diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/0_trainer_state.json b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/0_trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..f0bee82b498540bef3897a3b0cee4dd10ad88dc5
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/0_trainer_state.json
@@ -0,0 +1,15020 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 4996,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0008006405124099279,
+      "learning_rate": 2.406842319175051e-06,
+      "loss": 0.1837,
+      "step": 2
+    },
+    {
+      "epoch": 0.0008006405124099279,
+      "learning_rate": 2.415943612351265e-06,
+      "loss": 0.137,
+      "step": 4
+    },
+    {
+      "epoch": 0.0016012810248198558,
+      "learning_rate": 2.4250597173539104e-06,
+      "loss": 0.1388,
+      "step": 6
+    },
+    {
+      "epoch": 0.0016012810248198558,
+      "learning_rate": 2.4341906163790364e-06,
+      "loss": 0.3247,
+      "step": 8
+    },
+    {
+      "epoch": 0.0024019215372297837,
+      "learning_rate": 2.443336291593801e-06,
+      "loss": 0.1085,
+      "step": 10
+    },
+    {
+      "epoch": 0.0024019215372297837,
+      "learning_rate": 2.4524967251364995e-06,
+      "loss": 0.1875,
+      "step": 12
+    },
+    {
+      "epoch": 0.0032025620496397116,
+      "learning_rate": 2.461671899116598e-06,
+      "loss": 0.164,
+      "step": 14
+    },
+    {
+      "epoch": 0.0032025620496397116,
+      "learning_rate": 2.4708617956148052e-06,
+      "loss": 0.079,
+      "step": 16
+    },
+    {
+      "epoch": 0.0040032025620496394,
+      "learning_rate": 2.4800663966830417e-06,
+      "loss": 0.2091,
+      "step": 18
+    },
+    {
+      "epoch": 0.0040032025620496394,
+      "learning_rate": 2.4892856843445236e-06,
+      "loss": 0.1944,
+      "step": 20
+    },
+    {
+      "epoch": 0.004803843074459567,
+      "learning_rate": 2.4985196405937807e-06,
+      "loss": 0.1327,
+      "step": 22
+    },
+    {
+      "epoch": 0.004803843074459567,
+      "learning_rate": 2.507768247396697e-06,
+      "loss": 0.2811,
+      "step": 24
+    },
+    {
+      "epoch": 0.005604483586869495,
+      "learning_rate": 2.5170314866905443e-06,
+      "loss": 0.1624,
+      "step": 26
+    },
+    {
+      "epoch": 0.005604483586869495,
+      "learning_rate": 2.5263093403840022e-06,
+      "loss": 0.1611,
+      "step": 28
+    },
+    {
+      "epoch": 0.006405124099279423,
+      "learning_rate": 2.535601790357246e-06,
+      "loss": 0.214,
+      "step": 30
+    },
+    {
+      "epoch": 0.006405124099279423,
+      "learning_rate": 2.5449088184619065e-06,
+      "loss": 0.0992,
+      "step": 32
+    },
+    {
+      "epoch": 0.007205764611689352,
+      "learning_rate": 2.5542304065211578e-06,
+      "loss": 0.0958,
+      "step": 34
+    },
+    {
+      "epoch": 0.007205764611689352,
+      "learning_rate": 2.5635665363297356e-06,
+      "loss": 0.2021,
+      "step": 36
+    },
+    {
+      "epoch": 0.008006405124099279,
+      "learning_rate": 2.5729171896539763e-06,
+      "loss": 0.3269,
+      "step": 38
+    },
+    {
+      "epoch": 0.008006405124099279,
+      "learning_rate": 2.5822823482318517e-06,
+      "loss": 0.0858,
+      "step": 40
+    },
+    {
+      "epoch": 0.008807045636509208,
+      "learning_rate": 2.5916619937729915e-06,
+      "loss": 0.1152,
+      "step": 42
+    },
+    {
+      "epoch": 0.008807045636509208,
+      "learning_rate": 2.6010561079587694e-06,
+      "loss": 0.078,
+      "step": 44
+    },
+    {
+      "epoch": 0.009607686148919135,
+      "learning_rate": 2.6104646724422643e-06,
+      "loss": 0.2577,
+      "step": 46
+    },
+    {
+      "epoch": 0.009607686148919135,
+      "learning_rate": 2.6198876688483453e-06,
+      "loss": 0.083,
+      "step": 48
+    },
+    {
+      "epoch": 0.010408326661329063,
+      "learning_rate": 2.629325078773699e-06,
+      "loss": 0.6694,
+      "step": 50
+    },
+    {
+      "epoch": 0.010408326661329063,
+      "learning_rate": 2.6387768837868565e-06,
+      "loss": 0.1245,
+      "step": 52
+    },
+    {
+      "epoch": 0.01120896717373899,
+      "learning_rate": 2.648243065428239e-06,
+      "loss": 0.2879,
+      "step": 54
+    },
+    {
+      "epoch": 0.01120896717373899,
+      "learning_rate": 2.6577236052101764e-06,
+      "loss": 0.1754,
+      "step": 56
+    },
+    {
+      "epoch": 0.01200960768614892,
+      "learning_rate": 2.6672184846169934e-06,
+      "loss": 0.2798,
+      "step": 58
+    },
+    {
+      "epoch": 0.01200960768614892,
+      "learning_rate": 2.6767276851049716e-06,
+      "loss": 0.091,
+      "step": 60
+    },
+    {
+      "epoch": 0.012810248198558846,
+      "learning_rate": 2.686251188102439e-06,
+      "loss": 0.1506,
+      "step": 62
+    },
+    {
+      "epoch": 0.012810248198558846,
+      "learning_rate": 2.6957889750097866e-06,
+      "loss": 0.2014,
+      "step": 64
+    },
+    {
+      "epoch": 0.013610888710968775,
+      "learning_rate": 2.7053410271995085e-06,
+      "loss": 0.1241,
+      "step": 66
+    },
+    {
+      "epoch": 0.013610888710968775,
+      "learning_rate": 2.7149073260162416e-06,
+      "loss": 0.219,
+      "step": 68
+    },
+    {
+      "epoch": 0.014411529223378704,
+      "learning_rate": 2.724487852776785e-06,
+      "loss": 0.1256,
+      "step": 70
+    },
+    {
+      "epoch": 0.014411529223378704,
+      "learning_rate": 2.7340825887701848e-06,
+      "loss": 0.3652,
+      "step": 72
+    },
+    {
+      "epoch": 0.01521216973578863,
+      "learning_rate": 2.7436915152577038e-06,
+      "loss": 0.1133,
+      "step": 74
+    },
+    {
+      "epoch": 0.01521216973578863,
+      "learning_rate": 2.7533146134728993e-06,
+      "loss": 0.1085,
+      "step": 76
+    },
+    {
+      "epoch": 0.016012810248198558,
+      "learning_rate": 2.7629518646216522e-06,
+      "loss": 0.5842,
+      "step": 78
+    },
+    {
+      "epoch": 0.016012810248198558,
+      "learning_rate": 2.772603249882202e-06,
+      "loss": 0.6253,
+      "step": 80
+    },
+    {
+      "epoch": 0.016813450760608487,
+      "learning_rate": 2.782268750405185e-06,
+      "loss": 0.1312,
+      "step": 82
+    },
+    {
+      "epoch": 0.016813450760608487,
+      "learning_rate": 2.7919483473136555e-06,
+      "loss": 0.1036,
+      "step": 84
+    },
+    {
+      "epoch": 0.017614091273018415,
+      "learning_rate": 2.801642021703177e-06,
+      "loss": 0.1071,
+      "step": 86
+    },
+    {
+      "epoch": 0.017614091273018415,
+      "learning_rate": 2.81134975464178e-06,
+      "loss": 0.209,
+      "step": 88
+    },
+    {
+      "epoch": 0.018414731785428344,
+      "learning_rate": 2.821071527170053e-06,
+      "loss": 0.1211,
+      "step": 90
+    },
+    {
+      "epoch": 0.018414731785428344,
+      "learning_rate": 2.8308073203011634e-06,
+      "loss": 0.1012,
+      "step": 92
+    },
+    {
+      "epoch": 0.01921537229783827,
+      "learning_rate": 2.8405571150208945e-06,
+      "loss": 0.3513,
+      "step": 94
+    },
+    {
+      "epoch": 0.01921537229783827,
+      "learning_rate": 2.850320892287688e-06,
+      "loss": 0.2698,
+      "step": 96
+    },
+    {
+      "epoch": 0.020016012810248198,
+      "learning_rate": 2.860098633032663e-06,
+      "loss": 0.2184,
+      "step": 98
+    },
+    {
+      "epoch": 0.020016012810248198,
+      "learning_rate": 2.8698903181597026e-06,
+      "loss": 0.1107,
+      "step": 100
+    },
+    {
+      "epoch": 0.020816653322658127,
+      "learning_rate": 2.879695928545424e-06,
+      "loss": 0.1169,
+      "step": 102
+    },
+    {
+      "epoch": 0.020816653322658127,
+      "learning_rate": 2.889515445039256e-06,
+      "loss": 0.1433,
+      "step": 104
+    },
+    {
+      "epoch": 0.021617293835068056,
+      "learning_rate": 2.899348848463471e-06,
+      "loss": 0.0968,
+      "step": 106
+    },
+    {
+      "epoch": 0.021617293835068056,
+      "learning_rate": 2.909196119613218e-06,
+      "loss": 0.1201,
+      "step": 108
+    },
+    {
+      "epoch": 0.02241793434747798,
+      "learning_rate": 2.9190572392565643e-06,
+      "loss": 0.2138,
+      "step": 110
+    },
+    {
+      "epoch": 0.02241793434747798,
+      "learning_rate": 2.928932188134529e-06,
+      "loss": 0.1597,
+      "step": 112
+    },
+    {
+      "epoch": 0.02321857485988791,
+      "learning_rate": 2.9388209469611093e-06,
+      "loss": 0.0958,
+      "step": 114
+    },
+    {
+      "epoch": 0.02321857485988791,
+      "learning_rate": 2.9487234964233724e-06,
+      "loss": 0.1017,
+      "step": 116
+    },
+    {
+      "epoch": 0.02401921537229784,
+      "learning_rate": 2.9586398171814114e-06,
+      "loss": 0.1107,
+      "step": 118
+    },
+    {
+      "epoch": 0.02401921537229784,
+      "learning_rate": 2.9685698898684355e-06,
+      "loss": 0.04,
+      "step": 120
+    },
+    {
+      "epoch": 0.024819855884707767,
+      "learning_rate": 2.9785136950907987e-06,
+      "loss": 0.3738,
+      "step": 122
+    },
+    {
+      "epoch": 0.024819855884707767,
+      "learning_rate": 2.988471213428035e-06,
+      "loss": 0.0265,
+      "step": 124
+    },
+    {
+      "epoch": 0.025620496397117692,
+      "learning_rate": 2.9984424254328936e-06,
+      "loss": 0.1569,
+      "step": 126
+    },
+    {
+      "epoch": 0.025620496397117692,
+      "learning_rate": 3.00842731163137e-06,
+      "loss": 0.5498,
+      "step": 128
+    },
+    {
+      "epoch": 0.02642113690952762,
+      "learning_rate": 3.0184258525227895e-06,
+      "loss": 0.6811,
+      "step": 130
+    },
+    {
+      "epoch": 0.02642113690952762,
+      "learning_rate": 3.0284380285797733e-06,
+      "loss": 0.0504,
+      "step": 132
+    },
+    {
+      "epoch": 0.02722177742193755,
+      "learning_rate": 3.038463820248324e-06,
+      "loss": 0.2216,
+      "step": 134
+    },
+    {
+      "epoch": 0.02722177742193755,
+      "learning_rate": 3.048503207947854e-06,
+      "loss": 0.0719,
+      "step": 136
+    },
+    {
+      "epoch": 0.02802241793434748,
+      "learning_rate": 3.0585561720712207e-06,
+      "loss": 0.1581,
+      "step": 138
+    },
+    {
+      "epoch": 0.02802241793434748,
+      "learning_rate": 3.068622692984767e-06,
+      "loss": 0.0228,
+      "step": 140
+    },
+    {
+      "epoch": 0.028823058446757407,
+      "learning_rate": 3.0787027510283495e-06,
+      "loss": 0.1019,
+      "step": 142
+    },
+    {
+      "epoch": 0.028823058446757407,
+      "learning_rate": 3.0887963265154187e-06,
+      "loss": 0.209,
+      "step": 144
+    },
+    {
+      "epoch": 0.029623698959167333,
+      "learning_rate": 3.098903399732992e-06,
+      "loss": 0.4918,
+      "step": 146
+    },
+    {
+      "epoch": 0.029623698959167333,
+      "learning_rate": 3.1090239509417364e-06,
+      "loss": 0.0516,
+      "step": 148
+    },
+    {
+      "epoch": 0.03042433947157726,
+      "learning_rate": 3.1191579603759946e-06,
+      "loss": 0.1138,
+      "step": 150
+    },
+    {
+      "epoch": 0.03042433947157726,
+      "learning_rate": 3.129305408243829e-06,
+      "loss": 0.0958,
+      "step": 152
+    },
+    {
+      "epoch": 0.03122497998398719,
+      "learning_rate": 3.139466274727052e-06,
+      "loss": 0.1987,
+      "step": 154
+    },
+    {
+      "epoch": 0.03122497998398719,
+      "learning_rate": 3.1496405399812602e-06,
+      "loss": 0.1433,
+      "step": 156
+    },
+    {
+      "epoch": 0.032025620496397116,
+      "learning_rate": 3.159828184135917e-06,
+      "loss": 0.3302,
+      "step": 158
+    },
+    {
+      "epoch": 0.032025620496397116,
+      "learning_rate": 3.17002918729432e-06,
+      "loss": 0.04,
+      "step": 160
+    },
+    {
+      "epoch": 0.03282626100880705,
+      "learning_rate": 3.1802435295336908e-06,
+      "loss": 0.1231,
+      "step": 162
+    },
+    {
+      "epoch": 0.03282626100880705,
+      "learning_rate": 3.1904711909051967e-06,
+      "loss": 0.2238,
+      "step": 164
+    },
+    {
+      "epoch": 0.03362690152121697,
+      "learning_rate": 3.2007121514339924e-06,
+      "loss": 0.1847,
+      "step": 166
+    },
+    {
+      "epoch": 0.03362690152121697,
+      "learning_rate": 3.2109663911192622e-06,
+      "loss": 0.4912,
+      "step": 168
+    },
+    {
+      "epoch": 0.0344275420336269,
+      "learning_rate": 3.221233889934239e-06,
+      "loss": 0.4588,
+      "step": 170
+    },
+    {
+      "epoch": 0.0344275420336269,
+      "learning_rate": 3.231514627826302e-06,
+      "loss": 0.2498,
+      "step": 172
+    },
+    {
+      "epoch": 0.03522818254603683,
+      "learning_rate": 3.2418085847169344e-06,
+      "loss": 0.0887,
+      "step": 174
+    },
+    {
+      "epoch": 0.03522818254603683,
+      "learning_rate": 3.2521157405018146e-06,
+      "loss": 0.0328,
+      "step": 176
+    },
+    {
+      "epoch": 0.036028823058446756,
+      "learning_rate": 3.2624360750508457e-06,
+      "loss": 0.2531,
+      "step": 178
+    },
+    {
+      "epoch": 0.036028823058446756,
+      "learning_rate": 3.2727695682081897e-06,
+      "loss": 0.4032,
+      "step": 180
+    },
+    {
+      "epoch": 0.03682946357085669,
+      "learning_rate": 3.28311619979231e-06,
+      "loss": 0.2236,
+      "step": 182
+    },
+    {
+      "epoch": 0.03682946357085669,
+      "learning_rate": 3.293475949595998e-06,
+      "loss": 0.6358,
+      "step": 184
+    },
+    {
+      "epoch": 0.03763010408326661,
+      "learning_rate": 3.303848797386465e-06,
+      "loss": 0.1072,
+      "step": 186
+    },
+    {
+      "epoch": 0.03763010408326661,
+      "learning_rate": 3.314234722905302e-06,
+      "loss": 0.1338,
+      "step": 188
+    },
+    {
+      "epoch": 0.03843074459567654,
+      "learning_rate": 3.3246337058685697e-06,
+      "loss": 0.3992,
+      "step": 190
+    },
+    {
+      "epoch": 0.03843074459567654,
+      "learning_rate": 3.335045725966829e-06,
+      "loss": 0.0423,
+      "step": 192
+    },
+    {
+      "epoch": 0.03923138510808647,
+      "learning_rate": 3.3454707628651806e-06,
+      "loss": 0.3688,
+      "step": 194
+    },
+    {
+      "epoch": 0.03923138510808647,
+      "learning_rate": 3.355908796203301e-06,
+      "loss": 0.0216,
+      "step": 196
+    },
+    {
+      "epoch": 0.040032025620496396,
+      "learning_rate": 3.3663598055954716e-06,
+      "loss": 0.1035,
+      "step": 198
+    },
+    {
+      "epoch": 0.040032025620496396,
+      "learning_rate": 3.3768237706306716e-06,
+      "loss": 0.4789,
+      "step": 200
+    },
+    {
+      "epoch": 0.04083266613290633,
+      "learning_rate": 3.3873006708725365e-06,
+      "loss": 0.1658,
+      "step": 202
+    },
+    {
+      "epoch": 0.04083266613290633,
+      "learning_rate": 3.3977904858594534e-06,
+      "loss": 0.4219,
+      "step": 204
+    },
+    {
+      "epoch": 0.041633306645316254,
+      "learning_rate": 3.408293195104586e-06,
+      "loss": 0.1471,
+      "step": 206
+    },
+    {
+      "epoch": 0.041633306645316254,
+      "learning_rate": 3.418808778095917e-06,
+      "loss": 0.3087,
+      "step": 208
+    },
+    {
+      "epoch": 0.04243394715772618,
+      "learning_rate": 3.4293372142962845e-06,
+      "loss": 0.1581,
+      "step": 210
+    },
+    {
+      "epoch": 0.04243394715772618,
+      "learning_rate": 3.4398784831434097e-06,
+      "loss": 0.1016,
+      "step": 212
+    },
+    {
+      "epoch": 0.04323458767013611,
+      "learning_rate": 3.4504325640499936e-06,
+      "loss": 0.2027,
+      "step": 214
+    },
+    {
+      "epoch": 0.04323458767013611,
+      "learning_rate": 3.460999436403676e-06,
+      "loss": 0.105,
+      "step": 216
+    },
+    {
+      "epoch": 0.044035228182546036,
+      "learning_rate": 3.4715790795671232e-06,
+      "loss": 0.144,
+      "step": 218
+    },
+    {
+      "epoch": 0.044035228182546036,
+      "learning_rate": 3.4821714728780654e-06,
+      "loss": 0.4032,
+      "step": 220
+    },
+    {
+      "epoch": 0.04483586869495596,
+      "learning_rate": 3.4927765956493276e-06,
+      "loss": 0.0924,
+      "step": 222
+    },
+    {
+      "epoch": 0.04483586869495596,
+      "learning_rate": 3.5033944271688624e-06,
+      "loss": 0.1438,
+      "step": 224
+    },
+    {
+      "epoch": 0.045636509207365894,
+      "learning_rate": 3.514024946699842e-06,
+      "loss": 0.2157,
+      "step": 226
+    },
+    {
+      "epoch": 0.045636509207365894,
+      "learning_rate": 3.5246681334806177e-06,
+      "loss": 0.2133,
+      "step": 228
+    },
+    {
+      "epoch": 0.04643714971977582,
+      "learning_rate": 3.535323966724814e-06,
+      "loss": 0.0215,
+      "step": 230
+    },
+    {
+      "epoch": 0.04643714971977582,
+      "learning_rate": 3.5459924256213596e-06,
+      "loss": 0.028,
+      "step": 232
+    },
+    {
+      "epoch": 0.04723779023218575,
+      "learning_rate": 3.556673489334522e-06,
+      "loss": 0.1222,
+      "step": 234
+    },
+    {
+      "epoch": 0.04723779023218575,
+      "learning_rate": 3.567367137003953e-06,
+      "loss": 0.2998,
+      "step": 236
+    },
+    {
+      "epoch": 0.04803843074459568,
+      "learning_rate": 3.5780733477447127e-06,
+      "loss": 0.6442,
+      "step": 238
+    },
+    {
+      "epoch": 0.04803843074459568,
+      "learning_rate": 3.588792100647368e-06,
+      "loss": 0.3422,
+      "step": 240
+    },
+    {
+      "epoch": 0.0488390712570056,
+      "learning_rate": 3.5995233747779467e-06,
+      "loss": 0.078,
+      "step": 242
+    },
+    {
+      "epoch": 0.0488390712570056,
+      "learning_rate": 3.6102671491780393e-06,
+      "loss": 0.2073,
+      "step": 244
+    },
+    {
+      "epoch": 0.049639711769415534,
+      "learning_rate": 3.6210234028648216e-06,
+      "loss": 0.0957,
+      "step": 246
+    },
+    {
+      "epoch": 0.049639711769415534,
+      "learning_rate": 3.6317921148310965e-06,
+      "loss": 0.0095,
+      "step": 248
+    },
+    {
+      "epoch": 0.05044035228182546,
+      "learning_rate": 3.6425732640453235e-06,
+      "loss": 0.1492,
+      "step": 250
+    },
+    {
+      "epoch": 0.05044035228182546,
+      "learning_rate": 3.653366829451711e-06,
+      "loss": 0.0762,
+      "step": 252
+    },
+    {
+      "epoch": 0.051240992794235385,
+      "learning_rate": 3.6641727899701795e-06,
+      "loss": 0.209,
+      "step": 254
+    },
+    {
+      "epoch": 0.051240992794235385,
+      "learning_rate": 3.674991124496452e-06,
+      "loss": 0.1668,
+      "step": 256
+    },
+    {
+      "epoch": 0.05204163330664532,
+      "learning_rate": 3.6858218119020884e-06,
+      "loss": 0.078,
+      "step": 258
+    },
+    {
+      "epoch": 0.05204163330664532,
+      "learning_rate": 3.696664831034521e-06,
+      "loss": 0.0654,
+      "step": 260
+    },
+    {
+      "epoch": 0.05284227381905524,
+      "learning_rate": 3.7075201607170997e-06,
+      "loss": 0.0909,
+      "step": 262
+    },
+    {
+      "epoch": 0.05284227381905524,
+      "learning_rate": 3.7183877797491143e-06,
+      "loss": 0.0376,
+      "step": 264
+    },
+    {
+      "epoch": 0.053642914331465175,
+      "learning_rate": 3.729267666905899e-06,
+      "loss": 0.3431,
+      "step": 266
+    },
+    {
+      "epoch": 0.053642914331465175,
+      "learning_rate": 3.740159800938784e-06,
+      "loss": 0.0255,
+      "step": 268
+    },
+    {
+      "epoch": 0.0544435548438751,
+      "learning_rate": 3.751064160575195e-06,
+      "loss": 0.4538,
+      "step": 270
+    },
+    {
+      "epoch": 0.0544435548438751,
+      "learning_rate": 3.7619807245186824e-06,
+      "loss": 0.0347,
+      "step": 272
+    },
+    {
+      "epoch": 0.055244195356285025,
+      "learning_rate": 3.772909471448959e-06,
+      "loss": 0.6067,
+      "step": 274
+    },
+    {
+      "epoch": 0.055244195356285025,
+      "learning_rate": 3.783850380021933e-06,
+      "loss": 0.0212,
+      "step": 276
+    },
+    {
+      "epoch": 0.05604483586869496,
+      "learning_rate": 3.794803428869799e-06,
+      "loss": 0.7026,
+      "step": 278
+    },
+    {
+      "epoch": 0.05604483586869496,
+      "learning_rate": 3.8057685966010025e-06,
+      "loss": 0.3088,
+      "step": 280
+    },
+    {
+      "epoch": 0.05684547638110488,
+      "learning_rate": 3.816745861800334e-06,
+      "loss": 0.158,
+      "step": 282
+    },
+    {
+      "epoch": 0.05684547638110488,
+      "learning_rate": 3.827735203028956e-06,
+      "loss": 0.1093,
+      "step": 284
+    },
+    {
+      "epoch": 0.057646116893514815,
+      "learning_rate": 3.838736598824446e-06,
+      "loss": 0.3071,
+      "step": 286
+    },
+    {
+      "epoch": 0.057646116893514815,
+      "learning_rate": 3.849750027700842e-06,
+      "loss": 0.0633,
+      "step": 288
+    },
+    {
+      "epoch": 0.05844675740592474,
+      "learning_rate": 3.860775468148662e-06,
+      "loss": 0.3126,
+      "step": 290
+    },
+    {
+      "epoch": 0.05844675740592474,
+      "learning_rate": 3.871812898635011e-06,
+      "loss": 0.0558,
+      "step": 292
+    },
+    {
+      "epoch": 0.059247397918334666,
+      "learning_rate": 3.882862297603536e-06,
+      "loss": 0.6192,
+      "step": 294
+    },
+    {
+      "epoch": 0.059247397918334666,
+      "learning_rate": 3.8939236434745184e-06,
+      "loss": 0.1351,
+      "step": 296
+    },
+    {
+      "epoch": 0.0600480384307446,
+      "learning_rate": 3.904996914644913e-06,
+      "loss": 0.1579,
+      "step": 298
+    },
+    {
+      "epoch": 0.0600480384307446,
+      "learning_rate": 3.916082089488379e-06,
+      "loss": 0.0492,
+      "step": 300
+    },
+    {
+      "epoch": 0.06084867894315452,
+      "learning_rate": 3.927179146355317e-06,
+      "loss": 0.5427,
+      "step": 302
+    },
+    {
+      "epoch": 0.06084867894315452,
+      "learning_rate": 3.938288063572962e-06,
+      "loss": 0.0958,
+      "step": 304
+    },
+    {
+      "epoch": 0.06164931945556445,
+      "learning_rate": 3.949408819445345e-06,
+      "loss": 0.2609,
+      "step": 306
+    },
+    {
+      "epoch": 0.06164931945556445,
+      "learning_rate": 3.960541392253387e-06,
+      "loss": 0.27,
+      "step": 308
+    },
+    {
+      "epoch": 0.06244995996797438,
+      "learning_rate": 3.971685760254933e-06,
+      "loss": 0.4258,
+      "step": 310
+    },
+    {
+      "epoch": 0.06244995996797438,
+      "learning_rate": 3.982841901684792e-06,
+      "loss": 0.0762,
+      "step": 312
+    },
+    {
+      "epoch": 0.0632506004803843,
+      "learning_rate": 3.994009794754777e-06,
+      "loss": 0.0854,
+      "step": 314
+    },
+    {
+      "epoch": 0.0632506004803843,
+      "learning_rate": 4.005189417653737e-06,
+      "loss": 0.2072,
+      "step": 316
+    },
+    {
+      "epoch": 0.06405124099279423,
+      "learning_rate": 4.016380748547654e-06,
+      "loss": 0.1933,
+      "step": 318
+    },
+    {
+      "epoch": 0.06405124099279423,
+      "learning_rate": 4.027583765579601e-06,
+      "loss": 0.1992,
+      "step": 320
+    },
+    {
+      "epoch": 0.06485188150520416,
+      "learning_rate": 4.038798446869847e-06,
+      "loss": 0.2866,
+      "step": 322
+    },
+    {
+      "epoch": 0.06485188150520416,
+      "learning_rate": 4.050024770515873e-06,
+      "loss": 0.0506,
+      "step": 324
+    },
+    {
+      "epoch": 0.0656525220176141,
+      "learning_rate": 4.061262714592426e-06,
+      "loss": 0.1326,
+      "step": 326
+    },
+    {
+      "epoch": 0.0656525220176141,
+      "learning_rate": 4.072512257151546e-06,
+      "loss": 0.0592,
+      "step": 328
+    },
+    {
+      "epoch": 0.06645316253002402,
+      "learning_rate": 4.0837733762226584e-06,
+      "loss": 0.1291,
+      "step": 330
+    },
+    {
+      "epoch": 0.06645316253002402,
+      "learning_rate": 4.095046049812541e-06,
+      "loss": 0.2255,
+      "step": 332
+    },
+    {
+      "epoch": 0.06725380304243395,
+      "learning_rate": 4.106330255905417e-06,
+      "loss": 0.4503,
+      "step": 334
+    },
+    {
+      "epoch": 0.06725380304243395,
+      "learning_rate": 4.117625972462988e-06,
+      "loss": 0.1545,
+      "step": 336
+    },
+    {
+      "epoch": 0.06805444355484387,
+      "learning_rate": 4.128933177424475e-06,
+      "loss": 0.0849,
+      "step": 338
+    },
+    {
+      "epoch": 0.06805444355484387,
+      "learning_rate": 4.1402518487066624e-06,
+      "loss": 0.005,
+      "step": 340
+    },
+    {
+      "epoch": 0.0688550840672538,
+      "learning_rate": 4.151581964203924e-06,
+      "loss": 0.1231,
+      "step": 342
+    },
+    {
+      "epoch": 0.0688550840672538,
+      "learning_rate": 4.1629235017883285e-06,
+      "loss": 0.0066,
+      "step": 344
+    },
+    {
+      "epoch": 0.06965572457966374,
+      "learning_rate": 4.174276439309593e-06,
+      "loss": 0.2019,
+      "step": 346
+    },
+    {
+      "epoch": 0.06965572457966374,
+      "learning_rate": 4.1856407545951825e-06,
+      "loss": 0.0196,
+      "step": 348
+    },
+    {
+      "epoch": 0.07045636509207366,
+      "learning_rate": 4.197016425450347e-06,
+      "loss": 0.1949,
+      "step": 350
+    },
+    {
+      "epoch": 0.07045636509207366,
+      "learning_rate": 4.208403429658151e-06,
+      "loss": 0.3205,
+      "step": 352
+    },
+    {
+      "epoch": 0.07125700560448359,
+      "learning_rate": 4.219801744979517e-06,
+      "loss": 0.098,
+      "step": 354
+    },
+    {
+      "epoch": 0.07125700560448359,
+      "learning_rate": 4.2312113491533145e-06,
+      "loss": 0.0923,
+      "step": 356
+    },
+    {
+      "epoch": 0.07205764611689351,
+      "learning_rate": 4.242632219896328e-06,
+      "loss": 0.1154,
+      "step": 358
+    },
+    {
+      "epoch": 0.07205764611689351,
+      "learning_rate": 4.254064334903347e-06,
+      "loss": 0.0383,
+      "step": 360
+    },
+    {
+      "epoch": 0.07285828662930344,
+      "learning_rate": 4.2655076718472045e-06,
+      "loss": 0.2207,
+      "step": 362
+    },
+    {
+      "epoch": 0.07285828662930344,
+      "learning_rate": 4.276962208378814e-06,
+      "loss": 0.0514,
+      "step": 364
+    },
+    {
+      "epoch": 0.07365892714171338,
+      "learning_rate": 4.28842792212722e-06,
+      "loss": 0.5628,
+      "step": 366
+    },
+    {
+      "epoch": 0.07365892714171338,
+      "learning_rate": 4.299904790699619e-06,
+      "loss": 0.3137,
+      "step": 368
+    },
+    {
+      "epoch": 0.0744595676541233,
+      "learning_rate": 4.3113927916814665e-06,
+      "loss": 0.3749,
+      "step": 370
+    },
+    {
+      "epoch": 0.0744595676541233,
+      "learning_rate": 4.3228919026364345e-06,
+      "loss": 0.2242,
+      "step": 372
+    },
+    {
+      "epoch": 0.07526020816653323,
+      "learning_rate": 4.33440210110651e-06,
+      "loss": 0.3419,
+      "step": 374
+    },
+    {
+      "epoch": 0.07526020816653323,
+      "learning_rate": 4.345923364612024e-06,
+      "loss": 0.0688,
+      "step": 376
+    },
+    {
+      "epoch": 0.07606084867894315,
+      "learning_rate": 4.3574556706517035e-06,
+      "loss": 0.0983,
+      "step": 378
+    },
+    {
+      "epoch": 0.07606084867894315,
+      "learning_rate": 4.368998996702686e-06,
+      "loss": 0.0157,
+      "step": 380
+    },
+    {
+      "epoch": 0.07686148919135308,
+      "learning_rate": 4.380553320220638e-06,
+      "loss": 0.0847,
+      "step": 382
+    },
+    {
+      "epoch": 0.07686148919135308,
+      "learning_rate": 4.392118618639698e-06,
+      "loss": 0.4016,
+      "step": 384
+    },
+    {
+      "epoch": 0.07766212970376302,
+      "learning_rate": 4.403694869372589e-06,
+      "loss": 0.138,
+      "step": 386
+    },
+    {
+      "epoch": 0.07766212970376302,
+      "learning_rate": 4.415282049810643e-06,
+      "loss": 0.0347,
+      "step": 388
+    },
+    {
+      "epoch": 0.07846277021617294,
+      "learning_rate": 4.4268801373238454e-06,
+      "loss": 0.0908,
+      "step": 390
+    },
+    {
+      "epoch": 0.07846277021617294,
+      "learning_rate": 4.4384891092608795e-06,
+      "loss": 0.0011,
+      "step": 392
+    },
+    {
+      "epoch": 0.07926341072858287,
+      "learning_rate": 4.450108942949158e-06,
+      "loss": 0.5788,
+      "step": 394
+    },
+    {
+      "epoch": 0.07926341072858287,
+      "learning_rate": 4.461739615694921e-06,
+      "loss": 0.667,
+      "step": 396
+    },
+    {
+      "epoch": 0.08006405124099279,
+      "learning_rate": 4.473381104783201e-06,
+      "loss": 0.167,
+      "step": 398
+    },
+    {
+      "epoch": 0.08006405124099279,
+      "learning_rate": 4.485033387477915e-06,
+      "loss": 0.1626,
+      "step": 400
+    },
+    {
+      "epoch": 0.08086469175340272,
+      "learning_rate": 4.496696441021904e-06,
+      "loss": 1.0366,
+      "step": 402
+    },
+    {
+      "epoch": 0.08086469175340272,
+      "learning_rate": 4.5083702426369715e-06,
+      "loss": 0.3269,
+      "step": 404
+    },
+    {
+      "epoch": 0.08166533226581266,
+      "learning_rate": 4.520054769523929e-06,
+      "loss": 0.3985,
+      "step": 406
+    },
+    {
+      "epoch": 0.08166533226581266,
+      "learning_rate": 4.531749998862628e-06,
+      "loss": 0.1938,
+      "step": 408
+    },
+    {
+      "epoch": 0.08246597277822258,
+      "learning_rate": 4.543455907812063e-06,
+      "loss": 0.3644,
+      "step": 410
+    },
+    {
+      "epoch": 0.08246597277822258,
+      "learning_rate": 4.555172473510324e-06,
+      "loss": 0.0716,
+      "step": 412
+    },
+    {
+      "epoch": 0.08326661329063251,
+      "learning_rate": 4.566899673074706e-06,
+      "loss": 0.1137,
+      "step": 414
+    },
+    {
+      "epoch": 0.08326661329063251,
+      "learning_rate": 4.578637483601732e-06,
+      "loss": 0.4658,
+      "step": 416
+    },
+    {
+      "epoch": 0.08406725380304243,
+      "learning_rate": 4.590385882167206e-06,
+      "loss": 0.3881,
+      "step": 418
+    },
+    {
+      "epoch": 0.08406725380304243,
+      "learning_rate": 4.602144845826234e-06,
+      "loss": 0.5698,
+      "step": 420
+    },
+    {
+      "epoch": 0.08486789431545236,
+      "learning_rate": 4.613914351613337e-06,
+      "loss": 0.1149,
+      "step": 422
+    },
+    {
+      "epoch": 0.08486789431545236,
+      "learning_rate": 4.625694376542399e-06,
+      "loss": 0.149,
+      "step": 424
+    },
+    {
+      "epoch": 0.08566853482786228,
+      "learning_rate": 4.637484897606777e-06,
+      "loss": 0.4396,
+      "step": 426
+    },
+    {
+      "epoch": 0.08566853482786228,
+      "learning_rate": 4.649285891779326e-06,
+      "loss": 0.5493,
+      "step": 428
+    },
+    {
+      "epoch": 0.08646917534027222,
+      "learning_rate": 4.661097336012451e-06,
+      "loss": 0.4238,
+      "step": 430
+    },
+    {
+      "epoch": 0.08646917534027222,
+      "learning_rate": 4.672919207238145e-06,
+      "loss": 0.4509,
+      "step": 432
+    },
+    {
+      "epoch": 0.08726981585268215,
+      "learning_rate": 4.684751482368022e-06,
+      "loss": 0.2188,
+      "step": 434
+    },
+    {
+      "epoch": 0.08726981585268215,
+      "learning_rate": 4.696594138293421e-06,
+      "loss": 0.1277,
+      "step": 436
+    },
+    {
+      "epoch": 0.08807045636509207,
+      "learning_rate": 4.7084471518853656e-06,
+      "loss": 0.4742,
+      "step": 438
+    },
+    {
+      "epoch": 0.08807045636509207,
+      "learning_rate": 4.720310499994664e-06,
+      "loss": 0.3137,
+      "step": 440
+    },
+    {
+      "epoch": 0.088871096877502,
+      "learning_rate": 4.732184159451937e-06,
+      "loss": 0.3476,
+      "step": 442
+    },
+    {
+      "epoch": 0.088871096877502,
+      "learning_rate": 4.744068107067673e-06,
+      "loss": 0.141,
+      "step": 444
+    },
+    {
+      "epoch": 0.08967173738991192,
+      "learning_rate": 4.755962319632249e-06,
+      "loss": 0.3123,
+      "step": 446
+    },
+    {
+      "epoch": 0.08967173738991192,
+      "learning_rate": 4.767866773916041e-06,
+      "loss": 0.0884,
+      "step": 448
+    },
+    {
+      "epoch": 0.09047237790232186,
+      "learning_rate": 4.779781446669376e-06,
+      "loss": 0.2609,
+      "step": 450
+    },
+    {
+      "epoch": 0.09047237790232186,
+      "learning_rate": 4.79170631462264e-06,
+      "loss": 0.2777,
+      "step": 452
+    },
+    {
+      "epoch": 0.09127301841473179,
+      "learning_rate": 4.8036413544863095e-06,
+      "loss": 0.3376,
+      "step": 454
+    },
+    {
+      "epoch": 0.09127301841473179,
+      "learning_rate": 4.81558654295099e-06,
+      "loss": 0.3037,
+      "step": 456
+    },
+    {
+      "epoch": 0.09207365892714171,
+      "learning_rate": 4.827541856687471e-06,
+      "loss": 0.2645,
+      "step": 458
+    },
+    {
+      "epoch": 0.09207365892714171,
+      "learning_rate": 4.839507272346751e-06,
+      "loss": 0.2243,
+      "step": 460
+    },
+    {
+      "epoch": 0.09287429943955164,
+      "learning_rate": 4.8514827665601425e-06,
+      "loss": 0.1133,
+      "step": 462
+    },
+    {
+      "epoch": 0.09287429943955164,
+      "learning_rate": 4.863468315939234e-06,
+      "loss": 0.0677,
+      "step": 464
+    },
+    {
+      "epoch": 0.09367493995196156,
+      "learning_rate": 4.875463897075985e-06,
+      "loss": 0.1072,
+      "step": 466
+    },
+    {
+      "epoch": 0.09367493995196156,
+      "learning_rate": 4.8874694865427676e-06,
+      "loss": 0.1586,
+      "step": 468
+    },
+    {
+      "epoch": 0.0944755804643715,
+      "learning_rate": 4.899485060892404e-06,
+      "loss": 0.2633,
+      "step": 470
+    },
+    {
+      "epoch": 0.0944755804643715,
+      "learning_rate": 4.911510596658202e-06,
+      "loss": 0.0976,
+      "step": 472
+    },
+    {
+      "epoch": 0.09527622097678143,
+      "learning_rate": 4.9235460703540615e-06,
+      "loss": 0.1016,
+      "step": 474
+    },
+    {
+      "epoch": 0.09527622097678143,
+      "learning_rate": 4.935591458474425e-06,
+      "loss": 0.7014,
+      "step": 476
+    },
+    {
+      "epoch": 0.09607686148919135,
+      "learning_rate": 4.947646737494389e-06,
+      "loss": 0.5678,
+      "step": 478
+    },
+    {
+      "epoch": 0.09607686148919135,
+      "learning_rate": 4.959711883869734e-06,
+      "loss": 0.267,
+      "step": 480
+    },
+    {
+      "epoch": 0.09687750200160128,
+      "learning_rate": 4.9717868740369645e-06,
+      "loss": 0.2347,
+      "step": 482
+    },
+    {
+      "epoch": 0.09687750200160128,
+      "learning_rate": 4.9838716844133665e-06,
+      "loss": 0.1177,
+      "step": 484
+    },
+    {
+      "epoch": 0.0976781425140112,
+      "learning_rate": 4.9959662913970254e-06,
+      "loss": 0.1108,
+      "step": 486
+    },
+    {
+      "epoch": 0.0976781425140112,
+      "learning_rate": 5.0080706713669435e-06,
+      "loss": 0.0902,
+      "step": 488
+    },
+    {
+      "epoch": 0.09847878302642114,
+      "learning_rate": 5.02018480068299e-06,
+      "loss": 0.1663,
+      "step": 490
+    },
+    {
+      "epoch": 0.09847878302642114,
+      "learning_rate": 5.032308655686007e-06,
+      "loss": 0.105,
+      "step": 492
+    },
+    {
+      "epoch": 0.09927942353883107,
+      "learning_rate": 5.044442212697842e-06,
+      "loss": 0.2408,
+      "step": 494
+    },
+    {
+      "epoch": 0.09927942353883107,
+      "learning_rate": 5.056585448021398e-06,
+      "loss": 0.0514,
+      "step": 496
+    },
+    {
+      "epoch": 0.100080064051241,
+      "learning_rate": 5.068738337940655e-06,
+      "loss": 0.2519,
+      "step": 498
+    },
+    {
+      "epoch": 0.100080064051241,
+      "learning_rate": 5.080900858720789e-06,
+      "loss": 0.158,
+      "step": 500
+    },
+    {
+      "epoch": 0.10088070456365092,
+      "learning_rate": 5.093072986608116e-06,
+      "loss": 0.3834,
+      "step": 502
+    },
+    {
+      "epoch": 0.10088070456365092,
+      "learning_rate": 5.105254697830208e-06,
+      "loss": 0.0758,
+      "step": 504
+    },
+    {
+      "epoch": 0.10168134507606084,
+      "learning_rate": 5.1174459685959175e-06,
+      "loss": 0.5506,
+      "step": 506
+    },
+    {
+      "epoch": 0.10168134507606084,
+      "learning_rate": 5.129646775095432e-06,
+      "loss": 0.2481,
+      "step": 508
+    },
+    {
+      "epoch": 0.10248198558847077,
+      "learning_rate": 5.141857093500307e-06,
+      "loss": 0.4852,
+      "step": 510
+    },
+    {
+      "epoch": 0.10248198558847077,
+      "learning_rate": 5.154076899963514e-06,
+      "loss": 0.1585,
+      "step": 512
+    },
+    {
+      "epoch": 0.10328262610088071,
+      "learning_rate": 5.166306170619537e-06,
+      "loss": 0.1898,
+      "step": 514
+    },
+    {
+      "epoch": 0.10328262610088071,
+      "learning_rate": 5.178544881584328e-06,
+      "loss": 0.1211,
+      "step": 516
+    },
+    {
+      "epoch": 0.10408326661329063,
+      "learning_rate": 5.190793008955421e-06,
+      "loss": 0.2775,
+      "step": 518
+    },
+    {
+      "epoch": 0.10408326661329063,
+      "learning_rate": 5.203050528811959e-06,
+      "loss": 0.0987,
+      "step": 520
+    },
+    {
+      "epoch": 0.10488390712570056,
+      "learning_rate": 5.215317417214739e-06,
+      "loss": 0.0789,
+      "step": 522
+    },
+    {
+      "epoch": 0.10488390712570056,
+      "learning_rate": 5.227593650206246e-06,
+      "loss": 0.0017,
+      "step": 524
+    },
+    {
+      "epoch": 0.10568454763811048,
+      "learning_rate": 5.239879203810763e-06,
+      "loss": 0.0573,
+      "step": 526
+    },
+    {
+      "epoch": 0.10568454763811048,
+      "learning_rate": 5.2521740540343205e-06,
+      "loss": 0.0253,
+      "step": 528
+    },
+    {
+      "epoch": 0.10648518815052041,
+      "learning_rate": 5.264478176864811e-06,
+      "loss": 0.2421,
+      "step": 530
+    },
+    {
+      "epoch": 0.10648518815052041,
+      "learning_rate": 5.2767915482720164e-06,
+      "loss": 0.117,
+      "step": 532
+    },
+    {
+      "epoch": 0.10728582866293035,
+      "learning_rate": 5.289114144207656e-06,
+      "loss": 0.1657,
+      "step": 534
+    },
+    {
+      "epoch": 0.10728582866293035,
+      "learning_rate": 5.3014459406054295e-06,
+      "loss": 0.3285,
+      "step": 536
+    },
+    {
+      "epoch": 0.10808646917534027,
+      "learning_rate": 5.313786913381061e-06,
+      "loss": 0.1668,
+      "step": 538
+    },
+    {
+      "epoch": 0.10808646917534027,
+      "learning_rate": 5.3261370384323904e-06,
+      "loss": 0.2319,
+      "step": 540
+    },
+    {
+      "epoch": 0.1088871096877502,
+      "learning_rate": 5.338496291639341e-06,
+      "loss": 0.956,
+      "step": 542
+    },
+    {
+      "epoch": 0.1088871096877502,
+      "learning_rate": 5.350864648864026e-06,
+      "loss": 0.2241,
+      "step": 544
+    },
+    {
+      "epoch": 0.10968775020016013,
+      "learning_rate": 5.363242085950773e-06,
+      "loss": 0.1962,
+      "step": 546
+    },
+    {
+      "epoch": 0.10968775020016013,
+      "learning_rate": 5.375628578726181e-06,
+      "loss": 0.5747,
+      "step": 548
+    },
+    {
+      "epoch": 0.11048839071257005,
+      "learning_rate": 5.3880241029991434e-06,
+      "loss": 0.232,
+      "step": 550
+    },
+    {
+      "epoch": 0.11048839071257005,
+      "learning_rate": 5.4004286345609665e-06,
+      "loss": 0.4421,
+      "step": 552
+    },
+    {
+      "epoch": 0.11128903122497999,
+      "learning_rate": 5.412842149185316e-06,
+      "loss": 0.1223,
+      "step": 554
+    },
+    {
+      "epoch": 0.11128903122497999,
+      "learning_rate": 5.425264622628326e-06,
+      "loss": 0.1072,
+      "step": 556
+    },
+    {
+      "epoch": 0.11208967173738991,
+      "learning_rate": 5.437696030628639e-06,
+      "loss": 0.1282,
+      "step": 558
+    },
+    {
+      "epoch": 0.11208967173738991,
+      "learning_rate": 5.450136348907444e-06,
+      "loss": 0.046,
+      "step": 560
+    },
+    {
+      "epoch": 0.11289031224979984,
+      "learning_rate": 5.462585553168532e-06,
+      "loss": 0.1456,
+      "step": 562
+    },
+    {
+      "epoch": 0.11289031224979984,
+      "learning_rate": 5.475043619098321e-06,
+      "loss": 0.2147,
+      "step": 564
+    },
+    {
+      "epoch": 0.11369095276220977,
+      "learning_rate": 5.487510522365969e-06,
+      "loss": 0.2151,
+      "step": 566
+    },
+    {
+      "epoch": 0.11369095276220977,
+      "learning_rate": 5.499986238623329e-06,
+      "loss": 0.2098,
+      "step": 568
+    },
+    {
+      "epoch": 0.11449159327461969,
+      "learning_rate": 5.512470743505057e-06,
+      "loss": 0.1212,
+      "step": 570
+    },
+    {
+      "epoch": 0.11449159327461969,
+      "learning_rate": 5.524964012628644e-06,
+      "loss": 0.1138,
+      "step": 572
+    },
+    {
+      "epoch": 0.11529223378702963,
+      "learning_rate": 5.537466021594464e-06,
+      "loss": 0.3512,
+      "step": 574
+    },
+    {
+      "epoch": 0.11529223378702963,
+      "learning_rate": 5.549976745985809e-06,
+      "loss": 0.0874,
+      "step": 576
+    },
+    {
+      "epoch": 0.11609287429943956,
+      "learning_rate": 5.5624961613689934e-06,
+      "loss": 0.2841,
+      "step": 578
+    },
+    {
+      "epoch": 0.11609287429943956,
+      "learning_rate": 5.57502424329331e-06,
+      "loss": 0.0634,
+      "step": 580
+    },
+    {
+      "epoch": 0.11689351481184948,
+      "learning_rate": 5.5875609672911465e-06,
+      "loss": 0.1715,
+      "step": 582
+    },
+    {
+      "epoch": 0.11689351481184948,
+      "learning_rate": 5.6001063088780085e-06,
+      "loss": 0.684,
+      "step": 584
+    },
+    {
+      "epoch": 0.1176941553242594,
+      "learning_rate": 5.6126602435525725e-06,
+      "loss": 0.2229,
+      "step": 586
+    },
+    {
+      "epoch": 0.1176941553242594,
+      "learning_rate": 5.62522274679673e-06,
+      "loss": 0.7892,
+      "step": 588
+    },
+    {
+      "epoch": 0.11849479583666933,
+      "learning_rate": 5.637793794075625e-06,
+      "loss": 0.1324,
+      "step": 590
+    },
+    {
+      "epoch": 0.11849479583666933,
+      "learning_rate": 5.650373360837763e-06,
+      "loss": 0.2225,
+      "step": 592
+    },
+    {
+      "epoch": 0.11929543634907927,
+      "learning_rate": 5.662961422514961e-06,
+      "loss": 0.6194,
+      "step": 594
+    },
+    {
+      "epoch": 0.11929543634907927,
+      "learning_rate": 5.675557954522462e-06,
+      "loss": 0.0789,
+      "step": 596
+    },
+    {
+      "epoch": 0.1200960768614892,
+      "learning_rate": 5.688162932258965e-06,
+      "loss": 0.1673,
+      "step": 598
+    },
+    {
+      "epoch": 0.1200960768614892,
+      "learning_rate": 5.700776331106674e-06,
+      "loss": 0.3653,
+      "step": 600
+    },
+    {
+      "epoch": 0.12089671737389912,
+      "learning_rate": 5.713398126431353e-06,
+      "loss": 0.3828,
+      "step": 602
+    },
+    {
+      "epoch": 0.12089671737389912,
+      "learning_rate": 5.726028293582342e-06,
+      "loss": 0.6325,
+      "step": 604
+    },
+    {
+      "epoch": 0.12169735788630905,
+      "learning_rate": 5.738666807892684e-06,
+      "loss": 0.2609,
+      "step": 606
+    },
+    {
+      "epoch": 0.12169735788630905,
+      "learning_rate": 5.751313644679071e-06,
+      "loss": 0.0512,
+      "step": 608
+    },
+    {
+      "epoch": 0.12249799839871897,
+      "learning_rate": 5.763968779241957e-06,
+      "loss": 0.1663,
+      "step": 610
+    },
+    {
+      "epoch": 0.12249799839871897,
+      "learning_rate": 5.776632186865589e-06,
+      "loss": 0.4288,
+      "step": 612
+    },
+    {
+      "epoch": 0.1232986389111289,
+      "learning_rate": 5.7893038428180584e-06,
+      "loss": 0.3353,
+      "step": 614
+    },
+    {
+      "epoch": 0.1232986389111289,
+      "learning_rate": 5.8019837223513295e-06,
+      "loss": 0.1313,
+      "step": 616
+    },
+    {
+      "epoch": 0.12409927942353884,
+      "learning_rate": 5.814671800701357e-06,
+      "loss": 0.3143,
+      "step": 618
+    },
+    {
+      "epoch": 0.12409927942353884,
+      "learning_rate": 5.827368053088032e-06,
+      "loss": 0.1862,
+      "step": 620
+    },
+    {
+      "epoch": 0.12489991993594876,
+      "learning_rate": 5.840072454715297e-06,
+      "loss": 0.5415,
+      "step": 622
+    },
+    {
