update

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "t5-small",
+  "_name_or_path": "longcld/t5_small_qg_ae_hl",
   "architectures": [
     "T5ForConditionalGeneration"
   ],

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:98ca37835bcddf48996503f7299836e2a04857820a5ec456d26f28e72482837b
+oid sha256:168e33246477e3a85744128633fb3dfd79b076404ce3a29d6c2346111006b1a0
 size 352532601

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f71978eee39d0add93aaa7bf352d064f47c5b9459b744cfbf1c8465f8c94f5fc
+oid sha256:7db79b4ce90d22e321850f42ab2daa15bd8415d9927db2430e9162f1f050d229
 size 688496379

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:da0ccaa71ad9f8a40a5e05c29bc9d194a6c545c6f823a6102e265e33e2a4d730
+oid sha256:86b89038cec6aad84c2a0ec30e8837c3126fc83e216383d5c7050db0c0bb6603
 size 14503

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:312d4f5e5fefe6779de712dbbac5451ceaf780a569f970e562db2a2bf796906f
+oid sha256:6170a5404e788cd60947c893748928a641e3aee90b31a565d273eb3f39aaf38e
 size 623

tokenizer_config.json

@@ -1 +1 @@
-{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "extra_ids": 0, "additional_special_tokens": null, "sp_model_kwargs": {}, "special_tokens_map_file": "C:\\Users\\hydra/.cache\\huggingface\\transformers\\bde22e2f08347211bf47b72fb4730b4b9a2e3323a6a1254932cf0fb969eb18cd.b02fba13e4584f40329b20be0bad88c2ef97b5fda7d9f4f69b1cb3a620568a86", "tokenizer_file": null, "name_or_path": "NlpHUST/t5-vi-en-small", "tokenizer_class": "T5Tokenizer"}
+{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>", "extra_ids": 0, "additional_special_tokens": null, "sp_model_kwargs": {}, "special_tokens_map_file": "C:\\Users\\hydra/.cache\\huggingface\\transformers\\bde22e2f08347211bf47b72fb4730b4b9a2e3323a6a1254932cf0fb969eb18cd.b02fba13e4584f40329b20be0bad88c2ef97b5fda7d9f4f69b1cb3a620568a86", "tokenizer_file": null, "name_or_path": "longcld/t5_small_qg_ae_hl", "tokenizer_class": "T5Tokenizer"}

trainer_state.json

@@ -1,1306 +1,2596 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 9.392683592683593,
-  "global_step": 21500,
+  "epoch": 5.366926677067083,
+  "global_step": 43000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.04,
-      "learning_rate": 9.956312800349498e-05,
-      "loss": 8.326,
+      "epoch": 0.01,
+      "learning_rate": 9.987518721917126e-06,
+      "loss": 2.674,
       "step": 100
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 9.912625600698995e-05,
-      "loss": 5.3566,
+      "epoch": 0.02,
+      "learning_rate": 9.97503744383425e-06,
+      "loss": 2.5884,
       "step": 200
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 9.868938401048493e-05,
-      "loss": 4.887,
+      "epoch": 0.04,
+      "learning_rate": 9.962556165751374e-06,
+      "loss": 2.614,
       "step": 300
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 9.825251201397991e-05,
-      "loss": 4.6188,
+      "epoch": 0.05,
+      "learning_rate": 9.950074887668498e-06,
+      "loss": 2.6229,
       "step": 400
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 9.781564001747489e-05,
-      "loss": 4.4699,
+      "epoch": 0.06,
+      "learning_rate": 9.937593609585623e-06,
+      "loss": 2.6261,
       "step": 500
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 9.737876802096986e-05,
-      "loss": 4.371,
+      "epoch": 0.07,
+      "learning_rate": 9.925112331502746e-06,
+      "loss": 2.6,
       "step": 600
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 9.694189602446484e-05,
-      "loss": 4.1883,
+      "epoch": 0.09,
+      "learning_rate": 9.912631053419871e-06,
+      "loss": 2.6558,
       "step": 700
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 9.650502402795981e-05,
-      "loss": 4.1138,
+      "epoch": 0.1,
+      "learning_rate": 9.900149775336996e-06,
+      "loss": 2.5839,
       "step": 800
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 9.606815203145478e-05,
-      "loss": 4.0091,
+      "epoch": 0.11,
+      "learning_rate": 9.88766849725412e-06,
+      "loss": 2.6585,
       "step": 900
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 9.563128003494977e-05,
-      "loss": 3.9716,
+      "epoch": 0.12,
+      "learning_rate": 9.875187219171243e-06,
+      "loss": 2.5986,
       "step": 1000
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 9.519440803844473e-05,
-      "loss": 3.8813,
+      "epoch": 0.14,
+      "learning_rate": 9.862705941088368e-06,
+      "loss": 2.6437,
       "step": 1100
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 9.475753604193972e-05,
-      "loss": 3.8122,
+      "epoch": 0.15,
+      "learning_rate": 9.850224663005492e-06,
