,
+      "grad_norm": 1.5869940859426193,
+      "learning_rate": 5.104580499692992e-05,
+      "loss": 0.7845,
+      "step": 449
+    },
+    {
+      "epoch": 2.346805736636245,
+      "grad_norm": 2.0260440621389364,
+      "learning_rate": 5.090512911890715e-05,
+      "loss": 0.7857,
+      "step": 450
+    },
+    {
+      "epoch": 2.3520208604954367,
+      "grad_norm": 1.7167308373113537,
+      "learning_rate": 5.076430737835318e-05,
+      "loss": 0.7815,
+      "step": 451
+    },
+    {
+      "epoch": 2.3572359843546282,
+      "grad_norm": 1.881887746447444,
+      "learning_rate": 5.062334165884182e-05,
+      "loss": 0.7811,
+      "step": 452
+    },
+    {
+      "epoch": 2.36245110821382,
+      "grad_norm": 1.6824543517537593,
+      "learning_rate": 5.0482233845872674e-05,
+      "loss": 0.776,
+      "step": 453
+    },
+    {
+      "epoch": 2.3676662320730117,
+      "grad_norm": 1.8755854127797766,
+      "learning_rate": 5.034098582684595e-05,
+      "loss": 0.7803,
+      "step": 454
+    },
+    {
+      "epoch": 2.3728813559322033,
+      "grad_norm": 1.6777197590484048,
+      "learning_rate": 5.019959949103715e-05,
+      "loss": 0.7848,
+      "step": 455
+    },
+    {
+      "epoch": 2.3780964797913953,
+      "grad_norm": 1.81809003546367,
+      "learning_rate": 5.005807672957188e-05,
+      "loss": 0.7858,
+      "step": 456
+    },
+    {
+      "epoch": 2.383311603650587,
+      "grad_norm": 1.6133734346606823,
+      "learning_rate": 4.9916419435400516e-05,
+      "loss": 0.7808,
+      "step": 457
+    },
+    {
+      "epoch": 2.3885267275097783,
+      "grad_norm": 1.8363482997949565,
+      "learning_rate": 4.9774629503272874e-05,
+      "loss": 0.7796,
+      "step": 458
+    },
+    {
+      "epoch": 2.39374185136897,
+      "grad_norm": 1.5541908015103185,
+      "learning_rate": 4.96327088297129e-05,
+      "loss": 0.7771,
+      "step": 459
+    },
+    {
+      "epoch": 2.3989569752281614,
+      "grad_norm": 1.754540259013617,
+      "learning_rate": 4.949065931299328e-05,
+      "loss": 0.7858,
+      "step": 460
+    },
+    {
+      "epoch": 2.4041720990873534,
+      "grad_norm": 1.539083403172259,
+      "learning_rate": 4.934848285311002e-05,
+      "loss": 0.778,
+      "step": 461
+    },
+    {
+      "epoch": 2.409387222946545,
+      "grad_norm": 1.8702666142803812,
+      "learning_rate": 4.920618135175712e-05,
+      "loss": 0.7815,
+      "step": 462
+    },
+    {
+      "epoch": 2.4146023468057365,
+      "grad_norm": 1.644430924102753,
+      "learning_rate": 4.9063756712301036e-05,
+      "loss": 0.773,
+      "step": 463
+    },
+    {
+      "epoch": 2.4198174706649285,
+      "grad_norm": 1.6756469842658643,
+      "learning_rate": 4.8921210839755304e-05,
+      "loss": 0.7737,
+      "step": 464
+    },
+    {
+      "epoch": 2.42503259452412,
+      "grad_norm": 1.4353920458126117,
+      "learning_rate": 4.877854564075499e-05,
+      "loss": 0.7861,
+      "step": 465
+    },
+    {
+      "epoch": 2.4302477183833116,
+      "grad_norm": 1.8834730757505287,
+      "learning_rate": 4.863576302353125e-05,
+      "loss": 0.7638,
+      "step": 466
+    },
+    {
+      "epoch": 2.435462842242503,
+      "grad_norm": 1.6285082262536497,
+      "learning_rate": 4.849286489788579e-05,
+      "loss": 0.7719,
+      "step": 467
+    },
+    {
+      "epoch": 2.440677966101695,
+      "grad_norm": 1.6841805396754381,
+      "learning_rate": 4.834985317516525e-05,
+      "loss": 0.7833,
+      "step": 468
+    },
+    {
+      "epoch": 2.4458930899608866,
+      "grad_norm": 1.5254872325101743,
+      "learning_rate": 4.8206729768235756e-05,
+      "loss": 0.7766,
+      "step": 469
+    },
+    {
+      "epoch": 2.451108213820078,
+      "grad_norm": 1.5946022371035584,
+      "learning_rate": 4.8063496591457256e-05,
+      "loss": 0.7688,
+      "step": 470
+    },
+    {
+      "epoch": 2.4563233376792697,
+      "grad_norm": 1.3044832158013293,
+      "learning_rate": 4.792015556065793e-05,
+      "loss": 0.7745,
+      "step": 471
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 1.822367193607616,
+      "learning_rate": 4.777670859310857e-05,
+      "loss": 0.7839,
+      "step": 472
+    },
+    {
+      "epoch": 2.466753585397653,
+      "grad_norm": 1.5933250932152585,
+      "learning_rate": 4.763315760749695e-05,
+      "loss": 0.7886,
+      "step": 473
+    },
+    {
+      "epoch": 2.4719687092568448,
+      "grad_norm": 1.588436894654484,
+      "learning_rate": 4.748950452390212e-05,
+      "loss": 0.7742,
+      "step": 474
+    },
+    {
+      "epoch": 2.4771838331160367,
+      "grad_norm": 1.4522634410247943,
+      "learning_rate": 4.734575126376876e-05,
+      "loss": 0.7823,
+      "step": 475
+    },
+    {
+      "epoch": 2.4823989569752283,
+      "grad_norm": 1.586237253805528,
+      "learning_rate": 4.7201899749881504e-05,
+      "loss": 0.7926,
+      "step": 476
+    },
+    {
+      "epoch": 2.48761408083442,
+      "grad_norm": 1.3312635236777999,
+      "learning_rate": 4.705795190633915e-05,
+      "loss": 0.786,
+      "step": 477
+    },
+    {
+      "epoch": 2.4928292046936114,
+      "grad_norm": 1.7378087684763845,
+      "learning_rate": 4.691390965852893e-05,
+      "loss": 0.7693,
+      "step": 478
+    },
+    {
+      "epoch": 2.498044328552803,
+      "grad_norm": 1.554126764803909,
+      "learning_rate": 4.676977493310088e-05,
+      "loss": 0.7911,
+      "step": 479
+    },
+    {
+      "epoch": 2.503259452411995,
+      "grad_norm": 1.4897288575191927,
+      "learning_rate": 4.662554965794192e-05,
+      "loss": 0.7809,
+      "step": 480
+    },
+    {
+      "epoch": 2.5084745762711864,
+      "grad_norm": 1.2742943025314304,
+      "learning_rate": 4.648123576215011e-05,
+      "loss": 0.7811,
+      "step": 481
+    },
+    {
+      "epoch": 2.513689700130378,
+      "grad_norm": 1.714780336353614,
+      "learning_rate": 4.633683517600891e-05,
+      "loss": 0.7707,
+      "step": 482
+    },
+    {
+      "epoch": 2.51890482398957,
+      "grad_norm": 1.514593536382248,
+      "learning_rate": 4.61923498309613e-05,
+      "loss": 0.7817,
+      "step": 483
+    },
+    {
+      "epoch": 2.5241199478487615,
+      "grad_norm": 1.4058404125363677,
+      "learning_rate": 4.604778165958392e-05,
+      "loss": 0.7691,
+      "step": 484
+    },
+    {
+      "epoch": 2.529335071707953,
+      "grad_norm": 1.1465871672288677,
+      "learning_rate": 4.590313259556132e-05,
+      "loss": 0.7635,
+      "step": 485
+    },
+    {
+      "epoch": 2.5345501955671446,
+      "grad_norm": 1.6402163570623103,
+      "learning_rate": 4.575840457366001e-05,
+      "loss": 0.7876,
+      "step": 486
+    },
+    {
+      "epoch": 2.539765319426336,
+      "grad_norm": 1.3178513882789409,
+      "learning_rate": 4.561359952970259e-05,
+      "loss": 0.7664,
+      "step": 487
+    },
+    {
+      "epoch": 2.544980443285528,
+      "grad_norm": 1.6022661486448093,
+      "learning_rate": 4.546871940054191e-05,
+      "loss": 0.7905,
+      "step": 488
+    },
+    {
+      "epoch": 2.5501955671447196,
+      "grad_norm": 1.4090335919146642,
+      "learning_rate": 4.5323766124035115e-05,
+      "loss": 0.7819,
+      "step": 489
+    },
+    {
+      "epoch": 2.555410691003911,
+      "grad_norm": 1.324862817997592,
+      "learning_rate": 4.5178741639017736e-05,
+      "loss": 0.7787,
+      "step": 490
+    },
+    {
+      "epoch": 2.560625814863103,
+      "grad_norm": 1.1347550661741326,
+      "learning_rate": 4.503364788527774e-05,
+      "loss": 0.7831,
+      "step": 491
+    },
+    {
+      "epoch": 2.5658409387222947,
+      "grad_norm": 1.5760247018602986,
+      "learning_rate": 4.488848680352965e-05,
+      "loss": 0.7796,
+      "step": 492
+    },
+    {
+      "epoch": 2.5710560625814862,
+      "grad_norm": 1.3090981394784642,
+      "learning_rate": 4.4743260335388516e-05,
+      "loss": 0.7724,
+      "step": 493
+    },
+    {
+      "epoch": 2.576271186440678,
+      "grad_norm": 1.3548998817619167,
+      "learning_rate": 4.459797042334395e-05,
+      "loss": 0.781,
+      "step": 494
+    },
+    {
+      "epoch": 2.5814863102998697,
+      "grad_norm": 1.2234336308095453,
+      "learning_rate": 4.445261901073419e-05,
+      "loss": 0.7743,
+      "step": 495
+    },
+    {
+      "epoch": 2.5867014341590613,
+      "grad_norm": 1.2594794059995964,
+      "learning_rate": 4.430720804172008e-05,
+      "loss": 0.7878,
+      "step": 496
+    },
+    {
+      "epoch": 2.591916558018253,
+      "grad_norm": 1.0475475763925866,
+      "learning_rate": 4.416173946125906e-05,
+      "loss": 0.7851,
+      "step": 497
+    },
+    {
+      "epoch": 2.5971316818774444,
+      "grad_norm": 1.3034377416715606,
+      "learning_rate": 4.401621521507914e-05,
+      "loss": 0.7661,
+      "step": 498
+    },
+    {
+      "epoch": 2.6023468057366363,
+      "grad_norm": 1.0178177147229814,
+      "learning_rate": 4.387063724965292e-05,
+      "loss": 0.768,
+      "step": 499
+    },
+    {
+      "epoch": 2.607561929595828,
+      "grad_norm": 1.4358534072422087,
+      "learning_rate": 4.372500751217153e-05,
+      "loss": 0.7886,
+      "step": 500
+    },
+    {
+      "epoch": 2.6127770534550194,
+      "grad_norm": 1.150853401699166,
+      "learning_rate": 4.357932795051852e-05,
+      "loss": 0.7934,
+      "step": 501
+    },
+    {
+      "epoch": 2.6179921773142114,
+      "grad_norm": 1.1810592463532203,
+      "learning_rate": 4.3433600513243965e-05,
+      "loss": 0.7563,
+      "step": 502
+    },
+    {
+      "epoch": 2.623207301173403,
+      "grad_norm": 1.0034045366911863,
+      "learning_rate": 4.328782714953823e-05,
+      "loss": 0.79,
+      "step": 503
+    },
+    {
+      "epoch": 2.6284224250325945,
+      "grad_norm": 1.0980643139376323,
+      "learning_rate": 4.3142009809205986e-05,
+      "loss": 0.7861,
+      "step": 504
+    },
+    {
+      "epoch": 2.633637548891786,
+      "grad_norm": 0.8881990742171157,
+      "learning_rate": 4.2996150442640163e-05,
+      "loss": 0.7758,
+      "step": 505
+    },
+    {
+      "epoch": 2.6388526727509776,
+      "grad_norm": 0.8675797674117819,
+      "learning_rate": 4.2850251000795735e-05,
+      "loss": 0.7801,
+      "step": 506
+    },
+    {
+      "epoch": 2.6440677966101696,
+      "grad_norm": 0.7883090264802906,
+      "learning_rate": 4.270431343516379e-05,
+      "loss": 0.7745,
+      "step": 507
+    },
+    {
+      "epoch": 2.649282920469361,
+      "grad_norm": 0.7583525803187019,
+      "learning_rate": 4.2558339697745297e-05,
+      "loss": 0.7827,
+      "step": 508
+    },
+    {
+      "epoch": 2.6544980443285526,
+      "grad_norm": 0.7762877342404452,
+      "learning_rate": 4.2412331741025045e-05,
+      "loss": 0.7967,
+      "step": 509
+    },
+    {
+      "epoch": 2.6597131681877446,
+      "grad_norm": 0.7014761796032362,
+      "learning_rate": 4.226629151794555e-05,
+      "loss": 0.7858,
+      "step": 510
+    },
+    {
+      "epoch": 2.664928292046936,
+      "grad_norm": 0.7396414732409832,
+      "learning_rate": 4.2120220981880875e-05,
+      "loss": 0.7794,
+      "step": 511
+    },
+    {
+      "epoch": 2.6701434159061277,
+      "grad_norm": 0.6998054830397707,
+      "learning_rate": 4.197412208661058e-05,
+      "loss": 0.7777,
+      "step": 512
+    },
+    {
+      "epoch": 2.6753585397653197,