+      "epoch": 0.12489991993594876,
+      "learning_rate": 5.852784980771182e-06,
+      "loss": 0.7485,
+      "step": 624
+    },
+    {
+      "epoch": 0.1257005604483587,
+      "learning_rate": 5.865505606427848e-06,
+      "loss": 0.3478,
+      "step": 626
+    },
+    {
+      "epoch": 0.1257005604483587,
+      "learning_rate": 5.878234306841637e-06,
+      "loss": 0.1675,
+      "step": 628
+    },
+    {
+      "epoch": 0.1265012009607686,
+      "learning_rate": 5.890971057153105e-06,
+      "loss": 0.2724,
+      "step": 630
+    },
+    {
+      "epoch": 0.1265012009607686,
+      "learning_rate": 5.903715832487138e-06,
+      "loss": 0.0508,
+      "step": 632
+    },
+    {
+      "epoch": 0.12730184147317855,
+      "learning_rate": 5.916468607952892e-06,
+      "loss": 0.3178,
+      "step": 634
+    },
+    {
+      "epoch": 0.12730184147317855,
+      "learning_rate": 5.929229358643925e-06,
+      "loss": 0.1944,
+      "step": 636
+    },
+    {
+      "epoch": 0.12810248198558846,
+      "learning_rate": 5.941998059638212e-06,
+      "loss": 0.1959,
+      "step": 638
+    },
+    {
+      "epoch": 0.12810248198558846,
+      "learning_rate": 5.954774685998206e-06,
+      "loss": 0.6868,
+      "step": 640
+    },
+    {
+      "epoch": 0.1289031224979984,
+      "learning_rate": 5.9675592127708585e-06,
+      "loss": 0.1083,
+      "step": 642
+    },
+    {
+      "epoch": 0.1289031224979984,
+      "learning_rate": 5.9803516149877475e-06,
+      "loss": 0.2595,
+      "step": 644
+    },
+    {
+      "epoch": 0.1297037630104083,
+      "learning_rate": 5.993151867665015e-06,
+      "loss": 0.5629,
+      "step": 646
+    },
+    {
+      "epoch": 0.1297037630104083,
+      "learning_rate": 6.005959945803494e-06,
+      "loss": 0.1435,
+      "step": 648
+    },
+    {
+      "epoch": 0.13050440352281825,
+      "learning_rate": 6.01877582438873e-06,
+      "loss": 0.4032,
+      "step": 650
+    },
+    {
+      "epoch": 0.13050440352281825,
+      "learning_rate": 6.03159947839103e-06,
+      "loss": 0.0958,
+      "step": 652
+    },
+    {
+      "epoch": 0.1313050440352282,
+      "learning_rate": 6.0444308827655265e-06,
+      "loss": 2.0009,
+      "step": 654
+    },
+    {
+      "epoch": 0.1313050440352282,
+      "learning_rate": 6.057270012452186e-06,
+      "loss": 0.1283,
+      "step": 656
+    },
+    {
+      "epoch": 0.1321056845476381,
+      "learning_rate": 6.070116842375947e-06,
+      "loss": 0.4762,
+      "step": 658
+    },
+    {
+      "epoch": 0.1321056845476381,
+      "learning_rate": 6.082971347446654e-06,
+      "loss": 0.1662,
+      "step": 660
+    },
+    {
+      "epoch": 0.13290632506004804,
+      "learning_rate": 6.095833502559182e-06,
+      "loss": 0.2811,
+      "step": 662
+    },
+    {
+      "epoch": 0.13290632506004804,
+      "learning_rate": 6.108703282593461e-06,
+      "loss": 0.142,
+      "step": 664
+    },
+    {
+      "epoch": 0.13370696557245795,
+      "learning_rate": 6.121580662414533e-06,
+      "loss": 0.2816,
+      "step": 666
+    },
+    {
+      "epoch": 0.13370696557245795,
+      "learning_rate": 6.13446561687258e-06,
+      "loss": 0.1697,
+      "step": 668
+    },
+    {
+      "epoch": 0.1345076060848679,
+      "learning_rate": 6.147358120803041e-06,
+      "loss": 0.2465,
+      "step": 670
+    },
+    {
+      "epoch": 0.1345076060848679,
+      "learning_rate": 6.160258149026557e-06,
+      "loss": 0.2944,
+      "step": 672
+    },
+    {
+      "epoch": 0.13530824659727783,
+      "learning_rate": 6.173165676349095e-06,
+      "loss": 0.1375,
+      "step": 674
+    },
+    {
+      "epoch": 0.13530824659727783,
+      "learning_rate": 6.186080677561974e-06,
+      "loss": 0.1281,
+      "step": 676
+    },
+    {
+      "epoch": 0.13610888710968774,
+      "learning_rate": 6.1990031274419186e-06,
+      "loss": 0.3145,
+      "step": 678
+    },
+    {
+      "epoch": 0.13610888710968774,
+      "learning_rate": 6.2119330007511014e-06,
+      "loss": 0.5827,
+      "step": 680
+    },
+    {
+      "epoch": 0.13690952762209768,
+      "learning_rate": 6.224870272237185e-06,
+      "loss": 0.1274,
+      "step": 682
+    },
+    {
+      "epoch": 0.13690952762209768,
+      "learning_rate": 6.237814916633431e-06,
+      "loss": 0.1176,
+      "step": 684
+    },
+    {
+      "epoch": 0.1377101681345076,
+      "learning_rate": 6.250766908658652e-06,
+      "loss": 0.232,
+      "step": 686
+    },
+    {
+      "epoch": 0.1377101681345076,
+      "learning_rate": 6.263726223017326e-06,
+      "loss": 0.1949,
+      "step": 688
+    },
+    {
+      "epoch": 0.13851080864691753,
+      "learning_rate": 6.2766928343996314e-06,
+      "loss": 0.4071,
+      "step": 690
+    },
+    {
+      "epoch": 0.13851080864691753,
+      "learning_rate": 6.289666717481496e-06,
+      "loss": 0.1494,
+      "step": 692
+    },
+    {
+      "epoch": 0.13931144915932747,
+      "learning_rate": 6.3026478469246285e-06,
+      "loss": 0.237,
+      "step": 694
+    },
+    {
+      "epoch": 0.13931144915932747,
+      "learning_rate": 6.315636197376634e-06,
+      "loss": 0.2303,
+      "step": 696
+    },
+    {
+      "epoch": 0.14011208967173738,
+      "learning_rate": 6.328631743470968e-06,
+      "loss": 0.3638,
+      "step": 698
+    },
+    {
+      "epoch": 0.14011208967173738,
+      "learning_rate": 6.341634459827044e-06,
+      "loss": 0.2009,
+      "step": 700
+    },
+    {
+      "epoch": 0.14091273018414732,
+      "learning_rate": 6.354644321050279e-06,
+      "loss": 0.1447,
+      "step": 702
+    },
+    {
+      "epoch": 0.14091273018414732,
+      "learning_rate": 6.3676613017321305e-06,
+      "loss": 0.2465,
+      "step": 704
+    },
+    {
+      "epoch": 0.14171337069655723,
+      "learning_rate": 6.380685376450153e-06,
+      "loss": 0.1154,
+      "step": 706
+    },
+    {
+      "epoch": 0.14171337069655723,
+      "learning_rate": 6.393716519768032e-06,
+      "loss": 0.2026,
+      "step": 708
+    },
+    {
+      "epoch": 0.14251401120896717,
+      "learning_rate": 6.406754706235692e-06,
+      "loss": 0.2759,
+      "step": 710
+    },
+    {
+      "epoch": 0.14251401120896717,
+      "learning_rate": 6.419799910389257e-06,
+      "loss": 0.2683,
+      "step": 712
+    },
+    {
+      "epoch": 0.1433146517213771,
+      "learning_rate": 6.432852106751162e-06,
+      "loss": 0.0986,
+      "step": 714
+    },
+    {
+      "epoch": 0.1433146517213771,
+      "learning_rate": 6.445911269830183e-06,
+      "loss": 0.3629,
+      "step": 716
+    },
+    {
+      "epoch": 0.14411529223378702,
+      "learning_rate": 6.458977374121492e-06,
+      "loss": 0.4507,
+      "step": 718
+    },
+    {
+      "epoch": 0.14411529223378702,
+      "learning_rate": 6.472050394106689e-06,
+      "loss": 0.1346,
+      "step": 720
+    },
+    {
+      "epoch": 0.14491593274619696,
+      "learning_rate": 6.485130304253915e-06,
+      "loss": 0.1281,
+      "step": 722
+    },
+    {
+      "epoch": 0.14491593274619696,
+      "learning_rate": 6.498217079017806e-06,
+      "loss": 0.3212,
+      "step": 724
+    },
+    {
+      "epoch": 0.14571657325860687,
+      "learning_rate": 6.511310692839605e-06,
+      "loss": 0.4745,
+      "step": 726
+    },
+    {
+      "epoch": 0.14571657325860687,
+      "learning_rate": 6.524411120147204e-06,
+      "loss": 0.232,
+      "step": 728
+    },
+    {
+      "epoch": 0.1465172137710168,
+      "learning_rate": 6.537518335355182e-06,
+      "loss": 0.3634,
+      "step": 730
+    },
+    {
+      "epoch": 0.1465172137710168,
+      "learning_rate": 6.5506323128648654e-06,
+      "loss": 0.1944,
+      "step": 732
+    },
+    {
+      "epoch": 0.14731785428342675,
+      "learning_rate": 6.563753027064355e-06,
+      "loss": 0.2973,
+      "step": 734
+    },
+    {
+      "epoch": 0.14731785428342675,
+      "learning_rate": 6.576880452328645e-06,
+      "loss": 0.2412,
+      "step": 736
+    },
+    {
+      "epoch": 0.14811849479583666,
+      "learning_rate": 6.590014563019571e-06,
+      "loss": 0.3632,
+      "step": 738
+    },
+    {
+      "epoch": 0.14811849479583666,
+      "learning_rate": 6.603155333485934e-06,
+      "loss": 0.4961,
+      "step": 740
+    },
+    {
+      "epoch": 0.1489191353082466,
+      "learning_rate": 6.61630273806352e-06,
+      "loss": 0.2074,
+      "step": 742
+    },
+    {
+      "epoch": 0.1489191353082466,
+      "learning_rate": 6.6294567510751675e-06,
+      "loss": 0.2529,
+      "step": 744
+    },
+    {
+      "epoch": 0.14971977582065651,
+      "learning_rate": 6.642617346830784e-06,
+      "loss": 0.3805,
+      "step": 746
+    },
+    {
+      "epoch": 0.14971977582065651,
+      "learning_rate": 6.655784499627476e-06,
+      "loss": 0.1138,
+      "step": 748
+    },
+    {
+      "epoch": 0.15052041633306645,
+      "learning_rate": 6.6689581837494925e-06,
+      "loss": 0.5477,
+      "step": 750
+    },
+    {
+      "epoch": 0.15052041633306645,
+      "learning_rate": 6.682138373468341e-06,
+      "loss": 0.5494,
+      "step": 752
+    },
+    {
+      "epoch": 0.1513210568454764,
+      "learning_rate": 6.695325043042827e-06,
+      "loss": 0.1011,
+      "step": 754
+    },
+    {
+      "epoch": 0.1513210568454764,
+      "learning_rate": 6.7085181667191e-06,
+      "loss": 0.1012,
+      "step": 756
+    },
+    {
+      "epoch": 0.1521216973578863,
+      "learning_rate": 6.7217177187307e-06,
+      "loss": 0.1505,
+      "step": 758
+    },
+    {
+      "epoch": 0.1521216973578863,
+      "learning_rate": 6.734923673298605e-06,
+      "loss": 0.2461,
+      "step": 760
+    },
+    {
+      "epoch": 0.15292233787029624,
+      "learning_rate": 6.748136004631327e-06,
+      "loss": 0.0736,
+      "step": 762
+    },
+    {
+      "epoch": 0.15292233787029624,
+      "learning_rate": 6.761354686924883e-06,
+      "loss": 0.0537,
+      "step": 764
+    },
+    {
+      "epoch": 0.15372297838270615,
+      "learning_rate": 6.774579694362902e-06,
+      "loss": 0.3171,
+      "step": 766
+    },
+    {
+      "epoch": 0.15372297838270615,
+      "learning_rate": 6.787811001116654e-06,
+      "loss": 0.2944,
+      "step": 768
+    },
+    {
+      "epoch": 0.1545236188951161,
+      "learning_rate": 6.801048581345113e-06,
+      "loss": 0.4731,
+      "step": 770
+    },
+    {
+      "epoch": 0.1545236188951161,
+      "learning_rate": 6.8142924091949955e-06,
+      "loss": 0.215,
+      "step": 772
+    },
+    {
+      "epoch": 0.15532425940752603,
+      "learning_rate": 6.827542458800804e-06,
+      "loss": 0.1031,
+      "step": 774
+    },
+    {
+      "epoch": 0.15532425940752603,
+      "learning_rate": 6.840798704284939e-06,
+      "loss": 0.0111,
+      "step": 776
+    },
+    {
+      "epoch": 0.15612489991993594,
+      "learning_rate": 6.854061119757647e-06,
+      "loss": 0.129,
+      "step": 778
+    },
+    {
+      "epoch": 0.15612489991993594,
+      "learning_rate": 6.867329679317144e-06,
+      "loss": 0.3653,
+      "step": 780
+    },
+    {
+      "epoch": 0.15692554043234588,
+      "learning_rate": 6.880604357049646e-06,
+      "loss": 0.6752,
+      "step": 782
+    },
+    {
+      "epoch": 0.15692554043234588,
+      "learning_rate": 6.893885127029419e-06,
+      "loss": 0.0583,
+      "step": 784
+    },
+    {
+      "epoch": 0.1577261809447558,
+      "learning_rate": 6.907171963318815e-06,
+      "loss": 0.2816,
+      "step": 786
+    },
+    {
+      "epoch": 0.1577261809447558,
+      "learning_rate": 6.920464839968391e-06,
+      "loss": 0.2771,
+      "step": 788
+    },
+    {
+      "epoch": 0.15852682145716573,
+      "learning_rate": 6.9337637310168494e-06,
+      "loss": 0.263,
+      "step": 790
+    },
+    {
+      "epoch": 0.15852682145716573,
+      "learning_rate": 6.94706861049117e-06,
+      "loss": 0.052,
+      "step": 792
+    },
+    {
+      "epoch": 0.15932746196957567,
+      "learning_rate": 6.960379452406636e-06,
+      "loss": 0.1439,
+      "step": 794
+    },
+    {
+      "epoch": 0.15932746196957567,
+      "learning_rate": 6.973696230766884e-06,
+      "loss": 0.0985,
+      "step": 796
+    },
+    {
+      "epoch": 0.16012810248198558,
+      "learning_rate": 6.9870189195639595e-06,
+      "loss": 0.6513,
+      "step": 798
+    },
+    {
+      "epoch": 0.16012810248198558,
+      "learning_rate": 7.000347492778341e-06,
+      "loss": 0.2266,
+      "step": 800
+    },
+    {
+      "epoch": 0.16092874299439552,
+      "learning_rate": 7.013681924379073e-06,
+      "loss": 0.1196,
+      "step": 802
+    },
+    {
+      "epoch": 0.16092874299439552,
+      "learning_rate": 7.027022188323704e-06,
+      "loss": 0.0512,
+      "step": 804
+    },
+    {
+      "epoch": 0.16172938350680544,
+      "learning_rate": 7.040368258558412e-06,
+      "loss": 0.2521,
+      "step": 806
+    },
+    {
+      "epoch": 0.16172938350680544,
+      "learning_rate": 7.05372010901803e-06,
+      "loss": 0.0772,
+      "step": 808
+    },
+    {
+      "epoch": 0.16253002401921537,
+      "learning_rate": 7.0670777136261035e-06,
+      "loss": 0.0958,
+      "step": 810
+    },
+    {
+      "epoch": 0.16253002401921537,
+      "learning_rate": 7.080441046294945e-06,
+      "loss": 0.4974,
+      "step": 812
+    },
+    {
+      "epoch": 0.1633306645316253,
+      "learning_rate": 7.093810080925657e-06,
+      "loss": 0.6324,
+      "step": 814
+    },
+    {
+      "epoch": 0.1633306645316253,
+      "learning_rate": 7.1071847914082605e-06,
+      "loss": 0.1765,
+      "step": 816
+    },
+    {
+      "epoch": 0.16413130504403523,
+      "learning_rate": 7.120565151621638e-06,
+      "loss": 0.2464,
+      "step": 818
+    },
+    {
+      "epoch": 0.16413130504403523,
+      "learning_rate": 7.133951135433656e-06,
+      "loss": 0.0276,
+      "step": 820
+    },
+    {
+      "epoch": 0.16493194555644516,
+      "learning_rate": 7.1473427167012e-06,
+      "loss": 0.166,
+      "step": 822
+    },
+    {
+      "epoch": 0.16493194555644516,
+      "learning_rate": 7.160739869270219e-06,
+      "loss": 0.7568,
+      "step": 824
+    },
+    {
+      "epoch": 0.16573258606885508,
+      "learning_rate": 7.1741425669757854e-06,
+      "loss": 0.5495,
+      "step": 826
+    },
+    {
+      "epoch": 0.16573258606885508,
+      "learning_rate": 7.18755078364214e-06,
+      "loss": 0.2303,
+      "step": 828
+    },
+    {
+      "epoch": 0.16653322658126501,
+      "learning_rate": 7.200964493082727e-06,
+      "loss": 0.1223,
+      "step": 830
+    },
+    {
+      "epoch": 0.16653322658126501,
+      "learning_rate": 7.214383669100317e-06,
+      "loss": 0.1789,
+      "step": 832
+    },
+    {
+      "epoch": 0.16733386709367493,
+      "learning_rate": 7.227808285486952e-06,
+      "loss": 0.1132,
+      "step": 834
+    },
+    {
+      "epoch": 0.16733386709367493,
+      "learning_rate": 7.241238316024064e-06,
+      "loss": 0.2302,
+      "step": 836
+    },
+    {
+      "epoch": 0.16813450760608487,
+      "learning_rate": 7.254673734482513e-06,
+      "loss": 0.1941,
+      "step": 838
+    },
+    {
+      "epoch": 0.16813450760608487,
+      "learning_rate": 7.268114514622635e-06,
+      "loss": 0.1538,
+      "step": 840
+    },
+    {
+      "epoch": 0.1689351481184948,
+      "learning_rate": 7.2815606301942945e-06,
+      "loss": 0.246,
+      "step": 842
+    },
+    {
+      "epoch": 0.1689351481184948,
+      "learning_rate": 7.2950120549369204e-06,
+      "loss": 0.1976,
+      "step": 844
+    },
+    {
+      "epoch": 0.16973578863090472,
+      "learning_rate": 7.308468762579623e-06,
+      "loss": 0.3776,
+      "step": 846
+    },
+    {
+      "epoch": 0.16973578863090472,
+      "learning_rate": 7.321930726841144e-06,
+      "loss": 0.229,
+      "step": 848
+    },
+    {
+      "epoch": 0.17053642914331466,
+      "learning_rate": 7.3353979214299765e-06,
+      "loss": 0.1439,
+      "step": 850
+    },
+    {
+      "epoch": 0.17053642914331466,
+      "learning_rate": 7.348870320044395e-06,
+      "loss": 0.2369,
+      "step": 852
+    },
+    {
+      "epoch": 0.17133706965572457,
+      "learning_rate": 7.362347896372515e-06,
+      "loss": 0.2018,
+      "step": 854
+    },
+    {
+      "epoch": 0.17133706965572457,
+      "learning_rate": 7.375830624092336e-06,
+      "loss": 0.0117,
+      "step": 856
+    },
+    {
+      "epoch": 0.1721377101681345,
+      "learning_rate": 7.389318476871784e-06,
+      "loss": 0.2814,
+      "step": 858
+    },
+    {
+      "epoch": 0.1721377101681345,
+      "learning_rate": 7.402811428368824e-06,
+      "loss": 0.0235,
+      "step": 860
+    },
+    {
+      "epoch": 0.17293835068054444,
+      "learning_rate": 7.416309452231411e-06,
+      "loss": 0.1867,
+      "step": 862
+    },
+    {
+      "epoch": 0.17293835068054444,
+      "learning_rate": 7.429812522097613e-06,
+      "loss": 0.2095,
+      "step": 864
+    },
+    {
+      "epoch": 0.17373899119295436,
+      "learning_rate": 7.443320611595641e-06,
+      "loss": 0.307,
+      "step": 866
+    },
+    {
+      "epoch": 0.17373899119295436,
+      "learning_rate": 7.4568336943439055e-06,
+      "loss": 0.0347,
+      "step": 868
+    },
+    {
+      "epoch": 0.1745396317053643,
+      "learning_rate": 7.470351743951061e-06,
+      "loss": 0.1052,
+      "step": 870
+    },
+    {
+      "epoch": 0.1745396317053643,
+      "learning_rate": 7.4838747340160475e-06,
+      "loss": 0.0678,
+      "step": 872
+    },
+    {
+      "epoch": 0.1753402722177742,
+      "learning_rate": 7.497402638128209e-06,
+      "loss": 0.3855,
+      "step": 874
+    },
+    {
+      "epoch": 0.1753402722177742,
+      "learning_rate": 7.510935429867233e-06,
+      "loss": 0.2218,
+      "step": 876
+    },
+    {
+      "epoch": 0.17614091273018415,
+      "learning_rate": 7.52447308280329e-06,
+      "loss": 0.2914,
+      "step": 878
+    },
+    {
+      "epoch": 0.17614091273018415,
+      "learning_rate": 7.538015570497046e-06,
+      "loss": 0.0428,
+      "step": 880
+    },
+    {
+      "epoch": 0.17694155324259409,
+      "learning_rate": 7.551562866499732e-06,
+      "loss": 0.2521,
+      "step": 882
+    },
+    {
+      "epoch": 0.17694155324259409,
+      "learning_rate": 7.5651149443531846e-06,
+      "loss": 0.1713,
+      "step": 884
+    },
+    {
+      "epoch": 0.177742193755004,
+      "learning_rate": 7.578671777589884e-06,
+      "loss": 0.114,
+      "step": 886
+    },
+    {
+      "epoch": 0.177742193755004,
+      "learning_rate": 7.592233339733077e-06,
+      "loss": 0.0543,
+      "step": 888
+    },
+    {
+      "epoch": 0.17854283426741394,
+      "learning_rate": 7.605799604296721e-06,
+      "loss": 0.2854,
+      "step": 890
+    },
+    {
+      "epoch": 0.17854283426741394,
+      "learning_rate": 7.619370544785608e-06,
+      "loss": 0.4686,
+      "step": 892
+    },
+    {
+      "epoch": 0.17934347477982385,
+      "learning_rate": 7.632946134695396e-06,
+      "loss": 0.6396,
+      "step": 894
+    },
+    {
+      "epoch": 0.17934347477982385,
+      "learning_rate": 7.646526347512665e-06,
+      "loss": 0.0534,
+      "step": 896
+    },
+    {
+      "epoch": 0.1801441152922338,
+      "learning_rate": 7.660111156714964e-06,
+      "loss": 0.6108,
+      "step": 898
+    },
+    {
+      "epoch": 0.1801441152922338,
+      "learning_rate": 7.67370053577085e-06,
+      "loss": 0.2915,
+      "step": 900
+    },
+    {
+      "epoch": 0.18094475580464373,
+      "learning_rate": 7.687294458140006e-06,
+      "loss": 0.281,
+      "step": 902
+    },
+    {
+      "epoch": 0.18094475580464373,
+      "learning_rate": 7.70089289727319e-06,
+      "loss": 0.1215,
+      "step": 904
+    },
+    {
+      "epoch": 0.18174539631705364,
+      "learning_rate": 7.714495826612353e-06,
+      "loss": 0.246,
+      "step": 906
+    },
+    {
+      "epoch": 0.18174539631705364,
+      "learning_rate": 7.728103219590684e-06,
+      "loss": 0.123,
+      "step": 908
+    },
+    {
+      "epoch": 0.18254603682946358,
+      "learning_rate": 7.741715049632646e-06,
+      "loss": 0.1279,
+      "step": 910
+    },
+    {
+      "epoch": 0.18254603682946358,
+      "learning_rate": 7.755331290154041e-06,
+      "loss": 0.0398,
+      "step": 912
+    },
+    {
+      "epoch": 0.1833466773418735,
+      "learning_rate": 7.76895191456204e-06,
+      "loss": 0.2773,
+      "step": 914
+    },
+    {
+      "epoch": 0.1833466773418735,
+      "learning_rate": 7.7825768962553e-06,
+      "loss": 0.0247,
+      "step": 916
+    },
+    {
+      "epoch": 0.18414731785428343,
+      "learning_rate": 7.796206208623925e-06,
+      "loss": 0.3029,
+      "step": 918
+    },
+    {
+      "epoch": 0.18414731785428343,
+      "learning_rate": 7.809839825049565e-06,
+      "loss": 0.016,
+      "step": 920
+    },
+    {
+      "epoch": 0.18494795836669337,
+      "learning_rate": 7.82347771890548e-06,
+      "loss": 0.1394,
+      "step": 922
+    },
+    {
+      "epoch": 0.18494795836669337,
+      "learning_rate": 7.83711986355656e-06,
+      "loss": 0.0625,
+      "step": 924
+    },
+    {
+      "epoch": 0.18574859887910328,
+      "learning_rate": 7.850766232359408e-06,
+      "loss": 0.2673,
+      "step": 926
+    },
+    {
+      "epoch": 0.18574859887910328,
+      "learning_rate": 7.864416798662347e-06,
+      "loss": 0.0296,
+      "step": 928
+    },
+    {
+      "epoch": 0.18654923939151322,
+      "learning_rate": 7.878071535805564e-06,
+      "loss": 0.2151,
+      "step": 930
+    },
+    {
+      "epoch": 0.18654923939151322,
+      "learning_rate": 7.891730417121043e-06,
+      "loss": 0.0957,
+      "step": 932
+    },
+    {
+      "epoch": 0.18734987990392313,
+      "learning_rate": 7.90539341593269e-06,
+      "loss": 0.1187,
+      "step": 934
+    },
+    {
+      "epoch": 0.18734987990392313,
+      "learning_rate": 7.919060505556376e-06,
+      "loss": 0.0193,
+      "step": 936
+    },
+    {
+      "epoch": 0.18815052041633307,
+      "learning_rate": 7.932731659299978e-06,
+      "loss": 0.2859,
+      "step": 938
+    },
+    {
+      "epoch": 0.18815052041633307,
+      "learning_rate": 7.946406850463435e-06,
+      "loss": 0.0515,
+      "step": 940
+    },
+    {
+      "epoch": 0.188951160928743,
+      "learning_rate": 7.960086052338788e-06,
+      "loss": 0.3203,
+      "step": 942
+    },
+    {
+      "epoch": 0.188951160928743,
+      "learning_rate": 7.973769238210291e-06,
+      "loss": 0.028,
+      "step": 944
+    },
+    {
+      "epoch": 0.18975180144115292,
+      "learning_rate": 7.987456381354371e-06,
+      "loss": 0.5219,
+      "step": 946
+    },
+    {
+      "epoch": 0.18975180144115292,
+      "learning_rate": 8.001147455039737e-06,
+      "loss": 0.0784,
+      "step": 948
+    },
+    {
+      "epoch": 0.19055244195356286,
+      "learning_rate": 8.01484243252743e-06,
+      "loss": 0.4221,
+      "step": 950
+    },
+    {
+      "epoch": 0.19055244195356286,
+      "learning_rate": 8.028541287070858e-06,
+      "loss": 0.0357,
+      "step": 952
+    },
+    {
+      "epoch": 0.19135308246597277,
+      "learning_rate": 8.042243991915866e-06,
+      "loss": 0.0908,
+      "step": 954
+    },
+    {
+      "epoch": 0.19135308246597277,
+      "learning_rate": 8.055950520300756e-06,
+      "loss": 0.0167,
+      "step": 956
+    },
+    {
+      "epoch": 0.1921537229783827,
+      "learning_rate": 8.069660845456411e-06,
+      "loss": 0.6893,
+      "step": 958
+    },
+    {
+      "epoch": 0.1921537229783827,
+      "learning_rate": 8.083374940606256e-06,
+      "loss": 0.3342,
+      "step": 960
+    },
+    {
+      "epoch": 0.19295436349079265,
+      "learning_rate": 8.097092778966364e-06,
+      "loss": 0.5962,
+      "step": 962
+    },
+    {
+      "epoch": 0.19295436349079265,
+      "learning_rate": 8.110814333745503e-06,
+      "loss": 0.4965,
+      "step": 964
+    },
+    {
+      "epoch": 0.19375500400320256,
+      "learning_rate": 8.124539578145176e-06,
+      "loss": 0.1091,
+      "step": 966
+    },
+    {
+      "epoch": 0.19375500400320256,
+      "learning_rate": 8.138268485359684e-06,
+      "loss": 0.0374,
+      "step": 968
+    },
+    {
+      "epoch": 0.1945556445156125,
+      "learning_rate": 8.152001028576158e-06,
+      "loss": 0.2603,
+      "step": 970
+    },
+    {
+      "epoch": 0.1945556445156125,
+      "learning_rate": 8.165737180974676e-06,
+      "loss": 0.2151,
+      "step": 972
+    },
+    {
+      "epoch": 0.1953562850280224,
+      "learning_rate": 8.179476915728217e-06,
+      "loss": 0.4335,
+      "step": 974
+    },
+    {
+      "epoch": 0.1953562850280224,
+      "learning_rate": 8.193220206002785e-06,
+      "loss": 0.0152,
+      "step": 976
+    },
+    {
+      "epoch": 0.19615692554043235,
+      "learning_rate": 8.206967024957432e-06,
+      "loss": 0.3212,
+      "step": 978
+    },
+    {
+      "epoch": 0.19615692554043235,
+      "learning_rate": 8.220717345744326e-06,
+      "loss": 0.079,
+      "step": 980
+    },
+    {
+      "epoch": 0.1969575660528423,
+      "learning_rate": 8.234471141508773e-06,
+      "loss": 0.3126,
+      "step": 982
+    },
+    {
+      "epoch": 0.1969575660528423,
+      "learning_rate": 8.248228385389349e-06,
+      "loss": 0.145,
+      "step": 984
+    },
+    {
+      "epoch": 0.1977582065652522,
+      "learning_rate": 8.261989050517841e-06,
+      "loss": 0.6945,
+      "step": 986
+    },
+    {
+      "epoch": 0.1977582065652522,
+      "learning_rate": 8.275753110019367e-06,
+      "loss": 0.2941,
+      "step": 988
+    },
+    {
+      "epoch": 0.19855884707766214,
+      "learning_rate": 8.289520537012428e-06,
+      "loss": 0.2412,
+      "step": 990
+    },
+    {
+      "epoch": 0.19855884707766214,
+      "learning_rate": 8.303291304608936e-06,
+      "loss": 0.0904,
+      "step": 992
+    },
+    {
+      "epoch": 0.19935948759007205,
+      "learning_rate": 8.317065385914285e-06,
+      "loss": 0.5552,
+      "step": 994
+    },
+    {
+      "epoch": 0.19935948759007205,
+      "learning_rate": 8.330842754027378e-06,
+      "loss": 0.4889,
+      "step": 996
+    },
+    {
+      "epoch": 0.200160128102482,
+      "learning_rate": 8.344623382040752e-06,
+      "loss": 0.1383,
+      "step": 998
+    },
+    {
+      "epoch": 0.200160128102482,
+      "learning_rate": 8.358407243040524e-06,
+      "loss": 0.88,
+      "step": 1000
+    },
+    {
+      "epoch": 0.20096076861489193,
+      "learning_rate": 8.372194310106515e-06,
+      "loss": 0.1949,
+      "step": 1002
+    },
+    {
+      "epoch": 0.20096076861489193,
+      "learning_rate": 8.385984556312285e-06,
+      "loss": 0.1013,
+      "step": 1004
+    },
+    {
+      "epoch": 0.20176140912730184,
+      "learning_rate": 8.399777954725183e-06,
+      "loss": 0.158,
+      "step": 1006
+    },
+    {
+      "epoch": 0.20176140912730184,
+      "learning_rate": 8.413574478406386e-06,
+      "loss": 0.1358,
+      "step": 1008
+    },
+    {
+      "epoch": 0.20256204963971178,
+      "learning_rate": 8.427374100411022e-06,
+      "loss": 0.1601,
+      "step": 1010
+    },
+    {
+      "epoch": 0.20256204963971178,
+      "learning_rate": 8.441176793788106e-06,
+      "loss": 0.0809,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2033626901521217,
+      "learning_rate": 8.454982531580687e-06,
+      "loss": 0.1624,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2033626901521217,
+      "learning_rate": 8.468791286825856e-06,
+      "loss": 0.8361,
+      "step": 1016
+    },
+    {
+      "epoch": 0.20416333066453163,
+      "learning_rate": 8.482603032554812e-06,
+      "loss": 0.2148,
+      "step": 1018
+    },
+    {
+      "epoch": 0.20416333066453163,
+      "learning_rate": 8.496417741792922e-06,
+      "loss": 0.5845,
+      "step": 1020
+    },
+    {
+      "epoch": 0.20496397117694154,
+      "learning_rate": 8.510235387559738e-06,
+      "loss": 0.1581,
+      "step": 1022
+    },
+    {
+      "epoch": 0.20496397117694154,
+      "learning_rate": 8.524055942869135e-06,
+      "loss": 0.1794,
+      "step": 1024
+    },
+    {
+      "epoch": 0.20576461168935148,
+      "learning_rate": 8.537879380729254e-06,
+      "loss": 0.2084,
+      "step": 1026
+    },
+    {
+      "epoch": 0.20576461168935148,
+      "learning_rate": 8.551705674142616e-06,
+      "loss": 0.2956,
+      "step": 1028
+    },
+    {
+      "epoch": 0.20656525220176142,
+      "learning_rate": 8.565534796106175e-06,
+      "loss": 0.5345,
+      "step": 1030
+    },
+    {
+      "epoch": 0.20656525220176142,
+      "learning_rate": 8.579366719611353e-06,
+      "loss": 0.2775,
+      "step": 1032
+    },
+    {
+      "epoch": 0.20736589271417133,
+      "learning_rate": 8.593201417644091e-06,
+      "loss": 0.1895,
+      "step": 1034
+    },
+    {
+      "epoch": 0.20736589271417133,
+      "learning_rate": 8.607038863184952e-06,
+      "loss": 0.069,
+      "step": 1036
+    },
+    {
+      "epoch": 0.20816653322658127,
+      "learning_rate": 8.620879029209093e-06,
+      "loss": 0.1286,
+      "step": 1038
+    },
+    {
+      "epoch": 0.20816653322658127,
+      "learning_rate": 8.634721888686368e-06,
+      "loss": 0.4137,
+      "step": 1040
+    },
+    {
+      "epoch": 0.20896717373899118,
+      "learning_rate": 8.648567414581372e-06,
+      "loss": 0.2577,
+      "step": 1042
+    },
+    {
+      "epoch": 0.20896717373899118,
+      "learning_rate": 8.662415579853495e-06,
+      "loss": 0.1521,
+      "step": 1044
+    },
+    {
+      "epoch": 0.20976781425140112,
+      "learning_rate": 8.676266357456968e-06,
+      "loss": 0.2616,
+      "step": 1046
+    },
+    {
+      "epoch": 0.20976781425140112,
+      "learning_rate": 8.690119720340907e-06,
+      "loss": 0.114,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21056845476381106,
+      "learning_rate": 8.703975641449426e-06,
+      "loss": 0.1495,
+      "step": 1050
+    },
+    {
+      "epoch": 0.21056845476381106,
+      "learning_rate": 8.717834093721598e-06,
+      "loss": 0.6946,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21136909527622097,
+      "learning_rate": 8.731695050091561e-06,
+      "loss": 0.1582,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21136909527622097,
+      "learning_rate": 8.74555848348857e-06,
+      "loss": 0.0481,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2121697357886309,
+      "learning_rate": 8.759424366837035e-06,
+      "loss": 0.1945,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2121697357886309,
+      "learning_rate": 8.773292673056572e-06,
+      "loss": 0.0642,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21297037630104082,
+      "learning_rate": 8.787163375062113e-06,
+      "loss": 0.3856,
+      "step": 1062
+    },
+    {
+      "epoch": 0.21297037630104082,
+      "learning_rate": 8.801036445763858e-06,
+      "loss": 0.2244,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21377101681345076,
+      "learning_rate": 8.8149118580674e-06,
+      "loss": 0.0743,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21377101681345076,
+      "learning_rate": 8.828789584873757e-06,
+      "loss": 0.0157,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2145716573258607,
+      "learning_rate": 8.84266959907943e-06,
+      "loss": 0.3323,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2145716573258607,
+      "learning_rate": 8.856551873576448e-06,
+      "loss": 0.0831,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2153722978382706,
+      "learning_rate": 8.870436381252412e-06,
+      "loss": 0.3486,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2153722978382706,
+      "learning_rate": 8.884323094990613e-06,
+      "loss": 1.0135,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21617293835068055,
+      "learning_rate": 8.89821198766998e-06,
+      "loss": 0.2306,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21617293835068055,
+      "learning_rate": 8.912103032165206e-06,
+      "loss": 0.4392,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21697357886309046,
+      "learning_rate": 8.925996201346779e-06,
+      "loss": 0.2244,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21697357886309046,
+      "learning_rate": 8.939891468081036e-06,
+      "loss": 0.4113,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2177742193755004,
+      "learning_rate": 8.953788805230209e-06,
+      "loss": 0.2755,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2177742193755004,
+      "learning_rate": 8.967688185652527e-06,
+      "loss": 0.0494,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21857485988791034,
+      "learning_rate": 8.981589582202184e-06,
+      "loss": 0.1752,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21857485988791034,
+      "learning_rate": 8.995492967729449e-06,
+      "loss": 0.1019,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21937550040032025,
+      "learning_rate": 9.009398315080712e-06,
+      "loss": 0.1083,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21937550040032025,
+      "learning_rate": 9.023305597098526e-06,
+      "loss": 0.2046,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2201761409127302,
+      "learning_rate": 9.037214786621669e-06,
+      "loss": 0.1159,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2201761409127302,
+      "learning_rate": 9.051125856485175e-06,
+      "loss": 0.1266,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2209767814251401,
+      "learning_rate": 9.065038779520457e-06,
+      "loss": 0.381,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2209767814251401,
+      "learning_rate": 9.078953528555258e-06,
+      "loss": 0.4569,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22177742193755004,
+      "learning_rate": 9.092870076413771e-06,
+      "loss": 0.1823,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22177742193755004,
+      "learning_rate": 9.106788395916682e-06,
+      "loss": 0.155,
+      "step": 1108
+    },
+    {
+      "epoch": 0.22257806244995998,
+      "learning_rate": 9.120708459881203e-06,
+      "loss": 0.2775,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22257806244995998,
+      "learning_rate": 9.134630241121135e-06,
+      "loss": 0.235,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2233787029623699,
+      "learning_rate": 9.148553712446971e-06,
+      "loss": 0.0803,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2233787029623699,
+      "learning_rate": 9.162478846665854e-06,
+      "loss": 0.3807,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22417934347477983,
+      "learning_rate": 9.176405616581694e-06,
+      "loss": 0.3145,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22417934347477983,
+      "learning_rate": 9.190333994995208e-06,
+      "loss": 0.4833,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22497998398718974,
+      "learning_rate": 9.20426395470397e-06,
+      "loss": 0.3565,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22497998398718974,
+      "learning_rate": 9.218195468502469e-06,
+      "loss": 0.222,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22578062449959968,
+      "learning_rate": 9.232128509182136e-06,
+      "loss": 0.1821,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22578062449959968,
+      "learning_rate": 9.24606304953148e-06,
+      "loss": 0.2193,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22658126501200962,
+      "learning_rate": 9.259999062336021e-06,
+      "loss": 0.2552,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22658126501200962,
+      "learning_rate": 9.273936520378426e-06,
+      "loss": 0.1217,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22738190552441953,
+      "learning_rate": 9.287875396438536e-06,
+      "loss": 0.223,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22738190552441953,
+      "learning_rate": 9.301815663293426e-06,
+      "loss": 0.6569,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22818254603682947,
+      "learning_rate": 9.315757293717432e-06,
+      "loss": 0.2378,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22818254603682947,
+      "learning_rate": 9.329700260482286e-06,
+      "loss": 0.153,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22898318654923938,
+      "learning_rate": 9.343644536357053e-06,
+      "loss": 0.4707,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22898318654923938,
+      "learning_rate": 9.35759009410826e-06,
+      "loss": 0.1035,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22978382706164932,
+      "learning_rate": 9.37153690649993e-06,
+      "loss": 0.2942,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22978382706164932,
+      "learning_rate": 9.38548494629364e-06,
+      "loss": 0.252,
+      "step": 1148
+    },
+    {
+      "epoch": 0.23058446757405926,
+      "learning_rate": 9.39943418624856e-06,
+      "loss": 0.2149,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23058446757405926,
+      "learning_rate": 9.41338459912151e-06,
+      "loss": 0.2158,
+      "step": 1152
+    },
+    {
+      "epoch": 0.23138510808646917,
+      "learning_rate": 9.427336157667062e-06,
+      "loss": 0.3848,
+      "step": 1154
+    },
+    {
+      "epoch": 0.23138510808646917,
+      "learning_rate": 9.441288834637507e-06,
+      "loss": 0.6956,
+      "step": 1156
+    },
+    {
+      "epoch": 0.2321857485988791,
+      "learning_rate": 9.45524260278296e-06,
+      "loss": 0.182,
+      "step": 1158
+    },
+    {
+      "epoch": 0.2321857485988791,
+      "learning_rate": 9.469197434851414e-06,
+      "loss": 0.1821,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23298638911128902,
+      "learning_rate": 9.483153303588777e-06,
+      "loss": 0.2151,
+      "step": 1162
+    },
+    {
+      "epoch": 0.23298638911128902,
+      "learning_rate": 9.497110181738935e-06,
+      "loss": 0.1823,
+      "step": 1164
+    },
+    {
+      "epoch": 0.23378702962369896,
+      "learning_rate": 9.511068042043785e-06,
+      "loss": 0.2086,
+      "step": 1166
+    },
+    {
+      "epoch": 0.23378702962369896,
+      "learning_rate": 9.52502685724336e-06,
+      "loss": 0.1339,
+      "step": 1168
+    },
+    {
+      "epoch": 0.2345876701361089,
+      "learning_rate": 9.538986600075773e-06,
+      "loss": 0.1976,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2345876701361089,
+      "learning_rate": 9.552947243277342e-06,
+      "loss": 0.2006,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2353883106485188,
+      "learning_rate": 9.566908759582633e-06,
+      "loss": 0.2605,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2353883106485188,
+      "learning_rate": 9.580871121724498e-06,
+      "loss": 0.0686,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23618895116092875,
+      "learning_rate": 9.594834302434123e-06,
+      "loss": 0.5887,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23618895116092875,
+      "learning_rate": 9.608798274441153e-06,
+      "loss": 0.6413,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23698959167333866,
+      "learning_rate": 9.622763010473628e-06,
+      "loss": 0.2964,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23698959167333866,
+      "learning_rate": 9.636728483258116e-06,
+      "loss": 0.0442,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2377902321857486,
+      "learning_rate": 9.650694665519747e-06,
+      "loss": 0.2219,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2377902321857486,
+      "learning_rate": 9.664661529982263e-06,
+      "loss": 0.3407,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23859087269815854,
+      "learning_rate": 9.678629049368077e-06,
+      "loss": 0.1783,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23859087269815854,
+      "learning_rate": 9.692597196398302e-06,
+      "loss": 0.0235,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23939151321056845,
+      "learning_rate": 9.706565943792879e-06,
+      "loss": 0.1508,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23939151321056845,
+      "learning_rate": 9.720535264270526e-06,
+      "loss": 0.0139,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2401921537229784,
+      "learning_rate": 9.734505130548855e-06,
+      "loss": 0.1419,
+      "step": 1198
+    },
+    {
+      "epoch": 0.2401921537229784,
+      "learning_rate": 9.748475515344416e-06,
+      "loss": 0.7775,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2409927942353883,
+      "learning_rate": 9.762446391372746e-06,
+      "loss": 0.2755,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2409927942353883,
+      "learning_rate": 9.776417731348403e-06,
+      "loss": 0.253,
+      "step": 1204
+    },
+    {
+      "epoch": 0.24179343474779824,
+      "learning_rate": 9.790389507985091e-06,
+      "loss": 0.508,
+      "step": 1206
+    },
+    {
+      "epoch": 0.24179343474779824,
+      "learning_rate": 9.80436169399561e-06,
+      "loss": 0.6494,
+      "step": 1208
+    },
+    {
+      "epoch": 0.24259407526020815,
+      "learning_rate": 9.81833426209198e-06,
+      "loss": 0.2596,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24259407526020815,
+      "learning_rate": 9.832307184985473e-06,
+      "loss": 0.235,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2433947157726181,
+      "learning_rate": 9.846280435386668e-06,
+      "loss": 0.0825,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2433947157726181,
+      "learning_rate": 9.8602539860055e-06,
+      "loss": 0.0159,
+      "step": 1216
+    },
+    {
+      "epoch": 0.24419535628502803,
+      "learning_rate": 9.874227809551307e-06,
+      "loss": 0.6583,
+      "step": 1218
+    },
+    {
+      "epoch": 0.24419535628502803,
+      "learning_rate": 9.888201878732946e-06,
+      "loss": 0.0958,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24499599679743794,
+      "learning_rate": 9.902176166258738e-06,
+      "loss": 0.2326,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24499599679743794,
+      "learning_rate": 9.916150644836596e-06,
+      "loss": 0.3552,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24579663730984788,
+      "learning_rate": 9.930125287174061e-06,
+      "loss": 0.5497,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24579663730984788,
+      "learning_rate": 9.944100065978354e-06,
+      "loss": 0.1444,
+      "step": 1228
+    },
+    {
+      "epoch": 0.2465972778222578,
+      "learning_rate": 9.958074953956413e-06,
+      "loss": 0.1661,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2465972778222578,
+      "learning_rate": 9.972049923815011e-06,
+      "loss": 0.5448,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24739791833466773,
+      "learning_rate": 9.986024948260714e-06,
+      "loss": 0.2388,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24739791833466773,
+      "learning_rate": 9.999999999999996e-06,
+      "loss": 0.2446,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24819855884707767,
+      "learning_rate": 1.0013975051739277e-05,
+      "loss": 0.2243,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24819855884707767,
+      "learning_rate": 1.0027950076184982e-05,
+      "loss": 0.3724,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24899919935948758,
+      "learning_rate": 1.004192504604358e-05,
+      "loss": 0.1837,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24899919935948758,
+      "learning_rate": 1.0055899934021637e-05,
+      "loss": 0.3612,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24979983987189752,
+      "learning_rate": 1.006987471282593e-05,
+      "loss": 0.124,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24979983987189752,
+      "learning_rate": 1.0083849355163397e-05,
+      "loss": 0.3383,
+      "step": 1248
+    },
+    {
+      "epoch": 0.25060048038430743,
+      "learning_rate": 1.0097823833741255e-05,
+      "loss": 0.3006,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25060048038430743,
+      "learning_rate": 1.0111798121267047e-05,
+      "loss": 0.3046,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2514011208967174,
+      "learning_rate": 1.0125772190448686e-05,