+      "loss": 2.5858,
       "step": 1200
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 9.432066404543469e-05,
-      "loss": 3.7892,
+      "epoch": 0.16,
+      "learning_rate": 9.837743384922617e-06,
+      "loss": 2.6142,
       "step": 1300
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 9.388379204892967e-05,
-      "loss": 3.7502,
+      "epoch": 0.17,
+      "learning_rate": 9.825262106839742e-06,
+      "loss": 2.5436,
       "step": 1400
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 9.344692005242464e-05,
-      "loss": 3.6924,
+      "epoch": 0.19,
+      "learning_rate": 9.812780828756865e-06,
+      "loss": 2.5879,
       "step": 1500
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 9.301004805591961e-05,
-      "loss": 3.6578,
+      "epoch": 0.2,
+      "learning_rate": 9.80029955067399e-06,
+      "loss": 2.6459,
       "step": 1600
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 9.25731760594146e-05,
-      "loss": 3.6732,
+      "epoch": 0.21,
+      "learning_rate": 9.787818272591114e-06,
+      "loss": 2.6185,
       "step": 1700
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 9.213630406290958e-05,
-      "loss": 3.5946,
+      "epoch": 0.22,
+      "learning_rate": 9.775336994508239e-06,
+      "loss": 2.6672,
       "step": 1800
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 9.169943206640455e-05,
-      "loss": 3.5663,
+      "epoch": 0.24,
+      "learning_rate": 9.762855716425362e-06,
+      "loss": 2.6541,
       "step": 1900
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 9.126256006989952e-05,
-      "loss": 3.4855,
+      "epoch": 0.25,
+      "learning_rate": 9.750374438342487e-06,
+      "loss": 2.6669,
       "step": 2000
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 9.08256880733945e-05,
-      "loss": 3.4893,
+      "epoch": 0.26,
+      "learning_rate": 9.737893160259612e-06,
+      "loss": 2.6323,
       "step": 2100
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 9.038881607688947e-05,
-      "loss": 3.4512,
+      "epoch": 0.27,
+      "learning_rate": 9.725411882176736e-06,
+      "loss": 2.6764,
       "step": 2200
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 8.995194408038446e-05,
-      "loss": 3.444,
+      "epoch": 0.29,
+      "learning_rate": 9.712930604093859e-06,
+      "loss": 2.5925,
       "step": 2300
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 8.951507208387942e-05,
-      "loss": 3.4151,
+      "epoch": 0.3,
+      "learning_rate": 9.700449326010984e-06,
+      "loss": 2.5392,
       "step": 2400
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 8.90782000873744e-05,
-      "loss": 3.3786,
+      "epoch": 0.31,
+      "learning_rate": 9.68796804792811e-06,
+      "loss": 2.5843,
       "step": 2500
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.864132809086938e-05,
-      "loss": 3.3509,
+      "epoch": 0.32,
+      "learning_rate": 9.675486769845233e-06,
+      "loss": 2.6209,
       "step": 2600
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 8.820445609436435e-05,
-      "loss": 3.3527,
+      "epoch": 0.34,
+      "learning_rate": 9.663005491762358e-06,
+      "loss": 2.6575,
       "step": 2700
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 8.776758409785933e-05,
-      "loss": 3.2981,
+      "epoch": 0.35,
+      "learning_rate": 9.650524213679481e-06,
+      "loss": 2.6499,
       "step": 2800
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 8.73307121013543e-05,
-      "loss": 3.2862,
+      "epoch": 0.36,
+      "learning_rate": 9.638042935596606e-06,
+      "loss": 2.5762,
       "step": 2900
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 8.689384010484928e-05,
-      "loss": 3.2791,
+      "epoch": 0.37,
+      "learning_rate": 9.62556165751373e-06,
+      "loss": 2.5637,
       "step": 3000
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 8.645696810834426e-05,
-      "loss": 3.2725,
+      "epoch": 0.39,
+      "learning_rate": 9.613080379430855e-06,
+      "loss": 2.5902,
       "step": 3100
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 8.602009611183924e-05,
-      "loss": 3.2218,
+      "epoch": 0.4,
+      "learning_rate": 9.60059910134798e-06,
+      "loss": 2.5784,
       "step": 3200
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 8.558322411533421e-05,
-      "loss": 3.2012,
+      "epoch": 0.41,
+      "learning_rate": 9.588117823265103e-06,
+      "loss": 2.6246,
       "step": 3300
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 8.514635211882919e-05,
-      "loss": 3.1842,
+      "epoch": 0.42,