+      "grad_norm": 0.6386491344974894,
+      "learning_rate": 4.182799678629351e-05,
+      "loss": 0.7823,
+      "step": 513
+    },
+    {
+      "epoch": 2.680573663624511,
+      "grad_norm": 0.5877328765456284,
+      "learning_rate": 4.168184703544171e-05,
+      "loss": 0.7744,
+      "step": 514
+    },
+    {
+      "epoch": 2.6857887874837028,
+      "grad_norm": 0.5920983354896419,
+      "learning_rate": 4.153567478889426e-05,
+      "loss": 0.7769,
+      "step": 515
+    },
+    {
+      "epoch": 2.6910039113428943,
+      "grad_norm": 0.4867319202645289,
+      "learning_rate": 4.138948200179115e-05,
+      "loss": 0.7807,
+      "step": 516
+    },
+    {
+      "epoch": 2.696219035202086,
+      "grad_norm": 0.5746181168372155,
+      "learning_rate": 4.124327062954707e-05,
+      "loss": 0.7708,
+      "step": 517
+    },
+    {
+      "epoch": 2.701434159061278,
+      "grad_norm": 0.35394623638423245,
+      "learning_rate": 4.1097042627825325e-05,
+      "loss": 0.7823,
+      "step": 518
+    },
+    {
+      "epoch": 2.7066492829204694,
+      "grad_norm": 0.6245548216599335,
+      "learning_rate": 4.095079995251168e-05,
+      "loss": 0.784,
+      "step": 519
+    },
+    {
+      "epoch": 2.711864406779661,
+      "grad_norm": 0.3951096697035697,
+      "learning_rate": 4.080454455968812e-05,
+      "loss": 0.7711,
+      "step": 520
+    },
+    {
+      "epoch": 2.717079530638853,
+      "grad_norm": 0.5759028178578062,
+      "learning_rate": 4.065827840560674e-05,
+      "loss": 0.7773,
+      "step": 521
+    },
+    {
+      "epoch": 2.7222946544980444,
+      "grad_norm": 0.39680693057913397,
+      "learning_rate": 4.0512003446663576e-05,
+      "loss": 0.7736,
+      "step": 522
+    },
+    {
+      "epoch": 2.727509778357236,
+      "grad_norm": 0.49858076462377626,
+      "learning_rate": 4.0365721639372476e-05,
+      "loss": 0.7884,
+      "step": 523
+    },
+    {
+      "epoch": 2.7327249022164275,
+      "grad_norm": 0.4117350165519155,
+      "learning_rate": 4.021943494033882e-05,
+      "loss": 0.7707,
+      "step": 524
+    },
+    {
+      "epoch": 2.737940026075619,
+      "grad_norm": 0.4515042047115184,
+      "learning_rate": 4.007314530623348e-05,
+      "loss": 0.7778,
+      "step": 525
+    },
+    {
+      "epoch": 2.743155149934811,
+      "grad_norm": 0.42547404001436967,
+      "learning_rate": 3.9926854693766536e-05,
+      "loss": 0.7688,
+      "step": 526
+    },
+    {
+      "epoch": 2.7483702737940026,
+      "grad_norm": 0.35069455268010413,
+      "learning_rate": 3.978056505966119e-05,
+      "loss": 0.7744,
+      "step": 527
+    },
+    {
+      "epoch": 2.753585397653194,
+      "grad_norm": 0.4444789741021684,
+      "learning_rate": 3.963427836062753e-05,
+      "loss": 0.775,
+      "step": 528
+    },
+    {
+      "epoch": 2.758800521512386,
+      "grad_norm": 0.43329608911430956,
+      "learning_rate": 3.948799655333644e-05,
+      "loss": 0.7748,
+      "step": 529
+    },
+    {
+      "epoch": 2.7640156453715776,
+      "grad_norm": 0.4209440665741905,
+      "learning_rate": 3.9341721594393276e-05,
+      "loss": 0.7762,
+      "step": 530
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.4322531684473173,
+      "learning_rate": 3.919545544031189e-05,
+      "loss": 0.7878,
+      "step": 531
+    },
+    {
+      "epoch": 2.774445893089961,
+      "grad_norm": 0.4720335982176997,
+      "learning_rate": 3.904920004748833e-05,
+      "loss": 0.777,
+      "step": 532
+    },
+    {
+      "epoch": 2.7796610169491527,
+      "grad_norm": 0.37675833736688985,
+      "learning_rate": 3.8902957372174675e-05,
+      "loss": 0.7728,
+      "step": 533
+    },
+    {
+      "epoch": 2.7848761408083442,
+      "grad_norm": 0.3961055004093581,
+      "learning_rate": 3.8756729370452936e-05,
+      "loss": 0.7745,
+      "step": 534
+    },
+    {
+      "epoch": 2.7900912646675358,
+      "grad_norm": 0.39787870628829797,
+      "learning_rate": 3.8610517998208866e-05,
+      "loss": 0.779,
+      "step": 535
+    },
+    {
+      "epoch": 2.7953063885267273,
+      "grad_norm": 0.366576904221954,
+      "learning_rate": 3.846432521110574e-05,
+      "loss": 0.7823,
+      "step": 536
+    },
+    {
+      "epoch": 2.8005215123859193,
+      "grad_norm": 0.39669396154921643,
+      "learning_rate": 3.831815296455829e-05,
+      "loss": 0.7776,
+      "step": 537
+    },
+    {
+      "epoch": 2.805736636245111,
+      "grad_norm": 0.5485537263803792,
+      "learning_rate": 3.817200321370649e-05,
+      "loss": 0.7821,
+      "step": 538
+    },
+    {
+      "epoch": 2.8109517601043024,
+      "grad_norm": 0.5046596554915849,
+      "learning_rate": 3.802587791338943e-05,
+      "loss": 0.757,
+      "step": 539
+    },
+    {
+      "epoch": 2.8161668839634943,
+      "grad_norm": 0.4297088800078496,
+      "learning_rate": 3.787977901811914e-05,
+      "loss": 0.7819,
+      "step": 540
+    },
+    {
+      "epoch": 2.821382007822686,
+      "grad_norm": 0.2792196466948258,
+      "learning_rate": 3.7733708482054476e-05,
+      "loss": 0.7908,
+      "step": 541
+    },
+    {
+      "epoch": 2.8265971316818774,
+      "grad_norm": 0.3389535808547031,
+      "learning_rate": 3.758766825897497e-05,
+      "loss": 0.7789,
+      "step": 542
+    },
+    {
+      "epoch": 2.831812255541069,
+      "grad_norm": 0.33504887483040935,
+      "learning_rate": 3.7441660302254724e-05,
+      "loss": 0.7641,
+      "step": 543
+    },
+    {
+      "epoch": 2.8370273794002605,
+      "grad_norm": 0.35811016749224023,
+      "learning_rate": 3.729568656483623e-05,
+      "loss": 0.7711,
+      "step": 544
+    },
+    {
+      "epoch": 2.8422425032594525,
+      "grad_norm": 0.3304033008238814,
+      "learning_rate": 3.714974899920428e-05,
+      "loss": 0.781,
+      "step": 545
+    },
+    {
+      "epoch": 2.847457627118644,
+      "grad_norm": 0.42274779776394655,
+      "learning_rate": 3.7003849557359863e-05,
+      "loss": 0.7738,
+      "step": 546
+    },
+    {
+      "epoch": 2.8526727509778356,
+      "grad_norm": 0.33608657379975826,
+      "learning_rate": 3.685799019079402e-05,
+      "loss": 0.7858,
+      "step": 547
+    },
+    {
+      "epoch": 2.8578878748370276,
+      "grad_norm": 0.4402897321734762,
+      "learning_rate": 3.6712172850461785e-05,
+      "loss": 0.7679,
+      "step": 548
+    },
+    {
+      "epoch": 2.863102998696219,
+      "grad_norm": 0.29934112633294296,
+      "learning_rate": 3.6566399486756055e-05,
+      "loss": 0.7889,
+      "step": 549
+    },
+    {
+      "epoch": 2.8683181225554106,
+      "grad_norm": 0.38015963993433893,
+      "learning_rate": 3.642067204948149e-05,
+      "loss": 0.7695,
+      "step": 550
+    },
+    {
+      "epoch": 2.8735332464146026,
+      "grad_norm": 0.38644710923105946,
+      "learning_rate": 3.627499248782849e-05,
+      "loss": 0.7745,
+      "step": 551
+    },
+    {
+      "epoch": 2.878748370273794,
+      "grad_norm": 0.44283035266374077,
+      "learning_rate": 3.612936275034709e-05,
+      "loss": 0.7829,
+      "step": 552
+    },
+    {
+      "epoch": 2.8839634941329857,
+      "grad_norm": 0.426691177545099,
+      "learning_rate": 3.598378478492087e-05,
+      "loss": 0.7775,
+      "step": 553
+    },
+    {
+      "epoch": 2.8891786179921772,
+      "grad_norm": 0.41202751066509746,
+      "learning_rate": 3.583826053874096e-05,
+      "loss": 0.7792,
+      "step": 554
+    },
+    {
+      "epoch": 2.8943937418513688,
+      "grad_norm": 0.40337608056255875,
+      "learning_rate": 3.5692791958279924e-05,
+      "loss": 0.7737,
+      "step": 555
+    },
+    {
+      "epoch": 2.8996088657105608,
+      "grad_norm": 0.42109772590116307,
+      "learning_rate": 3.554738098926582e-05,
+      "loss": 0.7757,
+      "step": 556
+    },
+    {
+      "epoch": 2.9048239895697523,
+      "grad_norm": 0.6458006850893471,
+      "learning_rate": 3.540202957665606e-05,
+      "loss": 0.7775,
+      "step": 557
+    },
+    {
+      "epoch": 2.910039113428944,
+      "grad_norm": 0.5129008593035667,
+      "learning_rate": 3.525673966461149e-05,
+      "loss": 0.7793,
+      "step": 558
+    },
+    {
+      "epoch": 2.915254237288136,
+      "grad_norm": 0.37986325064356524,
+      "learning_rate": 3.5111513196470356e-05,
+      "loss": 0.7706,
+      "step": 559
+    },
+    {
+      "epoch": 2.9204693611473274,
+      "grad_norm": 0.3127213742675476,
+      "learning_rate": 3.4966352114722264e-05,
+      "loss": 0.779,
+      "step": 560
+    },
+    {
+      "epoch": 2.925684485006519,
+      "grad_norm": 0.44150964872249004,
+      "learning_rate": 3.482125836098227e-05,
+      "loss": 0.7726,
+      "step": 561
+    },
+    {
+      "epoch": 2.9308996088657104,
+      "grad_norm": 0.37918325106659184,
+      "learning_rate": 3.46762338759649e-05,
+      "loss": 0.7763,
+      "step": 562
+    },
+    {
+      "epoch": 2.936114732724902,
+      "grad_norm": 0.34801331615309594,
+      "learning_rate": 3.4531280599458096e-05,
+      "loss": 0.7792,
+      "step": 563
+    },
+    {
+      "epoch": 2.941329856584094,
+      "grad_norm": 0.3417071688239163,
+      "learning_rate": 3.438640047029741e-05,
+      "loss": 0.783,
+      "step": 564
+    },
+    {
+      "epoch": 2.9465449804432855,
+      "grad_norm": 0.4027724323607999,
+      "learning_rate": 3.424159542634001e-05,
+      "loss": 0.7799,
+      "step": 565
+    },
+    {
+      "epoch": 2.951760104302477,
+      "grad_norm": 0.359846745274617,
+      "learning_rate": 3.409686740443869e-05,
+      "loss": 0.7874,
+      "step": 566
+    },
+    {
+      "epoch": 2.956975228161669,
+      "grad_norm": 0.39877444581949706,
+      "learning_rate": 3.395221834041608e-05,
+      "loss": 0.7796,
+      "step": 567
+    },
+    {
+      "epoch": 2.9621903520208606,
+      "grad_norm": 0.31627637641393286,
+      "learning_rate": 3.3807650169038714e-05,
+      "loss": 0.7828,
+      "step": 568
+    },
+    {
+      "epoch": 2.967405475880052,
+      "grad_norm": 0.2915329688447508,
+      "learning_rate": 3.366316482399111e-05,
+      "loss": 0.7611,
+      "step": 569
+    },
+    {
+      "epoch": 2.9726205997392436,
+      "grad_norm": 0.3003806981610255,
+      "learning_rate": 3.351876423784991e-05,
+      "loss": 0.7753,
+      "step": 570
+    },
+    {
+      "epoch": 2.9778357235984356,
+      "grad_norm": 0.32825206856441885,
+      "learning_rate": 3.33744503420581e-05,
+      "loss": 0.7805,
+      "step": 571
+    },
+    {
+      "epoch": 2.983050847457627,
+      "grad_norm": 0.28811623986201,
+      "learning_rate": 3.323022506689913e-05,
+      "loss": 0.7787,
+      "step": 572
+    },
+    {
+      "epoch": 2.9882659713168187,
+      "grad_norm": 0.3007386340259681,
+      "learning_rate": 3.308609034147109e-05,
+      "loss": 0.7693,
+      "step": 573
+    },
+    {
+      "epoch": 2.9934810951760102,
+      "grad_norm": 0.28601621207061495,
+      "learning_rate": 3.294204809366088e-05,
+      "loss": 0.7811,
+      "step": 574
+    },
+    {
+      "epoch": 2.9986962190352022,
+      "grad_norm": 0.38527843914278115,
+      "learning_rate": 3.27981002501185e-05,
+      "loss": 0.8962,
+      "step": 575
+    },
+    {
+      "epoch": 3.0039113428943938,
+      "grad_norm": 0.3891353205481252,
+      "learning_rate": 3.265424873623125e-05,
+      "loss": 0.8279,