+      "loss": 0.3415,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2514011208967174,
+      "learning_rate": 1.0139746013994493e-05,
+      "loss": 0.2615,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2522017614091273,
+      "learning_rate": 1.0153719564613327e-05,
+      "loss": 0.0314,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2522017614091273,
+      "learning_rate": 1.016769281501452e-05,
+      "loss": 0.0604,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2530024019215372,
+      "learning_rate": 1.018166573790801e-05,
+      "loss": 0.315,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2530024019215372,
+      "learning_rate": 1.0195638306004383e-05,
+      "loss": 0.0887,
+      "step": 1264
+    },
+    {
+      "epoch": 0.25380304243394713,
+      "learning_rate": 1.0209610492014904e-05,
+      "loss": 0.5316,
+      "step": 1266
+    },
+    {
+      "epoch": 0.25380304243394713,
+      "learning_rate": 1.022358226865159e-05,
+      "loss": 0.3689,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2546036829463571,
+      "learning_rate": 1.0237553608627247e-05,
+      "loss": 0.1677,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2546036829463571,
+      "learning_rate": 1.0251524484655577e-05,
+      "loss": 0.3043,
+      "step": 1272
+    },
+    {
+      "epoch": 0.255404323458767,
+      "learning_rate": 1.0265494869451138e-05,
+      "loss": 0.2319,
+      "step": 1274
+    },
+    {
+      "epoch": 0.255404323458767,
+      "learning_rate": 1.0279464735729467e-05,
+      "loss": 0.2709,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2562049639711769,
+      "learning_rate": 1.0293434056207114e-05,
+      "loss": 0.215,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2562049639711769,
+      "learning_rate": 1.0307402803601691e-05,
+      "loss": 0.1662,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2570056044835869,
+      "learning_rate": 1.0321370950631918e-05,
+      "loss": 0.1603,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2570056044835869,
+      "learning_rate": 1.033533847001773e-05,
+      "loss": 0.3312,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2578062449959968,
+      "learning_rate": 1.0349305334480246e-05,
+      "loss": 0.3263,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2578062449959968,
+      "learning_rate": 1.0363271516741877e-05,
+      "loss": 0.2638,
+      "step": 1288
+    },
+    {
+      "epoch": 0.2586068855084067,
+      "learning_rate": 1.0377236989526366e-05,
+      "loss": 0.1641,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2586068855084067,
+      "learning_rate": 1.039120172555884e-05,
+      "loss": 0.4289,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2594075260208166,
+      "learning_rate": 1.0405165697565868e-05,
+      "loss": 0.0813,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2594075260208166,
+      "learning_rate": 1.0419128878275495e-05,
+      "loss": 0.0068,
+      "step": 1296
+    },
+    {
+      "epoch": 0.2602081665332266,
+      "learning_rate": 1.0433091240417362e-05,
+      "loss": 0.1943,
+      "step": 1298
+    },
+    {
+      "epoch": 0.2602081665332266,
+      "learning_rate": 1.0447052756722651e-05,
+      "loss": 0.1301,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2610088070456365,
+      "learning_rate": 1.046101339992422e-05,
+      "loss": 0.1204,
+      "step": 1302
+    },
+    {
+      "epoch": 0.2610088070456365,
+      "learning_rate": 1.0474973142756632e-05,
+      "loss": 0.3643,
+      "step": 1304
+    },
+    {
+      "epoch": 0.2618094475580464,
+      "learning_rate": 1.0488931957956208e-05,
+      "loss": 0.5458,
+      "step": 1306
+    },
+    {
+      "epoch": 0.2618094475580464,
+      "learning_rate": 1.0502889818261058e-05,
+      "loss": 0.0403,
+      "step": 1308
+    },
+    {
+      "epoch": 0.2626100880704564,
+      "learning_rate": 1.0516846696411216e-05,
+      "loss": 0.2424,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2626100880704564,
+      "learning_rate": 1.053080256514858e-05,
+      "loss": 0.1271,
+      "step": 1312
+    },
+    {
+      "epoch": 0.2634107285828663,
+      "learning_rate": 1.054475739721703e-05,
+      "loss": 0.8296,
+      "step": 1314
+    },
+    {
+      "epoch": 0.2634107285828663,
+      "learning_rate": 1.0558711165362488e-05,
+      "loss": 0.0591,
+      "step": 1316
+    },
+    {
+      "epoch": 0.2642113690952762,
+      "learning_rate": 1.0572663842332931e-05,
+      "loss": 0.142,
+      "step": 1318
+    },
+    {
+      "epoch": 0.2642113690952762,
+      "learning_rate": 1.0586615400878484e-05,
+      "loss": 0.1677,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26501200960768617,
+      "learning_rate": 1.0600565813751433e-05,
+      "loss": 0.2612,
+      "step": 1322
+    },
+    {
+      "epoch": 0.26501200960768617,
+      "learning_rate": 1.0614515053706354e-05,
+      "loss": 0.11,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2658126501200961,
+      "learning_rate": 1.0628463093500063e-05,
+      "loss": 0.4709,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2658126501200961,
+      "learning_rate": 1.0642409905891733e-05,
+      "loss": 0.1879,
+      "step": 1328
+    },
+    {
+      "epoch": 0.266613290632506,
+      "learning_rate": 1.065635546364294e-05,
+      "loss": 0.1199,
+      "step": 1330
+    },
+    {
+      "epoch": 0.266613290632506,
+      "learning_rate": 1.0670299739517706e-05,
+      "loss": 0.109,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2674139311449159,
+      "learning_rate": 1.0684242706282562e-05,
+      "loss": 0.3833,
+      "step": 1334
+    },
+    {
+      "epoch": 0.2674139311449159,
+      "learning_rate": 1.0698184336706567e-05,
+      "loss": 0.2179,
+      "step": 1336
+    },
+    {
+      "epoch": 0.2682145716573259,
+      "learning_rate": 1.0712124603561457e-05,
+      "loss": 0.2701,
+      "step": 1338
+    },
+    {
+      "epoch": 0.2682145716573259,
+      "learning_rate": 1.0726063479621567e-05,
+      "loss": 0.3134,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2690152121697358,
+      "learning_rate": 1.0740000937663972e-05,
+      "loss": 0.3418,
+      "step": 1342
+    },
+    {
+      "epoch": 0.2690152121697358,
+      "learning_rate": 1.0753936950468513e-05,
+      "loss": 0.1581,
+      "step": 1344
+    },
+    {
+      "epoch": 0.2698158526821457,
+      "learning_rate": 1.0767871490817856e-05,
+      "loss": 0.2163,
+      "step": 1346
+    },
+    {
+      "epoch": 0.2698158526821457,
+      "learning_rate": 1.0781804531497525e-05,
+      "loss": 0.138,
+      "step": 1348
+    },
+    {
+      "epoch": 0.27061649319455566,
+      "learning_rate": 1.0795736045296023e-05,
+      "loss": 0.3642,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27061649319455566,
+      "learning_rate": 1.0809666005004787e-05,
+      "loss": 0.0183,
+      "step": 1352
+    },
+    {
+      "epoch": 0.2714171337069656,
+      "learning_rate": 1.08235943834183e-05,
+      "loss": 0.5841,
+      "step": 1354
+    },
+    {
+      "epoch": 0.2714171337069656,
+      "learning_rate": 1.083752115333414e-05,
+      "loss": 0.201,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2722177742193755,
+      "learning_rate": 1.0851446287553022e-05,
+      "loss": 0.4889,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2722177742193755,
+      "learning_rate": 1.0865369758878858e-05,
+      "loss": 0.0572,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27301841473178545,
+      "learning_rate": 1.087929154011879e-05,
+      "loss": 0.0833,
+      "step": 1362
+    },
+    {
+      "epoch": 0.27301841473178545,
+      "learning_rate": 1.0893211604083311e-05,
+      "loss": 0.0278,
+      "step": 1364
+    },
+    {
+      "epoch": 0.27381905524419536,
+      "learning_rate": 1.090712992358622e-05,
+      "loss": 0.1132,
+      "step": 1366
+    },
+    {
+      "epoch": 0.27381905524419536,
+      "learning_rate": 1.0921046471444737e-05,
+      "loss": 0.4376,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2746196957566053,
+      "learning_rate": 1.0934961220479537e-05,
+      "loss": 0.1689,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2746196957566053,
+      "learning_rate": 1.0948874143514818e-05,
+      "loss": 0.8235,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2754203362690152,
+      "learning_rate": 1.0962785213378325e-05,
+      "loss": 0.3316,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2754203362690152,
+      "learning_rate": 1.0976694402901467e-05,
+      "loss": 0.492,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27622097678142515,
+      "learning_rate": 1.0990601684919282e-05,
+      "loss": 0.1865,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27622097678142515,
+      "learning_rate": 1.1004507032270544e-05,
+      "loss": 0.3552,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27702161729383507,
+      "learning_rate": 1.1018410417797809e-05,
+      "loss": 0.7587,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27702161729383507,
+      "learning_rate": 1.1032311814347467e-05,
+      "loss": 0.3838,
+      "step": 1384
+    },
+    {
+      "epoch": 0.277822257806245,
+      "learning_rate": 1.1046211194769784e-05,
+      "loss": 0.3149,
+      "step": 1386
+    },
+    {
+      "epoch": 0.277822257806245,
+      "learning_rate": 1.1060108531918955e-05,
+      "loss": 0.46,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27862289831865494,
+      "learning_rate": 1.1074003798653215e-05,
+      "loss": 0.1758,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27862289831865494,
+      "learning_rate": 1.1087896967834787e-05,
+      "loss": 0.0706,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27942353883106485,
+      "learning_rate": 1.1101788012330013e-05,
+      "loss": 0.1362,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27942353883106485,
+      "learning_rate": 1.111567690500938e-05,
+      "loss": 0.1505,
+      "step": 1396
+    },
+    {
+      "epoch": 0.28022417934347477,
+      "learning_rate": 1.1129563618747581e-05,
+      "loss": 0.1617,
+      "step": 1398
+    },
+    {
+      "epoch": 0.28022417934347477,
+      "learning_rate": 1.1143448126423545e-05,
+      "loss": 0.1319,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28102481985588473,
+      "learning_rate": 1.1157330400920563e-05,
+      "loss": 0.3119,
+      "step": 1402
+    },
+    {
+      "epoch": 0.28102481985588473,
+      "learning_rate": 1.1171210415126238e-05,
+      "loss": 0.1945,
+      "step": 1404
+    },
+    {
+      "epoch": 0.28182546036829464,
+      "learning_rate": 1.1185088141932594e-05,
+      "loss": 0.363,
+      "step": 1406
+    },
+    {
+      "epoch": 0.28182546036829464,
+      "learning_rate": 1.1198963554236135e-05,
+      "loss": 0.1512,
+      "step": 1408
+    },
+    {
+      "epoch": 0.28262610088070456,
+      "learning_rate": 1.121283662493788e-05,
+      "loss": 0.6766,
+      "step": 1410
+    },
+    {
+      "epoch": 0.28262610088070456,
+      "learning_rate": 1.122670732694342e-05,
+      "loss": 0.3273,
+      "step": 1412
+    },
+    {
+      "epoch": 0.28342674139311447,
+      "learning_rate": 1.1240575633162958e-05,
+      "loss": 0.791,
+      "step": 1414
+    },
+    {
+      "epoch": 0.28342674139311447,
+      "learning_rate": 1.1254441516511425e-05,
+      "loss": 0.0908,
+      "step": 1416
+    },
+    {
+      "epoch": 0.28422738190552443,
+      "learning_rate": 1.1268304949908434e-05,
+      "loss": 0.4785,
+      "step": 1418
+    },
+    {
+      "epoch": 0.28422738190552443,
+      "learning_rate": 1.1282165906278395e-05,
+      "loss": 0.1865,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28502802241793435,
+      "learning_rate": 1.1296024358550565e-05,
+      "loss": 0.5856,
+      "step": 1422
+    },
+    {
+      "epoch": 0.28502802241793435,
+      "learning_rate": 1.1309880279659087e-05,
+      "loss": 0.3426,
+      "step": 1424
+    },
+    {
+      "epoch": 0.28582866293034426,
+      "learning_rate": 1.1323733642543024e-05,
+      "loss": 0.2391,
+      "step": 1426
+    },
+    {
+      "epoch": 0.28582866293034426,
+      "learning_rate": 1.1337584420146496e-05,
+      "loss": 0.1511,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2866293034427542,
+      "learning_rate": 1.135143258541862e-05,
+      "loss": 0.2774,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2866293034427542,
+      "learning_rate": 1.1365278111313625e-05,
+      "loss": 0.3206,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28742994395516414,
+      "learning_rate": 1.13791209707909e-05,
+      "loss": 0.3635,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28742994395516414,
+      "learning_rate": 1.1392961136815041e-05,
+      "loss": 0.2231,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28823058446757405,
+      "learning_rate": 1.1406798582355902e-05,
+      "loss": 0.1714,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28823058446757405,
+      "learning_rate": 1.142063328038864e-05,
+      "loss": 0.134,
+      "step": 1440
+    },
+    {
+      "epoch": 0.289031224979984,
+      "learning_rate": 1.1434465203893818e-05,
+      "loss": 0.2081,
+      "step": 1442
+    },
+    {
+      "epoch": 0.289031224979984,
+      "learning_rate": 1.1448294325857377e-05,
+      "loss": 0.2605,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2898318654923939,
+      "learning_rate": 1.146212061927074e-05,
+      "loss": 0.2458,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2898318654923939,
+      "learning_rate": 1.1475944057130856e-05,
+      "loss": 0.1125,
+      "step": 1448
+    },
+    {
+      "epoch": 0.29063250600480384,
+      "learning_rate": 1.1489764612440255e-05,
+      "loss": 0.4706,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29063250600480384,
+      "learning_rate": 1.150358225820707e-05,
+      "loss": 0.1494,
+      "step": 1452
+    },
+    {
+      "epoch": 0.29143314651721375,
+      "learning_rate": 1.151739696744518e-05,
+      "loss": 0.1354,
+      "step": 1454
+    },
+    {
+      "epoch": 0.29143314651721375,
+      "learning_rate": 1.1531208713174138e-05,
+      "loss": 0.2099,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2922337870296237,
+      "learning_rate": 1.1545017468419307e-05,
+      "loss": 0.2054,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2922337870296237,
+      "learning_rate": 1.1558823206211887e-05,
+      "loss": 0.0943,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2930344275420336,
+      "learning_rate": 1.1572625899588972e-05,
+      "loss": 0.3127,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2930344275420336,
+      "learning_rate": 1.1586425521593607e-05,
+      "loss": 0.1961,
+      "step": 1464
+    },
+    {
+      "epoch": 0.29383506805444354,
+      "learning_rate": 1.1600222045274809e-05,
+      "loss": 0.2268,
+      "step": 1466
+    },
+    {
+      "epoch": 0.29383506805444354,
+      "learning_rate": 1.1614015443687708e-05,
+      "loss": 0.0103,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2946357085668535,
+      "learning_rate": 1.1627805689893478e-05,
+      "loss": 0.2393,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2946357085668535,
+      "learning_rate": 1.1641592756959467e-05,
+      "loss": 0.5504,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2954363490792634,
+      "learning_rate": 1.1655376617959239e-05,
+      "loss": 0.7182,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2954363490792634,
+      "learning_rate": 1.1669157245972616e-05,
+      "loss": 0.6106,
+      "step": 1476
+    },
+    {
+      "epoch": 0.2962369895916733,
+      "learning_rate": 1.1682934614085708e-05,
+      "loss": 0.4883,
+      "step": 1478
+    },
+    {
+      "epoch": 0.2962369895916733,
+      "learning_rate": 1.1696708695391057e-05,
+      "loss": 0.2193,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29703763010408324,
+      "learning_rate": 1.1710479462987565e-05,
+      "loss": 0.188,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29703763010408324,
+      "learning_rate": 1.1724246889980626e-05,
+      "loss": 0.2708,
+      "step": 1484
+    },
+    {
+      "epoch": 0.2978382706164932,
+      "learning_rate": 1.1738010949482152e-05,
+      "loss": 0.2946,
+      "step": 1486
+    },
+    {
+      "epoch": 0.2978382706164932,
+      "learning_rate": 1.1751771614610643e-05,
+      "loss": 0.4407,
+      "step": 1488
+    },
+    {
+      "epoch": 0.2986389111289031,
+      "learning_rate": 1.176552885849122e-05,
+      "loss": 0.7374,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2986389111289031,
+      "learning_rate": 1.1779282654255668e-05,
+      "loss": 0.2638,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29943955164131303,
+      "learning_rate": 1.1793032975042563e-05,
+      "loss": 0.5854,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29943955164131303,
+      "learning_rate": 1.180677979399721e-05,
+      "loss": 0.344,
+      "step": 1496
+    },
+    {
+      "epoch": 0.300240192153723,
+      "learning_rate": 1.1820523084271775e-05,
+      "loss": 0.3481,
+      "step": 1498
+    },
+    {
+      "epoch": 0.300240192153723,
+      "learning_rate": 1.1834262819025317e-05,
+      "loss": 0.202,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3010408326661329,
+      "learning_rate": 1.1847998971423835e-05,
+      "loss": 0.2194,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3010408326661329,
+      "learning_rate": 1.1861731514640309e-05,
+      "loss": 0.1423,
+      "step": 1504
+    },
+    {
+      "epoch": 0.3018414731785428,
+      "learning_rate": 1.1875460421854816e-05,
+      "loss": 0.5414,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3018414731785428,
+      "learning_rate": 1.188918566625449e-05,
+      "loss": 0.2507,
+      "step": 1508
+    },
+    {
+      "epoch": 0.3026421136909528,
+      "learning_rate": 1.1902907221033629e-05,
+      "loss": 0.2284,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3026421136909528,
+      "learning_rate": 1.1916625059393739e-05,
+      "loss": 0.3953,
+      "step": 1512
+    },
+    {
+      "epoch": 0.3034427542033627,
+      "learning_rate": 1.1930339154543582e-05,
+      "loss": 0.1421,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3034427542033627,
+      "learning_rate": 1.1944049479699241e-05,
+      "loss": 0.1371,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3042433947157726,
+      "learning_rate": 1.1957756008084127e-05,
+      "loss": 0.134,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3042433947157726,
+      "learning_rate": 1.1971458712929133e-05,
+      "loss": 0.1444,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3050440352281825,
+      "learning_rate": 1.1985157567472563e-05,
+      "loss": 0.3414,
+      "step": 1522
+    },
+    {
+      "epoch": 0.3050440352281825,
+      "learning_rate": 1.1998852544960256e-05,
+      "loss": 0.1708,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3058446757405925,
+      "learning_rate": 1.2012543618645622e-05,
+      "loss": 0.4088,
+      "step": 1526
+    },
+    {
+      "epoch": 0.3058446757405925,
+      "learning_rate": 1.2026230761789702e-05,
+      "loss": 0.5856,
+      "step": 1528
+    },
+    {
+      "epoch": 0.3066453162530024,
+      "learning_rate": 1.2039913947661205e-05,
+      "loss": 0.1267,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3066453162530024,
+      "learning_rate": 1.2053593149536557e-05,
+      "loss": 0.1749,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3074459567654123,
+      "learning_rate": 1.2067268340700016e-05,
+      "loss": 0.171,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3074459567654123,
+      "learning_rate": 1.2080939494443618e-05,
+      "loss": 0.5205,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3082465972778223,
+      "learning_rate": 1.2094606584067304e-05,
+      "loss": 0.2663,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3082465972778223,
+      "learning_rate": 1.210826958287895e-05,
+      "loss": 0.5045,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3090472377902322,
+      "learning_rate": 1.212192846419443e-05,
+      "loss": 0.2819,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3090472377902322,
+      "learning_rate": 1.2135583201337646e-05,
+      "loss": 0.5187,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3098478783026421,
+      "learning_rate": 1.2149233767640587e-05,
+      "loss": 0.1424,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3098478783026421,
+      "learning_rate": 1.2162880136443434e-05,
+      "loss": 0.0775,
+      "step": 1548
+    },
+    {
+      "epoch": 0.31064851881505207,
+      "learning_rate": 1.2176522281094514e-05,
+      "loss": 0.2619,
+      "step": 1550
+    },
+    {
+      "epoch": 0.31064851881505207,
+      "learning_rate": 1.2190160174950428e-05,
+      "loss": 0.3382,
+      "step": 1552
+    },
+    {
+      "epoch": 0.311449159327462,
+      "learning_rate": 1.220379379137607e-05,
+      "loss": 0.1415,
+      "step": 1554
+    },
+    {
+      "epoch": 0.311449159327462,
+      "learning_rate": 1.2217423103744692e-05,
+      "loss": 0.8941,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3122497998398719,
+      "learning_rate": 1.2231048085437953e-05,
+      "loss": 0.3341,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3122497998398719,
+      "learning_rate": 1.2244668709845952e-05,
+      "loss": 0.1666,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3130504403522818,
+      "learning_rate": 1.2258284950367347e-05,
+      "loss": 0.557,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3130504403522818,
+      "learning_rate": 1.2271896780409309e-05,
+      "loss": 0.3639,
+      "step": 1564
+    },
+    {
+      "epoch": 0.31385108086469177,
+      "learning_rate": 1.228550417338764e-05,
+      "loss": 0.2868,
+      "step": 1566
+    },
+    {
+      "epoch": 0.31385108086469177,
+      "learning_rate": 1.2299107102726804e-05,
+      "loss": 0.4484,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3146517213771017,
+      "learning_rate": 1.2312705541859985e-05,
+      "loss": 0.4792,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3146517213771017,
+      "learning_rate": 1.2326299464229143e-05,
+      "loss": 0.4259,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3154523618895116,
+      "learning_rate": 1.2339888843285029e-05,
+      "loss": 0.2203,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3154523618895116,
+      "learning_rate": 1.2353473652487329e-05,
+      "loss": 0.2445,
+      "step": 1576
+    },
+    {
+      "epoch": 0.31625300240192156,
+      "learning_rate": 1.2367053865304597e-05,
+      "loss": 0.1356,
+      "step": 1578
+    },
+    {
+      "epoch": 0.31625300240192156,
+      "learning_rate": 1.2380629455214385e-05,
+      "loss": 0.1783,
+      "step": 1580
+    },
+    {
+      "epoch": 0.31705364291433147,
+      "learning_rate": 1.2394200395703273e-05,
+      "loss": 0.4181,
+      "step": 1582
+    },
+    {
+      "epoch": 0.31705364291433147,
+      "learning_rate": 1.2407766660266916e-05,
+      "loss": 0.5074,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3178542834267414,
+      "learning_rate": 1.2421328222410109e-05,
+      "loss": 0.2392,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3178542834267414,
+      "learning_rate": 1.2434885055646808e-05,
+      "loss": 0.2884,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31865492393915135,
+      "learning_rate": 1.2448437133500262e-05,
+      "loss": 0.2466,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31865492393915135,
+      "learning_rate": 1.2461984429502947e-05,
+      "loss": 0.2529,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31945556445156126,
+      "learning_rate": 1.2475526917196703e-05,
+      "loss": 0.204,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31945556445156126,
+      "learning_rate": 1.2489064570132761e-05,
+      "loss": 0.1597,
+      "step": 1596
+    },
+    {
+      "epoch": 0.32025620496397117,
+      "learning_rate": 1.2502597361871787e-05,
+      "loss": 0.1865,
+      "step": 1598
+    },
+    {
+      "epoch": 0.32025620496397117,
+      "learning_rate": 1.2516125265983945e-05,
+      "loss": 0.5956,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3210568454763811,
+      "learning_rate": 1.2529648256048931e-05,
+      "loss": 0.4371,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3210568454763811,
+      "learning_rate": 1.2543166305656089e-05,
+      "loss": 0.4267,
+      "step": 1604
+    },
+    {
+      "epoch": 0.32185748598879105,
+      "learning_rate": 1.2556679388404351e-05,
+      "loss": 0.1836,
+      "step": 1606
+    },
+    {
+      "epoch": 0.32185748598879105,
+      "learning_rate": 1.257018747790238e-05,
+      "loss": 0.107,
+      "step": 1608
+    },
+    {
+      "epoch": 0.32265812650120096,
+      "learning_rate": 1.2583690547768584e-05,
+      "loss": 0.2521,
+      "step": 1610
+    },
+    {
+      "epoch": 0.32265812650120096,
+      "learning_rate": 1.259718857163117e-05,
+      "loss": 0.0604,
+      "step": 1612
+    },
+    {
+      "epoch": 0.32345876701361087,
+      "learning_rate": 1.261068152312821e-05,
+      "loss": 0.4678,
+      "step": 1614
+    },
+    {
+      "epoch": 0.32345876701361087,
+      "learning_rate": 1.2624169375907657e-05,
+      "loss": 0.3044,
+      "step": 1616
+    },
+    {
+      "epoch": 0.32425940752602084,
+      "learning_rate": 1.2637652103627481e-05,
+      "loss": 0.2477,
+      "step": 1618
+    },
+    {
+      "epoch": 0.32425940752602084,
+      "learning_rate": 1.2651129679955598e-05,
+      "loss": 0.1585,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32506004803843075,
+      "learning_rate": 1.2664602078570017e-05,
+      "loss": 0.1879,
+      "step": 1622
+    },
+    {
+      "epoch": 0.32506004803843075,
+      "learning_rate": 1.2678069273158849e-05,
+      "loss": 0.1581,
+      "step": 1624
+    },
+    {
+      "epoch": 0.32586068855084066,
+      "learning_rate": 1.2691531237420369e-05,
+      "loss": 0.352,
+      "step": 1626
+    },
+    {
+      "epoch": 0.32586068855084066,
+      "learning_rate": 1.2704987945063073e-05,
+      "loss": 0.2161,
+      "step": 1628
+    },
+    {
+      "epoch": 0.3266613290632506,
+      "learning_rate": 1.27184393698057e-05,
+      "loss": 0.2447,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3266613290632506,
+      "learning_rate": 1.273188548537736e-05,
+      "loss": 0.1586,
+      "step": 1632
+    },
+    {
+      "epoch": 0.32746196957566054,
+      "learning_rate": 1.2745326265517481e-05,
+      "loss": 0.2529,
+      "step": 1634
+    },
+    {
+      "epoch": 0.32746196957566054,
+      "learning_rate": 1.2758761683975929e-05,
+      "loss": 0.2019,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32826261008807045,
+      "learning_rate": 1.277219171451304e-05,
+      "loss": 0.2159,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32826261008807045,
+      "learning_rate": 1.2785616330899676e-05,
+      "loss": 0.347,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32906325060048036,
+      "learning_rate": 1.2799035506917265e-05,
+      "loss": 0.2066,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32906325060048036,
+      "learning_rate": 1.2812449216357855e-05,
+      "loss": 0.0917,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32986389111289033,
+      "learning_rate": 1.2825857433024208e-05,
+      "loss": 0.4873,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32986389111289033,
+      "learning_rate": 1.2839260130729776e-05,
+      "loss": 0.151,
+      "step": 1648
+    },
+    {
+      "epoch": 0.33066453162530024,
+      "learning_rate": 1.2852657283298794e-05,
+      "loss": 0.4022,
+      "step": 1650
+    },
+    {
+      "epoch": 0.33066453162530024,
+      "learning_rate": 1.2866048864566336e-05,
+      "loss": 0.0321,
+      "step": 1652
+    },
+    {
+      "epoch": 0.33146517213771015,
+      "learning_rate": 1.2879434848378356e-05,
+      "loss": 0.3497,
+      "step": 1654
+    },
+    {
+      "epoch": 0.33146517213771015,
+      "learning_rate": 1.2892815208591734e-05,
+      "loss": 0.4356,
+      "step": 1656
+    },
+    {
+      "epoch": 0.3322658126501201,
+      "learning_rate": 1.2906189919074336e-05,
+      "loss": 0.2773,
+      "step": 1658
+    },
+    {
+      "epoch": 0.3322658126501201,
+      "learning_rate": 1.2919558953705047e-05,
+      "loss": 0.7321,
+      "step": 1660
+    },
+    {
+      "epoch": 0.33306645316253003,
+      "learning_rate": 1.293292228637389e-05,
+      "loss": 0.5336,
+      "step": 1662
+    },
+    {
+      "epoch": 0.33306645316253003,
+      "learning_rate": 1.2946279890981966e-05,
+      "loss": 0.1866,
+      "step": 1664
+    },
+    {
+      "epoch": 0.33386709367493994,
+      "learning_rate": 1.2959631741441583e-05,
+      "loss": 0.4319,
+      "step": 1666
+    },
+    {
+      "epoch": 0.33386709367493994,
+      "learning_rate": 1.2972977811676289e-05,
+      "loss": 0.0908,
+      "step": 1668
+    },
+    {
+      "epoch": 0.33466773418734985,
+      "learning_rate": 1.298631807562092e-05,
+      "loss": 0.1757,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33466773418734985,
+      "learning_rate": 1.2999652507221652e-05,
+      "loss": 0.1134,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3354683746997598,
+      "learning_rate": 1.3012981080436036e-05,
+      "loss": 0.9963,
+      "step": 1674
+    },
+    {
+      "epoch": 0.3354683746997598,
+      "learning_rate": 1.3026303769233109e-05,
+      "loss": 0.1905,
+      "step": 1676
+    },
+    {
+      "epoch": 0.33626901521216973,
+      "learning_rate": 1.3039620547593357e-05,
+      "loss": 0.1582,
+      "step": 1678
+    },
+    {
+      "epoch": 0.33626901521216973,
+      "learning_rate": 1.3052931389508822e-05,
+      "loss": 0.3927,
+      "step": 1680
+    },
+    {
+      "epoch": 0.33706965572457964,
+      "learning_rate": 1.3066236268983143e-05,
+      "loss": 0.2775,
+      "step": 1682
+    },
+    {
+      "epoch": 0.33706965572457964,
+      "learning_rate": 1.3079535160031601e-05,
+      "loss": 0.3273,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3378702962369896,
+      "learning_rate": 1.3092828036681178e-05,
+      "loss": 0.3939,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3378702962369896,
+      "learning_rate": 1.3106114872970575e-05,
+      "loss": 0.6645,
+      "step": 1688
+    },
+    {
+      "epoch": 0.3386709367493995,
+      "learning_rate": 1.3119395642950348e-05,
+      "loss": 0.5927,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3386709367493995,
+      "learning_rate": 1.313267032068285e-05,
+      "loss": 0.1866,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33947157726180943,
+      "learning_rate": 1.3145938880242346e-05,
+      "loss": 0.5036,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33947157726180943,
+      "learning_rate": 1.3159201295715054e-05,
+      "loss": 0.2308,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3402722177742194,
+      "learning_rate": 1.3172457541199188e-05,
+      "loss": 0.1594,
+      "step": 1698
+    },
+    {
+      "epoch": 0.3402722177742194,
+      "learning_rate": 1.3185707590804997e-05,
+      "loss": 0.1036,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3410728582866293,
+      "learning_rate": 1.3198951418654882e-05,
+      "loss": 0.1588,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3410728582866293,
+      "learning_rate": 1.321218899888334e-05,
+      "loss": 0.3016,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3418734987990392,
+      "learning_rate": 1.322542030563709e-05,
+      "loss": 0.1278,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3418734987990392,
+      "learning_rate": 1.3238645313075109e-05,
+      "loss": 0.198,
+      "step": 1708
+    },
+    {
+      "epoch": 0.34267413931144913,
+      "learning_rate": 1.3251863995368665e-05,
+      "loss": 0.3702,
+      "step": 1710
+    },
+    {
+      "epoch": 0.34267413931144913,
+      "learning_rate": 1.326507632670139e-05,
+      "loss": 0.3422,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3434747798238591,
+      "learning_rate": 1.3278282281269293e-05,
+      "loss": 0.1581,
+      "step": 1714
+    },
+    {
+      "epoch": 0.3434747798238591,
+      "learning_rate": 1.3291481833280894e-05,
+      "loss": 0.1753,
+      "step": 1716
+    },
+    {
+      "epoch": 0.344275420336269,
+      "learning_rate": 1.3304674956957167e-05,
+      "loss": 0.1498,
+      "step": 1718
+    },
+    {
+      "epoch": 0.344275420336269,
+      "learning_rate": 1.3317861626531652e-05,
+      "loss": 0.9579,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3450760608486789,
+      "learning_rate": 1.3331041816250503e-05,
+      "loss": 0.3371,
+      "step": 1722
+    },
+    {
+      "epoch": 0.3450760608486789,
+      "learning_rate": 1.3344215500372517e-05,
+      "loss": 0.29,
+      "step": 1724
+    },
+    {
+      "epoch": 0.3458767013610889,
+      "learning_rate": 1.335738265316921e-05,
+      "loss": 0.1687,
+      "step": 1726
+    },
+    {
+      "epoch": 0.3458767013610889,
+      "learning_rate": 1.3370543248924826e-05,
+      "loss": 0.0662,
+      "step": 1728
+    },
+    {
+      "epoch": 0.3466773418734988,
+      "learning_rate": 1.3383697261936472e-05,
+      "loss": 0.1648,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3466773418734988,
+      "learning_rate": 1.3396844666514062e-05,
+      "loss": 0.1584,
+      "step": 1732
+    },
+    {
+      "epoch": 0.3474779823859087,
+      "learning_rate": 1.3409985436980422e-05,
+      "loss": 0.6786,
+      "step": 1734
+    },
+    {
+      "epoch": 0.3474779823859087,
+      "learning_rate": 1.3423119547671348e-05,
+      "loss": 0.1425,
+      "step": 1736
+    },
+    {
+      "epoch": 0.3482786228983187,
+      "learning_rate": 1.3436246972935638e-05,
+      "loss": 0.1852,
+      "step": 1738
+    },
+    {
+      "epoch": 0.3482786228983187,
+      "learning_rate": 1.344936768713513e-05,
+      "loss": 0.1847,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3490792634107286,
+      "learning_rate": 1.346248166464481e-05,
+      "loss": 0.3442,
+      "step": 1742
+    },
+    {
+      "epoch": 0.3490792634107286,
+      "learning_rate": 1.347558887985279e-05,
+      "loss": 0.2026,
+      "step": 1744
+    },
+    {
+      "epoch": 0.3498799039231385,
+      "learning_rate": 1.348868930716039e-05,
+      "loss": 0.1953,
+      "step": 1746
+    },
+    {
+      "epoch": 0.3498799039231385,
+      "learning_rate": 1.3501782920982189e-05,
+      "loss": 0.2775,
+      "step": 1748
+    },
+    {
+      "epoch": 0.3506805444355484,
+      "learning_rate": 1.3514869695746078e-05,
+      "loss": 0.7842,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3506805444355484,
+      "learning_rate": 1.3527949605893305e-05,
+      "loss": 0.1103,
+      "step": 1752
+    },
+    {
+      "epoch": 0.3514811849479584,
+      "learning_rate": 1.3541022625878501e-05,
+      "loss": 0.4597,
+      "step": 1754
+    },
+    {
+      "epoch": 0.3514811849479584,
+      "learning_rate": 1.3554088730169812e-05,
+      "loss": 0.155,
+      "step": 1756
+    },
+    {
+      "epoch": 0.3522818254603683,
+      "learning_rate": 1.3567147893248833e-05,
+      "loss": 0.3697,
+      "step": 1758
+    },
+    {
+      "epoch": 0.3522818254603683,
+      "learning_rate": 1.3580200089610739e-05,
+      "loss": 0.3635,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3530824659727782,
+      "learning_rate": 1.3593245293764303e-05,
+      "loss": 0.2324,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3530824659727782,
+      "learning_rate": 1.3606283480231962e-05,
+      "loss": 0.1206,
+      "step": 1764
+    },
+    {
+      "epoch": 0.35388310648518817,
+      "learning_rate": 1.361931462354984e-05,
+      "loss": 0.4571,
+      "step": 1766
+    },
+    {
+      "epoch": 0.35388310648518817,
+      "learning_rate": 1.3632338698267863e-05,
+      "loss": 0.2093,
+      "step": 1768
+    },
+    {
+      "epoch": 0.3546837469975981,
+      "learning_rate": 1.3645355678949715e-05,
+      "loss": 0.2104,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3546837469975981,
+      "learning_rate": 1.3658365540172948e-05,
+      "loss": 0.1509,
+      "step": 1772
+    },
+    {
+      "epoch": 0.355484387510008,
+      "learning_rate": 1.3671368256529026e-05,
+      "loss": 0.3641,
+      "step": 1774
+    },
+    {
+      "epoch": 0.355484387510008,
+      "learning_rate": 1.368436380262336e-05,
+      "loss": 0.4026,
+      "step": 1776
+    },
+    {
+      "epoch": 0.35628502802241796,
+      "learning_rate": 1.3697352153075365e-05,
+      "loss": 0.2054,
+      "step": 1778
+    },
+    {
+      "epoch": 0.35628502802241796,
+      "learning_rate": 1.3710333282518497e-05,
+      "loss": 0.3873,
+      "step": 1780
+    },
+    {
+      "epoch": 0.35708566853482787,
+      "learning_rate": 1.3723307165600361e-05,
+      "loss": 0.2098,
+      "step": 1782
+    },
+    {
+      "epoch": 0.35708566853482787,
+      "learning_rate": 1.3736273776982667e-05,
+      "loss": 0.5413,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3578863090472378,
+      "learning_rate": 1.3749233091341344e-05,
+      "loss": 0.6091,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3578863090472378,
+      "learning_rate": 1.3762185083366562e-05,
+      "loss": 0.363,
+      "step": 1788
+    },
+    {
+      "epoch": 0.3586869495596477,
+      "learning_rate": 1.3775129727762808e-05,
+      "loss": 0.1509,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3586869495596477,
+      "learning_rate": 1.3788066999248893e-05,
+      "loss": 0.5257,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35948759007205766,
+      "learning_rate": 1.3800996872558075e-05,
+      "loss": 0.1339,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35948759007205766,
+      "learning_rate": 1.3813919322438018e-05,
+      "loss": 0.2294,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3602882305844676,
+      "learning_rate": 1.3826834323650899e-05,
+      "loss": 0.1752,
+      "step": 1798
+    },
+    {
+      "epoch": 0.3602882305844676,
+      "learning_rate": 1.3839741850973435e-05,
+      "loss": 0.2616,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3610888710968775,
+      "learning_rate": 1.3852641879196952e-05,
+      "loss": 0.223,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3610888710968775,
+      "learning_rate": 1.3865534383127413e-05,
+      "loss": 0.2532,
+      "step": 1804
+    },
+    {
+      "epoch": 0.36188951160928745,
+      "learning_rate": 1.387841933758546e-05,
+      "loss": 0.6906,
+      "step": 1806
+    },
+    {
+      "epoch": 0.36188951160928745,
+      "learning_rate": 1.3891296717406533e-05,
+      "loss": 0.1677,
+      "step": 1808
+    },
+    {
+      "epoch": 0.36269015212169736,
+      "learning_rate": 1.3904166497440812e-05,
+      "loss": 0.382,
+      "step": 1810
+    },
+    {
+      "epoch": 0.36269015212169736,
+      "learning_rate": 1.391702865255334e-05,
+      "loss": 0.0419,
+      "step": 1812
+    },
+    {
+      "epoch": 0.3634907926341073,
+      "learning_rate": 1.3929883157624046e-05,
+      "loss": 0.2186,
+      "step": 1814
+    },
+    {
+      "epoch": 0.3634907926341073,
+      "learning_rate": 1.3942729987547808e-05,
+      "loss": 0.1339,
+      "step": 1816
+    },
+    {
+      "epoch": 0.36429143314651724,
+      "learning_rate": 1.3955569117234468e-05,
+      "loss": 0.2884,
+      "step": 1818
+    },
+    {
+      "epoch": 0.36429143314651724,
+      "learning_rate": 1.3968400521608962e-05,
+      "loss": 0.4636,
+      "step": 1820
+    },
+    {
+      "epoch": 0.36509207365892715,
+      "learning_rate": 1.3981224175611265e-05,
+      "loss": 0.1977,
+      "step": 1822
+    },
+    {
+      "epoch": 0.36509207365892715,
+      "learning_rate": 1.3994040054196498e-05,
+      "loss": 0.1427,
+      "step": 1824
+    },
+    {
+      "epoch": 0.36589271417133706,
+      "learning_rate": 1.4006848132334979e-05,
+      "loss": 0.2076,
+      "step": 1826
+    },
+    {
+      "epoch": 0.36589271417133706,
+      "learning_rate": 1.4019648385012245e-05,
+      "loss": 0.2774,
+      "step": 1828
+    },
+    {
+      "epoch": 0.366693354683747,
+      "learning_rate": 1.4032440787229135e-05,
+      "loss": 0.2814,
+      "step": 1830
+    },
+    {
+      "epoch": 0.366693354683747,
+      "learning_rate": 1.4045225314001789e-05,
+      "loss": 0.0472,
+      "step": 1832
+    },
+    {
+      "epoch": 0.36749399519615694,
+      "learning_rate": 1.4058001940361781e-05,
+      "loss": 0.1134,
+      "step": 1834
+    },
+    {
+      "epoch": 0.36749399519615694,
+      "learning_rate": 1.4070770641356069e-05,
+      "loss": 0.1165,
+      "step": 1836
+    },
+    {
+      "epoch": 0.36829463570856685,
+      "learning_rate": 1.40835313920471e-05,
+      "loss": 0.5222,
+      "step": 1838
+    },
+    {
+      "epoch": 0.36829463570856685,
+      "learning_rate": 1.4096284167512856e-05,
+      "loss": 0.093,
+      "step": 1840
+    },
+    {
+      "epoch": 0.36909527622097676,
+      "learning_rate": 1.4109028942846888e-05,
+      "loss": 0.3191,
+      "step": 1842
+    },
+    {
+      "epoch": 0.36909527622097676,
+      "learning_rate": 1.4121765693158355e-05,
+      "loss": 0.0727,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36989591673338673,
+      "learning_rate": 1.4134494393572146e-05,
+      "loss": 0.4688,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36989591673338673,
+      "learning_rate": 1.4147215019228813e-05,
+      "loss": 0.473,
+      "step": 1848
+    },
+    {
+      "epoch": 0.37069655724579664,
+      "learning_rate": 1.4159927545284697e-05,
+      "loss": 0.3694,
+      "step": 1850
+    },
+    {
+      "epoch": 0.37069655724579664,
+      "learning_rate": 1.4172631946911964e-05,
+      "loss": 0.3039,
+      "step": 1852
+    },
+    {
+      "epoch": 0.37149719775820655,
+      "learning_rate": 1.4185328199298636e-05,
+      "loss": 0.5704,
+      "step": 1854
+    },
+    {
+      "epoch": 0.37149719775820655,
+      "learning_rate": 1.4198016277648665e-05,
+      "loss": 0.2662,
+      "step": 1856
+    },
+    {
+      "epoch": 0.37229783827061647,
+      "learning_rate": 1.4210696157181936e-05,
+      "loss": 0.1665,
+      "step": 1858
+    },
+    {
+      "epoch": 0.37229783827061647,
+      "learning_rate": 1.4223367813134406e-05,
+      "loss": 0.3147,
+      "step": 1860
+    },
+    {
+      "epoch": 0.37309847878302643,
+      "learning_rate": 1.4236031220758037e-05,
+      "loss": 0.3042,
+      "step": 1862
+    },
+    {
+      "epoch": 0.37309847878302643,
+      "learning_rate": 1.4248686355320922e-05,
+      "loss": 0.508,
+      "step": 1864
+    },
+    {
+      "epoch": 0.37389911929543634,
+      "learning_rate": 1.426133319210731e-05,
+      "loss": 0.382,
+      "step": 1866
+    },
+    {
+      "epoch": 0.37389911929543634,
+      "learning_rate": 1.4273971706417653e-05,
+      "loss": 0.2046,
+      "step": 1868
+    },
+    {
+      "epoch": 0.37469975980784626,
+      "learning_rate": 1.4286601873568642e-05,
+      "loss": 0.2612,
+      "step": 1870
+    },
+    {
+      "epoch": 0.37469975980784626,
+      "learning_rate": 1.429922366889332e-05,
+      "loss": 0.2237,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3755004003202562,
+      "learning_rate": 1.431183706774103e-05,
+      "loss": 0.1515,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3755004003202562,
+      "learning_rate": 1.4324442045477534e-05,
+      "loss": 0.1314,
+      "step": 1876
+    },
+    {
+      "epoch": 0.37630104083266613,
+      "learning_rate": 1.4337038577485035e-05,
+      "loss": 0.3572,
+      "step": 1878
+    },
+    {
+      "epoch": 0.37630104083266613,
+      "learning_rate": 1.4349626639162231e-05,
