diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/0_trainer_state.json b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/0_trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..6025478b3bec60311c264b70d834225d3af06f74
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/0_trainer_state.json
@@ -0,0 +1,7526 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2498,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0008006405124099279,
+      "learning_rate": 2.415943612351265e-06,
+      "loss": 0.2467,
+      "step": 2
+    },
+    {
+      "epoch": 0.0016012810248198558,
+      "learning_rate": 2.4341906163790364e-06,
+      "loss": 0.0573,
+      "step": 4
+    },
+    {
+      "epoch": 0.0024019215372297837,
+      "learning_rate": 2.4524967251364995e-06,
+      "loss": 0.1694,
+      "step": 6
+    },
+    {
+      "epoch": 0.0032025620496397116,
+      "learning_rate": 2.4708617956148052e-06,
+      "loss": 0.2059,
+      "step": 8
+    },
+    {
+      "epoch": 0.0040032025620496394,
+      "learning_rate": 2.4892856843445236e-06,
+      "loss": 0.2581,
+      "step": 10
+    },
+    {
+      "epoch": 0.004803843074459567,
+      "learning_rate": 2.507768247396697e-06,
+      "loss": 0.0719,
+      "step": 12
+    },
+    {
+      "epoch": 0.005604483586869495,
+      "learning_rate": 2.5263093403840022e-06,
+      "loss": 0.3515,
+      "step": 14
+    },
+    {
+      "epoch": 0.006405124099279423,
+      "learning_rate": 2.5449088184619065e-06,
+      "loss": 0.0693,
+      "step": 16
+    },
+    {
+      "epoch": 0.007205764611689352,
+      "learning_rate": 2.5635665363297356e-06,
+      "loss": 0.1825,
+      "step": 18
+    },
+    {
+      "epoch": 0.008006405124099279,
+      "learning_rate": 2.5822823482318517e-06,
+      "loss": 0.1901,
+      "step": 20
+    },
+    {
+      "epoch": 0.008807045636509208,
+      "learning_rate": 2.6010561079587694e-06,
+      "loss": 0.6459,
+      "step": 22
+    },
+    {
+      "epoch": 0.009607686148919135,
+      "learning_rate": 2.6198876688483453e-06,
+      "loss": 0.1767,
+      "step": 24
+    },
+    {
+      "epoch": 0.010408326661329063,
+      "learning_rate": 2.6387768837868565e-06,
+      "loss": 0.4669,
+      "step": 26
+    },
+    {
+      "epoch": 0.01120896717373899,
+      "learning_rate": 2.6577236052101764e-06,
+      "loss": 0.1539,
+      "step": 28
+    },
+    {
+      "epoch": 0.01200960768614892,
+      "learning_rate": 2.6767276851049716e-06,
+      "loss": 0.1946,
+      "step": 30
+    },
+    {
+      "epoch": 0.012810248198558846,
+      "learning_rate": 2.6957889750097866e-06,
+      "loss": 0.2724,
+      "step": 32
+    },
+    {
+      "epoch": 0.013610888710968775,
+      "learning_rate": 2.7149073260162416e-06,
+      "loss": 0.0462,
+      "step": 34
+    },
+    {
+      "epoch": 0.014411529223378704,
+      "learning_rate": 2.7340825887701848e-06,
+      "loss": 0.0152,
+      "step": 36
+    },
+    {
+      "epoch": 0.01521216973578863,
+      "learning_rate": 2.7533146134728993e-06,
+      "loss": 0.1016,
+      "step": 38
+    },
+    {
+      "epoch": 0.016012810248198558,
+      "learning_rate": 2.772603249882202e-06,
+      "loss": 0.048,
+      "step": 40
+    },
+    {
+      "epoch": 0.016813450760608487,
+      "learning_rate": 2.7919483473136555e-06,
+      "loss": 0.1376,
+      "step": 42
+    },
+    {
+      "epoch": 0.017614091273018415,
+      "learning_rate": 2.81134975464178e-06,
+      "loss": 0.034,
+      "step": 44
+    },
+    {
+      "epoch": 0.018414731785428344,
+      "learning_rate": 2.8308073203011634e-06,
+      "loss": 0.1538,
+      "step": 46
+    },
+    {
+      "epoch": 0.01921537229783827,
+      "learning_rate": 2.850320892287688e-06,
+      "loss": 0.1127,
+      "step": 48
+    },
+    {
+      "epoch": 0.020016012810248198,
+      "learning_rate": 2.8698903181597026e-06,
+      "loss": 0.175,
+      "step": 50
+    },
+    {
+      "epoch": 0.020816653322658127,
+      "learning_rate": 2.889515445039256e-06,
+      "loss": 0.0457,
+      "step": 52
+    },
+    {
+      "epoch": 0.021617293835068056,
+      "learning_rate": 2.909196119613218e-06,
+      "loss": 0.4392,
+      "step": 54
+    },
+    {
+      "epoch": 0.02241793434747798,
+      "learning_rate": 2.928932188134529e-06,
+      "loss": 0.0658,
+      "step": 56
+    },
+    {
+      "epoch": 0.02321857485988791,
+      "learning_rate": 2.9487234964233724e-06,
+      "loss": 0.1111,
+      "step": 58
+    },
+    {
+      "epoch": 0.02401921537229784,
+      "learning_rate": 2.9685698898684355e-06,
+      "loss": 0.0388,
+      "step": 60
+    },
+    {
+      "epoch": 0.024819855884707767,
+      "learning_rate": 2.988471213428035e-06,
+      "loss": 0.1192,
+      "step": 62
+    },
+    {
+      "epoch": 0.025620496397117692,
+      "learning_rate": 3.00842731163137e-06,
+      "loss": 0.3528,
+      "step": 64
+    },
+    {
+      "epoch": 0.02642113690952762,
+      "learning_rate": 3.0284380285797733e-06,
+      "loss": 0.1812,
+      "step": 66
+    },
+    {
+      "epoch": 0.02722177742193755,
+      "learning_rate": 3.048503207947854e-06,
+      "loss": 0.0245,
+      "step": 68
+    },
+    {
+      "epoch": 0.02802241793434748,
+      "learning_rate": 3.068622692984767e-06,
+      "loss": 0.1994,
+      "step": 70
+    },
+    {
+      "epoch": 0.028823058446757407,
+      "learning_rate": 3.0887963265154187e-06,
+      "loss": 0.0103,
+      "step": 72
+    },
+    {
+      "epoch": 0.029623698959167333,
+      "learning_rate": 3.1090239509417364e-06,
+      "loss": 0.0254,
+      "step": 74
+    },
+    {
+      "epoch": 0.03042433947157726,
+      "learning_rate": 3.129305408243829e-06,
+      "loss": 0.0651,
+      "step": 76
+    },
+    {
+      "epoch": 0.03122497998398719,
+      "learning_rate": 3.1496405399812602e-06,
+      "loss": 0.0946,
+      "step": 78
+    },
+    {
+      "epoch": 0.032025620496397116,
+      "learning_rate": 3.17002918729432e-06,
+      "loss": 0.0301,
+      "step": 80
+    },
+    {
+      "epoch": 0.03282626100880705,
+      "learning_rate": 3.1904711909051967e-06,
+      "loss": 0.4394,
+      "step": 82
+    },
+    {
+      "epoch": 0.03362690152121697,
+      "learning_rate": 3.2109663911192622e-06,
+      "loss": 0.1513,
+      "step": 84
+    },
+    {
+      "epoch": 0.0344275420336269,
+      "learning_rate": 3.231514627826302e-06,
+      "loss": 0.112,
+      "step": 86
+    },
+    {
+      "epoch": 0.03522818254603683,
+      "learning_rate": 3.2521157405018146e-06,
+      "loss": 0.0642,
+      "step": 88
+    },
+    {
+      "epoch": 0.036028823058446756,
+      "learning_rate": 3.2727695682081897e-06,
+      "loss": 0.1087,
+      "step": 90
+    },
+    {
+      "epoch": 0.03682946357085669,
+      "learning_rate": 3.293475949595998e-06,
+      "loss": 0.3814,
+      "step": 92
+    },
+    {
+      "epoch": 0.03763010408326661,
+      "learning_rate": 3.314234722905302e-06,
+      "loss": 0.0535,
+      "step": 94
+    },
+    {
+      "epoch": 0.03843074459567654,
+      "learning_rate": 3.335045725966829e-06,
+      "loss": 0.3592,
+      "step": 96
+    },
+    {
+      "epoch": 0.03923138510808647,
+      "learning_rate": 3.355908796203301e-06,
+      "loss": 0.0818,
+      "step": 98
+    },
+    {
+      "epoch": 0.040032025620496396,
+      "learning_rate": 3.3768237706306716e-06,
+      "loss": 0.1316,
+      "step": 100
+    },
+    {
+      "epoch": 0.04083266613290633,
+      "learning_rate": 3.3977904858594534e-06,
+      "loss": 0.1642,
+      "step": 102
+    },
+    {
+      "epoch": 0.041633306645316254,
+      "learning_rate": 3.418808778095917e-06,
+      "loss": 0.1075,
+      "step": 104
+    },
+    {
+      "epoch": 0.04243394715772618,
+      "learning_rate": 3.4398784831434097e-06,
+      "loss": 0.3196,
+      "step": 106
+    },
+    {
+      "epoch": 0.04323458767013611,
+      "learning_rate": 3.460999436403676e-06,
+      "loss": 0.1469,
+      "step": 108
+    },
+    {
+      "epoch": 0.044035228182546036,
+      "learning_rate": 3.4821714728780654e-06,
+      "loss": 0.1458,
+      "step": 110
+    },
+    {
+      "epoch": 0.04483586869495596,
+      "learning_rate": 3.5033944271688624e-06,
+      "loss": 0.0183,
+      "step": 112
+    },
+    {
+      "epoch": 0.045636509207365894,
+      "learning_rate": 3.5246681334806177e-06,
+      "loss": 0.1096,
+      "step": 114
+    },
+    {
+      "epoch": 0.04643714971977582,
+      "learning_rate": 3.5459924256213596e-06,
+      "loss": 0.2373,
+      "step": 116
+    },
+    {
+      "epoch": 0.04723779023218575,
+      "learning_rate": 3.567367137003953e-06,
+      "loss": 0.9613,
+      "step": 118
+    },
+    {
+      "epoch": 0.04803843074459568,
+      "learning_rate": 3.588792100647368e-06,
+      "loss": 0.1215,
+      "step": 120
+    },
+    {
+      "epoch": 0.0488390712570056,
+      "learning_rate": 3.6102671491780393e-06,
+      "loss": 0.2998,
+      "step": 122
+    },
+    {
+      "epoch": 0.049639711769415534,
+      "learning_rate": 3.6317921148310965e-06,
+      "loss": 0.0343,
+      "step": 124
+    },
+    {
+      "epoch": 0.05044035228182546,
+      "learning_rate": 3.653366829451711e-06,
+      "loss": 0.256,
+      "step": 126
+    },
+    {
+      "epoch": 0.051240992794235385,
+      "learning_rate": 3.674991124496452e-06,
+      "loss": 0.0248,
+      "step": 128
+    },
+    {
+      "epoch": 0.05204163330664532,
+      "learning_rate": 3.696664831034521e-06,
+      "loss": 0.1913,
+      "step": 130
+    },
+    {
+      "epoch": 0.05284227381905524,
+      "learning_rate": 3.7183877797491143e-06,
+      "loss": 0.1421,
+      "step": 132
+    },
+    {
+      "epoch": 0.053642914331465175,
+      "learning_rate": 3.740159800938784e-06,
+      "loss": 0.0973,
+      "step": 134
+    },
+    {
+      "epoch": 0.0544435548438751,
+      "learning_rate": 3.7619807245186824e-06,
+      "loss": 0.008,
+      "step": 136
+    },
+    {
+      "epoch": 0.055244195356285025,
+      "learning_rate": 3.783850380021933e-06,
+      "loss": 0.1147,
+      "step": 138
+    },
+    {
+      "epoch": 0.05604483586869496,
+      "learning_rate": 3.8057685966010025e-06,
+      "loss": 0.6627,
+      "step": 140
+    },
+    {
+      "epoch": 0.05684547638110488,
+      "learning_rate": 3.827735203028956e-06,
+      "loss": 0.2326,
+      "step": 142
+    },
+    {
+      "epoch": 0.057646116893514815,
+      "learning_rate": 3.849750027700842e-06,
+      "loss": 0.0308,
+      "step": 144
+    },
+    {
+      "epoch": 0.05844675740592474,
+      "learning_rate": 3.871812898635011e-06,
+      "loss": 0.0637,
+      "step": 146
+    },
+    {
+      "epoch": 0.059247397918334666,
+      "learning_rate": 3.8939236434745184e-06,
+      "loss": 0.1526,
+      "step": 148
+    },
+    {
+      "epoch": 0.0600480384307446,
+      "learning_rate": 3.916082089488379e-06,
+      "loss": 0.0016,
+      "step": 150
+    },
+    {
+      "epoch": 0.06084867894315452,
+      "learning_rate": 3.938288063572962e-06,
+      "loss": 0.0926,
+      "step": 152
+    },
+    {
+      "epoch": 0.06164931945556445,
+      "learning_rate": 3.960541392253387e-06,
+      "loss": 0.1669,
+      "step": 154
+    },
+    {
+      "epoch": 0.06244995996797438,
+      "learning_rate": 3.982841901684792e-06,
+      "loss": 0.2158,
+      "step": 156
+    },
+    {
+      "epoch": 0.0632506004803843,
+      "learning_rate": 4.005189417653737e-06,
+      "loss": 0.148,
+      "step": 158
+    },
+    {
+      "epoch": 0.06405124099279423,
+      "learning_rate": 4.027583765579601e-06,
+      "loss": 0.0577,
+      "step": 160
+    },
+    {
+      "epoch": 0.06485188150520416,
+      "learning_rate": 4.050024770515873e-06,
+      "loss": 0.1165,
+      "step": 162
+    },
+    {
+      "epoch": 0.0656525220176141,
+      "learning_rate": 4.072512257151546e-06,
+      "loss": 0.1286,
+      "step": 164
+    },
+    {
+      "epoch": 0.06645316253002402,
+      "learning_rate": 4.095046049812541e-06,
+      "loss": 0.1151,
+      "step": 166
+    },
+    {
+      "epoch": 0.06725380304243395,
+      "learning_rate": 4.117625972462988e-06,
+      "loss": 0.0402,
+      "step": 168
+    },
+    {
+      "epoch": 0.06805444355484387,
+      "learning_rate": 4.1402518487066624e-06,
+      "loss": 0.2551,
+      "step": 170
+    },
+    {
+      "epoch": 0.0688550840672538,
+      "learning_rate": 4.1629235017883285e-06,
+      "loss": 0.4012,
+      "step": 172
+    },
+    {
+      "epoch": 0.06965572457966374,
+      "learning_rate": 4.1856407545951825e-06,
+      "loss": 0.2387,
+      "step": 174
+    },
+    {
+      "epoch": 0.07045636509207366,
+      "learning_rate": 4.208403429658151e-06,
+      "loss": 0.024,
+      "step": 176
+    },
+    {
+      "epoch": 0.07125700560448359,
+      "learning_rate": 4.2312113491533145e-06,
+      "loss": 0.1358,
+      "step": 178
+    },
+    {
+      "epoch": 0.07205764611689351,
+      "learning_rate": 4.254064334903347e-06,
+      "loss": 0.0094,
+      "step": 180
+    },
+    {
+      "epoch": 0.07285828662930344,
+      "learning_rate": 4.276962208378814e-06,
+      "loss": 0.1583,
+      "step": 182
+    },
+    {
+      "epoch": 0.07365892714171338,
+      "learning_rate": 4.299904790699619e-06,
+      "loss": 0.1238,
+      "step": 184
+    },
+    {
+      "epoch": 0.0744595676541233,
+      "learning_rate": 4.3228919026364345e-06,
+      "loss": 0.4382,
+      "step": 186
+    },
+    {
+      "epoch": 0.07526020816653323,
+      "learning_rate": 4.345923364612024e-06,
+      "loss": 0.3528,
+      "step": 188
+    },
+    {
+      "epoch": 0.07606084867894315,
+      "learning_rate": 4.368998996702686e-06,
+      "loss": 0.3852,
+      "step": 190
+    },
+    {
+      "epoch": 0.07686148919135308,
+      "learning_rate": 4.392118618639698e-06,
+      "loss": 0.0947,
+      "step": 192
+    },
+    {
+      "epoch": 0.07766212970376302,
+      "learning_rate": 4.415282049810643e-06,
+      "loss": 0.0537,
+      "step": 194
+    },
+    {
+      "epoch": 0.07846277021617294,
+      "learning_rate": 4.4384891092608795e-06,
+      "loss": 0.0766,
+      "step": 196
+    },
+    {
+      "epoch": 0.07926341072858287,
+      "learning_rate": 4.461739615694921e-06,
+      "loss": 0.0362,
+      "step": 198
+    },
+    {
+      "epoch": 0.08006405124099279,
+      "learning_rate": 4.485033387477915e-06,
+      "loss": 0.4005,
+      "step": 200
+    },
+    {
+      "epoch": 0.08086469175340272,
+      "learning_rate": 4.5083702426369715e-06,
+      "loss": 0.0108,
+      "step": 202
+    },
+    {
+      "epoch": 0.08166533226581266,
+      "learning_rate": 4.531749998862628e-06,
+      "loss": 0.3322,
+      "step": 204
+    },
+    {
+      "epoch": 0.08246597277822258,
+      "learning_rate": 4.555172473510324e-06,
+      "loss": 0.3022,
+      "step": 206
+    },
+    {
+      "epoch": 0.08326661329063251,
+      "learning_rate": 4.578637483601732e-06,
+      "loss": 0.1124,
+      "step": 208
+    },
+    {
+      "epoch": 0.08406725380304243,
+      "learning_rate": 4.602144845826234e-06,
+      "loss": 0.5849,
+      "step": 210
+    },
+    {
+      "epoch": 0.08486789431545236,
+      "learning_rate": 4.625694376542399e-06,
+      "loss": 0.0004,
+      "step": 212
+    },
+    {
+      "epoch": 0.08566853482786228,
+      "learning_rate": 4.649285891779326e-06,
+      "loss": 0.3145,
+      "step": 214
+    },
+    {
+      "epoch": 0.08646917534027222,
+      "learning_rate": 4.672919207238145e-06,
+      "loss": 0.2548,
+      "step": 216
+    },
+    {
+      "epoch": 0.08726981585268215,
+      "learning_rate": 4.696594138293421e-06,
+      "loss": 0.079,
+      "step": 218
+    },
+    {
+      "epoch": 0.08807045636509207,
+      "learning_rate": 4.720310499994664e-06,
+      "loss": 0.4485,
+      "step": 220
+    },
+    {
+      "epoch": 0.088871096877502,
+      "learning_rate": 4.744068107067673e-06,
+      "loss": 0.1557,
+      "step": 222
+    },
+    {
+      "epoch": 0.08967173738991192,
+      "learning_rate": 4.767866773916041e-06,
+      "loss": 0.1359,
+      "step": 224
+    },
+    {
+      "epoch": 0.09047237790232186,
+      "learning_rate": 4.79170631462264e-06,
+      "loss": 0.3935,
+      "step": 226
+    },
+    {
+      "epoch": 0.09127301841473179,
+      "learning_rate": 4.81558654295099e-06,
+      "loss": 0.1582,
+      "step": 228
+    },
+    {
+      "epoch": 0.09207365892714171,
+      "learning_rate": 4.839507272346751e-06,
+      "loss": 0.1028,
+      "step": 230
+    },
+    {
+      "epoch": 0.09287429943955164,
+      "learning_rate": 4.863468315939234e-06,
+      "loss": 0.0717,
+      "step": 232
+    },
+    {
+      "epoch": 0.09367493995196156,
+      "learning_rate": 4.8874694865427676e-06,
+      "loss": 0.3559,
+      "step": 234
+    },
+    {
+      "epoch": 0.0944755804643715,
+      "learning_rate": 4.911510596658202e-06,
+      "loss": 0.1527,
+      "step": 236
+    },
+    {
+      "epoch": 0.09527622097678143,
+      "learning_rate": 4.935591458474425e-06,
+      "loss": 0.1192,
+      "step": 238
+    },
+    {
+      "epoch": 0.09607686148919135,
+      "learning_rate": 4.959711883869734e-06,
+      "loss": 0.2082,
+      "step": 240
+    },
+    {
+      "epoch": 0.09687750200160128,
+      "learning_rate": 4.9838716844133665e-06,
+      "loss": 0.0934,
+      "step": 242
+    },
+    {
+      "epoch": 0.0976781425140112,
+      "learning_rate": 5.0080706713669435e-06,
+      "loss": 0.7068,
+      "step": 244
+    },
+    {
+      "epoch": 0.09847878302642114,
+      "learning_rate": 5.032308655686007e-06,
+      "loss": 0.1441,
+      "step": 246
+    },
+    {
+      "epoch": 0.09927942353883107,
+      "learning_rate": 5.056585448021398e-06,
+      "loss": 0.2424,
+      "step": 248
+    },
+    {
+      "epoch": 0.100080064051241,
+      "learning_rate": 5.080900858720789e-06,
+      "loss": 0.562,
+      "step": 250
+    },
+    {
+      "epoch": 0.10088070456365092,
+      "learning_rate": 5.105254697830208e-06,
+      "loss": 0.2064,
+      "step": 252
+    },
+    {
+      "epoch": 0.10168134507606084,
+      "learning_rate": 5.129646775095432e-06,
+      "loss": 0.1076,
+      "step": 254
+    },
+    {
+      "epoch": 0.10248198558847077,
+      "learning_rate": 5.154076899963514e-06,
+      "loss": 0.3241,
+      "step": 256
+    },
+    {
+      "epoch": 0.10328262610088071,
+      "learning_rate": 5.178544881584328e-06,
+      "loss": 0.1282,
+      "step": 258
+    },
+    {
+      "epoch": 0.10408326661329063,
+      "learning_rate": 5.203050528811959e-06,
+      "loss": 0.2354,
+      "step": 260
+    },
+    {
+      "epoch": 0.10488390712570056,
+      "learning_rate": 5.227593650206246e-06,
+      "loss": 0.2725,
+      "step": 262
+    },
+    {
+      "epoch": 0.10568454763811048,
+      "learning_rate": 5.2521740540343205e-06,
+      "loss": 0.096,
+      "step": 264
+    },
+    {
+      "epoch": 0.10648518815052041,
+      "learning_rate": 5.2767915482720164e-06,
+      "loss": 0.2634,
+      "step": 266
+    },
+    {
+      "epoch": 0.10728582866293035,
+      "learning_rate": 5.3014459406054295e-06,
+      "loss": 0.2161,
+      "step": 268
+    },
+    {
+      "epoch": 0.10808646917534027,
+      "learning_rate": 5.3261370384323904e-06,
+      "loss": 0.0682,
+      "step": 270
+    },
+    {
+      "epoch": 0.1088871096877502,
+      "learning_rate": 5.350864648864026e-06,
+      "loss": 0.0461,
+      "step": 272
+    },
+    {
+      "epoch": 0.10968775020016013,
+      "learning_rate": 5.375628578726181e-06,
+      "loss": 0.0681,
+      "step": 274
+    },
+    {
+      "epoch": 0.11048839071257005,
+      "learning_rate": 5.4004286345609665e-06,
+      "loss": 0.1726,
+      "step": 276
+    },
+    {
+      "epoch": 0.11128903122497999,
+      "learning_rate": 5.425264622628326e-06,
+      "loss": 0.0297,
+      "step": 278
+    },
+    {
+      "epoch": 0.11208967173738991,
+      "learning_rate": 5.450136348907444e-06,
+      "loss": 0.3649,
+      "step": 280
+    },
+    {
+      "epoch": 0.11289031224979984,
+      "learning_rate": 5.475043619098321e-06,
+      "loss": 0.1461,
+      "step": 282
+    },
+    {
+      "epoch": 0.11369095276220977,
+      "learning_rate": 5.499986238623329e-06,
+      "loss": 0.1071,
+      "step": 284
+    },
+    {
+      "epoch": 0.11449159327461969,
+      "learning_rate": 5.524964012628644e-06,
+      "loss": 0.0333,
+      "step": 286
+    },
+    {
+      "epoch": 0.11529223378702963,
+      "learning_rate": 5.549976745985809e-06,
+      "loss": 0.0659,
+      "step": 288
+    },
+    {
+      "epoch": 0.11609287429943956,
+      "learning_rate": 5.57502424329331e-06,
+      "loss": 0.1361,
+      "step": 290
+    },
+    {
+      "epoch": 0.11689351481184948,
+      "learning_rate": 5.6001063088780085e-06,
+      "loss": 0.6853,
+      "step": 292
+    },
+    {
+      "epoch": 0.1176941553242594,
+      "learning_rate": 5.62522274679673e-06,
+      "loss": 0.9823,
+      "step": 294
+    },
+    {
+      "epoch": 0.11849479583666933,
+      "learning_rate": 5.650373360837763e-06,
+      "loss": 0.2493,
+      "step": 296
+    },
+    {
+      "epoch": 0.11929543634907927,
+      "learning_rate": 5.675557954522462e-06,
+      "loss": 0.016,
+      "step": 298
+    },
+    {
+      "epoch": 0.1200960768614892,
+      "learning_rate": 5.700776331106674e-06,
+      "loss": 0.0424,
+      "step": 300
+    },
+    {
+      "epoch": 0.12089671737389912,
+      "learning_rate": 5.726028293582342e-06,
+      "loss": 0.0034,
+      "step": 302
+    },
+    {
+      "epoch": 0.12169735788630905,
+      "learning_rate": 5.751313644679071e-06,
+      "loss": 0.0171,
+      "step": 304
+    },
+    {
+      "epoch": 0.12249799839871897,
+      "learning_rate": 5.776632186865589e-06,
+      "loss": 0.0232,
+      "step": 306
+    },
+    {
+      "epoch": 0.1232986389111289,
+      "learning_rate": 5.8019837223513295e-06,
+      "loss": 0.1032,
+      "step": 308
+    },
+    {
+      "epoch": 0.12409927942353884,
+      "learning_rate": 5.827368053088032e-06,
+      "loss": 0.0758,
+      "step": 310
+    },
+    {