+      "step": 576
+    },
+    {
+      "epoch": 3.0091264667535853,
+      "grad_norm": 0.34677276173333527,
+      "learning_rate": 3.251049547609789e-05,
+      "loss": 0.7617,
+      "step": 577
+    },
+    {
+      "epoch": 3.014341590612777,
+      "grad_norm": 0.3887463573130963,
+      "learning_rate": 3.2366842392503065e-05,
+      "loss": 0.7404,
+      "step": 578
+    },
+    {
+      "epoch": 3.019556714471969,
+      "grad_norm": 0.45850333580212727,
+      "learning_rate": 3.222329140689144e-05,
+      "loss": 0.7509,
+      "step": 579
+    },
+    {
+      "epoch": 3.0247718383311604,
+      "grad_norm": 0.42139454484056155,
+      "learning_rate": 3.207984443934208e-05,
+      "loss": 0.7474,
+      "step": 580
+    },
+    {
+      "epoch": 3.029986962190352,
+      "grad_norm": 0.32213156347235194,
+      "learning_rate": 3.193650340854275e-05,
+      "loss": 0.7449,
+      "step": 581
+    },
+    {
+      "epoch": 3.035202086049544,
+      "grad_norm": 0.2882489783740716,
+      "learning_rate": 3.179327023176426e-05,
+      "loss": 0.7565,
+      "step": 582
+    },
+    {
+      "epoch": 3.0404172099087354,
+      "grad_norm": 0.31398533972182546,
+      "learning_rate": 3.1650146824834765e-05,
+      "loss": 0.7663,
+      "step": 583
+    },
+    {
+      "epoch": 3.045632333767927,
+      "grad_norm": 0.42483377951795925,
+      "learning_rate": 3.1507135102114224e-05,
+      "loss": 0.7524,
+      "step": 584
+    },
+    {
+      "epoch": 3.0508474576271185,
+      "grad_norm": 0.4203137878607459,
+      "learning_rate": 3.1364236976468755e-05,
+      "loss": 0.7476,
+      "step": 585
+    },
+    {
+      "epoch": 3.0560625814863105,
+      "grad_norm": 0.3536224112892052,
+      "learning_rate": 3.122145435924502e-05,
+      "loss": 0.7479,
+      "step": 586
+    },
+    {
+      "epoch": 3.061277705345502,
+      "grad_norm": 0.3355710189906583,
+      "learning_rate": 3.107878916024471e-05,
+      "loss": 0.7475,
+      "step": 587
+    },
+    {
+      "epoch": 3.0664928292046936,
+      "grad_norm": 0.3578880245402906,
+      "learning_rate": 3.093624328769898e-05,
+      "loss": 0.7542,
+      "step": 588
+    },
+    {
+      "epoch": 3.071707953063885,
+      "grad_norm": 0.3807896298844658,
+      "learning_rate": 3.079381864824289e-05,
+      "loss": 0.7617,
+      "step": 589
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.37153630367778145,
+      "learning_rate": 3.065151714688998e-05,
+      "loss": 0.7581,
+      "step": 590
+    },
+    {
+      "epoch": 3.0821382007822686,
+      "grad_norm": 0.41123425605078595,
+      "learning_rate": 3.0509340687006735e-05,
+      "loss": 0.7538,
+      "step": 591
+    },
+    {
+      "epoch": 3.08735332464146,
+      "grad_norm": 0.33981481049719536,
+      "learning_rate": 3.0367291170287102e-05,
+      "loss": 0.7524,
+      "step": 592
+    },
+    {
+      "epoch": 3.0925684485006517,
+      "grad_norm": 0.3651990332335353,
+      "learning_rate": 3.0225370496727125e-05,
+      "loss": 0.7584,
+      "step": 593
+    },
+    {
+      "epoch": 3.0977835723598437,
+      "grad_norm": 0.2918824973106232,
+      "learning_rate": 3.0083580564599484e-05,
+      "loss": 0.7563,
+      "step": 594
+    },
+    {
+      "epoch": 3.1029986962190352,
+      "grad_norm": 0.3423081259716029,
+      "learning_rate": 2.9941923270428125e-05,
+      "loss": 0.7471,
+      "step": 595
+    },
+    {
+      "epoch": 3.1082138200782268,
+      "grad_norm": 0.3604784090698713,
+      "learning_rate": 2.9800400508962854e-05,
+      "loss": 0.7458,
+      "step": 596
+    },
+    {
+      "epoch": 3.1134289439374183,
+      "grad_norm": 0.44107810877163156,
+      "learning_rate": 2.9659014173154053e-05,
+      "loss": 0.761,
+      "step": 597
+    },
+    {
+      "epoch": 3.1186440677966103,
+      "grad_norm": 0.36068201923449394,
+      "learning_rate": 2.9517766154127332e-05,
+      "loss": 0.7495,
+      "step": 598
+    },
+    {
+      "epoch": 3.123859191655802,
+      "grad_norm": 0.32775307660721575,
+      "learning_rate": 2.93766583411582e-05,
+      "loss": 0.7463,
+      "step": 599
+    },
+    {
+      "epoch": 3.1290743155149934,
+      "grad_norm": 0.4150847775843986,
+      "learning_rate": 2.923569262164684e-05,
+      "loss": 0.7459,
+      "step": 600
+    },
+    {
+      "epoch": 3.134289439374185,
+      "grad_norm": 0.3453247625813097,
+      "learning_rate": 2.909487088109287e-05,
+      "loss": 0.756,
+      "step": 601
+    },
+    {
+      "epoch": 3.139504563233377,
+      "grad_norm": 0.3884807540864722,
+      "learning_rate": 2.8954195003070106e-05,
+      "loss": 0.7469,
+      "step": 602
+    },
+    {
+      "epoch": 3.1447196870925684,
+      "grad_norm": 0.32406875922905615,
+      "learning_rate": 2.8813666869201323e-05,
+      "loss": 0.7492,
+      "step": 603
+    },
+    {
+      "epoch": 3.14993481095176,
+      "grad_norm": 0.3124660936977556,
+      "learning_rate": 2.867328835913319e-05,
+      "loss": 0.7412,
+      "step": 604
+    },
+    {
+      "epoch": 3.155149934810952,
+      "grad_norm": 0.402564278458608,
+      "learning_rate": 2.853306135051103e-05,
+      "loss": 0.7628,
+      "step": 605
+    },
+    {
+      "epoch": 3.1603650586701435,
+      "grad_norm": 0.3914753370029631,
+      "learning_rate": 2.8392987718953748e-05,
+      "loss": 0.7467,
+      "step": 606
+    },
+    {
+      "epoch": 3.165580182529335,
+      "grad_norm": 0.36094618715771837,
+      "learning_rate": 2.8253069338028756e-05,
+      "loss": 0.7528,
+      "step": 607
+    },
+    {
+      "epoch": 3.1707953063885266,
+      "grad_norm": 0.4160569360480524,
+      "learning_rate": 2.8113308079226897e-05,
+      "loss": 0.753,
+      "step": 608
+    },
+    {
+      "epoch": 3.1760104302477186,
+      "grad_norm": 0.36846901981179736,
+      "learning_rate": 2.7973705811937387e-05,
+      "loss": 0.762,
+      "step": 609
+    },
+    {
+      "epoch": 3.18122555410691,
+      "grad_norm": 0.3181951199229672,
+      "learning_rate": 2.7834264403422864e-05,
+      "loss": 0.7581,
+      "step": 610
+    },
+    {
+      "epoch": 3.1864406779661016,
+      "grad_norm": 0.28604043043751265,
+      "learning_rate": 2.769498571879438e-05,
+      "loss": 0.7591,
+      "step": 611
+    },
+    {
+      "epoch": 3.191655801825293,
+      "grad_norm": 0.2634604899316907,
+      "learning_rate": 2.7555871620986453e-05,
+      "loss": 0.7554,
+      "step": 612
+    },
+    {
+      "epoch": 3.196870925684485,
+      "grad_norm": 0.3039718480846403,
+      "learning_rate": 2.7416923970732146e-05,
+      "loss": 0.7588,
+      "step": 613
+    },
+    {
+      "epoch": 3.2020860495436767,
+      "grad_norm": 0.288103576580876,
+      "learning_rate": 2.7278144626538212e-05,
+      "loss": 0.7487,
+      "step": 614
+    },
+    {
+      "epoch": 3.2073011734028682,
+      "grad_norm": 0.32239905156935256,
+      "learning_rate": 2.71395354446602e-05,
+      "loss": 0.7459,
+      "step": 615
+    },
+    {
+      "epoch": 3.21251629726206,
+      "grad_norm": 0.29806440969996567,
+      "learning_rate": 2.7001098279077623e-05,
+      "loss": 0.7548,
+      "step": 616
+    },
+    {
+      "epoch": 3.2177314211212518,
+      "grad_norm": 0.25998375637060805,
+      "learning_rate": 2.6862834981469198e-05,
+      "loss": 0.7486,
+      "step": 617
+    },
+    {
+      "epoch": 3.2229465449804433,
+      "grad_norm": 0.39237992094856566,
+      "learning_rate": 2.6724747401188055e-05,
+      "loss": 0.7532,
+      "step": 618
+    },
+    {
+      "epoch": 3.228161668839635,
+      "grad_norm": 0.2548892235121485,
+      "learning_rate": 2.6586837385236966e-05,
+      "loss": 0.7484,
+      "step": 619
+    },
+    {
+      "epoch": 3.2333767926988264,
+      "grad_norm": 0.3225804950825967,
+      "learning_rate": 2.6449106778243708e-05,
+      "loss": 0.7529,
+      "step": 620
+    },
+    {
+      "epoch": 3.2385919165580184,
+      "grad_norm": 0.45928173129349054,
+      "learning_rate": 2.631155742243637e-05,
+      "loss": 0.7549,
+      "step": 621
+    },
+    {
+      "epoch": 3.24380704041721,
+      "grad_norm": 0.3308854647358001,
+      "learning_rate": 2.6174191157618652e-05,
+      "loss": 0.7554,
+      "step": 622
+    },
+    {
+      "epoch": 3.2490221642764014,
+      "grad_norm": 0.2795489427507208,
+      "learning_rate": 2.6037009821145346e-05,
+      "loss": 0.7593,
+      "step": 623
+    },
+    {
+      "epoch": 3.2542372881355934,
+      "grad_norm": 0.4430769575631773,
+      "learning_rate": 2.5900015247897714e-05,
+      "loss": 0.7545,
+      "step": 624
+    },
+    {
+      "epoch": 3.259452411994785,
+      "grad_norm": 0.33341725027604363,
+      "learning_rate": 2.576320927025892e-05,
+      "loss": 0.753,
+      "step": 625
+    },
+    {
+      "epoch": 3.2646675358539765,
+      "grad_norm": 0.33122719602795536,
+      "learning_rate": 2.562659371808959e-05,
+      "loss": 0.7436,
+      "step": 626
+    },
+    {
+      "epoch": 3.269882659713168,
+      "grad_norm": 0.3963689954079041,
+      "learning_rate": 2.549017041870325e-05,
+      "loss": 0.7676,
+      "step": 627
+    },
+    {
+      "epoch": 3.27509778357236,
+      "grad_norm": 0.3538575496536462,
+      "learning_rate": 2.535394119684197e-05,
+      "loss": 0.7548,
+      "step": 628
+    },
+    {
+      "epoch": 3.2803129074315516,
+      "grad_norm": 0.24662943382751912,
+      "learning_rate": 2.521790787465193e-05,
+      "loss": 0.7517,
+      "step": 629
+    },
+    {
+      "epoch": 3.285528031290743,
+      "grad_norm": 0.39858795017503024,
+      "learning_rate": 2.5082072271658982e-05,
+      "loss": 0.7482,
+      "step": 630
+    },
+    {
+      "epoch": 3.2907431551499347,
+      "grad_norm": 0.24293852218374687,
+      "learning_rate": 2.494643620474442e-05,
+      "loss": 0.7445,
+      "step": 631
+    },
+    {
+      "epoch": 3.2959582790091266,
+      "grad_norm": 0.3499814647894016,
+      "learning_rate": 2.4811001488120598e-05,
+      "loss": 0.7555,
+      "step": 632
+    },
+    {
+      "epoch": 3.301173402868318,
+      "grad_norm": 0.27161500960796325,
+      "learning_rate": 2.4675769933306696e-05,
+      "loss": 0.7568,
+      "step": 633
+    },
+    {
+      "epoch": 3.3063885267275097,
+      "grad_norm": 0.2881787773315299,
+      "learning_rate": 2.4540743349104485e-05,
+      "loss": 0.7442,
+      "step": 634
+    },
+    {
+      "epoch": 3.3116036505867013,
+      "grad_norm": 0.2740026111414901,
+      "learning_rate": 2.440592354157415e-05,
+      "loss": 0.7579,
+      "step": 635
+    },
+    {
+      "epoch": 3.3168187744458932,
+      "grad_norm": 0.2605377452426752,
+      "learning_rate": 2.427131231401008e-05,
+      "loss": 0.7518,
+      "step": 636
+    },
+    {
+      "epoch": 3.3220338983050848,
+      "grad_norm": 0.2482525542273476,
+      "learning_rate": 2.413691146691681e-05,
+      "loss": 0.7615,
+      "step": 637
+    },
+    {
+      "epoch": 3.3272490221642763,
+      "grad_norm": 0.33206747294953354,
+      "learning_rate": 2.400272279798492e-05,
+      "loss": 0.7498,
+      "step": 638
+    },
+    {
+      "epoch": 3.332464146023468,
+      "grad_norm": 0.29080523675344855,
+      "learning_rate": 2.3868748102066964e-05,
+      "loss": 0.7457,
+      "step": 639
+    },
+    {
+      "epoch": 3.33767926988266,
+      "grad_norm": 0.31636040314912883,