+      "loss": 0.2457,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37710168134507605,
+      "learning_rate": 1.436220620592437e-05,
+      "loss": 0.2285,
+      "step": 1882
+    },
+    {
+      "epoch": 0.37710168134507605,
+      "learning_rate": 1.4374777253203265e-05,
+      "loss": 0.1202,
+      "step": 1884
+    },
+    {
+      "epoch": 0.377902321857486,
+      "learning_rate": 1.4387339756447422e-05,
+      "loss": 0.1896,
+      "step": 1886
+    },
+    {
+      "epoch": 0.377902321857486,
+      "learning_rate": 1.4399893691121985e-05,
+      "loss": 0.2798,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3787029623698959,
+      "learning_rate": 1.4412439032708848e-05,
+      "loss": 0.4395,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3787029623698959,
+      "learning_rate": 1.4424975756706684e-05,
+      "loss": 0.7018,
+      "step": 1892
+    },
+    {
+      "epoch": 0.37950360288230583,
+      "learning_rate": 1.4437503838631002e-05,
+      "loss": 0.1509,
+      "step": 1894
+    },
+    {
+      "epoch": 0.37950360288230583,
+      "learning_rate": 1.4450023254014185e-05,
+      "loss": 0.2884,
+      "step": 1896
+    },
+    {
+      "epoch": 0.38030424339471575,
+      "learning_rate": 1.4462533978405529e-05,
+      "loss": 0.1681,
+      "step": 1898
+    },
+    {
+      "epoch": 0.38030424339471575,
+      "learning_rate": 1.4475035987371348e-05,
+      "loss": 0.6103,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3811048839071257,
+      "learning_rate": 1.4487529256494937e-05,
+      "loss": 0.4381,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3811048839071257,
+      "learning_rate": 1.4500013761376663e-05,
+      "loss": 0.0341,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3819055244195356,
+      "learning_rate": 1.4512489477634024e-05,
+      "loss": 0.2503,
+      "step": 1906
+    },
+    {
+      "epoch": 0.3819055244195356,
+      "learning_rate": 1.4524956380901674e-05,
+      "loss": 0.5279,
+      "step": 1908
+    },
+    {
+      "epoch": 0.38270616493194554,
+      "learning_rate": 1.4537414446831461e-05,
+      "loss": 0.3645,
+      "step": 1910
+    },
+    {
+      "epoch": 0.38270616493194554,
+      "learning_rate": 1.454986365109255e-05,
+      "loss": 0.2444,
+      "step": 1912
+    },
+    {
+      "epoch": 0.3835068054443555,
+      "learning_rate": 1.4562303969371357e-05,
+      "loss": 0.6764,
+      "step": 1914
+    },
+    {
+      "epoch": 0.3835068054443555,
+      "learning_rate": 1.4574735377371669e-05,
+      "loss": 0.1493,
+      "step": 1916
+    },
+    {
+      "epoch": 0.3843074459567654,
+      "learning_rate": 1.4587157850814679e-05,
+      "loss": 0.4251,
+      "step": 1918
+    },
+    {
+      "epoch": 0.3843074459567654,
+      "learning_rate": 1.4599571365439027e-05,
+      "loss": 0.5829,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3851080864691753,
+      "learning_rate": 1.4611975897000849e-05,
+      "loss": 0.1838,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3851080864691753,
+      "learning_rate": 1.4624371421273812e-05,
+      "loss": 0.4492,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3859087269815853,
+      "learning_rate": 1.463675791404922e-05,
+      "loss": 0.1978,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3859087269815853,
+      "learning_rate": 1.4649135351135968e-05,
+      "loss": 0.2612,
+      "step": 1928
+    },
+    {
+      "epoch": 0.3867093674939952,
+      "learning_rate": 1.4661503708360652e-05,
+      "loss": 0.2757,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3867093674939952,
+      "learning_rate": 1.4673862961567604e-05,
+      "loss": 0.1295,
+      "step": 1932
+    },
+    {
+      "epoch": 0.3875100080064051,
+      "learning_rate": 1.4686213086618932e-05,
+      "loss": 0.1662,
+      "step": 1934
+    },
+    {
+      "epoch": 0.3875100080064051,
+      "learning_rate": 1.4698554059394563e-05,
+      "loss": 0.5028,
+      "step": 1936
+    },
+    {
+      "epoch": 0.388310648518815,
+      "learning_rate": 1.4710885855792338e-05,
+      "loss": 0.3159,
+      "step": 1938
+    },
+    {
+      "epoch": 0.388310648518815,
+      "learning_rate": 1.4723208451727977e-05,
+      "loss": 0.2233,
+      "step": 1940
+    },
+    {
+      "epoch": 0.389111289031225,
+      "learning_rate": 1.4735521823135184e-05,
+      "loss": 0.7031,
+      "step": 1942
+    },
+    {
+      "epoch": 0.389111289031225,
+      "learning_rate": 1.4747825945965675e-05,
+      "loss": 0.5282,
+      "step": 1944
+    },
+    {
+      "epoch": 0.3899119295436349,
+      "learning_rate": 1.4760120796189233e-05,
+      "loss": 0.2355,
+      "step": 1946
+    },
+    {
+      "epoch": 0.3899119295436349,
+      "learning_rate": 1.4772406349793749e-05,
+      "loss": 0.2714,
+      "step": 1948
+    },
+    {
+      "epoch": 0.3907125700560448,
+      "learning_rate": 1.4784682582785254e-05,
+      "loss": 0.098,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3907125700560448,
+      "learning_rate": 1.4796949471188033e-05,
+      "loss": 0.0441,
+      "step": 1952
+    },
+    {
+      "epoch": 0.3915132105684548,
+      "learning_rate": 1.4809206991044571e-05,
+      "loss": 0.4596,
+      "step": 1954
+    },
+    {
+      "epoch": 0.3915132105684548,
+      "learning_rate": 1.4821455118415666e-05,
+      "loss": 0.48,
+      "step": 1956
+    },
+    {
+      "epoch": 0.3923138510808647,
+      "learning_rate": 1.4833693829380458e-05,
+      "loss": 0.602,
+      "step": 1958
+    },
+    {
+      "epoch": 0.3923138510808647,
+      "learning_rate": 1.4845923100036479e-05,
+      "loss": 0.2665,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3931144915932746,
+      "learning_rate": 1.4858142906499686e-05,
+      "loss": 0.2872,
+      "step": 1962
+    },
+    {
+      "epoch": 0.3931144915932746,
+      "learning_rate": 1.4870353224904563e-05,
+      "loss": 0.2957,
+      "step": 1964
+    },
+    {
+      "epoch": 0.3939151321056846,
+      "learning_rate": 1.4882554031404075e-05,
+      "loss": 0.3213,
+      "step": 1966
+    },
+    {
+      "epoch": 0.3939151321056846,
+      "learning_rate": 1.4894745302169786e-05,
+      "loss": 0.4493,
+      "step": 1968
+    },
+    {
+      "epoch": 0.3947157726180945,
+      "learning_rate": 1.4906927013391879e-05,
+      "loss": 0.1186,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3947157726180945,
+      "learning_rate": 1.4919099141279205e-05,
+      "loss": 0.0231,
+      "step": 1972
+    },
+    {
+      "epoch": 0.3955164131305044,
+      "learning_rate": 1.4931261662059338e-05,
+      "loss": 0.5704,
+      "step": 1974
+    },
+    {
+      "epoch": 0.3955164131305044,
+      "learning_rate": 1.4943414551978597e-05,
+      "loss": 0.1527,
+      "step": 1976
+    },
+    {
+      "epoch": 0.3963170536429143,
+      "learning_rate": 1.4955557787302151e-05,
+      "loss": 0.2607,
+      "step": 1978
+    },
+    {
+      "epoch": 0.3963170536429143,
+      "learning_rate": 1.4967691344313988e-05,
+      "loss": 0.4968,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3971176941553243,
+      "learning_rate": 1.4979815199317005e-05,
+      "loss": 0.3033,
+      "step": 1982
+    },
+    {
+      "epoch": 0.3971176941553243,
+      "learning_rate": 1.499192932863305e-05,
+      "loss": 0.3226,
+      "step": 1984
+    },
+    {
+      "epoch": 0.3979183346677342,
+      "learning_rate": 1.5004033708602967e-05,
+      "loss": 0.2159,
+      "step": 1986
+    },
+    {
+      "epoch": 0.3979183346677342,
+      "learning_rate": 1.5016128315586626e-05,
+      "loss": 0.3821,
+      "step": 1988
+    },
+    {
+      "epoch": 0.3987189751801441,
+      "learning_rate": 1.5028213125963029e-05,
+      "loss": 0.3776,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3987189751801441,
+      "learning_rate": 1.5040288116130261e-05,
+      "loss": 0.3598,
+      "step": 1992
+    },
+    {
+      "epoch": 0.39951961569255406,
+      "learning_rate": 1.5052353262505603e-05,
+      "loss": 0.1422,
+      "step": 1994
+    },
+    {
+      "epoch": 0.39951961569255406,
+      "learning_rate": 1.5064408541525568e-05,
+      "loss": 0.2193,
+      "step": 1996
+    },
+    {
+      "epoch": 0.400320256204964,
+      "learning_rate": 1.5076453929645933e-05,
+      "loss": 0.2446,
+      "step": 1998
+    },
+    {
+      "epoch": 0.400320256204964,
+      "learning_rate": 1.5088489403341793e-05,
+      "loss": 0.091,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4011208967173739,
+      "learning_rate": 1.510051493910759e-05,
+      "loss": 0.1945,
+      "step": 2002
+    },
+    {
+      "epoch": 0.4011208967173739,
+      "learning_rate": 1.5112530513457229e-05,
+      "loss": 0.4264,
+      "step": 2004
+    },
+    {
+      "epoch": 0.40192153722978385,
+      "learning_rate": 1.512453610292401e-05,
+      "loss": 0.3213,
+      "step": 2006
+    },
+    {
+      "epoch": 0.40192153722978385,
+      "learning_rate": 1.513653168406076e-05,
+      "loss": 0.4588,
+      "step": 2008
+    },
+    {
+      "epoch": 0.40272217774219377,
+      "learning_rate": 1.514851723343985e-05,
+      "loss": 0.2394,
+      "step": 2010
+    },
+    {
+      "epoch": 0.40272217774219377,
+      "learning_rate": 1.5160492727653245e-05,
+      "loss": 0.2162,
+      "step": 2012
+    },
+    {
+      "epoch": 0.4035228182546037,
+      "learning_rate": 1.5172458143312522e-05,
+      "loss": 0.2277,
+      "step": 2014
+    },
+    {
+      "epoch": 0.4035228182546037,
+      "learning_rate": 1.5184413457049006e-05,
+      "loss": 0.1239,
+      "step": 2016
+    },
+    {
+      "epoch": 0.4043234587670136,
+      "learning_rate": 1.5196358645513685e-05,
+      "loss": 0.3698,
+      "step": 2018
+    },
+    {
+      "epoch": 0.4043234587670136,
+      "learning_rate": 1.5208293685377354e-05,
+      "loss": 0.6308,
+      "step": 2020
+    },
+    {
+      "epoch": 0.40512409927942356,
+      "learning_rate": 1.5220218553330618e-05,
+      "loss": 0.1513,
+      "step": 2022
+    },
+    {
+      "epoch": 0.40512409927942356,
+      "learning_rate": 1.5232133226083954e-05,
+      "loss": 0.1977,
+      "step": 2024
+    },
+    {
+      "epoch": 0.40592473979183347,
+      "learning_rate": 1.5244037680367744e-05,
+      "loss": 0.1753,
+      "step": 2026
+    },
+    {
+      "epoch": 0.40592473979183347,
+      "learning_rate": 1.5255931892932322e-05,
+      "loss": 0.154,
+      "step": 2028
+    },
+    {
+      "epoch": 0.4067253803042434,
+      "learning_rate": 1.5267815840548057e-05,
+      "loss": 0.352,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4067253803042434,
+      "learning_rate": 1.527968950000533e-05,
+      "loss": 0.1966,
+      "step": 2032
+    },
+    {
+      "epoch": 0.40752602081665334,
+      "learning_rate": 1.529155284811463e-05,
+      "loss": 0.2521,
+      "step": 2034
+    },
+    {
+      "epoch": 0.40752602081665334,
+      "learning_rate": 1.5303405861706574e-05,
+      "loss": 0.3046,
+      "step": 2036
+    },
+    {
+      "epoch": 0.40832666132906326,
+      "learning_rate": 1.5315248517631975e-05,
+      "loss": 0.6125,
+      "step": 2038
+    },
+    {
+      "epoch": 0.40832666132906326,
+      "learning_rate": 1.532708079276185e-05,
+      "loss": 0.0514,
+      "step": 2040
+    },
+    {
+      "epoch": 0.40912730184147317,
+      "learning_rate": 1.5338902663987544e-05,
+      "loss": 0.1347,
+      "step": 2042
+    },
+    {
+      "epoch": 0.40912730184147317,
+      "learning_rate": 1.5350714108220667e-05,
+      "loss": 0.0644,
+      "step": 2044
+    },
+    {
+      "epoch": 0.4099279423538831,
+      "learning_rate": 1.5362515102393217e-05,
+      "loss": 0.7223,
+      "step": 2046
+    },
+    {
+      "epoch": 0.4099279423538831,
+      "learning_rate": 1.5374305623457594e-05,
+      "loss": 0.0498,
+      "step": 2048
+    },
+    {
+      "epoch": 0.41072858286629305,
+      "learning_rate": 1.5386085648386656e-05,
+      "loss": 0.151,
+      "step": 2050
+    },
+    {
+      "epoch": 0.41072858286629305,
+      "learning_rate": 1.539785515417376e-05,
+      "loss": 0.2163,
+      "step": 2052
+    },
+    {
+      "epoch": 0.41152922337870296,
+      "learning_rate": 1.540961411783279e-05,
+      "loss": 0.3696,
+      "step": 2054
+    },
+    {
+      "epoch": 0.41152922337870296,
+      "learning_rate": 1.542136251639826e-05,
+      "loss": 0.4409,
+      "step": 2056
+    },
+    {
+      "epoch": 0.41232986389111287,
+      "learning_rate": 1.5433100326925288e-05,
+      "loss": 0.4636,
+      "step": 2058
+    },
+    {
+      "epoch": 0.41232986389111287,
+      "learning_rate": 1.5444827526489668e-05,
+      "loss": 0.4201,
+      "step": 2060
+    },
+    {
+      "epoch": 0.41313050440352284,
+      "learning_rate": 1.545654409218793e-05,
+      "loss": 0.1682,
+      "step": 2062
+    },
+    {
+      "epoch": 0.41313050440352284,
+      "learning_rate": 1.5468250001137368e-05,
+      "loss": 0.133,
+      "step": 2064
+    },
+    {
+      "epoch": 0.41393114491593275,
+      "learning_rate": 1.5479945230476066e-05,
+      "loss": 0.1679,
+      "step": 2066
+    },
+    {
+      "epoch": 0.41393114491593275,
+      "learning_rate": 1.5491629757363026e-05,
+      "loss": 0.0951,
+      "step": 2068
+    },
+    {
+      "epoch": 0.41473178542834266,
+      "learning_rate": 1.550330355897809e-05,
+      "loss": 0.1947,
+      "step": 2070
+    },
+    {
+      "epoch": 0.41473178542834266,
+      "learning_rate": 1.551496661252208e-05,
+      "loss": 0.3252,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4155324259407526,
+      "learning_rate": 1.5526618895216793e-05,
+      "loss": 0.3031,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4155324259407526,
+      "learning_rate": 1.5538260384305073e-05,
+      "loss": 0.6415,
+      "step": 2076
+    },
+    {
+      "epoch": 0.41633306645316254,
+      "learning_rate": 1.5549891057050837e-05,
+      "loss": 0.2067,
+      "step": 2078
+    },
+    {
+      "epoch": 0.41633306645316254,
+      "learning_rate": 1.5561510890739113e-05,
+      "loss": 0.0631,
+      "step": 2080
+    },
+    {
+      "epoch": 0.41713370696557245,
+      "learning_rate": 1.557311986267615e-05,
+      "loss": 0.5288,
+      "step": 2082
+    },
+    {
+      "epoch": 0.41713370696557245,
+      "learning_rate": 1.5584717950189353e-05,
+      "loss": 0.5654,
+      "step": 2084
+    },
+    {
+      "epoch": 0.41793434747798236,
+      "learning_rate": 1.5596305130627404e-05,
+      "loss": 0.3229,
+      "step": 2086
+    },
+    {
+      "epoch": 0.41793434747798236,
+      "learning_rate": 1.5607881381360296e-05,
+      "loss": 0.2014,
+      "step": 2088
+    },
+    {
+      "epoch": 0.4187349879903923,
+      "learning_rate": 1.5619446679779357e-05,
+      "loss": 0.2087,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4187349879903923,
+      "learning_rate": 1.563100100329731e-05,
+      "loss": 0.4383,
+      "step": 2092
+    },
+    {
+      "epoch": 0.41953562850280224,
+      "learning_rate": 1.564254432934829e-05,
+      "loss": 0.8945,
+      "step": 2094
+    },
+    {
+      "epoch": 0.41953562850280224,
+      "learning_rate": 1.565407663538797e-05,
+      "loss": 0.0433,
+      "step": 2096
+    },
+    {
+      "epoch": 0.42033626901521215,
+      "learning_rate": 1.5665597898893484e-05,
+      "loss": 0.4326,
+      "step": 2098
+    },
+    {
+      "epoch": 0.42033626901521215,
+      "learning_rate": 1.567710809736356e-05,
+      "loss": 0.2864,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4211369095276221,
+      "learning_rate": 1.568860720831853e-05,
+      "loss": 0.3651,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4211369095276221,
+      "learning_rate": 1.5700095209300376e-05,
+      "loss": 0.1667,
+      "step": 2104
+    },
+    {
+      "epoch": 0.42193755004003203,
+      "learning_rate": 1.5711572077872774e-05,
+      "loss": 0.3225,
+      "step": 2106
+    },
+    {
+      "epoch": 0.42193755004003203,
+      "learning_rate": 1.572303779162118e-05,
+      "loss": 0.2241,
+      "step": 2108
+    },
+    {
+      "epoch": 0.42273819055244194,
+      "learning_rate": 1.573449232815279e-05,
+      "loss": 0.3348,
+      "step": 2110
+    },
+    {
+      "epoch": 0.42273819055244194,
+      "learning_rate": 1.5745935665096647e-05,
+      "loss": 0.1496,
+      "step": 2112
+    },
+    {
+      "epoch": 0.4235388310648519,
+      "learning_rate": 1.5757367780103666e-05,
+      "loss": 0.2714,
+      "step": 2114
+    },
+    {
+      "epoch": 0.4235388310648519,
+      "learning_rate": 1.5768788650846677e-05,
+      "loss": 0.4636,
+      "step": 2116
+    },
+    {
+      "epoch": 0.4243394715772618,
+      "learning_rate": 1.5780198255020478e-05,
+      "loss": 0.397,
+      "step": 2118
+    },
+    {
+      "epoch": 0.4243394715772618,
+      "learning_rate": 1.5791596570341844e-05,
+      "loss": 0.1761,
+      "step": 2120
+    },
+    {
+      "epoch": 0.42514011208967173,
+      "learning_rate": 1.580298357454965e-05,
+      "loss": 0.2564,
+      "step": 2122
+    },
+    {
+      "epoch": 0.42514011208967173,
+      "learning_rate": 1.581435924540481e-05,
+      "loss": 0.4445,
+      "step": 2124
+    },
+    {
+      "epoch": 0.42594075260208164,
+      "learning_rate": 1.5825723560690403e-05,
+      "loss": 0.238,
+      "step": 2126
+    },
+    {
+      "epoch": 0.42594075260208164,
+      "learning_rate": 1.5837076498211666e-05,
+      "loss": 0.3033,
+      "step": 2128
+    },
+    {
+      "epoch": 0.4267413931144916,
+      "learning_rate": 1.5848418035796068e-05,
+      "loss": 0.2188,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4267413931144916,
+      "learning_rate": 1.5859748151293333e-05,
+      "loss": 0.3723,
+      "step": 2132
+    },
+    {
+      "epoch": 0.4275420336269015,
+      "learning_rate": 1.587106682257552e-05,
+      "loss": 0.2064,
+      "step": 2134
+    },
+    {
+      "epoch": 0.4275420336269015,
+      "learning_rate": 1.5882374027537005e-05,
+      "loss": 0.226,
+      "step": 2136
+    },
+    {
+      "epoch": 0.42834267413931143,
+      "learning_rate": 1.5893669744094577e-05,
+      "loss": 0.1439,
+      "step": 2138
+    },
+    {
+      "epoch": 0.42834267413931143,
+      "learning_rate": 1.5904953950187455e-05,
+      "loss": 0.0465,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4291433146517214,
+      "learning_rate": 1.591622662377734e-05,
+      "loss": 0.3633,
+      "step": 2142
+    },
+    {
+      "epoch": 0.4291433146517214,
+      "learning_rate": 1.5927487742848448e-05,
+      "loss": 0.1977,
+      "step": 2144
+    },
+    {
+      "epoch": 0.4299439551641313,
+      "learning_rate": 1.5938737285407567e-05,
+      "loss": 0.3223,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4299439551641313,
+      "learning_rate": 1.594997522948412e-05,
+      "loss": 0.8772,
+      "step": 2148
+    },
+    {
+      "epoch": 0.4307445956765412,
+      "learning_rate": 1.5961201553130148e-05,
+      "loss": 0.1245,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4307445956765412,
+      "learning_rate": 1.5972416234420393e-05,
+      "loss": 0.0522,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4315452361889512,
+      "learning_rate": 1.598361925145234e-05,
+      "loss": 0.1581,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4315452361889512,
+      "learning_rate": 1.599481058234626e-05,
+      "loss": 0.2856,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4323458767013611,
+      "learning_rate": 1.6005990205245216e-05,
+      "loss": 0.415,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4323458767013611,
+      "learning_rate": 1.60171580983152e-05,
+      "loss": 0.5578,
+      "step": 2160
+    },
+    {
+      "epoch": 0.433146517213771,
+      "learning_rate": 1.602831423974506e-05,
+      "loss": 0.1504,
+      "step": 2162
+    },
+    {
+      "epoch": 0.433146517213771,
+      "learning_rate": 1.6039458607746607e-05,
+      "loss": 0.4326,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4339471577261809,
+      "learning_rate": 1.6050591180554648e-05,
+      "loss": 0.1494,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4339471577261809,
+      "learning_rate": 1.606171193642703e-05,
+      "loss": 0.1186,
+      "step": 2168
+    },
+    {
+      "epoch": 0.4347477982385909,
+      "learning_rate": 1.6072820853644677e-05,
+      "loss": 0.1952,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4347477982385909,
+      "learning_rate": 1.6083917910511616e-05,
+      "loss": 0.5925,
+      "step": 2172
+    },
+    {
+      "epoch": 0.4355484387510008,
+      "learning_rate": 1.6095003085355082e-05,
+      "loss": 0.2538,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4355484387510008,
+      "learning_rate": 1.6106076356525474e-05,
+      "loss": 0.343,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4363490792634107,
+      "learning_rate": 1.611713770239646e-05,
+      "loss": 0.2065,
+      "step": 2178
+    },
+    {
+      "epoch": 0.4363490792634107,
+      "learning_rate": 1.6128187101364982e-05,
+      "loss": 0.5187,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4371497197758207,
+      "learning_rate": 1.6139224531851332e-05,
+      "loss": 0.1274,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4371497197758207,
+      "learning_rate": 1.6150249972299153e-05,
+      "loss": 0.0868,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4379503602882306,
+      "learning_rate": 1.616126340117555e-05,
+      "loss": 0.3844,
+      "step": 2186
+    },
+    {
+      "epoch": 0.4379503602882306,
+      "learning_rate": 1.617226479697104e-05,
+      "loss": 0.3451,
+      "step": 2188
+    },
+    {
+      "epoch": 0.4387510008006405,
+      "learning_rate": 1.618325413819966e-05,
+      "loss": 0.1991,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4387510008006405,
+      "learning_rate": 1.6194231403398994e-05,
+      "loss": 0.0863,
+      "step": 2192
+    },
+    {
+      "epoch": 0.43955164131305047,
+      "learning_rate": 1.6205196571130194e-05,
+      "loss": 0.2099,
+      "step": 2194
+    },
+    {
+      "epoch": 0.43955164131305047,
+      "learning_rate": 1.621614961997806e-05,
+      "loss": 0.6277,
+      "step": 2196
+    },
+    {
+      "epoch": 0.4403522818254604,
+      "learning_rate": 1.6227090528551034e-05,
+      "loss": 0.0712,
+      "step": 2198
+    },
+    {
+      "epoch": 0.4403522818254604,
+      "learning_rate": 1.6238019275481313e-05,
+      "loss": 0.0873,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4411529223378703,
+      "learning_rate": 1.62489358394248e-05,
+      "loss": 0.256,
+      "step": 2202
+    },
+    {
+      "epoch": 0.4411529223378703,
+      "learning_rate": 1.6259840199061212e-05,
+      "loss": 0.0645,
+      "step": 2204
+    },
+    {
+      "epoch": 0.4419535628502802,
+      "learning_rate": 1.6270732333094095e-05,
+      "loss": 0.2258,
+      "step": 2206
+    },
+    {
+      "epoch": 0.4419535628502802,
+      "learning_rate": 1.6281612220250883e-05,
+      "loss": 0.4092,
+      "step": 2208
+    },
+    {
+      "epoch": 0.44275420336269017,
+      "learning_rate": 1.6292479839282897e-05,
+      "loss": 0.2853,
+      "step": 2210
+    },
+    {
+      "epoch": 0.44275420336269017,
+      "learning_rate": 1.6303335168965474e-05,
+      "loss": 0.7046,
+      "step": 2212
+    },
+    {
+      "epoch": 0.4435548438751001,
+      "learning_rate": 1.6314178188097907e-05,
+      "loss": 0.5072,
+      "step": 2214
+    },
+    {
+      "epoch": 0.4435548438751001,
+      "learning_rate": 1.6325008875503543e-05,
+      "loss": 0.2541,
+      "step": 2216
+    },
+    {
+      "epoch": 0.44435548438751,
+      "learning_rate": 1.6335827210029816e-05,
+      "loss": 0.6665,
+      "step": 2218
+    },
+    {
+      "epoch": 0.44435548438751,
+      "learning_rate": 1.6346633170548285e-05,
+      "loss": 0.5315,
+      "step": 2220
+    },
+    {
+      "epoch": 0.44515612489991996,
+      "learning_rate": 1.635742673595467e-05,
+      "loss": 0.545,
+      "step": 2222
+    },
+    {
+      "epoch": 0.44515612489991996,
+      "learning_rate": 1.6368207885168897e-05,
+      "loss": 0.0607,
+      "step": 2224
+    },
+    {
+      "epoch": 0.44595676541232987,
+      "learning_rate": 1.6378976597135173e-05,
+      "loss": 0.1846,
+      "step": 2226
+    },
+    {
+      "epoch": 0.44595676541232987,
+      "learning_rate": 1.6389732850821957e-05,
+      "loss": 0.4225,
+      "step": 2228
+    },
+    {
+      "epoch": 0.4467574059247398,
+      "learning_rate": 1.640047662522205e-05,
+      "loss": 0.3091,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4467574059247398,
+      "learning_rate": 1.641120789935263e-05,
+      "loss": 0.2481,
+      "step": 2232
+    },
+    {
+      "epoch": 0.4475580464371497,
+      "learning_rate": 1.6421926652255282e-05,
+      "loss": 0.2102,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4475580464371497,
+      "learning_rate": 1.6432632862996042e-05,
+      "loss": 0.1568,
+      "step": 2236
+    },
+    {
+      "epoch": 0.44835868694955966,
+      "learning_rate": 1.6443326510665474e-05,
+      "loss": 0.2707,
+      "step": 2238
+    },
+    {
+      "epoch": 0.44835868694955966,
+      "learning_rate": 1.6454007574378637e-05,
+      "loss": 0.0835,
+      "step": 2240
+    },
+    {
+      "epoch": 0.44915932746196957,
+      "learning_rate": 1.646467603327518e-05,
+      "loss": 0.219,
+      "step": 2242
+    },
+    {
+      "epoch": 0.44915932746196957,
+      "learning_rate": 1.6475331866519377e-05,
+      "loss": 0.2712,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4499599679743795,
+      "learning_rate": 1.6485975053300154e-05,
+      "loss": 0.4228,
+      "step": 2246
+    },
+    {
+      "epoch": 0.4499599679743795,
+      "learning_rate": 1.6496605572831134e-05,
+      "loss": 0.4714,
+      "step": 2248
+    },
+    {
+      "epoch": 0.45076060848678945,
+      "learning_rate": 1.650722340435067e-05,
+      "loss": 0.5088,
+      "step": 2250
+    },
+    {
+      "epoch": 0.45076060848678945,
+      "learning_rate": 1.6517828527121928e-05,
+      "loss": 0.1424,
+      "step": 2252
+    },
+    {
+      "epoch": 0.45156124899919936,
+      "learning_rate": 1.652842092043287e-05,
+      "loss": 0.1763,
+      "step": 2254
+    },
+    {
+      "epoch": 0.45156124899919936,
+      "learning_rate": 1.6539000563596318e-05,
+      "loss": 0.1936,
+      "step": 2256
+    },
+    {
+      "epoch": 0.45236188951160927,
+      "learning_rate": 1.6549567435950004e-05,
+      "loss": 0.8804,
+      "step": 2258
+    },
+    {
+      "epoch": 0.45236188951160927,
+      "learning_rate": 1.6560121516856586e-05,
+      "loss": 0.5306,
+      "step": 2260
+    },
+    {
+      "epoch": 0.45316253002401924,
+      "learning_rate": 1.6570662785703713e-05,
+      "loss": 0.2049,
+      "step": 2262
+    },
+    {
+      "epoch": 0.45316253002401924,
+      "learning_rate": 1.6581191221904077e-05,
+      "loss": 0.8036,
+      "step": 2264
+    },
+    {
+      "epoch": 0.45396317053642915,
+      "learning_rate": 1.6591706804895408e-05,
+      "loss": 0.2407,
+      "step": 2266
+    },
+    {
+      "epoch": 0.45396317053642915,
+      "learning_rate": 1.6602209514140542e-05,
+      "loss": 0.1464,
+      "step": 2268
+    },
+    {
+      "epoch": 0.45476381104883906,
+      "learning_rate": 1.6612699329127457e-05,
+      "loss": 0.198,
+      "step": 2270
+    },
+    {
+      "epoch": 0.45476381104883906,
+      "learning_rate": 1.6623176229369324e-05,
+      "loss": 0.1756,
+      "step": 2272
+    },
+    {
+      "epoch": 0.455564451561249,
+      "learning_rate": 1.6633640194404523e-05,
+      "loss": 0.5481,
+      "step": 2274
+    },
+    {
+      "epoch": 0.455564451561249,
+      "learning_rate": 1.6644091203796694e-05,
+      "loss": 0.0634,
+      "step": 2276
+    },
+    {
+      "epoch": 0.45636509207365894,
+      "learning_rate": 1.6654529237134816e-05,
+      "loss": 0.1461,
+      "step": 2278
+    },
+    {
+      "epoch": 0.45636509207365894,
+      "learning_rate": 1.6664954274033168e-05,
+      "loss": 0.217,
+      "step": 2280
+    },
+    {
+      "epoch": 0.45716573258606885,
+      "learning_rate": 1.667536629413143e-05,
+      "loss": 0.1683,
+      "step": 2282
+    },
+    {
+      "epoch": 0.45716573258606885,
+      "learning_rate": 1.6685765277094695e-05,
+      "loss": 0.4839,
+      "step": 2284
+    },
+    {
+      "epoch": 0.45796637309847876,
+      "learning_rate": 1.6696151202613527e-05,
+      "loss": 0.1511,
+      "step": 2286
+    },
+    {
+      "epoch": 0.45796637309847876,
+      "learning_rate": 1.6706524050403996e-05,
+      "loss": 0.5398,
+      "step": 2288
+    },
+    {
+      "epoch": 0.45876701361088873,
+      "learning_rate": 1.6716883800207685e-05,
+      "loss": 0.2436,
+      "step": 2290
+    },
+    {
+      "epoch": 0.45876701361088873,
+      "learning_rate": 1.6727230431791806e-05,
+      "loss": 0.2422,
+      "step": 2292
+    },
+    {
+      "epoch": 0.45956765412329864,
+      "learning_rate": 1.673756392494915e-05,
+      "loss": 0.1677,
+      "step": 2294
+    },
+    {
+      "epoch": 0.45956765412329864,
+      "learning_rate": 1.674788425949818e-05,
+      "loss": 0.1576,
+      "step": 2296
+    },
+    {
+      "epoch": 0.46036829463570855,
+      "learning_rate": 1.6758191415283063e-05,
+      "loss": 0.4196,
+      "step": 2298
+    },
+    {
+      "epoch": 0.46036829463570855,
+      "learning_rate": 1.6768485372173696e-05,
+      "loss": 0.1937,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4611689351481185,
+      "learning_rate": 1.6778766110065755e-05,
+      "loss": 0.1729,
+      "step": 2302
+    },
+    {
+      "epoch": 0.4611689351481185,
+      "learning_rate": 1.6789033608880735e-05,
+      "loss": 0.2618,
+      "step": 2304
+    },
+    {
+      "epoch": 0.46196957566052843,
+      "learning_rate": 1.6799287848566e-05,
+      "loss": 0.5152,
+      "step": 2306
+    },
+    {
+      "epoch": 0.46196957566052843,
+      "learning_rate": 1.6809528809094798e-05,
+      "loss": 0.6898,
+      "step": 2308
+    },
+    {
+      "epoch": 0.46277021617293834,
+      "learning_rate": 1.6819756470466305e-05,
+      "loss": 0.4229,
+      "step": 2310
+    },
+    {
+      "epoch": 0.46277021617293834,
+      "learning_rate": 1.6829970812705674e-05,
+      "loss": 0.2503,
+      "step": 2312
+    },
+    {
+      "epoch": 0.46357085668534825,
+      "learning_rate": 1.684017181586408e-05,
+      "loss": 0.1583,
+      "step": 2314
+    },
+    {
+      "epoch": 0.46357085668534825,
+      "learning_rate": 1.6850359460018733e-05,
+      "loss": 0.2153,
+      "step": 2316
+    },
+    {
+      "epoch": 0.4643714971977582,
+      "learning_rate": 1.6860533725272943e-05,
+      "loss": 0.1977,
+      "step": 2318
+    },
+    {
+      "epoch": 0.4643714971977582,
+      "learning_rate": 1.6870694591756165e-05,
+      "loss": 0.1865,
+      "step": 2320
+    },
+    {
+      "epoch": 0.46517213771016813,
+      "learning_rate": 1.6880842039624e-05,
+      "loss": 0.1416,
+      "step": 2322
+    },
+    {
+      "epoch": 0.46517213771016813,
+      "learning_rate": 1.689097604905826e-05,
+      "loss": 0.3322,
+      "step": 2324
+    },
+    {
+      "epoch": 0.46597277822257804,
+      "learning_rate": 1.6901096600267e-05,
+      "loss": 0.202,
+      "step": 2326
+    },
+    {
+      "epoch": 0.46597277822257804,
+      "learning_rate": 1.6911203673484577e-05,
+      "loss": 0.206,
+      "step": 2328
+    },
+    {
+      "epoch": 0.466773418734988,
+      "learning_rate": 1.6921297248971645e-05,
+      "loss": 0.1342,
+      "step": 2330
+    },
+    {
+      "epoch": 0.466773418734988,
+      "learning_rate": 1.6931377307015226e-05,
+      "loss": 0.2286,
+      "step": 2332
+    },
+    {
+      "epoch": 0.4675740592473979,
+      "learning_rate": 1.6941443827928778e-05,
+      "loss": 0.3483,
+      "step": 2334
+    },
+    {
+      "epoch": 0.4675740592473979,
+      "learning_rate": 1.695149679205214e-05,
+      "loss": 0.0824,
+      "step": 2336
+    },
+    {
+      "epoch": 0.46837469975980783,
+      "learning_rate": 1.6961536179751672e-05,
+      "loss": 0.5755,
+      "step": 2338
+    },
+    {
+      "epoch": 0.46837469975980783,
+      "learning_rate": 1.6971561971420222e-05,
+      "loss": 0.1356,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4691753402722178,
+      "learning_rate": 1.6981574147477204e-05,
+      "loss": 0.3308,
+      "step": 2342
+    },
+    {
+      "epoch": 0.4691753402722178,
+      "learning_rate": 1.6991572688368628e-05,
+      "loss": 0.255,
+      "step": 2344
+    },
+    {
+      "epoch": 0.4699759807846277,
+      "learning_rate": 1.70015575745671e-05,
+      "loss": 0.8065,
+      "step": 2346
+    },
+    {
+      "epoch": 0.4699759807846277,
+      "learning_rate": 1.701152878657196e-05,
+      "loss": 0.3164,
+      "step": 2348
+    },
+    {
+      "epoch": 0.4707766212970376,
+      "learning_rate": 1.7021486304909196e-05,
+      "loss": 0.2188,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4707766212970376,
+      "learning_rate": 1.7031430110131562e-05,
+      "loss": 0.4395,
+      "step": 2352
+    },
+    {
+      "epoch": 0.47157726180944753,
+      "learning_rate": 1.7041360182818583e-05,
+      "loss": 0.1784,
+      "step": 2354
+    },
+    {
+      "epoch": 0.47157726180944753,
+      "learning_rate": 1.705127650357662e-05,
+      "loss": 0.246,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4723779023218575,
+      "learning_rate": 1.7061179053038887e-05,
+      "loss": 0.6699,
+      "step": 2358
+    },
+    {
+      "epoch": 0.4723779023218575,
+      "learning_rate": 1.7071067811865467e-05,
+      "loss": 0.4094,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4731785428342674,
+      "learning_rate": 1.708094276074343e-05,
+      "loss": 0.9815,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4731785428342674,
+      "learning_rate": 1.7090803880386778e-05,
+      "loss": 0.0879,
+      "step": 2364
+    },
+    {
+      "epoch": 0.4739791833466773,
+      "learning_rate": 1.7100651151536525e-05,
+      "loss": 0.1681,
+      "step": 2366
+    },
+    {
+      "epoch": 0.4739791833466773,
+      "learning_rate": 1.7110484554960738e-05,
+      "loss": 0.2034,
+      "step": 2368
+    },
+    {
+      "epoch": 0.4747798238590873,
+      "learning_rate": 1.712030407145457e-05,
+      "loss": 0.4398,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4747798238590873,
+      "learning_rate": 1.713010968184029e-05,
+      "loss": 0.1583,
+      "step": 2372
+    },
+    {
+      "epoch": 0.4755804643714972,
+      "learning_rate": 1.7139901366967332e-05,
+      "loss": 0.2618,
+      "step": 2374
+    },
+    {
+      "epoch": 0.4755804643714972,
+      "learning_rate": 1.7149679107712306e-05,
+      "loss": 0.0996,
+      "step": 2376
+    },
+    {
+      "epoch": 0.4763811048839071,
+      "learning_rate": 1.71594428849791e-05,
+      "loss": 0.3188,
+      "step": 2378
+    },
+    {
+      "epoch": 0.4763811048839071,
+      "learning_rate": 1.716919267969883e-05,
+      "loss": 0.3078,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4771817453963171,
+      "learning_rate": 1.717892847282994e-05,
+      "loss": 0.298,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4771817453963171,
+      "learning_rate": 1.7188650245358215e-05,
+      "loss": 0.4512,
+      "step": 2384
+    },
+    {
+      "epoch": 0.477982385908727,
+      "learning_rate": 1.7198357978296817e-05,
+      "loss": 0.2024,
+      "step": 2386
+    },
+    {
+      "epoch": 0.477982385908727,
+      "learning_rate": 1.7208051652686338e-05,
+      "loss": 0.1444,
+      "step": 2388
+    },
+    {
+      "epoch": 0.4787830264211369,
+      "learning_rate": 1.721773124959481e-05,
+      "loss": 0.3035,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4787830264211369,
+      "learning_rate": 1.722739675011779e-05,
+      "loss": 0.1482,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4795836669335468,
+      "learning_rate": 1.723704813537834e-05,
+      "loss": 0.2527,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4795836669335468,
+      "learning_rate": 1.7246685386527095e-05,
+      "loss": 0.2582,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4803843074459568,
+      "learning_rate": 1.725630848474229e-05,
+      "loss": 0.5233,
+      "step": 2398
+    },
+    {
+      "epoch": 0.4803843074459568,
+      "learning_rate": 1.726591741122981e-05,
+      "loss": 0.0761,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4811849479583667,
+      "learning_rate": 1.727551214722321e-05,
+      "loss": 0.232,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4811849479583667,
+      "learning_rate": 1.7285092673983753e-05,
+      "loss": 0.74,
+      "step": 2404
+    },
+    {
+      "epoch": 0.4819855884707766,
+      "learning_rate": 1.7294658972800488e-05,
+      "loss": 0.2872,
+      "step": 2406
+    },
+    {
+      "epoch": 0.4819855884707766,
+      "learning_rate": 1.730421102499021e-05,
+      "loss": 0.1952,
+      "step": 2408
+    },
+    {
+      "epoch": 0.48278622898318657,
+      "learning_rate": 1.7313748811897558e-05,
+      "loss": 0.4487,
+      "step": 2410
+    },
+    {
+      "epoch": 0.48278622898318657,
+      "learning_rate": 1.7323272314895022e-05,
+      "loss": 0.1421,
+      "step": 2412
+    },
+    {
+      "epoch": 0.4835868694955965,
+      "learning_rate": 1.7332781515383003e-05,
+      "loss": 0.2526,
+      "step": 2414
+    },
+    {
+      "epoch": 0.4835868694955965,
+      "learning_rate": 1.734227639478982e-05,
+      "loss": 0.1204,
+      "step": 2416
+    },
+    {
+      "epoch": 0.4843875100080064,
+      "learning_rate": 1.7351756934571758e-05,
+      "loss": 0.3345,
+      "step": 2418
+    },
+    {
+      "epoch": 0.4843875100080064,
+      "learning_rate": 1.736122311621314e-05,
+      "loss": 0.3129,
+      "step": 2420
+    },
+    {
+      "epoch": 0.4851881505204163,
+      "learning_rate": 1.7370674921226296e-05,
+      "loss": 0.4167,
+      "step": 2422
+    },
+    {
+      "epoch": 0.4851881505204163,
+      "learning_rate": 1.738011233115165e-05,
+      "loss": 0.4551,
+      "step": 2424
+    },
+    {
+      "epoch": 0.4859887910328263,
+      "learning_rate": 1.7389535327557733e-05,
+      "loss": 0.554,
+      "step": 2426
+    },
+    {
+      "epoch": 0.4859887910328263,
+      "learning_rate": 1.7398943892041227e-05,
+      "loss": 0.4983,
+      "step": 2428
+    },
+    {
+      "epoch": 0.4867894315452362,
+      "learning_rate": 1.7408338006227005e-05,
+      "loss": 0.4439,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4867894315452362,
+      "learning_rate": 1.7417717651768144e-05,
+      "loss": 0.1658,
+      "step": 2432
+    },
+    {
+      "epoch": 0.4875900720576461,
+      "learning_rate": 1.7427082810346018e-05,
+      "loss": 0.2466,
+      "step": 2434
+    },
+    {
+      "epoch": 0.4875900720576461,
+      "learning_rate": 1.743643346367026e-05,
+      "loss": 0.2512,
+      "step": 2436
+    },
+    {
+      "epoch": 0.48839071257005606,
+      "learning_rate": 1.744576959347884e-05,
+      "loss": 0.238,
+      "step": 2438
+    },
+    {
+      "epoch": 0.48839071257005606,
+      "learning_rate": 1.7455091181538087e-05,
+      "loss": 0.2271,
+      "step": 2440
+    },
+    {
+      "epoch": 0.489191353082466,
+      "learning_rate": 1.746439820964275e-05,
+      "loss": 0.1682,
+      "step": 2442
+    },
+    {
+      "epoch": 0.489191353082466,
+      "learning_rate": 1.7473690659615992e-05,
+      "loss": 0.1304,
+      "step": 2444
+    },
+    {
+      "epoch": 0.4899919935948759,
+      "learning_rate": 1.748296851330945e-05,
+      "loss": 0.205,
+      "step": 2446
+    },
+    {
+      "epoch": 0.4899919935948759,
+      "learning_rate": 1.74922317526033e-05,
+      "loss": 0.1145,
+      "step": 2448
+    },
+    {
+      "epoch": 0.49079263410728585,
+      "learning_rate": 1.7501480359406217e-05,
+      "loss": 0.1598,
+      "step": 2450
+    },
+    {
+      "epoch": 0.49079263410728585,
+      "learning_rate": 1.7510714315655474e-05,
+      "loss": 0.1798,
+      "step": 2452
+    },
+    {
+      "epoch": 0.49159327461969576,
+      "learning_rate": 1.7519933603316955e-05,
+      "loss": 0.4794,
+      "step": 2454
+    },
+    {
+      "epoch": 0.49159327461969576,
+      "learning_rate": 1.752913820438519e-05,
+      "loss": 0.0358,
+      "step": 2456
+    },
+    {
+      "epoch": 0.4923939151321057,
+      "learning_rate": 1.7538328100883397e-05,
+      "loss": 0.5099,
+      "step": 2458
+    },
+    {
+      "epoch": 0.4923939151321057,
+      "learning_rate": 1.7547503274863495e-05,
+      "loss": 0.4349,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4931945556445156,
+      "learning_rate": 1.7556663708406193e-05,
+      "loss": 0.1685,
+      "step": 2462
+    },
+    {
+      "epoch": 0.4931945556445156,
+      "learning_rate": 1.756580938362096e-05,
+      "loss": 0.202,
+      "step": 2464
+    },
+    {
+      "epoch": 0.49399519615692555,
+      "learning_rate": 1.7574940282646085e-05,
+      "loss": 0.4309,
+      "step": 2466
+    },
+    {
+      "epoch": 0.49399519615692555,
+      "learning_rate": 1.758405638764873e-05,
+      "loss": 0.5622,
+      "step": 2468
+    },
+    {
+      "epoch": 0.49479583666933546,
+      "learning_rate": 1.7593157680824946e-05,
+      "loss": 0.1961,
+      "step": 2470
+    },
+    {
+      "epoch": 0.49479583666933546,
+      "learning_rate": 1.7602244144399693e-05,
+      "loss": 0.1869,
+      "step": 2472
+    },
+    {
+      "epoch": 0.4955964771817454,
+      "learning_rate": 1.761131576062694e-05,
+      "loss": 0.1668,
+      "step": 2474
+    },
+    {
+      "epoch": 0.4955964771817454,
+      "learning_rate": 1.7620372511789604e-05,
+      "loss": 0.2749,
+      "step": 2476
+    },
+    {
+      "epoch": 0.49639711769415534,
+      "learning_rate": 1.7629414380199662e-05,
+      "loss": 0.2235,
+      "step": 2478
+    },
+    {
+      "epoch": 0.49639711769415534,
+      "learning_rate": 1.7638441348198147e-05,
+      "loss": 0.1031,
+      "step": 2480
+    },
+    {
+      "epoch": 0.49719775820656525,
+      "learning_rate": 1.7647453398155194e-05,
+      "loss": 0.3745,
+      "step": 2482
+    },
+    {
+      "epoch": 0.49719775820656525,
+      "learning_rate": 1.7656450512470077e-05,
+      "loss": 0.1008,
+      "step": 2484
+    },
+    {
+      "epoch": 0.49799839871897517,
+      "learning_rate": 1.7665432673571218e-05,
+      "loss": 0.7476,
+      "step": 2486
+    },
+    {
+      "epoch": 0.49799839871897517,
+      "learning_rate": 1.7674399863916295e-05,
+      "loss": 1.0966,
+      "step": 2488
+    },
+    {
+      "epoch": 0.49879903923138513,
+      "learning_rate": 1.768335206599217e-05,
+      "loss": 0.775,
+      "step": 2490
+    },
+    {
+      "epoch": 0.49879903923138513,
+      "learning_rate": 1.7692289262315e-05,
+      "loss": 0.2311,
+      "step": 2492
+    },
+    {
+      "epoch": 0.49959967974379504,
+      "learning_rate": 1.7701211435430256e-05,
+      "loss": 0.391,
+      "step": 2494
+    },
+    {
+      "epoch": 0.49959967974379504,
+      "learning_rate": 1.771011856791273e-05,
+      "loss": 0.5734,
+      "step": 2496
+    },
+    {
+      "epoch": 0.500400320256205,
+      "learning_rate": 1.771901064236659e-05,
+      "loss": 0.2815,
+      "step": 2498
+    },
+    {
+      "epoch": 0.500400320256205,
+      "learning_rate": 1.7727887641425448e-05,
+      "loss": 0.2396,
+      "step": 2500
+    },
+    {
+      "epoch": 0.5012009607686149,
+      "learning_rate": 1.773674954775232e-05,
+      "loss": 0.1344,
+      "step": 2502
+    },
+    {
+      "epoch": 0.5012009607686149,
+      "learning_rate": 1.7745596344039712e-05,
+      "loss": 0.3452,
+      "step": 2504
+    },
+    {
+      "epoch": 0.5020016012810248,
+      "learning_rate": 1.7754428013009637e-05,