+      "epoch": 0.12489991993594876,
+      "learning_rate": 5.852784980771182e-06,
+      "loss": 0.0282,
+      "step": 312
+    },
+    {
+      "epoch": 0.1257005604483587,
+      "learning_rate": 5.878234306841637e-06,
+      "loss": 0.5472,
+      "step": 314
+    },
+    {
+      "epoch": 0.1265012009607686,
+      "learning_rate": 5.903715832487138e-06,
+      "loss": 0.0415,
+      "step": 316
+    },
+    {
+      "epoch": 0.12730184147317855,
+      "learning_rate": 5.929229358643925e-06,
+      "loss": 0.2705,
+      "step": 318
+    },
+    {
+      "epoch": 0.12810248198558846,
+      "learning_rate": 5.954774685998206e-06,
+      "loss": 0.11,
+      "step": 320
+    },
+    {
+      "epoch": 0.1289031224979984,
+      "learning_rate": 5.9803516149877475e-06,
+      "loss": 0.2075,
+      "step": 322
+    },
+    {
+      "epoch": 0.1297037630104083,
+      "learning_rate": 6.005959945803494e-06,
+      "loss": 0.0062,
+      "step": 324
+    },
+    {
+      "epoch": 0.13050440352281825,
+      "learning_rate": 6.03159947839103e-06,
+      "loss": 0.1936,
+      "step": 326
+    },
+    {
+      "epoch": 0.1313050440352282,
+      "learning_rate": 6.057270012452186e-06,
+      "loss": 0.2279,
+      "step": 328
+    },
+    {
+      "epoch": 0.1321056845476381,
+      "learning_rate": 6.082971347446654e-06,
+      "loss": 0.565,
+      "step": 330
+    },
+    {
+      "epoch": 0.13290632506004804,
+      "learning_rate": 6.108703282593461e-06,
+      "loss": 0.0678,
+      "step": 332
+    },
+    {
+      "epoch": 0.13370696557245795,
+      "learning_rate": 6.13446561687258e-06,
+      "loss": 0.0736,
+      "step": 334
+    },
+    {
+      "epoch": 0.1345076060848679,
+      "learning_rate": 6.160258149026557e-06,
+      "loss": 0.0189,
+      "step": 336
+    },
+    {
+      "epoch": 0.13530824659727783,
+      "learning_rate": 6.186080677561974e-06,
+      "loss": 0.3389,
+      "step": 338
+    },
+    {
+      "epoch": 0.13610888710968774,
+      "learning_rate": 6.2119330007511014e-06,
+      "loss": 0.037,
+      "step": 340
+    },
+    {
+      "epoch": 0.13690952762209768,
+      "learning_rate": 6.237814916633431e-06,
+      "loss": 0.1,
+      "step": 342
+    },
+    {
+      "epoch": 0.1377101681345076,
+      "learning_rate": 6.263726223017326e-06,
+      "loss": 0.0294,
+      "step": 344
+    },
+    {
+      "epoch": 0.13851080864691753,
+      "learning_rate": 6.289666717481496e-06,
+      "loss": 0.2595,
+      "step": 346
+    },
+    {
+      "epoch": 0.13931144915932747,
+      "learning_rate": 6.315636197376634e-06,
+      "loss": 0.04,
+      "step": 348
+    },
+    {
+      "epoch": 0.14011208967173738,
+      "learning_rate": 6.341634459827044e-06,
+      "loss": 0.0864,
+      "step": 350
+    },
+    {
+      "epoch": 0.14091273018414732,
+      "learning_rate": 6.3676613017321305e-06,
+      "loss": 0.101,
+      "step": 352
+    },
+    {
+      "epoch": 0.14171337069655723,
+      "learning_rate": 6.393716519768032e-06,
+      "loss": 0.0951,
+      "step": 354
+    },
+    {
+      "epoch": 0.14251401120896717,
+      "learning_rate": 6.419799910389257e-06,
+      "loss": 0.0108,
+      "step": 356
+    },
+    {
+      "epoch": 0.1433146517213771,
+      "learning_rate": 6.445911269830183e-06,
+      "loss": 0.1673,
+      "step": 358
+    },
+    {
+      "epoch": 0.14411529223378702,
+      "learning_rate": 6.472050394106689e-06,
+      "loss": 0.1196,
+      "step": 360
+    },
+    {
+      "epoch": 0.14491593274619696,
+      "learning_rate": 6.498217079017806e-06,
+      "loss": 0.1911,
+      "step": 362
+    },
+    {
+      "epoch": 0.14571657325860687,
+      "learning_rate": 6.524411120147204e-06,
+      "loss": 0.1379,
+      "step": 364
+    },
+    {
+      "epoch": 0.1465172137710168,
+      "learning_rate": 6.5506323128648654e-06,
+      "loss": 0.1161,
+      "step": 366
+    },
+    {
+      "epoch": 0.14731785428342675,
+      "learning_rate": 6.576880452328645e-06,
+      "loss": 0.399,
+      "step": 368
+    },
+    {
+      "epoch": 0.14811849479583666,
+      "learning_rate": 6.603155333485934e-06,
+      "loss": 0.1525,
+      "step": 370
+    },
+    {
+      "epoch": 0.1489191353082466,
+      "learning_rate": 6.6294567510751675e-06,
+      "loss": 0.3517,
+      "step": 372
+    },
+    {
+      "epoch": 0.14971977582065651,
+      "learning_rate": 6.655784499627476e-06,
+      "loss": 0.3868,
+      "step": 374
+    },
+    {
+      "epoch": 0.15052041633306645,
+      "learning_rate": 6.682138373468341e-06,
+      "loss": 0.0292,
+      "step": 376
+    },
+    {
+      "epoch": 0.1513210568454764,
+      "learning_rate": 6.7085181667191e-06,
+      "loss": 0.0172,
+      "step": 378
+    },
+    {
+      "epoch": 0.1521216973578863,
+      "learning_rate": 6.734923673298605e-06,
+      "loss": 0.1275,
+      "step": 380
+    },
+    {
+      "epoch": 0.15292233787029624,
+      "learning_rate": 6.761354686924883e-06,
+      "loss": 0.1193,
+      "step": 382
+    },
+    {
+      "epoch": 0.15372297838270615,
+      "learning_rate": 6.787811001116654e-06,
+      "loss": 0.2931,
+      "step": 384
+    },
+    {
+      "epoch": 0.1545236188951161,
+      "learning_rate": 6.8142924091949955e-06,
+      "loss": 0.6562,
+      "step": 386
+    },
+    {
+      "epoch": 0.15532425940752603,
+      "learning_rate": 6.840798704284939e-06,
+      "loss": 0.13,
+      "step": 388
+    },
+    {
+      "epoch": 0.15612489991993594,
+      "learning_rate": 6.867329679317144e-06,
+      "loss": 0.6252,
+      "step": 390
+    },
+    {
+      "epoch": 0.15692554043234588,
+      "learning_rate": 6.893885127029419e-06,
+      "loss": 0.2444,
+      "step": 392
+    },
+    {
+      "epoch": 0.1577261809447558,
+      "learning_rate": 6.920464839968391e-06,
+      "loss": 0.0536,
+      "step": 394
+    },
+    {
+      "epoch": 0.15852682145716573,
+      "learning_rate": 6.94706861049117e-06,
+      "loss": 0.2237,
+      "step": 396
+    },
+    {
+      "epoch": 0.15932746196957567,
+      "learning_rate": 6.973696230766884e-06,
+      "loss": 0.397,
+      "step": 398
+    },
+    {
+      "epoch": 0.16012810248198558,
+      "learning_rate": 7.000347492778341e-06,
+      "loss": 0.4228,
+      "step": 400
+    },
+    {
+      "epoch": 0.16092874299439552,
+      "learning_rate": 7.027022188323704e-06,
+      "loss": 0.2557,
+      "step": 402
+    },
+    {
+      "epoch": 0.16172938350680544,
+      "learning_rate": 7.05372010901803e-06,
+      "loss": 0.0858,
+      "step": 404
+    },
+    {
+      "epoch": 0.16253002401921537,
+      "learning_rate": 7.080441046294945e-06,
+      "loss": 0.1755,
+      "step": 406
+    },
+    {
+      "epoch": 0.1633306645316253,
+      "learning_rate": 7.1071847914082605e-06,
+      "loss": 0.0102,
+      "step": 408
+    },
+    {
+      "epoch": 0.16413130504403523,
+      "learning_rate": 7.133951135433656e-06,
+      "loss": 0.06,
+      "step": 410
+    },
+    {
+      "epoch": 0.16493194555644516,
+      "learning_rate": 7.160739869270219e-06,
+      "loss": 0.2886,
+      "step": 412
+    },
+    {
+      "epoch": 0.16573258606885508,
+      "learning_rate": 7.18755078364214e-06,
+      "loss": 0.3412,
+      "step": 414
+    },
+    {
+      "epoch": 0.16653322658126501,
+      "learning_rate": 7.214383669100317e-06,
+      "loss": 0.2079,
+      "step": 416
+    },
+    {
+      "epoch": 0.16733386709367493,
+      "learning_rate": 7.241238316024064e-06,
+      "loss": 0.5073,
+      "step": 418
+    },
+    {
+      "epoch": 0.16813450760608487,
+      "learning_rate": 7.268114514622635e-06,
+      "loss": 0.0016,
+      "step": 420
+    },
+    {
+      "epoch": 0.1689351481184948,
+      "learning_rate": 7.2950120549369204e-06,
+      "loss": 0.0156,
+      "step": 422
+    },
+    {
+      "epoch": 0.16973578863090472,
+      "learning_rate": 7.321930726841144e-06,
+      "loss": 0.0325,
+      "step": 424
+    },
+    {
+      "epoch": 0.17053642914331466,
+      "learning_rate": 7.348870320044395e-06,
+      "loss": 0.1904,
+      "step": 426
+    },
+    {
+      "epoch": 0.17133706965572457,
+      "learning_rate": 7.375830624092336e-06,
+      "loss": 0.1332,
+      "step": 428
+    },
+    {
+      "epoch": 0.1721377101681345,
+      "learning_rate": 7.402811428368824e-06,
+      "loss": 0.2039,
+      "step": 430
+    },
+    {
+      "epoch": 0.17293835068054444,
+      "learning_rate": 7.429812522097613e-06,
+      "loss": 0.1391,
+      "step": 432
+    },
+    {
+      "epoch": 0.17373899119295436,
+      "learning_rate": 7.4568336943439055e-06,
+      "loss": 0.0247,
+      "step": 434
+    },
+    {
+      "epoch": 0.1745396317053643,
+      "learning_rate": 7.4838747340160475e-06,
+      "loss": 0.5593,
+      "step": 436
+    },
+    {
+      "epoch": 0.1753402722177742,
+      "learning_rate": 7.510935429867233e-06,
+      "loss": 0.5083,
+      "step": 438
+    },
+    {
+      "epoch": 0.17614091273018415,
+      "learning_rate": 7.538015570497046e-06,
+      "loss": 0.0494,
+      "step": 440
+    },
+    {
+      "epoch": 0.17694155324259409,
+      "learning_rate": 7.5651149443531846e-06,
+      "loss": 0.0339,
+      "step": 442
+    },
+    {
+      "epoch": 0.177742193755004,
+      "learning_rate": 7.592233339733077e-06,
+      "loss": 0.103,
+      "step": 444
+    },
+    {
+      "epoch": 0.17854283426741394,
+      "learning_rate": 7.619370544785608e-06,
+      "loss": 0.0907,
+      "step": 446
+    },
+    {
+      "epoch": 0.17934347477982385,
+      "learning_rate": 7.646526347512665e-06,
+      "loss": 0.1347,
+      "step": 448
+    },
+    {
+      "epoch": 0.1801441152922338,
+      "learning_rate": 7.67370053577085e-06,
+      "loss": 0.5048,
+      "step": 450
+    },
+    {
+      "epoch": 0.18094475580464373,
+      "learning_rate": 7.70089289727319e-06,
+      "loss": 0.0254,
+      "step": 452
+    },
+    {
+      "epoch": 0.18174539631705364,
+      "learning_rate": 7.728103219590684e-06,
+      "loss": 0.2381,
+      "step": 454
+    },
+    {
+      "epoch": 0.18254603682946358,
+      "learning_rate": 7.755331290154041e-06,
+      "loss": 0.1525,
+      "step": 456
+    },
+    {
+      "epoch": 0.1833466773418735,
+      "learning_rate": 7.7825768962553e-06,
+      "loss": 0.0364,
+      "step": 458
+    },
+    {
+      "epoch": 0.18414731785428343,
+      "learning_rate": 7.809839825049565e-06,
+      "loss": 0.4224,
+      "step": 460
+    },
+    {
+      "epoch": 0.18494795836669337,
+      "learning_rate": 7.83711986355656e-06,
+      "loss": 0.1089,
+      "step": 462
+    },
+    {
+      "epoch": 0.18574859887910328,
+      "learning_rate": 7.864416798662347e-06,
+      "loss": 0.4363,
+      "step": 464
+    },
+    {
+      "epoch": 0.18654923939151322,
+      "learning_rate": 7.891730417121043e-06,
+      "loss": 0.0321,
+      "step": 466
+    },
+    {
+      "epoch": 0.18734987990392313,
+      "learning_rate": 7.919060505556376e-06,
+      "loss": 0.008,
+      "step": 468
+    },
+    {
+      "epoch": 0.18815052041633307,
+      "learning_rate": 7.946406850463435e-06,
+      "loss": 0.1374,
+      "step": 470
+    },
+    {
+      "epoch": 0.188951160928743,
+      "learning_rate": 7.973769238210291e-06,
+      "loss": 0.2525,
+      "step": 472
+    },
+    {
+      "epoch": 0.18975180144115292,
+      "learning_rate": 8.001147455039737e-06,
+      "loss": 0.2124,
+      "step": 474
+    },
+    {
+      "epoch": 0.19055244195356286,
+      "learning_rate": 8.028541287070858e-06,
+      "loss": 0.0513,
+      "step": 476
+    },
+    {
+      "epoch": 0.19135308246597277,
+      "learning_rate": 8.055950520300756e-06,
+      "loss": 0.2265,
+      "step": 478
+    },
+    {
+      "epoch": 0.1921537229783827,
+      "learning_rate": 8.083374940606256e-06,
+      "loss": 0.0198,
+      "step": 480
+    },
+    {
+      "epoch": 0.19295436349079265,
+      "learning_rate": 8.110814333745503e-06,
+      "loss": 0.5947,
+      "step": 482
+    },
+    {
+      "epoch": 0.19375500400320256,
+      "learning_rate": 8.138268485359684e-06,
+      "loss": 0.1962,
+      "step": 484
+    },
+    {
+      "epoch": 0.1945556445156125,
+      "learning_rate": 8.165737180974676e-06,
+      "loss": 0.1806,
+      "step": 486
+    },
+    {
+      "epoch": 0.1953562850280224,
+      "learning_rate": 8.193220206002785e-06,
+      "loss": 0.4964,
+      "step": 488
+    },
+    {
+      "epoch": 0.19615692554043235,
+      "learning_rate": 8.220717345744326e-06,
+      "loss": 0.1587,
+      "step": 490
+    },
+    {
+      "epoch": 0.1969575660528423,
+      "learning_rate": 8.248228385389349e-06,
+      "loss": 0.1606,
+      "step": 492
+    },
+    {
+      "epoch": 0.1977582065652522,
+      "learning_rate": 8.275753110019367e-06,
+      "loss": 0.1103,
+      "step": 494
+    },
+    {
+      "epoch": 0.19855884707766214,
+      "learning_rate": 8.303291304608936e-06,
+      "loss": 0.021,
+      "step": 496
+    },
+    {
+      "epoch": 0.19935948759007205,
+      "learning_rate": 8.330842754027378e-06,
+      "loss": 0.0049,
+      "step": 498
+    },
+    {
+      "epoch": 0.200160128102482,
+      "learning_rate": 8.358407243040524e-06,
+      "loss": 0.2578,
+      "step": 500
+    },
+    {
+      "epoch": 0.20096076861489193,
+      "learning_rate": 8.385984556312285e-06,
+      "loss": 0.1362,
+      "step": 502
+    },
+    {
+      "epoch": 0.20176140912730184,
+      "learning_rate": 8.413574478406386e-06,
+      "loss": 0.4179,
+      "step": 504
+    },
+    {
+      "epoch": 0.20256204963971178,
+      "learning_rate": 8.441176793788106e-06,
+      "loss": 0.1503,
+      "step": 506
+    },
+    {
+      "epoch": 0.2033626901521217,
+      "learning_rate": 8.468791286825856e-06,
+      "loss": 0.1755,
+      "step": 508
+    },
+    {
+      "epoch": 0.20416333066453163,
+      "learning_rate": 8.496417741792922e-06,
+      "loss": 0.1431,
+      "step": 510
+    },
+    {
+      "epoch": 0.20496397117694154,
+      "learning_rate": 8.524055942869135e-06,
+      "loss": 0.011,
+      "step": 512
+    },
+    {
+      "epoch": 0.20576461168935148,
+      "learning_rate": 8.551705674142616e-06,
+      "loss": 0.3312,
+      "step": 514
+    },
+    {
+      "epoch": 0.20656525220176142,
+      "learning_rate": 8.579366719611353e-06,
+      "loss": 0.019,
+      "step": 516
+    },
+    {
+      "epoch": 0.20736589271417133,
+      "learning_rate": 8.607038863184952e-06,
+      "loss": 0.0045,
+      "step": 518
+    },
+    {
+      "epoch": 0.20816653322658127,
+      "learning_rate": 8.634721888686368e-06,
+      "loss": 0.1427,
+      "step": 520
+    },
+    {
+      "epoch": 0.20896717373899118,
+      "learning_rate": 8.662415579853495e-06,
+      "loss": 0.0707,
+      "step": 522
+    },
+    {
+      "epoch": 0.20976781425140112,
+      "learning_rate": 8.690119720340907e-06,
+      "loss": 0.1926,
+      "step": 524
+    },
+    {
+      "epoch": 0.21056845476381106,
+      "learning_rate": 8.717834093721598e-06,
+      "loss": 0.1663,
+      "step": 526
+    },
+    {
+      "epoch": 0.21136909527622097,
+      "learning_rate": 8.74555848348857e-06,
+      "loss": 0.1228,
+      "step": 528
+    },
+    {
+      "epoch": 0.2121697357886309,
+      "learning_rate": 8.773292673056572e-06,
+      "loss": 0.235,
+      "step": 530
+    },
+    {
+      "epoch": 0.21297037630104082,
+      "learning_rate": 8.801036445763858e-06,
+      "loss": 0.0872,
+      "step": 532
+    },
+    {
+      "epoch": 0.21377101681345076,
+      "learning_rate": 8.828789584873757e-06,
+      "loss": 0.2412,
+      "step": 534
+    },
+    {
+      "epoch": 0.2145716573258607,
+      "learning_rate": 8.856551873576448e-06,
+      "loss": 0.1103,
+      "step": 536
+    },
+    {
+      "epoch": 0.2153722978382706,
+      "learning_rate": 8.884323094990613e-06,
+      "loss": 0.0017,
+      "step": 538
+    },
+    {
+      "epoch": 0.21617293835068055,
+      "learning_rate": 8.912103032165206e-06,
+      "loss": 0.0287,
+      "step": 540
+    },
+    {
+      "epoch": 0.21697357886309046,
+      "learning_rate": 8.939891468081036e-06,
+      "loss": 0.1293,
+      "step": 542
+    },
+    {
+      "epoch": 0.2177742193755004,
+      "learning_rate": 8.967688185652527e-06,
+      "loss": 0.0015,
+      "step": 544
+    },
+    {
+      "epoch": 0.21857485988791034,
+      "learning_rate": 8.995492967729449e-06,
+      "loss": 0.2504,
+      "step": 546
+    },
+    {
+      "epoch": 0.21937550040032025,
+      "learning_rate": 9.023305597098526e-06,
+      "loss": 0.1533,
+      "step": 548
+    },
+    {
+      "epoch": 0.2201761409127302,
+      "learning_rate": 9.051125856485175e-06,
+      "loss": 0.0213,
+      "step": 550
+    },
+    {
+      "epoch": 0.2209767814251401,
+      "learning_rate": 9.078953528555258e-06,
+      "loss": 0.9907,
+      "step": 552
+    },
+    {
+      "epoch": 0.22177742193755004,
+      "learning_rate": 9.106788395916682e-06,
+      "loss": 1.0512,
+      "step": 554
+    },
+    {
+      "epoch": 0.22257806244995998,
+      "learning_rate": 9.134630241121135e-06,
+      "loss": 0.0298,
+      "step": 556
+    },
+    {
+      "epoch": 0.2233787029623699,
+      "learning_rate": 9.162478846665854e-06,
+      "loss": 0.4309,
+      "step": 558
+    },
+    {
+      "epoch": 0.22417934347477983,
+      "learning_rate": 9.190333994995208e-06,
+      "loss": 0.1522,
+      "step": 560
+    },
+    {
+      "epoch": 0.22497998398718974,
+      "learning_rate": 9.218195468502469e-06,
+      "loss": 0.0953,
+      "step": 562
+    },
+    {
+      "epoch": 0.22578062449959968,
+      "learning_rate": 9.24606304953148e-06,
+      "loss": 0.5407,
+      "step": 564
+    },
+    {
+      "epoch": 0.22658126501200962,
+      "learning_rate": 9.273936520378426e-06,
+      "loss": 0.0858,
+      "step": 566
+    },
+    {
+      "epoch": 0.22738190552441953,
+      "learning_rate": 9.301815663293426e-06,
+      "loss": 0.2854,
+      "step": 568
+    },
+    {
+      "epoch": 0.22818254603682947,
+      "learning_rate": 9.329700260482286e-06,
+      "loss": 1.9709,
+      "step": 570
+    },
+    {
+      "epoch": 0.22898318654923938,
+      "learning_rate": 9.35759009410826e-06,
+      "loss": 0.0383,
+      "step": 572
+    },
+    {
+      "epoch": 0.22978382706164932,
+      "learning_rate": 9.38548494629364e-06,
+      "loss": 0.0332,
+      "step": 574
+    },
+    {
+      "epoch": 0.23058446757405926,
+      "learning_rate": 9.41338459912151e-06,
+      "loss": 0.1591,
+      "step": 576
+    },
+    {
+      "epoch": 0.23138510808646917,
+      "learning_rate": 9.441288834637507e-06,
+      "loss": 0.4677,
+      "step": 578
+    },
+    {
+      "epoch": 0.2321857485988791,
+      "learning_rate": 9.469197434851414e-06,
+      "loss": 0.108,
+      "step": 580
+    },
+    {
+      "epoch": 0.23298638911128902,
+      "learning_rate": 9.497110181738935e-06,
+      "loss": 0.254,
+      "step": 582
+    },
+    {
+      "epoch": 0.23378702962369896,
+      "learning_rate": 9.52502685724336e-06,
+      "loss": 0.2554,
+      "step": 584
+    },
+    {
+      "epoch": 0.2345876701361089,
+      "learning_rate": 9.552947243277342e-06,
+      "loss": 0.1302,
+      "step": 586
+    },
+    {
+      "epoch": 0.2353883106485188,
+      "learning_rate": 9.580871121724498e-06,
+      "loss": 0.2466,
+      "step": 588
+    },
+    {
+      "epoch": 0.23618895116092875,
+      "learning_rate": 9.608798274441153e-06,
+      "loss": 0.0709,
+      "step": 590
+    },
+    {
+      "epoch": 0.23698959167333866,
+      "learning_rate": 9.636728483258116e-06,
+      "loss": 0.2866,
+      "step": 592
+    },
+    {
+      "epoch": 0.2377902321857486,
+      "learning_rate": 9.664661529982263e-06,
+      "loss": 0.1067,
+      "step": 594
+    },
+    {
+      "epoch": 0.23859087269815854,
+      "learning_rate": 9.692597196398302e-06,
+      "loss": 0.2506,
+      "step": 596
+    },
+    {
+      "epoch": 0.23939151321056845,
+      "learning_rate": 9.720535264270526e-06,
+      "loss": 0.1105,
+      "step": 598
+    },
+    {
+      "epoch": 0.2401921537229784,
+      "learning_rate": 9.748475515344416e-06,
+      "loss": 0.2434,
+      "step": 600
+    },
+    {
+      "epoch": 0.2409927942353883,
+      "learning_rate": 9.776417731348403e-06,
+      "loss": 0.1571,
+      "step": 602
+    },
+    {
+      "epoch": 0.24179343474779824,
+      "learning_rate": 9.80436169399561e-06,
+      "loss": 0.5832,
+      "step": 604
+    },
+    {
+      "epoch": 0.24259407526020815,
+      "learning_rate": 9.832307184985473e-06,
+      "loss": 0.0777,
+      "step": 606
+    },
+    {
+      "epoch": 0.2433947157726181,
+      "learning_rate": 9.8602539860055e-06,
+      "loss": 0.2253,
+      "step": 608
+    },
+    {
+      "epoch": 0.24419535628502803,
+      "learning_rate": 9.888201878732946e-06,
+      "loss": 0.3303,
+      "step": 610
+    },
+    {
+      "epoch": 0.24499599679743794,
+      "learning_rate": 9.916150644836596e-06,
+      "loss": 0.233,
+      "step": 612
+    },
+    {
+      "epoch": 0.24579663730984788,
+      "learning_rate": 9.944100065978354e-06,
+      "loss": 0.1816,
+      "step": 614
+    },
+    {
+      "epoch": 0.2465972778222578,
+      "learning_rate": 9.972049923815011e-06,
+      "loss": 0.0678,
+      "step": 616
+    },
+    {
+      "epoch": 0.24739791833466773,
+      "learning_rate": 9.999999999999996e-06,
+      "loss": 0.2618,
+      "step": 618
+    },
+    {
+      "epoch": 0.24819855884707767,
+      "learning_rate": 1.0027950076184982e-05,
+      "loss": 0.0673,
+      "step": 620
+    },
+    {
+      "epoch": 0.24899919935948758,
+      "learning_rate": 1.0055899934021637e-05,
+      "loss": 0.1078,
+      "step": 622
+    },
+    {
+      "epoch": 0.24979983987189752,
+      "learning_rate": 1.0083849355163397e-05,
+      "loss": 0.0413,
+      "step": 624
+    },
+    {
+      "epoch": 0.25060048038430743,
+      "learning_rate": 1.0111798121267047e-05,
+      "loss": 0.0515,
+      "step": 626
+    },
+    {
+      "epoch": 0.2514011208967174,
+      "learning_rate": 1.0139746013994493e-05,
+      "loss": 0.2504,