+      "learning_rate": 9.575636545182228e-06,
+      "loss": 2.6543,
       "step": 3400
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 8.470948012232416e-05,
-      "loss": 3.2005,
+      "epoch": 0.44,
+      "learning_rate": 9.563155267099352e-06,
+      "loss": 2.581,
       "step": 3500
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 8.427260812581915e-05,
-      "loss": 3.1484,
+      "epoch": 0.45,
+      "learning_rate": 9.550673989016475e-06,
+      "loss": 2.6039,
       "step": 3600
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 8.383573612931411e-05,
-      "loss": 3.1521,
+      "epoch": 0.46,
+      "learning_rate": 9.5381927109336e-06,
+      "loss": 2.5848,
       "step": 3700
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 8.33988641328091e-05,
-      "loss": 3.1207,
+      "epoch": 0.47,
+      "learning_rate": 9.525711432850725e-06,
+      "loss": 2.6026,
       "step": 3800
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 8.296199213630407e-05,
-      "loss": 3.1528,
+      "epoch": 0.49,
+      "learning_rate": 9.513230154767849e-06,
+      "loss": 2.6178,
       "step": 3900
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.252512013979904e-05,
-      "loss": 3.1201,
+      "epoch": 0.5,
+      "learning_rate": 9.500748876684974e-06,
+      "loss": 2.5908,
       "step": 4000
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 8.208824814329402e-05,
-      "loss": 3.1162,
+      "epoch": 0.51,
+      "learning_rate": 9.488267598602097e-06,
+      "loss": 2.599,
       "step": 4100
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 8.165137614678899e-05,
-      "loss": 3.1381,
+      "epoch": 0.52,
+      "learning_rate": 9.475786320519222e-06,
+      "loss": 2.5937,
       "step": 4200
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 8.121450415028396e-05,
-      "loss": 3.0811,
+      "epoch": 0.54,
+      "learning_rate": 9.463305042436346e-06,
+      "loss": 2.6038,
       "step": 4300
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.077763215377895e-05,
-      "loss": 3.0746,
+      "epoch": 0.55,
+      "learning_rate": 9.45082376435347e-06,
+      "loss": 2.5825,
       "step": 4400
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 8.034076015727393e-05,
-      "loss": 3.0438,
+      "epoch": 0.56,
+      "learning_rate": 9.438342486270596e-06,
+      "loss": 2.5668,
       "step": 4500
     },
     {
-      "epoch": 2.01,
-      "learning_rate": 7.99038881607689e-05,
-      "loss": 3.0947,
+      "epoch": 0.57,
+      "learning_rate": 9.42586120818772e-06,
+      "loss": 2.6231,
       "step": 4600
     },
     {
-      "epoch": 2.05,
-      "learning_rate": 7.946701616426387e-05,
-      "loss": 2.9938,
+      "epoch": 0.59,
+      "learning_rate": 9.413379930104844e-06,
+      "loss": 2.5955,
       "step": 4700
     },
     {
-      "epoch": 2.1,
-      "learning_rate": 7.903014416775885e-05,
-      "loss": 3.0187,
+      "epoch": 0.6,
+      "learning_rate": 9.400898652021968e-06,
+      "loss": 2.6169,
       "step": 4800
     },
     {
-      "epoch": 2.14,
-      "learning_rate": 7.859327217125383e-05,
-      "loss": 3.0137,
+      "epoch": 0.61,
+      "learning_rate": 9.388417373939091e-06,
+      "loss": 2.5865,
       "step": 4900
     },
     {
-      "epoch": 2.18,
-      "learning_rate": 7.81564001747488e-05,
-      "loss": 2.9832,
+      "epoch": 0.62,
+      "learning_rate": 9.375936095856216e-06,
+      "loss": 2.5837,
       "step": 5000
     },
     {
-      "epoch": 2.23,
-      "learning_rate": 7.771952817824378e-05,
-      "loss": 2.9913,
+      "epoch": 0.64,
+      "learning_rate": 9.363454817773341e-06,
+      "loss": 2.5994,
       "step": 5100
     },
     {
-      "epoch": 2.27,
-      "learning_rate": 7.728265618173876e-05,
-      "loss": 2.9714,
+      "epoch": 0.65,
+      "learning_rate": 9.350973539690466e-06,
+      "loss": 2.6152,
       "step": 5200
     },
     {
-      "epoch": 2.32,
-      "learning_rate": 7.684578418523373e-05,
-      "loss": 2.9774,
+      "epoch": 0.66,
+      "learning_rate": 9.33849226160759e-06,
+      "loss": 2.5968,
       "step": 5300
     },
     {
-      "epoch": 2.36,
-      "learning_rate": 7.64089121887287e-05,
-      "loss": 3.0163,
+      "epoch": 0.67,
+      "learning_rate": 9.326010983524713e-06,
+      "loss": 2.5876,
       "step": 5400
     },
     {
-      "epoch": 2.4,
-      "learning_rate": 7.597204019222368e-05,
-      "loss": 2.9823,
+      "epoch": 0.69,
+      "learning_rate": 9.313529705441838e-06,
+      "loss": 2.5356,
       "step": 5500
     },
     {
-      "epoch": 2.45,