+      "loss": 0.1867,
+      "step": 2506
+    },
+    {
+      "epoch": 0.5020016012810248,
+      "learning_rate": 1.7763244537413657e-05,
+      "loss": 0.3293,
+      "step": 2508
+    },
+    {
+      "epoch": 0.5028022417934348,
+      "learning_rate": 1.77720459000329e-05,
+      "loss": 0.2863,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5028022417934348,
+      "learning_rate": 1.7780832083678116e-05,
+      "loss": 0.3385,
+      "step": 2512
+    },
+    {
+      "epoch": 0.5036028823058447,
+      "learning_rate": 1.7789603071189712e-05,
+      "loss": 0.3413,
+      "step": 2514
+    },
+    {
+      "epoch": 0.5036028823058447,
+      "learning_rate": 1.7798358845437754e-05,
+      "loss": 0.3642,
+      "step": 2516
+    },
+    {
+      "epoch": 0.5044035228182546,
+      "learning_rate": 1.780709938932202e-05,
+      "loss": 0.2378,
+      "step": 2518
+    },
+    {
+      "epoch": 0.5044035228182546,
+      "learning_rate": 1.7815824685772035e-05,
+      "loss": 0.2243,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5052041633306645,
+      "learning_rate": 1.7824534717747115e-05,
+      "loss": 0.1958,
+      "step": 2522
+    },
+    {
+      "epoch": 0.5052041633306645,
+      "learning_rate": 1.7833229468236364e-05,
+      "loss": 0.4071,
+      "step": 2524
+    },
+    {
+      "epoch": 0.5060048038430744,
+      "learning_rate": 1.7841908920258767e-05,
+      "loss": 0.5398,
+      "step": 2526
+    },
+    {
+      "epoch": 0.5060048038430744,
+      "learning_rate": 1.7850573056863156e-05,
+      "loss": 0.3774,
+      "step": 2528
+    },
+    {
+      "epoch": 0.5068054443554844,
+      "learning_rate": 1.7859221861128284e-05,
+      "loss": 0.1713,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5068054443554844,
+      "learning_rate": 1.786785531616285e-05,
+      "loss": 0.535,
+      "step": 2532
+    },
+    {
+      "epoch": 0.5076060848678943,
+      "learning_rate": 1.7876473405105528e-05,
+      "loss": 0.3927,
+      "step": 2534
+    },
+    {
+      "epoch": 0.5076060848678943,
+      "learning_rate": 1.7885076111125004e-05,
+      "loss": 0.1844,
+      "step": 2536
+    },
+    {
+      "epoch": 0.5084067253803043,
+      "learning_rate": 1.7893663417419995e-05,
+      "loss": 0.3155,
+      "step": 2538
+    },
+    {
+      "epoch": 0.5084067253803043,
+      "learning_rate": 1.790223530721933e-05,
+      "loss": 0.2506,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5092073658927142,
+      "learning_rate": 1.791079176378191e-05,
+      "loss": 0.2709,
+      "step": 2542
+    },
+    {
+      "epoch": 0.5092073658927142,
+      "learning_rate": 1.791933277039679e-05,
+      "loss": 0.349,
+      "step": 2544
+    },
+    {
+      "epoch": 0.5100080064051241,
+      "learning_rate": 1.7927858310383202e-05,
+      "loss": 0.3345,
+      "step": 2546
+    },
+    {
+      "epoch": 0.5100080064051241,
+      "learning_rate": 1.7936368367090577e-05,
+      "loss": 0.1962,
+      "step": 2548
+    },
+    {
+      "epoch": 0.510808646917534,
+      "learning_rate": 1.794486292389858e-05,
+      "loss": 0.3233,
+      "step": 2550
+    },
+    {
+      "epoch": 0.510808646917534,
+      "learning_rate": 1.7953341964217183e-05,
+      "loss": 0.2564,
+      "step": 2552
+    },
+    {
+      "epoch": 0.5116092874299439,
+      "learning_rate": 1.7961805471486618e-05,
+      "loss": 0.2419,
+      "step": 2554
+    },
+    {
+      "epoch": 0.5116092874299439,
+      "learning_rate": 1.7970253429177477e-05,
+      "loss": 0.4263,
+      "step": 2556
+    },
+    {
+      "epoch": 0.5124099279423538,
+      "learning_rate": 1.797868582079072e-05,
+      "loss": 0.5227,
+      "step": 2558
+    },
+    {
+      "epoch": 0.5124099279423538,
+      "learning_rate": 1.7987102629857696e-05,
+      "loss": 0.091,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5132105684547638,
+      "learning_rate": 1.7995503839940197e-05,
+      "loss": 0.3153,
+      "step": 2562
+    },
+    {
+      "epoch": 0.5132105684547638,
+      "learning_rate": 1.800388943463047e-05,
+      "loss": 0.0395,
+      "step": 2564
+    },
+    {
+      "epoch": 0.5140112089671738,
+      "learning_rate": 1.8012259397551283e-05,
+      "loss": 0.1586,
+      "step": 2566
+    },
+    {
+      "epoch": 0.5140112089671738,
+      "learning_rate": 1.8020613712355912e-05,
+      "loss": 0.1021,
+      "step": 2568
+    },
+    {
+      "epoch": 0.5148118494795837,
+      "learning_rate": 1.8028952362728197e-05,
+      "loss": 0.0406,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5148118494795837,
+      "learning_rate": 1.803727533238257e-05,
+      "loss": 0.0778,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5156124899919936,
+      "learning_rate": 1.804558260506409e-05,
+      "loss": 0.194,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5156124899919936,
+      "learning_rate": 1.805387416454847e-05,
+      "loss": 0.1681,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5164131305044035,
+      "learning_rate": 1.8062149994642135e-05,
+      "loss": 0.1868,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5164131305044035,
+      "learning_rate": 1.8070410079182195e-05,
+      "loss": 0.0327,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5172137710168134,
+      "learning_rate": 1.8078654402036526e-05,
+      "loss": 0.1116,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5172137710168134,
+      "learning_rate": 1.8086882947103787e-05,
+      "loss": 0.0425,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5180144115292233,
+      "learning_rate": 1.8095095698313452e-05,
+      "loss": 0.1516,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5180144115292233,
+      "learning_rate": 1.8103292639625842e-05,
+      "loss": 0.1115,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5188150520416333,
+      "learning_rate": 1.811147375503214e-05,
+      "loss": 1.0942,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5188150520416333,
+      "learning_rate": 1.811963902855447e-05,
+      "loss": 0.0614,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5196156925540433,
+      "learning_rate": 1.812778844424587e-05,
+      "loss": 0.2463,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5196156925540433,
+      "learning_rate": 1.813592198619035e-05,
+      "loss": 0.472,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5204163330664532,
+      "learning_rate": 1.814403963850293e-05,
+      "loss": 0.1681,
+      "step": 2598
+    },
+    {
+      "epoch": 0.5204163330664532,
+      "learning_rate": 1.8152141385329658e-05,
+      "loss": 0.1759,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5212169735788631,
+      "learning_rate": 1.8160227210847636e-05,
+      "loss": 0.144,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5212169735788631,
+      "learning_rate": 1.816829709926509e-05,
+      "loss": 0.1328,
+      "step": 2604
+    },
+    {
+      "epoch": 0.522017614091273,
+      "learning_rate": 1.8176351034821345e-05,
+      "loss": 0.3637,
+      "step": 2606
+    },
+    {
+      "epoch": 0.522017614091273,
+      "learning_rate": 1.8184389001786895e-05,
+      "loss": 0.3628,
+      "step": 2608
+    },
+    {
+      "epoch": 0.5228182546036829,
+      "learning_rate": 1.819241098446341e-05,
+      "loss": 0.5551,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5228182546036829,
+      "learning_rate": 1.8200416967183785e-05,
+      "loss": 0.1671,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5236188951160928,
+      "learning_rate": 1.8208406934312167e-05,
+      "loss": 0.3377,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5236188951160928,
+      "learning_rate": 1.821638087024396e-05,
+      "loss": 0.1464,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5244195356285029,
+      "learning_rate": 1.8224338759405917e-05,
+      "loss": 0.733,
+      "step": 2618
+    },
+    {
+      "epoch": 0.5244195356285029,
+      "learning_rate": 1.8232280586256097e-05,
+      "loss": 0.1686,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5252201761409128,
+      "learning_rate": 1.8240206335283947e-05,
+      "loss": 0.29,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5252201761409128,
+      "learning_rate": 1.8248115991010296e-05,
+      "loss": 0.362,
+      "step": 2624
+    },
+    {
+      "epoch": 0.5260208166533227,
+      "learning_rate": 1.825600953798743e-05,
+      "loss": 0.4373,
+      "step": 2626
+    },
+    {
+      "epoch": 0.5260208166533227,
+      "learning_rate": 1.8263886960799055e-05,
+      "loss": 0.038,
+      "step": 2628
+    },
+    {
+      "epoch": 0.5268214571657326,
+      "learning_rate": 1.8271748244060426e-05,
+      "loss": 0.2377,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5268214571657326,
+      "learning_rate": 1.8279593372418264e-05,
+      "loss": 0.2393,
+      "step": 2632
+    },
+    {
+      "epoch": 0.5276220976781425,
+      "learning_rate": 1.8287422330550878e-05,
+      "loss": 0.6519,
+      "step": 2634
+    },
+    {
+      "epoch": 0.5276220976781425,
+      "learning_rate": 1.829523510316813e-05,
+      "loss": 0.4658,
+      "step": 2636
+    },
+    {
+      "epoch": 0.5284227381905524,
+      "learning_rate": 1.8303031675011515e-05,
+      "loss": 0.3502,
+      "step": 2638
+    },
+    {
+      "epoch": 0.5284227381905524,
+      "learning_rate": 1.8310812030854155e-05,
+      "loss": 0.3361,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5292233787029623,
+      "learning_rate": 1.8318576155500838e-05,
+      "loss": 0.2481,
+      "step": 2642
+    },
+    {
+      "epoch": 0.5292233787029623,
+      "learning_rate": 1.832632403378808e-05,
+      "loss": 0.3521,
+      "step": 2644
+    },
+    {
+      "epoch": 0.5300240192153723,
+      "learning_rate": 1.8334055650584094e-05,
+      "loss": 0.1969,
+      "step": 2646
+    },
+    {
+      "epoch": 0.5300240192153723,
+      "learning_rate": 1.834177099078887e-05,
+      "loss": 0.1755,
+      "step": 2648
+    },
+    {
+      "epoch": 0.5308246597277823,
+      "learning_rate": 1.8349470039334173e-05,
+      "loss": 0.5506,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5308246597277823,
+      "learning_rate": 1.8357152781183606e-05,
+      "loss": 0.2311,
+      "step": 2652
+    },
+    {
+      "epoch": 0.5316253002401922,
+      "learning_rate": 1.83648192013326e-05,
+      "loss": 0.2634,
+      "step": 2654
+    },
+    {
+      "epoch": 0.5316253002401922,
+      "learning_rate": 1.8372469284808465e-05,
+      "loss": 0.3706,
+      "step": 2656
+    },
+    {
+      "epoch": 0.5324259407526021,
+      "learning_rate": 1.8380103016670437e-05,
+      "loss": 0.5143,
+      "step": 2658
+    },
+    {
+      "epoch": 0.5324259407526021,
+      "learning_rate": 1.8387720382009665e-05,
+      "loss": 0.1756,
+      "step": 2660
+    },
+    {
+      "epoch": 0.533226581265012,
+      "learning_rate": 1.839532136594927e-05,
+      "loss": 0.2278,
+      "step": 2662
+    },
+    {
+      "epoch": 0.533226581265012,
+      "learning_rate": 1.840290595364436e-05,
+      "loss": 0.1966,
+      "step": 2664
+    },
+    {
+      "epoch": 0.5340272217774219,
+      "learning_rate": 1.8410474130282085e-05,
+      "loss": 0.5232,
+      "step": 2666
+    },
+    {
+      "epoch": 0.5340272217774219,
+      "learning_rate": 1.8418025881081606e-05,
+      "loss": 0.2224,
+      "step": 2668
+    },
+    {
+      "epoch": 0.5348278622898318,
+      "learning_rate": 1.8425561191294217e-05,
+      "loss": 0.1512,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5348278622898318,
+      "learning_rate": 1.8433080046203286e-05,
+      "loss": 0.2844,
+      "step": 2672
+    },
+    {
+      "epoch": 0.5356285028022418,
+      "learning_rate": 1.8440582431124325e-05,
+      "loss": 0.1787,
+      "step": 2674
+    },
+    {
+      "epoch": 0.5356285028022418,
+      "learning_rate": 1.844806833140501e-05,
+      "loss": 0.2778,
+      "step": 2676
+    },
+    {
+      "epoch": 0.5364291433146517,
+      "learning_rate": 1.8455537732425223e-05,
+      "loss": 0.2382,
+      "step": 2678
+    },
+    {
+      "epoch": 0.5364291433146517,
+      "learning_rate": 1.8462990619597054e-05,
+      "loss": 0.1623,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5372297838270617,
+      "learning_rate": 1.847042697836485e-05,
+      "loss": 0.2065,
+      "step": 2682
+    },
+    {
+      "epoch": 0.5372297838270617,
+      "learning_rate": 1.8477846794205258e-05,
+      "loss": 0.1673,
+      "step": 2684
+    },
+    {
+      "epoch": 0.5380304243394716,
+      "learning_rate": 1.84852500526272e-05,
+      "loss": 0.4928,
+      "step": 2686
+    },
+    {
+      "epoch": 0.5380304243394716,
+      "learning_rate": 1.8492636739171966e-05,
+      "loss": 0.1676,
+      "step": 2688
+    },
+    {
+      "epoch": 0.5388310648518815,
+      "learning_rate": 1.8500006839413183e-05,
+      "loss": 0.1712,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5388310648518815,
+      "learning_rate": 1.85073603389569e-05,
+      "loss": 0.0793,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5396317053642914,
+      "learning_rate": 1.851469722344155e-05,
+      "loss": 0.1865,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5396317053642914,
+      "learning_rate": 1.8522017478538067e-05,
+      "loss": 0.1241,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5404323458767014,
+      "learning_rate": 1.8529321089949817e-05,
+      "loss": 1.0657,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5404323458767014,
+      "learning_rate": 1.8536608043412695e-05,
+      "loss": 0.3722,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5412329863891113,
+      "learning_rate": 1.8543878324695122e-05,
+      "loss": 0.2222,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5412329863891113,
+      "learning_rate": 1.855113191959808e-05,
+      "loss": 0.1297,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5420336269015212,
+      "learning_rate": 1.8558368813955143e-05,
+      "loss": 0.1843,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5420336269015212,
+      "learning_rate": 1.856558899363248e-05,
+      "loss": 0.0395,
+      "step": 2708
+    },
+    {
+      "epoch": 0.5428342674139311,
+      "learning_rate": 1.857279244452896e-05,
+      "loss": 0.177,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5428342674139311,
+      "learning_rate": 1.8579979152576063e-05,
+      "loss": 0.1148,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5436349079263411,
+      "learning_rate": 1.8587149103738e-05,
+      "loss": 0.5083,
+      "step": 2714
+    },
+    {
+      "epoch": 0.5436349079263411,
+      "learning_rate": 1.85943022840117e-05,
+      "loss": 0.0513,
+      "step": 2716
+    },
+    {
+      "epoch": 0.544435548438751,
+      "learning_rate": 1.8601438679426847e-05,
+      "loss": 0.5416,
+      "step": 2718
+    },
+    {
+      "epoch": 0.544435548438751,
+      "learning_rate": 1.8608558276045895e-05,
+      "loss": 0.1071,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5452361889511609,
+      "learning_rate": 1.8615661059964134e-05,
+      "loss": 0.3568,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5452361889511609,
+      "learning_rate": 1.862274701730967e-05,
+      "loss": 0.2019,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5460368294635709,
+      "learning_rate": 1.862981613424347e-05,
+      "loss": 0.2101,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5460368294635709,
+      "learning_rate": 1.86368683969594e-05,
+      "loss": 0.1669,
+      "step": 2728
+    },
+    {
+      "epoch": 0.5468374699759808,
+      "learning_rate": 1.864390379168423e-05,
+      "loss": 0.3327,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5468374699759808,
+      "learning_rate": 1.865092230467769e-05,
+      "loss": 0.4152,
+      "step": 2732
+    },
+    {
+      "epoch": 0.5476381104883907,
+      "learning_rate": 1.8657923922232464e-05,
+      "loss": 0.6162,
+      "step": 2734
+    },
+    {
+      "epoch": 0.5476381104883907,
+      "learning_rate": 1.866490863067425e-05,
+      "loss": 0.1787,
+      "step": 2736
+    },
+    {
+      "epoch": 0.5484387510008006,
+      "learning_rate": 1.8671876416361763e-05,
+      "loss": 0.1753,
+      "step": 2738
+    },
+    {
+      "epoch": 0.5484387510008006,
+      "learning_rate": 1.8678827265686753e-05,
+      "loss": 0.1863,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5492393915132106,
+      "learning_rate": 1.8685761165074073e-05,
+      "loss": 0.2162,
+      "step": 2742
+    },
+    {
+      "epoch": 0.5492393915132106,
+      "learning_rate": 1.8692678100981663e-05,
+      "loss": 0.3516,
+      "step": 2744
+    },
+    {
+      "epoch": 0.5500400320256205,
+      "learning_rate": 1.869957805990059e-05,
+      "loss": 0.4707,
+      "step": 2746
+    },
+    {
+      "epoch": 0.5500400320256205,
+      "learning_rate": 1.87064610283551e-05,
+      "loss": 0.4434,
+      "step": 2748
+    },
+    {
+      "epoch": 0.5508406725380304,
+      "learning_rate": 1.87133269929026e-05,
+      "loss": 0.3514,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5508406725380304,
+      "learning_rate": 1.8720175940133705e-05,
+      "loss": 0.4731,
+      "step": 2752
+    },
+    {
+      "epoch": 0.5516413130504404,
+      "learning_rate": 1.8727007856672285e-05,
+      "loss": 0.1661,
+      "step": 2754
+    },
+    {
+      "epoch": 0.5516413130504404,
+      "learning_rate": 1.873382272917545e-05,
+      "loss": 0.2822,
+      "step": 2756
+    },
+    {
+      "epoch": 0.5524419535628503,
+      "learning_rate": 1.8740620544333607e-05,
+      "loss": 0.1664,
+      "step": 2758
+    },
+    {
+      "epoch": 0.5524419535628503,
+      "learning_rate": 1.8747401288870472e-05,
+      "loss": 0.6414,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5532425940752602,
+      "learning_rate": 1.875416494954312e-05,
+      "loss": 0.1342,
+      "step": 2762
+    },
+    {
+      "epoch": 0.5532425940752602,
+      "learning_rate": 1.876091151314196e-05,
+      "loss": 0.256,
+      "step": 2764
+    },
+    {
+      "epoch": 0.5540432345876701,
+      "learning_rate": 1.8767640966490813e-05,
+      "loss": 0.6766,
+      "step": 2766
+    },
+    {
+      "epoch": 0.5540432345876701,
+      "learning_rate": 1.877435329644691e-05,
+      "loss": 0.0962,
+      "step": 2768
+    },
+    {
+      "epoch": 0.55484387510008,
+      "learning_rate": 1.878104848990093e-05,
+      "loss": 0.4713,
+      "step": 2770
+    },
+    {
+      "epoch": 0.55484387510008,
+      "learning_rate": 1.8787726533776996e-05,
+      "loss": 0.042,
+      "step": 2772
+    },
+    {
+      "epoch": 0.55564451561249,
+      "learning_rate": 1.879438741503277e-05,
+      "loss": 0.6157,
+      "step": 2774
+    },
+    {
+      "epoch": 0.55564451561249,
+      "learning_rate": 1.8801031120659393e-05,
+      "loss": 0.3977,
+      "step": 2776
+    },
+    {
+      "epoch": 0.5564451561248999,
+      "learning_rate": 1.8807657637681563e-05,
+      "loss": 0.3495,
+      "step": 2778
+    },
+    {
+      "epoch": 0.5564451561248999,
+      "learning_rate": 1.8814266953157557e-05,
+      "loss": 0.3486,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5572457966373099,
+      "learning_rate": 1.8820859054179225e-05,
+      "loss": 0.3993,
+      "step": 2782
+    },
+    {
+      "epoch": 0.5572457966373099,
+      "learning_rate": 1.8827433927872066e-05,
+      "loss": 0.1276,
+      "step": 2784
+    },
+    {
+      "epoch": 0.5580464371497198,
+      "learning_rate": 1.883399156139519e-05,
+      "loss": 0.1528,
+      "step": 2786
+    },
+    {
+      "epoch": 0.5580464371497198,
+      "learning_rate": 1.8840531941941415e-05,
+      "loss": 0.0514,
+      "step": 2788
+    },
+    {
+      "epoch": 0.5588470776621297,
+      "learning_rate": 1.8847055056737233e-05,
+      "loss": 0.1769,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5588470776621297,
+      "learning_rate": 1.8853560893042854e-05,
+      "loss": 0.1586,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5596477181745396,
+      "learning_rate": 1.8860049438152244e-05,
+      "loss": 0.1765,
+      "step": 2794
+    },
+    {
+      "epoch": 0.5596477181745396,
+      "learning_rate": 1.8866520679393127e-05,
+      "loss": 0.27,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5604483586869495,
+      "learning_rate": 1.8872974604127025e-05,
+      "loss": 0.1669,
+      "step": 2798
+    },
+    {
+      "epoch": 0.5604483586869495,
+      "learning_rate": 1.8879411199749303e-05,
+      "loss": 0.1201,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5612489991993594,
+      "learning_rate": 1.8885830453689132e-05,
+      "loss": 0.5391,
+      "step": 2802
+    },
+    {
+      "epoch": 0.5612489991993594,
+      "learning_rate": 1.889223235340958e-05,
+      "loss": 0.1809,
+      "step": 2804
+    },
+    {
+      "epoch": 0.5620496397117695,
+      "learning_rate": 1.889861688640759e-05,
+      "loss": 0.2612,
+      "step": 2806
+    },
+    {
+      "epoch": 0.5620496397117695,
+      "learning_rate": 1.8904984040214037e-05,
+      "loss": 0.5066,
+      "step": 2808
+    },
+    {
+      "epoch": 0.5628502802241794,
+      "learning_rate": 1.891133380239373e-05,
+      "loss": 0.2505,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5628502802241794,
+      "learning_rate": 1.8917666160545436e-05,
+      "loss": 0.2059,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5636509207365893,
+      "learning_rate": 1.892398110230194e-05,
+      "loss": 0.5197,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5636509207365893,
+      "learning_rate": 1.893027861533002e-05,
+      "loss": 0.2614,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5644515612489992,
+      "learning_rate": 1.8936558687330485e-05,
+      "loss": 0.2859,
+      "step": 2818
+    },
+    {
+      "epoch": 0.5644515612489992,
+      "learning_rate": 1.894282130603823e-05,
+      "loss": 0.1955,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5652522017614091,
+      "learning_rate": 1.8949066459222217e-05,
+      "loss": 0.4254,
+      "step": 2822
+    },
+    {
+      "epoch": 0.5652522017614091,
+      "learning_rate": 1.8955294134685528e-05,
+      "loss": 0.3996,
+      "step": 2824
+    },
+    {
+      "epoch": 0.566052842273819,
+      "learning_rate": 1.8961504320265382e-05,
+      "loss": 0.3123,
+      "step": 2826
+    },
+    {
+      "epoch": 0.566052842273819,
+      "learning_rate": 1.896769700383315e-05,
+      "loss": 0.5108,
+      "step": 2828
+    },
+    {
+      "epoch": 0.5668534827862289,
+      "learning_rate": 1.897387217329439e-05,
+      "loss": 0.2395,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5668534827862289,
+      "learning_rate": 1.898002981658886e-05,
+      "loss": 0.1866,
+      "step": 2832
+    },
+    {
+      "epoch": 0.567654123298639,
+      "learning_rate": 1.8986169921690543e-05,
+      "loss": 0.224,
+      "step": 2834
+    },
+    {
+      "epoch": 0.567654123298639,
+      "learning_rate": 1.899229247660769e-05,
+      "loss": 0.0974,
+      "step": 2836
+    },
+    {
+      "epoch": 0.5684547638110489,
+      "learning_rate": 1.899839746938281e-05,
+      "loss": 0.2569,
+      "step": 2838
+    },
+    {
+      "epoch": 0.5684547638110489,
+      "learning_rate": 1.9004484888092724e-05,
+      "loss": 0.1579,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5692554043234588,
+      "learning_rate": 1.9010554720848577e-05,
+      "loss": 0.3427,
+      "step": 2842
+    },
+    {
+      "epoch": 0.5692554043234588,
+      "learning_rate": 1.901660695579585e-05,
+      "loss": 0.0411,
+      "step": 2844
+    },
+    {
+      "epoch": 0.5700560448358687,
+      "learning_rate": 1.9022641581114392e-05,
+      "loss": 0.5222,
+      "step": 2846
+    },
+    {
+      "epoch": 0.5700560448358687,
+      "learning_rate": 1.9028658585018455e-05,
+      "loss": 0.3942,
+      "step": 2848
+    },
+    {
+      "epoch": 0.5708566853482786,
+      "learning_rate": 1.9034657955756695e-05,
+      "loss": 0.2259,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5708566853482786,
+      "learning_rate": 1.9040639681612212e-05,
+      "loss": 0.2525,
+      "step": 2852
+    },
+    {
+      "epoch": 0.5716573258606885,
+      "learning_rate": 1.904660375090257e-05,
+      "loss": 0.5569,
+      "step": 2854
+    },
+    {
+      "epoch": 0.5716573258606885,
+      "learning_rate": 1.9052550151979816e-05,
+      "loss": 0.2262,
+      "step": 2856
+    },
+    {
+      "epoch": 0.5724579663730984,
+      "learning_rate": 1.905847887323049e-05,
+      "loss": 0.2463,
+      "step": 2858
+    },
+    {
+      "epoch": 0.5724579663730984,
+      "learning_rate": 1.9064389903075676e-05,
+      "loss": 0.2776,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5732586068855084,
+      "learning_rate": 1.9070283229971007e-05,
+      "loss": 0.5323,
+      "step": 2862
+    },
+    {
+      "epoch": 0.5732586068855084,
+      "learning_rate": 1.9076158842406674e-05,
+      "loss": 0.2051,
+      "step": 2864
+    },
+    {
+      "epoch": 0.5740592473979184,
+      "learning_rate": 1.9082016728907496e-05,
+      "loss": 0.9686,
+      "step": 2866
+    },
+    {
+      "epoch": 0.5740592473979184,
+      "learning_rate": 1.9087856878032886e-05,
+      "loss": 0.3131,
+      "step": 2868
+    },
+    {
+      "epoch": 0.5748598879103283,
+      "learning_rate": 1.909367927837691e-05,
+      "loss": 0.4471,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5748598879103283,
+      "learning_rate": 1.909948391856829e-05,
+      "loss": 0.2458,
+      "step": 2872
+    },
+    {
+      "epoch": 0.5756605284227382,
+      "learning_rate": 1.910527078727044e-05,
+      "loss": 0.2416,
+      "step": 2874
+    },
+    {
+      "epoch": 0.5756605284227382,
+      "learning_rate": 1.911103987318148e-05,
+      "loss": 0.2777,
+      "step": 2876
+    },
+    {
+      "epoch": 0.5764611689351481,
+      "learning_rate": 1.911679116503425e-05,
+      "loss": 0.3449,
+      "step": 2878
+    },
+    {
+      "epoch": 0.5764611689351481,
+      "learning_rate": 1.912252465159637e-05,
+      "loss": 0.2106,
+      "step": 2880
+    },
+    {
+      "epoch": 0.577261809447558,
+      "learning_rate": 1.9128240321670208e-05,
+      "loss": 0.2166,
+      "step": 2882
+    },
+    {
+      "epoch": 0.577261809447558,
+      "learning_rate": 1.913393816409294e-05,
+      "loss": 0.3659,
+      "step": 2884
+    },
+    {
+      "epoch": 0.578062449959968,
+      "learning_rate": 1.913961816773655e-05,
+      "loss": 0.4227,
+      "step": 2886
+    },
+    {
+      "epoch": 0.578062449959968,
+      "learning_rate": 1.9145280321507872e-05,
+      "loss": 0.1546,
+      "step": 2888
+    },
+    {
+      "epoch": 0.5788630904723779,
+      "learning_rate": 1.9150924614348588e-05,
+      "loss": 0.1763,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5788630904723779,
+      "learning_rate": 1.9156551035235288e-05,
+      "loss": 0.55,
+      "step": 2892
+    },
+    {
+      "epoch": 0.5796637309847879,
+      "learning_rate": 1.916215957317944e-05,
+      "loss": 0.6128,
+      "step": 2894
+    },
+    {
+      "epoch": 0.5796637309847879,
+      "learning_rate": 1.9167750217227454e-05,
+      "loss": 0.0501,
+      "step": 2896
+    },
+    {
+      "epoch": 0.5804643714971978,
+      "learning_rate": 1.9173322956460675e-05,
+      "loss": 0.1496,
+      "step": 2898
+    },
+    {
+      "epoch": 0.5804643714971978,
+      "learning_rate": 1.9178877779995423e-05,
+      "loss": 0.0263,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5812650120096077,
+      "learning_rate": 1.9184414676983006e-05,
+      "loss": 0.4598,
+      "step": 2902
+    },
+    {
+      "epoch": 0.5812650120096077,
+      "learning_rate": 1.9189933636609747e-05,
+      "loss": 0.2697,
+      "step": 2904
+    },
+    {
+      "epoch": 0.5820656525220176,
+      "learning_rate": 1.9195434648097003e-05,
+      "loss": 0.4029,
+      "step": 2906
+    },
+    {
+      "epoch": 0.5820656525220176,
+      "learning_rate": 1.9200917700701173e-05,
+      "loss": 0.1956,
+      "step": 2908
+    },
+    {
+      "epoch": 0.5828662930344275,
+      "learning_rate": 1.9206382783713738e-05,
+      "loss": 0.2976,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5828662930344275,
+      "learning_rate": 1.9211829886461274e-05,
+      "loss": 0.1694,
+      "step": 2912
+    },
+    {
+      "epoch": 0.5836669335468375,
+      "learning_rate": 1.921725899830547e-05,
+      "loss": 0.3611,
+      "step": 2914
+    },
+    {
+      "epoch": 0.5836669335468375,
+      "learning_rate": 1.9222670108643146e-05,
+      "loss": 0.2782,
+      "step": 2916
+    },
+    {
+      "epoch": 0.5844675740592474,
+      "learning_rate": 1.92280632069063e-05,
+      "loss": 0.2271,
+      "step": 2918
+    },
+    {
+      "epoch": 0.5844675740592474,
+      "learning_rate": 1.9233438282562085e-05,
+      "loss": 0.135,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5852682145716573,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.1275,
+      "step": 2922
+    },
+    {
+      "epoch": 0.5852682145716573,
+      "learning_rate": 1.924413432409622e-05,
+      "loss": 0.3228,
+      "step": 2924
+    },
+    {
+      "epoch": 0.5860688550840673,
+      "learning_rate": 1.924945526908497e-05,
+      "loss": 0.519,
+      "step": 2926
+    },
+    {
+      "epoch": 0.5860688550840673,
+      "learning_rate": 1.925475814968719e-05,
+      "loss": 0.121,
+      "step": 2928
+    },
+    {
+      "epoch": 0.5868694955964772,
+      "learning_rate": 1.9260042955546237e-05,
+      "loss": 0.6254,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5868694955964772,
+      "learning_rate": 1.926530967634078e-05,
+      "loss": 0.1059,
+      "step": 2932
+    },
+    {
+      "epoch": 0.5876701361088871,
+      "learning_rate": 1.9270558301784795e-05,
+      "loss": 0.4969,
+      "step": 2934
+    },
+    {
+      "epoch": 0.5876701361088871,
+      "learning_rate": 1.9275788821627607e-05,
+      "loss": 0.1946,
+      "step": 2936
+    },
+    {
+      "epoch": 0.588470776621297,
+      "learning_rate": 1.9281001225653887e-05,
+      "loss": 0.3149,
+      "step": 2938
+    },
+    {
+      "epoch": 0.588470776621297,
+      "learning_rate": 1.9286195503683705e-05,
+      "loss": 0.4259,
+      "step": 2940
+    },
+    {
+      "epoch": 0.589271417133707,
+      "learning_rate": 1.9291371645572517e-05,
+      "loss": 0.4693,
+      "step": 2942
+    },
+    {
+      "epoch": 0.589271417133707,
+      "learning_rate": 1.9296529641211215e-05,
+      "loss": 0.0127,
+      "step": 2944
+    },
+    {
+      "epoch": 0.5900720576461169,
+      "learning_rate": 1.9301669480526115e-05,
+      "loss": 0.4233,
+      "step": 2946
+    },
+    {
+      "epoch": 0.5900720576461169,
+      "learning_rate": 1.9306791153479004e-05,
+      "loss": 0.984,
+      "step": 2948
+    },
+    {
+      "epoch": 0.5908726981585268,
+      "learning_rate": 1.931189465006714e-05,
+      "loss": 0.2328,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5908726981585268,
+      "learning_rate": 1.9316979960323286e-05,
+      "loss": 0.2282,
+      "step": 2952
+    },
+    {
+      "epoch": 0.5916733386709367,
+      "learning_rate": 1.9322047074315717e-05,
+      "loss": 0.6037,
+      "step": 2954
+    },
+    {
+      "epoch": 0.5916733386709367,
+      "learning_rate": 1.932709598214825e-05,
+      "loss": 0.144,
+      "step": 2956
+    },
+    {
+      "epoch": 0.5924739791833467,
+      "learning_rate": 1.9332126673960262e-05,
+      "loss": 0.2022,
+      "step": 2958
+    },
+    {
+      "epoch": 0.5924739791833467,
+      "learning_rate": 1.9337139139926707e-05,
+      "loss": 0.1427,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5932746196957566,
+      "learning_rate": 1.934213337025812e-05,
+      "loss": 0.0921,
+      "step": 2962
+    },
+    {
+      "epoch": 0.5932746196957566,
+      "learning_rate": 1.9347109355200672e-05,
+      "loss": 0.0734,
+      "step": 2964
+    },
+    {
+      "epoch": 0.5940752602081665,
+      "learning_rate": 1.9352067085036145e-05,
+      "loss": 0.1581,
+      "step": 2966
+    },
+    {
+      "epoch": 0.5940752602081665,
+      "learning_rate": 1.935700655008199e-05,
+      "loss": 0.1153,
+      "step": 2968
+    },
+    {
+      "epoch": 0.5948759007205765,
+      "learning_rate": 1.9361927740691327e-05,
+      "loss": 0.2135,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5948759007205765,
+      "learning_rate": 1.9366830647252967e-05,
+      "loss": 0.1675,
+      "step": 2972
+    },
+    {
+      "epoch": 0.5956765412329864,
+      "learning_rate": 1.937171526019142e-05,
+      "loss": 0.4079,
+      "step": 2974
+    },
+    {
+      "epoch": 0.5956765412329864,
+      "learning_rate": 1.9376581569966933e-05,
+      "loss": 0.1283,
+      "step": 2976
+    },
+    {
+      "epoch": 0.5964771817453963,
+      "learning_rate": 1.9381429567075504e-05,
+      "loss": 0.2776,
+      "step": 2978
+    },
+    {
+      "epoch": 0.5964771817453963,
+      "learning_rate": 1.9386259242048883e-05,
+      "loss": 0.3433,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5972778222578062,
+      "learning_rate": 1.93910705854546e-05,
+      "loss": 0.5167,
+      "step": 2982
+    },
+    {
+      "epoch": 0.5972778222578062,
+      "learning_rate": 1.939586358789602e-05,
+      "loss": 0.2453,
+      "step": 2984
+    },
+    {
+      "epoch": 0.5980784627702161,
+      "learning_rate": 1.9400638240012294e-05,
+      "loss": 0.2479,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5980784627702161,
+      "learning_rate": 1.940539453247842e-05,
+      "loss": 0.1008,
+      "step": 2988
+    },
+    {
+      "epoch": 0.5988791032826261,
+      "learning_rate": 1.9410132456005262e-05,
+      "loss": 0.8446,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5988791032826261,
+      "learning_rate": 1.9414852001339547e-05,
+      "loss": 0.0862,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5996797437950361,
+      "learning_rate": 1.9419553159263896e-05,
+      "loss": 0.2936,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5996797437950361,
+      "learning_rate": 1.9424235920596863e-05,
+      "loss": 0.0187,
+      "step": 2996
+    },
+    {
+      "epoch": 0.600480384307446,
+      "learning_rate": 1.94289002761929e-05,
+      "loss": 0.3349,
+      "step": 2998
+    },
+    {
+      "epoch": 0.600480384307446,
+      "learning_rate": 1.9433546216942423e-05,
+      "loss": 0.0936,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6012810248198559,
+      "learning_rate": 1.943817373377181e-05,
+      "loss": 0.3229,
+      "step": 3002
+    },
+    {
+      "epoch": 0.6012810248198559,
+      "learning_rate": 1.944278281764342e-05,
+      "loss": 0.1426,
+      "step": 3004
+    },
+    {
+      "epoch": 0.6020816653322658,
+      "learning_rate": 1.944737345955561e-05,
+      "loss": 0.2377,
+      "step": 3006
+    },
+    {
+      "epoch": 0.6020816653322658,
+      "learning_rate": 1.945194565054276e-05,
+      "loss": 0.1163,
+      "step": 3008
+    },
+    {
+      "epoch": 0.6028823058446757,
+      "learning_rate": 1.945649938167528e-05,
+      "loss": 0.1426,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6028823058446757,
+      "learning_rate": 1.9461034644059637e-05,
+      "loss": 0.1146,
+      "step": 3012
+    },
+    {
+      "epoch": 0.6036829463570856,
+      "learning_rate": 1.946555142883836e-05,
+      "loss": 0.3716,
+      "step": 3014
+    },
+    {
+      "epoch": 0.6036829463570856,
+      "learning_rate": 1.9470049727190073e-05,
+      "loss": 0.1954,
+      "step": 3016
+    },
+    {
+      "epoch": 0.6044835868694955,
+      "learning_rate": 1.9474529530329507e-05,
+      "loss": 0.4179,
+      "step": 3018
+    },
+    {
+      "epoch": 0.6044835868694955,
+      "learning_rate": 1.9478990829507504e-05,
+      "loss": 0.2286,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6052842273819056,
+      "learning_rate": 1.9483433616011047e-05,
+      "loss": 0.5044,
+      "step": 3022
+    },
+    {
+      "epoch": 0.6052842273819056,
+      "learning_rate": 1.948785788116329e-05,
+      "loss": 0.4513,
+      "step": 3024
+    },
+    {
+      "epoch": 0.6060848678943155,
+      "learning_rate": 1.9492263616323533e-05,
+      "loss": 0.3038,
+      "step": 3026
+    },
+    {
+      "epoch": 0.6060848678943155,
+      "learning_rate": 1.9496650812887286e-05,
+      "loss": 0.2418,
+      "step": 3028
+    },
+    {
+      "epoch": 0.6068855084067254,
+      "learning_rate": 1.9501019462286263e-05,
+      "loss": 0.6364,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6068855084067254,
+      "learning_rate": 1.9505369555988395e-05,
+      "loss": 0.0595,
+      "step": 3032
+    },
+    {
+      "epoch": 0.6076861489191353,
+      "learning_rate": 1.9509701085497842e-05,
+      "loss": 0.4716,
+      "step": 3034
+    },
+    {
+      "epoch": 0.6076861489191353,
+      "learning_rate": 1.951401404235505e-05,
+      "loss": 0.749,
+      "step": 3036
+    },
+    {
+      "epoch": 0.6084867894315452,
+      "learning_rate": 1.9518308418136718e-05,
+      "loss": 0.3639,
+      "step": 3038
+    },
+    {
+      "epoch": 0.6084867894315452,
+      "learning_rate": 1.952258420445583e-05,
+      "loss": 0.7821,
+      "step": 3040
+    },
+    {
+      "epoch": 0.6092874299439551,
+      "learning_rate": 1.952684139296169e-05,
+      "loss": 0.2778,
+      "step": 3042
+    },
+    {
+      "epoch": 0.6092874299439551,
+      "learning_rate": 1.9531079975339912e-05,
+      "loss": 0.0734,
+      "step": 3044
+    },
+    {
+      "epoch": 0.610088070456365,
+      "learning_rate": 1.9535299943312455e-05,
+      "loss": 0.3615,
+      "step": 3046
+    },
+    {
+      "epoch": 0.610088070456365,
+      "learning_rate": 1.953950128863762e-05,
+      "loss": 0.1937,
+      "step": 3048
+    },
+    {
+      "epoch": 0.6108887109687751,
+      "learning_rate": 1.9543684003110105e-05,
+      "loss": 0.1667,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6108887109687751,
+      "learning_rate": 1.9547848078560975e-05,
+      "loss": 0.261,
+      "step": 3052
+    },
+    {
+      "epoch": 0.611689351481185,
+      "learning_rate": 1.9551993506857688e-05,
+      "loss": 0.1349,
+      "step": 3054
+    },
+    {
+      "epoch": 0.611689351481185,
+      "learning_rate": 1.9556120279904144e-05,
+      "loss": 0.5711,
+      "step": 3056
+    },
+    {
+      "epoch": 0.6124899919935949,
+      "learning_rate": 1.9560228389640664e-05,
+      "loss": 0.2478,
+      "step": 3058
+    },
+    {
+      "epoch": 0.6124899919935949,
+      "learning_rate": 1.956431782804402e-05,
+      "loss": 0.1349,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6132906325060048,
+      "learning_rate": 1.956838858712744e-05,
+      "loss": 0.2274,
+      "step": 3062
+    },
+    {
+      "epoch": 0.6132906325060048,
+      "learning_rate": 1.957244065894066e-05,
+      "loss": 0.1054,
+      "step": 3064
+    },
+    {
+      "epoch": 0.6140912730184147,
+      "learning_rate": 1.9576474035569892e-05,
+      "loss": 0.3092,
+      "step": 3066
+    },
+    {
+      "epoch": 0.6140912730184147,
+      "learning_rate": 1.9580488709137858e-05,
+      "loss": 0.1817,
+      "step": 3068
+    },
+    {
+      "epoch": 0.6148919135308246,
+      "learning_rate": 1.9584484671803818e-05,
+      "loss": 0.2106,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6148919135308246,
+      "learning_rate": 1.9588461915763566e-05,
+      "loss": 0.1778,
+      "step": 3072
+    },
+    {
+      "epoch": 0.6156925540432346,
+      "learning_rate": 1.9592420433249462e-05,
+      "loss": 0.2018,
+      "step": 3074
+    },
+    {
+      "epoch": 0.6156925540432346,
+      "learning_rate": 1.9596360216530436e-05,
+      "loss": 0.5225,
+      "step": 3076
+    },
+    {
+      "epoch": 0.6164931945556446,
+      "learning_rate": 1.9600281257912e-05,
+      "loss": 0.1496,
+      "step": 3078
+    },
+    {
+      "epoch": 0.6164931945556446,
+      "learning_rate": 1.9604183549736283e-05,
+      "loss": 0.208,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6172938350680545,
+      "learning_rate": 1.960806708438202e-05,
+      "loss": 0.2119,
+      "step": 3082
+    },
+    {
+      "epoch": 0.6172938350680545,
+      "learning_rate": 1.961193185426459e-05,
+      "loss": 0.5395,
+      "step": 3084
+    },
+    {
+      "epoch": 0.6180944755804644,
+      "learning_rate": 1.9615777851836003e-05,
+      "loss": 0.2192,
+      "step": 3086
+    },
+    {
+      "epoch": 0.6180944755804644,
+      "learning_rate": 1.9619605069584954e-05,
+      "loss": 0.4713,
+      "step": 3088
+    },
+    {
+      "epoch": 0.6188951160928743,
+      "learning_rate": 1.962341350003679e-05,
+      "loss": 0.6812,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6188951160928743,
+      "learning_rate": 1.9627203135753573e-05,
+      "loss": 0.1647,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6196957566052842,
+      "learning_rate": 1.9630973969334068e-05,
+      "loss": 0.4365,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6196957566052842,
+      "learning_rate": 1.9634725993413744e-05,
+      "loss": 0.205,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6204963971176941,
+      "learning_rate": 1.9638459200664822e-05,
+      "loss": 0.1851,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6204963971176941,
+      "learning_rate": 1.9642173583796265e-05,
+      "loss": 0.1681,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6212970376301041,
+      "learning_rate": 1.9645869135553806e-05,
+      "loss": 0.2529,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6212970376301041,
+      "learning_rate": 1.964954584871995e-05,
+      "loss": 0.1761,
+      "step": 3104
+    },
+    {
+      "epoch": 0.622097678142514,
+      "learning_rate": 1.965320371611399e-05,
+      "loss": 0.2018,
+      "step": 3106
+    },
+    {
+      "epoch": 0.622097678142514,
+      "learning_rate": 1.9656842730592046e-05,
+      "loss": 0.0646,
+      "step": 3108
+    },
+    {
+      "epoch": 0.622898318654924,
+      "learning_rate": 1.9660462885047032e-05,
+      "loss": 0.1787,
+      "step": 3110
+    },
+    {
+      "epoch": 0.622898318654924,
+      "learning_rate": 1.966406417240872e-05,
+      "loss": 0.2307,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6236989591673339,
+      "learning_rate": 1.9667646585643703e-05,
+      "loss": 0.5142,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6236989591673339,
+      "learning_rate": 1.967121011775546e-05,
+      "loss": 0.0361,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6244995996797438,
+      "learning_rate": 1.967475476178433e-05,
+      "loss": 0.4309,
+      "step": 3118
+    },
+    {
+      "epoch": 0.6244995996797438,
+      "learning_rate": 1.967828051080755e-05,
+      "loss": 0.2464,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6253002401921537,
+      "learning_rate": 1.9681787357939254e-05,
+      "loss": 0.6014,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6253002401921537,
+      "learning_rate": 1.9685275296330497e-05,
+      "loss": 0.5234,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6261008807045636,
+      "learning_rate": 1.968874431916926e-05,
+      "loss": 0.4965,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6261008807045636,
+      "learning_rate": 1.969219441968046e-05,
+      "loss": 0.091,
+      "step": 3128
+    },
+    {
+      "epoch": 0.6269015212169736,
+      "learning_rate": 1.969562559112598e-05,
+      "loss": 0.2164,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6269015212169736,
+      "learning_rate": 1.969903782680467e-05,
+      "loss": 0.716,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6277021617293835,
+      "learning_rate": 1.970243112005235e-05,
+      "loss": 0.2961,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6277021617293835,