+      "step": 628
+    },
+    {
+      "epoch": 0.2522017614091273,
+      "learning_rate": 1.016769281501452e-05,
+      "loss": 0.1164,
+      "step": 630
+    },
+    {
+      "epoch": 0.2530024019215372,
+      "learning_rate": 1.0195638306004383e-05,
+      "loss": 0.2182,
+      "step": 632
+    },
+    {
+      "epoch": 0.25380304243394713,
+      "learning_rate": 1.022358226865159e-05,
+      "loss": 0.1589,
+      "step": 634
+    },
+    {
+      "epoch": 0.2546036829463571,
+      "learning_rate": 1.0251524484655577e-05,
+      "loss": 0.1725,
+      "step": 636
+    },
+    {
+      "epoch": 0.255404323458767,
+      "learning_rate": 1.0279464735729467e-05,
+      "loss": 0.0579,
+      "step": 638
+    },
+    {
+      "epoch": 0.2562049639711769,
+      "learning_rate": 1.0307402803601691e-05,
+      "loss": 0.0485,
+      "step": 640
+    },
+    {
+      "epoch": 0.2570056044835869,
+      "learning_rate": 1.033533847001773e-05,
+      "loss": 0.4735,
+      "step": 642
+    },
+    {
+      "epoch": 0.2578062449959968,
+      "learning_rate": 1.0363271516741877e-05,
+      "loss": 0.0575,
+      "step": 644
+    },
+    {
+      "epoch": 0.2586068855084067,
+      "learning_rate": 1.039120172555884e-05,
+      "loss": 0.0828,
+      "step": 646
+    },
+    {
+      "epoch": 0.2594075260208166,
+      "learning_rate": 1.0419128878275495e-05,
+      "loss": 0.0701,
+      "step": 648
+    },
+    {
+      "epoch": 0.2602081665332266,
+      "learning_rate": 1.0447052756722651e-05,
+      "loss": 0.2886,
+      "step": 650
+    },
+    {
+      "epoch": 0.2610088070456365,
+      "learning_rate": 1.0474973142756632e-05,
+      "loss": 0.1189,
+      "step": 652
+    },
+    {
+      "epoch": 0.2618094475580464,
+      "learning_rate": 1.0502889818261058e-05,
+      "loss": 0.3615,
+      "step": 654
+    },
+    {
+      "epoch": 0.2626100880704564,
+      "learning_rate": 1.053080256514858e-05,
+      "loss": 0.177,
+      "step": 656
+    },
+    {
+      "epoch": 0.2634107285828663,
+      "learning_rate": 1.0558711165362488e-05,
+      "loss": 0.0059,
+      "step": 658
+    },
+    {
+      "epoch": 0.2642113690952762,
+      "learning_rate": 1.0586615400878484e-05,
+      "loss": 0.394,
+      "step": 660
+    },
+    {
+      "epoch": 0.26501200960768617,
+      "learning_rate": 1.0614515053706354e-05,
+      "loss": 0.177,
+      "step": 662
+    },
+    {
+      "epoch": 0.2658126501200961,
+      "learning_rate": 1.0642409905891733e-05,
+      "loss": 0.6303,
+      "step": 664
+    },
+    {
+      "epoch": 0.266613290632506,
+      "learning_rate": 1.0670299739517706e-05,
+      "loss": 0.3272,
+      "step": 666
+    },
+    {
+      "epoch": 0.2674139311449159,
+      "learning_rate": 1.0698184336706567e-05,
+      "loss": 0.2295,
+      "step": 668
+    },
+    {
+      "epoch": 0.2682145716573259,
+      "learning_rate": 1.0726063479621567e-05,
+      "loss": 0.1106,
+      "step": 670
+    },
+    {
+      "epoch": 0.2690152121697358,
+      "learning_rate": 1.0753936950468513e-05,
+      "loss": 0.0988,
+      "step": 672
+    },
+    {
+      "epoch": 0.2698158526821457,
+      "learning_rate": 1.0781804531497525e-05,
+      "loss": 0.5348,
+      "step": 674
+    },
+    {
+      "epoch": 0.27061649319455566,
+      "learning_rate": 1.0809666005004787e-05,
+      "loss": 0.1596,
+      "step": 676
+    },
+    {
+      "epoch": 0.2714171337069656,
+      "learning_rate": 1.083752115333414e-05,
+      "loss": 0.0666,
+      "step": 678
+    },
+    {
+      "epoch": 0.2722177742193755,
+      "learning_rate": 1.0865369758878858e-05,
+      "loss": 0.3488,
+      "step": 680
+    },
+    {
+      "epoch": 0.27301841473178545,
+      "learning_rate": 1.0893211604083311e-05,
+      "loss": 0.2719,
+      "step": 682
+    },
+    {
+      "epoch": 0.27381905524419536,
+      "learning_rate": 1.0921046471444737e-05,
+      "loss": 0.1288,
+      "step": 684
+    },
+    {
+      "epoch": 0.2746196957566053,
+      "learning_rate": 1.0948874143514818e-05,
+      "loss": 0.427,
+      "step": 686
+    },
+    {
+      "epoch": 0.2754203362690152,
+      "learning_rate": 1.0976694402901467e-05,
+      "loss": 0.1688,
+      "step": 688
+    },
+    {
+      "epoch": 0.27622097678142515,
+      "learning_rate": 1.1004507032270544e-05,
+      "loss": 0.2892,
+      "step": 690
+    },
+    {
+      "epoch": 0.27702161729383507,
+      "learning_rate": 1.1032311814347467e-05,
+      "loss": 0.1668,
+      "step": 692
+    },
+    {
+      "epoch": 0.277822257806245,
+      "learning_rate": 1.1060108531918955e-05,
+      "loss": 0.0432,
+      "step": 694
+    },
+    {
+      "epoch": 0.27862289831865494,
+      "learning_rate": 1.1087896967834787e-05,
+      "loss": 0.5289,
+      "step": 696
+    },
+    {
+      "epoch": 0.27942353883106485,
+      "learning_rate": 1.111567690500938e-05,
+      "loss": 0.1471,
+      "step": 698
+    },
+    {
+      "epoch": 0.28022417934347477,
+      "learning_rate": 1.1143448126423545e-05,
+      "loss": 0.0306,
+      "step": 700
+    },
+    {
+      "epoch": 0.28102481985588473,
+      "learning_rate": 1.1171210415126238e-05,
+      "loss": 0.2731,
+      "step": 702
+    },
+    {
+      "epoch": 0.28182546036829464,
+      "learning_rate": 1.1198963554236135e-05,
+      "loss": 0.5657,
+      "step": 704
+    },
+    {
+      "epoch": 0.28262610088070456,
+      "learning_rate": 1.122670732694342e-05,
+      "loss": 0.0508,
+      "step": 706
+    },
+    {
+      "epoch": 0.28342674139311447,
+      "learning_rate": 1.1254441516511425e-05,
+      "loss": 0.0586,
+      "step": 708
+    },
+    {
+      "epoch": 0.28422738190552443,
+      "learning_rate": 1.1282165906278395e-05,
+      "loss": 0.0144,
+      "step": 710
+    },
+    {
+      "epoch": 0.28502802241793435,
+      "learning_rate": 1.1309880279659087e-05,
+      "loss": 0.3224,
+      "step": 712
+    },
+    {
+      "epoch": 0.28582866293034426,
+      "learning_rate": 1.1337584420146496e-05,
+      "loss": 0.278,
+      "step": 714
+    },
+    {
+      "epoch": 0.2866293034427542,
+      "learning_rate": 1.1365278111313625e-05,
+      "loss": 0.5086,
+      "step": 716
+    },
+    {
+      "epoch": 0.28742994395516414,
+      "learning_rate": 1.1392961136815041e-05,
+      "loss": 0.2386,
+      "step": 718
+    },
+    {
+      "epoch": 0.28823058446757405,
+      "learning_rate": 1.142063328038864e-05,
+      "loss": 0.5037,
+      "step": 720
+    },
+    {
+      "epoch": 0.289031224979984,
+      "learning_rate": 1.1448294325857377e-05,
+      "loss": 0.1244,
+      "step": 722
+    },
+    {
+      "epoch": 0.2898318654923939,
+      "learning_rate": 1.1475944057130856e-05,
+      "loss": 0.1801,
+      "step": 724
+    },
+    {
+      "epoch": 0.29063250600480384,
+      "learning_rate": 1.150358225820707e-05,
+      "loss": 0.1845,
+      "step": 726
+    },
+    {
+      "epoch": 0.29143314651721375,
+      "learning_rate": 1.1531208713174138e-05,
+      "loss": 0.3542,
+      "step": 728
+    },
+    {
+      "epoch": 0.2922337870296237,
+      "learning_rate": 1.1558823206211887e-05,
+      "loss": 0.0561,
+      "step": 730
+    },
+    {
+      "epoch": 0.2930344275420336,
+      "learning_rate": 1.1586425521593607e-05,
+      "loss": 0.3418,
+      "step": 732
+    },
+    {
+      "epoch": 0.29383506805444354,
+      "learning_rate": 1.1614015443687708e-05,
+      "loss": 0.2323,
+      "step": 734
+    },
+    {
+      "epoch": 0.2946357085668535,
+      "learning_rate": 1.1641592756959467e-05,
+      "loss": 0.152,
+      "step": 736
+    },
+    {
+      "epoch": 0.2954363490792634,
+      "learning_rate": 1.1669157245972616e-05,
+      "loss": 0.1304,
+      "step": 738
+    },
+    {
+      "epoch": 0.2962369895916733,
+      "learning_rate": 1.1696708695391057e-05,
+      "loss": 0.4447,
+      "step": 740
+    },
+    {
+      "epoch": 0.29703763010408324,
+      "learning_rate": 1.1724246889980626e-05,
+      "loss": 0.0264,
+      "step": 742
+    },
+    {
+      "epoch": 0.2978382706164932,
+      "learning_rate": 1.1751771614610643e-05,
+      "loss": 0.0659,
+      "step": 744
+    },
+    {
+      "epoch": 0.2986389111289031,
+      "learning_rate": 1.1779282654255668e-05,
+      "loss": 0.1796,
+      "step": 746
+    },
+    {
+      "epoch": 0.29943955164131303,
+      "learning_rate": 1.180677979399721e-05,
+      "loss": 0.1283,
+      "step": 748
+    },
+    {
+      "epoch": 0.300240192153723,
+      "learning_rate": 1.1834262819025317e-05,
+      "loss": 0.0365,
+      "step": 750
+    },
+    {
+      "epoch": 0.3010408326661329,
+      "learning_rate": 1.1861731514640309e-05,
+      "loss": 0.2157,
+      "step": 752
+    },
+    {
+      "epoch": 0.3018414731785428,
+      "learning_rate": 1.188918566625449e-05,
+      "loss": 0.5453,
+      "step": 754
+    },
+    {
+      "epoch": 0.3026421136909528,
+      "learning_rate": 1.1916625059393739e-05,
+      "loss": 0.3992,
+      "step": 756
+    },
+    {
+      "epoch": 0.3034427542033627,
+      "learning_rate": 1.1944049479699241e-05,
+      "loss": 0.0596,
+      "step": 758
+    },
+    {
+      "epoch": 0.3042433947157726,
+      "learning_rate": 1.1971458712929133e-05,
+      "loss": 0.0521,
+      "step": 760
+    },
+    {
+      "epoch": 0.3050440352281825,
+      "learning_rate": 1.1998852544960256e-05,
+      "loss": 0.2155,
+      "step": 762
+    },
+    {
+      "epoch": 0.3058446757405925,
+      "learning_rate": 1.2026230761789702e-05,
+      "loss": 0.1666,
+      "step": 764
+    },
+    {
+      "epoch": 0.3066453162530024,
+      "learning_rate": 1.2053593149536557e-05,
+      "loss": 0.0528,
+      "step": 766
+    },
+    {
+      "epoch": 0.3074459567654123,
+      "learning_rate": 1.2080939494443618e-05,
+      "loss": 0.2306,
+      "step": 768
+    },
+    {
+      "epoch": 0.3082465972778223,
+      "learning_rate": 1.210826958287895e-05,
+      "loss": 0.0265,
+      "step": 770
+    },
+    {
+      "epoch": 0.3090472377902322,
+      "learning_rate": 1.2135583201337646e-05,
+      "loss": 0.04,
+      "step": 772
+    },
+    {
+      "epoch": 0.3098478783026421,
+      "learning_rate": 1.2162880136443434e-05,
+      "loss": 0.2041,
+      "step": 774
+    },
+    {
+      "epoch": 0.31064851881505207,
+      "learning_rate": 1.2190160174950428e-05,
+      "loss": 0.3985,
+      "step": 776
+    },
+    {
+      "epoch": 0.311449159327462,
+      "learning_rate": 1.2217423103744692e-05,
+      "loss": 0.4338,
+      "step": 778
+    },
+    {
+      "epoch": 0.3122497998398719,
+      "learning_rate": 1.2244668709845952e-05,
+      "loss": 0.0033,
+      "step": 780
+    },
+    {
+      "epoch": 0.3130504403522818,
+      "learning_rate": 1.2271896780409309e-05,
+      "loss": 0.1383,
+      "step": 782
+    },
+    {
+      "epoch": 0.31385108086469177,
+      "learning_rate": 1.2299107102726804e-05,
+      "loss": 0.1044,
+      "step": 784
+    },
+    {
+      "epoch": 0.3146517213771017,
+      "learning_rate": 1.2326299464229143e-05,
+      "loss": 0.6256,
+      "step": 786
+    },
+    {
+      "epoch": 0.3154523618895116,
+      "learning_rate": 1.2353473652487329e-05,
+      "loss": 0.1302,
+      "step": 788
+    },
+    {
+      "epoch": 0.31625300240192156,
+      "learning_rate": 1.2380629455214385e-05,
+      "loss": 0.1811,
+      "step": 790
+    },
+    {
+      "epoch": 0.31705364291433147,
+      "learning_rate": 1.2407766660266916e-05,
+      "loss": 0.7897,
+      "step": 792
+    },
+    {
+      "epoch": 0.3178542834267414,
+      "learning_rate": 1.2434885055646808e-05,
+      "loss": 0.0882,
+      "step": 794
+    },
+    {
+      "epoch": 0.31865492393915135,
+      "learning_rate": 1.2461984429502947e-05,
+      "loss": 0.2129,
+      "step": 796
+    },
+    {
+      "epoch": 0.31945556445156126,
+      "learning_rate": 1.2489064570132761e-05,
+      "loss": 0.2419,
+      "step": 798
+    },
+    {
+      "epoch": 0.32025620496397117,
+      "learning_rate": 1.2516125265983945e-05,
+      "loss": 0.0982,
+      "step": 800
+    },
+    {
+      "epoch": 0.3210568454763811,
+      "learning_rate": 1.2543166305656089e-05,
+      "loss": 0.0043,
+      "step": 802
+    },
+    {
+      "epoch": 0.32185748598879105,
+      "learning_rate": 1.257018747790238e-05,
+      "loss": 0.0432,
+      "step": 804
+    },
+    {
+      "epoch": 0.32265812650120096,
+      "learning_rate": 1.259718857163117e-05,
+      "loss": 0.3037,
+      "step": 806
+    },
+    {
+      "epoch": 0.32345876701361087,
+      "learning_rate": 1.2624169375907657e-05,
+      "loss": 0.1702,
+      "step": 808
+    },
+    {
+      "epoch": 0.32425940752602084,
+      "learning_rate": 1.2651129679955598e-05,
+      "loss": 0.3856,
+      "step": 810
+    },
+    {
+      "epoch": 0.32506004803843075,
+      "learning_rate": 1.2678069273158849e-05,
+      "loss": 0.133,
+      "step": 812
+    },
+    {
+      "epoch": 0.32586068855084066,
+      "learning_rate": 1.2704987945063073e-05,
+      "loss": 0.2291,
+      "step": 814
+    },
+    {
+      "epoch": 0.3266613290632506,
+      "learning_rate": 1.273188548537736e-05,
+      "loss": 0.6004,
+      "step": 816
+    },
+    {
+      "epoch": 0.32746196957566054,
+      "learning_rate": 1.2758761683975929e-05,
+      "loss": 0.2873,
+      "step": 818
+    },
+    {
+      "epoch": 0.32826261008807045,
+      "learning_rate": 1.2785616330899676e-05,
+      "loss": 0.0601,
+      "step": 820
+    },
+    {
+      "epoch": 0.32906325060048036,
+      "learning_rate": 1.2812449216357855e-05,
+      "loss": 0.023,
+      "step": 822
+    },
+    {
+      "epoch": 0.32986389111289033,
+      "learning_rate": 1.2839260130729776e-05,
+      "loss": 0.3317,
+      "step": 824
+    },
+    {
+      "epoch": 0.33066453162530024,
+      "learning_rate": 1.2866048864566336e-05,
+      "loss": 0.0976,
+      "step": 826
+    },
+    {
+      "epoch": 0.33146517213771015,
+      "learning_rate": 1.2892815208591734e-05,
+      "loss": 0.2702,
+      "step": 828
+    },
+    {
+      "epoch": 0.3322658126501201,
+      "learning_rate": 1.2919558953705047e-05,
+      "loss": 0.1872,
+      "step": 830
+    },
+    {
+      "epoch": 0.33306645316253003,
+      "learning_rate": 1.2946279890981966e-05,
+      "loss": 0.0484,
+      "step": 832
+    },
+    {
+      "epoch": 0.33386709367493994,
+      "learning_rate": 1.2972977811676289e-05,
+      "loss": 0.2853,
+      "step": 834
+    },
+    {
+      "epoch": 0.33466773418734985,
+      "learning_rate": 1.2999652507221652e-05,
+      "loss": 0.0551,
+      "step": 836
+    },
+    {
+      "epoch": 0.3354683746997598,
+      "learning_rate": 1.3026303769233109e-05,
+      "loss": 0.3515,
+      "step": 838
+    },
+    {
+      "epoch": 0.33626901521216973,
+      "learning_rate": 1.3052931389508822e-05,
+      "loss": 0.3938,
+      "step": 840
+    },
+    {
+      "epoch": 0.33706965572457964,
+      "learning_rate": 1.3079535160031601e-05,
+      "loss": 0.0773,
+      "step": 842
+    },
+    {
+      "epoch": 0.3378702962369896,
+      "learning_rate": 1.3106114872970575e-05,
+      "loss": 0.25,
+      "step": 844
+    },
+    {
+      "epoch": 0.3386709367493995,
+      "learning_rate": 1.313267032068285e-05,
+      "loss": 0.2299,
+      "step": 846
+    },
+    {
+      "epoch": 0.33947157726180943,
+      "learning_rate": 1.3159201295715054e-05,
+      "loss": 0.8993,
+      "step": 848
+    },
+    {
+      "epoch": 0.3402722177742194,
+      "learning_rate": 1.3185707590804997e-05,
+      "loss": 0.0636,
+      "step": 850
+    },
+    {
+      "epoch": 0.3410728582866293,
+      "learning_rate": 1.321218899888334e-05,
+      "loss": 0.2269,
+      "step": 852
+    },
+    {
+      "epoch": 0.3418734987990392,
+      "learning_rate": 1.3238645313075109e-05,
+      "loss": 0.3642,
+      "step": 854
+    },
+    {
+      "epoch": 0.34267413931144913,
+      "learning_rate": 1.326507632670139e-05,
+      "loss": 0.5907,
+      "step": 856
+    },
+    {
+      "epoch": 0.3434747798238591,
+      "learning_rate": 1.3291481833280894e-05,
+      "loss": 0.6678,
+      "step": 858
+    },
+    {
+      "epoch": 0.344275420336269,
+      "learning_rate": 1.3317861626531652e-05,
+      "loss": 0.0843,
+      "step": 860
+    },
+    {
+      "epoch": 0.3450760608486789,
+      "learning_rate": 1.3344215500372517e-05,
+      "loss": 0.1857,
+      "step": 862
+    },
+    {
+      "epoch": 0.3458767013610889,
+      "learning_rate": 1.3370543248924826e-05,
+      "loss": 0.2602,
+      "step": 864
+    },
+    {
+      "epoch": 0.3466773418734988,
+      "learning_rate": 1.3396844666514062e-05,
+      "loss": 0.2764,
+      "step": 866
+    },
+    {
+      "epoch": 0.3474779823859087,
+      "learning_rate": 1.3423119547671348e-05,
+      "loss": 0.2288,
+      "step": 868
+    },
+    {
+      "epoch": 0.3482786228983187,
+      "learning_rate": 1.344936768713513e-05,
+      "loss": 0.143,
+      "step": 870
+    },
+    {
+      "epoch": 0.3490792634107286,
+      "learning_rate": 1.347558887985279e-05,
+      "loss": 0.2076,
+      "step": 872
+    },
+    {
+      "epoch": 0.3498799039231385,
+      "learning_rate": 1.3501782920982189e-05,
+      "loss": 0.0369,
+      "step": 874
+    },
+    {
+      "epoch": 0.3506805444355484,
+      "learning_rate": 1.3527949605893305e-05,
+      "loss": 0.0689,
+      "step": 876
+    },
+    {
+      "epoch": 0.3514811849479584,
+      "learning_rate": 1.3554088730169812e-05,
+      "loss": 0.0862,
+      "step": 878
+    },
+    {
+      "epoch": 0.3522818254603683,
+      "learning_rate": 1.3580200089610739e-05,
+      "loss": 0.1249,
+      "step": 880
+    },
+    {
+      "epoch": 0.3530824659727782,
+      "learning_rate": 1.3606283480231962e-05,
+      "loss": 0.0659,
+      "step": 882
+    },
+    {
+      "epoch": 0.35388310648518817,
+      "learning_rate": 1.3632338698267863e-05,
+      "loss": 0.0604,
+      "step": 884
+    },
+    {
+      "epoch": 0.3546837469975981,
+      "learning_rate": 1.3658365540172948e-05,
+      "loss": 0.6569,
+      "step": 886
+    },
+    {
+      "epoch": 0.355484387510008,
+      "learning_rate": 1.368436380262336e-05,
+      "loss": 0.1914,
+      "step": 888
+    },
+    {
+      "epoch": 0.35628502802241796,
+      "learning_rate": 1.3710333282518497e-05,
+      "loss": 0.1759,
+      "step": 890
+    },
+    {
+      "epoch": 0.35708566853482787,
+      "learning_rate": 1.3736273776982667e-05,
+      "loss": 0.0608,
+      "step": 892
+    },
+    {
+      "epoch": 0.3578863090472378,
+      "learning_rate": 1.3762185083366562e-05,
+      "loss": 0.4837,
+      "step": 894
+    },
+    {
+      "epoch": 0.3586869495596477,
+      "learning_rate": 1.3788066999248893e-05,
+      "loss": 0.0775,
+      "step": 896
+    },
+    {
+      "epoch": 0.35948759007205766,
+      "learning_rate": 1.3813919322438018e-05,
+      "loss": 0.2043,
+      "step": 898
+    },
+    {
+      "epoch": 0.3602882305844676,
+      "learning_rate": 1.3839741850973435e-05,
+      "loss": 0.3837,
+      "step": 900
+    },
+    {
+      "epoch": 0.3610888710968775,
+      "learning_rate": 1.3865534383127413e-05,
+      "loss": 0.2239,
+      "step": 902
+    },
+    {
+      "epoch": 0.36188951160928745,
+      "learning_rate": 1.3891296717406533e-05,
+      "loss": 0.083,
+      "step": 904
+    },
+    {
+      "epoch": 0.36269015212169736,
+      "learning_rate": 1.391702865255334e-05,
+      "loss": 0.3167,
+      "step": 906
+    },
+    {
+      "epoch": 0.3634907926341073,
+      "learning_rate": 1.3942729987547808e-05,
+      "loss": 0.0993,
+      "step": 908
+    },
+    {
+      "epoch": 0.36429143314651724,
+      "learning_rate": 1.3968400521608962e-05,
+      "loss": 0.2865,
+      "step": 910
+    },
+    {
+      "epoch": 0.36509207365892715,
+      "learning_rate": 1.3994040054196498e-05,
+      "loss": 0.2562,
+      "step": 912
+    },
+    {
+      "epoch": 0.36589271417133706,
+      "learning_rate": 1.4019648385012245e-05,
+      "loss": 0.1762,
+      "step": 914
+    },
+    {
+      "epoch": 0.366693354683747,
+      "learning_rate": 1.4045225314001789e-05,
+      "loss": 0.1536,
+      "step": 916
+    },
+    {
+      "epoch": 0.36749399519615694,
+      "learning_rate": 1.4070770641356069e-05,
+      "loss": 0.4599,
+      "step": 918
+    },
+    {
+      "epoch": 0.36829463570856685,
+      "learning_rate": 1.4096284167512856e-05,
+      "loss": 0.2153,
+      "step": 920
+    },
+    {
+      "epoch": 0.36909527622097676,
+      "learning_rate": 1.4121765693158355e-05,
+      "loss": 0.0877,
+      "step": 922
+    },
+    {
+      "epoch": 0.36989591673338673,
+      "learning_rate": 1.4147215019228813e-05,
+      "loss": 0.5322,
+      "step": 924
+    },
+    {
+      "epoch": 0.37069655724579664,
+      "learning_rate": 1.4172631946911964e-05,
+      "loss": 0.3526,
+      "step": 926
+    },
+    {
+      "epoch": 0.37149719775820655,
+      "learning_rate": 1.4198016277648665e-05,
+      "loss": 0.323,
+      "step": 928
+    },
+    {
+      "epoch": 0.37229783827061647,
+      "learning_rate": 1.4223367813134406e-05,
+      "loss": 0.0472,
+      "step": 930
+    },
+    {
+      "epoch": 0.37309847878302643,
+      "learning_rate": 1.4248686355320922e-05,
+      "loss": 0.1587,
+      "step": 932
+    },
+    {
+      "epoch": 0.37389911929543634,
+      "learning_rate": 1.4273971706417653e-05,
+      "loss": 0.4038,
+      "step": 934
+    },
+    {
+      "epoch": 0.37469975980784626,
+      "learning_rate": 1.429922366889332e-05,
+      "loss": 0.2765,
+      "step": 936
+    },
+    {
+      "epoch": 0.3755004003202562,
+      "learning_rate": 1.4324442045477534e-05,
+      "loss": 0.0974,
+      "step": 938
+    },
+    {
+      "epoch": 0.37630104083266613,
+      "learning_rate": 1.4349626639162231e-05,
+      "loss": 0.2454,
+      "step": 940
+    },
+    {
+      "epoch": 0.37710168134507605,
+      "learning_rate": 1.4374777253203265e-05,
+      "loss": 0.1364,
+      "step": 942
+    },
+    {
+      "epoch": 0.377902321857486,
+      "learning_rate": 1.4399893691121985e-05,
+      "loss": 0.2448,