-      "learning_rate": 7.553516819571865e-05,
-      "loss": 2.9861,
+      "epoch": 0.7,
+      "learning_rate": 9.301048427358962e-06,
+      "loss": 2.6401,
       "step": 5600
     },
     {
-      "epoch": 2.49,
-      "learning_rate": 7.509829619921364e-05,
-      "loss": 2.942,
+      "epoch": 0.71,
+      "learning_rate": 9.288567149276087e-06,
+      "loss": 2.6072,
       "step": 5700
     },
     {
-      "epoch": 2.53,
-      "learning_rate": 7.466142420270861e-05,
-      "loss": 2.9326,
+      "epoch": 0.72,
+      "learning_rate": 9.276085871193212e-06,
+      "loss": 2.6391,
       "step": 5800
     },
     {
-      "epoch": 2.58,
-      "learning_rate": 7.422455220620359e-05,
-      "loss": 2.9434,
+      "epoch": 0.74,
+      "learning_rate": 9.263604593110335e-06,
+      "loss": 2.6492,
       "step": 5900
     },
     {
-      "epoch": 2.62,
-      "learning_rate": 7.378768020969856e-05,
-      "loss": 2.9462,
+      "epoch": 0.75,
+      "learning_rate": 9.251123315027458e-06,
+      "loss": 2.5952,
       "step": 6000
     },
     {
-      "epoch": 2.66,
-      "learning_rate": 7.335080821319354e-05,
-      "loss": 2.9233,
+      "epoch": 0.76,
+      "learning_rate": 9.238642036944584e-06,
+      "loss": 2.606,
       "step": 6100
     },
     {
-      "epoch": 2.71,
-      "learning_rate": 7.291393621668852e-05,
-      "loss": 2.9345,
+      "epoch": 0.77,
+      "learning_rate": 9.226160758861709e-06,
+      "loss": 2.6346,
       "step": 6200
     },
     {
-      "epoch": 2.75,
-      "learning_rate": 7.247706422018348e-05,
-      "loss": 2.9106,
+      "epoch": 0.79,
+      "learning_rate": 9.213679480778832e-06,
+      "loss": 2.6238,
       "step": 6300
     },
     {
-      "epoch": 2.8,
-      "learning_rate": 7.204019222367847e-05,
-      "loss": 2.8783,
+      "epoch": 0.8,
+      "learning_rate": 9.201198202695957e-06,
+      "loss": 2.5924,
       "step": 6400
     },
     {
-      "epoch": 2.84,
-      "learning_rate": 7.160332022717344e-05,
-      "loss": 2.9428,
+      "epoch": 0.81,
+      "learning_rate": 9.188716924613082e-06,
+      "loss": 2.6511,
       "step": 6500
     },
     {
-      "epoch": 2.88,
-      "learning_rate": 7.116644823066842e-05,
-      "loss": 2.9045,
+      "epoch": 0.82,
+      "learning_rate": 9.176235646530206e-06,
+      "loss": 2.5397,
       "step": 6600
     },
     {
-      "epoch": 2.93,
-      "learning_rate": 7.072957623416339e-05,
-      "loss": 2.8842,
+      "epoch": 0.84,
+      "learning_rate": 9.163754368447329e-06,
+      "loss": 2.6166,
       "step": 6700
     },
     {
-      "epoch": 2.97,
-      "learning_rate": 7.029270423765837e-05,
-      "loss": 2.8984,
+      "epoch": 0.85,
+      "learning_rate": 9.151273090364454e-06,
+      "loss": 2.5976,
       "step": 6800
     },
     {
-      "epoch": 3.01,
-      "learning_rate": 6.985583224115334e-05,
-      "loss": 2.9212,
+      "epoch": 0.86,
+      "learning_rate": 9.13879181228158e-06,
+      "loss": 2.5556,
       "step": 6900
     },
     {
-      "epoch": 3.06,
-      "learning_rate": 6.941896024464833e-05,
-      "loss": 2.8536,
+      "epoch": 0.87,
+      "learning_rate": 9.126310534198703e-06,
+      "loss": 2.5568,
       "step": 7000
     },
     {
-      "epoch": 3.1,
-      "learning_rate": 6.89820882481433e-05,
-      "loss": 2.8642,
+      "epoch": 0.89,
+      "learning_rate": 9.113829256115828e-06,
+      "loss": 2.5994,
       "step": 7100
     },
     {
-      "epoch": 3.15,
-      "learning_rate": 6.854521625163828e-05,
-      "loss": 2.8586,
+      "epoch": 0.9,
+      "learning_rate": 9.101347978032951e-06,
+      "loss": 2.5802,
       "step": 7200
     },
     {
-      "epoch": 3.19,
-      "learning_rate": 6.810834425513325e-05,
-      "loss": 2.8457,
+      "epoch": 0.91,
+      "learning_rate": 9.088866699950074e-06,
+      "loss": 2.5724,
       "step": 7300
     },
     {
-      "epoch": 3.23,
-      "learning_rate": 6.767147225862822e-05,
-      "loss": 2.8056,
+      "epoch": 0.92,
+      "learning_rate": 9.0763854218672e-06,
+      "loss": 2.5317,
       "step": 7400
     },
     {
-      "epoch": 3.28,
-      "learning_rate": 6.723460026212321e-05,
-      "loss": 2.8255,
+      "epoch": 0.94,
+      "learning_rate": 9.063904143784325e-06,
+      "loss": 2.6202,
       "step": 7500
     },
     {
-      "epoch": 3.32,
-      "learning_rate": 6.679772826561817e-05,
-      "loss": 2.8473,
+      "epoch": 0.95,
+      "learning_rate": 9.051422865701448e-06,
+      "loss": 2.5479,
       "step": 7600
     },
     {
-      "epoch": 3.36,
-      "learning_rate": 6.636085626911316e-05,