+      "learning_rate": 1.9705805464241856e-05,
+      "loss": 0.3265,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6285028022417934,
+      "learning_rate": 1.970916085278302e-05,
+      "loss": 0.2236,
+      "step": 3138
+    },
+    {
+      "epoch": 0.6285028022417934,
+      "learning_rate": 1.9712497279122692e-05,
+      "loss": 0.5319,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6293034427542034,
+      "learning_rate": 1.9715814736744755e-05,
+      "loss": 0.1162,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6293034427542034,
+      "learning_rate": 1.971911321917015e-05,
+      "loss": 0.0386,
+      "step": 3144
+    },
+    {
+      "epoch": 0.6301040832666133,
+      "learning_rate": 1.9722392719956864e-05,
+      "loss": 0.1987,
+      "step": 3146
+    },
+    {
+      "epoch": 0.6301040832666133,
+      "learning_rate": 1.9725653232699962e-05,
+      "loss": 0.2876,
+      "step": 3148
+    },
+    {
+      "epoch": 0.6309047237790232,
+      "learning_rate": 1.9728894751031595e-05,
+      "loss": 0.3842,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6309047237790232,
+      "learning_rate": 1.9732117268621005e-05,
+      "loss": 0.7951,
+      "step": 3152
+    },
+    {
+      "epoch": 0.6317053642914331,
+      "learning_rate": 1.9735320779174545e-05,
+      "loss": 0.3069,
+      "step": 3154
+    },
+    {
+      "epoch": 0.6317053642914331,
+      "learning_rate": 1.9738505276435692e-05,
+      "loss": 0.1682,
+      "step": 3156
+    },
+    {
+      "epoch": 0.6325060048038431,
+      "learning_rate": 1.974167075418505e-05,
+      "loss": 0.2534,
+      "step": 3158
+    },
+    {
+      "epoch": 0.6325060048038431,
+      "learning_rate": 1.9744817206240377e-05,
+      "loss": 0.2263,
+      "step": 3160
+    },
+    {
+      "epoch": 0.633306645316253,
+      "learning_rate": 1.9747944626456577e-05,
+      "loss": 0.1977,
+      "step": 3162
+    },
+    {
+      "epoch": 0.633306645316253,
+      "learning_rate": 1.9751053008725736e-05,
+      "loss": 0.1496,
+      "step": 3164
+    },
+    {
+      "epoch": 0.6341072858286629,
+      "learning_rate": 1.975414234697712e-05,
+      "loss": 0.0671,
+      "step": 3166
+    },
+    {
+      "epoch": 0.6341072858286629,
+      "learning_rate": 1.9757212635177177e-05,
+      "loss": 0.0523,
+      "step": 3168
+    },
+    {
+      "epoch": 0.6349079263410728,
+      "learning_rate": 1.9760263867329568e-05,
+      "loss": 0.1675,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6349079263410728,
+      "learning_rate": 1.9763296037475174e-05,
+      "loss": 0.6791,
+      "step": 3172
+    },
+    {
+      "epoch": 0.6357085668534828,
+      "learning_rate": 1.97663091396921e-05,
+      "loss": 0.1423,
+      "step": 3174
+    },
+    {
+      "epoch": 0.6357085668534828,
+      "learning_rate": 1.976930316809569e-05,
+      "loss": 0.3445,
+      "step": 3176
+    },
+    {
+      "epoch": 0.6365092073658927,
+      "learning_rate": 1.9772278116838543e-05,
+      "loss": 0.2066,
+      "step": 3178
+    },
+    {
+      "epoch": 0.6365092073658927,
+      "learning_rate": 1.9775233980110524e-05,
+      "loss": 0.1515,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6373098478783027,
+      "learning_rate": 1.977817075213876e-05,
+      "loss": 0.4071,
+      "step": 3182
+    },
+    {
+      "epoch": 0.6373098478783027,
+      "learning_rate": 1.978108842718768e-05,
+      "loss": 0.2463,
+      "step": 3184
+    },
+    {
+      "epoch": 0.6381104883907126,
+      "learning_rate": 1.9783986999558994e-05,
+      "loss": 0.1663,
+      "step": 3186
+    },
+    {
+      "epoch": 0.6381104883907126,
+      "learning_rate": 1.9786866463591732e-05,
+      "loss": 0.62,
+      "step": 3188
+    },
+    {
+      "epoch": 0.6389111289031225,
+      "learning_rate": 1.9789726813662233e-05,
+      "loss": 0.1948,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6389111289031225,
+      "learning_rate": 1.9792568044184176e-05,
+      "loss": 0.4769,
+      "step": 3192
+    },
+    {
+      "epoch": 0.6397117694155324,
+      "learning_rate": 1.979539014960858e-05,
+      "loss": 0.1599,
+      "step": 3194
+    },
+    {
+      "epoch": 0.6397117694155324,
+      "learning_rate": 1.9798193124423804e-05,
+      "loss": 0.1502,
+      "step": 3196
+    },
+    {
+      "epoch": 0.6405124099279423,
+      "learning_rate": 1.9800976963155584e-05,
+      "loss": 0.549,
+      "step": 3198
+    },
+    {
+      "epoch": 0.6405124099279423,
+      "learning_rate": 1.9803741660367015e-05,
+      "loss": 0.1827,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6413130504403523,
+      "learning_rate": 1.980648721065859e-05,
+      "loss": 0.2668,
+      "step": 3202
+    },
+    {
+      "epoch": 0.6413130504403523,
+      "learning_rate": 1.9809213608668185e-05,
+      "loss": 0.2382,
+      "step": 3204
+    },
+    {
+      "epoch": 0.6421136909527622,
+      "learning_rate": 1.9811920849071092e-05,
+      "loss": 0.2692,
+      "step": 3206
+    },
+    {
+      "epoch": 0.6421136909527622,
+      "learning_rate": 1.9814608926580007e-05,
+      "loss": 0.1033,
+      "step": 3208
+    },
+    {
+      "epoch": 0.6429143314651722,
+      "learning_rate": 1.9817277835945057e-05,
+      "loss": 0.5423,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6429143314651722,
+      "learning_rate": 1.9819927571953807e-05,
+      "loss": 0.4381,
+      "step": 3212
+    },
+    {
+      "epoch": 0.6437149719775821,
+      "learning_rate": 1.9822558129431263e-05,
+      "loss": 0.929,
+      "step": 3214
+    },
+    {
+      "epoch": 0.6437149719775821,
+      "learning_rate": 1.9825169503239885e-05,
+      "loss": 0.1673,
+      "step": 3216
+    },
+    {
+      "epoch": 0.644515612489992,
+      "learning_rate": 1.9827761688279606e-05,
+      "loss": 0.2194,
+      "step": 3218
+    },
+    {
+      "epoch": 0.644515612489992,
+      "learning_rate": 1.983033467948784e-05,
+      "loss": 0.0304,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6453162530024019,
+      "learning_rate": 1.983288847183947e-05,
+      "loss": 0.2982,
+      "step": 3222
+    },
+    {
+      "epoch": 0.6453162530024019,
+      "learning_rate": 1.9835423060346892e-05,
+      "loss": 0.2106,
+      "step": 3224
+    },
+    {
+      "epoch": 0.6461168935148118,
+      "learning_rate": 1.9837938440059992e-05,
+      "loss": 0.4267,
+      "step": 3226
+    },
+    {
+      "epoch": 0.6461168935148118,
+      "learning_rate": 1.9840434606066182e-05,
+      "loss": 0.1768,
+      "step": 3228
+    },
+    {
+      "epoch": 0.6469175340272217,
+      "learning_rate": 1.9842911553490392e-05,
+      "loss": 0.5597,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6469175340272217,
+      "learning_rate": 1.9845369277495102e-05,
+      "loss": 0.1497,
+      "step": 3232
+    },
+    {
+      "epoch": 0.6477181745396317,
+      "learning_rate": 1.984780777328031e-05,
+      "loss": 0.2465,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6477181745396317,
+      "learning_rate": 1.9850227036083592e-05,
+      "loss": 0.1608,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6485188150520417,
+      "learning_rate": 1.985262706118007e-05,
+      "loss": 0.2411,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6485188150520417,
+      "learning_rate": 1.985500784388244e-05,
+      "loss": 0.2069,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6493194555644516,
+      "learning_rate": 1.9857369379540982e-05,
+      "loss": 0.3757,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6493194555644516,
+      "learning_rate": 1.985971166354357e-05,
+      "loss": 0.3775,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6501200960768615,
+      "learning_rate": 1.986203469131567e-05,
+      "loss": 0.5834,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6501200960768615,
+      "learning_rate": 1.9864338458320366e-05,
+      "loss": 0.5152,
+      "step": 3248
+    },
+    {
+      "epoch": 0.6509207365892714,
+      "learning_rate": 1.986662296005834e-05,
+      "loss": 0.5086,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6509207365892714,
+      "learning_rate": 1.986888819206792e-05,
+      "loss": 0.5967,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6517213771016813,
+      "learning_rate": 1.987113414992505e-05,
+      "loss": 0.4372,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6517213771016813,
+      "learning_rate": 1.9873360829243323e-05,
+      "loss": 0.2861,
+      "step": 3256
+    },
+    {
+      "epoch": 0.6525220176140912,
+      "learning_rate": 1.9875568225674e-05,
+      "loss": 0.1581,
+      "step": 3258
+    },
+    {
+      "epoch": 0.6525220176140912,
+      "learning_rate": 1.9877756334905983e-05,
+      "loss": 0.1872,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6533226581265013,
+      "learning_rate": 1.9879925152665845e-05,
+      "loss": 0.5262,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6533226581265013,
+      "learning_rate": 1.9882074674717836e-05,
+      "loss": 0.183,
+      "step": 3264
+    },
+    {
+      "epoch": 0.6541232986389112,
+      "learning_rate": 1.9884204896863895e-05,
+      "loss": 0.3277,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6541232986389112,
+      "learning_rate": 1.988631581494365e-05,
+      "loss": 0.1612,
+      "step": 3268
+    },
+    {
+      "epoch": 0.6549239391513211,
+      "learning_rate": 1.9888407424834433e-05,
+      "loss": 0.4648,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6549239391513211,
+      "learning_rate": 1.989047972245129e-05,
+      "loss": 0.3959,
+      "step": 3272
+    },
+    {
+      "epoch": 0.655724579663731,
+      "learning_rate": 1.989253270374697e-05,
+      "loss": 0.2024,
+      "step": 3274
+    },
+    {
+      "epoch": 0.655724579663731,
+      "learning_rate": 1.9894566364711965e-05,
+      "loss": 0.0771,
+      "step": 3276
+    },
+    {
+      "epoch": 0.6565252201761409,
+      "learning_rate": 1.9896580701374482e-05,
+      "loss": 0.4497,
+      "step": 3278
+    },
+    {
+      "epoch": 0.6565252201761409,
+      "learning_rate": 1.989857570980049e-05,
+      "loss": 0.2759,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6573258606885508,
+      "learning_rate": 1.9900551386093677e-05,
+      "loss": 0.1272,
+      "step": 3282
+    },
+    {
+      "epoch": 0.6573258606885508,
+      "learning_rate": 1.990250772639552e-05,
+      "loss": 0.3815,
+      "step": 3284
+    },
+    {
+      "epoch": 0.6581265012009607,
+      "learning_rate": 1.9904444726885236e-05,
+      "loss": 0.4654,
+      "step": 3286
+    },
+    {
+      "epoch": 0.6581265012009607,
+      "learning_rate": 1.9906362383779826e-05,
+      "loss": 0.4346,
+      "step": 3288
+    },
+    {
+      "epoch": 0.6589271417133707,
+      "learning_rate": 1.990826069333406e-05,
+      "loss": 0.3647,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6589271417133707,
+      "learning_rate": 1.99101396518405e-05,
+      "loss": 0.4385,
+      "step": 3292
+    },
+    {
+      "epoch": 0.6597277822257807,
+      "learning_rate": 1.99119992556295e-05,
+      "loss": 0.352,
+      "step": 3294
+    },
+    {
+      "epoch": 0.6597277822257807,
+      "learning_rate": 1.9913839501069213e-05,
+      "loss": 0.3491,
+      "step": 3296
+    },
+    {
+      "epoch": 0.6605284227381906,
+      "learning_rate": 1.99156603845656e-05,
+      "loss": 0.3054,
+      "step": 3298
+    },
+    {
+      "epoch": 0.6605284227381906,
+      "learning_rate": 1.9917461902562435e-05,
+      "loss": 0.2773,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6613290632506005,
+      "learning_rate": 1.9919244051541315e-05,
+      "loss": 0.2163,
+      "step": 3302
+    },
+    {
+      "epoch": 0.6613290632506005,
+      "learning_rate": 1.9921006828021666e-05,
+      "loss": 0.264,
+      "step": 3304
+    },
+    {
+      "epoch": 0.6621297037630104,
+      "learning_rate": 1.9922750228560746e-05,
+      "loss": 0.2482,
+      "step": 3306
+    },
+    {
+      "epoch": 0.6621297037630104,
+      "learning_rate": 1.9924474249753652e-05,
+      "loss": 0.696,
+      "step": 3308
+    },
+    {
+      "epoch": 0.6629303442754203,
+      "learning_rate": 1.9926178888233344e-05,
+      "loss": 0.1611,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6629303442754203,
+      "learning_rate": 1.9927864140670615e-05,
+      "loss": 0.5024,
+      "step": 3312
+    },
+    {
+      "epoch": 0.6637309847878302,
+      "learning_rate": 1.9929530003774133e-05,
+      "loss": 0.3944,
+      "step": 3314
+    },
+    {
+      "epoch": 0.6637309847878302,
+      "learning_rate": 1.9931176474290438e-05,
+      "loss": 0.2715,
+      "step": 3316
+    },
+    {
+      "epoch": 0.6645316253002402,
+      "learning_rate": 1.993280354900393e-05,
+      "loss": 0.4165,
+      "step": 3318
+    },
+    {
+      "epoch": 0.6645316253002402,
+      "learning_rate": 1.99344112247369e-05,
+      "loss": 0.4235,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6653322658126501,
+      "learning_rate": 1.9935999498349518e-05,
+      "loss": 0.1793,
+      "step": 3322
+    },
+    {
+      "epoch": 0.6653322658126501,
+      "learning_rate": 1.9937568366739858e-05,
+      "loss": 0.1954,
+      "step": 3324
+    },
+    {
+      "epoch": 0.6661329063250601,
+      "learning_rate": 1.9939117826843883e-05,
+      "loss": 0.1881,
+      "step": 3326
+    },
+    {
+      "epoch": 0.6661329063250601,
+      "learning_rate": 1.9940647875635463e-05,
+      "loss": 0.3484,
+      "step": 3328
+    },
+    {
+      "epoch": 0.66693354683747,
+      "learning_rate": 1.9942158510126384e-05,
+      "loss": 0.165,
+      "step": 3330
+    },
+    {
+      "epoch": 0.66693354683747,
+      "learning_rate": 1.9943649727366335e-05,
+      "loss": 0.2826,
+      "step": 3332
+    },
+    {
+      "epoch": 0.6677341873498799,
+      "learning_rate": 1.9945121524442944e-05,
+      "loss": 0.2468,
+      "step": 3334
+    },
+    {
+      "epoch": 0.6677341873498799,
+      "learning_rate": 1.994657389848176e-05,
+      "loss": 0.1852,
+      "step": 3336
+    },
+    {
+      "epoch": 0.6685348278622898,
+      "learning_rate": 1.9948006846646262e-05,
+      "loss": 0.6371,
+      "step": 3338
+    },
+    {
+      "epoch": 0.6685348278622898,
+      "learning_rate": 1.994942036613787e-05,
+      "loss": 0.1481,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6693354683746997,
+      "learning_rate": 1.9950814454195953e-05,
+      "loss": 0.1688,
+      "step": 3342
+    },
+    {
+      "epoch": 0.6693354683746997,
+      "learning_rate": 1.9952189108097825e-05,
+      "loss": 0.6612,
+      "step": 3344
+    },
+    {
+      "epoch": 0.6701361088871097,
+      "learning_rate": 1.9953544325158755e-05,
+      "loss": 0.7063,
+      "step": 3346
+    },
+    {
+      "epoch": 0.6701361088871097,
+      "learning_rate": 1.995488010273198e-05,
+      "loss": 0.0452,
+      "step": 3348
+    },
+    {
+      "epoch": 0.6709367493995196,
+      "learning_rate": 1.9956196438208693e-05,
+      "loss": 0.5025,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6709367493995196,
+      "learning_rate": 1.9957493329018064e-05,
+      "loss": 0.1038,
+      "step": 3352
+    },
+    {
+      "epoch": 0.6717373899119295,
+      "learning_rate": 1.9958770772627236e-05,
+      "loss": 0.1697,
+      "step": 3354
+    },
+    {
+      "epoch": 0.6717373899119295,
+      "learning_rate": 1.9960028766541336e-05,
+      "loss": 0.1717,
+      "step": 3356
+    },
+    {
+      "epoch": 0.6725380304243395,
+      "learning_rate": 1.9961267308303473e-05,
+      "loss": 0.2192,
+      "step": 3358
+    },
+    {
+      "epoch": 0.6725380304243395,
+      "learning_rate": 1.996248639549475e-05,
+      "loss": 0.1963,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6733386709367494,
+      "learning_rate": 1.9963686025734262e-05,
+      "loss": 0.2069,
+      "step": 3362
+    },
+    {
+      "epoch": 0.6733386709367494,
+      "learning_rate": 1.9964866196679105e-05,
+      "loss": 0.2617,
+      "step": 3364
+    },
+    {
+      "epoch": 0.6741393114491593,
+      "learning_rate": 1.9966026906024377e-05,
+      "loss": 0.3077,
+      "step": 3366
+    },
+    {
+      "epoch": 0.6741393114491593,
+      "learning_rate": 1.9967168151503196e-05,
+      "loss": 0.2767,
+      "step": 3368
+    },
+    {
+      "epoch": 0.6749399519615693,
+      "learning_rate": 1.9968289930886675e-05,
+      "loss": 0.2471,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6749399519615693,
+      "learning_rate": 1.9969392241983957e-05,
+      "loss": 0.0995,
+      "step": 3372
+    },
+    {
+      "epoch": 0.6757405924739792,
+      "learning_rate": 1.997047508264221e-05,
+      "loss": 0.3756,
+      "step": 3374
+    },
+    {
+      "epoch": 0.6757405924739792,
+      "learning_rate": 1.997153845074662e-05,
+      "loss": 0.4216,
+      "step": 3376
+    },
+    {
+      "epoch": 0.6765412329863891,
+      "learning_rate": 1.99725823442204e-05,
+      "loss": 0.5415,
+      "step": 3378
+    },
+    {
+      "epoch": 0.6765412329863891,
+      "learning_rate": 1.9973606761024813e-05,
+      "loss": 0.1883,
+      "step": 3380
+    },
+    {
+      "epoch": 0.677341873498799,
+      "learning_rate": 1.9974611699159142e-05,
+      "loss": 0.1124,
+      "step": 3382
+    },
+    {
+      "epoch": 0.677341873498799,
+      "learning_rate": 1.997559715666073e-05,
+      "loss": 0.0647,
+      "step": 3384
+    },
+    {
+      "epoch": 0.678142514011209,
+      "learning_rate": 1.9976563131604945e-05,
+      "loss": 0.1507,
+      "step": 3386
+    },
+    {
+      "epoch": 0.678142514011209,
+      "learning_rate": 1.9977509622105233e-05,
+      "loss": 0.2612,
+      "step": 3388
+    },
+    {
+      "epoch": 0.6789431545236189,
+      "learning_rate": 1.9978436626313065e-05,
+      "loss": 0.7532,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6789431545236189,
+      "learning_rate": 1.9979344142417986e-05,
+      "loss": 0.3349,
+      "step": 3392
+    },
+    {
+      "epoch": 0.6797437950360288,
+      "learning_rate": 1.99802321686476e-05,
+      "loss": 0.5495,
+      "step": 3394
+    },
+    {
+      "epoch": 0.6797437950360288,
+      "learning_rate": 1.9981100703267567e-05,
+      "loss": 0.0274,
+      "step": 3396
+    },
+    {
+      "epoch": 0.6805444355484388,
+      "learning_rate": 1.9981949744581622e-05,
+      "loss": 0.542,
+      "step": 3398
+    },
+    {
+      "epoch": 0.6805444355484388,
+      "learning_rate": 1.998277929093157e-05,
+      "loss": 0.0597,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6813450760608487,
+      "learning_rate": 1.9983589340697288e-05,
+      "loss": 0.6511,
+      "step": 3402
+    },
+    {
+      "epoch": 0.6813450760608487,
+      "learning_rate": 1.998437989229673e-05,
+      "loss": 0.3231,
+      "step": 3404
+    },
+    {
+      "epoch": 0.6821457165732586,
+      "learning_rate": 1.998515094418594e-05,
+      "loss": 0.1761,
+      "step": 3406
+    },
+    {
+      "epoch": 0.6821457165732586,
+      "learning_rate": 1.9985902494859023e-05,
+      "loss": 0.239,
+      "step": 3408
+    },
+    {
+      "epoch": 0.6829463570856685,
+      "learning_rate": 1.99866345428482e-05,
+      "loss": 0.2069,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6829463570856685,
+      "learning_rate": 1.998734708672375e-05,
+      "loss": 0.2792,
+      "step": 3412
+    },
+    {
+      "epoch": 0.6837469975980784,
+      "learning_rate": 1.998804012509407e-05,
+      "loss": 0.4495,
+      "step": 3414
+    },
+    {
+      "epoch": 0.6837469975980784,
+      "learning_rate": 1.9988713656605635e-05,
+      "loss": 0.5576,
+      "step": 3416
+    },
+    {
+      "epoch": 0.6845476381104884,
+      "learning_rate": 1.9989367679943025e-05,
+      "loss": 0.3214,
+      "step": 3418
+    },
+    {
+      "epoch": 0.6845476381104884,
+      "learning_rate": 1.9990002193828923e-05,
+      "loss": 0.3858,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6853482786228983,
+      "learning_rate": 1.9990617197024103e-05,
+      "loss": 0.4545,
+      "step": 3422
+    },
+    {
+      "epoch": 0.6853482786228983,
+      "learning_rate": 1.9991212688327456e-05,
+      "loss": 0.2006,
+      "step": 3424
+    },
+    {
+      "epoch": 0.6861489191353083,
+      "learning_rate": 1.999178866657597e-05,
+      "loss": 0.1961,
+      "step": 3426
+    },
+    {
+      "epoch": 0.6861489191353083,
+      "learning_rate": 1.9992345130644747e-05,
+      "loss": 0.4886,
+      "step": 3428
+    },
+    {
+      "epoch": 0.6869495596477182,
+      "learning_rate": 1.999288207944701e-05,
+      "loss": 0.1273,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6869495596477182,
+      "learning_rate": 1.999339951193407e-05,
+      "loss": 0.1237,
+      "step": 3432
+    },
+    {
+      "epoch": 0.6877502001601281,
+      "learning_rate": 1.999389742709538e-05,
+      "loss": 0.2531,
+      "step": 3434
+    },
+    {
+      "epoch": 0.6877502001601281,
+      "learning_rate": 1.9994375823958504e-05,
+      "loss": 0.1505,
+      "step": 3436
+    },
+    {
+      "epoch": 0.688550840672538,
+      "learning_rate": 1.9994834701589113e-05,
+      "loss": 0.5149,
+      "step": 3438
+    },
+    {
+      "epoch": 0.688550840672538,
+      "learning_rate": 1.9995274059091018e-05,
+      "loss": 0.2244,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6893514811849479,
+      "learning_rate": 1.999569389560614e-05,
+      "loss": 0.4143,
+      "step": 3442
+    },
+    {
+      "epoch": 0.6893514811849479,
+      "learning_rate": 1.999609421031453e-05,
+      "loss": 0.2074,
+      "step": 3444
+    },
+    {
+      "epoch": 0.6901521216973578,
+      "learning_rate": 1.9996475002434365e-05,
+      "loss": 0.1683,
+      "step": 3446
+    },
+    {
+      "epoch": 0.6901521216973578,
+      "learning_rate": 1.999683627122195e-05,
+      "loss": 0.5138,
+      "step": 3448
+    },
+    {
+      "epoch": 0.6909527622097679,
+      "learning_rate": 1.999717801597172e-05,
+      "loss": 0.5701,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6909527622097679,
+      "learning_rate": 1.9997500236016233e-05,
+      "loss": 0.423,
+      "step": 3452
+    },
+    {
+      "epoch": 0.6917534027221778,
+      "learning_rate": 1.9997802930726195e-05,
+      "loss": 0.3856,
+      "step": 3454
+    },
+    {
+      "epoch": 0.6917534027221778,
+      "learning_rate": 1.9998086099510433e-05,
+      "loss": 0.2978,
+      "step": 3456
+    },
+    {
+      "epoch": 0.6925540432345877,
+      "learning_rate": 1.9998349741815916e-05,
+      "loss": 0.3094,
+      "step": 3458
+    },
+    {
+      "epoch": 0.6925540432345877,
+      "learning_rate": 1.9998593857127736e-05,
+      "loss": 0.1577,
+      "step": 3460
+    },
+    {
+      "epoch": 0.6933546837469976,
+      "learning_rate": 1.999881844496914e-05,
+      "loss": 0.211,
+      "step": 3462
+    },
+    {
+      "epoch": 0.6933546837469976,
+      "learning_rate": 1.99990235049015e-05,
+      "loss": 0.2351,
+      "step": 3464
+    },
+    {
+      "epoch": 0.6941553242594075,
+      "learning_rate": 1.9999209036524326e-05,
+      "loss": 0.3274,
+      "step": 3466
+    },
+    {
+      "epoch": 0.6941553242594075,
+      "learning_rate": 1.9999375039475275e-05,
+      "loss": 0.3728,
+      "step": 3468
+    },
+    {
+      "epoch": 0.6949559647718174,
+      "learning_rate": 1.999952151343014e-05,
+      "loss": 0.1854,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6949559647718174,
+      "learning_rate": 1.999964845810285e-05,
+      "loss": 0.4345,
+      "step": 3472
+    },
+    {
+      "epoch": 0.6957566052842273,
+      "learning_rate": 1.9999755873245484e-05,
+      "loss": 0.2237,
+      "step": 3474
+    },
+    {
+      "epoch": 0.6957566052842273,
+      "learning_rate": 1.9999843758648253e-05,
+      "loss": 0.1767,
+      "step": 3476
+    },
+    {
+      "epoch": 0.6965572457966374,
+      "learning_rate": 1.999991211413952e-05,
+      "loss": 0.5157,
+      "step": 3478
+    },
+    {
+      "epoch": 0.6965572457966374,
+      "learning_rate": 1.999996093958578e-05,
+      "loss": 0.3625,
+      "step": 3480
+    },
+    {
+      "epoch": 0.6973578863090473,
+      "learning_rate": 1.9999990234891677e-05,
+      "loss": 0.4346,
+      "step": 3482
+    },
+    {
+      "epoch": 0.6973578863090473,
+      "learning_rate": 2e-05,
+      "loss": 0.1571,
+      "step": 3484
+    },
+    {
+      "epoch": 0.6981585268214572,
+      "learning_rate": 1.999999023489168e-05,
+      "loss": 0.1579,
+      "step": 3486
+    },
+    {
+      "epoch": 0.6981585268214572,
+      "learning_rate": 1.999996093958578e-05,
+      "loss": 0.1955,
+      "step": 3488
+    },
+    {
+      "epoch": 0.6989591673338671,
+      "learning_rate": 1.999991211413952e-05,
+      "loss": 0.1647,
+      "step": 3490
+    },
+    {
+      "epoch": 0.6989591673338671,
+      "learning_rate": 1.9999843758648253e-05,
+      "loss": 0.0961,
+      "step": 3492
+    },
+    {
+      "epoch": 0.699759807846277,
+      "learning_rate": 1.9999755873245484e-05,
+      "loss": 0.44,
+      "step": 3494
+    },
+    {
+      "epoch": 0.699759807846277,
+      "learning_rate": 1.999964845810285e-05,
+      "loss": 0.9857,
+      "step": 3496
+    },
+    {
+      "epoch": 0.7005604483586869,
+      "learning_rate": 1.999952151343014e-05,
+      "loss": 0.1852,
+      "step": 3498
+    },
+    {
+      "epoch": 0.7005604483586869,
+      "learning_rate": 1.9999375039475278e-05,
+      "loss": 0.3013,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7013610888710968,
+      "learning_rate": 1.9999209036524326e-05,
+      "loss": 0.6867,
+      "step": 3502
+    },
+    {
+      "epoch": 0.7013610888710968,
+      "learning_rate": 1.99990235049015e-05,
+      "loss": 0.2531,
+      "step": 3504
+    },
+    {
+      "epoch": 0.7021617293835068,
+      "learning_rate": 1.999881844496914e-05,
+      "loss": 0.2156,
+      "step": 3506
+    },
+    {
+      "epoch": 0.7021617293835068,
+      "learning_rate": 1.9998593857127736e-05,
+      "loss": 0.2191,
+      "step": 3508
+    },
+    {
+      "epoch": 0.7029623698959168,
+      "learning_rate": 1.9998349741815916e-05,
+      "loss": 0.282,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7029623698959168,
+      "learning_rate": 1.9998086099510433e-05,
+      "loss": 0.3974,
+      "step": 3512
+    },
+    {
+      "epoch": 0.7037630104083267,
+      "learning_rate": 1.9997802930726195e-05,
+      "loss": 0.4265,
+      "step": 3514
+    },
+    {
+      "epoch": 0.7037630104083267,
+      "learning_rate": 1.9997500236016233e-05,
+      "loss": 0.1428,
+      "step": 3516
+    },
+    {
+      "epoch": 0.7045636509207366,
+      "learning_rate": 1.999717801597172e-05,
+      "loss": 0.3773,
+      "step": 3518
+    },
+    {
+      "epoch": 0.7045636509207366,
+      "learning_rate": 1.999683627122195e-05,
+      "loss": 0.1995,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7053642914331465,
+      "learning_rate": 1.9996475002434365e-05,
+      "loss": 0.4602,
+      "step": 3522
+    },
+    {
+      "epoch": 0.7053642914331465,
+      "learning_rate": 1.999609421031453e-05,
+      "loss": 0.2064,
+      "step": 3524
+    },
+    {
+      "epoch": 0.7061649319455564,
+      "learning_rate": 1.999569389560614e-05,
+      "loss": 0.1755,
+      "step": 3526
+    },
+    {
+      "epoch": 0.7061649319455564,
+      "learning_rate": 1.999527405909102e-05,
+      "loss": 0.5047,
+      "step": 3528
+    },
+    {
+      "epoch": 0.7069655724579663,
+      "learning_rate": 1.9994834701589113e-05,
+      "loss": 0.3721,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7069655724579663,
+      "learning_rate": 1.9994375823958504e-05,
+      "loss": 0.1685,
+      "step": 3532
+    },
+    {
+      "epoch": 0.7077662129703763,
+      "learning_rate": 1.9993897427095378e-05,
+      "loss": 0.1963,
+      "step": 3534
+    },
+    {
+      "epoch": 0.7077662129703763,
+      "learning_rate": 1.999339951193407e-05,
+      "loss": 0.3616,
+      "step": 3536
+    },
+    {
+      "epoch": 0.7085668534827863,
+      "learning_rate": 1.999288207944701e-05,
+      "loss": 0.4263,
+      "step": 3538
+    },
+    {
+      "epoch": 0.7085668534827863,
+      "learning_rate": 1.999234513064475e-05,
+      "loss": 0.1666,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7093674939951962,
+      "learning_rate": 1.999178866657597e-05,
+      "loss": 0.5756,
+      "step": 3542
+    },
+    {
+      "epoch": 0.7093674939951962,
+      "learning_rate": 1.9991212688327456e-05,
+      "loss": 0.1601,
+      "step": 3544
+    },
+    {
+      "epoch": 0.7101681345076061,
+      "learning_rate": 1.9990617197024103e-05,
+      "loss": 0.176,
+      "step": 3546
+    },
+    {
+      "epoch": 0.7101681345076061,
+      "learning_rate": 1.9990002193828923e-05,
+      "loss": 0.1501,
+      "step": 3548
+    },
+    {
+      "epoch": 0.710968775020016,
+      "learning_rate": 1.998936767994303e-05,
+      "loss": 0.1513,
+      "step": 3550
+    },
+    {
+      "epoch": 0.710968775020016,
+      "learning_rate": 1.9988713656605635e-05,
+      "loss": 0.0448,
+      "step": 3552
+    },
+    {
+      "epoch": 0.7117694155324259,
+      "learning_rate": 1.998804012509407e-05,
+      "loss": 0.2052,
+      "step": 3554
+    },
+    {
+      "epoch": 0.7117694155324259,
+      "learning_rate": 1.998734708672375e-05,
+      "loss": 0.1983,
+      "step": 3556
+    },
+    {
+      "epoch": 0.7125700560448359,
+      "learning_rate": 1.99866345428482e-05,
+      "loss": 0.1946,
+      "step": 3558
+    },
+    {
+      "epoch": 0.7125700560448359,
+      "learning_rate": 1.9985902494859026e-05,
+      "loss": 0.1585,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7133706965572458,
+      "learning_rate": 1.998515094418594e-05,
+      "loss": 0.278,
+      "step": 3562
+    },
+    {
+      "epoch": 0.7133706965572458,
+      "learning_rate": 1.9984379892296735e-05,
+      "loss": 0.0863,
+      "step": 3564
+    },
+    {
+      "epoch": 0.7141713370696557,
+      "learning_rate": 1.9983589340697288e-05,
+      "loss": 0.5755,
+      "step": 3566
+    },
+    {
+      "epoch": 0.7141713370696557,
+      "learning_rate": 1.9982779290931572e-05,
+      "loss": 0.0929,
+      "step": 3568
+    },
+    {
+      "epoch": 0.7149719775820657,
+      "learning_rate": 1.9981949744581622e-05,
+      "loss": 0.1784,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7149719775820657,
+      "learning_rate": 1.9981100703267567e-05,
+      "loss": 0.0969,
+      "step": 3572
+    },
+    {
+      "epoch": 0.7157726180944756,
+      "learning_rate": 1.99802321686476e-05,
+      "loss": 0.229,
+      "step": 3574
+    },
+    {
+      "epoch": 0.7157726180944756,
+      "learning_rate": 1.997934414241799e-05,
+      "loss": 0.4519,
+      "step": 3576
+    },
+    {
+      "epoch": 0.7165732586068855,
+      "learning_rate": 1.9978436626313068e-05,
+      "loss": 0.9369,
+      "step": 3578
+    },
+    {
+      "epoch": 0.7165732586068855,
+      "learning_rate": 1.9977509622105236e-05,
+      "loss": 0.0366,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7173738991192954,
+      "learning_rate": 1.997656313160495e-05,
+      "loss": 0.059,
+      "step": 3582
+    },
+    {
+      "epoch": 0.7173738991192954,
+      "learning_rate": 1.997559715666073e-05,
+      "loss": 0.0156,
+      "step": 3584
+    },
+    {
+      "epoch": 0.7181745396317054,
+      "learning_rate": 1.9974611699159142e-05,
+      "loss": 0.4294,
+      "step": 3586
+    },
+    {
+      "epoch": 0.7181745396317054,
+      "learning_rate": 1.9973606761024813e-05,
+      "loss": 0.233,
+      "step": 3588
+    },
+    {
+      "epoch": 0.7189751801441153,
+      "learning_rate": 1.99725823442204e-05,
+      "loss": 0.168,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7189751801441153,
+      "learning_rate": 1.997153845074662e-05,
+      "loss": 0.0714,
+      "step": 3592
+    },
+    {
+      "epoch": 0.7197758206565252,
+      "learning_rate": 1.9970475082642212e-05,
+      "loss": 0.1761,
+      "step": 3594
+    },
+    {
+      "epoch": 0.7197758206565252,
+      "learning_rate": 1.9969392241983957e-05,
+      "loss": 0.198,
+      "step": 3596
+    },
+    {
+      "epoch": 0.7205764611689351,
+      "learning_rate": 1.9968289930886675e-05,
+      "loss": 0.0076,
+      "step": 3598
+    },
+    {
+      "epoch": 0.7205764611689351,
+      "learning_rate": 1.9967168151503193e-05,
+      "loss": 0.0173,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7213771016813451,
+      "learning_rate": 1.9966026906024377e-05,
+      "loss": 0.2869,
+      "step": 3602
+    },
+    {
+      "epoch": 0.7213771016813451,
+      "learning_rate": 1.996486619667911e-05,
+      "loss": 0.5903,
+      "step": 3604
+    },
+    {
+      "epoch": 0.722177742193755,
+      "learning_rate": 1.9963686025734262e-05,
+      "loss": 0.1797,
+      "step": 3606
+    },
+    {
+      "epoch": 0.722177742193755,
+      "learning_rate": 1.9962486395494753e-05,
+      "loss": 0.044,
+      "step": 3608
+    },
+    {
+      "epoch": 0.7229783827061649,
+      "learning_rate": 1.9961267308303473e-05,
+      "loss": 0.6855,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7229783827061649,
+      "learning_rate": 1.9960028766541336e-05,
+      "loss": 0.6294,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7237790232185749,
+      "learning_rate": 1.9958770772627236e-05,
+      "loss": 0.3414,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7237790232185749,
+      "learning_rate": 1.9957493329018064e-05,
+      "loss": 0.2379,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7245796637309848,
+      "learning_rate": 1.9956196438208693e-05,
+      "loss": 0.455,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7245796637309848,
+      "learning_rate": 1.995488010273198e-05,
+      "loss": 0.2024,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7253803042433947,
+      "learning_rate": 1.9953544325158755e-05,
+      "loss": 0.1667,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7253803042433947,
+      "learning_rate": 1.9952189108097825e-05,
+      "loss": 0.471,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7261809447558046,
+      "learning_rate": 1.9950814454195953e-05,
+      "loss": 0.3317,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7261809447558046,
+      "learning_rate": 1.9949420366137873e-05,
+      "loss": 0.3127,
+      "step": 3628
+    },
+    {
+      "epoch": 0.7269815852682145,
+      "learning_rate": 1.9948006846646262e-05,
+      "loss": 0.168,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7269815852682145,
+      "learning_rate": 1.994657389848176e-05,
+      "loss": 0.7609,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7277822257806245,
+      "learning_rate": 1.9945121524442947e-05,
+      "loss": 0.4163,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7277822257806245,
+      "learning_rate": 1.994364972736634e-05,
+      "loss": 0.3349,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7285828662930345,
+      "learning_rate": 1.9942158510126384e-05,
+      "loss": 0.5226,
+      "step": 3638
+    },
+    {
+      "epoch": 0.7285828662930345,
+      "learning_rate": 1.9940647875635466e-05,
+      "loss": 0.2286,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7293835068054444,
+      "learning_rate": 1.9939117826843887e-05,
+      "loss": 0.5597,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7293835068054444,
+      "learning_rate": 1.993756836673986e-05,
+      "loss": 0.2156,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7301841473178543,
+      "learning_rate": 1.9935999498349525e-05,
+      "loss": 0.3924,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7301841473178543,
+      "learning_rate": 1.99344112247369e-05,
+      "loss": 0.5538,
+      "step": 3648
+    },
+    {
+      "epoch": 0.7309847878302642,
+      "learning_rate": 1.9932803549003932e-05,
+      "loss": 0.2051,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7309847878302642,
+      "learning_rate": 1.9931176474290438e-05,
+      "loss": 0.3971,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7317854283426741,
+      "learning_rate": 1.9929530003774136e-05,
+      "loss": 0.1989,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7317854283426741,
+      "learning_rate": 1.9927864140670618e-05,
+      "loss": 0.2877,
+      "step": 3656
+    },
+    {
+      "epoch": 0.732586068855084,
+      "learning_rate": 1.9926178888233344e-05,
+      "loss": 0.3857,
+      "step": 3658
+    },
+    {
+      "epoch": 0.732586068855084,
+      "learning_rate": 1.9924474249753656e-05,
+      "loss": 0.4158,
+      "step": 3660
+    },
+    {
+      "epoch": 0.733386709367494,
+      "learning_rate": 1.9922750228560746e-05,
+      "loss": 0.3614,
+      "step": 3662
+    },
+    {
+      "epoch": 0.733386709367494,
+      "learning_rate": 1.9921006828021666e-05,
+      "loss": 0.2104,
+      "step": 3664
+    },
+    {
+      "epoch": 0.734187349879904,
+      "learning_rate": 1.9919244051541315e-05,
+      "loss": 0.3535,
+      "step": 3666
+    },
+    {
+      "epoch": 0.734187349879904,
+      "learning_rate": 1.9917461902562435e-05,
+      "loss": 0.1195,
+      "step": 3668
+    },
+    {
+      "epoch": 0.7349879903923139,
+      "learning_rate": 1.9915660384565603e-05,
+      "loss": 0.3613,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7349879903923139,
+      "learning_rate": 1.9913839501069213e-05,
+      "loss": 0.166,
+      "step": 3672
+    },
+    {
+      "epoch": 0.7357886309047238,
+      "learning_rate": 1.9911999255629504e-05,
+      "loss": 0.1989,
+      "step": 3674
+    },
+    {
+      "epoch": 0.7357886309047238,
+      "learning_rate": 1.9910139651840497e-05,
+      "loss": 0.2196,
+      "step": 3676
+    },
+    {
+      "epoch": 0.7365892714171337,
+      "learning_rate": 1.990826069333406e-05,
+      "loss": 0.4604,
+      "step": 3678
+    },
+    {
+      "epoch": 0.7365892714171337,
+      "learning_rate": 1.9906362383779826e-05,
+      "loss": 0.1984,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7373899119295436,
+      "learning_rate": 1.9904444726885236e-05,
+      "loss": 0.1669,
+      "step": 3682
+    },
+    {
+      "epoch": 0.7373899119295436,
+      "learning_rate": 1.9902507726395524e-05,
+      "loss": 0.082,
+      "step": 3684
+    },
+    {
+      "epoch": 0.7381905524419535,
+      "learning_rate": 1.9900551386093677e-05,
+      "loss": 0.2931,
+      "step": 3686
+    },
+    {
+      "epoch": 0.7381905524419535,
+      "learning_rate": 1.989857570980049e-05,
+      "loss": 0.2164,
+      "step": 3688
+    },
+    {
+      "epoch": 0.7389911929543634,
+      "learning_rate": 1.9896580701374482e-05,
+      "loss": 0.1678,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7389911929543634,
+      "learning_rate": 1.9894566364711965e-05,
+      "loss": 0.0201,
+      "step": 3692
+    },
+    {
+      "epoch": 0.7397918334667735,
+      "learning_rate": 1.9892532703746977e-05,
+      "loss": 0.3052,
+      "step": 3694
+    },
+    {
+      "epoch": 0.7397918334667735,
+      "learning_rate": 1.9890479722451292e-05,
+      "loss": 0.3427,
+      "step": 3696
+    },
+    {
+      "epoch": 0.7405924739791834,
+      "learning_rate": 1.9888407424834437e-05,
+      "loss": 0.4775,
+      "step": 3698
+    },
+    {
+      "epoch": 0.7405924739791834,
+      "learning_rate": 1.988631581494365e-05,
+      "loss": 0.4975,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7413931144915933,
+      "learning_rate": 1.9884204896863895e-05,
+      "loss": 0.4598,
+      "step": 3702
+    },
+    {
+      "epoch": 0.7413931144915933,
+      "learning_rate": 1.9882074674717832e-05,
+      "loss": 0.3838,
+      "step": 3704
+    },
+    {
+      "epoch": 0.7421937550040032,
+      "learning_rate": 1.9879925152665845e-05,
+      "loss": 0.1849,
+      "step": 3706
+    },
+    {
+      "epoch": 0.7421937550040032,
+      "learning_rate": 1.987775633490599e-05,
+      "loss": 0.2017,
+      "step": 3708
+    },
+    {
+      "epoch": 0.7429943955164131,
+      "learning_rate": 1.9875568225674005e-05,
+      "loss": 0.1008,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7429943955164131,
+      "learning_rate": 1.987336082924333e-05,
+      "loss": 0.0474,
+      "step": 3712
+    },
+    {
+      "epoch": 0.743795036028823,
+      "learning_rate": 1.987113414992505e-05,
+      "loss": 0.1762,
+      "step": 3714
+    },
+    {
+      "epoch": 0.743795036028823,
+      "learning_rate": 1.986888819206792e-05,
+      "loss": 0.432,
+      "step": 3716
+    },
+    {
+      "epoch": 0.7445956765412329,
+      "learning_rate": 1.986662296005834e-05,
+      "loss": 0.465,
+      "step": 3718
+    },
+    {
+      "epoch": 0.7445956765412329,
+      "learning_rate": 1.986433845832037e-05,
+      "loss": 0.0209,
+      "step": 3720
+    },
+    {
+      "epoch": 0.745396317053643,
+      "learning_rate": 1.9862034691315678e-05,
+      "loss": 0.2399,
+      "step": 3722
+    },
+    {
+      "epoch": 0.745396317053643,
+      "learning_rate": 1.9859711663543573e-05,
+      "loss": 0.1987,
+      "step": 3724
+    },
+    {
+      "epoch": 0.7461969575660529,
+      "learning_rate": 1.9857369379540985e-05,
+      "loss": 0.5046,
+      "step": 3726
+    },
+    {
+      "epoch": 0.7461969575660529,
+      "learning_rate": 1.9855007843882437e-05,
+      "loss": 0.2266,
+      "step": 3728
+    },
+    {
+      "epoch": 0.7469975980784628,
+      "learning_rate": 1.985262706118007e-05,
+      "loss": 0.2417,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7469975980784628,
+      "learning_rate": 1.9850227036083592e-05,
+      "loss": 0.1269,
+      "step": 3732
+    },
+    {
+      "epoch": 0.7477982385908727,
+      "learning_rate": 1.9847807773280314e-05,
+      "loss": 0.1664,
+      "step": 3734
+    },
+    {
+      "epoch": 0.7477982385908727,
+      "learning_rate": 1.9845369277495105e-05,
+      "loss": 0.1948,
+      "step": 3736
+    },
+    {
+      "epoch": 0.7485988791032826,
+      "learning_rate": 1.9842911553490396e-05,
+      "loss": 0.3393,
+      "step": 3738
+    },
+    {
+      "epoch": 0.7485988791032826,
+      "learning_rate": 1.9840434606066186e-05,
+      "loss": 1.2607,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7493995196156925,
+      "learning_rate": 1.983793844005999e-05,
+      "loss": 0.0891,
+      "step": 3742
+    },
+    {
+      "epoch": 0.7493995196156925,
+      "learning_rate": 1.9835423060346892e-05,
+      "loss": 0.0716,
+      "step": 3744
+    },
+    {
+      "epoch": 0.7502001601281025,
+      "learning_rate": 1.9832888471839475e-05,
+      "loss": 0.3133,
+      "step": 3746
+    },
+    {
+      "epoch": 0.7502001601281025,
+      "learning_rate": 1.983033467948784e-05,
+      "loss": 0.3731,
+      "step": 3748
+    },
+    {
+      "epoch": 0.7510008006405124,
+      "learning_rate": 1.9827761688279613e-05,
+      "loss": 0.2168,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7510008006405124,
+      "learning_rate": 1.9825169503239885e-05,
+      "loss": 0.3082,
+      "step": 3752
+    },
+    {
+      "epoch": 0.7518014411529224,
+      "learning_rate": 1.9822558129431263e-05,
+      "loss": 0.5453,
+      "step": 3754
+    },
+    {
+      "epoch": 0.7518014411529224,
+      "learning_rate": 1.9819927571953804e-05,
+      "loss": 0.2924,
+      "step": 3756
+    },
+    {
+      "epoch": 0.7526020816653323,
+      "learning_rate": 1.981727783594506e-05,
+      "loss": 0.1903,
+      "step": 3758
+    },
+    {
+      "epoch": 0.7526020816653323,
+      "learning_rate": 1.9814608926580007e-05,
+      "loss": 0.0273,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7534027221777422,
+      "learning_rate": 1.9811920849071092e-05,
+      "loss": 0.2325,
+      "step": 3762
+    },
+    {
+      "epoch": 0.7534027221777422,
+      "learning_rate": 1.980921360866819e-05,