+      "learning_rate": 2.3734989171153475e-05,
+      "loss": 0.7605,
+      "step": 640
+    },
+    {
+      "epoch": 3.3428943937418514,
+      "grad_norm": 0.3319918116811582,
+      "learning_rate": 2.3601447794349035e-05,
+      "loss": 0.7601,
+      "step": 641
+    },
+    {
+      "epoch": 3.348109517601043,
+      "grad_norm": 0.24931361371615676,
+      "learning_rate": 2.346812575784826e-05,
+      "loss": 0.752,
+      "step": 642
+    },
+    {
+      "epoch": 3.353324641460235,
+      "grad_norm": 0.30718347134905344,
+      "learning_rate": 2.333502484491202e-05,
+      "loss": 0.7616,
+      "step": 643
+    },
+    {
+      "epoch": 3.3585397653194264,
+      "grad_norm": 0.282987877244827,
+      "learning_rate": 2.3202146835843458e-05,
+      "loss": 0.7462,
+      "step": 644
+    },
+    {
+      "epoch": 3.363754889178618,
+      "grad_norm": 0.24823892221787283,
+      "learning_rate": 2.3069493507964328e-05,
+      "loss": 0.7508,
+      "step": 645
+    },
+    {
+      "epoch": 3.3689700130378095,
+      "grad_norm": 0.2747764846839421,
+      "learning_rate": 2.2937066635591082e-05,
+      "loss": 0.757,
+      "step": 646
+    },
+    {
+      "epoch": 3.374185136897001,
+      "grad_norm": 0.2496756674179984,
+      "learning_rate": 2.2804867990011186e-05,
+      "loss": 0.7578,
+      "step": 647
+    },
+    {
+      "epoch": 3.379400260756193,
+      "grad_norm": 0.2395621072609135,
+      "learning_rate": 2.26728993394595e-05,
+      "loss": 0.765,
+      "step": 648
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.25006905141016617,
+      "learning_rate": 2.2541162449094517e-05,
+      "loss": 0.7558,
+      "step": 649
+    },
+    {
+      "epoch": 3.389830508474576,
+      "grad_norm": 0.22593196210127933,
+      "learning_rate": 2.2409659080974792e-05,
+      "loss": 0.7441,
+      "step": 650
+    },
+    {
+      "epoch": 3.395045632333768,
+      "grad_norm": 0.25905615560125267,
+      "learning_rate": 2.2278390994035437e-05,
+      "loss": 0.759,
+      "step": 651
+    },
+    {
+      "epoch": 3.4002607561929596,
+      "grad_norm": 0.21460137257291464,
+      "learning_rate": 2.214735994406449e-05,
+      "loss": 0.7356,
+      "step": 652
+    },
+    {
+      "epoch": 3.405475880052151,
+      "grad_norm": 0.2844041034563339,
+      "learning_rate": 2.2016567683679474e-05,
+      "loss": 0.7611,
+      "step": 653
+    },
+    {
+      "epoch": 3.4106910039113427,
+      "grad_norm": 0.2193094602636308,
+      "learning_rate": 2.188601596230402e-05,
+      "loss": 0.7628,
+      "step": 654
+    },
+    {
+      "epoch": 3.4159061277705347,
+      "grad_norm": 0.2688886990907944,
+      "learning_rate": 2.1755706526144346e-05,
+      "loss": 0.7555,
+      "step": 655
+    },
+    {
+      "epoch": 3.4211212516297262,
+      "grad_norm": 0.22979387277456514,
+      "learning_rate": 2.1625641118165993e-05,
+      "loss": 0.7506,
+      "step": 656
+    },
+    {
+      "epoch": 3.426336375488918,
+      "grad_norm": 0.2575836540395883,
+      "learning_rate": 2.1495821478070463e-05,
+      "loss": 0.7454,
+      "step": 657
+    },
+    {
+      "epoch": 3.4315514993481093,
+      "grad_norm": 0.22113251166352366,
+      "learning_rate": 2.1366249342271983e-05,
+      "loss": 0.7607,
+      "step": 658
+    },
+    {
+      "epoch": 3.4367666232073013,
+      "grad_norm": 0.2541994114755118,
+      "learning_rate": 2.123692644387427e-05,
+      "loss": 0.7529,
+      "step": 659
+    },
+    {
+      "epoch": 3.441981747066493,
+      "grad_norm": 0.17984867947865277,
+      "learning_rate": 2.110785451264733e-05,
+      "loss": 0.7584,
+      "step": 660
+    },
+    {
+      "epoch": 3.4471968709256844,
+      "grad_norm": 0.2623173957981238,
+      "learning_rate": 2.0979035275004326e-05,
+      "loss": 0.7589,
+      "step": 661
+    },
+    {
+      "epoch": 3.4524119947848764,
+      "grad_norm": 0.17573507602846267,
+      "learning_rate": 2.0850470453978527e-05,
+      "loss": 0.7575,
+      "step": 662
+    },
+    {
+      "epoch": 3.457627118644068,
+      "grad_norm": 0.23281049598480352,
+      "learning_rate": 2.07221617692002e-05,
+      "loss": 0.7495,
+      "step": 663
+    },
+    {
+      "epoch": 3.4628422425032594,
+      "grad_norm": 0.19792669337262203,
+      "learning_rate": 2.0594110936873646e-05,
+      "loss": 0.7554,
+      "step": 664
+    },
+    {
+      "epoch": 3.468057366362451,
+      "grad_norm": 0.2348813479171187,
+      "learning_rate": 2.0466319669754257e-05,
+      "loss": 0.7459,
+      "step": 665
+    },
+    {
+      "epoch": 3.4732724902216425,
+      "grad_norm": 0.17805434883842863,
+      "learning_rate": 2.033878967712556e-05,
+      "loss": 0.7482,
+      "step": 666
+    },
+    {
+      "epoch": 3.4784876140808345,
+      "grad_norm": 0.2324152396008409,
+      "learning_rate": 2.0211522664776378e-05,
+      "loss": 0.7627,
+      "step": 667
+    },
+    {
+      "epoch": 3.483702737940026,
+      "grad_norm": 0.21569790923016002,
+      "learning_rate": 2.0084520334978066e-05,
+      "loss": 0.7475,
+      "step": 668
+    },
+    {
+      "epoch": 3.4889178617992176,
+      "grad_norm": 0.22787366870500253,
+      "learning_rate": 1.9957784386461642e-05,
+      "loss": 0.7539,
+      "step": 669
+    },
+    {
+      "epoch": 3.4941329856584096,
+      "grad_norm": 0.19346838428327817,
+      "learning_rate": 1.9831316514395125e-05,
+      "loss": 0.7482,
+      "step": 670
+    },
+    {
+      "epoch": 3.499348109517601,
+      "grad_norm": 0.22386565166114242,
+      "learning_rate": 1.9705118410360888e-05,
+      "loss": 0.7516,
+      "step": 671
+    },
+    {
+      "epoch": 3.5045632333767927,
+      "grad_norm": 0.19045550096072955,
+      "learning_rate": 1.9579191762332962e-05,
+      "loss": 0.7524,
+      "step": 672
+    },
+    {
+      "epoch": 3.509778357235984,
+      "grad_norm": 0.2568792722114023,
+      "learning_rate": 1.9453538254654492e-05,
+      "loss": 0.7608,
+      "step": 673
+    },
+    {
+      "epoch": 3.514993481095176,
+      "grad_norm": 0.2309953251505408,
+      "learning_rate": 1.9328159568015253e-05,
+      "loss": 0.761,
+      "step": 674
+    },
+    {
+      "epoch": 3.5202086049543677,
+      "grad_norm": 0.20090449798296708,
+      "learning_rate": 1.920305737942908e-05,
+      "loss": 0.7498,
+      "step": 675
+    },
+    {
+      "epoch": 3.5254237288135593,
+      "grad_norm": 0.1995433496289568,
+      "learning_rate": 1.9078233362211488e-05,
+      "loss": 0.7635,
+      "step": 676
+    },
+    {
+      "epoch": 3.530638852672751,
+      "grad_norm": 0.20023193260506214,
+      "learning_rate": 1.895368918595733e-05,
+      "loss": 0.7497,
+      "step": 677
+    },
+    {
+      "epoch": 3.5358539765319428,
+      "grad_norm": 0.19621170906491942,
+      "learning_rate": 1.882942651651835e-05,
+      "loss": 0.7434,
+      "step": 678
+    },
+    {
+      "epoch": 3.5410691003911343,
+      "grad_norm": 0.20574042364789344,
+      "learning_rate": 1.8705447015981056e-05,
+      "loss": 0.7436,
+      "step": 679
+    },
+    {
+      "epoch": 3.546284224250326,
+      "grad_norm": 0.18782121755639544,
+      "learning_rate": 1.858175234264434e-05,
+      "loss": 0.7556,
+      "step": 680
+    },
+    {
+      "epoch": 3.551499348109518,
+      "grad_norm": 0.1936150072481402,
+      "learning_rate": 1.845834415099736e-05,
+      "loss": 0.757,
+      "step": 681
+    },
+    {
+      "epoch": 3.5567144719687094,
+      "grad_norm": 0.1926949621185055,
+      "learning_rate": 1.833522409169748e-05,
+      "loss": 0.756,
+      "step": 682
+    },
+    {
+      "epoch": 3.561929595827901,
+      "grad_norm": 0.20476384841537432,
+      "learning_rate": 1.8212393811548074e-05,
+      "loss": 0.7341,
+      "step": 683
+    },
+    {
+      "epoch": 3.5671447196870925,
+      "grad_norm": 0.1696138534253153,
+      "learning_rate": 1.8089854953476503e-05,
+      "loss": 0.7521,
+      "step": 684
+    },
+    {
+      "epoch": 3.572359843546284,
+      "grad_norm": 0.22686575356319796,
+      "learning_rate": 1.796760915651229e-05,
+      "loss": 0.7621,
+      "step": 685
+    },
+    {
+      "epoch": 3.577574967405476,
+      "grad_norm": 0.18860960425351292,
+      "learning_rate": 1.784565805576503e-05,
+      "loss": 0.7537,
+      "step": 686
+    },
+    {
+      "epoch": 3.5827900912646675,
+      "grad_norm": 0.19640065748239693,
+      "learning_rate": 1.7724003282402567e-05,
+      "loss": 0.7592,
+      "step": 687
+    },
+    {
+      "epoch": 3.588005215123859,
+      "grad_norm": 0.18973579313916158,
+      "learning_rate": 1.760264646362926e-05,
+      "loss": 0.7477,
+      "step": 688
+    },
+    {
+      "epoch": 3.593220338983051,
+      "grad_norm": 0.20298759538425024,
+      "learning_rate": 1.7481589222664076e-05,
+      "loss": 0.7486,
+      "step": 689
+    },
+    {
+      "epoch": 3.5984354628422426,
+      "grad_norm": 0.17790150008707342,
+      "learning_rate": 1.736083317871898e-05,
+      "loss": 0.7458,
+      "step": 690
+    },
+    {
+      "epoch": 3.603650586701434,
+      "grad_norm": 0.20575599062083585,
+      "learning_rate": 1.7240379946977283e-05,
+      "loss": 0.7544,
+      "step": 691
+    },
+    {
+      "epoch": 3.6088657105606257,
+      "grad_norm": 0.17855398739046716,
+      "learning_rate": 1.712023113857195e-05,
+      "loss": 0.756,
+      "step": 692
+    },
+    {
+      "epoch": 3.614080834419817,
+      "grad_norm": 0.1947379717329557,
+      "learning_rate": 1.700038836056413e-05,
+      "loss": 0.7547,
+      "step": 693
+    },
+    {
+      "epoch": 3.619295958279009,
+      "grad_norm": 0.17740265903389008,
+      "learning_rate": 1.6880853215921642e-05,
+      "loss": 0.7532,
+      "step": 694
+    },
+    {
+      "epoch": 3.6245110821382007,
+      "grad_norm": 0.18082866727173452,
+      "learning_rate": 1.676162730349751e-05,
+      "loss": 0.7466,
+      "step": 695
+    },
+    {
+      "epoch": 3.6297262059973923,
+      "grad_norm": 0.18963190902387492,
+      "learning_rate": 1.6642712218008613e-05,
+      "loss": 0.742,
+      "step": 696
+    },
+    {
+      "epoch": 3.6349413298565842,
+      "grad_norm": 0.18716923070682592,
+      "learning_rate": 1.6524109550014323e-05,
+      "loss": 0.7515,
+      "step": 697
+    },
+    {
+      "epoch": 3.640156453715776,
+      "grad_norm": 0.20784228257641144,
+      "learning_rate": 1.6405820885895242e-05,
+      "loss": 0.7578,
+      "step": 698
+    },
+    {
+      "epoch": 3.6453715775749673,
+      "grad_norm": 0.1920361900700138,
+      "learning_rate": 1.6287847807832e-05,
+      "loss": 0.7538,
+      "step": 699
+    },
+    {
+      "epoch": 3.6505867014341593,
+      "grad_norm": 0.1652320731622616,
+      "learning_rate": 1.617019189378407e-05,
+      "loss": 0.7511,
+      "step": 700
+    },
+    {
+      "epoch": 3.655801825293351,
+      "grad_norm": 0.19431451990268614,
+      "learning_rate": 1.6052854717468644e-05,
+      "loss": 0.7515,
+      "step": 701
+    },
+    {
+      "epoch": 3.6610169491525424,
+      "grad_norm": 0.19087250562535932,
+      "learning_rate": 1.5935837848339667e-05,
+      "loss": 0.7568,
+      "step": 702
+    },
+    {
+      "epoch": 3.666232073011734,
+      "grad_norm": 0.20029152301458664,
+      "learning_rate": 1.581914285156673e-05,
+      "loss": 0.7465,
+      "step": 703
+    },