+      "step": 944
+    },
+    {
+      "epoch": 0.3787029623698959,
+      "learning_rate": 1.4424975756706684e-05,
+      "loss": 0.2071,
+      "step": 946
+    },
+    {
+      "epoch": 0.37950360288230583,
+      "learning_rate": 1.4450023254014185e-05,
+      "loss": 0.1957,
+      "step": 948
+    },
+    {
+      "epoch": 0.38030424339471575,
+      "learning_rate": 1.4475035987371348e-05,
+      "loss": 0.2507,
+      "step": 950
+    },
+    {
+      "epoch": 0.3811048839071257,
+      "learning_rate": 1.4500013761376663e-05,
+      "loss": 0.8129,
+      "step": 952
+    },
+    {
+      "epoch": 0.3819055244195356,
+      "learning_rate": 1.4524956380901674e-05,
+      "loss": 0.0641,
+      "step": 954
+    },
+    {
+      "epoch": 0.38270616493194554,
+      "learning_rate": 1.454986365109255e-05,
+      "loss": 0.1356,
+      "step": 956
+    },
+    {
+      "epoch": 0.3835068054443555,
+      "learning_rate": 1.4574735377371669e-05,
+      "loss": 0.3025,
+      "step": 958
+    },
+    {
+      "epoch": 0.3843074459567654,
+      "learning_rate": 1.4599571365439027e-05,
+      "loss": 0.2261,
+      "step": 960
+    },
+    {
+      "epoch": 0.3851080864691753,
+      "learning_rate": 1.4624371421273812e-05,
+      "loss": 0.0825,
+      "step": 962
+    },
+    {
+      "epoch": 0.3859087269815853,
+      "learning_rate": 1.4649135351135968e-05,
+      "loss": 0.1754,
+      "step": 964
+    },
+    {
+      "epoch": 0.3867093674939952,
+      "learning_rate": 1.4673862961567604e-05,
+      "loss": 0.0762,
+      "step": 966
+    },
+    {
+      "epoch": 0.3875100080064051,
+      "learning_rate": 1.4698554059394563e-05,
+      "loss": 0.2071,
+      "step": 968
+    },
+    {
+      "epoch": 0.388310648518815,
+      "learning_rate": 1.4723208451727977e-05,
+      "loss": 0.0323,
+      "step": 970
+    },
+    {
+      "epoch": 0.389111289031225,
+      "learning_rate": 1.4747825945965675e-05,
+      "loss": 0.0896,
+      "step": 972
+    },
+    {
+      "epoch": 0.3899119295436349,
+      "learning_rate": 1.4772406349793749e-05,
+      "loss": 0.3451,
+      "step": 974
+    },
+    {
+      "epoch": 0.3907125700560448,
+      "learning_rate": 1.4796949471188033e-05,
+      "loss": 0.0705,
+      "step": 976
+    },
+    {
+      "epoch": 0.3915132105684548,
+      "learning_rate": 1.4821455118415666e-05,
+      "loss": 0.2977,
+      "step": 978
+    },
+    {
+      "epoch": 0.3923138510808647,
+      "learning_rate": 1.4845923100036479e-05,
+      "loss": 0.0907,
+      "step": 980
+    },
+    {
+      "epoch": 0.3931144915932746,
+      "learning_rate": 1.4870353224904563e-05,
+      "loss": 0.0208,
+      "step": 982
+    },
+    {
+      "epoch": 0.3939151321056846,
+      "learning_rate": 1.4894745302169786e-05,
+      "loss": 0.0376,
+      "step": 984
+    },
+    {
+      "epoch": 0.3947157726180945,
+      "learning_rate": 1.4919099141279205e-05,
+      "loss": 0.8712,
+      "step": 986
+    },
+    {
+      "epoch": 0.3955164131305044,
+      "learning_rate": 1.4943414551978597e-05,
+      "loss": 0.5081,
+      "step": 988
+    },
+    {
+      "epoch": 0.3963170536429143,
+      "learning_rate": 1.4967691344313988e-05,
+      "loss": 0.2639,
+      "step": 990
+    },
+    {
+      "epoch": 0.3971176941553243,
+      "learning_rate": 1.499192932863305e-05,
+      "loss": 0.4575,
+      "step": 992
+    },
+    {
+      "epoch": 0.3979183346677342,
+      "learning_rate": 1.5016128315586626e-05,
+      "loss": 0.0815,
+      "step": 994
+    },
+    {
+      "epoch": 0.3987189751801441,
+      "learning_rate": 1.5040288116130261e-05,
+      "loss": 0.015,
+      "step": 996
+    },
+    {
+      "epoch": 0.39951961569255406,
+      "learning_rate": 1.5064408541525568e-05,
+      "loss": 0.3493,
+      "step": 998
+    },
+    {
+      "epoch": 0.400320256204964,
+      "learning_rate": 1.5088489403341793e-05,
+      "loss": 0.5574,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4011208967173739,
+      "learning_rate": 1.5112530513457229e-05,
+      "loss": 0.0433,
+      "step": 1002
+    },
+    {
+      "epoch": 0.40192153722978385,
+      "learning_rate": 1.513653168406076e-05,
+      "loss": 0.4451,
+      "step": 1004
+    },
+    {
+      "epoch": 0.40272217774219377,
+      "learning_rate": 1.5160492727653245e-05,
+      "loss": 0.1424,
+      "step": 1006
+    },
+    {
+      "epoch": 0.4035228182546037,
+      "learning_rate": 1.5184413457049006e-05,
+      "loss": 0.0179,
+      "step": 1008
+    },
+    {
+      "epoch": 0.4043234587670136,
+      "learning_rate": 1.5208293685377354e-05,
+      "loss": 0.1899,
+      "step": 1010
+    },
+    {
+      "epoch": 0.40512409927942356,
+      "learning_rate": 1.5232133226083954e-05,
+      "loss": 0.1627,
+      "step": 1012
+    },
+    {
+      "epoch": 0.40592473979183347,
+      "learning_rate": 1.5255931892932322e-05,
+      "loss": 0.128,
+      "step": 1014
+    },
+    {
+      "epoch": 0.4067253803042434,
+      "learning_rate": 1.527968950000533e-05,
+      "loss": 0.0551,
+      "step": 1016
+    },
+    {
+      "epoch": 0.40752602081665334,
+      "learning_rate": 1.5303405861706574e-05,
+      "loss": 0.3644,
+      "step": 1018
+    },
+    {
+      "epoch": 0.40832666132906326,
+      "learning_rate": 1.532708079276185e-05,
+      "loss": 0.0188,
+      "step": 1020
+    },
+    {
+      "epoch": 0.40912730184147317,
+      "learning_rate": 1.5350714108220667e-05,
+      "loss": 0.0302,
+      "step": 1022
+    },
+    {
+      "epoch": 0.4099279423538831,
+      "learning_rate": 1.5374305623457594e-05,
+      "loss": 0.1886,
+      "step": 1024
+    },
+    {
+      "epoch": 0.41072858286629305,
+      "learning_rate": 1.539785515417376e-05,
+      "loss": 0.0798,
+      "step": 1026
+    },
+    {
+      "epoch": 0.41152922337870296,
+      "learning_rate": 1.542136251639826e-05,
+      "loss": 0.8809,
+      "step": 1028
+    },
+    {
+      "epoch": 0.41232986389111287,
+      "learning_rate": 1.5444827526489668e-05,
+      "loss": 0.1639,
+      "step": 1030
+    },
+    {
+      "epoch": 0.41313050440352284,
+      "learning_rate": 1.5468250001137368e-05,
+      "loss": 0.0516,
+      "step": 1032
+    },
+    {
+      "epoch": 0.41393114491593275,
+      "learning_rate": 1.5491629757363026e-05,
+      "loss": 0.2273,
+      "step": 1034
+    },
+    {
+      "epoch": 0.41473178542834266,
+      "learning_rate": 1.551496661252208e-05,
+      "loss": 0.3641,
+      "step": 1036
+    },
+    {
+      "epoch": 0.4155324259407526,
+      "learning_rate": 1.5538260384305073e-05,
+      "loss": 0.0152,
+      "step": 1038
+    },
+    {
+      "epoch": 0.41633306645316254,
+      "learning_rate": 1.5561510890739113e-05,
+      "loss": 0.1172,
+      "step": 1040
+    },
+    {
+      "epoch": 0.41713370696557245,
+      "learning_rate": 1.5584717950189353e-05,
+      "loss": 0.0946,
+      "step": 1042
+    },
+    {
+      "epoch": 0.41793434747798236,
+      "learning_rate": 1.5607881381360296e-05,
+      "loss": 0.0921,
+      "step": 1044
+    },
+    {
+      "epoch": 0.4187349879903923,
+      "learning_rate": 1.563100100329731e-05,
+      "loss": 0.1605,
+      "step": 1046
+    },
+    {
+      "epoch": 0.41953562850280224,
+      "learning_rate": 1.565407663538797e-05,
+      "loss": 0.0204,
+      "step": 1048
+    },
+    {
+      "epoch": 0.42033626901521215,
+      "learning_rate": 1.567710809736356e-05,
+      "loss": 0.0842,
+      "step": 1050
+    },
+    {
+      "epoch": 0.4211369095276221,
+      "learning_rate": 1.5700095209300376e-05,
+      "loss": 0.0796,
+      "step": 1052
+    },
+    {
+      "epoch": 0.42193755004003203,
+      "learning_rate": 1.572303779162118e-05,
+      "loss": 0.0992,
+      "step": 1054
+    },
+    {
+      "epoch": 0.42273819055244194,
+      "learning_rate": 1.5745935665096647e-05,
+      "loss": 0.0317,
+      "step": 1056
+    },
+    {
+      "epoch": 0.4235388310648519,
+      "learning_rate": 1.5768788650846677e-05,
+      "loss": 0.0394,
+      "step": 1058
+    },
+    {
+      "epoch": 0.4243394715772618,
+      "learning_rate": 1.5791596570341844e-05,
+      "loss": 0.1129,
+      "step": 1060
+    },
+    {
+      "epoch": 0.42514011208967173,
+      "learning_rate": 1.581435924540481e-05,
+      "loss": 0.0158,
+      "step": 1062
+    },
+    {
+      "epoch": 0.42594075260208164,
+      "learning_rate": 1.5837076498211666e-05,
+      "loss": 0.0748,
+      "step": 1064
+    },
+    {
+      "epoch": 0.4267413931144916,
+      "learning_rate": 1.5859748151293333e-05,
+      "loss": 0.4398,
+      "step": 1066
+    },
+    {
+      "epoch": 0.4275420336269015,
+      "learning_rate": 1.5882374027537005e-05,
+      "loss": 0.1629,
+      "step": 1068
+    },
+    {
+      "epoch": 0.42834267413931143,
+      "learning_rate": 1.5904953950187455e-05,
+      "loss": 0.4817,
+      "step": 1070
+    },
+    {
+      "epoch": 0.4291433146517214,
+      "learning_rate": 1.5927487742848448e-05,
+      "loss": 0.8606,
+      "step": 1072
+    },
+    {
+      "epoch": 0.4299439551641313,
+      "learning_rate": 1.594997522948412e-05,
+      "loss": 0.2745,
+      "step": 1074
+    },
+    {
+      "epoch": 0.4307445956765412,
+      "learning_rate": 1.5972416234420393e-05,
+      "loss": 0.1034,
+      "step": 1076
+    },
+    {
+      "epoch": 0.4315452361889512,
+      "learning_rate": 1.599481058234626e-05,
+      "loss": 0.6513,
+      "step": 1078
+    },
+    {
+      "epoch": 0.4323458767013611,
+      "learning_rate": 1.60171580983152e-05,
+      "loss": 0.0517,
+      "step": 1080
+    },
+    {
+      "epoch": 0.433146517213771,
+      "learning_rate": 1.6039458607746607e-05,
+      "loss": 0.011,
+      "step": 1082
+    },
+    {
+      "epoch": 0.4339471577261809,
+      "learning_rate": 1.606171193642703e-05,
+      "loss": 0.0504,
+      "step": 1084
+    },
+    {
+      "epoch": 0.4347477982385909,
+      "learning_rate": 1.6083917910511616e-05,
+      "loss": 0.1601,
+      "step": 1086
+    },
+    {
+      "epoch": 0.4355484387510008,
+      "learning_rate": 1.6106076356525474e-05,
+      "loss": 0.0823,
+      "step": 1088
+    },
+    {
+      "epoch": 0.4363490792634107,
+      "learning_rate": 1.6128187101364982e-05,
+      "loss": 0.1996,
+      "step": 1090
+    },
+    {
+      "epoch": 0.4371497197758207,
+      "learning_rate": 1.6150249972299153e-05,
+      "loss": 0.0065,
+      "step": 1092
+    },
+    {
+      "epoch": 0.4379503602882306,
+      "learning_rate": 1.617226479697104e-05,
+      "loss": 0.0838,
+      "step": 1094
+    },
+    {
+      "epoch": 0.4387510008006405,
+      "learning_rate": 1.6194231403398994e-05,
+      "loss": 0.0665,
+      "step": 1096
+    },
+    {
+      "epoch": 0.43955164131305047,
+      "learning_rate": 1.621614961997806e-05,
+      "loss": 0.2692,
+      "step": 1098
+    },
+    {
+      "epoch": 0.4403522818254604,
+      "learning_rate": 1.6238019275481313e-05,
+      "loss": 0.1905,
+      "step": 1100
+    },
+    {
+      "epoch": 0.4411529223378703,
+      "learning_rate": 1.6259840199061212e-05,
+      "loss": 0.0288,
+      "step": 1102
+    },
+    {
+      "epoch": 0.4419535628502802,
+      "learning_rate": 1.6281612220250883e-05,
+      "loss": 0.0064,
+      "step": 1104
+    },
+    {
+      "epoch": 0.44275420336269017,
+      "learning_rate": 1.6303335168965474e-05,
+      "loss": 0.2056,
+      "step": 1106
+    },
+    {
+      "epoch": 0.4435548438751001,
+      "learning_rate": 1.6325008875503543e-05,
+      "loss": 0.1396,
+      "step": 1108
+    },
+    {
+      "epoch": 0.44435548438751,
+      "learning_rate": 1.6346633170548285e-05,
+      "loss": 0.0021,
+      "step": 1110
+    },
+    {
+      "epoch": 0.44515612489991996,
+      "learning_rate": 1.6368207885168897e-05,
+      "loss": 0.2049,
+      "step": 1112
+    },
+    {
+      "epoch": 0.44595676541232987,
+      "learning_rate": 1.6389732850821957e-05,
+      "loss": 0.4087,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4467574059247398,
+      "learning_rate": 1.641120789935263e-05,
+      "loss": 0.0183,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4475580464371497,
+      "learning_rate": 1.6432632862996042e-05,
+      "loss": 0.194,
+      "step": 1118
+    },
+    {
+      "epoch": 0.44835868694955966,
+      "learning_rate": 1.6454007574378637e-05,
+      "loss": 0.195,
+      "step": 1120
+    },
+    {
+      "epoch": 0.44915932746196957,
+      "learning_rate": 1.6475331866519377e-05,
+      "loss": 0.0973,
+      "step": 1122
+    },
+    {
+      "epoch": 0.4499599679743795,
+      "learning_rate": 1.6496605572831134e-05,
+      "loss": 0.2849,
+      "step": 1124
+    },
+    {
+      "epoch": 0.45076060848678945,
+      "learning_rate": 1.6517828527121928e-05,
+      "loss": 0.0539,
+      "step": 1126
+    },
+    {
+      "epoch": 0.45156124899919936,
+      "learning_rate": 1.6539000563596318e-05,
+      "loss": 0.0263,
+      "step": 1128
+    },
+    {
+      "epoch": 0.45236188951160927,
+      "learning_rate": 1.6560121516856586e-05,
+      "loss": 0.0043,
+      "step": 1130
+    },
+    {
+      "epoch": 0.45316253002401924,
+      "learning_rate": 1.6581191221904077e-05,
+      "loss": 0.093,
+      "step": 1132
+    },
+    {
+      "epoch": 0.45396317053642915,
+      "learning_rate": 1.6602209514140542e-05,
+      "loss": 0.0711,
+      "step": 1134
+    },
+    {
+      "epoch": 0.45476381104883906,
+      "learning_rate": 1.6623176229369324e-05,
+      "loss": 0.2443,
+      "step": 1136
+    },
+    {
+      "epoch": 0.455564451561249,
+      "learning_rate": 1.6644091203796694e-05,
+      "loss": 0.4891,
+      "step": 1138
+    },
+    {
+      "epoch": 0.45636509207365894,
+      "learning_rate": 1.6664954274033168e-05,
+      "loss": 0.1636,
+      "step": 1140
+    },
+    {
+      "epoch": 0.45716573258606885,
+      "learning_rate": 1.6685765277094695e-05,
+      "loss": 0.2698,
+      "step": 1142
+    },
+    {
+      "epoch": 0.45796637309847876,
+      "learning_rate": 1.6706524050403996e-05,
+      "loss": 0.7748,
+      "step": 1144
+    },
+    {
+      "epoch": 0.45876701361088873,
+      "learning_rate": 1.6727230431791806e-05,
+      "loss": 0.1163,
+      "step": 1146
+    },
+    {
+      "epoch": 0.45956765412329864,
+      "learning_rate": 1.674788425949818e-05,
+      "loss": 0.0006,
+      "step": 1148
+    },
+    {
+      "epoch": 0.46036829463570855,
+      "learning_rate": 1.6768485372173696e-05,
+      "loss": 0.39,
+      "step": 1150
+    },
+    {
+      "epoch": 0.4611689351481185,
+      "learning_rate": 1.6789033608880735e-05,
+      "loss": 0.064,
+      "step": 1152
+    },
+    {
+      "epoch": 0.46196957566052843,
+      "learning_rate": 1.6809528809094798e-05,
+      "loss": 0.7769,
+      "step": 1154
+    },
+    {
+      "epoch": 0.46277021617293834,
+      "learning_rate": 1.6829970812705674e-05,
+      "loss": 0.151,
+      "step": 1156
+    },
+    {
+      "epoch": 0.46357085668534825,
+      "learning_rate": 1.6850359460018733e-05,
+      "loss": 0.5947,
+      "step": 1158
+    },
+    {
+      "epoch": 0.4643714971977582,
+      "learning_rate": 1.6870694591756165e-05,
+      "loss": 0.0909,
+      "step": 1160
+    },
+    {
+      "epoch": 0.46517213771016813,
+      "learning_rate": 1.689097604905826e-05,
+      "loss": 0.1042,
+      "step": 1162
+    },
+    {
+      "epoch": 0.46597277822257804,
+      "learning_rate": 1.6911203673484577e-05,
+      "loss": 0.4073,
+      "step": 1164
+    },
+    {
+      "epoch": 0.466773418734988,
+      "learning_rate": 1.6931377307015226e-05,
+      "loss": 0.0709,
+      "step": 1166
+    },
+    {
+      "epoch": 0.4675740592473979,
+      "learning_rate": 1.695149679205214e-05,
+      "loss": 0.1424,
+      "step": 1168
+    },
+    {
+      "epoch": 0.46837469975980783,
+      "learning_rate": 1.6971561971420222e-05,
+      "loss": 0.2204,
+      "step": 1170
+    },
+    {
+      "epoch": 0.4691753402722178,
+      "learning_rate": 1.6991572688368628e-05,
+      "loss": 0.5136,
+      "step": 1172
+    },
+    {
+      "epoch": 0.4699759807846277,
+      "learning_rate": 1.701152878657196e-05,
+      "loss": 0.1587,
+      "step": 1174
+    },
+    {
+      "epoch": 0.4707766212970376,
+      "learning_rate": 1.7031430110131562e-05,
+      "loss": 0.2293,
+      "step": 1176
+    },
+    {
+      "epoch": 0.47157726180944753,
+      "learning_rate": 1.705127650357662e-05,
+      "loss": 0.2638,
+      "step": 1178
+    },
+    {
+      "epoch": 0.4723779023218575,
+      "learning_rate": 1.7071067811865467e-05,
+      "loss": 0.4434,
+      "step": 1180
+    },
+    {
+      "epoch": 0.4731785428342674,
+      "learning_rate": 1.7090803880386778e-05,
+      "loss": 0.0871,
+      "step": 1182
+    },
+    {
+      "epoch": 0.4739791833466773,
+      "learning_rate": 1.7110484554960738e-05,
+      "loss": 0.2391,
+      "step": 1184
+    },
+    {
+      "epoch": 0.4747798238590873,
+      "learning_rate": 1.713010968184029e-05,
+      "loss": 0.7516,
+      "step": 1186
+    },
+    {
+      "epoch": 0.4755804643714972,
+      "learning_rate": 1.7149679107712306e-05,
+      "loss": 0.0321,
+      "step": 1188
+    },
+    {
+      "epoch": 0.4763811048839071,
+      "learning_rate": 1.716919267969883e-05,
+      "loss": 0.2413,
+      "step": 1190
+    },
+    {
+      "epoch": 0.4771817453963171,
+      "learning_rate": 1.7188650245358215e-05,
+      "loss": 0.3946,
+      "step": 1192
+    },
+    {
+      "epoch": 0.477982385908727,
+      "learning_rate": 1.7208051652686338e-05,
+      "loss": 0.2789,
+      "step": 1194
+    },
+    {
+      "epoch": 0.4787830264211369,
+      "learning_rate": 1.722739675011779e-05,
+      "loss": 0.0135,
+      "step": 1196
+    },
+    {
+      "epoch": 0.4795836669335468,
+      "learning_rate": 1.7246685386527095e-05,
+      "loss": 0.1909,
+      "step": 1198
+    },
+    {
+      "epoch": 0.4803843074459568,
+      "learning_rate": 1.726591741122981e-05,
+      "loss": 0.2864,
+      "step": 1200
+    },
+    {
+      "epoch": 0.4811849479583667,
+      "learning_rate": 1.7285092673983753e-05,
+      "loss": 0.3336,
+      "step": 1202
+    },
+    {
+      "epoch": 0.4819855884707766,
+      "learning_rate": 1.730421102499021e-05,
+      "loss": 0.1287,
+      "step": 1204
+    },
+    {
+      "epoch": 0.48278622898318657,
+      "learning_rate": 1.7323272314895022e-05,
+      "loss": 0.3104,
+      "step": 1206
+    },
+    {
+      "epoch": 0.4835868694955965,
+      "learning_rate": 1.734227639478982e-05,
+      "loss": 0.2498,
+      "step": 1208
+    },
+    {
+      "epoch": 0.4843875100080064,
+      "learning_rate": 1.736122311621314e-05,
+      "loss": 0.2466,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4851881505204163,
+      "learning_rate": 1.738011233115165e-05,
+      "loss": 0.7192,
+      "step": 1212
+    },
+    {
+      "epoch": 0.4859887910328263,
+      "learning_rate": 1.7398943892041227e-05,
+      "loss": 0.1451,
+      "step": 1214
+    },
+    {
+      "epoch": 0.4867894315452362,
+      "learning_rate": 1.7417717651768144e-05,
+      "loss": 0.5076,
+      "step": 1216
+    },
+    {
+      "epoch": 0.4875900720576461,
+      "learning_rate": 1.743643346367026e-05,
+      "loss": 0.1385,
+      "step": 1218
+    },
+    {
+      "epoch": 0.48839071257005606,
+      "learning_rate": 1.7455091181538087e-05,
+      "loss": 0.192,
+      "step": 1220
+    },
+    {
+      "epoch": 0.489191353082466,
+      "learning_rate": 1.7473690659615992e-05,
+      "loss": 0.328,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4899919935948759,
+      "learning_rate": 1.74922317526033e-05,
+      "loss": 0.1104,
+      "step": 1224
+    },
+    {
+      "epoch": 0.49079263410728585,
+      "learning_rate": 1.7510714315655474e-05,
+      "loss": 0.1754,
+      "step": 1226
+    },
+    {
+      "epoch": 0.49159327461969576,
+      "learning_rate": 1.752913820438519e-05,
+      "loss": 0.2398,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4923939151321057,
+      "learning_rate": 1.7547503274863495e-05,
+      "loss": 0.136,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4931945556445156,
+      "learning_rate": 1.756580938362096e-05,
+      "loss": 0.132,
+      "step": 1232
+    },
+    {
+      "epoch": 0.49399519615692555,
+      "learning_rate": 1.758405638764873e-05,
+      "loss": 0.2823,
+      "step": 1234
+    },
+    {
+      "epoch": 0.49479583666933546,
+      "learning_rate": 1.7602244144399693e-05,
+      "loss": 0.069,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4955964771817454,
+      "learning_rate": 1.7620372511789604e-05,
+      "loss": 0.4614,
+      "step": 1238
+    },
+    {
+      "epoch": 0.49639711769415534,
+      "learning_rate": 1.7638441348198147e-05,
+      "loss": 0.2106,
+      "step": 1240
+    },
+    {
+      "epoch": 0.49719775820656525,
+      "learning_rate": 1.7656450512470077e-05,
+      "loss": 0.0966,
+      "step": 1242
+    },
+    {
+      "epoch": 0.49799839871897517,
+      "learning_rate": 1.7674399863916295e-05,
+      "loss": 0.1319,
+      "step": 1244
+    },
+    {
+      "epoch": 0.49879903923138513,
+      "learning_rate": 1.7692289262315e-05,
+      "loss": 0.2828,
+      "step": 1246
+    },
+    {
+      "epoch": 0.49959967974379504,
+      "learning_rate": 1.771011856791273e-05,
+      "loss": 0.5292,
+      "step": 1248
+    },
+    {
+      "epoch": 0.500400320256205,
+      "learning_rate": 1.7727887641425448e-05,
+      "loss": 0.1641,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5012009607686149,
+      "learning_rate": 1.7745596344039712e-05,
+      "loss": 0.2002,
+      "step": 1252
+    },
+    {
+      "epoch": 0.5020016012810248,
+      "learning_rate": 1.7763244537413657e-05,
+      "loss": 0.2159,
+      "step": 1254
+    },
+    {
+      "epoch": 0.5028022417934348,
+      "learning_rate": 1.7780832083678116e-05,
+      "loss": 0.2908,
+      "step": 1256
+    },
+    {
+      "epoch": 0.5036028823058447,
+      "learning_rate": 1.7798358845437754e-05,
+      "loss": 0.1535,
+      "step": 1258
+    },
+    {