+      "loss": 0.4156,
+      "step": 3764
+    },
+    {
+      "epoch": 0.7542033626901521,
+      "learning_rate": 1.980648721065859e-05,
+      "loss": 0.3152,
+      "step": 3766
+    },
+    {
+      "epoch": 0.7542033626901521,
+      "learning_rate": 1.9803741660367018e-05,
+      "loss": 0.1907,
+      "step": 3768
+    },
+    {
+      "epoch": 0.755004003202562,
+      "learning_rate": 1.980097696315558e-05,
+      "loss": 0.3841,
+      "step": 3770
+    },
+    {
+      "epoch": 0.755004003202562,
+      "learning_rate": 1.9798193124423804e-05,
+      "loss": 0.0873,
+      "step": 3772
+    },
+    {
+      "epoch": 0.755804643714972,
+      "learning_rate": 1.979539014960858e-05,
+      "loss": 0.4258,
+      "step": 3774
+    },
+    {
+      "epoch": 0.755804643714972,
+      "learning_rate": 1.979256804418418e-05,
+      "loss": 0.1078,
+      "step": 3776
+    },
+    {
+      "epoch": 0.7566052842273819,
+      "learning_rate": 1.9789726813662233e-05,
+      "loss": 0.3728,
+      "step": 3778
+    },
+    {
+      "epoch": 0.7566052842273819,
+      "learning_rate": 1.978686646359173e-05,
+      "loss": 0.3641,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7574059247397918,
+      "learning_rate": 1.9783986999558994e-05,
+      "loss": 0.5143,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7574059247397918,
+      "learning_rate": 1.9781088427187677e-05,
+      "loss": 0.4478,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7582065652522018,
+      "learning_rate": 1.9778170752138763e-05,
+      "loss": 0.1869,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7582065652522018,
+      "learning_rate": 1.9775233980110524e-05,
+      "loss": 0.064,
+      "step": 3788
+    },
+    {
+      "epoch": 0.7590072057646117,
+      "learning_rate": 1.9772278116838546e-05,
+      "loss": 0.1504,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7590072057646117,
+      "learning_rate": 1.976930316809569e-05,
+      "loss": 0.0338,
+      "step": 3792
+    },
+    {
+      "epoch": 0.7598078462770216,
+      "learning_rate": 1.97663091396921e-05,
+      "loss": 0.4835,
+      "step": 3794
+    },
+    {
+      "epoch": 0.7598078462770216,
+      "learning_rate": 1.9763296037475177e-05,
+      "loss": 1.1043,
+      "step": 3796
+    },
+    {
+      "epoch": 0.7606084867894315,
+      "learning_rate": 1.976026386732957e-05,
+      "loss": 0.3634,
+      "step": 3798
+    },
+    {
+      "epoch": 0.7606084867894315,
+      "learning_rate": 1.9757212635177177e-05,
+      "loss": 0.4011,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7614091273018415,
+      "learning_rate": 1.9754142346977122e-05,
+      "loss": 0.1789,
+      "step": 3802
+    },
+    {
+      "epoch": 0.7614091273018415,
+      "learning_rate": 1.9751053008725736e-05,
+      "loss": 0.0058,
+      "step": 3804
+    },
+    {
+      "epoch": 0.7622097678142514,
+      "learning_rate": 1.9747944626456577e-05,
+      "loss": 0.6204,
+      "step": 3806
+    },
+    {
+      "epoch": 0.7622097678142514,
+      "learning_rate": 1.9744817206240374e-05,
+      "loss": 0.2192,
+      "step": 3808
+    },
+    {
+      "epoch": 0.7630104083266613,
+      "learning_rate": 1.9741670754185054e-05,
+      "loss": 0.1905,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7630104083266613,
+      "learning_rate": 1.9738505276435695e-05,
+      "loss": 0.7769,
+      "step": 3812
+    },
+    {
+      "epoch": 0.7638110488390712,
+      "learning_rate": 1.9735320779174548e-05,
+      "loss": 0.571,
+      "step": 3814
+    },
+    {
+      "epoch": 0.7638110488390712,
+      "learning_rate": 1.9732117268621005e-05,
+      "loss": 0.5202,
+      "step": 3816
+    },
+    {
+      "epoch": 0.7646116893514812,
+      "learning_rate": 1.9728894751031595e-05,
+      "loss": 0.5054,
+      "step": 3818
+    },
+    {
+      "epoch": 0.7646116893514812,
+      "learning_rate": 1.9725653232699962e-05,
+      "loss": 0.2107,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7654123298638911,
+      "learning_rate": 1.972239271995686e-05,
+      "loss": 0.2162,
+      "step": 3822
+    },
+    {
+      "epoch": 0.7654123298638911,
+      "learning_rate": 1.9719113219170152e-05,
+      "loss": 0.6044,
+      "step": 3824
+    },
+    {
+      "epoch": 0.7662129703763011,
+      "learning_rate": 1.9715814736744758e-05,
+      "loss": 0.2619,
+      "step": 3826
+    },
+    {
+      "epoch": 0.7662129703763011,
+      "learning_rate": 1.9712497279122692e-05,
+      "loss": 0.1986,
+      "step": 3828
+    },
+    {
+      "epoch": 0.767013610888711,
+      "learning_rate": 1.9709160852783022e-05,
+      "loss": 0.229,
+      "step": 3830
+    },
+    {
+      "epoch": 0.767013610888711,
+      "learning_rate": 1.970580546424186e-05,
+      "loss": 0.4985,
+      "step": 3832
+    },
+    {
+      "epoch": 0.7678142514011209,
+      "learning_rate": 1.9702431120052352e-05,
+      "loss": 0.1767,
+      "step": 3834
+    },
+    {
+      "epoch": 0.7678142514011209,
+      "learning_rate": 1.969903782680467e-05,
+      "loss": 0.1224,
+      "step": 3836
+    },
+    {
+      "epoch": 0.7686148919135308,
+      "learning_rate": 1.9695625591125984e-05,
+      "loss": 0.4602,
+      "step": 3838
+    },
+    {
+      "epoch": 0.7686148919135308,
+      "learning_rate": 1.9692194419680463e-05,
+      "loss": 0.3979,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7694155324259407,
+      "learning_rate": 1.968874431916926e-05,
+      "loss": 0.3618,
+      "step": 3842
+    },
+    {
+      "epoch": 0.7694155324259407,
+      "learning_rate": 1.96852752963305e-05,
+      "loss": 0.1771,
+      "step": 3844
+    },
+    {
+      "epoch": 0.7702161729383507,
+      "learning_rate": 1.9681787357939257e-05,
+      "loss": 0.2169,
+      "step": 3846
+    },
+    {
+      "epoch": 0.7702161729383507,
+      "learning_rate": 1.9678280510807552e-05,
+      "loss": 0.6681,
+      "step": 3848
+    },
+    {
+      "epoch": 0.7710168134507606,
+      "learning_rate": 1.9674754761784334e-05,
+      "loss": 0.1674,
+      "step": 3850
+    },
+    {
+      "epoch": 0.7710168134507606,
+      "learning_rate": 1.9671210117755462e-05,
+      "loss": 0.3857,
+      "step": 3852
+    },
+    {
+      "epoch": 0.7718174539631706,
+      "learning_rate": 1.9667646585643706e-05,
+      "loss": 0.4017,
+      "step": 3854
+    },
+    {
+      "epoch": 0.7718174539631706,
+      "learning_rate": 1.966406417240872e-05,
+      "loss": 0.298,
+      "step": 3856
+    },
+    {
+      "epoch": 0.7726180944755805,
+      "learning_rate": 1.966046288504704e-05,
+      "loss": 0.2642,
+      "step": 3858
+    },
+    {
+      "epoch": 0.7726180944755805,
+      "learning_rate": 1.9656842730592046e-05,
+      "loss": 0.1519,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7734187349879904,
+      "learning_rate": 1.965320371611399e-05,
+      "loss": 0.1718,
+      "step": 3862
+    },
+    {
+      "epoch": 0.7734187349879904,
+      "learning_rate": 1.964954584871995e-05,
+      "loss": 0.393,
+      "step": 3864
+    },
+    {
+      "epoch": 0.7742193755004003,
+      "learning_rate": 1.964586913555381e-05,
+      "loss": 0.23,
+      "step": 3866
+    },
+    {
+      "epoch": 0.7742193755004003,
+      "learning_rate": 1.9642173583796265e-05,
+      "loss": 0.1498,
+      "step": 3868
+    },
+    {
+      "epoch": 0.7750200160128102,
+      "learning_rate": 1.9638459200664822e-05,
+      "loss": 0.1593,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7750200160128102,
+      "learning_rate": 1.9634725993413744e-05,
+      "loss": 0.2157,
+      "step": 3872
+    },
+    {
+      "epoch": 0.7758206565252201,
+      "learning_rate": 1.9630973969334068e-05,
+      "loss": 0.3213,
+      "step": 3874
+    },
+    {
+      "epoch": 0.7758206565252201,
+      "learning_rate": 1.9627203135753576e-05,
+      "loss": 0.0939,
+      "step": 3876
+    },
+    {
+      "epoch": 0.77662129703763,
+      "learning_rate": 1.9623413500036795e-05,
+      "loss": 0.3834,
+      "step": 3878
+    },
+    {
+      "epoch": 0.77662129703763,
+      "learning_rate": 1.9619605069584954e-05,
+      "loss": 0.253,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7774219375500401,
+      "learning_rate": 1.9615777851836007e-05,
+      "loss": 0.224,
+      "step": 3882
+    },
+    {
+      "epoch": 0.7774219375500401,
+      "learning_rate": 1.961193185426459e-05,
+      "loss": 0.3822,
+      "step": 3884
+    },
+    {
+      "epoch": 0.77822257806245,
+      "learning_rate": 1.9608067084382025e-05,
+      "loss": 0.2108,
+      "step": 3886
+    },
+    {
+      "epoch": 0.77822257806245,
+      "learning_rate": 1.9604183549736287e-05,
+      "loss": 0.1212,
+      "step": 3888
+    },
+    {
+      "epoch": 0.7790232185748599,
+      "learning_rate": 1.9600281257912002e-05,
+      "loss": 0.2463,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7790232185748599,
+      "learning_rate": 1.959636021653044e-05,
+      "loss": 0.198,
+      "step": 3892
+    },
+    {
+      "epoch": 0.7798238590872698,
+      "learning_rate": 1.9592420433249465e-05,
+      "loss": 0.3772,
+      "step": 3894
+    },
+    {
+      "epoch": 0.7798238590872698,
+      "learning_rate": 1.958846191576357e-05,
+      "loss": 0.1502,
+      "step": 3896
+    },
+    {
+      "epoch": 0.7806244995996797,
+      "learning_rate": 1.958448467180382e-05,
+      "loss": 0.3887,
+      "step": 3898
+    },
+    {
+      "epoch": 0.7806244995996797,
+      "learning_rate": 1.958048870913786e-05,
+      "loss": 1.0372,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7814251401120896,
+      "learning_rate": 1.9576474035569895e-05,
+      "loss": 0.2238,
+      "step": 3902
+    },
+    {
+      "epoch": 0.7814251401120896,
+      "learning_rate": 1.9572440658940667e-05,
+      "loss": 0.0693,
+      "step": 3904
+    },
+    {
+      "epoch": 0.7822257806244995,
+      "learning_rate": 1.9568388587127448e-05,
+      "loss": 0.5662,
+      "step": 3906
+    },
+    {
+      "epoch": 0.7822257806244995,
+      "learning_rate": 1.9564317828044022e-05,
+      "loss": 0.333,
+      "step": 3908
+    },
+    {
+      "epoch": 0.7830264211369096,
+      "learning_rate": 1.9560228389640668e-05,
+      "loss": 0.4041,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7830264211369096,
+      "learning_rate": 1.955612027990415e-05,
+      "loss": 0.2359,
+      "step": 3912
+    },
+    {
+      "epoch": 0.7838270616493195,
+      "learning_rate": 1.955199350685769e-05,
+      "loss": 0.4231,
+      "step": 3914
+    },
+    {
+      "epoch": 0.7838270616493195,
+      "learning_rate": 1.9547848078560982e-05,
+      "loss": 0.1687,
+      "step": 3916
+    },
+    {
+      "epoch": 0.7846277021617294,
+      "learning_rate": 1.954368400311011e-05,
+      "loss": 0.1581,
+      "step": 3918
+    },
+    {
+      "epoch": 0.7846277021617294,
+      "learning_rate": 1.953950128863763e-05,
+      "loss": 0.3495,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7854283426741393,
+      "learning_rate": 1.9535299943312455e-05,
+      "loss": 0.3632,
+      "step": 3922
+    },
+    {
+      "epoch": 0.7854283426741393,
+      "learning_rate": 1.9531079975339915e-05,
+      "loss": 0.2938,
+      "step": 3924
+    },
+    {
+      "epoch": 0.7862289831865492,
+      "learning_rate": 1.9526841392961694e-05,
+      "loss": 0.104,
+      "step": 3926
+    },
+    {
+      "epoch": 0.7862289831865492,
+      "learning_rate": 1.9522584204455835e-05,
+      "loss": 0.0706,
+      "step": 3928
+    },
+    {
+      "epoch": 0.7870296236989591,
+      "learning_rate": 1.9518308418136728e-05,
+      "loss": 0.2953,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7870296236989591,
+      "learning_rate": 1.9514014042355054e-05,
+      "loss": 0.2781,
+      "step": 3932
+    },
+    {
+      "epoch": 0.7878302642113691,
+      "learning_rate": 1.9509701085497852e-05,
+      "loss": 0.3613,
+      "step": 3934
+    },
+    {
+      "epoch": 0.7878302642113691,
+      "learning_rate": 1.9505369555988395e-05,
+      "loss": 0.8307,
+      "step": 3936
+    },
+    {
+      "epoch": 0.7886309047237791,
+      "learning_rate": 1.9501019462286266e-05,
+      "loss": 0.2325,
+      "step": 3938
+    },
+    {
+      "epoch": 0.7886309047237791,
+      "learning_rate": 1.9496650812887293e-05,
+      "loss": 0.3829,
+      "step": 3940
+    },
+    {
+      "epoch": 0.789431545236189,
+      "learning_rate": 1.9492263616323536e-05,
+      "loss": 0.2862,
+      "step": 3942
+    },
+    {
+      "epoch": 0.789431545236189,
+      "learning_rate": 1.9487857881163295e-05,
+      "loss": 0.1871,
+      "step": 3944
+    },
+    {
+      "epoch": 0.7902321857485989,
+      "learning_rate": 1.948343361601105e-05,
+      "loss": 0.2966,
+      "step": 3946
+    },
+    {
+      "epoch": 0.7902321857485989,
+      "learning_rate": 1.947899082950751e-05,
+      "loss": 0.3186,
+      "step": 3948
+    },
+    {
+      "epoch": 0.7910328262610088,
+      "learning_rate": 1.947452953032951e-05,
+      "loss": 0.2294,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7910328262610088,
+      "learning_rate": 1.947004972719008e-05,
+      "loss": 0.335,
+      "step": 3952
+    },
+    {
+      "epoch": 0.7918334667734187,
+      "learning_rate": 1.9465551428838363e-05,
+      "loss": 0.1959,
+      "step": 3954
+    },
+    {
+      "epoch": 0.7918334667734187,
+      "learning_rate": 1.9461034644059637e-05,
+      "loss": 0.534,
+      "step": 3956
+    },
+    {
+      "epoch": 0.7926341072858286,
+      "learning_rate": 1.9456499381675285e-05,
+      "loss": 0.1668,
+      "step": 3958
+    },
+    {
+      "epoch": 0.7926341072858286,
+      "learning_rate": 1.945194565054276e-05,
+      "loss": 0.1361,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7934347477982386,
+      "learning_rate": 1.9447373459555617e-05,
+      "loss": 0.2156,
+      "step": 3962
+    },
+    {
+      "epoch": 0.7934347477982386,
+      "learning_rate": 1.9442782817643425e-05,
+      "loss": 0.1359,
+      "step": 3964
+    },
+    {
+      "epoch": 0.7942353883106485,
+      "learning_rate": 1.9438173733771814e-05,
+      "loss": 0.168,
+      "step": 3966
+    },
+    {
+      "epoch": 0.7942353883106485,
+      "learning_rate": 1.9433546216942433e-05,
+      "loss": 0.596,
+      "step": 3968
+    },
+    {
+      "epoch": 0.7950360288230585,
+      "learning_rate": 1.9428900276192903e-05,
+      "loss": 0.2781,
+      "step": 3970
+    },
+    {
+      "epoch": 0.7950360288230585,
+      "learning_rate": 1.942423592059687e-05,
+      "loss": 0.7508,
+      "step": 3972
+    },
+    {
+      "epoch": 0.7958366693354684,
+      "learning_rate": 1.94195531592639e-05,
+      "loss": 0.2635,
+      "step": 3974
+    },
+    {
+      "epoch": 0.7958366693354684,
+      "learning_rate": 1.941485200133955e-05,
+      "loss": 0.0822,
+      "step": 3976
+    },
+    {
+      "epoch": 0.7966373098478783,
+      "learning_rate": 1.9410132456005262e-05,
+      "loss": 0.2761,
+      "step": 3978
+    },
+    {
+      "epoch": 0.7966373098478783,
+      "learning_rate": 1.9405394532478422e-05,
+      "loss": 0.5745,
+      "step": 3980
+    },
+    {
+      "epoch": 0.7974379503602882,
+      "learning_rate": 1.94006382400123e-05,
+      "loss": 0.3888,
+      "step": 3982
+    },
+    {
+      "epoch": 0.7974379503602882,
+      "learning_rate": 1.9395863587896025e-05,
+      "loss": 0.1216,
+      "step": 3984
+    },
+    {
+      "epoch": 0.7982385908726981,
+      "learning_rate": 1.939107058545461e-05,
+      "loss": 0.1513,
+      "step": 3986
+    },
+    {
+      "epoch": 0.7982385908726981,
+      "learning_rate": 1.938625924204888e-05,
+      "loss": 0.0319,
+      "step": 3988
+    },
+    {
+      "epoch": 0.7990392313851081,
+      "learning_rate": 1.9381429567075507e-05,
+      "loss": 0.503,
+      "step": 3990
+    },
+    {
+      "epoch": 0.7990392313851081,
+      "learning_rate": 1.937658156996694e-05,
+      "loss": 0.4072,
+      "step": 3992
+    },
+    {
+      "epoch": 0.799839871897518,
+      "learning_rate": 1.9371715260191425e-05,
+      "loss": 0.112,
+      "step": 3994
+    },
+    {
+      "epoch": 0.799839871897518,
+      "learning_rate": 1.9366830647252977e-05,
+      "loss": 0.1082,
+      "step": 3996
+    },
+    {
+      "epoch": 0.800640512409928,
+      "learning_rate": 1.936192774069133e-05,
+      "loss": 0.3999,
+      "step": 3998
+    },
+    {
+      "epoch": 0.800640512409928,
+      "learning_rate": 1.9357006550082e-05,
+      "loss": 0.1527,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8014411529223379,
+      "learning_rate": 1.9352067085036145e-05,
+      "loss": 0.3613,
+      "step": 4002
+    },
+    {
+      "epoch": 0.8014411529223379,
+      "learning_rate": 1.9347109355200676e-05,
+      "loss": 0.1511,
+      "step": 4004
+    },
+    {
+      "epoch": 0.8022417934347478,
+      "learning_rate": 1.9342133370258124e-05,
+      "loss": 0.2964,
+      "step": 4006
+    },
+    {
+      "epoch": 0.8022417934347478,
+      "learning_rate": 1.933713913992671e-05,
+      "loss": 0.0088,
+      "step": 4008
+    },
+    {
+      "epoch": 0.8030424339471577,
+      "learning_rate": 1.9332126673960276e-05,
+      "loss": 0.4076,
+      "step": 4010
+    },
+    {
+      "epoch": 0.8030424339471577,
+      "learning_rate": 1.9327095982148255e-05,
+      "loss": 0.2023,
+      "step": 4012
+    },
+    {
+      "epoch": 0.8038430744595677,
+      "learning_rate": 1.932204707431572e-05,
+      "loss": 0.3831,
+      "step": 4014
+    },
+    {
+      "epoch": 0.8038430744595677,
+      "learning_rate": 1.9316979960323283e-05,
+      "loss": 0.3714,
+      "step": 4016
+    },
+    {
+      "epoch": 0.8046437149719776,
+      "learning_rate": 1.9311894650067146e-05,
+      "loss": 0.6496,
+      "step": 4018
+    },
+    {
+      "epoch": 0.8046437149719776,
+      "learning_rate": 1.9306791153479017e-05,
+      "loss": 0.6367,
+      "step": 4020
+    },
+    {
+      "epoch": 0.8054443554843875,
+      "learning_rate": 1.9301669480526118e-05,
+      "loss": 0.8116,
+      "step": 4022
+    },
+    {
+      "epoch": 0.8054443554843875,
+      "learning_rate": 1.9296529641211226e-05,
+      "loss": 0.7964,
+      "step": 4024
+    },
+    {
+      "epoch": 0.8062449959967974,
+      "learning_rate": 1.929137164557252e-05,
+      "loss": 0.1502,
+      "step": 4026
+    },
+    {
+      "epoch": 0.8062449959967974,
+      "learning_rate": 1.928619550368371e-05,
+      "loss": 0.2104,
+      "step": 4028
+    },
+    {
+      "epoch": 0.8070456365092074,
+      "learning_rate": 1.9281001225653883e-05,
+      "loss": 0.7397,
+      "step": 4030
+    },
+    {
+      "epoch": 0.8070456365092074,
+      "learning_rate": 1.9275788821627607e-05,
+      "loss": 0.5979,
+      "step": 4032
+    },
+    {
+      "epoch": 0.8078462770216173,
+      "learning_rate": 1.9270558301784808e-05,
+      "loss": 0.3664,
+      "step": 4034
+    },
+    {
+      "epoch": 0.8078462770216173,
+      "learning_rate": 1.9265309676340783e-05,
+      "loss": 0.1687,
+      "step": 4036
+    },
+    {
+      "epoch": 0.8086469175340272,
+      "learning_rate": 1.9260042955546247e-05,
+      "loss": 0.5936,
+      "step": 4038
+    },
+    {
+      "epoch": 0.8086469175340272,
+      "learning_rate": 1.9254758149687187e-05,
+      "loss": 0.1326,
+      "step": 4040
+    },
+    {
+      "epoch": 0.8094475580464372,
+      "learning_rate": 1.9249455269084972e-05,
+      "loss": 0.3855,
+      "step": 4042
+    },
+    {
+      "epoch": 0.8094475580464372,
+      "learning_rate": 1.9244134324096216e-05,
+      "loss": 0.229,
+      "step": 4044
+    },
+    {
+      "epoch": 0.8102481985588471,
+      "learning_rate": 1.9238795325112867e-05,
+      "loss": 0.2195,
+      "step": 4046
+    },
+    {
+      "epoch": 0.8102481985588471,
+      "learning_rate": 1.9233438282562095e-05,
+      "loss": 0.4748,
+      "step": 4048
+    },
+    {
+      "epoch": 0.811048839071257,
+      "learning_rate": 1.9228063206906302e-05,
+      "loss": 0.2376,
+      "step": 4050
+    },
+    {
+      "epoch": 0.811048839071257,
+      "learning_rate": 1.9222670108643156e-05,
+      "loss": 0.2422,
+      "step": 4052
+    },
+    {
+      "epoch": 0.8118494795836669,
+      "learning_rate": 1.9217258998305464e-05,
+      "loss": 0.5496,
+      "step": 4054
+    },
+    {
+      "epoch": 0.8118494795836669,
+      "learning_rate": 1.9211829886461278e-05,
+      "loss": 0.1845,
+      "step": 4056
+    },
+    {
+      "epoch": 0.8126501200960768,
+      "learning_rate": 1.9206382783713735e-05,
+      "loss": 0.1581,
+      "step": 4058
+    },
+    {
+      "epoch": 0.8126501200960768,
+      "learning_rate": 1.9200917700701176e-05,
+      "loss": 0.3557,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8134507606084868,
+      "learning_rate": 1.9195434648097013e-05,
+      "loss": 0.2271,
+      "step": 4062
+    },
+    {
+      "epoch": 0.8134507606084868,
+      "learning_rate": 1.918993363660975e-05,
+      "loss": 0.2764,
+      "step": 4064
+    },
+    {
+      "epoch": 0.8142514011208967,
+      "learning_rate": 1.9184414676983013e-05,
+      "loss": 0.1428,
+      "step": 4066
+    },
+    {
+      "epoch": 0.8142514011208967,
+      "learning_rate": 1.9178877779995416e-05,
+      "loss": 0.2323,
+      "step": 4068
+    },
+    {
+      "epoch": 0.8150520416333067,
+      "learning_rate": 1.9173322956460678e-05,
+      "loss": 0.4734,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8150520416333067,
+      "learning_rate": 1.916775021722745e-05,
+      "loss": 0.6129,
+      "step": 4072
+    },
+    {
+      "epoch": 0.8158526821457166,
+      "learning_rate": 1.9162159573179446e-05,
+      "loss": 0.2513,
+      "step": 4074
+    },
+    {
+      "epoch": 0.8158526821457166,
+      "learning_rate": 1.9156551035235298e-05,
+      "loss": 0.0878,
+      "step": 4076
+    },
+    {
+      "epoch": 0.8166533226581265,
+      "learning_rate": 1.915092461434859e-05,
+      "loss": 0.1764,
+      "step": 4078
+    },
+    {
+      "epoch": 0.8166533226581265,
+      "learning_rate": 1.9145280321507872e-05,
+      "loss": 0.1586,
+      "step": 4080
+    },
+    {
+      "epoch": 0.8174539631705364,
+      "learning_rate": 1.9139618167736547e-05,
+      "loss": 0.3045,
+      "step": 4082
+    },
+    {
+      "epoch": 0.8174539631705364,
+      "learning_rate": 1.9133938164092942e-05,
+      "loss": 0.1982,
+      "step": 4084
+    },
+    {
+      "epoch": 0.8182546036829463,
+      "learning_rate": 1.912824032167022e-05,
+      "loss": 0.2082,
+      "step": 4086
+    },
+    {
+      "epoch": 0.8182546036829463,
+      "learning_rate": 1.9122524651596372e-05,
+      "loss": 0.2981,
+      "step": 4088
+    },
+    {
+      "epoch": 0.8190552441953562,
+      "learning_rate": 1.911679116503426e-05,
+      "loss": 0.2617,
+      "step": 4090
+    },
+    {
+      "epoch": 0.8190552441953562,
+      "learning_rate": 1.9111039873181475e-05,
+      "loss": 0.2895,
+      "step": 4092
+    },
+    {
+      "epoch": 0.8198558847077662,
+      "learning_rate": 1.9105270787270446e-05,
+      "loss": 0.5754,
+      "step": 4094
+    },
+    {
+      "epoch": 0.8198558847077662,
+      "learning_rate": 1.9099483918568287e-05,
+      "loss": 0.3637,
+      "step": 4096
+    },
+    {
+      "epoch": 0.8206565252201762,
+      "learning_rate": 1.9093679278376913e-05,
+      "loss": 0.2712,
+      "step": 4098
+    },
+    {
+      "epoch": 0.8206565252201762,
+      "learning_rate": 1.90878568780329e-05,
+      "loss": 0.3055,
+      "step": 4100
+    },
+    {
+      "epoch": 0.8214571657325861,
+      "learning_rate": 1.90820167289075e-05,
+      "loss": 0.1704,
+      "step": 4102
+    },
+    {
+      "epoch": 0.8214571657325861,
+      "learning_rate": 1.907615884240668e-05,
+      "loss": 0.1078,
+      "step": 4104
+    },
+    {
+      "epoch": 0.822257806244996,
+      "learning_rate": 1.9070283229971003e-05,
+      "loss": 0.1762,
+      "step": 4106
+    },
+    {
+      "epoch": 0.822257806244996,
+      "learning_rate": 1.9064389903075683e-05,
+      "loss": 0.2959,
+      "step": 4108
+    },
+    {
+      "epoch": 0.8230584467574059,
+      "learning_rate": 1.9058478873230487e-05,
+      "loss": 0.5579,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8230584467574059,
+      "learning_rate": 1.905255015197982e-05,
+      "loss": 0.1564,
+      "step": 4112
+    },
+    {
+      "epoch": 0.8238590872698158,
+      "learning_rate": 1.9046603750902585e-05,
+      "loss": 0.2448,
+      "step": 4114
+    },
+    {
+      "epoch": 0.8238590872698158,
+      "learning_rate": 1.9040639681612216e-05,
+      "loss": 0.1979,
+      "step": 4116
+    },
+    {
+      "epoch": 0.8246597277822257,
+      "learning_rate": 1.9034657955756702e-05,
+      "loss": 0.13,
+      "step": 4118
+    },
+    {
+      "epoch": 0.8246597277822257,
+      "learning_rate": 1.902865858501845e-05,
+      "loss": 0.1185,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8254603682946358,
+      "learning_rate": 1.9022641581114396e-05,
+      "loss": 0.3634,
+      "step": 4122
+    },
+    {
+      "epoch": 0.8254603682946358,
+      "learning_rate": 1.9016606955795843e-05,
+      "loss": 0.2813,
+      "step": 4124
+    },
+    {
+      "epoch": 0.8262610088070457,
+      "learning_rate": 1.901055472084858e-05,
+      "loss": 0.2416,
+      "step": 4126
+    },
+    {
+      "epoch": 0.8262610088070457,
+      "learning_rate": 1.9004484888092734e-05,
+      "loss": 0.2049,
+      "step": 4128
+    },
+    {
+      "epoch": 0.8270616493194556,
+      "learning_rate": 1.8998397469382812e-05,
+      "loss": 0.2322,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8270616493194556,
+      "learning_rate": 1.8992292476607695e-05,
+      "loss": 0.1342,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8278622898318655,
+      "learning_rate": 1.898616992169054e-05,
+      "loss": 0.2779,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8278622898318655,
+      "learning_rate": 1.8980029816588863e-05,
+      "loss": 0.1142,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8286629303442754,
+      "learning_rate": 1.89738721732944e-05,
+      "loss": 0.1825,
+      "step": 4138
+    },
+    {
+      "epoch": 0.8286629303442754,
+      "learning_rate": 1.8967697003833156e-05,
+      "loss": 0.1075,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8294635708566853,
+      "learning_rate": 1.8961504320265392e-05,
+      "loss": 0.6788,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8294635708566853,
+      "learning_rate": 1.8955294134685528e-05,
+      "loss": 0.0285,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8302642113690952,
+      "learning_rate": 1.8949066459222224e-05,
+      "loss": 0.1704,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8302642113690952,
+      "learning_rate": 1.8942821306038227e-05,
+      "loss": 0.0975,
+      "step": 4148
+    },
+    {
+      "epoch": 0.8310648518815053,
+      "learning_rate": 1.8936558687330492e-05,
+      "loss": 0.3483,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8310648518815053,
+      "learning_rate": 1.893027861533003e-05,
+      "loss": 0.0197,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8318654923939152,
+      "learning_rate": 1.8923981102301944e-05,
+      "loss": 0.5074,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8318654923939152,
+      "learning_rate": 1.891766616054545e-05,
+      "loss": 0.017,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8326661329063251,
+      "learning_rate": 1.8911333802393725e-05,
+      "loss": 0.3443,
+      "step": 4158
+    },
+    {
+      "epoch": 0.8326661329063251,
+      "learning_rate": 1.8904984040214043e-05,
+      "loss": 1.0473,
+      "step": 4160
+    },
+    {
+      "epoch": 0.833466773418735,
+      "learning_rate": 1.8898616886407588e-05,
+      "loss": 0.0773,
+      "step": 4162
+    },
+    {
+      "epoch": 0.833466773418735,
+      "learning_rate": 1.8892232353409582e-05,
+      "loss": 0.0464,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8342674139311449,
+      "learning_rate": 1.8885830453689146e-05,
+      "loss": 0.176,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8342674139311449,
+      "learning_rate": 1.8879411199749306e-05,
+      "loss": 0.0313,
+      "step": 4168
+    },
+    {
+      "epoch": 0.8350680544435548,
+      "learning_rate": 1.8872974604127038e-05,
+      "loss": 0.3711,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8350680544435548,
+      "learning_rate": 1.8866520679393124e-05,
+      "loss": 0.6589,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8358686949559647,
+      "learning_rate": 1.8860049438152247e-05,
+      "loss": 0.2154,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8358686949559647,
+      "learning_rate": 1.885356089304285e-05,
+      "loss": 0.2633,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8366693354683747,
+      "learning_rate": 1.8847055056737236e-05,
+      "loss": 0.5028,
+      "step": 4178
+    },
+    {
+      "epoch": 0.8366693354683747,
+      "learning_rate": 1.884053194194143e-05,
+      "loss": 0.3358,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8374699759807847,
+      "learning_rate": 1.8833991561395194e-05,
+      "loss": 0.2323,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8374699759807847,
+      "learning_rate": 1.882743392787207e-05,
+      "loss": 0.5659,
+      "step": 4184
+    },
+    {
+      "epoch": 0.8382706164931946,
+      "learning_rate": 1.8820859054179225e-05,
+      "loss": 0.0664,
+      "step": 4186
+    },
+    {
+      "epoch": 0.8382706164931946,
+      "learning_rate": 1.881426695315756e-05,
+      "loss": 0.0416,
+      "step": 4188
+    },
+    {
+      "epoch": 0.8390712570056045,
+      "learning_rate": 1.8807657637681577e-05,
+      "loss": 0.2099,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8390712570056045,
+      "learning_rate": 1.8801031120659396e-05,
+      "loss": 0.1769,
+      "step": 4192
+    },
+    {
+      "epoch": 0.8398718975180144,
+      "learning_rate": 1.8794387415032783e-05,
+      "loss": 0.2049,
+      "step": 4194
+    },
+    {
+      "epoch": 0.8398718975180144,
+      "learning_rate": 1.8787726533777003e-05,
+      "loss": 0.5707,
+      "step": 4196
+    },
+    {
+      "epoch": 0.8406725380304243,
+      "learning_rate": 1.8781048489900936e-05,
+      "loss": 0.3429,
+      "step": 4198
+    },
+    {
+      "epoch": 0.8406725380304243,
+      "learning_rate": 1.877435329644691e-05,
+      "loss": 0.5196,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8414731785428343,
+      "learning_rate": 1.876764096649082e-05,
+      "loss": 0.2776,
+      "step": 4202
+    },
+    {
+      "epoch": 0.8414731785428343,
+      "learning_rate": 1.8760911513141974e-05,
+      "loss": 0.1203,
+      "step": 4204
+    },
+    {
+      "epoch": 0.8422738190552442,
+      "learning_rate": 1.8754164949543123e-05,
+      "loss": 0.4077,
+      "step": 4206
+    },
+    {
+      "epoch": 0.8422738190552442,
+      "learning_rate": 1.8747401288870482e-05,
+      "loss": 0.629,
+      "step": 4208
+    },
+    {
+      "epoch": 0.8430744595676541,
+      "learning_rate": 1.8740620544333604e-05,
+      "loss": 0.0998,
+      "step": 4210
+    },
+    {
+      "epoch": 0.8430744595676541,
+      "learning_rate": 1.8733822729175455e-05,
+      "loss": 0.0205,
+      "step": 4212
+    },
+    {
+      "epoch": 0.8438751000800641,
+      "learning_rate": 1.872700785667228e-05,
+      "loss": 0.1761,
+      "step": 4214
+    },
+    {
+      "epoch": 0.8438751000800641,
+      "learning_rate": 1.8720175940133712e-05,
+      "loss": 0.4468,
+      "step": 4216
+    },
+    {
+      "epoch": 0.844675740592474,
+      "learning_rate": 1.8713326992902612e-05,
+      "loss": 0.2239,
+      "step": 4218
+    },
+    {
+      "epoch": 0.844675740592474,
+      "learning_rate": 1.8706461028355107e-05,
+      "loss": 0.1588,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8454763811048839,
+      "learning_rate": 1.8699578059900604e-05,
+      "loss": 0.8129,
+      "step": 4222
+    },
+    {
+      "epoch": 0.8454763811048839,
+      "learning_rate": 1.8692678100981663e-05,
+      "loss": 0.0581,
+      "step": 4224
+    },
+    {
+      "epoch": 0.8462770216172938,
+      "learning_rate": 1.868576116507408e-05,
+      "loss": 0.2173,
+      "step": 4226
+    },
+    {
+      "epoch": 0.8462770216172938,
+      "learning_rate": 1.8678827265686753e-05,
+      "loss": 0.0585,
+      "step": 4228
+    },
+    {
+      "epoch": 0.8470776621297038,
+      "learning_rate": 1.8671876416361767e-05,
+      "loss": 0.2385,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8470776621297038,
+      "learning_rate": 1.8664908630674264e-05,
+      "loss": 0.6322,
+      "step": 4232
+    },
+    {
+      "epoch": 0.8478783026421137,
+      "learning_rate": 1.8657923922232467e-05,
+      "loss": 0.3056,
+      "step": 4234
+    },
+    {
+      "epoch": 0.8478783026421137,
+      "learning_rate": 1.86509223046777e-05,
+      "loss": 0.3098,
+      "step": 4236
+    },
+    {
+      "epoch": 0.8486789431545236,
+      "learning_rate": 1.8643903791684228e-05,
+      "loss": 0.8131,
+      "step": 4238
+    },
+    {
+      "epoch": 0.8486789431545236,
+      "learning_rate": 1.8636868396959406e-05,
+      "loss": 0.4263,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8494795836669335,
+      "learning_rate": 1.8629816134243466e-05,
+      "loss": 0.5489,
+      "step": 4242
+    },
+    {
+      "epoch": 0.8494795836669335,
+      "learning_rate": 1.8622747017309676e-05,
+      "loss": 0.4834,
+      "step": 4244
+    },
+    {
+      "epoch": 0.8502802241793435,
+      "learning_rate": 1.8615661059964148e-05,
+      "loss": 0.3243,
+      "step": 4246
+    },
+    {
+      "epoch": 0.8502802241793435,
+      "learning_rate": 1.8608558276045898e-05,
+      "loss": 0.1349,
+      "step": 4248
+    },
+    {
+      "epoch": 0.8510808646917534,
+      "learning_rate": 1.860143867942685e-05,
+      "loss": 0.4736,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8510808646917534,
+      "learning_rate": 1.8594302284011697e-05,
+      "loss": 0.1684,
+      "step": 4252
+    },
+    {
+      "epoch": 0.8518815052041633,
+      "learning_rate": 1.8587149103738006e-05,
+      "loss": 0.2282,
+      "step": 4254
+    },
+    {
+      "epoch": 0.8518815052041633,
+      "learning_rate": 1.8579979152576076e-05,
+      "loss": 0.187,
+      "step": 4256
+    },
+    {
+      "epoch": 0.8526821457165733,
+      "learning_rate": 1.8572792444528963e-05,
+      "loss": 0.3356,
+      "step": 4258
+    },
+    {
+      "epoch": 0.8526821457165733,
+      "learning_rate": 1.8565588993632498e-05,
+      "loss": 0.1199,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8534827862289832,
+      "learning_rate": 1.8558368813955136e-05,
+      "loss": 0.4171,
+      "step": 4262
+    },
+    {
+      "epoch": 0.8534827862289832,
+      "learning_rate": 1.8551131919598084e-05,
+      "loss": 0.1853,
+      "step": 4264
+    },
+    {
+      "epoch": 0.8542834267413931,
+      "learning_rate": 1.854387832469512e-05,
+      "loss": 0.2303,
+      "step": 4266
+    },
+    {
+      "epoch": 0.8542834267413931,
+      "learning_rate": 1.8536608043412702e-05,
+      "loss": 0.3686,
+      "step": 4268
+    },
+    {
+      "epoch": 0.855084067253803,
+      "learning_rate": 1.8529321089949833e-05,
+      "loss": 0.7221,
+      "step": 4270
+    },
+    {
+      "epoch": 0.855084067253803,
+      "learning_rate": 1.852201747853807e-05,
+      "loss": 0.2138,
+      "step": 4272
+    },
+    {
+      "epoch": 0.855884707766213,
+      "learning_rate": 1.8514697223441565e-05,
+      "loss": 0.2426,
+      "step": 4274
+    },
+    {
+      "epoch": 0.855884707766213,
+      "learning_rate": 1.8507360338956896e-05,
+      "loss": 0.6411,
+      "step": 4276
+    },
+    {
+      "epoch": 0.8566853482786229,
+      "learning_rate": 1.850000683941319e-05,
+      "loss": 0.291,
+      "step": 4278
+    },
+    {
+      "epoch": 0.8566853482786229,
+      "learning_rate": 1.849263673917196e-05,
+      "loss": 0.0517,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8574859887910328,
+      "learning_rate": 1.8485250052627205e-05,
+      "loss": 0.217,
+      "step": 4282
+    },
+    {
+      "epoch": 0.8574859887910328,
+      "learning_rate": 1.847784679420527e-05,
+      "loss": 0.1541,
+      "step": 4284
+    },
+    {
+      "epoch": 0.8582866293034428,
+      "learning_rate": 1.8470426978364857e-05,
+      "loss": 0.1623,
+      "step": 4286
+    },
+    {
+      "epoch": 0.8582866293034428,
+      "learning_rate": 1.846299061959706e-05,
+      "loss": 0.3731,
+      "step": 4288
+    },
+    {
+      "epoch": 0.8590872698158527,
+      "learning_rate": 1.845553773242522e-05,
+      "loss": 0.2323,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8590872698158527,
+      "learning_rate": 1.8448068331405018e-05,
+      "loss": 0.2715,
+      "step": 4292
+    },
+    {
+      "epoch": 0.8598879103282626,
+      "learning_rate": 1.8440582431124322e-05,
+      "loss": 0.6596,
+      "step": 4294
+    },
+    {
+      "epoch": 0.8598879103282626,
+      "learning_rate": 1.8433080046203293e-05,
+      "loss": 0.2975,
+      "step": 4296
+    },
+    {
+      "epoch": 0.8606885508406725,
+      "learning_rate": 1.842556119129423e-05,
+      "loss": 0.1853,
+      "step": 4298
+    },
+    {
+      "epoch": 0.8606885508406725,
+      "learning_rate": 1.841802588108161e-05,
+      "loss": 0.0966,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8614891913530824,
+      "learning_rate": 1.841047413028209e-05,
+      "loss": 0.159,
+      "step": 4302
+    },
+    {
+      "epoch": 0.8614891913530824,
+      "learning_rate": 1.8402905953644356e-05,
+      "loss": 0.0753,
+      "step": 4304
+    },
+    {
+      "epoch": 0.8622898318654924,
+      "learning_rate": 1.8395321365949273e-05,
+      "loss": 0.5291,
+      "step": 4306
+    },
+    {
+      "epoch": 0.8622898318654924,
+      "learning_rate": 1.838772038200968e-05,
+      "loss": 0.1877,
+      "step": 4308
+    },
+    {
+      "epoch": 0.8630904723779024,
+      "learning_rate": 1.838010301667044e-05,
+      "loss": 0.1699,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8630904723779024,
+      "learning_rate": 1.837246928480848e-05,
+      "loss": 0.0826,
+      "step": 4312
+    },
+    {
+      "epoch": 0.8638911128903123,
+      "learning_rate": 1.8364819201332596e-05,
+      "loss": 0.2274,
+      "step": 4314
+    },
+    {
+      "epoch": 0.8638911128903123,
+      "learning_rate": 1.8357152781183613e-05,
+      "loss": 0.3852,
+      "step": 4316
+    },
+    {
+      "epoch": 0.8646917534027222,
+      "learning_rate": 1.834947003933417e-05,
+      "loss": 0.532,
+      "step": 4318
+    },
+    {
+      "epoch": 0.8646917534027222,
+      "learning_rate": 1.8341770990788874e-05,
+      "loss": 0.073,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8654923939151321,
+      "learning_rate": 1.8334055650584107e-05,
+      "loss": 0.2157,
+      "step": 4322
+    },
+    {
+      "epoch": 0.8654923939151321,
+      "learning_rate": 1.8326324033788087e-05,
+      "loss": 0.0201,
+      "step": 4324
+    },
+    {
+      "epoch": 0.866293034427542,
+      "learning_rate": 1.8318576155500855e-05,
+      "loss": 0.2419,
+      "step": 4326
+    },
+    {
+      "epoch": 0.866293034427542,
+      "learning_rate": 1.831081203085415e-05,
+      "loss": 0.0429,
+      "step": 4328
+    },
+    {
+      "epoch": 0.8670936749399519,
+      "learning_rate": 1.830303167501152e-05,
+      "loss": 0.4884,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8670936749399519,
+      "learning_rate": 1.8295235103168128e-05,
+      "loss": 0.0053,
+      "step": 4332
+    },
+    {
+      "epoch": 0.8678943154523618,
+      "learning_rate": 1.8287422330550885e-05,
+      "loss": 0.6323,
+      "step": 4334
+    },
+    {
+      "epoch": 0.8678943154523618,
+      "learning_rate": 1.8279593372418284e-05,
+      "loss": 0.0937,
+      "step": 4336
+    },
+    {
+      "epoch": 0.8686949559647719,
+      "learning_rate": 1.827174824406043e-05,
+      "loss": 0.3153,
+      "step": 4338
+    },
+    {
+      "epoch": 0.8686949559647719,
+      "learning_rate": 1.8263886960799072e-05,
+      "loss": 0.0579,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8694955964771818,
+      "learning_rate": 1.8256009537987424e-05,
+      "loss": 0.2198,
+      "step": 4342
+    },
+    {
+      "epoch": 0.8694955964771818,
+      "learning_rate": 1.8248115991010303e-05,
+      "loss": 0.3042,
+      "step": 4344
+    },
+    {
+      "epoch": 0.8702962369895917,
+      "learning_rate": 1.8240206335283943e-05,
+      "loss": 0.207,
+      "step": 4346
+    },
+    {
+      "epoch": 0.8702962369895917,
+      "learning_rate": 1.8232280586256104e-05,
+      "loss": 0.3416,
+      "step": 4348
+    },
+    {
+      "epoch": 0.8710968775020016,
+      "learning_rate": 1.8224338759405934e-05,
+      "loss": 0.9246,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8710968775020016,
+      "learning_rate": 1.8216380870243963e-05,
+      "loss": 0.0775,
+      "step": 4352
+    },
+    {
+      "epoch": 0.8718975180144115,
+      "learning_rate": 1.820840693431217e-05,
+      "loss": 0.2935,
+      "step": 4354
+    },
+    {
+      "epoch": 0.8718975180144115,
+      "learning_rate": 1.820041696718378e-05,
+      "loss": 0.6678,
+      "step": 4356
+    },
+    {
+      "epoch": 0.8726981585268214,
+      "learning_rate": 1.8192410984463416e-05,
+      "loss": 0.1903,
+      "step": 4358
+    },
+    {
+      "epoch": 0.8726981585268214,
+      "learning_rate": 1.8184389001786912e-05,
+      "loss": 0.6652,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8734987990392313,
+      "learning_rate": 1.8176351034821352e-05,
+      "loss": 0.2199,
+      "step": 4362
+    },
+    {
+      "epoch": 0.8734987990392313,
+      "learning_rate": 1.8168297099265108e-05,
+      "loss": 0.1135,
+      "step": 4364
+    },
+    {
+      "epoch": 0.8742994395516414,
+      "learning_rate": 1.8160227210847642e-05,
+      "loss": 0.2087,
+      "step": 4366
+    },
+    {
+      "epoch": 0.8742994395516414,
+      "learning_rate": 1.815214138532966e-05,
+      "loss": 0.4475,
+      "step": 4368
+    },
+    {
+      "epoch": 0.8751000800640513,
+      "learning_rate": 1.8144039638502927e-05,
+      "loss": 0.1587,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8751000800640513,
+      "learning_rate": 1.8135921986190358e-05,
+      "loss": 0.0433,
+      "step": 4372
+    },
+    {
+      "epoch": 0.8759007205764612,
+      "learning_rate": 1.8127788444245884e-05,
+      "loss": 0.6798,
+      "step": 4374
+    },
+    {
+      "epoch": 0.8759007205764612,
+      "learning_rate": 1.8119639028554475e-05,
+      "loss": 0.6588,
+      "step": 4376
+    },
+    {
+      "epoch": 0.8767013610888711,
+      "learning_rate": 1.8111473755032152e-05,
+      "loss": 0.4657,
+      "step": 4378
+    },
+    {
+      "epoch": 0.8767013610888711,
+      "learning_rate": 1.8103292639625835e-05,
+      "loss": 0.3322,
+      "step": 4380
+    },
+    {
+      "epoch": 0.877502001601281,
+      "learning_rate": 1.8095095698313456e-05,
+      "loss": 0.211,
+      "step": 4382
+    },
+    {
+      "epoch": 0.877502001601281,
+      "learning_rate": 1.808688294710378e-05,
+      "loss": 0.189,
+      "step": 4384
+    },
+    {
+      "epoch": 0.8783026421136909,
+      "learning_rate": 1.807865440203653e-05,
+      "loss": 0.8527,
+      "step": 4386
+    },
+    {
+      "epoch": 0.8783026421136909,
+      "learning_rate": 1.807041007918221e-05,
+      "loss": 0.5408,
+      "step": 4388
+    },
+    {
+      "epoch": 0.8791032826261009,
+      "learning_rate": 1.806214999464214e-05,
+      "loss": 0.1502,
+      "step": 4390
+    },
+    {
+      "epoch": 0.8791032826261009,
+      "learning_rate": 1.805387416454849e-05,