+    {
+      "epoch": 3.6714471968709255,
+      "grad_norm": 0.18770482135472025,
+      "learning_rate": 1.5702771288014206e-05,
+      "loss": 0.7516,
+      "step": 704
+    },
+    {
+      "epoch": 3.6766623207301175,
+      "grad_norm": 0.17463919163255975,
+      "learning_rate": 1.5586724714220397e-05,
+      "loss": 0.7498,
+      "step": 705
+    },
+    {
+      "epoch": 3.681877444589309,
+      "grad_norm": 0.1942555556221587,
+      "learning_rate": 1.5471004682376626e-05,
+      "loss": 0.743,
+      "step": 706
+    },
+    {
+      "epoch": 3.6870925684485005,
+      "grad_norm": 0.1946562516538543,
+      "learning_rate": 1.535561274030652e-05,
+      "loss": 0.7569,
+      "step": 707
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.17582331062620302,
+      "learning_rate": 1.5240550431445376e-05,
+      "loss": 0.7519,
+      "step": 708
+    },
+    {
+      "epoch": 3.697522816166884,
+      "grad_norm": 0.1861426528732765,
+      "learning_rate": 1.5125819294819391e-05,
+      "loss": 0.7482,
+      "step": 709
+    },
+    {
+      "epoch": 3.7027379400260756,
+      "grad_norm": 0.18269562874012577,
+      "learning_rate": 1.5011420865025143e-05,
+      "loss": 0.7523,
+      "step": 710
+    },
+    {
+      "epoch": 3.707953063885267,
+      "grad_norm": 0.18506228544951506,
+      "learning_rate": 1.4897356672209112e-05,
+      "loss": 0.7547,
+      "step": 711
+    },
+    {
+      "epoch": 3.7131681877444587,
+      "grad_norm": 0.20416348855073826,
+      "learning_rate": 1.4783628242047097e-05,
+      "loss": 0.7502,
+      "step": 712
+    },
+    {
+      "epoch": 3.7183833116036507,
+      "grad_norm": 0.16997764410348623,
+      "learning_rate": 1.467023709572391e-05,
+      "loss": 0.7512,
+      "step": 713
+    },
+    {
+      "epoch": 3.723598435462842,
+      "grad_norm": 0.18897244039596858,
+      "learning_rate": 1.4557184749912981e-05,
+      "loss": 0.7473,
+      "step": 714
+    },
+    {
+      "epoch": 3.7288135593220337,
+      "grad_norm": 0.19172065409291622,
+      "learning_rate": 1.4444472716756074e-05,
+      "loss": 0.7497,
+      "step": 715
+    },
+    {
+      "epoch": 3.7340286831812257,
+      "grad_norm": 0.171487618996453,
+      "learning_rate": 1.433210250384311e-05,
+      "loss": 0.7526,
+      "step": 716
+    },
+    {
+      "epoch": 3.7392438070404173,
+      "grad_norm": 0.17442752265085087,
+      "learning_rate": 1.4220075614191924e-05,
+      "loss": 0.7515,
+      "step": 717
+    },
+    {
+      "epoch": 3.744458930899609,
+      "grad_norm": 0.17508720890471086,
+      "learning_rate": 1.4108393546228194e-05,
+      "loss": 0.7578,
+      "step": 718
+    },
+    {
+      "epoch": 3.749674054758801,
+      "grad_norm": 0.19876602903764187,
+      "learning_rate": 1.3997057793765452e-05,
+      "loss": 0.7483,
+      "step": 719
+    },
+    {
+      "epoch": 3.7548891786179923,
+      "grad_norm": 0.17214365448338131,
+      "learning_rate": 1.3886069845985013e-05,
+      "loss": 0.7568,
+      "step": 720
+    },
+    {
+      "epoch": 3.760104302477184,
+      "grad_norm": 0.1920697005442559,
+      "learning_rate": 1.3775431187416097e-05,
+      "loss": 0.7504,
+      "step": 721
+    },
+    {
+      "epoch": 3.7653194263363754,
+      "grad_norm": 0.18503507230287503,
+      "learning_rate": 1.3665143297916018e-05,
+      "loss": 0.7636,
+      "step": 722
+    },
+    {
+      "epoch": 3.770534550195567,
+      "grad_norm": 0.19252969937704967,
+      "learning_rate": 1.3555207652650308e-05,
+      "loss": 0.7526,
+      "step": 723
+    },
+    {
+      "epoch": 3.775749674054759,
+      "grad_norm": 0.1554286223570067,
+      "learning_rate": 1.3445625722073023e-05,
+      "loss": 0.7512,
+      "step": 724
+    },
+    {
+      "epoch": 3.7809647979139505,
+      "grad_norm": 0.16978176727895614,
+      "learning_rate": 1.3336398971907114e-05,
+      "loss": 0.7556,
+      "step": 725
+    },
+    {
+      "epoch": 3.786179921773142,
+      "grad_norm": 0.15542348945118697,
+      "learning_rate": 1.3227528863124745e-05,
+      "loss": 0.7512,
+      "step": 726
+    },
+    {
+      "epoch": 3.791395045632334,
+      "grad_norm": 0.16399605746487972,
+      "learning_rate": 1.3119016851927798e-05,
+      "loss": 0.756,
+      "step": 727
+    },
+    {
+      "epoch": 3.7966101694915255,
+      "grad_norm": 0.16707283374852486,
+      "learning_rate": 1.301086438972842e-05,
+      "loss": 0.7475,
+      "step": 728
+    },
+    {
+      "epoch": 3.801825293350717,
+      "grad_norm": 0.15983694259778197,
+      "learning_rate": 1.2903072923129547e-05,
+      "loss": 0.7474,
+      "step": 729
+    },
+    {
+      "epoch": 3.8070404172099086,
+      "grad_norm": 0.14876964520461528,
+      "learning_rate": 1.2795643893905575e-05,
+      "loss": 0.7471,
+      "step": 730
+    },
+    {
+      "epoch": 3.8122555410691,
+      "grad_norm": 0.19924804420654435,
+      "learning_rate": 1.2688578738983131e-05,
+      "loss": 0.7464,
+      "step": 731
+    },
+    {
+      "epoch": 3.817470664928292,
+      "grad_norm": 0.19191423801767207,
+      "learning_rate": 1.2581878890421777e-05,
+      "loss": 0.7562,
+      "step": 732
+    },
+    {
+      "epoch": 3.8226857887874837,
+      "grad_norm": 0.20765129848679256,
+      "learning_rate": 1.2475545775394879e-05,
+      "loss": 0.7565,
+      "step": 733
+    },
+    {
+      "epoch": 3.827900912646675,
+      "grad_norm": 0.1697957267489159,
+      "learning_rate": 1.2369580816170563e-05,
+      "loss": 0.7586,
+      "step": 734
+    },
+    {
+      "epoch": 3.833116036505867,
+      "grad_norm": 0.1716215315213691,
+      "learning_rate": 1.2263985430092618e-05,
+      "loss": 0.7524,
+      "step": 735
+    },
+    {
+      "epoch": 3.8383311603650587,
+      "grad_norm": 0.18944814607450408,
+      "learning_rate": 1.2158761029561582e-05,
+      "loss": 0.7547,
+      "step": 736
+    },
+    {
+      "epoch": 3.8435462842242503,
+      "grad_norm": 0.14989091598529683,
+      "learning_rate": 1.2053909022015863e-05,
+      "loss": 0.7446,
+      "step": 737
+    },
+    {
+      "epoch": 3.8487614080834422,
+      "grad_norm": 0.1858141017872855,
+      "learning_rate": 1.194943080991287e-05,
+      "loss": 0.7503,
+      "step": 738
+    },
+    {
+      "epoch": 3.853976531942634,
+      "grad_norm": 0.14994883200639647,
+      "learning_rate": 1.1845327790710276e-05,
+      "loss": 0.747,
+      "step": 739
+    },
+    {
+      "epoch": 3.8591916558018253,
+      "grad_norm": 0.14941004189650142,
+      "learning_rate": 1.1741601356847365e-05,
+      "loss": 0.7541,
+      "step": 740
+    },
+    {
+      "epoch": 3.864406779661017,
+      "grad_norm": 0.16071992101640972,
+      "learning_rate": 1.1638252895726327e-05,
+      "loss": 0.7496,
+      "step": 741
+    },
+    {
+      "epoch": 3.8696219035202084,
+      "grad_norm": 0.1374205892222837,
+      "learning_rate": 1.1535283789693756e-05,
+      "loss": 0.753,
+      "step": 742
+    },
+    {
+      "epoch": 3.8748370273794004,
+      "grad_norm": 0.1394433360371365,
+      "learning_rate": 1.1432695416022158e-05,
+      "loss": 0.7535,
+      "step": 743
+    },
+    {
+      "epoch": 3.880052151238592,
+      "grad_norm": 0.16866985633081483,
+      "learning_rate": 1.1330489146891494e-05,
+      "loss": 0.7517,
+      "step": 744
+    },
+    {
+      "epoch": 3.8852672750977835,
+      "grad_norm": 0.15730043046261316,
+      "learning_rate": 1.1228666349370897e-05,
+      "loss": 0.7474,
+      "step": 745
+    },
+    {
+      "epoch": 3.8904823989569755,
+      "grad_norm": 0.1848086766193644,
+      "learning_rate": 1.112722838540031e-05,
+      "loss": 0.7615,
+      "step": 746
+    },
+    {
+      "epoch": 3.895697522816167,
+      "grad_norm": 0.16625357168861077,
+      "learning_rate": 1.1026176611772286e-05,
+      "loss": 0.7488,
+      "step": 747
+    },
+    {
+      "epoch": 3.9009126466753585,
+      "grad_norm": 0.1629255000824365,
+      "learning_rate": 1.0925512380113892e-05,
+      "loss": 0.7577,
+      "step": 748
+    },
+    {
+      "epoch": 3.90612777053455,
+      "grad_norm": 0.18675770122521296,
+      "learning_rate": 1.0825237036868575e-05,
+      "loss": 0.7531,
+      "step": 749
+    },
+    {
+      "epoch": 3.9113428943937416,
+      "grad_norm": 0.15755029689979122,
+      "learning_rate": 1.0725351923278144e-05,
+      "loss": 0.7548,
+      "step": 750
+    },
+    {
+      "epoch": 3.9165580182529336,
+      "grad_norm": 0.18913533326276255,
+      "learning_rate": 1.0625858375364917e-05,
+      "loss": 0.754,
+      "step": 751
+    },
+    {
+      "epoch": 3.921773142112125,
+      "grad_norm": 0.18170716935082054,
+      "learning_rate": 1.0526757723913735e-05,
+      "loss": 0.7441,
+      "step": 752
+    },
+    {
+      "epoch": 3.9269882659713167,
+      "grad_norm": 0.13608488064261914,
+      "learning_rate": 1.042805129445423e-05,
+      "loss": 0.7508,
+      "step": 753
+    },
+    {
+      "epoch": 3.9322033898305087,
+      "grad_norm": 0.17632467806971025,
+      "learning_rate": 1.0329740407243105e-05,
+      "loss": 0.7747,
+      "step": 754
+    },
+    {
+      "epoch": 3.9374185136897,
+      "grad_norm": 0.17072961754308208,
+      "learning_rate": 1.0231826377246437e-05,
+      "loss": 0.7462,
+      "step": 755
+    },
+    {
+      "epoch": 3.9426336375488917,
+      "grad_norm": 0.16478645139827036,
+      "learning_rate": 1.0134310514122082e-05,
+      "loss": 0.7563,
+      "step": 756
+    },
+    {
+      "epoch": 3.9478487614080837,
+      "grad_norm": 0.17542376854887573,
+      "learning_rate": 1.0037194122202227e-05,
+      "loss": 0.7567,
+      "step": 757
+    },
+    {
+      "epoch": 3.9530638852672753,
+      "grad_norm": 0.17159268136488617,
+      "learning_rate": 9.940478500475858e-06,
+      "loss": 0.7503,
+      "step": 758
+    },
+    {
+      "epoch": 3.958279009126467,
+      "grad_norm": 0.15300026372913983,
+      "learning_rate": 9.844164942571424e-06,
+      "loss": 0.7523,
+      "step": 759
+    },
+    {
+      "epoch": 3.9634941329856583,
+      "grad_norm": 0.159680621058249,
+      "learning_rate": 9.748254736739571e-06,
+      "loss": 0.757,
+      "step": 760
+    },
+    {
+      "epoch": 3.96870925684485,
+      "grad_norm": 0.15768593487660648,
+      "learning_rate": 9.652749165835828e-06,
+      "loss": 0.7458,
+      "step": 761
+    },
+    {
+      "epoch": 3.973924380704042,
+      "grad_norm": 0.16943127341342135,
+      "learning_rate": 9.557649507303508e-06,
+      "loss": 0.7456,
+      "step": 762
+    },
+    {
+      "epoch": 3.9791395045632334,
+      "grad_norm": 0.14842747623637131,
+      "learning_rate": 9.462957033156632e-06,
+      "loss": 0.7497,
+      "step": 763
+    },
+    {
+      "epoch": 3.984354628422425,
+      "grad_norm": 0.1501263870664962,
+      "learning_rate": 9.368673009962852e-06,
+      "loss": 0.7514,
+      "step": 764
+    },
+    {
+      "epoch": 3.989569752281617,
+      "grad_norm": 0.17717065342340108,
+      "learning_rate": 9.274798698826557e-06,
+      "loss": 0.7531,
+      "step": 765
+    },
+    {
+      "epoch": 3.9947848761408085,
+      "grad_norm": 0.15866607398462376,
+      "learning_rate": 9.181335355372028e-06,
+      "loss": 0.7512,
+      "step": 766
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.17952277081095838,