-      "loss": 2.848,
+      "epoch": 0.96,
+      "learning_rate": 9.038941587618573e-06,
+      "loss": 2.5918,
       "step": 7700
     },
     {
-      "epoch": 3.41,
-      "learning_rate": 6.592398427260813e-05,
-      "loss": 2.8232,
+      "epoch": 0.97,
+      "learning_rate": 9.026460309535696e-06,
+      "loss": 2.5706,
       "step": 7800
     },
     {
-      "epoch": 3.45,
-      "learning_rate": 6.54871122761031e-05,
-      "loss": 2.811,
+      "epoch": 0.99,
+      "learning_rate": 9.013979031452822e-06,
+      "loss": 2.5578,
       "step": 7900
     },
     {
-      "epoch": 3.49,
-      "learning_rate": 6.505024027959808e-05,
-      "loss": 2.8616,
+      "epoch": 1.0,
+      "learning_rate": 9.001497753369945e-06,
+      "loss": 2.6592,
       "step": 8000
     },
     {
-      "epoch": 3.54,
-      "learning_rate": 6.461336828309305e-05,
-      "loss": 2.8362,
+      "epoch": 1.01,
+      "learning_rate": 8.98901647528707e-06,
+      "loss": 2.6137,
       "step": 8100
     },
     {
-      "epoch": 3.58,
-      "learning_rate": 6.417649628658803e-05,
-      "loss": 2.8052,
+      "epoch": 1.02,
+      "learning_rate": 8.976535197204195e-06,
+      "loss": 2.5963,
       "step": 8200
     },
     {
-      "epoch": 3.63,
-      "learning_rate": 6.373962429008302e-05,
-      "loss": 2.8091,
+      "epoch": 1.04,
+      "learning_rate": 8.964053919121319e-06,
+      "loss": 2.5891,
       "step": 8300
     },
     {
-      "epoch": 3.67,
-      "learning_rate": 6.330275229357799e-05,
-      "loss": 2.7864,
+      "epoch": 1.05,
+      "learning_rate": 8.951572641038444e-06,
+      "loss": 2.5719,
       "step": 8400
     },
     {
-      "epoch": 3.71,
-      "learning_rate": 6.286588029707296e-05,
-      "loss": 2.832,
+      "epoch": 1.06,
+      "learning_rate": 8.939091362955567e-06,
+      "loss": 2.5967,
       "step": 8500
     },
     {
-      "epoch": 3.76,
-      "learning_rate": 6.242900830056794e-05,
-      "loss": 2.8012,
+      "epoch": 1.07,
+      "learning_rate": 8.926610084872692e-06,
+      "loss": 2.5656,
       "step": 8600
     },
     {
-      "epoch": 3.8,
-      "learning_rate": 6.199213630406291e-05,
-      "loss": 2.784,
+      "epoch": 1.09,
+      "learning_rate": 8.914128806789815e-06,
+      "loss": 2.6024,
       "step": 8700
     },
     {
-      "epoch": 3.84,
-      "learning_rate": 6.155526430755788e-05,
-      "loss": 2.777,
+      "epoch": 1.1,
+      "learning_rate": 8.90164752870694e-06,
+      "loss": 2.566,
       "step": 8800
     },
     {
-      "epoch": 3.89,
-      "learning_rate": 6.111839231105286e-05,
-      "loss": 2.7777,
+      "epoch": 1.11,
+      "learning_rate": 8.889166250624066e-06,
+      "loss": 2.5757,
       "step": 8900
     },
     {
-      "epoch": 3.93,
-      "learning_rate": 6.068152031454783e-05,
-      "loss": 2.7256,
+      "epoch": 1.12,
+      "learning_rate": 8.876684972541189e-06,
+      "loss": 2.5595,
       "step": 9000
     },
     {
-      "epoch": 3.98,
-      "learning_rate": 6.024464831804282e-05,
-      "loss": 2.7666,
+      "epoch": 1.14,
+      "learning_rate": 8.864203694458312e-06,
+      "loss": 2.605,
       "step": 9100
     },
     {
-      "epoch": 4.02,
-      "learning_rate": 5.980777632153779e-05,
-      "loss": 2.7656,
+      "epoch": 1.15,
+      "learning_rate": 8.851722416375437e-06,
+      "loss": 2.594,
       "step": 9200
     },
     {
-      "epoch": 4.06,
-      "learning_rate": 5.9370904325032775e-05,
-      "loss": 2.7678,
+      "epoch": 1.16,
+      "learning_rate": 8.839241138292561e-06,
+      "loss": 2.6241,
       "step": 9300
     },
     {
-      "epoch": 4.11,
-      "learning_rate": 5.893403232852774e-05,
-      "loss": 2.7277,
+      "epoch": 1.17,
+      "learning_rate": 8.826759860209686e-06,
+      "loss": 2.6328,
       "step": 9400
     },
     {
-      "epoch": 4.15,
-      "learning_rate": 5.8497160332022716e-05,
-      "loss": 2.7571,
+      "epoch": 1.19,
+      "learning_rate": 8.814278582126811e-06,
+      "loss": 2.5948,
       "step": 9500
     },
     {
-      "epoch": 4.19,
-      "learning_rate": 5.8060288335517697e-05,
-      "loss": 2.7641,
+      "epoch": 1.2,
+      "learning_rate": 8.801797304043934e-06,
+      "loss": 2.5232,
       "step": 9600
     },
     {
-      "epoch": 4.24,
-      "learning_rate": 5.762341633901267e-05,
-      "loss": 2.757,
+      "epoch": 1.21,
+      "learning_rate": 8.78931602596106e-06,
+      "loss": 2.614,
       "step": 9700
     },
     {
-      "epoch": 4.28,
-      "learning_rate": 5.718654434250765e-05,
-      "loss": 2.7147,
+      "epoch": 1.22,