+      "epoch": 0.5044035228182546,
+      "learning_rate": 1.7815824685772035e-05,
+      "loss": 0.3213,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5052041633306645,
+      "learning_rate": 1.7833229468236364e-05,
+      "loss": 0.0867,
+      "step": 1262
+    },
+    {
+      "epoch": 0.5060048038430744,
+      "learning_rate": 1.7850573056863156e-05,
+      "loss": 0.1262,
+      "step": 1264
+    },
+    {
+      "epoch": 0.5068054443554844,
+      "learning_rate": 1.786785531616285e-05,
+      "loss": 0.1487,
+      "step": 1266
+    },
+    {
+      "epoch": 0.5076060848678943,
+      "learning_rate": 1.7885076111125004e-05,
+      "loss": 0.2642,
+      "step": 1268
+    },
+    {
+      "epoch": 0.5084067253803043,
+      "learning_rate": 1.790223530721933e-05,
+      "loss": 0.0003,
+      "step": 1270
+    },
+    {
+      "epoch": 0.5092073658927142,
+      "learning_rate": 1.791933277039679e-05,
+      "loss": 0.145,
+      "step": 1272
+    },
+    {
+      "epoch": 0.5100080064051241,
+      "learning_rate": 1.7936368367090577e-05,
+      "loss": 0.174,
+      "step": 1274
+    },
+    {
+      "epoch": 0.510808646917534,
+      "learning_rate": 1.7953341964217183e-05,
+      "loss": 0.0573,
+      "step": 1276
+    },
+    {
+      "epoch": 0.5116092874299439,
+      "learning_rate": 1.7970253429177477e-05,
+      "loss": 0.6106,
+      "step": 1278
+    },
+    {
+      "epoch": 0.5124099279423538,
+      "learning_rate": 1.7987102629857696e-05,
+      "loss": 0.1791,
+      "step": 1280
+    },
+    {
+      "epoch": 0.5132105684547638,
+      "learning_rate": 1.800388943463047e-05,
+      "loss": 0.2116,
+      "step": 1282
+    },
+    {
+      "epoch": 0.5140112089671738,
+      "learning_rate": 1.8020613712355912e-05,
+      "loss": 0.1299,
+      "step": 1284
+    },
+    {
+      "epoch": 0.5148118494795837,
+      "learning_rate": 1.803727533238257e-05,
+      "loss": 0.1081,
+      "step": 1286
+    },
+    {
+      "epoch": 0.5156124899919936,
+      "learning_rate": 1.805387416454847e-05,
+      "loss": 0.3832,
+      "step": 1288
+    },
+    {
+      "epoch": 0.5164131305044035,
+      "learning_rate": 1.8070410079182195e-05,
+      "loss": 0.1825,
+      "step": 1290
+    },
+    {
+      "epoch": 0.5172137710168134,
+      "learning_rate": 1.8086882947103787e-05,
+      "loss": 0.142,
+      "step": 1292
+    },
+    {
+      "epoch": 0.5180144115292233,
+      "learning_rate": 1.8103292639625842e-05,
+      "loss": 0.0229,
+      "step": 1294
+    },
+    {
+      "epoch": 0.5188150520416333,
+      "learning_rate": 1.811963902855447e-05,
+      "loss": 0.1252,
+      "step": 1296
+    },
+    {
+      "epoch": 0.5196156925540433,
+      "learning_rate": 1.813592198619035e-05,
+      "loss": 0.0974,
+      "step": 1298
+    },
+    {
+      "epoch": 0.5204163330664532,
+      "learning_rate": 1.8152141385329658e-05,
+      "loss": 0.1453,
+      "step": 1300
+    },
+    {
+      "epoch": 0.5212169735788631,
+      "learning_rate": 1.816829709926509e-05,
+      "loss": 0.2697,
+      "step": 1302
+    },
+    {
+      "epoch": 0.522017614091273,
+      "learning_rate": 1.8184389001786895e-05,
+      "loss": 0.1717,
+      "step": 1304
+    },
+    {
+      "epoch": 0.5228182546036829,
+      "learning_rate": 1.8200416967183785e-05,
+      "loss": 0.0476,
+      "step": 1306
+    },
+    {
+      "epoch": 0.5236188951160928,
+      "learning_rate": 1.821638087024396e-05,
+      "loss": 0.0677,
+      "step": 1308
+    },
+    {
+      "epoch": 0.5244195356285029,
+      "learning_rate": 1.8232280586256097e-05,
+      "loss": 0.3593,
+      "step": 1310
+    },
+    {
+      "epoch": 0.5252201761409128,
+      "learning_rate": 1.8248115991010296e-05,
+      "loss": 0.0539,
+      "step": 1312
+    },
+    {
+      "epoch": 0.5260208166533227,
+      "learning_rate": 1.8263886960799055e-05,
+      "loss": 0.0399,
+      "step": 1314
+    },
+    {
+      "epoch": 0.5268214571657326,
+      "learning_rate": 1.8279593372418264e-05,
+      "loss": 1.1203,
+      "step": 1316
+    },
+    {
+      "epoch": 0.5276220976781425,
+      "learning_rate": 1.829523510316813e-05,
+      "loss": 0.018,
+      "step": 1318
+    },
+    {
+      "epoch": 0.5284227381905524,
+      "learning_rate": 1.8310812030854155e-05,
+      "loss": 0.4293,
+      "step": 1320
+    },
+    {
+      "epoch": 0.5292233787029623,
+      "learning_rate": 1.832632403378808e-05,
+      "loss": 0.2101,
+      "step": 1322
+    },
+    {
+      "epoch": 0.5300240192153723,
+      "learning_rate": 1.834177099078887e-05,
+      "loss": 0.3111,
+      "step": 1324
+    },
+    {
+      "epoch": 0.5308246597277823,
+      "learning_rate": 1.8357152781183606e-05,
+      "loss": 0.1155,
+      "step": 1326
+    },
+    {
+      "epoch": 0.5316253002401922,
+      "learning_rate": 1.8372469284808465e-05,
+      "loss": 0.2028,
+      "step": 1328
+    },
+    {
+      "epoch": 0.5324259407526021,
+      "learning_rate": 1.8387720382009665e-05,
+      "loss": 0.2867,
+      "step": 1330
+    },
+    {
+      "epoch": 0.533226581265012,
+      "learning_rate": 1.840290595364436e-05,
+      "loss": 0.2181,
+      "step": 1332
+    },
+    {
+      "epoch": 0.5340272217774219,
+      "learning_rate": 1.8418025881081606e-05,
+      "loss": 0.2585,
+      "step": 1334
+    },
+    {
+      "epoch": 0.5348278622898318,
+      "learning_rate": 1.8433080046203286e-05,
+      "loss": 0.662,
+      "step": 1336
+    },
+    {
+      "epoch": 0.5356285028022418,
+      "learning_rate": 1.844806833140501e-05,
+      "loss": 0.1319,
+      "step": 1338
+    },
+    {
+      "epoch": 0.5364291433146517,
+      "learning_rate": 1.8462990619597054e-05,
+      "loss": 0.0692,
+      "step": 1340
+    },
+    {
+      "epoch": 0.5372297838270617,
+      "learning_rate": 1.8477846794205258e-05,
+      "loss": 0.0794,
+      "step": 1342
+    },
+    {
+      "epoch": 0.5380304243394716,
+      "learning_rate": 1.8492636739171966e-05,
+      "loss": 0.0022,
+      "step": 1344
+    },
+    {
+      "epoch": 0.5388310648518815,
+      "learning_rate": 1.85073603389569e-05,
+      "loss": 0.0033,
+      "step": 1346
+    },
+    {
+      "epoch": 0.5396317053642914,
+      "learning_rate": 1.8522017478538067e-05,
+      "loss": 0.0789,
+      "step": 1348
+    },
+    {
+      "epoch": 0.5404323458767014,
+      "learning_rate": 1.8536608043412695e-05,
+      "loss": 0.0815,
+      "step": 1350
+    },
+    {
+      "epoch": 0.5412329863891113,
+      "learning_rate": 1.855113191959808e-05,
+      "loss": 0.1315,
+      "step": 1352
+    },
+    {
+      "epoch": 0.5420336269015212,
+      "learning_rate": 1.856558899363248e-05,
+      "loss": 0.3248,
+      "step": 1354
+    },
+    {
+      "epoch": 0.5428342674139311,
+      "learning_rate": 1.8579979152576063e-05,
+      "loss": 0.0728,
+      "step": 1356
+    },
+    {
+      "epoch": 0.5436349079263411,
+      "learning_rate": 1.85943022840117e-05,
+      "loss": 1.013,
+      "step": 1358
+    },
+    {
+      "epoch": 0.544435548438751,
+      "learning_rate": 1.8608558276045895e-05,
+      "loss": 0.4096,
+      "step": 1360
+    },
+    {
+      "epoch": 0.5452361889511609,
+      "learning_rate": 1.862274701730967e-05,
+      "loss": 0.2427,
+      "step": 1362
+    },
+    {
+      "epoch": 0.5460368294635709,
+      "learning_rate": 1.86368683969594e-05,
+      "loss": 0.3155,
+      "step": 1364
+    },
+    {
+      "epoch": 0.5468374699759808,
+      "learning_rate": 1.865092230467769e-05,
+      "loss": 0.4296,
+      "step": 1366
+    },
+    {
+      "epoch": 0.5476381104883907,
+      "learning_rate": 1.866490863067425e-05,
+      "loss": 0.2626,
+      "step": 1368
+    },
+    {
+      "epoch": 0.5484387510008006,
+      "learning_rate": 1.8678827265686753e-05,
+      "loss": 0.1588,
+      "step": 1370
+    },
+    {
+      "epoch": 0.5492393915132106,
+      "learning_rate": 1.8692678100981663e-05,
+      "loss": 0.1276,
+      "step": 1372
+    },
+    {
+      "epoch": 0.5500400320256205,
+      "learning_rate": 1.87064610283551e-05,
+      "loss": 0.102,
+      "step": 1374
+    },
+    {
+      "epoch": 0.5508406725380304,
+      "learning_rate": 1.8720175940133705e-05,
+      "loss": 0.1099,
+      "step": 1376
+    },
+    {
+      "epoch": 0.5516413130504404,
+      "learning_rate": 1.873382272917545e-05,
+      "loss": 0.3393,
+      "step": 1378
+    },
+    {
+      "epoch": 0.5524419535628503,
+      "learning_rate": 1.8747401288870472e-05,
+      "loss": 0.08,
+      "step": 1380
+    },
+    {
+      "epoch": 0.5532425940752602,
+      "learning_rate": 1.876091151314196e-05,
+      "loss": 0.2856,
+      "step": 1382
+    },
+    {
+      "epoch": 0.5540432345876701,
+      "learning_rate": 1.877435329644691e-05,
+      "loss": 0.2085,
+      "step": 1384
+    },
+    {
+      "epoch": 0.55484387510008,
+      "learning_rate": 1.8787726533776996e-05,
+      "loss": 0.4122,
+      "step": 1386
+    },
+    {
+      "epoch": 0.55564451561249,
+      "learning_rate": 1.8801031120659393e-05,
+      "loss": 0.1118,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5564451561248999,
+      "learning_rate": 1.8814266953157557e-05,
+      "loss": 0.0669,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5572457966373099,
+      "learning_rate": 1.8827433927872066e-05,
+      "loss": 0.411,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5580464371497198,
+      "learning_rate": 1.8840531941941415e-05,
+      "loss": 0.0661,
+      "step": 1394
+    },
+    {
+      "epoch": 0.5588470776621297,
+      "learning_rate": 1.8853560893042854e-05,
+      "loss": 0.4688,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5596477181745396,
+      "learning_rate": 1.8866520679393127e-05,
+      "loss": 0.0438,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5604483586869495,
+      "learning_rate": 1.8879411199749303e-05,
+      "loss": 0.0017,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5612489991993594,
+      "learning_rate": 1.889223235340958e-05,
+      "loss": 0.0591,
+      "step": 1402
+    },
+    {
+      "epoch": 0.5620496397117695,
+      "learning_rate": 1.8904984040214037e-05,
+      "loss": 0.4956,
+      "step": 1404
+    },
+    {
+      "epoch": 0.5628502802241794,
+      "learning_rate": 1.8917666160545436e-05,
+      "loss": 0.029,
+      "step": 1406
+    },
+    {
+      "epoch": 0.5636509207365893,
+      "learning_rate": 1.893027861533002e-05,
+      "loss": 0.1913,
+      "step": 1408
+    },
+    {
+      "epoch": 0.5644515612489992,
+      "learning_rate": 1.894282130603823e-05,
+      "loss": 0.2488,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5652522017614091,
+      "learning_rate": 1.8955294134685528e-05,
+      "loss": 0.0893,
+      "step": 1412
+    },
+    {
+      "epoch": 0.566052842273819,
+      "learning_rate": 1.896769700383315e-05,
+      "loss": 0.3816,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5668534827862289,
+      "learning_rate": 1.898002981658886e-05,
+      "loss": 0.3141,
+      "step": 1416
+    },
+    {
+      "epoch": 0.567654123298639,
+      "learning_rate": 1.899229247660769e-05,
+      "loss": 0.0574,
+      "step": 1418
+    },
+    {
+      "epoch": 0.5684547638110489,
+      "learning_rate": 1.9004484888092724e-05,
+      "loss": 0.5658,
+      "step": 1420
+    },
+    {
+      "epoch": 0.5692554043234588,
+      "learning_rate": 1.901660695579585e-05,
+      "loss": 0.4091,
+      "step": 1422
+    },
+    {
+      "epoch": 0.5700560448358687,
+      "learning_rate": 1.9028658585018455e-05,
+      "loss": 0.2643,
+      "step": 1424
+    },
+    {
+      "epoch": 0.5708566853482786,
+      "learning_rate": 1.9040639681612212e-05,
+      "loss": 0.0455,
+      "step": 1426
+    },
+    {
+      "epoch": 0.5716573258606885,
+      "learning_rate": 1.9052550151979816e-05,
+      "loss": 0.2142,
+      "step": 1428
+    },
+    {
+      "epoch": 0.5724579663730984,
+      "learning_rate": 1.9064389903075676e-05,
+      "loss": 0.2532,
+      "step": 1430
+    },
+    {
+      "epoch": 0.5732586068855084,
+      "learning_rate": 1.9076158842406674e-05,
+      "loss": 1.2189,
+      "step": 1432
+    },
+    {
+      "epoch": 0.5740592473979184,
+      "learning_rate": 1.9087856878032886e-05,
+      "loss": 0.1078,
+      "step": 1434
+    },
+    {
+      "epoch": 0.5748598879103283,
+      "learning_rate": 1.909948391856829e-05,
+      "loss": 0.3039,
+      "step": 1436
+    },
+    {
+      "epoch": 0.5756605284227382,
+      "learning_rate": 1.911103987318148e-05,
+      "loss": 0.2527,
+      "step": 1438
+    },
+    {
+      "epoch": 0.5764611689351481,
+      "learning_rate": 1.912252465159637e-05,
+      "loss": 0.1365,
+      "step": 1440
+    },
+    {
+      "epoch": 0.577261809447558,
+      "learning_rate": 1.913393816409294e-05,
+      "loss": 0.2057,
+      "step": 1442
+    },
+    {
+      "epoch": 0.578062449959968,
+      "learning_rate": 1.9145280321507872e-05,
+      "loss": 0.5137,
+      "step": 1444
+    },
+    {
+      "epoch": 0.5788630904723779,
+      "learning_rate": 1.9156551035235288e-05,
+      "loss": 0.0381,
+      "step": 1446
+    },
+    {
+      "epoch": 0.5796637309847879,
+      "learning_rate": 1.9167750217227454e-05,
+      "loss": 0.3175,
+      "step": 1448
+    },
+    {
+      "epoch": 0.5804643714971978,
+      "learning_rate": 1.9178877779995423e-05,
+      "loss": 0.3451,
+      "step": 1450
+    },
+    {
+      "epoch": 0.5812650120096077,
+      "learning_rate": 1.9189933636609747e-05,
+      "loss": 0.4345,
+      "step": 1452
+    },
+    {
+      "epoch": 0.5820656525220176,
+      "learning_rate": 1.9200917700701173e-05,
+      "loss": 0.2699,
+      "step": 1454
+    },
+    {
+      "epoch": 0.5828662930344275,
+      "learning_rate": 1.9211829886461274e-05,
+      "loss": 0.3656,
+      "step": 1456
+    },
+    {
+      "epoch": 0.5836669335468375,
+      "learning_rate": 1.9222670108643146e-05,
+      "loss": 0.0408,
+      "step": 1458
+    },
+    {
+      "epoch": 0.5844675740592474,
+      "learning_rate": 1.9233438282562085e-05,
+      "loss": 0.0889,
+      "step": 1460
+    },
+    {
+      "epoch": 0.5852682145716573,
+      "learning_rate": 1.924413432409622e-05,
+      "loss": 0.4903,
+      "step": 1462
+    },
+    {
+      "epoch": 0.5860688550840673,
+      "learning_rate": 1.925475814968719e-05,
+      "loss": 0.1423,
+      "step": 1464
+    },
+    {
+      "epoch": 0.5868694955964772,
+      "learning_rate": 1.926530967634078e-05,
+      "loss": 0.4906,
+      "step": 1466
+    },
+    {
+      "epoch": 0.5876701361088871,
+      "learning_rate": 1.9275788821627607e-05,
+      "loss": 0.2471,
+      "step": 1468
+    },
+    {
+      "epoch": 0.588470776621297,
+      "learning_rate": 1.9286195503683705e-05,
+      "loss": 0.3457,
+      "step": 1470
+    },
+    {
+      "epoch": 0.589271417133707,
+      "learning_rate": 1.9296529641211215e-05,
+      "loss": 0.2819,
+      "step": 1472
+    },
+    {
+      "epoch": 0.5900720576461169,
+      "learning_rate": 1.9306791153479004e-05,
+      "loss": 0.2074,
+      "step": 1474
+    },
+    {
+      "epoch": 0.5908726981585268,
+      "learning_rate": 1.9316979960323286e-05,
+      "loss": 0.1451,
+      "step": 1476
+    },
+    {
+      "epoch": 0.5916733386709367,
+      "learning_rate": 1.932709598214825e-05,
+      "loss": 0.2294,
+      "step": 1478
+    },
+    {
+      "epoch": 0.5924739791833467,
+      "learning_rate": 1.9337139139926707e-05,
+      "loss": 0.4399,
+      "step": 1480
+    },
+    {
+      "epoch": 0.5932746196957566,
+      "learning_rate": 1.9347109355200672e-05,
+      "loss": 0.2047,
+      "step": 1482
+    },
+    {
+      "epoch": 0.5940752602081665,
+      "learning_rate": 1.935700655008199e-05,
+      "loss": 0.2363,
+      "step": 1484
+    },
+    {
+      "epoch": 0.5948759007205765,
+      "learning_rate": 1.9366830647252967e-05,
+      "loss": 0.1318,
+      "step": 1486
+    },
+    {
+      "epoch": 0.5956765412329864,
+      "learning_rate": 1.9376581569966933e-05,
+      "loss": 0.1546,
+      "step": 1488
+    },
+    {
+      "epoch": 0.5964771817453963,
+      "learning_rate": 1.9386259242048883e-05,
+      "loss": 0.1202,
+      "step": 1490
+    },
+    {
+      "epoch": 0.5972778222578062,
+      "learning_rate": 1.939586358789602e-05,
+      "loss": 0.0085,
+      "step": 1492
+    },
+    {
+      "epoch": 0.5980784627702161,
+      "learning_rate": 1.940539453247842e-05,
+      "loss": 0.0263,
+      "step": 1494
+    },
+    {
+      "epoch": 0.5988791032826261,
+      "learning_rate": 1.9414852001339547e-05,
+      "loss": 0.2422,
+      "step": 1496
+    },
+    {
+      "epoch": 0.5996797437950361,
+      "learning_rate": 1.9424235920596863e-05,
+      "loss": 0.0332,
+      "step": 1498
+    },
+    {
+      "epoch": 0.600480384307446,
+      "learning_rate": 1.9433546216942423e-05,
+      "loss": 1.3784,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6012810248198559,
+      "learning_rate": 1.944278281764342e-05,
+      "loss": 0.0198,
+      "step": 1502
+    },
+    {
+      "epoch": 0.6020816653322658,
+      "learning_rate": 1.945194565054276e-05,
+      "loss": 0.0887,
+      "step": 1504
+    },
+    {
+      "epoch": 0.6028823058446757,
+      "learning_rate": 1.9461034644059637e-05,
+      "loss": 0.3157,
+      "step": 1506
+    },
+    {
+      "epoch": 0.6036829463570856,
+      "learning_rate": 1.9470049727190073e-05,
+      "loss": 0.3134,
+      "step": 1508
+    },
+    {
+      "epoch": 0.6044835868694955,
+      "learning_rate": 1.9478990829507504e-05,
+      "loss": 0.1176,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6052842273819056,
+      "learning_rate": 1.948785788116329e-05,
+      "loss": 0.0093,
+      "step": 1512
+    },
+    {
+      "epoch": 0.6060848678943155,
+      "learning_rate": 1.9496650812887286e-05,
+      "loss": 0.0659,
+      "step": 1514
+    },
+    {
+      "epoch": 0.6068855084067254,
+      "learning_rate": 1.9505369555988395e-05,
+      "loss": 0.1703,
+      "step": 1516
+    },
+    {
+      "epoch": 0.6076861489191353,
+      "learning_rate": 1.951401404235505e-05,
+      "loss": 0.0626,
+      "step": 1518
+    },
+    {
+      "epoch": 0.6084867894315452,
+      "learning_rate": 1.952258420445583e-05,
+      "loss": 0.1296,
+      "step": 1520
+    },
+    {
+      "epoch": 0.6092874299439551,
+      "learning_rate": 1.9531079975339912e-05,
+      "loss": 0.1258,
+      "step": 1522
+    },
+    {
+      "epoch": 0.610088070456365,
+      "learning_rate": 1.953950128863762e-05,
+      "loss": 0.0125,
+      "step": 1524
+    },
+    {
+      "epoch": 0.6108887109687751,
+      "learning_rate": 1.9547848078560975e-05,
+      "loss": 0.555,
+      "step": 1526
+    },
+    {
+      "epoch": 0.611689351481185,
+      "learning_rate": 1.9556120279904144e-05,
+      "loss": 0.0572,
+      "step": 1528
+    },
+    {
+      "epoch": 0.6124899919935949,
+      "learning_rate": 1.956431782804402e-05,
+      "loss": 0.2353,
+      "step": 1530
+    },
+    {
+      "epoch": 0.6132906325060048,
+      "learning_rate": 1.957244065894066e-05,
+      "loss": 0.4456,
+      "step": 1532
+    },
+    {
+      "epoch": 0.6140912730184147,
+      "learning_rate": 1.9580488709137858e-05,
+      "loss": 0.2388,
+      "step": 1534
+    },
+    {
+      "epoch": 0.6148919135308246,
+      "learning_rate": 1.9588461915763566e-05,
+      "loss": 0.1931,
+      "step": 1536
+    },
+    {
+      "epoch": 0.6156925540432346,
+      "learning_rate": 1.9596360216530436e-05,
+      "loss": 0.0611,
+      "step": 1538
+    },
+    {
+      "epoch": 0.6164931945556446,
+      "learning_rate": 1.9604183549736283e-05,
+      "loss": 0.067,
+      "step": 1540
+    },
+    {
+      "epoch": 0.6172938350680545,
+      "learning_rate": 1.961193185426459e-05,
+      "loss": 0.2984,
+      "step": 1542
+    },
+    {
+      "epoch": 0.6180944755804644,
+      "learning_rate": 1.9619605069584954e-05,
+      "loss": 0.0881,
+      "step": 1544
+    },
+    {
+      "epoch": 0.6188951160928743,
+      "learning_rate": 1.9627203135753573e-05,
+      "loss": 0.2874,
+      "step": 1546
+    },
+    {
+      "epoch": 0.6196957566052842,
+      "learning_rate": 1.9634725993413744e-05,
+      "loss": 0.4476,
+      "step": 1548
+    },
+    {
+      "epoch": 0.6204963971176941,
+      "learning_rate": 1.9642173583796265e-05,
+      "loss": 0.031,
+      "step": 1550
+    },
+    {
+      "epoch": 0.6212970376301041,
+      "learning_rate": 1.964954584871995e-05,
+      "loss": 0.0747,
+      "step": 1552
+    },
+    {
+      "epoch": 0.622097678142514,
+      "learning_rate": 1.9656842730592046e-05,
+      "loss": 0.4271,
+      "step": 1554
+    },
+    {
+      "epoch": 0.622898318654924,
+      "learning_rate": 1.966406417240872e-05,
+      "loss": 0.2952,
+      "step": 1556
+    },
+    {
+      "epoch": 0.6236989591673339,
+      "learning_rate": 1.967121011775546e-05,
+      "loss": 0.2226,
+      "step": 1558
+    },
+    {
+      "epoch": 0.6244995996797438,
+      "learning_rate": 1.967828051080755e-05,
+      "loss": 0.3785,
+      "step": 1560
+    },
+    {
+      "epoch": 0.6253002401921537,
+      "learning_rate": 1.9685275296330497e-05,
+      "loss": 0.0303,
+      "step": 1562
+    },
+    {
+      "epoch": 0.6261008807045636,
+      "learning_rate": 1.969219441968046e-05,
+      "loss": 0.3485,
+      "step": 1564
+    },
+    {
+      "epoch": 0.6269015212169736,
+      "learning_rate": 1.969903782680467e-05,
+      "loss": 0.2553,
+      "step": 1566
+    },
+    {
+      "epoch": 0.6277021617293835,
+      "learning_rate": 1.9705805464241856e-05,
+      "loss": 0.1104,
+      "step": 1568
+    },
+    {
+      "epoch": 0.6285028022417934,
+      "learning_rate": 1.9712497279122692e-05,
+      "loss": 0.0624,
+      "step": 1570
+    },
+    {
+      "epoch": 0.6293034427542034,
+      "learning_rate": 1.971911321917015e-05,
+      "loss": 0.0821,
+      "step": 1572
+    },
+    {
+      "epoch": 0.6301040832666133,
+      "learning_rate": 1.9725653232699962e-05,
+      "loss": 0.1675,
+      "step": 1574