+      "loss": 0.5762,
+      "step": 4392
+    },
+    {
+      "epoch": 0.8799039231385108,
+      "learning_rate": 1.8045582605064087e-05,
+      "loss": 0.2893,
+      "step": 4394
+    },
+    {
+      "epoch": 0.8799039231385108,
+      "learning_rate": 1.8037275332382575e-05,
+      "loss": 0.376,
+      "step": 4396
+    },
+    {
+      "epoch": 0.8807045636509208,
+      "learning_rate": 1.802895236272819e-05,
+      "loss": 0.3249,
+      "step": 4398
+    },
+    {
+      "epoch": 0.8807045636509208,
+      "learning_rate": 1.802061371235592e-05,
+      "loss": 0.3931,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8815052041633307,
+      "learning_rate": 1.80122593975513e-05,
+      "loss": 0.2533,
+      "step": 4402
+    },
+    {
+      "epoch": 0.8815052041633307,
+      "learning_rate": 1.8003889434630476e-05,
+      "loss": 0.2072,
+      "step": 4404
+    },
+    {
+      "epoch": 0.8823058446757406,
+      "learning_rate": 1.7995503839940204e-05,
+      "loss": 0.1587,
+      "step": 4406
+    },
+    {
+      "epoch": 0.8823058446757406,
+      "learning_rate": 1.7987102629857692e-05,
+      "loss": 0.3521,
+      "step": 4408
+    },
+    {
+      "epoch": 0.8831064851881505,
+      "learning_rate": 1.7978685820790725e-05,
+      "loss": 0.3734,
+      "step": 4410
+    },
+    {
+      "epoch": 0.8831064851881505,
+      "learning_rate": 1.7970253429177494e-05,
+      "loss": 0.2753,
+      "step": 4412
+    },
+    {
+      "epoch": 0.8839071257005604,
+      "learning_rate": 1.796180547148662e-05,
+      "loss": 0.1587,
+      "step": 4414
+    },
+    {
+      "epoch": 0.8839071257005604,
+      "learning_rate": 1.7953341964217196e-05,
+      "loss": 0.4716,
+      "step": 4416
+    },
+    {
+      "epoch": 0.8847077662129704,
+      "learning_rate": 1.7944862923898586e-05,
+      "loss": 0.3863,
+      "step": 4418
+    },
+    {
+      "epoch": 0.8847077662129704,
+      "learning_rate": 1.7936368367090583e-05,
+      "loss": 0.1854,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8855084067253803,
+      "learning_rate": 1.7927858310383196e-05,
+      "loss": 0.3268,
+      "step": 4422
+    },
+    {
+      "epoch": 0.8855084067253803,
+      "learning_rate": 1.7919332770396798e-05,
+      "loss": 0.1508,
+      "step": 4424
+    },
+    {
+      "epoch": 0.8863090472377902,
+      "learning_rate": 1.7910791763781928e-05,
+      "loss": 0.1719,
+      "step": 4426
+    },
+    {
+      "epoch": 0.8863090472377902,
+      "learning_rate": 1.7902235307219336e-05,
+      "loss": 0.1763,
+      "step": 4428
+    },
+    {
+      "epoch": 0.8871096877502002,
+      "learning_rate": 1.789366341742001e-05,
+      "loss": 0.1335,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8871096877502002,
+      "learning_rate": 1.7885076111125e-05,
+      "loss": 0.0577,
+      "step": 4432
+    },
+    {
+      "epoch": 0.8879103282626101,
+      "learning_rate": 1.7876473405105535e-05,
+      "loss": 0.3524,
+      "step": 4434
+    },
+    {
+      "epoch": 0.8879103282626101,
+      "learning_rate": 1.7867855316162846e-05,
+      "loss": 0.6029,
+      "step": 4436
+    },
+    {
+      "epoch": 0.88871096877502,
+      "learning_rate": 1.785922186112829e-05,
+      "loss": 0.3894,
+      "step": 4438
+    },
+    {
+      "epoch": 0.88871096877502,
+      "learning_rate": 1.7850573056863173e-05,
+      "loss": 0.3637,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8895116092874299,
+      "learning_rate": 1.7841908920258774e-05,
+      "loss": 0.176,
+      "step": 4442
+    },
+    {
+      "epoch": 0.8895116092874299,
+      "learning_rate": 1.783322946823638e-05,
+      "loss": 0.2878,
+      "step": 4444
+    },
+    {
+      "epoch": 0.8903122497998399,
+      "learning_rate": 1.782453471774711e-05,
+      "loss": 0.1513,
+      "step": 4446
+    },
+    {
+      "epoch": 0.8903122497998399,
+      "learning_rate": 1.7815824685772042e-05,
+      "loss": 0.4897,
+      "step": 4448
+    },
+    {
+      "epoch": 0.8911128903122498,
+      "learning_rate": 1.7807099389322013e-05,
+      "loss": 0.3776,
+      "step": 4450
+    },
+    {
+      "epoch": 0.8911128903122498,
+      "learning_rate": 1.779835884543776e-05,
+      "loss": 0.2187,
+      "step": 4452
+    },
+    {
+      "epoch": 0.8919135308246597,
+      "learning_rate": 1.7789603071189733e-05,
+      "loss": 0.2069,
+      "step": 4454
+    },
+    {
+      "epoch": 0.8919135308246597,
+      "learning_rate": 1.7780832083678122e-05,
+      "loss": 0.1773,
+      "step": 4456
+    },
+    {
+      "epoch": 0.8927141713370697,
+      "learning_rate": 1.7772045900032912e-05,
+      "loss": 0.1587,
+      "step": 4458
+    },
+    {
+      "epoch": 0.8927141713370697,
+      "learning_rate": 1.776324453741365e-05,
+      "loss": 0.438,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8935148118494796,
+      "learning_rate": 1.7754428013009644e-05,
+      "loss": 0.4262,
+      "step": 4462
+    },
+    {
+      "epoch": 0.8935148118494796,
+      "learning_rate": 1.774559634403971e-05,
+      "loss": 0.122,
+      "step": 4464
+    },
+    {
+      "epoch": 0.8943154523618895,
+      "learning_rate": 1.7736749547752327e-05,
+      "loss": 0.2068,
+      "step": 4466
+    },
+    {
+      "epoch": 0.8943154523618895,
+      "learning_rate": 1.7727887641425465e-05,
+      "loss": 0.5509,
+      "step": 4468
+    },
+    {
+      "epoch": 0.8951160928742994,
+      "learning_rate": 1.7719010642366597e-05,
+      "loss": 0.1901,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8951160928742994,
+      "learning_rate": 1.7710118567912732e-05,
+      "loss": 0.418,
+      "step": 4472
+    },
+    {
+      "epoch": 0.8959167333867094,
+      "learning_rate": 1.770121143543025e-05,
+      "loss": 0.4342,
+      "step": 4474
+    },
+    {
+      "epoch": 0.8959167333867094,
+      "learning_rate": 1.7692289262315008e-05,
+      "loss": 0.4755,
+      "step": 4476
+    },
+    {
+      "epoch": 0.8967173738991193,
+      "learning_rate": 1.7683352065992174e-05,
+      "loss": 0.3034,
+      "step": 4478
+    },
+    {
+      "epoch": 0.8967173738991193,
+      "learning_rate": 1.7674399863916298e-05,
+      "loss": 0.2379,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8975180144115292,
+      "learning_rate": 1.7665432673571238e-05,
+      "loss": 0.2379,
+      "step": 4482
+    },
+    {
+      "epoch": 0.8975180144115292,
+      "learning_rate": 1.765645051247007e-05,
+      "loss": 0.1511,
+      "step": 4484
+    },
+    {
+      "epoch": 0.8983186549239391,
+      "learning_rate": 1.7647453398155204e-05,
+      "loss": 0.2849,
+      "step": 4486
+    },
+    {
+      "epoch": 0.8983186549239391,
+      "learning_rate": 1.7638441348198144e-05,
+      "loss": 0.0123,
+      "step": 4488
+    },
+    {
+      "epoch": 0.899119295436349,
+      "learning_rate": 1.7629414380199672e-05,
+      "loss": 0.1866,
+      "step": 4490
+    },
+    {
+      "epoch": 0.899119295436349,
+      "learning_rate": 1.762037251178961e-05,
+      "loss": 0.4598,
+      "step": 4492
+    },
+    {
+      "epoch": 0.899919935948759,
+      "learning_rate": 1.7611315760626943e-05,
+      "loss": 0.1496,
+      "step": 4494
+    },
+    {
+      "epoch": 0.899919935948759,
+      "learning_rate": 1.7602244144399713e-05,
+      "loss": 0.0817,
+      "step": 4496
+    },
+    {
+      "epoch": 0.900720576461169,
+      "learning_rate": 1.7593157680824943e-05,
+      "loss": 0.4548,
+      "step": 4498
+    },
+    {
+      "epoch": 0.900720576461169,
+      "learning_rate": 1.7584056387648738e-05,
+      "loss": 0.1285,
+      "step": 4500
+    },
+    {
+      "epoch": 0.9015212169735789,
+      "learning_rate": 1.757494028264608e-05,
+      "loss": 0.2961,
+      "step": 4502
+    },
+    {
+      "epoch": 0.9015212169735789,
+      "learning_rate": 1.7565809383620966e-05,
+      "loss": 0.1537,
+      "step": 4504
+    },
+    {
+      "epoch": 0.9023218574859888,
+      "learning_rate": 1.7556663708406203e-05,
+      "loss": 0.2138,
+      "step": 4506
+    },
+    {
+      "epoch": 0.9023218574859888,
+      "learning_rate": 1.7547503274863502e-05,
+      "loss": 0.13,
+      "step": 4508
+    },
+    {
+      "epoch": 0.9031224979983987,
+      "learning_rate": 1.7538328100883404e-05,
+      "loss": 0.4411,
+      "step": 4510
+    },
+    {
+      "epoch": 0.9031224979983987,
+      "learning_rate": 1.7529138204385186e-05,
+      "loss": 0.1029,
+      "step": 4512
+    },
+    {
+      "epoch": 0.9039231385108086,
+      "learning_rate": 1.7519933603316962e-05,
+      "loss": 0.4227,
+      "step": 4514
+    },
+    {
+      "epoch": 0.9039231385108086,
+      "learning_rate": 1.7510714315655467e-05,
+      "loss": 0.6255,
+      "step": 4516
+    },
+    {
+      "epoch": 0.9047237790232185,
+      "learning_rate": 1.750148035940622e-05,
+      "loss": 0.1786,
+      "step": 4518
+    },
+    {
+      "epoch": 0.9047237790232185,
+      "learning_rate": 1.7492231752603305e-05,
+      "loss": 0.426,
+      "step": 4520
+    },
+    {
+      "epoch": 0.9055244195356285,
+      "learning_rate": 1.7482968513309458e-05,
+      "loss": 0.5727,
+      "step": 4522
+    },
+    {
+      "epoch": 0.9055244195356285,
+      "learning_rate": 1.7473690659616e-05,
+      "loss": 0.0447,
+      "step": 4524
+    },
+    {
+      "epoch": 0.9063250600480385,
+      "learning_rate": 1.7464398209642744e-05,
+      "loss": 0.333,
+      "step": 4526
+    },
+    {
+      "epoch": 0.9063250600480385,
+      "learning_rate": 1.7455091181538094e-05,
+      "loss": 0.5703,
+      "step": 4528
+    },
+    {
+      "epoch": 0.9071257005604484,
+      "learning_rate": 1.7445769593478842e-05,
+      "loss": 0.2161,
+      "step": 4530
+    },
+    {
+      "epoch": 0.9071257005604484,
+      "learning_rate": 1.743643346367027e-05,
+      "loss": 0.2706,
+      "step": 4532
+    },
+    {
+      "epoch": 0.9079263410728583,
+      "learning_rate": 1.7427082810346024e-05,
+      "loss": 0.1665,
+      "step": 4534
+    },
+    {
+      "epoch": 0.9079263410728583,
+      "learning_rate": 1.741771765176815e-05,
+      "loss": 0.2637,
+      "step": 4536
+    },
+    {
+      "epoch": 0.9087269815852682,
+      "learning_rate": 1.740833800622701e-05,
+      "loss": 0.3771,
+      "step": 4538
+    },
+    {
+      "epoch": 0.9087269815852682,
+      "learning_rate": 1.739894389204122e-05,
+      "loss": 0.0067,
+      "step": 4540
+    },
+    {
+      "epoch": 0.9095276220976781,
+      "learning_rate": 1.738953532755774e-05,
+      "loss": 0.3853,
+      "step": 4542
+    },
+    {
+      "epoch": 0.9095276220976781,
+      "learning_rate": 1.7380112331151657e-05,
+      "loss": 0.2051,
+      "step": 4544
+    },
+    {
+      "epoch": 0.910328262610088,
+      "learning_rate": 1.7370674921226306e-05,
+      "loss": 0.1955,
+      "step": 4546
+    },
+    {
+      "epoch": 0.910328262610088,
+      "learning_rate": 1.7361223116213146e-05,
+      "loss": 0.2711,
+      "step": 4548
+    },
+    {
+      "epoch": 0.911128903122498,
+      "learning_rate": 1.7351756934571764e-05,
+      "loss": 0.2293,
+      "step": 4550
+    },
+    {
+      "epoch": 0.911128903122498,
+      "learning_rate": 1.7342276394789825e-05,
+      "loss": 0.1236,
+      "step": 4552
+    },
+    {
+      "epoch": 0.911929543634908,
+      "learning_rate": 1.7332781515382996e-05,
+      "loss": 0.51,
+      "step": 4554
+    },
+    {
+      "epoch": 0.911929543634908,
+      "learning_rate": 1.732327231489503e-05,
+      "loss": 0.7844,
+      "step": 4556
+    },
+    {
+      "epoch": 0.9127301841473179,
+      "learning_rate": 1.7313748811897564e-05,
+      "loss": 0.2334,
+      "step": 4558
+    },
+    {
+      "epoch": 0.9127301841473179,
+      "learning_rate": 1.7304211024990216e-05,
+      "loss": 0.2295,
+      "step": 4560
+    },
+    {
+      "epoch": 0.9135308246597278,
+      "learning_rate": 1.7294658972800495e-05,
+      "loss": 0.1496,
+      "step": 4562
+    },
+    {
+      "epoch": 0.9135308246597278,
+      "learning_rate": 1.728509267398376e-05,
+      "loss": 0.0667,
+      "step": 4564
+    },
+    {
+      "epoch": 0.9143314651721377,
+      "learning_rate": 1.727551214722322e-05,
+      "loss": 0.3487,
+      "step": 4566
+    },
+    {
+      "epoch": 0.9143314651721377,
+      "learning_rate": 1.7265917411229803e-05,
+      "loss": 0.0767,
+      "step": 4568
+    },
+    {
+      "epoch": 0.9151321056845476,
+      "learning_rate": 1.72563084847423e-05,
+      "loss": 0.3192,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9151321056845476,
+      "learning_rate": 1.7246685386527105e-05,
+      "loss": 0.3517,
+      "step": 4572
+    },
+    {
+      "epoch": 0.9159327461969575,
+      "learning_rate": 1.723704813537835e-05,
+      "loss": 0.2294,
+      "step": 4574
+    },
+    {
+      "epoch": 0.9159327461969575,
+      "learning_rate": 1.7227396750117802e-05,
+      "loss": 0.0792,
+      "step": 4576
+    },
+    {
+      "epoch": 0.9167333867093675,
+      "learning_rate": 1.7217731249594817e-05,
+      "loss": 0.6158,
+      "step": 4578
+    },
+    {
+      "epoch": 0.9167333867093675,
+      "learning_rate": 1.7208051652686348e-05,
+      "loss": 0.1384,
+      "step": 4580
+    },
+    {
+      "epoch": 0.9175340272217775,
+      "learning_rate": 1.7198357978296827e-05,
+      "loss": 0.4153,
+      "step": 4582
+    },
+    {
+      "epoch": 0.9175340272217775,
+      "learning_rate": 1.718865024535822e-05,
+      "loss": 0.0014,
+      "step": 4584
+    },
+    {
+      "epoch": 0.9183346677341874,
+      "learning_rate": 1.717892847282995e-05,
+      "loss": 0.2328,
+      "step": 4586
+    },
+    {
+      "epoch": 0.9183346677341874,
+      "learning_rate": 1.716919267969884e-05,
+      "loss": 0.3754,
+      "step": 4588
+    },
+    {
+      "epoch": 0.9191353082465973,
+      "learning_rate": 1.715944288497911e-05,
+      "loss": 0.1695,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9191353082465973,
+      "learning_rate": 1.7149679107712317e-05,
+      "loss": 0.2707,
+      "step": 4592
+    },
+    {
+      "epoch": 0.9199359487590072,
+      "learning_rate": 1.713990136696734e-05,
+      "loss": 0.3555,
+      "step": 4594
+    },
+    {
+      "epoch": 0.9199359487590072,
+      "learning_rate": 1.7130109681840298e-05,
+      "loss": 0.4731,
+      "step": 4596
+    },
+    {
+      "epoch": 0.9207365892714171,
+      "learning_rate": 1.7120304071454578e-05,
+      "loss": 0.0322,
+      "step": 4598
+    },
+    {
+      "epoch": 0.9207365892714171,
+      "learning_rate": 1.711048455496075e-05,
+      "loss": 0.0551,
+      "step": 4600
+    },
+    {
+      "epoch": 0.921537229783827,
+      "learning_rate": 1.7100651151536532e-05,
+      "loss": 0.8615,
+      "step": 4602
+    },
+    {
+      "epoch": 0.921537229783827,
+      "learning_rate": 1.7090803880386784e-05,
+      "loss": 0.1954,
+      "step": 4604
+    },
+    {
+      "epoch": 0.922337870296237,
+      "learning_rate": 1.708094276074344e-05,
+      "loss": 0.2265,
+      "step": 4606
+    },
+    {
+      "epoch": 0.922337870296237,
+      "learning_rate": 1.7071067811865474e-05,
+      "loss": 0.1821,
+      "step": 4608
+    },
+    {
+      "epoch": 0.923138510808647,
+      "learning_rate": 1.7061179053038894e-05,
+      "loss": 0.5494,
+      "step": 4610
+    },
+    {
+      "epoch": 0.923138510808647,
+      "learning_rate": 1.705127650357663e-05,
+      "loss": 0.1867,
+      "step": 4612
+    },
+    {
+      "epoch": 0.9239391513210569,
+      "learning_rate": 1.704136018281859e-05,
+      "loss": 0.1756,
+      "step": 4614
+    },
+    {
+      "epoch": 0.9239391513210569,
+      "learning_rate": 1.7031430110131566e-05,
+      "loss": 0.6573,
+      "step": 4616
+    },
+    {
+      "epoch": 0.9247397918334668,
+      "learning_rate": 1.7021486304909202e-05,
+      "loss": 0.1342,
+      "step": 4618
+    },
+    {
+      "epoch": 0.9247397918334668,
+      "learning_rate": 1.701152878657197e-05,
+      "loss": 0.1762,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9255404323458767,
+      "learning_rate": 1.700155757456711e-05,
+      "loss": 0.1346,
+      "step": 4622
+    },
+    {
+      "epoch": 0.9255404323458767,
+      "learning_rate": 1.699157268836863e-05,
+      "loss": 0.2233,
+      "step": 4624
+    },
+    {
+      "epoch": 0.9263410728582866,
+      "learning_rate": 1.6981574147477214e-05,
+      "loss": 0.1991,
+      "step": 4626
+    },
+    {
+      "epoch": 0.9263410728582866,
+      "learning_rate": 1.697156197142023e-05,
+      "loss": 0.0719,
+      "step": 4628
+    },
+    {
+      "epoch": 0.9271417133706965,
+      "learning_rate": 1.696153617975168e-05,
+      "loss": 0.2535,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9271417133706965,
+      "learning_rate": 1.6951496792052148e-05,
+      "loss": 0.0063,
+      "step": 4632
+    },
+    {
+      "epoch": 0.9279423538831065,
+      "learning_rate": 1.694144382792878e-05,
+      "loss": 0.2162,
+      "step": 4634
+    },
+    {
+      "epoch": 0.9279423538831065,
+      "learning_rate": 1.6931377307015236e-05,
+      "loss": 0.1016,
+      "step": 4636
+    },
+    {
+      "epoch": 0.9287429943955164,
+      "learning_rate": 1.6921297248971652e-05,
+      "loss": 0.3514,
+      "step": 4638
+    },
+    {
+      "epoch": 0.9287429943955164,
+      "learning_rate": 1.6911203673484583e-05,
+      "loss": 0.1953,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9295436349079264,
+      "learning_rate": 1.690109660026701e-05,
+      "loss": 0.3386,
+      "step": 4642
+    },
+    {
+      "epoch": 0.9295436349079264,
+      "learning_rate": 1.6890976049058267e-05,
+      "loss": 0.0428,
+      "step": 4644
+    },
+    {
+      "epoch": 0.9303442754203363,
+      "learning_rate": 1.688084203962401e-05,
+      "loss": 0.0296,
+      "step": 4646
+    },
+    {
+      "epoch": 0.9303442754203363,
+      "learning_rate": 1.687069459175619e-05,
+      "loss": 0.1098,
+      "step": 4648
+    },
+    {
+      "epoch": 0.9311449159327462,
+      "learning_rate": 1.6860533725272953e-05,
+      "loss": 0.4078,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9311449159327462,
+      "learning_rate": 1.6850359460018744e-05,
+      "loss": 0.3788,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9319455564451561,
+      "learning_rate": 1.6840171815864085e-05,
+      "loss": 0.4026,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9319455564451561,
+      "learning_rate": 1.682997081270568e-05,
+      "loss": 0.0298,
+      "step": 4656
+    },
+    {
+      "epoch": 0.932746196957566,
+      "learning_rate": 1.681975647046631e-05,
+      "loss": 0.3147,
+      "step": 4658
+    },
+    {
+      "epoch": 0.932746196957566,
+      "learning_rate": 1.6809528809094805e-05,
+      "loss": 0.3,
+      "step": 4660
+    },
+    {
+      "epoch": 0.933546837469976,
+      "learning_rate": 1.6799287848566024e-05,
+      "loss": 0.6517,
+      "step": 4662
+    },
+    {
+      "epoch": 0.933546837469976,
+      "learning_rate": 1.6789033608880742e-05,
+      "loss": 0.238,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9343474779823859,
+      "learning_rate": 1.6778766110065765e-05,
+      "loss": 0.1606,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9343474779823859,
+      "learning_rate": 1.67684853721737e-05,
+      "loss": 0.4368,
+      "step": 4668
+    },
+    {
+      "epoch": 0.9351481184947958,
+      "learning_rate": 1.6758191415283066e-05,
+      "loss": 0.3007,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9351481184947958,
+      "learning_rate": 1.6747884259498185e-05,
+      "loss": 0.0737,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9359487590072058,
+      "learning_rate": 1.673756392494916e-05,
+      "loss": 0.5567,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9359487590072058,
+      "learning_rate": 1.6727230431791826e-05,
+      "loss": 0.1951,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9367493995196157,
+      "learning_rate": 1.671688380020769e-05,
+      "loss": 0.202,
+      "step": 4678
+    },
+    {
+      "epoch": 0.9367493995196157,
+      "learning_rate": 1.6706524050404006e-05,
+      "loss": 0.2464,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9375500400320256,
+      "learning_rate": 1.6696151202613537e-05,
+      "loss": 0.6446,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9375500400320256,
+      "learning_rate": 1.6685765277094702e-05,
+      "loss": 0.2138,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9383506805444356,
+      "learning_rate": 1.6675366294131432e-05,
+      "loss": 0.0839,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9383506805444356,
+      "learning_rate": 1.6664954274033175e-05,
+      "loss": 0.0096,
+      "step": 4688
+    },
+    {
+      "epoch": 0.9391513210568455,
+      "learning_rate": 1.6654529237134833e-05,
+      "loss": 0.1432,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9391513210568455,
+      "learning_rate": 1.66440912037967e-05,
+      "loss": 0.0993,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9399519615692554,
+      "learning_rate": 1.663364019440453e-05,
+      "loss": 0.5957,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9399519615692554,
+      "learning_rate": 1.662317622936933e-05,
+      "loss": 0.1596,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9407526020816653,
+      "learning_rate": 1.6612699329127467e-05,
+      "loss": 0.4715,
+      "step": 4698
+    },
+    {
+      "epoch": 0.9407526020816653,
+      "learning_rate": 1.6602209514140562e-05,
+      "loss": 0.7418,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9415532425940752,
+      "learning_rate": 1.6591706804895415e-05,
+      "loss": 0.2233,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9415532425940752,
+      "learning_rate": 1.6581191221904098e-05,
+      "loss": 0.1497,
+      "step": 4704
+    },
+    {
+      "epoch": 0.9423538831064852,
+      "learning_rate": 1.6570662785703716e-05,
+      "loss": 0.1512,
+      "step": 4706
+    },
+    {
+      "epoch": 0.9423538831064852,
+      "learning_rate": 1.6560121516856592e-05,
+      "loss": 0.1666,
+      "step": 4708
+    },
+    {
+      "epoch": 0.9431545236188951,
+      "learning_rate": 1.654956743595001e-05,
+      "loss": 0.1528,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9431545236188951,
+      "learning_rate": 1.6539000563596328e-05,
+      "loss": 0.228,
+      "step": 4712
+    },
+    {
+      "epoch": 0.9439551641313051,
+      "learning_rate": 1.6528420920432893e-05,
+      "loss": 0.3298,
+      "step": 4714
+    },
+    {
+      "epoch": 0.9439551641313051,
+      "learning_rate": 1.651782852712194e-05,
+      "loss": 0.0243,
+      "step": 4716
+    },
+    {
+      "epoch": 0.944755804643715,
+      "learning_rate": 1.6507223404350686e-05,
+      "loss": 0.635,
+      "step": 4718
+    },
+    {
+      "epoch": 0.944755804643715,
+      "learning_rate": 1.6496605572831127e-05,
+      "loss": 0.1585,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9455564451561249,
+      "learning_rate": 1.648597505330016e-05,
+      "loss": 0.4467,
+      "step": 4722
+    },
+    {
+      "epoch": 0.9455564451561249,
+      "learning_rate": 1.6475331866519387e-05,
+      "loss": 0.0793,
+      "step": 4724
+    },
+    {
+      "epoch": 0.9463570856685348,
+      "learning_rate": 1.6464676033275187e-05,
+      "loss": 0.2946,
+      "step": 4726
+    },
+    {
+      "epoch": 0.9463570856685348,
+      "learning_rate": 1.6454007574378657e-05,
+      "loss": 0.0438,
+      "step": 4728
+    },
+    {
+      "epoch": 0.9471577261809447,
+      "learning_rate": 1.644332651066548e-05,
+      "loss": 0.5859,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9471577261809447,
+      "learning_rate": 1.6432632862996062e-05,
+      "loss": 0.2021,
+      "step": 4732
+    },
+    {
+      "epoch": 0.9479583666933546,
+      "learning_rate": 1.6421926652255275e-05,
+      "loss": 0.1627,
+      "step": 4734
+    },
+    {
+      "epoch": 0.9479583666933546,
+      "learning_rate": 1.6411207899352633e-05,
+      "loss": 0.3356,
+      "step": 4736
+    },
+    {
+      "epoch": 0.9487590072057646,
+      "learning_rate": 1.6400476625222057e-05,
+      "loss": 0.3043,
+      "step": 4738
+    },
+    {
+      "epoch": 0.9487590072057646,
+      "learning_rate": 1.6389732850821964e-05,
+      "loss": 0.256,
+      "step": 4740
+    },
+    {
+      "epoch": 0.9495596477181746,
+      "learning_rate": 1.6378976597135193e-05,
+      "loss": 0.3328,
+      "step": 4742
+    },
+    {
+      "epoch": 0.9495596477181746,
+      "learning_rate": 1.6368207885168904e-05,
+      "loss": 0.0391,
+      "step": 4744
+    },
+    {
+      "epoch": 0.9503602882305845,
+      "learning_rate": 1.635742673595468e-05,
+      "loss": 0.67,
+      "step": 4746
+    },
+    {
+      "epoch": 0.9503602882305845,
+      "learning_rate": 1.6346633170548275e-05,
+      "loss": 0.378,
+      "step": 4748
+    },
+    {
+      "epoch": 0.9511609287429944,
+      "learning_rate": 1.6335827210029823e-05,
+      "loss": 0.089,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9511609287429944,
+      "learning_rate": 1.6325008875503563e-05,
+      "loss": 0.0366,
+      "step": 4752
+    },
+    {
+      "epoch": 0.9519615692554043,
+      "learning_rate": 1.6314178188097917e-05,
+      "loss": 0.4012,
+      "step": 4754
+    },
+    {
+      "epoch": 0.9519615692554043,
+      "learning_rate": 1.6303335168965495e-05,
+      "loss": 0.1774,
+      "step": 4756
+    },
+    {
+      "epoch": 0.9527622097678142,
+      "learning_rate": 1.6292479839282904e-05,
+      "loss": 0.2536,
+      "step": 4758
+    },
+    {
+      "epoch": 0.9527622097678142,
+      "learning_rate": 1.628161222025089e-05,
+      "loss": 0.2885,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9535628502802241,
+      "learning_rate": 1.627073233309409e-05,
+      "loss": 0.2917,
+      "step": 4762
+    },
+    {
+      "epoch": 0.9535628502802241,
+      "learning_rate": 1.625984019906122e-05,
+      "loss": 0.0923,
+      "step": 4764
+    },
+    {
+      "epoch": 0.9543634907926342,
+      "learning_rate": 1.624893583942482e-05,
+      "loss": 0.3349,
+      "step": 4766
+    },
+    {
+      "epoch": 0.9543634907926342,
+      "learning_rate": 1.623801927548132e-05,
+      "loss": 0.0433,
+      "step": 4768
+    },
+    {
+      "epoch": 0.9551641313050441,
+      "learning_rate": 1.6227090528551058e-05,
+      "loss": 0.1419,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9551641313050441,
+      "learning_rate": 1.6216149619978057e-05,
+      "loss": 0.1378,
+      "step": 4772
+    },
+    {
+      "epoch": 0.955964771817454,
+      "learning_rate": 1.6205196571130204e-05,
+      "loss": 0.2879,
+      "step": 4774
+    },
+    {
+      "epoch": 0.955964771817454,
+      "learning_rate": 1.6194231403398987e-05,
+      "loss": 0.1541,
+      "step": 4776
+    },
+    {
+      "epoch": 0.9567654123298639,
+      "learning_rate": 1.618325413819967e-05,
+      "loss": 0.411,
+      "step": 4778
+    },
+    {
+      "epoch": 0.9567654123298639,
+      "learning_rate": 1.6172264796971063e-05,
+      "loss": 0.0572,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9575660528422738,
+      "learning_rate": 1.6161263401175555e-05,
+      "loss": 0.5453,
+      "step": 4782
+    },
+    {
+      "epoch": 0.9575660528422738,
+      "learning_rate": 1.6150249972299173e-05,
+      "loss": 0.1949,
+      "step": 4784
+    },
+    {
+      "epoch": 0.9583666933546837,
+      "learning_rate": 1.613922453185133e-05,
+      "loss": 0.7139,
+      "step": 4786
+    },
+    {
+      "epoch": 0.9583666933546837,
+      "learning_rate": 1.612818710136499e-05,
+      "loss": 0.4795,
+      "step": 4788
+    },
+    {
+      "epoch": 0.9591673338670936,
+      "learning_rate": 1.6117137702396454e-05,
+      "loss": 0.5209,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9591673338670936,
+      "learning_rate": 1.6106076356525484e-05,
+      "loss": 0.4714,
+      "step": 4792
+    },
+    {
+      "epoch": 0.9599679743795037,
+      "learning_rate": 1.6095003085355103e-05,
+      "loss": 0.3584,
+      "step": 4794
+    },
+    {
+      "epoch": 0.9599679743795037,
+      "learning_rate": 1.6083917910511623e-05,
+      "loss": 0.3463,
+      "step": 4796
+    },
+    {
+      "epoch": 0.9607686148919136,
+      "learning_rate": 1.6072820853644688e-05,
+      "loss": 0.7911,
+      "step": 4798
+    },
+    {
+      "epoch": 0.9607686148919136,
+      "learning_rate": 1.6061711936427028e-05,
+      "loss": 0.2156,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9615692554043235,
+      "learning_rate": 1.6050591180554658e-05,
+      "loss": 0.183,
+      "step": 4802
+    },
+    {
+      "epoch": 0.9615692554043235,
+      "learning_rate": 1.60394586077466e-05,
+      "loss": 0.1345,
+      "step": 4804
+    },
+    {
+      "epoch": 0.9623698959167334,
+      "learning_rate": 1.6028314239745068e-05,
+      "loss": 0.1276,
+      "step": 4806
+    },
+    {
+      "epoch": 0.9623698959167334,
+      "learning_rate": 1.6017158098315224e-05,
+      "loss": 0.2641,
+      "step": 4808
+    },
+    {
+      "epoch": 0.9631705364291433,
+      "learning_rate": 1.6005990205245226e-05,
+      "loss": 0.4003,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9631705364291433,
+      "learning_rate": 1.5994810582346266e-05,
+      "loss": 0.0812,
+      "step": 4812
+    },
+    {
+      "epoch": 0.9639711769415532,
+      "learning_rate": 1.5983619251452334e-05,
+      "loss": 0.1702,
+      "step": 4814
+    },
+    {
+      "epoch": 0.9639711769415532,
+      "learning_rate": 1.5972416234420404e-05,
+      "loss": 0.1665,
+      "step": 4816
+    },
+    {
+      "epoch": 0.9647718174539631,
+      "learning_rate": 1.596120155313017e-05,
+      "loss": 0.2824,
+      "step": 4818
+    },
+    {
+      "epoch": 0.9647718174539631,
+      "learning_rate": 1.594997522948413e-05,
+      "loss": 0.0134,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9655724579663731,
+      "learning_rate": 1.593873728540759e-05,
+      "loss": 0.1138,
+      "step": 4822
+    },
+    {
+      "epoch": 0.9655724579663731,
+      "learning_rate": 1.592748774284844e-05,
+      "loss": 0.1343,
+      "step": 4824
+    },
+    {
+      "epoch": 0.966373098478783,
+      "learning_rate": 1.5916226623777346e-05,
+      "loss": 0.2533,
+      "step": 4826
+    },
+    {
+      "epoch": 0.966373098478783,
+      "learning_rate": 1.5904953950187448e-05,
+      "loss": 0.1226,
+      "step": 4828
+    },
+    {
+      "epoch": 0.967173738991193,
+      "learning_rate": 1.5893669744094587e-05,
+      "loss": 0.1512,
+      "step": 4830
+    },
+    {
+      "epoch": 0.967173738991193,
+      "learning_rate": 1.588237402753703e-05,
+      "loss": 0.2555,
+      "step": 4832
+    },
+    {
+      "epoch": 0.9679743795036029,
+      "learning_rate": 1.5871066822575526e-05,
+      "loss": 0.2378,
+      "step": 4834
+    },
+    {
+      "epoch": 0.9679743795036029,
+      "learning_rate": 1.5859748151293354e-05,
+      "loss": 0.7478,
+      "step": 4836
+    },
+    {
+      "epoch": 0.9687750200160128,
+      "learning_rate": 1.5848418035796064e-05,
+      "loss": 0.152,
+      "step": 4838
+    },
+    {
+      "epoch": 0.9687750200160128,
+      "learning_rate": 1.5837076498211673e-05,
+      "loss": 0.0552,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9695756605284227,
+      "learning_rate": 1.5825723560690396e-05,
+      "loss": 0.4747,
+      "step": 4842
+    },
+    {
+      "epoch": 0.9695756605284227,
+      "learning_rate": 1.581435924540482e-05,
+      "loss": 0.2445,
+      "step": 4844
+    },
+    {
+      "epoch": 0.9703763010408326,
+      "learning_rate": 1.580298357454967e-05,
+      "loss": 0.3226,
+      "step": 4846
+    },
+    {
+      "epoch": 0.9703763010408326,
+      "learning_rate": 1.579159657034185e-05,
+      "loss": 0.0665,
+      "step": 4848
+    },
+    {
+      "epoch": 0.9711769415532426,
+      "learning_rate": 1.5780198255020485e-05,
+      "loss": 0.2761,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9711769415532426,
+      "learning_rate": 1.5768788650846674e-05,
+      "loss": 0.007,
+      "step": 4852
+    },
+    {
+      "epoch": 0.9719775820656525,
+      "learning_rate": 1.5757367780103672e-05,
+      "loss": 0.1608,
+      "step": 4854
+    },
+    {
+      "epoch": 0.9719775820656525,
+      "learning_rate": 1.574593566509664e-05,
+      "loss": 0.1789,
+      "step": 4856
+    },
+    {
+      "epoch": 0.9727782225780625,
+      "learning_rate": 1.5734492328152796e-05,
+      "loss": 0.0919,
+      "step": 4858
+    },
+    {
+      "epoch": 0.9727782225780625,
+      "learning_rate": 1.5723037791621203e-05,
+      "loss": 0.0056,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9735788630904724,
+      "learning_rate": 1.5711572077872784e-05,
+      "loss": 0.1635,
+      "step": 4862
+    },
+    {
+      "epoch": 0.9735788630904724,
+      "learning_rate": 1.5700095209300386e-05,
+      "loss": 0.0626,
+      "step": 4864
+    },
+    {
+      "epoch": 0.9743795036028823,
+      "learning_rate": 1.568860720831852e-05,
+      "loss": 0.5397,
+      "step": 4866
+    },
+    {
+      "epoch": 0.9743795036028823,
+      "learning_rate": 1.5677108097363565e-05,
+      "loss": 0.0692,
+      "step": 4868
+    },
+    {
+      "epoch": 0.9751801441152922,
+      "learning_rate": 1.5665597898893508e-05,
+      "loss": 0.1807,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9751801441152922,
+      "learning_rate": 1.5654076635387976e-05,
+      "loss": 0.029,
+      "step": 4872
+    },
+    {
+      "epoch": 0.9759807846277022,
+      "learning_rate": 1.5642544329348316e-05,
+      "loss": 0.3771,
+      "step": 4874
+    },
+    {
+      "epoch": 0.9759807846277022,
+      "learning_rate": 1.5631001003297302e-05,
+      "loss": 0.1662,
+      "step": 4876
+    },
+    {
+      "epoch": 0.9767814251401121,
+      "learning_rate": 1.5619446679779367e-05,
+      "loss": 0.5357,
+      "step": 4878
+    },
+    {
+      "epoch": 0.9767814251401121,
+      "learning_rate": 1.560788138136029e-05,
+      "loss": 0.5486,
+      "step": 4880
+    },
+    {
+      "epoch": 0.977582065652522,
+      "learning_rate": 1.5596305130627414e-05,
+      "loss": 0.2382,
+      "step": 4882
+    },
+    {
+      "epoch": 0.977582065652522,
+      "learning_rate": 1.5584717950189373e-05,
+      "loss": 0.1727,
+      "step": 4884
+    },
+    {
+      "epoch": 0.978382706164932,
+      "learning_rate": 1.5573119862676155e-05,
+      "loss": 0.1686,
+      "step": 4886
+    },
+    {
+      "epoch": 0.978382706164932,
+      "learning_rate": 1.5561510890739137e-05,
+      "loss": 0.0837,
+      "step": 4888
+    },
+    {
+      "epoch": 0.9791833466773419,
+      "learning_rate": 1.554989105705083e-05,
+      "loss": 0.1311,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9791833466773419,
+      "learning_rate": 1.5538260384305083e-05,
+      "loss": 0.0826,
+      "step": 4892
+    },
+    {
+      "epoch": 0.9799839871897518,
+      "learning_rate": 1.5526618895216786e-05,
+      "loss": 0.1105,
+      "step": 4894
+    },
+    {
+      "epoch": 0.9799839871897518,
+      "learning_rate": 1.5514966612522088e-05,
+      "loss": 0.0088,
+      "step": 4896
+    },
+    {
+      "epoch": 0.9807846277021617,
+      "learning_rate": 1.5503303558978112e-05,
+      "loss": 0.5397,
+      "step": 4898
+    },
+    {
+      "epoch": 0.9807846277021617,
+      "learning_rate": 1.5491629757363033e-05,
+      "loss": 0.6078,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9815852682145717,
+      "learning_rate": 1.547994523047609e-05,
+      "loss": 0.1693,
+      "step": 4902
+    },
+    {
+      "epoch": 0.9815852682145717,
+      "learning_rate": 1.546825000113736e-05,
+      "loss": 0.1275,
+      "step": 4904
+    },
+    {
+      "epoch": 0.9823859087269816,
+      "learning_rate": 1.545654409218794e-05,
+      "loss": 0.6113,
+      "step": 4906
+    },
+    {
+      "epoch": 0.9823859087269816,
+      "learning_rate": 1.544482752648966e-05,
+      "loss": 0.2083,
+      "step": 4908
+    },
+    {
+      "epoch": 0.9831865492393915,
+      "learning_rate": 1.5433100326925298e-05,
+      "loss": 0.0843,
+      "step": 4910
+    },
+    {
+      "epoch": 0.9831865492393915,
+      "learning_rate": 1.5421362516398285e-05,
+      "loss": 0.0569,
+      "step": 4912
+    },
+    {
+      "epoch": 0.9839871897518014,
+      "learning_rate": 1.5409614117832797e-05,
+      "loss": 0.4227,
+      "step": 4914
+    },
+    {
+      "epoch": 0.9839871897518014,
+      "learning_rate": 1.539785515417377e-05,
+      "loss": 0.0934,
+      "step": 4916
+    },
+    {
+      "epoch": 0.9847878302642114,
+      "learning_rate": 1.538608564838665e-05,
+      "loss": 0.153,
+      "step": 4918
+    },
+    {
+      "epoch": 0.9847878302642114,
+      "learning_rate": 1.5374305623457605e-05,
+      "loss": 0.0904,
+      "step": 4920
+    },
+    {
+      "epoch": 0.9855884707766213,
+      "learning_rate": 1.5362515102393244e-05,
+      "loss": 0.1268,
+      "step": 4922
+    },
+    {
+      "epoch": 0.9855884707766213,
+      "learning_rate": 1.5350714108220677e-05,
+      "loss": 0.5589,
+      "step": 4924
+    },
+    {
+      "epoch": 0.9863891112890312,
+      "learning_rate": 1.5338902663987564e-05,
+      "loss": 0.1444,
+      "step": 4926
+    },
+    {
+      "epoch": 0.9863891112890312,
+      "learning_rate": 1.532708079276186e-05,
+      "loss": 0.1909,
+      "step": 4928
+    },
+    {
+      "epoch": 0.9871897518014412,
+      "learning_rate": 1.531524851763198e-05,
+      "loss": 0.1268,
+      "step": 4930
+    },
+    {
+      "epoch": 0.9871897518014412,
+      "learning_rate": 1.5303405861706567e-05,
+      "loss": 0.1504,
+      "step": 4932
+    },
+    {
+      "epoch": 0.9879903923138511,
+      "learning_rate": 1.529155284811464e-05,
+      "loss": 0.1542,
+      "step": 4934
+    },
+    {
+      "epoch": 0.9879903923138511,
+      "learning_rate": 1.5279689500005353e-05,
+      "loss": 0.0221,
+      "step": 4936
+    },
+    {
+      "epoch": 0.988791032826261,
+      "learning_rate": 1.5267815840548067e-05,
+      "loss": 0.4188,
+      "step": 4938
+    },
+    {
+      "epoch": 0.988791032826261,
+      "learning_rate": 1.5255931892932344e-05,
+      "loss": 0.3513,
+      "step": 4940
+    },
+    {
+      "epoch": 0.9895916733386709,
+      "learning_rate": 1.5244037680367739e-05,
+      "loss": 0.1664,
+      "step": 4942
+    },
+    {
+      "epoch": 0.9895916733386709,
+      "learning_rate": 1.5232133226083962e-05,
+      "loss": 0.1242,
+      "step": 4944
+    },
+    {
+      "epoch": 0.9903923138510808,
+      "learning_rate": 1.522021855333061e-05,
+      "loss": 0.1204,
+      "step": 4946
+    },
+    {
+      "epoch": 0.9903923138510808,
+      "learning_rate": 1.5208293685377362e-05,
+      "loss": 0.2119,
+      "step": 4948
+    },
+    {
+      "epoch": 0.9911929543634908,
+      "learning_rate": 1.519635864551371e-05,
+      "loss": 0.1162,
+      "step": 4950
+    },
+    {
+      "epoch": 0.9911929543634908,
+      "learning_rate": 1.5184413457049014e-05,
+      "loss": 0.0457,
+      "step": 4952
+    },
+    {
+      "epoch": 0.9919935948759008,
+      "learning_rate": 1.5172458143312548e-05,
+      "loss": 0.2387,
+      "step": 4954
+    },
+    {
+      "epoch": 0.9919935948759008,
+      "learning_rate": 1.5160492727653238e-05,
+      "loss": 0.034,
+      "step": 4956
+    },
+    {
+      "epoch": 0.9927942353883107,
+      "learning_rate": 1.5148517233439858e-05,
+      "loss": 0.2563,
+      "step": 4958
+    },
+    {
+      "epoch": 0.9927942353883107,
+      "learning_rate": 1.5136531684060753e-05,
+      "loss": 0.2303,
+      "step": 4960
+    },
+    {
+      "epoch": 0.9935948759007206,
+      "learning_rate": 1.512453610292402e-05,
+      "loss": 0.3777,
+      "step": 4962
+    },
+    {
+      "epoch": 0.9935948759007206,
+      "learning_rate": 1.5112530513457251e-05,
+      "loss": 0.0183,
+      "step": 4964
+    },
+    {
+      "epoch": 0.9943955164131305,
+      "learning_rate": 1.5100514939107598e-05,
+      "loss": 0.1849,
+      "step": 4966
+    },
+    {
+      "epoch": 0.9943955164131305,
+      "learning_rate": 1.50884894033418e-05,
+      "loss": 0.0986,
+      "step": 4968
+    },
+    {
+      "epoch": 0.9951961569255404,
+      "learning_rate": 1.5076453929645927e-05,
+      "loss": 0.697,
+      "step": 4970
+    },
+    {
+      "epoch": 0.9951961569255404,
+      "learning_rate": 1.5064408541525578e-05,
+      "loss": 0.0108,
+      "step": 4972
+    },
+    {
+      "epoch": 0.9959967974379503,
+      "learning_rate": 1.505235326250563e-05,
+      "loss": 0.4589,
+      "step": 4974
+    },
+    {
+      "epoch": 0.9959967974379503,
+      "learning_rate": 1.504028811613027e-05,
+      "loss": 0.0098,
+      "step": 4976
+    },
+    {
+      "epoch": 0.9967974379503602,
+      "learning_rate": 1.5028213125963054e-05,
+      "loss": 0.3555,
+      "step": 4978
+    },
+    {
+      "epoch": 0.9967974379503602,
+      "learning_rate": 1.5016128315586636e-05,
+      "loss": 0.2235,
+      "step": 4980
+    },
+    {
+      "epoch": 0.9975980784627703,
+      "learning_rate": 1.5004033708602977e-05,
+      "loss": 0.1274,
+      "step": 4982
+    },
+    {
+      "epoch": 0.9975980784627703,
+      "learning_rate": 1.4991929328633043e-05,
+      "loss": 0.0136,
+      "step": 4984
+    },
+    {
+      "epoch": 0.9983987189751802,
+      "learning_rate": 1.4979815199317011e-05,
+      "loss": 0.9098,
+      "step": 4986
+    },
+    {
+      "epoch": 0.9983987189751802,
+      "learning_rate": 1.4967691344314012e-05,
+      "loss": 0.1939,
+      "step": 4988
+    },
+    {
+      "epoch": 0.9991993594875901,
+      "learning_rate": 1.495555778730216e-05,
+      "loss": 0.6106,
+      "step": 4990
+    },
+    {
+      "epoch": 0.9991993594875901,
+      "learning_rate": 1.4943414551978622e-05,
+      "loss": 0.017,
+      "step": 4992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4931261662059333e-05,
+      "loss": 0.1969,
+      "step": 4994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4919099141279214e-05,
+      "loss": 0.7851,
+      "step": 4996
+    },
+    {
+      "epoch": 1.0,
+      "step": 4996,
+      "total_flos": 2.927145290039296e+16,
+      "train_loss": 0.27889094776088064,
+      "train_runtime": 8503.2779,
+      "train_samples_per_second": 2.35,
+      "train_steps_per_second": 0.588
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 4996,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {},
+  "total_flos": 2.927145290039296e+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_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round0.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..4929dcb1326f376d0a13570f2ce34ce154591bef
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1429b8638f8f26d39670c2c450d26e04ee5e6aacd66d17bcbeb21b4bbed2dd21
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round1.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..2a86b1642ac9562118e2b50ee561acd5e307e618
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f433e41c0774412e2d54624d75e19252ea86c1417a4ab460e9092764d81bb2c
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round2.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..d1117fbe79c6733118e13c14e174db95b3f81b67
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c858add8c9c9bc9679ef6ea731c6e25961adf98717bd17aba1b4a143012839b
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round3.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..e4db193800c1d3a9c89a75437e6ba7485f96c043
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_infoBatch_scenario12_new_10000_random0_25_seed1/server_model_round3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a3b7595208c3e5400a1cf7e2264a15968c0770649c7ef2df76f2045cda7c2e8
+size 639793378