+      "learning_rate": 9.088284229726572e-06,
+      "loss": 0.9412,
+      "step": 767
+    },
+    {
+      "epoch": 4.005215123859192,
+      "grad_norm": 0.23274451897753248,
+      "learning_rate": 8.995646566503838e-06,
+      "loss": 0.7375,
+      "step": 768
+    },
+    {
+      "epoch": 4.010430247718383,
+      "grad_norm": 0.17288112228894242,
+      "learning_rate": 8.90342360478723e-06,
+      "loss": 0.7373,
+      "step": 769
+    },
+    {
+      "epoch": 4.015645371577575,
+      "grad_norm": 0.20174270563988336,
+      "learning_rate": 8.81161657811318e-06,
+      "loss": 0.7453,
+      "step": 770
+    },
+    {
+      "epoch": 4.020860495436767,
+      "grad_norm": 0.1726715461295719,
+      "learning_rate": 8.720226714454831e-06,
+      "loss": 0.7421,
+      "step": 771
+    },
+    {
+      "epoch": 4.026075619295958,
+      "grad_norm": 0.20664612564398913,
+      "learning_rate": 8.629255236205475e-06,
+      "loss": 0.7393,
+      "step": 772
+    },
+    {
+      "epoch": 4.03129074315515,
+      "grad_norm": 0.183032821290067,
+      "learning_rate": 8.538703360162267e-06,
+      "loss": 0.7262,
+      "step": 773
+    },
+    {
+      "epoch": 4.036505867014341,
+      "grad_norm": 0.19162204178734604,
+      "learning_rate": 8.44857229750994e-06,
+      "loss": 0.7229,
+      "step": 774
+    },
+    {
+      "epoch": 4.041720990873533,
+      "grad_norm": 0.19714648419474126,
+      "learning_rate": 8.3588632538046e-06,
+      "loss": 0.7406,
+      "step": 775
+    },
+    {
+      "epoch": 4.046936114732725,
+      "grad_norm": 0.18562646027448648,
+      "learning_rate": 8.269577428957571e-06,
+      "loss": 0.7375,
+      "step": 776
+    },
+    {
+      "epoch": 4.052151238591916,
+      "grad_norm": 0.19319081667416735,
+      "learning_rate": 8.180716017219433e-06,
+      "loss": 0.7315,
+      "step": 777
+    },
+    {
+      "epoch": 4.057366362451108,
+      "grad_norm": 0.17933639021135425,
+      "learning_rate": 8.092280207163941e-06,
+      "loss": 0.7439,
+      "step": 778
+    },
+    {
+      "epoch": 4.0625814863103,
+      "grad_norm": 0.16796265908006924,
+      "learning_rate": 8.004271181672178e-06,
+      "loss": 0.7219,
+      "step": 779
+    },
+    {
+      "epoch": 4.067796610169491,
+      "grad_norm": 0.16506690334266957,
+      "learning_rate": 7.916690117916772e-06,
+      "loss": 0.7409,
+      "step": 780
+    },
+    {
+      "epoch": 4.073011734028683,
+      "grad_norm": 0.16443977378096947,
+      "learning_rate": 7.829538187346077e-06,
+      "loss": 0.7377,
+      "step": 781
+    },
+    {
+      "epoch": 4.078226857887874,
+      "grad_norm": 0.1997012117535286,
+      "learning_rate": 7.742816555668532e-06,
+      "loss": 0.7399,
+      "step": 782
+    },
+    {
+      "epoch": 4.083441981747066,
+      "grad_norm": 0.14695927923208477,
+      "learning_rate": 7.656526382837106e-06,
+      "loss": 0.7327,
+      "step": 783
+    },
+    {
+      "epoch": 4.088657105606258,
+      "grad_norm": 0.1799749838620325,
+      "learning_rate": 7.570668823033727e-06,
+      "loss": 0.7309,
+      "step": 784
+    },
+    {
+      "epoch": 4.0938722294654495,
+      "grad_norm": 0.17989297489241943,
+      "learning_rate": 7.4852450246538685e-06,
+      "loss": 0.7343,
+      "step": 785
+    },
+    {
+      "epoch": 4.0990873533246415,
+      "grad_norm": 0.15506025585209593,
+      "learning_rate": 7.40025613029121e-06,
+      "loss": 0.736,
+      "step": 786
+    },
+    {
+      "epoch": 4.1043024771838335,
+      "grad_norm": 0.18105126244534,
+      "learning_rate": 7.315703276722317e-06,
+      "loss": 0.7342,
+      "step": 787
+    },
+    {
+      "epoch": 4.1095176010430245,
+      "grad_norm": 0.19050717050961422,
+      "learning_rate": 7.231587594891438e-06,
+      "loss": 0.7458,
+      "step": 788
+    },
+    {
+      "epoch": 4.1147327249022165,
+      "grad_norm": 0.14216335950493533,
+      "learning_rate": 7.147910209895435e-06,
+      "loss": 0.7318,
+      "step": 789
+    },
+    {
+      "epoch": 4.1199478487614085,
+      "grad_norm": 0.1655844728014387,
+      "learning_rate": 7.064672240968638e-06,
+      "loss": 0.7442,
+      "step": 790
+    },
+    {
+      "epoch": 4.1251629726206,
+      "grad_norm": 0.14883381171151777,
+      "learning_rate": 6.9818748014679785e-06,
+      "loss": 0.7403,
+      "step": 791
+    },
+    {
+      "epoch": 4.130378096479792,
+      "grad_norm": 0.1602685057342838,
+      "learning_rate": 6.899518998858017e-06,
+      "loss": 0.7405,
+      "step": 792
+    },
+    {
+      "epoch": 4.135593220338983,
+      "grad_norm": 0.1649659601993552,
+      "learning_rate": 6.81760593469615e-06,
+      "loss": 0.7346,
+      "step": 793
+    },
+    {
+      "epoch": 4.140808344198175,
+      "grad_norm": 0.12921893028933829,
+      "learning_rate": 6.7361367046179325e-06,
+      "loss": 0.7303,
+      "step": 794
+    },
+    {
+      "epoch": 4.146023468057367,
+      "grad_norm": 0.12489877860585255,
+      "learning_rate": 6.655112398322332e-06,
+      "loss": 0.731,
+      "step": 795
+    },
+    {
+      "epoch": 4.151238591916558,
+      "grad_norm": 0.17428621687246298,
+      "learning_rate": 6.574534099557204e-06,
+      "loss": 0.7307,
+      "step": 796
+    },
+    {
+      "epoch": 4.15645371577575,
+      "grad_norm": 0.1251987161714777,
+      "learning_rate": 6.49440288610482e-06,
+      "loss": 0.7275,
+      "step": 797
+    },
+    {
+      "epoch": 4.161668839634942,
+      "grad_norm": 0.15533418379259104,
+      "learning_rate": 6.414719829767406e-06,
+      "loss": 0.7339,
+      "step": 798
+    },
+    {
+      "epoch": 4.166883963494133,
+      "grad_norm": 0.16236363314856642,
+      "learning_rate": 6.335485996352782e-06,
+      "loss": 0.7418,
+      "step": 799
+    },
+    {
+      "epoch": 4.172099087353325,
+      "grad_norm": 0.13801981121095086,
+      "learning_rate": 6.256702445660221e-06,
+      "loss": 0.7304,
+      "step": 800
+    },
+    {
+      "epoch": 4.177314211212516,
+      "grad_norm": 0.14142425026968236,
+      "learning_rate": 6.178370231466142e-06,
+      "loss": 0.743,
+      "step": 801
+    },
+    {
+      "epoch": 4.182529335071708,
+      "grad_norm": 0.16216801700272557,
+      "learning_rate": 6.100490401510089e-06,
+      "loss": 0.7437,
+      "step": 802
+    },
+    {
+      "epoch": 4.1877444589309,
+      "grad_norm": 0.14030637701048476,
+      "learning_rate": 6.0230639974807206e-06,
+      "loss": 0.7383,
+      "step": 803
+    },
+    {
+      "epoch": 4.192959582790091,
+      "grad_norm": 0.13754640300037912,
+      "learning_rate": 5.946092055001824e-06,
+      "loss": 0.7329,
+      "step": 804
+    },
+    {
+      "epoch": 4.198174706649283,
+      "grad_norm": 0.16139415971113086,
+      "learning_rate": 5.869575603618507e-06,
+      "loss": 0.7417,
+      "step": 805
+    },
+    {
+      "epoch": 4.203389830508475,
+      "grad_norm": 0.15626356140455866,
+      "learning_rate": 5.793515666783429e-06,
+      "loss": 0.7337,
+      "step": 806
+    },
+    {
+      "epoch": 4.208604954367666,
+      "grad_norm": 0.12177414306535114,
+      "learning_rate": 5.717913261843078e-06,
+      "loss": 0.733,
+      "step": 807
+    },
+    {
+      "epoch": 4.213820078226858,
+      "grad_norm": 0.12926025156980447,
+      "learning_rate": 5.642769400024199e-06,
+      "loss": 0.7422,
+      "step": 808
+    },
+    {
+      "epoch": 4.219035202086049,
+      "grad_norm": 0.13704261747895924,
+      "learning_rate": 5.568085086420247e-06,
+      "loss": 0.7404,
+      "step": 809
+    },
+    {
+      "epoch": 4.224250325945241,
+      "grad_norm": 0.13842188890961235,
+      "learning_rate": 5.4938613199779334e-06,
+      "loss": 0.7334,
+      "step": 810
+    },
+    {
+      "epoch": 4.229465449804433,
+      "grad_norm": 0.13979962035636995,
+      "learning_rate": 5.420099093483911e-06,
+      "loss": 0.7222,
+      "step": 811
+    },
+    {
+      "epoch": 4.234680573663624,
+      "grad_norm": 0.12565163857223183,
+      "learning_rate": 5.346799393551436e-06,
+      "loss": 0.7307,
+      "step": 812
+    },
+    {
+      "epoch": 4.239895697522816,
+      "grad_norm": 0.12679243834545414,
+      "learning_rate": 5.273963200607206e-06,
+      "loss": 0.7387,
+      "step": 813
+    },
+    {
+      "epoch": 4.245110821382008,
+      "grad_norm": 0.13813903814909753,
+      "learning_rate": 5.201591488878257e-06,
+      "loss": 0.7251,
+      "step": 814
+    },
+    {
+      "epoch": 4.250325945241199,
+      "grad_norm": 0.1285263356419264,
+      "learning_rate": 5.129685226378893e-06,
+      "loss": 0.7234,
+      "step": 815
+    },
+    {
+      "epoch": 4.255541069100391,
+      "grad_norm": 0.12820199717697753,
+      "learning_rate": 5.058245374897763e-06,
+      "loss": 0.7271,
+      "step": 816
+    },
+    {
+      "epoch": 4.260756192959583,
+      "grad_norm": 0.12495602397125814,
+      "learning_rate": 4.987272889985009e-06,
+      "loss": 0.7393,
+      "step": 817
+    },
+    {
+      "epoch": 4.265971316818774,
+      "grad_norm": 0.15151715317554978,
+      "learning_rate": 4.916768720939451e-06,
+      "loss": 0.7332,
+      "step": 818
+    },
+    {
+      "epoch": 4.271186440677966,
+      "grad_norm": 0.14747133461014475,
+      "learning_rate": 4.846733810795914e-06,
+      "loss": 0.7297,
+      "step": 819
+    },
+    {
+      "epoch": 4.276401564537157,
+      "grad_norm": 0.14100514701144612,
+      "learning_rate": 4.777169096312624e-06,
+      "loss": 0.7276,
+      "step": 820
+    },
+    {
+      "epoch": 4.281616688396349,
+      "grad_norm": 0.1646113420856527,
+      "learning_rate": 4.708075507958642e-06,
+      "loss": 0.7281,
+      "step": 821
+    },
+    {
+      "epoch": 4.286831812255541,
+      "grad_norm": 0.13983624281016735,
+      "learning_rate": 4.639453969901442e-06,
+      "loss": 0.7276,
+      "step": 822
+    },
+    {
+      "epoch": 4.292046936114732,
+      "grad_norm": 0.13237794200876413,
+      "learning_rate": 4.571305399994561e-06,
+      "loss": 0.7409,
+      "step": 823
+    },
+    {
+      "epoch": 4.297262059973924,
+      "grad_norm": 0.180777223722747,
+      "learning_rate": 4.503630709765294e-06,
+      "loss": 0.741,
+      "step": 824
+    },
+    {
+      "epoch": 4.302477183833116,
+      "grad_norm": 0.1367694718240071,
+      "learning_rate": 4.436430804402521e-06,
+      "loss": 0.7331,
+      "step": 825
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.15547588436169807,
+      "learning_rate": 4.3697065827446e-06,
+      "loss": 0.7376,
+      "step": 826
+    },
+    {
+      "epoch": 4.3129074315514995,
+      "grad_norm": 0.17342743882348932,
+      "learning_rate": 4.303458937267326e-06,
+      "loss": 0.7267,
+      "step": 827
+    },
+    {
+      "epoch": 4.318122555410691,
+      "grad_norm": 0.16682655867364918,
+      "learning_rate": 4.237688754072022e-06,
+      "loss": 0.7309,
+      "step": 828
+    },
+    {
+      "epoch": 4.3233376792698825,
+      "grad_norm": 0.16450623559503522,
+      "learning_rate": 4.172396912873659e-06,
+      "loss": 0.7342,
+      "step": 829
+    },
+    {
+      "epoch": 4.3285528031290745,
+      "grad_norm": 0.19380979384775532,
+      "learning_rate": 4.107584286989097e-06,
+      "loss": 0.7392,
+      "step": 830
+    },
+    {
+      "epoch": 4.333767926988266,
+      "grad_norm": 0.18721173933947008,