+      "learning_rate": 8.776834747878183e-06,
+      "loss": 2.561,
       "step": 9800
     },
     {
-      "epoch": 4.32,
-      "learning_rate": 5.6749672346002625e-05,
-      "loss": 2.7301,
+      "epoch": 1.24,
+      "learning_rate": 8.764353469795308e-06,
+      "loss": 2.6071,
       "step": 9900
     },
     {
-      "epoch": 4.37,
-      "learning_rate": 5.63128003494976e-05,
-      "loss": 2.6818,
+      "epoch": 1.25,
+      "learning_rate": 8.751872191712431e-06,
+      "loss": 2.5902,
       "step": 10000
     },
     {
-      "epoch": 4.41,
-      "learning_rate": 5.587592835299258e-05,
-      "loss": 2.7148,
+      "epoch": 1.26,
+      "learning_rate": 8.739390913629556e-06,
+      "loss": 2.5653,
       "step": 10100
     },
     {
-      "epoch": 4.46,
-      "learning_rate": 5.5439056356487554e-05,
-      "loss": 2.7291,
+      "epoch": 1.27,
+      "learning_rate": 8.726909635546682e-06,
+      "loss": 2.5746,
       "step": 10200
     },
     {
-      "epoch": 4.5,
-      "learning_rate": 5.500218435998252e-05,
-      "loss": 2.7499,
+      "epoch": 1.29,
+      "learning_rate": 8.714428357463805e-06,
+      "loss": 2.525,
       "step": 10300
     },
     {
-      "epoch": 4.54,
-      "learning_rate": 5.456531236347751e-05,
-      "loss": 2.7162,
+      "epoch": 1.3,
+      "learning_rate": 8.701947079380928e-06,
+      "loss": 2.6418,
       "step": 10400
     },
     {
-      "epoch": 4.59,
-      "learning_rate": 5.4128440366972475e-05,
-      "loss": 2.7226,
+      "epoch": 1.31,
+      "learning_rate": 8.689465801298053e-06,
+      "loss": 2.6579,
       "step": 10500
     },
     {
-      "epoch": 4.63,
-      "learning_rate": 5.369156837046746e-05,
-      "loss": 2.7014,
+      "epoch": 1.32,
+      "learning_rate": 8.676984523215179e-06,
+      "loss": 2.5815,
       "step": 10600
     },
     {
-      "epoch": 4.67,
-      "learning_rate": 5.325469637396243e-05,
-      "loss": 2.7104,
+      "epoch": 1.34,
+      "learning_rate": 8.664503245132302e-06,
+      "loss": 2.6014,
       "step": 10700
     },
     {
-      "epoch": 4.72,
-      "learning_rate": 5.2817824377457404e-05,
-      "loss": 2.7589,
+      "epoch": 1.35,
+      "learning_rate": 8.652021967049427e-06,
+      "loss": 2.6233,
       "step": 10800
     },
     {
-      "epoch": 4.76,
-      "learning_rate": 5.2380952380952384e-05,
-      "loss": 2.7083,
+      "epoch": 1.36,
+      "learning_rate": 8.63954068896655e-06,
+      "loss": 2.5808,
       "step": 10900
     },
     {
-      "epoch": 4.81,
-      "learning_rate": 5.194408038444736e-05,
-      "loss": 2.7493,
+      "epoch": 1.37,
+      "learning_rate": 8.627059410883675e-06,
+      "loss": 2.6013,
       "step": 11000
     },
     {
-      "epoch": 4.85,
-      "learning_rate": 5.150720838794234e-05,
-      "loss": 2.7038,
+      "epoch": 1.39,
+      "learning_rate": 8.614578132800799e-06,
+      "loss": 2.5305,
       "step": 11100
     },
     {
-      "epoch": 4.89,
-      "learning_rate": 5.107033639143731e-05,
-      "loss": 2.7278,
+      "epoch": 1.4,
+      "learning_rate": 8.602096854717924e-06,
+      "loss": 2.5726,
       "step": 11200
     },
     {
-      "epoch": 4.94,
-      "learning_rate": 5.063346439493229e-05,
-      "loss": 2.7034,
+      "epoch": 1.41,
+      "learning_rate": 8.589615576635049e-06,
+      "loss": 2.5862,
       "step": 11300
     },
     {
-      "epoch": 4.98,
-      "learning_rate": 5.019659239842727e-05,
-      "loss": 2.676,
+      "epoch": 1.42,
+      "learning_rate": 8.577134298552172e-06,
+      "loss": 2.5728,
       "step": 11400
     },
     {
-      "epoch": 5.02,
-      "learning_rate": 4.975972040192224e-05,
-      "loss": 2.6757,
+      "epoch": 1.44,
+      "learning_rate": 8.564653020469298e-06,
+      "loss": 2.5863,
       "step": 11500
     },
     {
-      "epoch": 5.07,
-      "learning_rate": 4.9322848405417215e-05,
-      "loss": 2.6979,
+      "epoch": 1.45,
+      "learning_rate": 8.552171742386421e-06,
+      "loss": 2.5121,
       "step": 11600
     },
     {
-      "epoch": 5.11,
-      "learning_rate": 4.8885976408912196e-05,
-      "loss": 2.6961,
+      "epoch": 1.46,
+      "learning_rate": 8.539690464303544e-06,
+      "loss": 2.583,
       "step": 11700
     },
     {
-      "epoch": 5.16,
-      "learning_rate": 4.844910441240716e-05,
-      "loss": 2.6735,
+      "epoch": 1.47,
+      "learning_rate": 8.52720918622067e-06,
+      "loss": 2.6128,
       "step": 11800
     },
     {
-      "epoch": 5.2,
-      "learning_rate": 4.8012232415902144e-05,
-      "loss": 2.6777,
+      "epoch": 1.49,
+      "learning_rate": 8.514727908137794e-06,