+    },
+    {
+      "epoch": 0.6309047237790232,
+      "learning_rate": 1.9732117268621005e-05,
+      "loss": 0.0596,
+      "step": 1576
+    },
+    {
+      "epoch": 0.6317053642914331,
+      "learning_rate": 1.9738505276435692e-05,
+      "loss": 0.5358,
+      "step": 1578
+    },
+    {
+      "epoch": 0.6325060048038431,
+      "learning_rate": 1.9744817206240377e-05,
+      "loss": 0.0355,
+      "step": 1580
+    },
+    {
+      "epoch": 0.633306645316253,
+      "learning_rate": 1.9751053008725736e-05,
+      "loss": 0.1109,
+      "step": 1582
+    },
+    {
+      "epoch": 0.6341072858286629,
+      "learning_rate": 1.9757212635177177e-05,
+      "loss": 0.0341,
+      "step": 1584
+    },
+    {
+      "epoch": 0.6349079263410728,
+      "learning_rate": 1.9763296037475174e-05,
+      "loss": 0.025,
+      "step": 1586
+    },
+    {
+      "epoch": 0.6357085668534828,
+      "learning_rate": 1.976930316809569e-05,
+      "loss": 0.0736,
+      "step": 1588
+    },
+    {
+      "epoch": 0.6365092073658927,
+      "learning_rate": 1.9775233980110524e-05,
+      "loss": 0.0179,
+      "step": 1590
+    },
+    {
+      "epoch": 0.6373098478783027,
+      "learning_rate": 1.978108842718768e-05,
+      "loss": 0.0019,
+      "step": 1592
+    },
+    {
+      "epoch": 0.6381104883907126,
+      "learning_rate": 1.9786866463591732e-05,
+      "loss": 0.1085,
+      "step": 1594
+    },
+    {
+      "epoch": 0.6389111289031225,
+      "learning_rate": 1.9792568044184176e-05,
+      "loss": 0.0477,
+      "step": 1596
+    },
+    {
+      "epoch": 0.6397117694155324,
+      "learning_rate": 1.9798193124423804e-05,
+      "loss": 0.0333,
+      "step": 1598
+    },
+    {
+      "epoch": 0.6405124099279423,
+      "learning_rate": 1.9803741660367015e-05,
+      "loss": 0.2606,
+      "step": 1600
+    },
+    {
+      "epoch": 0.6413130504403523,
+      "learning_rate": 1.9809213608668185e-05,
+      "loss": 0.0667,
+      "step": 1602
+    },
+    {
+      "epoch": 0.6421136909527622,
+      "learning_rate": 1.9814608926580007e-05,
+      "loss": 0.7038,
+      "step": 1604
+    },
+    {
+      "epoch": 0.6429143314651722,
+      "learning_rate": 1.9819927571953807e-05,
+      "loss": 0.0585,
+      "step": 1606
+    },
+    {
+      "epoch": 0.6437149719775821,
+      "learning_rate": 1.9825169503239885e-05,
+      "loss": 0.167,
+      "step": 1608
+    },
+    {
+      "epoch": 0.644515612489992,
+      "learning_rate": 1.983033467948784e-05,
+      "loss": 0.0716,
+      "step": 1610
+    },
+    {
+      "epoch": 0.6453162530024019,
+      "learning_rate": 1.9835423060346892e-05,
+      "loss": 0.2851,
+      "step": 1612
+    },
+    {
+      "epoch": 0.6461168935148118,
+      "learning_rate": 1.9840434606066182e-05,
+      "loss": 0.0759,
+      "step": 1614
+    },
+    {
+      "epoch": 0.6469175340272217,
+      "learning_rate": 1.9845369277495102e-05,
+      "loss": 0.6611,
+      "step": 1616
+    },
+    {
+      "epoch": 0.6477181745396317,
+      "learning_rate": 1.9850227036083592e-05,
+      "loss": 0.2019,
+      "step": 1618
+    },
+    {
+      "epoch": 0.6485188150520417,
+      "learning_rate": 1.985500784388244e-05,
+      "loss": 0.4445,
+      "step": 1620
+    },
+    {
+      "epoch": 0.6493194555644516,
+      "learning_rate": 1.985971166354357e-05,
+      "loss": 0.0054,
+      "step": 1622
+    },
+    {
+      "epoch": 0.6501200960768615,
+      "learning_rate": 1.9864338458320366e-05,
+      "loss": 0.4025,
+      "step": 1624
+    },
+    {
+      "epoch": 0.6509207365892714,
+      "learning_rate": 1.986888819206792e-05,
+      "loss": 0.238,
+      "step": 1626
+    },
+    {
+      "epoch": 0.6517213771016813,
+      "learning_rate": 1.9873360829243323e-05,
+      "loss": 0.249,
+      "step": 1628
+    },
+    {
+      "epoch": 0.6525220176140912,
+      "learning_rate": 1.9877756334905983e-05,
+      "loss": 0.2378,
+      "step": 1630
+    },
+    {
+      "epoch": 0.6533226581265013,
+      "learning_rate": 1.9882074674717836e-05,
+      "loss": 0.265,
+      "step": 1632
+    },
+    {
+      "epoch": 0.6541232986389112,
+      "learning_rate": 1.988631581494365e-05,
+      "loss": 0.2361,
+      "step": 1634
+    },
+    {
+      "epoch": 0.6549239391513211,
+      "learning_rate": 1.989047972245129e-05,
+      "loss": 0.0343,
+      "step": 1636
+    },
+    {
+      "epoch": 0.655724579663731,
+      "learning_rate": 1.9894566364711965e-05,
+      "loss": 0.1859,
+      "step": 1638
+    },
+    {
+      "epoch": 0.6565252201761409,
+      "learning_rate": 1.989857570980049e-05,
+      "loss": 0.1679,
+      "step": 1640
+    },
+    {
+      "epoch": 0.6573258606885508,
+      "learning_rate": 1.990250772639552e-05,
+      "loss": 0.1719,
+      "step": 1642
+    },
+    {
+      "epoch": 0.6581265012009607,
+      "learning_rate": 1.9906362383779826e-05,
+      "loss": 0.1317,
+      "step": 1644
+    },
+    {
+      "epoch": 0.6589271417133707,
+      "learning_rate": 1.99101396518405e-05,
+      "loss": 0.313,
+      "step": 1646
+    },
+    {
+      "epoch": 0.6597277822257807,
+      "learning_rate": 1.9913839501069213e-05,
+      "loss": 0.1333,
+      "step": 1648
+    },
+    {
+      "epoch": 0.6605284227381906,
+      "learning_rate": 1.9917461902562435e-05,
+      "loss": 0.0822,
+      "step": 1650
+    },
+    {
+      "epoch": 0.6613290632506005,
+      "learning_rate": 1.9921006828021666e-05,
+      "loss": 0.1722,
+      "step": 1652
+    },
+    {
+      "epoch": 0.6621297037630104,
+      "learning_rate": 1.9924474249753652e-05,
+      "loss": 0.1906,
+      "step": 1654
+    },
+    {
+      "epoch": 0.6629303442754203,
+      "learning_rate": 1.9927864140670615e-05,
+      "loss": 0.1694,
+      "step": 1656
+    },
+    {
+      "epoch": 0.6637309847878302,
+      "learning_rate": 1.9931176474290438e-05,
+      "loss": 0.0376,
+      "step": 1658
+    },
+    {
+      "epoch": 0.6645316253002402,
+      "learning_rate": 1.99344112247369e-05,
+      "loss": 0.0567,
+      "step": 1660
+    },
+    {
+      "epoch": 0.6653322658126501,
+      "learning_rate": 1.9937568366739858e-05,
+      "loss": 0.2487,
+      "step": 1662
+    },
+    {
+      "epoch": 0.6661329063250601,
+      "learning_rate": 1.9940647875635463e-05,
+      "loss": 0.016,
+      "step": 1664
+    },
+    {
+      "epoch": 0.66693354683747,
+      "learning_rate": 1.9943649727366335e-05,
+      "loss": 0.9333,
+      "step": 1666
+    },
+    {
+      "epoch": 0.6677341873498799,
+      "learning_rate": 1.994657389848176e-05,
+      "loss": 0.4249,
+      "step": 1668
+    },
+    {
+      "epoch": 0.6685348278622898,
+      "learning_rate": 1.994942036613787e-05,
+      "loss": 0.6759,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6693354683746997,
+      "learning_rate": 1.9952189108097825e-05,
+      "loss": 0.0906,
+      "step": 1672
+    },
+    {
+      "epoch": 0.6701361088871097,
+      "learning_rate": 1.995488010273198e-05,
+      "loss": 0.0132,
+      "step": 1674
+    },
+    {
+      "epoch": 0.6709367493995196,
+      "learning_rate": 1.9957493329018064e-05,
+      "loss": 0.0983,
+      "step": 1676
+    },
+    {
+      "epoch": 0.6717373899119295,
+      "learning_rate": 1.9960028766541336e-05,
+      "loss": 0.4848,
+      "step": 1678
+    },
+    {
+      "epoch": 0.6725380304243395,
+      "learning_rate": 1.996248639549475e-05,
+      "loss": 0.1295,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6733386709367494,
+      "learning_rate": 1.9964866196679105e-05,
+      "loss": 0.8247,
+      "step": 1682
+    },
+    {
+      "epoch": 0.6741393114491593,
+      "learning_rate": 1.9967168151503196e-05,
+      "loss": 0.3808,
+      "step": 1684
+    },
+    {
+      "epoch": 0.6749399519615693,
+      "learning_rate": 1.9969392241983957e-05,
+      "loss": 0.0001,
+      "step": 1686
+    },
+    {
+      "epoch": 0.6757405924739792,
+      "learning_rate": 1.997153845074662e-05,
+      "loss": 0.4078,
+      "step": 1688
+    },
+    {
+      "epoch": 0.6765412329863891,
+      "learning_rate": 1.9973606761024813e-05,
+      "loss": 0.094,
+      "step": 1690
+    },
+    {
+      "epoch": 0.677341873498799,
+      "learning_rate": 1.997559715666073e-05,
+      "loss": 0.1079,
+      "step": 1692
+    },
+    {
+      "epoch": 0.678142514011209,
+      "learning_rate": 1.9977509622105233e-05,
+      "loss": 0.057,
+      "step": 1694
+    },
+    {
+      "epoch": 0.6789431545236189,
+      "learning_rate": 1.9979344142417986e-05,
+      "loss": 0.1165,
+      "step": 1696
+    },
+    {
+      "epoch": 0.6797437950360288,
+      "learning_rate": 1.9981100703267567e-05,
+      "loss": 0.2249,
+      "step": 1698
+    },
+    {
+      "epoch": 0.6805444355484388,
+      "learning_rate": 1.998277929093157e-05,
+      "loss": 0.3387,
+      "step": 1700
+    },
+    {
+      "epoch": 0.6813450760608487,
+      "learning_rate": 1.998437989229673e-05,
+      "loss": 0.0786,
+      "step": 1702
+    },
+    {
+      "epoch": 0.6821457165732586,
+      "learning_rate": 1.9985902494859023e-05,
+      "loss": 0.1919,
+      "step": 1704
+    },
+    {
+      "epoch": 0.6829463570856685,
+      "learning_rate": 1.998734708672375e-05,
+      "loss": 0.1849,
+      "step": 1706
+    },
+    {
+      "epoch": 0.6837469975980784,
+      "learning_rate": 1.9988713656605635e-05,
+      "loss": 0.0975,
+      "step": 1708
+    },
+    {
+      "epoch": 0.6845476381104884,
+      "learning_rate": 1.9990002193828923e-05,
+      "loss": 0.1772,
+      "step": 1710
+    },
+    {
+      "epoch": 0.6853482786228983,
+      "learning_rate": 1.9991212688327456e-05,
+      "loss": 0.218,
+      "step": 1712
+    },
+    {
+      "epoch": 0.6861489191353083,
+      "learning_rate": 1.9992345130644747e-05,
+      "loss": 0.2987,
+      "step": 1714
+    },
+    {
+      "epoch": 0.6869495596477182,
+      "learning_rate": 1.999339951193407e-05,
+      "loss": 0.067,
+      "step": 1716
+    },
+    {
+      "epoch": 0.6877502001601281,
+      "learning_rate": 1.9994375823958504e-05,
+      "loss": 0.1281,
+      "step": 1718
+    },
+    {
+      "epoch": 0.688550840672538,
+      "learning_rate": 1.9995274059091018e-05,
+      "loss": 0.2421,
+      "step": 1720
+    },
+    {
+      "epoch": 0.6893514811849479,
+      "learning_rate": 1.999609421031453e-05,
+      "loss": 0.4262,
+      "step": 1722
+    },
+    {
+      "epoch": 0.6901521216973578,
+      "learning_rate": 1.999683627122195e-05,
+      "loss": 0.2677,
+      "step": 1724
+    },
+    {
+      "epoch": 0.6909527622097679,
+      "learning_rate": 1.9997500236016233e-05,
+      "loss": 0.0343,
+      "step": 1726
+    },
+    {
+      "epoch": 0.6917534027221778,
+      "learning_rate": 1.9998086099510433e-05,
+      "loss": 0.1022,
+      "step": 1728
+    },
+    {
+      "epoch": 0.6925540432345877,
+      "learning_rate": 1.9998593857127736e-05,
+      "loss": 0.1873,
+      "step": 1730
+    },
+    {
+      "epoch": 0.6933546837469976,
+      "learning_rate": 1.99990235049015e-05,
+      "loss": 0.2784,
+      "step": 1732
+    },
+    {
+      "epoch": 0.6941553242594075,
+      "learning_rate": 1.9999375039475275e-05,
+      "loss": 0.0593,
+      "step": 1734
+    },
+    {
+      "epoch": 0.6949559647718174,
+      "learning_rate": 1.999964845810285e-05,
+      "loss": 0.0838,
+      "step": 1736
+    },
+    {
+      "epoch": 0.6957566052842273,
+      "learning_rate": 1.9999843758648253e-05,
+      "loss": 0.0984,
+      "step": 1738
+    },
+    {
+      "epoch": 0.6965572457966374,
+      "learning_rate": 1.999996093958578e-05,
+      "loss": 0.0006,
+      "step": 1740
+    },
+    {
+      "epoch": 0.6973578863090473,
+      "learning_rate": 2e-05,
+      "loss": 0.033,
+      "step": 1742
+    },
+    {
+      "epoch": 0.6981585268214572,
+      "learning_rate": 1.999996093958578e-05,
+      "loss": 0.264,
+      "step": 1744
+    },
+    {
+      "epoch": 0.6989591673338671,
+      "learning_rate": 1.9999843758648253e-05,
+      "loss": 0.2612,
+      "step": 1746
+    },
+    {
+      "epoch": 0.699759807846277,
+      "learning_rate": 1.999964845810285e-05,
+      "loss": 0.1759,
+      "step": 1748
+    },
+    {
+      "epoch": 0.7005604483586869,
+      "learning_rate": 1.9999375039475278e-05,
+      "loss": 0.2072,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7013610888710968,
+      "learning_rate": 1.99990235049015e-05,
+      "loss": 0.4149,
+      "step": 1752
+    },
+    {
+      "epoch": 0.7021617293835068,
+      "learning_rate": 1.9998593857127736e-05,
+      "loss": 0.0984,
+      "step": 1754
+    },
+    {
+      "epoch": 0.7029623698959168,
+      "learning_rate": 1.9998086099510433e-05,
+      "loss": 0.9108,
+      "step": 1756
+    },
+    {
+      "epoch": 0.7037630104083267,
+      "learning_rate": 1.9997500236016233e-05,
+      "loss": 0.0378,
+      "step": 1758
+    },
+    {
+      "epoch": 0.7045636509207366,
+      "learning_rate": 1.999683627122195e-05,
+      "loss": 0.3632,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7053642914331465,
+      "learning_rate": 1.999609421031453e-05,
+      "loss": 0.014,
+      "step": 1762
+    },
+    {
+      "epoch": 0.7061649319455564,
+      "learning_rate": 1.999527405909102e-05,
+      "loss": 0.3102,
+      "step": 1764
+    },
+    {
+      "epoch": 0.7069655724579663,
+      "learning_rate": 1.9994375823958504e-05,
+      "loss": 0.0602,
+      "step": 1766
+    },
+    {
+      "epoch": 0.7077662129703763,
+      "learning_rate": 1.999339951193407e-05,
+      "loss": 0.256,
+      "step": 1768
+    },
+    {
+      "epoch": 0.7085668534827863,
+      "learning_rate": 1.999234513064475e-05,
+      "loss": 0.2158,
+      "step": 1770
+    },
+    {
+      "epoch": 0.7093674939951962,
+      "learning_rate": 1.9991212688327456e-05,
+      "loss": 0.1384,
+      "step": 1772
+    },
+    {
+      "epoch": 0.7101681345076061,
+      "learning_rate": 1.9990002193828923e-05,
+      "loss": 0.1103,
+      "step": 1774
+    },
+    {
+      "epoch": 0.710968775020016,
+      "learning_rate": 1.9988713656605635e-05,
+      "loss": 0.1634,
+      "step": 1776
+    },
+    {
+      "epoch": 0.7117694155324259,
+      "learning_rate": 1.998734708672375e-05,
+      "loss": 0.0888,
+      "step": 1778
+    },
+    {
+      "epoch": 0.7125700560448359,
+      "learning_rate": 1.9985902494859026e-05,
+      "loss": 0.204,
+      "step": 1780
+    },
+    {
+      "epoch": 0.7133706965572458,
+      "learning_rate": 1.9984379892296735e-05,
+      "loss": 0.0226,
+      "step": 1782
+    },
+    {
+      "epoch": 0.7141713370696557,
+      "learning_rate": 1.9982779290931572e-05,
+      "loss": 0.3126,
+      "step": 1784
+    },
+    {
+      "epoch": 0.7149719775820657,
+      "learning_rate": 1.9981100703267567e-05,
+      "loss": 0.0467,
+      "step": 1786
+    },
+    {
+      "epoch": 0.7157726180944756,
+      "learning_rate": 1.997934414241799e-05,
+      "loss": 0.0295,
+      "step": 1788
+    },
+    {
+      "epoch": 0.7165732586068855,
+      "learning_rate": 1.9977509622105236e-05,
+      "loss": 0.5553,
+      "step": 1790
+    },
+    {
+      "epoch": 0.7173738991192954,
+      "learning_rate": 1.997559715666073e-05,
+      "loss": 0.6304,
+      "step": 1792
+    },
+    {
+      "epoch": 0.7181745396317054,
+      "learning_rate": 1.9973606761024813e-05,
+      "loss": 0.0665,
+      "step": 1794
+    },
+    {
+      "epoch": 0.7189751801441153,
+      "learning_rate": 1.997153845074662e-05,
+      "loss": 0.0303,
+      "step": 1796
+    },
+    {
+      "epoch": 0.7197758206565252,
+      "learning_rate": 1.9969392241983957e-05,
+      "loss": 0.0254,
+      "step": 1798
+    },
+    {
+      "epoch": 0.7205764611689351,
+      "learning_rate": 1.9967168151503193e-05,
+      "loss": 0.3363,
+      "step": 1800
+    },
+    {
+      "epoch": 0.7213771016813451,
+      "learning_rate": 1.996486619667911e-05,
+      "loss": 0.3805,
+      "step": 1802
+    },
+    {
+      "epoch": 0.722177742193755,
+      "learning_rate": 1.9962486395494753e-05,
+      "loss": 0.0144,
+      "step": 1804
+    },
+    {
+      "epoch": 0.7229783827061649,
+      "learning_rate": 1.9960028766541336e-05,
+      "loss": 0.2415,
+      "step": 1806
+    },
+    {
+      "epoch": 0.7237790232185749,
+      "learning_rate": 1.9957493329018064e-05,
+      "loss": 0.0346,
+      "step": 1808
+    },
+    {
+      "epoch": 0.7245796637309848,
+      "learning_rate": 1.995488010273198e-05,
+      "loss": 0.1156,
+      "step": 1810
+    },
+    {
+      "epoch": 0.7253803042433947,
+      "learning_rate": 1.9952189108097825e-05,
+      "loss": 0.2213,
+      "step": 1812
+    },
+    {
+      "epoch": 0.7261809447558046,
+      "learning_rate": 1.9949420366137873e-05,
+      "loss": 0.1772,
+      "step": 1814
+    },
+    {
+      "epoch": 0.7269815852682145,
+      "learning_rate": 1.994657389848176e-05,
+      "loss": 0.0259,
+      "step": 1816
+    },
+    {
+      "epoch": 0.7277822257806245,
+      "learning_rate": 1.994364972736634e-05,
+      "loss": 0.1447,
+      "step": 1818
+    },
+    {
+      "epoch": 0.7285828662930345,
+      "learning_rate": 1.9940647875635466e-05,
+      "loss": 0.0001,
+      "step": 1820
+    },
+    {
+      "epoch": 0.7293835068054444,
+      "learning_rate": 1.993756836673986e-05,
+      "loss": 0.1412,
+      "step": 1822
+    },
+    {
+      "epoch": 0.7301841473178543,
+      "learning_rate": 1.99344112247369e-05,
+      "loss": 0.3803,
+      "step": 1824
+    },
+    {
+      "epoch": 0.7309847878302642,
+      "learning_rate": 1.9931176474290438e-05,
+      "loss": 0.1956,
+      "step": 1826
+    },
+    {
+      "epoch": 0.7317854283426741,
+      "learning_rate": 1.9927864140670618e-05,
+      "loss": 0.1739,
+      "step": 1828
+    },
+    {
+      "epoch": 0.732586068855084,
+      "learning_rate": 1.9924474249753656e-05,
+      "loss": 0.3067,
+      "step": 1830
+    },
+    {
+      "epoch": 0.733386709367494,
+      "learning_rate": 1.9921006828021666e-05,
+      "loss": 0.1189,
+      "step": 1832
+    },
+    {
+      "epoch": 0.734187349879904,
+      "learning_rate": 1.9917461902562435e-05,
+      "loss": 0.2977,
+      "step": 1834
+    },
+    {
+      "epoch": 0.7349879903923139,
+      "learning_rate": 1.9913839501069213e-05,
+      "loss": 0.1836,
+      "step": 1836
+    },
+    {
+      "epoch": 0.7357886309047238,
+      "learning_rate": 1.9910139651840497e-05,
+      "loss": 0.3066,
+      "step": 1838
+    },
+    {
+      "epoch": 0.7365892714171337,
+      "learning_rate": 1.9906362383779826e-05,
+      "loss": 0.2902,
+      "step": 1840
+    },
+    {
+      "epoch": 0.7373899119295436,
+      "learning_rate": 1.9902507726395524e-05,
+      "loss": 0.1327,
+      "step": 1842
+    },
+    {
+      "epoch": 0.7381905524419535,
+      "learning_rate": 1.989857570980049e-05,
+      "loss": 0.034,
+      "step": 1844
+    },
+    {
+      "epoch": 0.7389911929543634,
+      "learning_rate": 1.9894566364711965e-05,
+      "loss": 0.2908,
+      "step": 1846
+    },
+    {
+      "epoch": 0.7397918334667735,
+      "learning_rate": 1.9890479722451292e-05,
+      "loss": 0.3809,
+      "step": 1848
+    },
+    {
+      "epoch": 0.7405924739791834,
+      "learning_rate": 1.988631581494365e-05,
+      "loss": 0.0179,
+      "step": 1850
+    },
+    {
+      "epoch": 0.7413931144915933,
+      "learning_rate": 1.9882074674717832e-05,
+      "loss": 0.4259,
+      "step": 1852
+    },
+    {
+      "epoch": 0.7421937550040032,
+      "learning_rate": 1.987775633490599e-05,
+      "loss": 0.5049,
+      "step": 1854
+    },
+    {
+      "epoch": 0.7429943955164131,
+      "learning_rate": 1.987336082924333e-05,
+      "loss": 0.0307,
+      "step": 1856
+    },
+    {
+      "epoch": 0.743795036028823,
+      "learning_rate": 1.986888819206792e-05,
+      "loss": 0.1995,
+      "step": 1858
+    },
+    {
+      "epoch": 0.7445956765412329,
+      "learning_rate": 1.986433845832037e-05,
+      "loss": 0.2614,
+      "step": 1860
+    },
+    {
+      "epoch": 0.745396317053643,
+      "learning_rate": 1.9859711663543573e-05,
+      "loss": 0.3215,
+      "step": 1862
+    },
+    {
+      "epoch": 0.7461969575660529,
+      "learning_rate": 1.9855007843882437e-05,
+      "loss": 0.0051,
+      "step": 1864
+    },
+    {
+      "epoch": 0.7469975980784628,
+      "learning_rate": 1.9850227036083592e-05,
+      "loss": 0.174,
+      "step": 1866
+    },
+    {
+      "epoch": 0.7477982385908727,
+      "learning_rate": 1.9845369277495105e-05,
+      "loss": 0.2507,
+      "step": 1868
+    },
+    {
+      "epoch": 0.7485988791032826,
+      "learning_rate": 1.9840434606066186e-05,
+      "loss": 0.2157,
+      "step": 1870
+    },
+    {
+      "epoch": 0.7493995196156925,
+      "learning_rate": 1.9835423060346892e-05,
+      "loss": 0.2949,
+      "step": 1872
+    },
+    {
+      "epoch": 0.7502001601281025,
+      "learning_rate": 1.983033467948784e-05,
+      "loss": 0.2882,
+      "step": 1874
+    },
+    {
+      "epoch": 0.7510008006405124,
+      "learning_rate": 1.9825169503239885e-05,
+      "loss": 0.1412,
+      "step": 1876
+    },
+    {
+      "epoch": 0.7518014411529224,
+      "learning_rate": 1.9819927571953804e-05,
+      "loss": 0.2432,
+      "step": 1878
+    },
+    {
+      "epoch": 0.7526020816653323,
+      "learning_rate": 1.9814608926580007e-05,
+      "loss": 0.1141,
+      "step": 1880
+    },
+    {
+      "epoch": 0.7534027221777422,
+      "learning_rate": 1.980921360866819e-05,
+      "loss": 0.0728,
+      "step": 1882
+    },
+    {
+      "epoch": 0.7542033626901521,
+      "learning_rate": 1.9803741660367018e-05,
+      "loss": 0.0084,
+      "step": 1884
+    },
+    {
+      "epoch": 0.755004003202562,
+      "learning_rate": 1.9798193124423804e-05,
+      "loss": 0.0052,
+      "step": 1886
+    },
+    {
+      "epoch": 0.755804643714972,
+      "learning_rate": 1.979256804418418e-05,
+      "loss": 1.0712,
+      "step": 1888
+    },
+    {
+      "epoch": 0.7566052842273819,
+      "learning_rate": 1.978686646359173e-05,
+      "loss": 0.0176,