+      "learning_rate": 4.043251743325436e-06,
+      "loss": 0.7303,
+      "step": 831
+    },
+    {
+      "epoch": 4.338983050847458,
+      "grad_norm": 0.15835333882238062,
+      "learning_rate": 3.979400142368368e-06,
+      "loss": 0.7301,
+      "step": 832
+    },
+    {
+      "epoch": 4.34419817470665,
+      "grad_norm": 0.18385561342339599,
+      "learning_rate": 3.916030338170696e-06,
+      "loss": 0.7374,
+      "step": 833
+    },
+    {
+      "epoch": 4.349413298565841,
+      "grad_norm": 0.2022205872843893,
+      "learning_rate": 3.8531431783409165e-06,
+      "loss": 0.7387,
+      "step": 834
+    },
+    {
+      "epoch": 4.354628422425033,
+      "grad_norm": 0.15316197738525242,
+      "learning_rate": 3.7907395040318685e-06,
+      "loss": 0.7385,
+      "step": 835
+    },
+    {
+      "epoch": 4.359843546284225,
+      "grad_norm": 0.1500484412205804,
+      "learning_rate": 3.7288201499294797e-06,
+      "loss": 0.7268,
+      "step": 836
+    },
+    {
+      "epoch": 4.365058670143416,
+      "grad_norm": 0.14189535258541974,
+      "learning_rate": 3.667385944241626e-06,
+      "loss": 0.7453,
+      "step": 837
+    },
+    {
+      "epoch": 4.370273794002608,
+      "grad_norm": 0.15984630251802126,
+      "learning_rate": 3.606437708687023e-06,
+      "loss": 0.7395,
+      "step": 838
+    },
+    {
+      "epoch": 4.375488917861799,
+      "grad_norm": 0.1642822923840156,
+      "learning_rate": 3.545976258484247e-06,
+      "loss": 0.7354,
+      "step": 839
+    },
+    {
+      "epoch": 4.380704041720991,
+      "grad_norm": 0.13221533894356902,
+      "learning_rate": 3.4860024023408577e-06,
+      "loss": 0.7459,
+      "step": 840
+    },
+    {
+      "epoch": 4.385919165580183,
+      "grad_norm": 0.18318603173559903,
+      "learning_rate": 3.426516942442524e-06,
+      "loss": 0.7509,
+      "step": 841
+    },
+    {
+      "epoch": 4.391134289439374,
+      "grad_norm": 0.17040203525878148,
+      "learning_rate": 3.3675206744423394e-06,
+      "loss": 0.7375,
+      "step": 842
+    },
+    {
+      "epoch": 4.396349413298566,
+      "grad_norm": 0.157205271311182,
+      "learning_rate": 3.309014387450189e-06,
+      "loss": 0.7302,
+      "step": 843
+    },
+    {
+      "epoch": 4.401564537157758,
+      "grad_norm": 0.15260420463119376,
+      "learning_rate": 3.2509988640221414e-06,
+      "loss": 0.7407,
+      "step": 844
+    },
+    {
+      "epoch": 4.406779661016949,
+      "grad_norm": 0.16154440296062286,
+      "learning_rate": 3.1934748801500184e-06,
+      "loss": 0.7255,
+      "step": 845
+    },
+    {
+      "epoch": 4.411994784876141,
+      "grad_norm": 0.14388929124722757,
+      "learning_rate": 3.136443205251034e-06,
+      "loss": 0.7369,
+      "step": 846
+    },
+    {
+      "epoch": 4.417209908735332,
+      "grad_norm": 0.14793139417234433,
+      "learning_rate": 3.0799046021574574e-06,
+      "loss": 0.7337,
+      "step": 847
+    },
+    {
+      "epoch": 4.422425032594524,
+      "grad_norm": 0.14906886650426293,
+      "learning_rate": 3.0238598271064323e-06,
+      "loss": 0.7319,
+      "step": 848
+    },
+    {
+      "epoch": 4.427640156453716,
+      "grad_norm": 0.18885017044608088,
+      "learning_rate": 2.9683096297298753e-06,
+      "loss": 0.7226,
+      "step": 849
+    },
+    {
+      "epoch": 4.432855280312907,
+      "grad_norm": 0.16698890562230234,
+      "learning_rate": 2.9132547530444254e-06,
+      "loss": 0.74,
+      "step": 850
+    },
+    {
+      "epoch": 4.438070404172099,
+      "grad_norm": 0.12976275259498832,
+      "learning_rate": 2.8586959334415064e-06,
+      "loss": 0.7325,
+      "step": 851
+    },
+    {
+      "epoch": 4.443285528031291,
+      "grad_norm": 0.14807620276113978,
+      "learning_rate": 2.8046339006775115e-06,
+      "loss": 0.7197,
+      "step": 852
+    },
+    {
+      "epoch": 4.448500651890482,
+      "grad_norm": 0.15235231363332793,
+      "learning_rate": 2.751069377863993e-06,
+      "loss": 0.7302,
+      "step": 853
+    },
+    {
+      "epoch": 4.453715775749674,
+      "grad_norm": 0.15202060711926385,
+      "learning_rate": 2.698003081458018e-06,
+      "loss": 0.7289,
+      "step": 854
+    },
+    {
+      "epoch": 4.458930899608866,
+      "grad_norm": 0.12630616368200856,
+      "learning_rate": 2.6454357212526026e-06,
+      "loss": 0.7279,
+      "step": 855
+    },
+    {
+      "epoch": 4.464146023468057,
+      "grad_norm": 0.11739628579126643,
+      "learning_rate": 2.5933680003671625e-06,
+      "loss": 0.7335,
+      "step": 856
+    },
+    {
+      "epoch": 4.469361147327249,
+      "grad_norm": 0.1434952804888338,
+      "learning_rate": 2.541800615238175e-06,
+      "loss": 0.7266,
+      "step": 857
+    },
+    {
+      "epoch": 4.47457627118644,
+      "grad_norm": 0.176277045965935,
+      "learning_rate": 2.4907342556098125e-06,
+      "loss": 0.7269,
+      "step": 858
+    },
+    {
+      "epoch": 4.479791395045632,
+      "grad_norm": 0.18491464302255675,
+      "learning_rate": 2.440169604524738e-06,
+      "loss": 0.7453,
+      "step": 859
+    },
+    {
+      "epoch": 4.485006518904824,
+      "grad_norm": 0.12415001369259007,
+      "learning_rate": 2.390107338314982e-06,
+      "loss": 0.7319,
+      "step": 860
+    },
+    {
+      "epoch": 4.490221642764015,
+      "grad_norm": 0.10478588253479926,
+      "learning_rate": 2.340548126592861e-06,
+      "loss": 0.737,
+      "step": 861
+    },
+    {
+      "epoch": 4.495436766623207,
+      "grad_norm": 0.14165809120057415,
+      "learning_rate": 2.2914926322420515e-06,
+      "loss": 0.7343,
+      "step": 862
+    },
+    {
+      "epoch": 4.500651890482399,
+      "grad_norm": 0.11590106390711405,
+      "learning_rate": 2.242941511408714e-06,
+      "loss": 0.737,
+      "step": 863
+    },
+    {
+      "epoch": 4.50586701434159,
+      "grad_norm": 0.12634034846215603,
+      "learning_rate": 2.1948954134927103e-06,
+      "loss": 0.7377,
+      "step": 864
+    },
+    {
+      "epoch": 4.511082138200782,
+      "grad_norm": 0.11268712032715957,
+      "learning_rate": 2.147354981138925e-06,
+      "loss": 0.7257,
+      "step": 865
+    },
+    {
+      "epoch": 4.5162972620599735,
+      "grad_norm": 0.11916219994594991,
+      "learning_rate": 2.1003208502286744e-06,
+      "loss": 0.729,
+      "step": 866
+    },
+    {
+      "epoch": 4.5215123859191655,
+      "grad_norm": 0.13119722935564196,
+      "learning_rate": 2.0537936498711898e-06,
+      "loss": 0.7316,
+      "step": 867
+    },
+    {
+      "epoch": 4.5267275097783575,
+      "grad_norm": 0.12883622307786546,
+      "learning_rate": 2.007774002395193e-06,
+      "loss": 0.7337,
+      "step": 868
+    },
+    {
+      "epoch": 4.531942633637549,
+      "grad_norm": 0.10574304590110693,
+      "learning_rate": 1.9622625233406143e-06,
+      "loss": 0.7268,
+      "step": 869
+    },
+    {
+      "epoch": 4.5371577574967406,
+      "grad_norm": 0.11189460238334994,
+      "learning_rate": 1.9172598214503146e-06,
+      "loss": 0.736,
+      "step": 870
+    },
+    {
+      "epoch": 4.5423728813559325,
+      "grad_norm": 0.12169459450452168,
+      "learning_rate": 1.8727664986619576e-06,
+      "loss": 0.7359,
+      "step": 871
+    },
+    {
+      "epoch": 4.547588005215124,
+      "grad_norm": 0.11602787776789947,
+      "learning_rate": 1.8287831500999775e-06,
+      "loss": 0.733,
+      "step": 872
+    },
+    {
+      "epoch": 4.552803129074316,
+      "grad_norm": 0.11542784974890338,
+      "learning_rate": 1.7853103640675852e-06,
+      "loss": 0.7274,
+      "step": 873
+    },
+    {
+      "epoch": 4.558018252933508,
+      "grad_norm": 0.10042034550407807,
+      "learning_rate": 1.74234872203892e-06,
+      "loss": 0.7346,
+      "step": 874
+    },
+    {
+      "epoch": 4.563233376792699,
+      "grad_norm": 0.11267524138272358,
+      "learning_rate": 1.699898798651285e-06,
+      "loss": 0.7355,
+      "step": 875
+    },
+    {
+      "epoch": 4.568448500651891,
+      "grad_norm": 0.14020374405591798,
+      "learning_rate": 1.6579611616974212e-06,
+      "loss": 0.7323,
+      "step": 876
+    },
+    {
+      "epoch": 4.573663624511082,
+      "grad_norm": 0.1329010216092144,
+      "learning_rate": 1.6165363721179471e-06,
+      "loss": 0.7333,
+      "step": 877
+    },
+    {
+      "epoch": 4.578878748370274,
+      "grad_norm": 0.1111745646565014,
+      "learning_rate": 1.575624983993853e-06,
+      "loss": 0.7198,
+      "step": 878
+    },
+    {
+      "epoch": 4.584093872229466,
+      "grad_norm": 0.12252637821649354,
+      "learning_rate": 1.5352275445390752e-06,
+      "loss": 0.7265,
+      "step": 879
+    },
+    {
+      "epoch": 4.589308996088657,
+      "grad_norm": 0.10588650578235918,
+      "learning_rate": 1.4953445940931688e-06,
+      "loss": 0.7328,
+      "step": 880
+    },
+    {
+      "epoch": 4.594524119947849,
+      "grad_norm": 0.10242904609486453,
+      "learning_rate": 1.4559766661141317e-06,
+      "loss": 0.7305,
+      "step": 881
+    },
+    {
+      "epoch": 4.599739243807041,
+      "grad_norm": 0.11484087780107081,
+      "learning_rate": 1.4171242871711922e-06,
+      "loss": 0.7432,
+      "step": 882
+    },
+    {
+      "epoch": 4.604954367666232,
+      "grad_norm": 0.10614866491620119,
+      "learning_rate": 1.3787879769378277e-06,
+      "loss": 0.7317,
+      "step": 883
+    },
+    {
+      "epoch": 4.610169491525424,
+      "grad_norm": 0.11216621248162341,
+      "learning_rate": 1.3409682481848063e-06,
+      "loss": 0.7379,
+      "step": 884
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.09993770692863328,
+      "learning_rate": 1.3036656067732679e-06,
+      "loss": 0.739,
+      "step": 885
+    },
+    {
+      "epoch": 4.620599739243807,
+      "grad_norm": 0.09930632699157292,
+      "learning_rate": 1.2668805516480577e-06,
+      "loss": 0.743,
+      "step": 886
+    },
+    {
+      "epoch": 4.625814863102999,
+      "grad_norm": 0.1008961163196706,
+      "learning_rate": 1.2306135748309633e-06,
+      "loss": 0.746,
+      "step": 887
+    },
+    {
+      "epoch": 4.63102998696219,
+      "grad_norm": 0.1255524490616452,
+      "learning_rate": 1.1948651614141783e-06,
+      "loss": 0.7317,
+      "step": 888
+    },
+    {
+      "epoch": 4.636245110821382,
+      "grad_norm": 0.10155914604030952,
+      "learning_rate": 1.1596357895538213e-06,
+      "loss": 0.7213,
+      "step": 889
+    },
+    {
+      "epoch": 4.641460234680574,
+      "grad_norm": 0.10501816031109035,
+      "learning_rate": 1.1249259304634986e-06,
+      "loss": 0.7398,
+      "step": 890
+    },
+    {
+      "epoch": 4.646675358539765,
+      "grad_norm": 0.13819236460373796,
+      "learning_rate": 1.0907360484080409e-06,
+      "loss": 0.7368,
+      "step": 891
+    },
+    {
+      "epoch": 4.651890482398957,
+      "grad_norm": 0.10830515113482954,
+      "learning_rate": 1.0570666006972875e-06,
+      "loss": 0.7543,
+      "step": 892
+    },
+    {
+      "epoch": 4.657105606258149,
+      "grad_norm": 0.10654993027327486,
+      "learning_rate": 1.023918037679943e-06,
+      "loss": 0.7243,
+      "step": 893
+    },
+    {
+      "epoch": 4.66232073011734,
+      "grad_norm": 0.10012139135715376,
+      "learning_rate": 9.912908027375745e-07,
+      "loss": 0.7373,
+      "step": 894
+    },
+    {