+      "loss": 2.5568,
       "step": 11900
     },
     {
-      "epoch": 5.24,
-      "learning_rate": 4.757536041939712e-05,
-      "loss": 2.652,
+      "epoch": 1.5,
+      "learning_rate": 8.502246630054918e-06,
+      "loss": 2.5913,
       "step": 12000
     },
     {
-      "epoch": 5.29,
-      "learning_rate": 4.71384884228921e-05,
-      "loss": 2.6824,
+      "epoch": 1.51,
+      "learning_rate": 8.489765351972043e-06,
+      "loss": 2.6139,
       "step": 12100
     },
     {
-      "epoch": 5.33,
-      "learning_rate": 4.6701616426387066e-05,
-      "loss": 2.6532,
+      "epoch": 1.52,
+      "learning_rate": 8.477284073889166e-06,
+      "loss": 2.5659,
       "step": 12200
     },
     {
-      "epoch": 5.37,
-      "learning_rate": 4.6264744429882046e-05,
-      "loss": 2.6828,
+      "epoch": 1.54,
+      "learning_rate": 8.464802795806291e-06,
+      "loss": 2.5676,
       "step": 12300
     },
     {
-      "epoch": 5.42,
-      "learning_rate": 4.582787243337702e-05,
-      "loss": 2.646,
+      "epoch": 1.55,
+      "learning_rate": 8.452321517723415e-06,
+      "loss": 2.5329,
       "step": 12400
     },
     {
-      "epoch": 5.46,
-      "learning_rate": 4.5391000436872e-05,
-      "loss": 2.6168,
+      "epoch": 1.56,
+      "learning_rate": 8.43984023964054e-06,
+      "loss": 2.5653,
       "step": 12500
     },
     {
-      "epoch": 5.5,
-      "learning_rate": 4.4954128440366975e-05,
-      "loss": 2.6751,
+      "epoch": 1.57,
+      "learning_rate": 8.427358961557665e-06,
+      "loss": 2.6023,
       "step": 12600
     },
     {
-      "epoch": 5.55,
-      "learning_rate": 4.451725644386195e-05,
-      "loss": 2.6647,
+      "epoch": 1.59,
+      "learning_rate": 8.414877683474788e-06,
+      "loss": 2.5692,
       "step": 12700
     },
     {
-      "epoch": 5.59,
-      "learning_rate": 4.408038444735693e-05,
-      "loss": 2.6263,
+      "epoch": 1.6,
+      "learning_rate": 8.402396405391913e-06,
+      "loss": 2.5947,
       "step": 12800
     },
     {
-      "epoch": 5.64,
-      "learning_rate": 4.36435124508519e-05,
-      "loss": 2.6676,
+      "epoch": 1.61,
+      "learning_rate": 8.389915127309037e-06,
+      "loss": 2.6291,
       "step": 12900
     },
     {
-      "epoch": 5.68,
-      "learning_rate": 4.3206640454346884e-05,
-      "loss": 2.6434,
+      "epoch": 1.62,
+      "learning_rate": 8.377433849226162e-06,
+      "loss": 2.5965,
       "step": 13000
     },
     {
-      "epoch": 5.72,
-      "learning_rate": 4.276976845784185e-05,
-      "loss": 2.6518,
+      "epoch": 1.64,
+      "learning_rate": 8.364952571143285e-06,
+      "loss": 2.5433,
       "step": 13100
     },
     {
-      "epoch": 5.77,
-      "learning_rate": 4.233289646133683e-05,
-      "loss": 2.6252,
+      "epoch": 1.65,
+      "learning_rate": 8.35247129306041e-06,
+      "loss": 2.5804,
       "step": 13200
     },
     {
-      "epoch": 5.81,
-      "learning_rate": 4.1896024464831806e-05,
-      "loss": 2.6726,
+      "epoch": 1.66,
+      "learning_rate": 8.339990014977535e-06,
+      "loss": 2.5296,
       "step": 13300
     },
     {
-      "epoch": 5.85,
-      "learning_rate": 4.1459152468326786e-05,
-      "loss": 2.6541,
+      "epoch": 1.67,
+      "learning_rate": 8.327508736894659e-06,
+      "loss": 2.6349,
       "step": 13400
     },
     {
-      "epoch": 5.9,
-      "learning_rate": 4.1022280471821753e-05,
-      "loss": 2.6694,
+      "epoch": 1.68,
+      "learning_rate": 8.315027458811782e-06,
+      "loss": 2.5424,
       "step": 13500
     },
     {
-      "epoch": 5.94,
-      "learning_rate": 4.0585408475316734e-05,
-      "loss": 2.6413,
+      "epoch": 1.7,
+      "learning_rate": 8.302546180728907e-06,
+      "loss": 2.6139,
       "step": 13600
     },
     {
-      "epoch": 5.98,
-      "learning_rate": 4.014853647881171e-05,
-      "loss": 2.6435,
+      "epoch": 1.71,
+      "learning_rate": 8.29006490264603e-06,
+      "loss": 2.585,
       "step": 13700
     },
     {
-      "epoch": 6.03,
-      "learning_rate": 3.971166448230669e-05,
-      "loss": 2.6185,
+      "epoch": 1.72,
+      "learning_rate": 8.277583624563156e-06,
+      "loss": 2.5734,
       "step": 13800
     },
     {
-      "epoch": 6.07,
-      "learning_rate": 3.927479248580166e-05,
-      "loss": 2.6527,
+      "epoch": 1.73,
+      "learning_rate": 8.265102346480281e-06,
+      "loss": 2.591,
       "step": 13900
     },
     {
-      "epoch": 6.12,
-      "learning_rate": 3.8837920489296637e-05,
-      "loss": 2.6014,
+      "epoch": 1.75,
+      "learning_rate": 8.252621068397404e-06,