+      "step": 1890
+    },
+    {
+      "epoch": 0.7574059247397918,
+      "learning_rate": 1.9781088427187677e-05,
+      "loss": 0.8224,
+      "step": 1892
+    },
+    {
+      "epoch": 0.7582065652522018,
+      "learning_rate": 1.9775233980110524e-05,
+      "loss": 0.0612,
+      "step": 1894
+    },
+    {
+      "epoch": 0.7590072057646117,
+      "learning_rate": 1.976930316809569e-05,
+      "loss": 1.2486,
+      "step": 1896
+    },
+    {
+      "epoch": 0.7598078462770216,
+      "learning_rate": 1.9763296037475177e-05,
+      "loss": 0.5712,
+      "step": 1898
+    },
+    {
+      "epoch": 0.7606084867894315,
+      "learning_rate": 1.9757212635177177e-05,
+      "loss": 0.253,
+      "step": 1900
+    },
+    {
+      "epoch": 0.7614091273018415,
+      "learning_rate": 1.9751053008725736e-05,
+      "loss": 0.3648,
+      "step": 1902
+    },
+    {
+      "epoch": 0.7622097678142514,
+      "learning_rate": 1.9744817206240374e-05,
+      "loss": 0.3473,
+      "step": 1904
+    },
+    {
+      "epoch": 0.7630104083266613,
+      "learning_rate": 1.9738505276435695e-05,
+      "loss": 0.2674,
+      "step": 1906
+    },
+    {
+      "epoch": 0.7638110488390712,
+      "learning_rate": 1.9732117268621005e-05,
+      "loss": 0.0053,
+      "step": 1908
+    },
+    {
+      "epoch": 0.7646116893514812,
+      "learning_rate": 1.9725653232699962e-05,
+      "loss": 0.0881,
+      "step": 1910
+    },
+    {
+      "epoch": 0.7654123298638911,
+      "learning_rate": 1.9719113219170152e-05,
+      "loss": 0.3642,
+      "step": 1912
+    },
+    {
+      "epoch": 0.7662129703763011,
+      "learning_rate": 1.9712497279122692e-05,
+      "loss": 0.041,
+      "step": 1914
+    },
+    {
+      "epoch": 0.767013610888711,
+      "learning_rate": 1.970580546424186e-05,
+      "loss": 0.3424,
+      "step": 1916
+    },
+    {
+      "epoch": 0.7678142514011209,
+      "learning_rate": 1.969903782680467e-05,
+      "loss": 0.2806,
+      "step": 1918
+    },
+    {
+      "epoch": 0.7686148919135308,
+      "learning_rate": 1.9692194419680463e-05,
+      "loss": 0.2913,
+      "step": 1920
+    },
+    {
+      "epoch": 0.7694155324259407,
+      "learning_rate": 1.96852752963305e-05,
+      "loss": 0.277,
+      "step": 1922
+    },
+    {
+      "epoch": 0.7702161729383507,
+      "learning_rate": 1.9678280510807552e-05,
+      "loss": 0.1203,
+      "step": 1924
+    },
+    {
+      "epoch": 0.7710168134507606,
+      "learning_rate": 1.9671210117755462e-05,
+      "loss": 0.03,
+      "step": 1926
+    },
+    {
+      "epoch": 0.7718174539631706,
+      "learning_rate": 1.966406417240872e-05,
+      "loss": 0.1387,
+      "step": 1928
+    },
+    {
+      "epoch": 0.7726180944755805,
+      "learning_rate": 1.9656842730592046e-05,
+      "loss": 0.196,
+      "step": 1930
+    },
+    {
+      "epoch": 0.7734187349879904,
+      "learning_rate": 1.964954584871995e-05,
+      "loss": 0.1097,
+      "step": 1932
+    },
+    {
+      "epoch": 0.7742193755004003,
+      "learning_rate": 1.9642173583796265e-05,
+      "loss": 0.0572,
+      "step": 1934
+    },
+    {
+      "epoch": 0.7750200160128102,
+      "learning_rate": 1.9634725993413744e-05,
+      "loss": 0.5186,
+      "step": 1936
+    },
+    {
+      "epoch": 0.7758206565252201,
+      "learning_rate": 1.9627203135753576e-05,
+      "loss": 0.0547,
+      "step": 1938
+    },
+    {
+      "epoch": 0.77662129703763,
+      "learning_rate": 1.9619605069584954e-05,
+      "loss": 0.4263,
+      "step": 1940
+    },
+    {
+      "epoch": 0.7774219375500401,
+      "learning_rate": 1.961193185426459e-05,
+      "loss": 0.0049,
+      "step": 1942
+    },
+    {
+      "epoch": 0.77822257806245,
+      "learning_rate": 1.9604183549736287e-05,
+      "loss": 0.3619,
+      "step": 1944
+    },
+    {
+      "epoch": 0.7790232185748599,
+      "learning_rate": 1.959636021653044e-05,
+      "loss": 0.3762,
+      "step": 1946
+    },
+    {
+      "epoch": 0.7798238590872698,
+      "learning_rate": 1.958846191576357e-05,
+      "loss": 1.2455,
+      "step": 1948
+    },
+    {
+      "epoch": 0.7806244995996797,
+      "learning_rate": 1.958048870913786e-05,
+      "loss": 0.6391,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7814251401120896,
+      "learning_rate": 1.9572440658940667e-05,
+      "loss": 0.301,
+      "step": 1952
+    },
+    {
+      "epoch": 0.7822257806244995,
+      "learning_rate": 1.9564317828044022e-05,
+      "loss": 0.2228,
+      "step": 1954
+    },
+    {
+      "epoch": 0.7830264211369096,
+      "learning_rate": 1.955612027990415e-05,
+      "loss": 0.047,
+      "step": 1956
+    },
+    {
+      "epoch": 0.7838270616493195,
+      "learning_rate": 1.9547848078560982e-05,
+      "loss": 0.0756,
+      "step": 1958
+    },
+    {
+      "epoch": 0.7846277021617294,
+      "learning_rate": 1.953950128863763e-05,
+      "loss": 0.1306,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7854283426741393,
+      "learning_rate": 1.9531079975339915e-05,
+      "loss": 0.1555,
+      "step": 1962
+    },
+    {
+      "epoch": 0.7862289831865492,
+      "learning_rate": 1.9522584204455835e-05,
+      "loss": 0.4522,
+      "step": 1964
+    },
+    {
+      "epoch": 0.7870296236989591,
+      "learning_rate": 1.9514014042355054e-05,
+      "loss": 0.4533,
+      "step": 1966
+    },
+    {
+      "epoch": 0.7878302642113691,
+      "learning_rate": 1.9505369555988395e-05,
+      "loss": 0.3241,
+      "step": 1968
+    },
+    {
+      "epoch": 0.7886309047237791,
+      "learning_rate": 1.9496650812887293e-05,
+      "loss": 0.0001,
+      "step": 1970
+    },
+    {
+      "epoch": 0.789431545236189,
+      "learning_rate": 1.9487857881163295e-05,
+      "loss": 0.1759,
+      "step": 1972
+    },
+    {
+      "epoch": 0.7902321857485989,
+      "learning_rate": 1.947899082950751e-05,
+      "loss": 0.1912,
+      "step": 1974
+    },
+    {
+      "epoch": 0.7910328262610088,
+      "learning_rate": 1.947004972719008e-05,
+      "loss": 0.0505,
+      "step": 1976
+    },
+    {
+      "epoch": 0.7918334667734187,
+      "learning_rate": 1.9461034644059637e-05,
+      "loss": 0.0515,
+      "step": 1978
+    },
+    {
+      "epoch": 0.7926341072858286,
+      "learning_rate": 1.945194565054276e-05,
+      "loss": 0.3746,
+      "step": 1980
+    },
+    {
+      "epoch": 0.7934347477982386,
+      "learning_rate": 1.9442782817643425e-05,
+      "loss": 0.0978,
+      "step": 1982
+    },
+    {
+      "epoch": 0.7942353883106485,
+      "learning_rate": 1.9433546216942433e-05,
+      "loss": 0.174,
+      "step": 1984
+    },
+    {
+      "epoch": 0.7950360288230585,
+      "learning_rate": 1.942423592059687e-05,
+      "loss": 0.115,
+      "step": 1986
+    },
+    {
+      "epoch": 0.7958366693354684,
+      "learning_rate": 1.941485200133955e-05,
+      "loss": 0.0653,
+      "step": 1988
+    },
+    {
+      "epoch": 0.7966373098478783,
+      "learning_rate": 1.9405394532478422e-05,
+      "loss": 0.1865,
+      "step": 1990
+    },
+    {
+      "epoch": 0.7974379503602882,
+      "learning_rate": 1.9395863587896025e-05,
+      "loss": 0.1241,
+      "step": 1992
+    },
+    {
+      "epoch": 0.7982385908726981,
+      "learning_rate": 1.938625924204888e-05,
+      "loss": 0.405,
+      "step": 1994
+    },
+    {
+      "epoch": 0.7990392313851081,
+      "learning_rate": 1.937658156996694e-05,
+      "loss": 0.2969,
+      "step": 1996
+    },
+    {
+      "epoch": 0.799839871897518,
+      "learning_rate": 1.9366830647252977e-05,
+      "loss": 0.3522,
+      "step": 1998
+    },
+    {
+      "epoch": 0.800640512409928,
+      "learning_rate": 1.9357006550082e-05,
+      "loss": 0.1591,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8014411529223379,
+      "learning_rate": 1.9347109355200676e-05,
+      "loss": 0.1505,
+      "step": 2002
+    },
+    {
+      "epoch": 0.8022417934347478,
+      "learning_rate": 1.933713913992671e-05,
+      "loss": 0.0585,
+      "step": 2004
+    },
+    {
+      "epoch": 0.8030424339471577,
+      "learning_rate": 1.9327095982148255e-05,
+      "loss": 0.1003,
+      "step": 2006
+    },
+    {
+      "epoch": 0.8038430744595677,
+      "learning_rate": 1.9316979960323283e-05,
+      "loss": 0.4811,
+      "step": 2008
+    },
+    {
+      "epoch": 0.8046437149719776,
+      "learning_rate": 1.9306791153479017e-05,
+      "loss": 0.5298,
+      "step": 2010
+    },
+    {
+      "epoch": 0.8054443554843875,
+      "learning_rate": 1.9296529641211226e-05,
+      "loss": 0.0979,
+      "step": 2012
+    },
+    {
+      "epoch": 0.8062449959967974,
+      "learning_rate": 1.928619550368371e-05,
+      "loss": 0.0981,
+      "step": 2014
+    },
+    {
+      "epoch": 0.8070456365092074,
+      "learning_rate": 1.9275788821627607e-05,
+      "loss": 0.2901,
+      "step": 2016
+    },
+    {
+      "epoch": 0.8078462770216173,
+      "learning_rate": 1.9265309676340783e-05,
+      "loss": 0.0537,
+      "step": 2018
+    },
+    {
+      "epoch": 0.8086469175340272,
+      "learning_rate": 1.9254758149687187e-05,
+      "loss": 0.8714,
+      "step": 2020
+    },
+    {
+      "epoch": 0.8094475580464372,
+      "learning_rate": 1.9244134324096216e-05,
+      "loss": 0.821,
+      "step": 2022
+    },
+    {
+      "epoch": 0.8102481985588471,
+      "learning_rate": 1.9233438282562095e-05,
+      "loss": 0.0681,
+      "step": 2024
+    },
+    {
+      "epoch": 0.811048839071257,
+      "learning_rate": 1.9222670108643156e-05,
+      "loss": 0.2082,
+      "step": 2026
+    },
+    {
+      "epoch": 0.8118494795836669,
+      "learning_rate": 1.9211829886461278e-05,
+      "loss": 0.5836,
+      "step": 2028
+    },
+    {
+      "epoch": 0.8126501200960768,
+      "learning_rate": 1.9200917700701176e-05,
+      "loss": 0.1669,
+      "step": 2030
+    },
+    {
+      "epoch": 0.8134507606084868,
+      "learning_rate": 1.918993363660975e-05,
+      "loss": 0.1299,
+      "step": 2032
+    },
+    {
+      "epoch": 0.8142514011208967,
+      "learning_rate": 1.9178877779995416e-05,
+      "loss": 0.0688,
+      "step": 2034
+    },
+    {
+      "epoch": 0.8150520416333067,
+      "learning_rate": 1.916775021722745e-05,
+      "loss": 0.4094,
+      "step": 2036
+    },
+    {
+      "epoch": 0.8158526821457166,
+      "learning_rate": 1.9156551035235298e-05,
+      "loss": 0.0134,
+      "step": 2038
+    },
+    {
+      "epoch": 0.8166533226581265,
+      "learning_rate": 1.9145280321507872e-05,
+      "loss": 0.2536,
+      "step": 2040
+    },
+    {
+      "epoch": 0.8174539631705364,
+      "learning_rate": 1.9133938164092942e-05,
+      "loss": 0.2122,
+      "step": 2042
+    },
+    {
+      "epoch": 0.8182546036829463,
+      "learning_rate": 1.9122524651596372e-05,
+      "loss": 0.0984,
+      "step": 2044
+    },
+    {
+      "epoch": 0.8190552441953562,
+      "learning_rate": 1.9111039873181475e-05,
+      "loss": 0.2875,
+      "step": 2046
+    },
+    {
+      "epoch": 0.8198558847077662,
+      "learning_rate": 1.9099483918568287e-05,
+      "loss": 0.5369,
+      "step": 2048
+    },
+    {
+      "epoch": 0.8206565252201762,
+      "learning_rate": 1.90878568780329e-05,
+      "loss": 0.3338,
+      "step": 2050
+    },
+    {
+      "epoch": 0.8214571657325861,
+      "learning_rate": 1.907615884240668e-05,
+      "loss": 0.2758,
+      "step": 2052
+    },
+    {
+      "epoch": 0.822257806244996,
+      "learning_rate": 1.9064389903075683e-05,
+      "loss": 0.2697,
+      "step": 2054
+    },
+    {
+      "epoch": 0.8230584467574059,
+      "learning_rate": 1.905255015197982e-05,
+      "loss": 0.1285,
+      "step": 2056
+    },
+    {
+      "epoch": 0.8238590872698158,
+      "learning_rate": 1.9040639681612216e-05,
+      "loss": 0.2542,
+      "step": 2058
+    },
+    {
+      "epoch": 0.8246597277822257,
+      "learning_rate": 1.902865858501845e-05,
+      "loss": 0.1639,
+      "step": 2060
+    },
+    {
+      "epoch": 0.8254603682946358,
+      "learning_rate": 1.9016606955795843e-05,
+      "loss": 0.0512,
+      "step": 2062
+    },
+    {
+      "epoch": 0.8262610088070457,
+      "learning_rate": 1.9004484888092734e-05,
+      "loss": 0.0898,
+      "step": 2064
+    },
+    {
+      "epoch": 0.8270616493194556,
+      "learning_rate": 1.8992292476607695e-05,
+      "loss": 0.3781,
+      "step": 2066
+    },
+    {
+      "epoch": 0.8278622898318655,
+      "learning_rate": 1.8980029816588863e-05,
+      "loss": 0.2964,
+      "step": 2068
+    },
+    {
+      "epoch": 0.8286629303442754,
+      "learning_rate": 1.8967697003833156e-05,
+      "loss": 0.1906,
+      "step": 2070
+    },
+    {
+      "epoch": 0.8294635708566853,
+      "learning_rate": 1.8955294134685528e-05,
+      "loss": 0.2854,
+      "step": 2072
+    },
+    {
+      "epoch": 0.8302642113690952,
+      "learning_rate": 1.8942821306038227e-05,
+      "loss": 0.1701,
+      "step": 2074
+    },
+    {
+      "epoch": 0.8310648518815053,
+      "learning_rate": 1.893027861533003e-05,
+      "loss": 0.236,
+      "step": 2076
+    },
+    {
+      "epoch": 0.8318654923939152,
+      "learning_rate": 1.891766616054545e-05,
+      "loss": 0.2393,
+      "step": 2078
+    },
+    {
+      "epoch": 0.8326661329063251,
+      "learning_rate": 1.8904984040214043e-05,
+      "loss": 0.0004,
+      "step": 2080
+    },
+    {
+      "epoch": 0.833466773418735,
+      "learning_rate": 1.8892232353409582e-05,
+      "loss": 0.0981,
+      "step": 2082
+    },
+    {
+      "epoch": 0.8342674139311449,
+      "learning_rate": 1.8879411199749306e-05,
+      "loss": 0.1249,
+      "step": 2084
+    },
+    {
+      "epoch": 0.8350680544435548,
+      "learning_rate": 1.8866520679393124e-05,
+      "loss": 0.6806,
+      "step": 2086
+    },
+    {
+      "epoch": 0.8358686949559647,
+      "learning_rate": 1.885356089304285e-05,
+      "loss": 0.0441,
+      "step": 2088
+    },
+    {
+      "epoch": 0.8366693354683747,
+      "learning_rate": 1.884053194194143e-05,
+      "loss": 0.3056,
+      "step": 2090
+    },
+    {
+      "epoch": 0.8374699759807847,
+      "learning_rate": 1.882743392787207e-05,
+      "loss": 0.1716,
+      "step": 2092
+    },
+    {
+      "epoch": 0.8382706164931946,
+      "learning_rate": 1.881426695315756e-05,
+      "loss": 0.3304,
+      "step": 2094
+    },
+    {
+      "epoch": 0.8390712570056045,
+      "learning_rate": 1.8801031120659396e-05,
+      "loss": 0.1705,
+      "step": 2096
+    },
+    {
+      "epoch": 0.8398718975180144,
+      "learning_rate": 1.8787726533777003e-05,
+      "loss": 0.5106,
+      "step": 2098
+    },
+    {
+      "epoch": 0.8406725380304243,
+      "learning_rate": 1.877435329644691e-05,
+      "loss": 0.3935,
+      "step": 2100
+    },
+    {
+      "epoch": 0.8414731785428343,
+      "learning_rate": 1.8760911513141974e-05,
+      "loss": 0.0292,
+      "step": 2102
+    },
+    {
+      "epoch": 0.8422738190552442,
+      "learning_rate": 1.8747401288870482e-05,
+      "loss": 0.0359,
+      "step": 2104
+    },
+    {
+      "epoch": 0.8430744595676541,
+      "learning_rate": 1.8733822729175455e-05,
+      "loss": 0.0288,
+      "step": 2106
+    },
+    {
+      "epoch": 0.8438751000800641,
+      "learning_rate": 1.8720175940133712e-05,
+      "loss": 0.0732,
+      "step": 2108
+    },
+    {
+      "epoch": 0.844675740592474,
+      "learning_rate": 1.8706461028355107e-05,
+      "loss": 0.1904,
+      "step": 2110
+    },
+    {
+      "epoch": 0.8454763811048839,
+      "learning_rate": 1.8692678100981663e-05,
+      "loss": 0.1565,
+      "step": 2112
+    },
+    {
+      "epoch": 0.8462770216172938,
+      "learning_rate": 1.8678827265686753e-05,
+      "loss": 0.576,
+      "step": 2114
+    },
+    {
+      "epoch": 0.8470776621297038,
+      "learning_rate": 1.8664908630674264e-05,
+      "loss": 0.6103,
+      "step": 2116
+    },
+    {
+      "epoch": 0.8478783026421137,
+      "learning_rate": 1.86509223046777e-05,
+      "loss": 0.0349,
+      "step": 2118
+    },
+    {
+      "epoch": 0.8486789431545236,
+      "learning_rate": 1.8636868396959406e-05,
+      "loss": 0.2403,
+      "step": 2120
+    },
+    {
+      "epoch": 0.8494795836669335,
+      "learning_rate": 1.8622747017309676e-05,
+      "loss": 0.0048,
+      "step": 2122
+    },
+    {
+      "epoch": 0.8502802241793435,
+      "learning_rate": 1.8608558276045898e-05,
+      "loss": 0.0155,
+      "step": 2124
+    },
+    {
+      "epoch": 0.8510808646917534,
+      "learning_rate": 1.8594302284011697e-05,
+      "loss": 0.1246,
+      "step": 2126
+    },
+    {
+      "epoch": 0.8518815052041633,
+      "learning_rate": 1.8579979152576076e-05,
+      "loss": 0.1598,
+      "step": 2128
+    },
+    {
+      "epoch": 0.8526821457165733,
+      "learning_rate": 1.8565588993632498e-05,
+      "loss": 0.0322,
+      "step": 2130
+    },
+    {
+      "epoch": 0.8534827862289832,
+      "learning_rate": 1.8551131919598084e-05,
+      "loss": 0.3569,
+      "step": 2132
+    },
+    {
+      "epoch": 0.8542834267413931,
+      "learning_rate": 1.8536608043412702e-05,
+      "loss": 0.1195,
+      "step": 2134
+    },
+    {
+      "epoch": 0.855084067253803,
+      "learning_rate": 1.852201747853807e-05,
+      "loss": 0.0451,
+      "step": 2136
+    },
+    {
+      "epoch": 0.855884707766213,
+      "learning_rate": 1.8507360338956896e-05,
+      "loss": 0.0938,
+      "step": 2138
+    },
+    {
+      "epoch": 0.8566853482786229,
+      "learning_rate": 1.849263673917196e-05,
+      "loss": 0.3296,
+      "step": 2140
+    },
+    {
+      "epoch": 0.8574859887910328,
+      "learning_rate": 1.847784679420527e-05,
+      "loss": 0.0642,
+      "step": 2142
+    },
+    {
+      "epoch": 0.8582866293034428,
+      "learning_rate": 1.846299061959706e-05,
+      "loss": 0.1166,
+      "step": 2144
+    },
+    {
+      "epoch": 0.8590872698158527,
+      "learning_rate": 1.8448068331405018e-05,
+      "loss": 0.0455,
+      "step": 2146
+    },
+    {
+      "epoch": 0.8598879103282626,
+      "learning_rate": 1.8433080046203293e-05,
+      "loss": 0.0201,
+      "step": 2148
+    },
+    {
+      "epoch": 0.8606885508406725,
+      "learning_rate": 1.841802588108161e-05,
+      "loss": 0.0726,
+      "step": 2150
+    },
+    {
+      "epoch": 0.8614891913530824,
+      "learning_rate": 1.8402905953644356e-05,
+      "loss": 0.0901,
+      "step": 2152
+    },
+    {
+      "epoch": 0.8622898318654924,
+      "learning_rate": 1.838772038200968e-05,
+      "loss": 0.0186,
+      "step": 2154
+    },
+    {
+      "epoch": 0.8630904723779024,
+      "learning_rate": 1.837246928480848e-05,
+      "loss": 0.2786,
+      "step": 2156
+    },
+    {
+      "epoch": 0.8638911128903123,
+      "learning_rate": 1.8357152781183613e-05,
+      "loss": 0.0711,
+      "step": 2158
+    },
+    {
+      "epoch": 0.8646917534027222,
+      "learning_rate": 1.8341770990788874e-05,
+      "loss": 0.021,
+      "step": 2160
+    },
+    {
+      "epoch": 0.8654923939151321,
+      "learning_rate": 1.8326324033788087e-05,
+      "loss": 0.5279,
+      "step": 2162
+    },
+    {
+      "epoch": 0.866293034427542,
+      "learning_rate": 1.831081203085415e-05,
+      "loss": 0.1609,
+      "step": 2164
+    },
+    {
+      "epoch": 0.8670936749399519,
+      "learning_rate": 1.8295235103168128e-05,
+      "loss": 0.1316,
+      "step": 2166
+    },
+    {
+      "epoch": 0.8678943154523618,
+      "learning_rate": 1.8279593372418284e-05,
+      "loss": 0.2905,
+      "step": 2168
+    },
+    {
+      "epoch": 0.8686949559647719,
+      "learning_rate": 1.8263886960799072e-05,
+      "loss": 0.0491,
+      "step": 2170
+    },
+    {
+      "epoch": 0.8694955964771818,
+      "learning_rate": 1.8248115991010303e-05,
+      "loss": 0.3533,
+      "step": 2172
+    },
+    {
+      "epoch": 0.8702962369895917,
+      "learning_rate": 1.8232280586256104e-05,
+      "loss": 0.1822,
+      "step": 2174
+    },
+    {
+      "epoch": 0.8710968775020016,
+      "learning_rate": 1.8216380870243963e-05,
+      "loss": 0.079,
+      "step": 2176
+    },
+    {
+      "epoch": 0.8718975180144115,
+      "learning_rate": 1.820041696718378e-05,
+      "loss": 0.1392,
+      "step": 2178
+    },
+    {
+      "epoch": 0.8726981585268214,
+      "learning_rate": 1.8184389001786912e-05,
+      "loss": 0.2052,
+      "step": 2180
+    },
+    {
+      "epoch": 0.8734987990392313,
+      "learning_rate": 1.8168297099265108e-05,
+      "loss": 0.3134,
+      "step": 2182
+    },
+    {
+      "epoch": 0.8742994395516414,
+      "learning_rate": 1.815214138532966e-05,
+      "loss": 0.0791,
+      "step": 2184
+    },
+    {
+      "epoch": 0.8751000800640513,
+      "learning_rate": 1.8135921986190358e-05,
+      "loss": 0.1473,
+      "step": 2186
+    },
+    {
+      "epoch": 0.8759007205764612,
+      "learning_rate": 1.8119639028554475e-05,
+      "loss": 0.0087,
+      "step": 2188
+    },
+    {
+      "epoch": 0.8767013610888711,
+      "learning_rate": 1.8103292639625835e-05,
+      "loss": 0.1448,
+      "step": 2190
+    },
+    {
+      "epoch": 0.877502001601281,
+      "learning_rate": 1.808688294710378e-05,
+      "loss": 0.0578,
+      "step": 2192
+    },
+    {
+      "epoch": 0.8783026421136909,
+      "learning_rate": 1.807041007918221e-05,
+      "loss": 0.0475,
+      "step": 2194
+    },
+    {
+      "epoch": 0.8791032826261009,
+      "learning_rate": 1.805387416454849e-05,
+      "loss": 0.5049,
+      "step": 2196
+    },
+    {
+      "epoch": 0.8799039231385108,
+      "learning_rate": 1.8037275332382575e-05,
+      "loss": 0.4241,
+      "step": 2198
+    },
+    {
+      "epoch": 0.8807045636509208,
+      "learning_rate": 1.802061371235592e-05,
+      "loss": 0.0,
+      "step": 2200
+    },
+    {
+      "epoch": 0.8815052041633307,
+      "learning_rate": 1.8003889434630476e-05,
+      "loss": 0.427,
+      "step": 2202
+    },
+    {
+      "epoch": 0.8823058446757406,
+      "learning_rate": 1.7987102629857692e-05,
+      "loss": 0.0413,
+      "step": 2204
+    },
+    {
+      "epoch": 0.8831064851881505,