+      "epoch": 4.667535853976532,
+      "grad_norm": 0.0950421289561811,
+      "learning_rate": 9.591853322787003e-07,
+      "loss": 0.7412,
+      "step": 895
+    },
+    {
+      "epoch": 4.672750977835723,
+      "grad_norm": 0.09904664343265647,
+      "learning_rate": 9.276020557329101e-07,
+      "loss": 0.729,
+      "step": 896
+    },
+    {
+      "epoch": 4.677966101694915,
+      "grad_norm": 0.1051253462879674,
+      "learning_rate": 8.965413955451363e-07,
+      "loss": 0.7329,
+      "step": 897
+    },
+    {
+      "epoch": 4.683181225554107,
+      "grad_norm": 0.10272032718312339,
+      "learning_rate": 8.660037671700405e-07,
+      "loss": 0.7387,
+      "step": 898
+    },
+    {
+      "epoch": 4.688396349413298,
+      "grad_norm": 0.09813662944001211,
+      "learning_rate": 8.359895790663963e-07,
+      "loss": 0.7377,
+      "step": 899
+    },
+    {
+      "epoch": 4.69361147327249,
+      "grad_norm": 0.10536763446890161,
+      "learning_rate": 8.064992326916577e-07,
+      "loss": 0.7353,
+      "step": 900
+    },
+    {
+      "epoch": 4.698826597131681,
+      "grad_norm": 0.10236023154880665,
+      "learning_rate": 7.775331224966076e-07,
+      "loss": 0.7272,
+      "step": 901
+    },
+    {
+      "epoch": 4.704041720990873,
+      "grad_norm": 0.09414675460502064,
+      "learning_rate": 7.49091635920034e-07,
+      "loss": 0.7292,
+      "step": 902
+    },
+    {
+      "epoch": 4.709256844850065,
+      "grad_norm": 0.0943501127919075,
+      "learning_rate": 7.211751533835998e-07,
+      "loss": 0.7416,
+      "step": 903
+    },
+    {
+      "epoch": 4.7144719687092564,
+      "grad_norm": 0.1021023935601688,
+      "learning_rate": 6.937840482867142e-07,
+      "loss": 0.7337,
+      "step": 904
+    },
+    {
+      "epoch": 4.719687092568448,
+      "grad_norm": 0.09881423844437276,
+      "learning_rate": 6.669186870015454e-07,
+      "loss": 0.7324,
+      "step": 905
+    },
+    {
+      "epoch": 4.72490221642764,
+      "grad_norm": 0.09987028627197758,
+      "learning_rate": 6.405794288681577e-07,
+      "loss": 0.7374,
+      "step": 906
+    },
+    {
+      "epoch": 4.7301173402868315,
+      "grad_norm": 0.09574731953052935,
+      "learning_rate": 6.147666261896445e-07,
+      "loss": 0.7326,
+      "step": 907
+    },
+    {
+      "epoch": 4.7353324641460235,
+      "grad_norm": 0.09726177695264816,
+      "learning_rate": 5.894806242274565e-07,
+      "loss": 0.7375,
+      "step": 908
+    },
+    {
+      "epoch": 4.7405475880052155,
+      "grad_norm": 0.09663506260906218,
+      "learning_rate": 5.647217611967914e-07,
+      "loss": 0.7366,
+      "step": 909
+    },
+    {
+      "epoch": 4.745762711864407,
+      "grad_norm": 0.09898203274690626,
+      "learning_rate": 5.404903682620299e-07,
+      "loss": 0.7282,
+      "step": 910
+    },
+    {
+      "epoch": 4.7509778357235986,
+      "grad_norm": 0.09888749932985194,
+      "learning_rate": 5.167867695323426e-07,
+      "loss": 0.7234,
+      "step": 911
+    },
+    {
+      "epoch": 4.7561929595827905,
+      "grad_norm": 0.09677300838881348,
+      "learning_rate": 4.936112820573335e-07,
+      "loss": 0.734,
+      "step": 912
+    },
+    {
+      "epoch": 4.761408083441982,
+      "grad_norm": 0.09338358111347181,
+      "learning_rate": 4.709642158228134e-07,
+      "loss": 0.7319,
+      "step": 913
+    },
+    {
+      "epoch": 4.766623207301174,
+      "grad_norm": 0.09475677222529419,
+      "learning_rate": 4.4884587374663727e-07,
+      "loss": 0.7345,
+      "step": 914
+    },
+    {
+      "epoch": 4.771838331160365,
+      "grad_norm": 0.09660842094296682,
+      "learning_rate": 4.2725655167468626e-07,
+      "loss": 0.7328,
+      "step": 915
+    },
+    {
+      "epoch": 4.777053455019557,
+      "grad_norm": 0.09522844691697419,
+      "learning_rate": 4.061965383768529e-07,
+      "loss": 0.7395,
+      "step": 916
+    },
+    {
+      "epoch": 4.782268578878749,
+      "grad_norm": 0.09658310921334537,
+      "learning_rate": 3.856661155432484e-07,
+      "loss": 0.7282,
+      "step": 917
+    },
+    {
+      "epoch": 4.78748370273794,
+      "grad_norm": 0.09209868347782292,
+      "learning_rate": 3.656655577803703e-07,
+      "loss": 0.7321,
+      "step": 918
+    },
+    {
+      "epoch": 4.792698826597132,
+      "grad_norm": 0.09809162403374437,
+      "learning_rate": 3.4619513260746973e-07,
+      "loss": 0.721,
+      "step": 919
+    },
+    {
+      "epoch": 4.797913950456323,
+      "grad_norm": 0.09728296962426303,
+      "learning_rate": 3.2725510045297225e-07,
+      "loss": 0.7383,
+      "step": 920
+    },
+    {
+      "epoch": 4.803129074315515,
+      "grad_norm": 0.0982190410036559,
+      "learning_rate": 3.0884571465096936e-07,
+      "loss": 0.7356,
+      "step": 921
+    },
+    {
+      "epoch": 4.808344198174707,
+      "grad_norm": 0.09138351849890962,
+      "learning_rate": 2.909672214378434e-07,
+      "loss": 0.7375,
+      "step": 922
+    },
+    {
+      "epoch": 4.813559322033898,
+      "grad_norm": 0.08931700971308501,
+      "learning_rate": 2.736198599489903e-07,
+      "loss": 0.7399,
+      "step": 923
+    },
+    {
+      "epoch": 4.81877444589309,
+      "grad_norm": 0.09559406138911163,
+      "learning_rate": 2.568038622155955e-07,
+      "loss": 0.7509,
+      "step": 924
+    },
+    {
+      "epoch": 4.823989569752282,
+      "grad_norm": 0.09150057561208538,
+      "learning_rate": 2.4051945316153846e-07,
+      "loss": 0.7284,
+      "step": 925
+    },
+    {
+      "epoch": 4.829204693611473,
+      "grad_norm": 0.09349065890924081,
+      "learning_rate": 2.2476685060039527e-07,
+      "loss": 0.7299,
+      "step": 926
+    },
+    {
+      "epoch": 4.834419817470665,
+      "grad_norm": 0.09556536713909304,
+      "learning_rate": 2.0954626523251197e-07,
+      "loss": 0.7276,
+      "step": 927
+    },
+    {
+      "epoch": 4.839634941329857,
+      "grad_norm": 0.09471713249250732,
+      "learning_rate": 1.948579006421847e-07,
+      "loss": 0.7428,
+      "step": 928
+    },
+    {
+      "epoch": 4.844850065189048,
+      "grad_norm": 0.09109503005212934,
+      "learning_rate": 1.8070195329495944e-07,
+      "loss": 0.7401,
+      "step": 929
+    },
+    {
+      "epoch": 4.85006518904824,
+      "grad_norm": 0.09359860047319951,
+      "learning_rate": 1.6707861253497214e-07,
+      "loss": 0.7409,
+      "step": 930
+    },
+    {
+      "epoch": 4.855280312907432,
+      "grad_norm": 0.09252775154456866,
+      "learning_rate": 1.539880605824351e-07,
+      "loss": 0.7362,
+      "step": 931
+    },
+    {
+      "epoch": 4.860495436766623,
+      "grad_norm": 0.09910390005741515,
+      "learning_rate": 1.4143047253120322e-07,
+      "loss": 0.7332,
+      "step": 932
+    },
+    {
+      "epoch": 4.865710560625815,
+      "grad_norm": 0.0900655265782037,
+      "learning_rate": 1.2940601634642059e-07,
+      "loss": 0.7233,
+      "step": 933
+    },
+    {
+      "epoch": 4.870925684485006,
+      "grad_norm": 0.09409378931281938,
+      "learning_rate": 1.1791485286227311e-07,
+      "loss": 0.7395,
+      "step": 934
+    },
+    {
+      "epoch": 4.876140808344198,
+      "grad_norm": 0.10015850940263663,
+      "learning_rate": 1.0695713577984824e-07,
+      "loss": 0.7231,
+      "step": 935
+    },
+    {
+      "epoch": 4.88135593220339,
+      "grad_norm": 0.0903358412925811,
+      "learning_rate": 9.653301166507422e-08,
+      "loss": 0.7351,
+      "step": 936
+    },
+    {
+      "epoch": 4.886571056062581,
+      "grad_norm": 0.0902382557022594,
+      "learning_rate": 8.664261994675738e-08,
+      "loss": 0.7333,
+      "step": 937
+    },
+    {
+      "epoch": 4.891786179921773,
+      "grad_norm": 0.09428682569360029,
+      "learning_rate": 7.728609291471678e-08,
+      "loss": 0.7353,
+      "step": 938
+    },
+    {
+      "epoch": 4.897001303780964,
+      "grad_norm": 0.09406554806150552,
+      "learning_rate": 6.846355571801688e-08,
+      "loss": 0.7387,
+      "step": 939
+    },
+    {
+      "epoch": 4.902216427640156,
+      "grad_norm": 0.10257884244588195,
+      "learning_rate": 6.017512636329325e-08,
+      "loss": 0.7327,
+      "step": 940
+    },
+    {
+      "epoch": 4.907431551499348,
+      "grad_norm": 0.08993907897520984,
+      "learning_rate": 5.242091571318053e-08,
+      "loss": 0.7323,
+      "step": 941
+    },
+    {
+      "epoch": 4.912646675358539,
+      "grad_norm": 0.09752580757548222,
+      "learning_rate": 4.5201027484811365e-08,
+      "loss": 0.7369,
+      "step": 942
+    },
+    {
+      "epoch": 4.917861799217731,
+      "grad_norm": 0.09295306846482325,
+      "learning_rate": 3.851555824844866e-08,
+      "loss": 0.7269,
+      "step": 943
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.09090791000944588,
+      "learning_rate": 3.23645974261888e-08,
+      "loss": 0.72,
+      "step": 944
+    },
+    {
+      "epoch": 4.9282920469361144,
+      "grad_norm": 0.0941176803792802,
+      "learning_rate": 2.674822729075377e-08,
+      "loss": 0.7302,
+      "step": 945
+    },
+    {
+      "epoch": 4.933507170795306,
+      "grad_norm": 0.09476591232745427,
+      "learning_rate": 2.166652296440752e-08,
+      "loss": 0.7389,
+      "step": 946
+    },
+    {
+      "epoch": 4.938722294654498,
+      "grad_norm": 0.09487039421752298,
+      "learning_rate": 1.7119552417943496e-08,
+      "loss": 0.7384,
+      "step": 947
+    },
+    {
+      "epoch": 4.9439374185136895,
+      "grad_norm": 0.09350946727423737,
+      "learning_rate": 1.3107376469769783e-08,
+      "loss": 0.7352,
+      "step": 948
+    },
+    {
+      "epoch": 4.9491525423728815,
+      "grad_norm": 0.09008974287657726,
+      "learning_rate": 9.630048785105318e-09,
+      "loss": 0.7202,
+      "step": 949
+    },
+    {
+      "epoch": 4.9543676662320735,
+      "grad_norm": 0.09424907250830968,
+      "learning_rate": 6.687615875264897e-09,
+      "loss": 0.7352,
+      "step": 950
+    },
+    {
+      "epoch": 4.959582790091265,
+      "grad_norm": 0.08996524126321254,
+      "learning_rate": 4.280117097015257e-09,
+      "loss": 0.7342,
+      "step": 951
+    },
+    {
+      "epoch": 4.9647979139504566,
+      "grad_norm": 0.09830845644976817,
+      "learning_rate": 2.407584652073247e-09,
+      "loss": 0.7357,
+      "step": 952
+    },
+    {
+      "epoch": 4.970013037809648,
+      "grad_norm": 0.0969409044778751,
+      "learning_rate": 1.0700435866706216e-09,
+      "loss": 0.7212,
+      "step": 953
+    },
+    {
+      "epoch": 4.97522816166884,
+      "grad_norm": 0.09285007352436725,
+      "learning_rate": 2.6751179119877124e-10,
+      "loss": 0.7353,
+      "step": 954
+    },
+    {
+      "epoch": 4.980443285528032,
+      "grad_norm": 0.09324700428805709,
+      "learning_rate": 0.0,
+      "loss": 0.7386,
+      "step": 955
+    },
+    {
+      "epoch": 4.980443285528032,
+      "step": 955,
+      "total_flos": 1.9663195796108476e+19,
+      "train_loss": 0.7990808349005215,
+      "train_runtime": 49935.3477,
+      "train_samples_per_second": 9.826,
+      "train_steps_per_second": 0.019
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 955,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "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": 1.9663195796108476e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}