+      "learning_rate": 1.7970253429177494e-05,
+      "loss": 0.3957,
+      "step": 2206
+    },
+    {
+      "epoch": 0.8839071257005604,
+      "learning_rate": 1.7953341964217196e-05,
+      "loss": 0.0366,
+      "step": 2208
+    },
+    {
+      "epoch": 0.8847077662129704,
+      "learning_rate": 1.7936368367090583e-05,
+      "loss": 0.0093,
+      "step": 2210
+    },
+    {
+      "epoch": 0.8855084067253803,
+      "learning_rate": 1.7919332770396798e-05,
+      "loss": 0.0087,
+      "step": 2212
+    },
+    {
+      "epoch": 0.8863090472377902,
+      "learning_rate": 1.7902235307219336e-05,
+      "loss": 0.1281,
+      "step": 2214
+    },
+    {
+      "epoch": 0.8871096877502002,
+      "learning_rate": 1.7885076111125e-05,
+      "loss": 0.0003,
+      "step": 2216
+    },
+    {
+      "epoch": 0.8879103282626101,
+      "learning_rate": 1.7867855316162846e-05,
+      "loss": 0.0592,
+      "step": 2218
+    },
+    {
+      "epoch": 0.88871096877502,
+      "learning_rate": 1.7850573056863173e-05,
+      "loss": 0.0222,
+      "step": 2220
+    },
+    {
+      "epoch": 0.8895116092874299,
+      "learning_rate": 1.783322946823638e-05,
+      "loss": 0.1386,
+      "step": 2222
+    },
+    {
+      "epoch": 0.8903122497998399,
+      "learning_rate": 1.7815824685772042e-05,
+      "loss": 0.2224,
+      "step": 2224
+    },
+    {
+      "epoch": 0.8911128903122498,
+      "learning_rate": 1.779835884543776e-05,
+      "loss": 0.0014,
+      "step": 2226
+    },
+    {
+      "epoch": 0.8919135308246597,
+      "learning_rate": 1.7780832083678122e-05,
+      "loss": 0.0365,
+      "step": 2228
+    },
+    {
+      "epoch": 0.8927141713370697,
+      "learning_rate": 1.776324453741365e-05,
+      "loss": 0.315,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8935148118494796,
+      "learning_rate": 1.774559634403971e-05,
+      "loss": 0.6204,
+      "step": 2232
+    },
+    {
+      "epoch": 0.8943154523618895,
+      "learning_rate": 1.7727887641425465e-05,
+      "loss": 0.0644,
+      "step": 2234
+    },
+    {
+      "epoch": 0.8951160928742994,
+      "learning_rate": 1.7710118567912732e-05,
+      "loss": 0.4255,
+      "step": 2236
+    },
+    {
+      "epoch": 0.8959167333867094,
+      "learning_rate": 1.7692289262315008e-05,
+      "loss": 0.1004,
+      "step": 2238
+    },
+    {
+      "epoch": 0.8967173738991193,
+      "learning_rate": 1.7674399863916298e-05,
+      "loss": 0.1559,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8975180144115292,
+      "learning_rate": 1.765645051247007e-05,
+      "loss": 0.0007,
+      "step": 2242
+    },
+    {
+      "epoch": 0.8983186549239391,
+      "learning_rate": 1.7638441348198144e-05,
+      "loss": 0.0329,
+      "step": 2244
+    },
+    {
+      "epoch": 0.899119295436349,
+      "learning_rate": 1.762037251178961e-05,
+      "loss": 0.0871,
+      "step": 2246
+    },
+    {
+      "epoch": 0.899919935948759,
+      "learning_rate": 1.7602244144399713e-05,
+      "loss": 0.0455,
+      "step": 2248
+    },
+    {
+      "epoch": 0.900720576461169,
+      "learning_rate": 1.7584056387648738e-05,
+      "loss": 0.5892,
+      "step": 2250
+    },
+    {
+      "epoch": 0.9015212169735789,
+      "learning_rate": 1.7565809383620966e-05,
+      "loss": 0.079,
+      "step": 2252
+    },
+    {
+      "epoch": 0.9023218574859888,
+      "learning_rate": 1.7547503274863502e-05,
+      "loss": 0.1274,
+      "step": 2254
+    },
+    {
+      "epoch": 0.9031224979983987,
+      "learning_rate": 1.7529138204385186e-05,
+      "loss": 0.4296,
+      "step": 2256
+    },
+    {
+      "epoch": 0.9039231385108086,
+      "learning_rate": 1.7510714315655467e-05,
+      "loss": 0.7326,
+      "step": 2258
+    },
+    {
+      "epoch": 0.9047237790232185,
+      "learning_rate": 1.7492231752603305e-05,
+      "loss": 0.2269,
+      "step": 2260
+    },
+    {
+      "epoch": 0.9055244195356285,
+      "learning_rate": 1.7473690659616e-05,
+      "loss": 0.5769,
+      "step": 2262
+    },
+    {
+      "epoch": 0.9063250600480385,
+      "learning_rate": 1.7455091181538094e-05,
+      "loss": 0.9878,
+      "step": 2264
+    },
+    {
+      "epoch": 0.9071257005604484,
+      "learning_rate": 1.743643346367027e-05,
+      "loss": 0.0126,
+      "step": 2266
+    },
+    {
+      "epoch": 0.9079263410728583,
+      "learning_rate": 1.741771765176815e-05,
+      "loss": 0.0573,
+      "step": 2268
+    },
+    {
+      "epoch": 0.9087269815852682,
+      "learning_rate": 1.739894389204122e-05,
+      "loss": 0.0123,
+      "step": 2270
+    },
+    {
+      "epoch": 0.9095276220976781,
+      "learning_rate": 1.7380112331151657e-05,
+      "loss": 0.2645,
+      "step": 2272
+    },
+    {
+      "epoch": 0.910328262610088,
+      "learning_rate": 1.7361223116213146e-05,
+      "loss": 0.1383,
+      "step": 2274
+    },
+    {
+      "epoch": 0.911128903122498,
+      "learning_rate": 1.7342276394789825e-05,
+      "loss": 0.1959,
+      "step": 2276
+    },
+    {
+      "epoch": 0.911929543634908,
+      "learning_rate": 1.732327231489503e-05,
+      "loss": 0.2246,
+      "step": 2278
+    },
+    {
+      "epoch": 0.9127301841473179,
+      "learning_rate": 1.7304211024990216e-05,
+      "loss": 0.1226,
+      "step": 2280
+    },
+    {
+      "epoch": 0.9135308246597278,
+      "learning_rate": 1.728509267398376e-05,
+      "loss": 0.117,
+      "step": 2282
+    },
+    {
+      "epoch": 0.9143314651721377,
+      "learning_rate": 1.7265917411229803e-05,
+      "loss": 0.1093,
+      "step": 2284
+    },
+    {
+      "epoch": 0.9151321056845476,
+      "learning_rate": 1.7246685386527105e-05,
+      "loss": 0.1817,
+      "step": 2286
+    },
+    {
+      "epoch": 0.9159327461969575,
+      "learning_rate": 1.7227396750117802e-05,
+      "loss": 0.2221,
+      "step": 2288
+    },
+    {
+      "epoch": 0.9167333867093675,
+      "learning_rate": 1.7208051652686348e-05,
+      "loss": 0.0549,
+      "step": 2290
+    },
+    {
+      "epoch": 0.9175340272217775,
+      "learning_rate": 1.718865024535822e-05,
+      "loss": 0.2853,
+      "step": 2292
+    },
+    {
+      "epoch": 0.9183346677341874,
+      "learning_rate": 1.716919267969884e-05,
+      "loss": 0.308,
+      "step": 2294
+    },
+    {
+      "epoch": 0.9191353082465973,
+      "learning_rate": 1.7149679107712317e-05,
+      "loss": 0.1333,
+      "step": 2296
+    },
+    {
+      "epoch": 0.9199359487590072,
+      "learning_rate": 1.7130109681840298e-05,
+      "loss": 0.2974,
+      "step": 2298
+    },
+    {
+      "epoch": 0.9207365892714171,
+      "learning_rate": 1.711048455496075e-05,
+      "loss": 0.022,
+      "step": 2300
+    },
+    {
+      "epoch": 0.921537229783827,
+      "learning_rate": 1.7090803880386784e-05,
+      "loss": 0.0724,
+      "step": 2302
+    },
+    {
+      "epoch": 0.922337870296237,
+      "learning_rate": 1.7071067811865474e-05,
+      "loss": 0.073,
+      "step": 2304
+    },
+    {
+      "epoch": 0.923138510808647,
+      "learning_rate": 1.705127650357663e-05,
+      "loss": 0.1961,
+      "step": 2306
+    },
+    {
+      "epoch": 0.9239391513210569,
+      "learning_rate": 1.7031430110131566e-05,
+      "loss": 0.9507,
+      "step": 2308
+    },
+    {
+      "epoch": 0.9247397918334668,
+      "learning_rate": 1.701152878657197e-05,
+      "loss": 0.2302,
+      "step": 2310
+    },
+    {
+      "epoch": 0.9255404323458767,
+      "learning_rate": 1.699157268836863e-05,
+      "loss": 0.353,
+      "step": 2312
+    },
+    {
+      "epoch": 0.9263410728582866,
+      "learning_rate": 1.697156197142023e-05,
+      "loss": 0.0348,
+      "step": 2314
+    },
+    {
+      "epoch": 0.9271417133706965,
+      "learning_rate": 1.6951496792052148e-05,
+      "loss": 0.3032,
+      "step": 2316
+    },
+    {
+      "epoch": 0.9279423538831065,
+      "learning_rate": 1.6931377307015236e-05,
+      "loss": 0.1601,
+      "step": 2318
+    },
+    {
+      "epoch": 0.9287429943955164,
+      "learning_rate": 1.6911203673484583e-05,
+      "loss": 0.2476,
+      "step": 2320
+    },
+    {
+      "epoch": 0.9295436349079264,
+      "learning_rate": 1.6890976049058267e-05,
+      "loss": 0.288,
+      "step": 2322
+    },
+    {
+      "epoch": 0.9303442754203363,
+      "learning_rate": 1.687069459175619e-05,
+      "loss": 0.1142,
+      "step": 2324
+    },
+    {
+      "epoch": 0.9311449159327462,
+      "learning_rate": 1.6850359460018744e-05,
+      "loss": 0.0748,
+      "step": 2326
+    },
+    {
+      "epoch": 0.9319455564451561,
+      "learning_rate": 1.682997081270568e-05,
+      "loss": 0.1398,
+      "step": 2328
+    },
+    {
+      "epoch": 0.932746196957566,
+      "learning_rate": 1.6809528809094805e-05,
+      "loss": 0.075,
+      "step": 2330
+    },
+    {
+      "epoch": 0.933546837469976,
+      "learning_rate": 1.6789033608880742e-05,
+      "loss": 0.3647,
+      "step": 2332
+    },
+    {
+      "epoch": 0.9343474779823859,
+      "learning_rate": 1.67684853721737e-05,
+      "loss": 0.0252,
+      "step": 2334
+    },
+    {
+      "epoch": 0.9351481184947958,
+      "learning_rate": 1.6747884259498185e-05,
+      "loss": 0.4783,
+      "step": 2336
+    },
+    {
+      "epoch": 0.9359487590072058,
+      "learning_rate": 1.6727230431791826e-05,
+      "loss": 0.2695,
+      "step": 2338
+    },
+    {
+      "epoch": 0.9367493995196157,
+      "learning_rate": 1.6706524050404006e-05,
+      "loss": 0.2075,
+      "step": 2340
+    },
+    {
+      "epoch": 0.9375500400320256,
+      "learning_rate": 1.6685765277094702e-05,
+      "loss": 0.0415,
+      "step": 2342
+    },
+    {
+      "epoch": 0.9383506805444356,
+      "learning_rate": 1.6664954274033175e-05,
+      "loss": 0.0542,
+      "step": 2344
+    },
+    {
+      "epoch": 0.9391513210568455,
+      "learning_rate": 1.66440912037967e-05,
+      "loss": 0.0237,
+      "step": 2346
+    },
+    {
+      "epoch": 0.9399519615692554,
+      "learning_rate": 1.662317622936933e-05,
+      "loss": 0.3304,
+      "step": 2348
+    },
+    {
+      "epoch": 0.9407526020816653,
+      "learning_rate": 1.6602209514140562e-05,
+      "loss": 0.5609,
+      "step": 2350
+    },
+    {
+      "epoch": 0.9415532425940752,
+      "learning_rate": 1.6581191221904098e-05,
+      "loss": 0.6125,
+      "step": 2352
+    },
+    {
+      "epoch": 0.9423538831064852,
+      "learning_rate": 1.6560121516856592e-05,
+      "loss": 0.1366,
+      "step": 2354
+    },
+    {
+      "epoch": 0.9431545236188951,
+      "learning_rate": 1.6539000563596328e-05,
+      "loss": 0.003,
+      "step": 2356
+    },
+    {
+      "epoch": 0.9439551641313051,
+      "learning_rate": 1.651782852712194e-05,
+      "loss": 0.0442,
+      "step": 2358
+    },
+    {
+      "epoch": 0.944755804643715,
+      "learning_rate": 1.6496605572831127e-05,
+      "loss": 0.052,
+      "step": 2360
+    },
+    {
+      "epoch": 0.9455564451561249,
+      "learning_rate": 1.6475331866519387e-05,
+      "loss": 0.0048,
+      "step": 2362
+    },
+    {
+      "epoch": 0.9463570856685348,
+      "learning_rate": 1.6454007574378657e-05,
+      "loss": 0.1963,
+      "step": 2364
+    },
+    {
+      "epoch": 0.9471577261809447,
+      "learning_rate": 1.6432632862996062e-05,
+      "loss": 0.2465,
+      "step": 2366
+    },
+    {
+      "epoch": 0.9479583666933546,
+      "learning_rate": 1.6411207899352633e-05,
+      "loss": 0.0489,
+      "step": 2368
+    },
+    {
+      "epoch": 0.9487590072057646,
+      "learning_rate": 1.6389732850821964e-05,
+      "loss": 0.0135,
+      "step": 2370
+    },
+    {
+      "epoch": 0.9495596477181746,
+      "learning_rate": 1.6368207885168904e-05,
+      "loss": 0.1357,
+      "step": 2372
+    },
+    {
+      "epoch": 0.9503602882305845,
+      "learning_rate": 1.6346633170548275e-05,
+      "loss": 0.2982,
+      "step": 2374
+    },
+    {
+      "epoch": 0.9511609287429944,
+      "learning_rate": 1.6325008875503563e-05,
+      "loss": 0.0393,
+      "step": 2376
+    },
+    {
+      "epoch": 0.9519615692554043,
+      "learning_rate": 1.6303335168965495e-05,
+      "loss": 0.2672,
+      "step": 2378
+    },
+    {
+      "epoch": 0.9527622097678142,
+      "learning_rate": 1.628161222025089e-05,
+      "loss": 0.0149,
+      "step": 2380
+    },
+    {
+      "epoch": 0.9535628502802241,
+      "learning_rate": 1.625984019906122e-05,
+      "loss": 0.2072,
+      "step": 2382
+    },
+    {
+      "epoch": 0.9543634907926342,
+      "learning_rate": 1.623801927548132e-05,
+      "loss": 0.3846,
+      "step": 2384
+    },
+    {
+      "epoch": 0.9551641313050441,
+      "learning_rate": 1.6216149619978057e-05,
+      "loss": 0.0548,
+      "step": 2386
+    },
+    {
+      "epoch": 0.955964771817454,
+      "learning_rate": 1.6194231403398987e-05,
+      "loss": 0.0021,
+      "step": 2388
+    },
+    {
+      "epoch": 0.9567654123298639,
+      "learning_rate": 1.6172264796971063e-05,
+      "loss": 0.5501,
+      "step": 2390
+    },
+    {
+      "epoch": 0.9575660528422738,
+      "learning_rate": 1.6150249972299173e-05,
+      "loss": 0.381,
+      "step": 2392
+    },
+    {
+      "epoch": 0.9583666933546837,
+      "learning_rate": 1.612818710136499e-05,
+      "loss": 0.2256,
+      "step": 2394
+    },
+    {
+      "epoch": 0.9591673338670936,
+      "learning_rate": 1.6106076356525484e-05,
+      "loss": 0.1745,
+      "step": 2396
+    },
+    {
+      "epoch": 0.9599679743795037,
+      "learning_rate": 1.6083917910511623e-05,
+      "loss": 0.4644,
+      "step": 2398
+    },
+    {
+      "epoch": 0.9607686148919136,
+      "learning_rate": 1.6061711936427028e-05,
+      "loss": 0.001,
+      "step": 2400
+    },
+    {
+      "epoch": 0.9615692554043235,
+      "learning_rate": 1.60394586077466e-05,
+      "loss": 0.2483,
+      "step": 2402
+    },
+    {
+      "epoch": 0.9623698959167334,
+      "learning_rate": 1.6017158098315224e-05,
+      "loss": 0.0512,
+      "step": 2404
+    },
+    {
+      "epoch": 0.9631705364291433,
+      "learning_rate": 1.5994810582346266e-05,
+      "loss": 0.0671,
+      "step": 2406
+    },
+    {
+      "epoch": 0.9639711769415532,
+      "learning_rate": 1.5972416234420404e-05,
+      "loss": 0.1952,
+      "step": 2408
+    },
+    {
+      "epoch": 0.9647718174539631,
+      "learning_rate": 1.594997522948413e-05,
+      "loss": 0.2874,
+      "step": 2410
+    },
+    {
+      "epoch": 0.9655724579663731,
+      "learning_rate": 1.592748774284844e-05,
+      "loss": 0.102,
+      "step": 2412
+    },
+    {
+      "epoch": 0.966373098478783,
+      "learning_rate": 1.5904953950187448e-05,
+      "loss": 0.1618,
+      "step": 2414
+    },
+    {
+      "epoch": 0.967173738991193,
+      "learning_rate": 1.588237402753703e-05,
+      "loss": 0.3621,
+      "step": 2416
+    },
+    {
+      "epoch": 0.9679743795036029,
+      "learning_rate": 1.5859748151293354e-05,
+      "loss": 0.1427,
+      "step": 2418
+    },
+    {
+      "epoch": 0.9687750200160128,
+      "learning_rate": 1.5837076498211673e-05,
+      "loss": 0.2411,
+      "step": 2420
+    },
+    {
+      "epoch": 0.9695756605284227,
+      "learning_rate": 1.581435924540482e-05,
+      "loss": 0.094,
+      "step": 2422
+    },
+    {
+      "epoch": 0.9703763010408326,
+      "learning_rate": 1.579159657034185e-05,
+      "loss": 0.0049,
+      "step": 2424
+    },
+    {
+      "epoch": 0.9711769415532426,
+      "learning_rate": 1.5768788650846674e-05,
+      "loss": 0.1452,
+      "step": 2426
+    },
+    {
+      "epoch": 0.9719775820656525,
+      "learning_rate": 1.574593566509664e-05,
+      "loss": 0.2861,
+      "step": 2428
+    },
+    {
+      "epoch": 0.9727782225780625,
+      "learning_rate": 1.5723037791621203e-05,
+      "loss": 0.0181,
+      "step": 2430
+    },
+    {
+      "epoch": 0.9735788630904724,
+      "learning_rate": 1.5700095209300386e-05,
+      "loss": 0.0083,
+      "step": 2432
+    },
+    {
+      "epoch": 0.9743795036028823,
+      "learning_rate": 1.5677108097363565e-05,
+      "loss": 0.4448,
+      "step": 2434
+    },
+    {
+      "epoch": 0.9751801441152922,
+      "learning_rate": 1.5654076635387976e-05,
+      "loss": 0.1516,
+      "step": 2436
+    },
+    {
+      "epoch": 0.9759807846277022,
+      "learning_rate": 1.5631001003297302e-05,
+      "loss": 0.1595,
+      "step": 2438
+    },
+    {
+      "epoch": 0.9767814251401121,
+      "learning_rate": 1.560788138136029e-05,
+      "loss": 0.1581,
+      "step": 2440
+    },
+    {
+      "epoch": 0.977582065652522,
+      "learning_rate": 1.5584717950189373e-05,
+      "loss": 0.0335,
+      "step": 2442
+    },
+    {
+      "epoch": 0.978382706164932,
+      "learning_rate": 1.5561510890739137e-05,
+      "loss": 0.1403,
+      "step": 2444
+    },
+    {
+      "epoch": 0.9791833466773419,
+      "learning_rate": 1.5538260384305083e-05,
+      "loss": 0.2236,
+      "step": 2446
+    },
+    {
+      "epoch": 0.9799839871897518,
+      "learning_rate": 1.5514966612522088e-05,
+      "loss": 0.1106,
+      "step": 2448
+    },
+    {
+      "epoch": 0.9807846277021617,
+      "learning_rate": 1.5491629757363033e-05,
+      "loss": 0.2088,
+      "step": 2450
+    },
+    {
+      "epoch": 0.9815852682145717,
+      "learning_rate": 1.546825000113736e-05,
+      "loss": 0.608,
+      "step": 2452
+    },
+    {
+      "epoch": 0.9823859087269816,
+      "learning_rate": 1.544482752648966e-05,
+      "loss": 0.06,
+      "step": 2454
+    },
+    {
+      "epoch": 0.9831865492393915,
+      "learning_rate": 1.5421362516398285e-05,
+      "loss": 0.5576,
+      "step": 2456
+    },
+    {
+      "epoch": 0.9839871897518014,
+      "learning_rate": 1.539785515417377e-05,
+      "loss": 0.1275,
+      "step": 2458
+    },
+    {
+      "epoch": 0.9847878302642114,
+      "learning_rate": 1.5374305623457605e-05,
+      "loss": 1.3793,
+      "step": 2460
+    },
+    {
+      "epoch": 0.9855884707766213,
+      "learning_rate": 1.5350714108220677e-05,
+      "loss": 0.0663,
+      "step": 2462
+    },
+    {
+      "epoch": 0.9863891112890312,
+      "learning_rate": 1.532708079276186e-05,
+      "loss": 0.1428,
+      "step": 2464
+    },
+    {
+      "epoch": 0.9871897518014412,
+      "learning_rate": 1.5303405861706567e-05,
+      "loss": 0.9145,
+      "step": 2466
+    },
+    {
+      "epoch": 0.9879903923138511,
+      "learning_rate": 1.5279689500005353e-05,
+      "loss": 0.0227,
+      "step": 2468
+    },
+    {
+      "epoch": 0.988791032826261,
+      "learning_rate": 1.5255931892932344e-05,
+      "loss": 0.0544,
+      "step": 2470
+    },
+    {
+      "epoch": 0.9895916733386709,
+      "learning_rate": 1.5232133226083962e-05,
+      "loss": 0.4662,
+      "step": 2472
+    },
+    {
+      "epoch": 0.9903923138510808,
+      "learning_rate": 1.5208293685377362e-05,
+      "loss": 0.064,
+      "step": 2474
+    },
+    {
+      "epoch": 0.9911929543634908,
+      "learning_rate": 1.5184413457049014e-05,
+      "loss": 0.0645,
+      "step": 2476
+    },
+    {
+      "epoch": 0.9919935948759008,
+      "learning_rate": 1.5160492727653238e-05,
+      "loss": 0.3333,
+      "step": 2478
+    },
+    {
+      "epoch": 0.9927942353883107,
+      "learning_rate": 1.5136531684060753e-05,
+      "loss": 0.3651,
+      "step": 2480
+    },
+    {
+      "epoch": 0.9935948759007206,
+      "learning_rate": 1.5112530513457251e-05,
+      "loss": 0.1784,
+      "step": 2482
+    },
+    {
+      "epoch": 0.9943955164131305,
+      "learning_rate": 1.50884894033418e-05,
+      "loss": 0.0366,
+      "step": 2484
+    },
+    {
+      "epoch": 0.9951961569255404,
+      "learning_rate": 1.5064408541525578e-05,
+      "loss": 0.0181,
+      "step": 2486
+    },
+    {
+      "epoch": 0.9959967974379503,
+      "learning_rate": 1.504028811613027e-05,
+      "loss": 0.1543,
+      "step": 2488
+    },
+    {
+      "epoch": 0.9967974379503602,
+      "learning_rate": 1.5016128315586636e-05,
+      "loss": 0.2302,
+      "step": 2490
+    },
+    {
+      "epoch": 0.9975980784627703,
+      "learning_rate": 1.4991929328633043e-05,
+      "loss": 0.1723,
+      "step": 2492
+    },
+    {
+      "epoch": 0.9983987189751802,
+      "learning_rate": 1.4967691344314012e-05,
+      "loss": 0.1521,
+      "step": 2494
+    },
+    {
+      "epoch": 0.9991993594875901,
+      "learning_rate": 1.4943414551978622e-05,
+      "loss": 0.4803,
+      "step": 2496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4919099141279214e-05,
+      "loss": 0.0601,
+      "step": 2498
+    },
+    {
+      "epoch": 1.0,
+      "step": 2498,
+      "total_flos": 1.5606287483011072e+16,
+      "train_loss": 0.21205857841617706,
+      "train_runtime": 3220.1664,
+      "train_samples_per_second": 6.206,
+      "train_steps_per_second": 0.776
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 2498,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {},
+  "total_flos": 1.5606287483011072e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round0.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..9b04a2edf8c97b507d03b9cf03ef8a34da46e1a8
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2d947126c628438ba9d02f915ddee955ca5a48797cf023adac0b3268e98b96e
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round1.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..e0c83c7f314ea55c2475252648c63d8b67653d54
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ffa64505aa4b1a82142e7c39ca1971c1f08f5d4b4e846e61cb8b56823401e22
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round2.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..0113a1d9af7522a739aa74aacd456565774ba0f7
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:753ef7369cb87ba16731ae8518682a71ee44ae2fbc264b3d02ed42cc158fb671
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round3.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..8830b839aacf624ea057f0038a5e76eea13dccbb
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_125_seed1/server_model_round3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:804e679959facc911c5746a9efdbcd738e2d0d047f73f20205748f9664d9ac65
+size 639793378