+      "loss": 2.6089,
       "step": 14000
     },
     {
-      "epoch": 6.16,
-      "learning_rate": 3.840104849279161e-05,
-      "loss": 2.5959,
+      "epoch": 1.76,
+      "learning_rate": 8.24013979031453e-06,
+      "loss": 2.6216,
       "step": 14100
     },
     {
-      "epoch": 6.2,
-      "learning_rate": 3.796417649628659e-05,
-      "loss": 2.6461,
+      "epoch": 1.77,
+      "learning_rate": 8.227658512231653e-06,
+      "loss": 2.5636,
       "step": 14200
     },
     {
-      "epoch": 6.25,
-      "learning_rate": 3.7527304499781565e-05,
-      "loss": 2.6259,
+      "epoch": 1.78,
+      "learning_rate": 8.215177234148778e-06,
+      "loss": 2.5922,
       "step": 14300
     },
     {
-      "epoch": 6.29,
-      "learning_rate": 3.709043250327654e-05,
-      "loss": 2.6419,
+      "epoch": 1.8,
+      "learning_rate": 8.202695956065901e-06,
+      "loss": 2.5404,
       "step": 14400
     },
     {
-      "epoch": 6.33,
-      "learning_rate": 3.665356050677152e-05,
-      "loss": 2.6126,
+      "epoch": 1.81,
+      "learning_rate": 8.190214677983026e-06,
+      "loss": 2.6366,
       "step": 14500
     },
     {
-      "epoch": 6.38,
-      "learning_rate": 3.6216688510266494e-05,
-      "loss": 2.6329,
+      "epoch": 1.82,
+      "learning_rate": 8.177733399900151e-06,
+      "loss": 2.6201,
       "step": 14600
     },
     {
-      "epoch": 6.42,
-      "learning_rate": 3.5779816513761474e-05,
-      "loss": 2.5802,
+      "epoch": 1.83,
+      "learning_rate": 8.165252121817275e-06,
+      "loss": 2.5972,
       "step": 14700
     },
     {
-      "epoch": 6.47,
-      "learning_rate": 3.534294451725644e-05,
-      "loss": 2.5999,
+      "epoch": 1.85,
+      "learning_rate": 8.152770843734398e-06,
+      "loss": 2.5356,
       "step": 14800
     },
     {
-      "epoch": 6.51,
-      "learning_rate": 3.490607252075142e-05,
-      "loss": 2.6306,
+      "epoch": 1.86,
+      "learning_rate": 8.140289565651523e-06,
+      "loss": 2.6053,
       "step": 14900
     },
     {
-      "epoch": 6.55,
-      "learning_rate": 3.4469200524246396e-05,
-      "loss": 2.633,
+      "epoch": 1.87,
+      "learning_rate": 8.127808287568648e-06,
+      "loss": 2.5312,
       "step": 15000
     },
     {
-      "epoch": 6.6,
-      "learning_rate": 3.403232852774138e-05,
-      "loss": 2.6396,
+      "epoch": 1.88,
+      "learning_rate": 8.115327009485772e-06,
+      "loss": 2.5696,
       "step": 15100
     },
     {
-      "epoch": 6.64,
-      "learning_rate": 3.3595456531236344e-05,
-      "loss": 2.5996,
+      "epoch": 1.9,
+      "learning_rate": 8.102845731402897e-06,
+      "loss": 2.5485,
       "step": 15200
     },
     {
-      "epoch": 6.68,
-      "learning_rate": 3.3158584534731324e-05,
-      "loss": 2.5878,
+      "epoch": 1.91,
+      "learning_rate": 8.09036445332002e-06,
+      "loss": 2.5444,
       "step": 15300
     },
     {
-      "epoch": 6.73,
-      "learning_rate": 3.27217125382263e-05,
-      "loss": 2.6027,
+      "epoch": 1.92,
+      "learning_rate": 8.077883175237144e-06,
+      "loss": 2.5953,
       "step": 15400
     },
     {
-      "epoch": 6.77,
-      "learning_rate": 3.228484054172128e-05,
-      "loss": 2.6267,
+      "epoch": 1.93,
+      "learning_rate": 8.065401897154269e-06,
+      "loss": 2.5588,
       "step": 15500
     },
     {
-      "epoch": 6.82,
-      "learning_rate": 3.184796854521625e-05,
-      "loss": 2.6092,
+      "epoch": 1.95,
+      "learning_rate": 8.052920619071394e-06,
+      "loss": 2.5685,
       "step": 15600
     },
     {
-      "epoch": 6.86,
-      "learning_rate": 3.141109654871123e-05,
-      "loss": 2.5901,
+      "epoch": 1.96,
+      "learning_rate": 8.040439340988519e-06,
+      "loss": 2.6248,
       "step": 15700
     },
     {
-      "epoch": 6.9,
-      "learning_rate": 3.097422455220621e-05,
-      "loss": 2.6066,
+      "epoch": 1.97,
+      "learning_rate": 8.027958062905642e-06,
+      "loss": 2.5816,
       "step": 15800
     },
     {
-      "epoch": 6.95,
-      "learning_rate": 3.053735255570118e-05,
-      "loss": 2.5548,
+      "epoch": 1.98,
+      "learning_rate": 8.015476784822767e-06,
+      "loss": 2.5558,
       "step": 15900
     },
     {
-      "epoch": 6.99,
-      "learning_rate": 3.010048055919616e-05,
-      "loss": 2.5848,
+      "epoch": 2.0,
+      "learning_rate": 8.00299550673989e-06,
+      "loss": 2.6264,
       "step": 16000
     },
     {
-      "epoch": 7.03,
-      "learning_rate": 2.966360856269113e-05,
-      "loss": 2.5954,
+      "epoch": 2.01,
+      "learning_rate": 7.990514228657014e-06,
+      "loss": 2.5924,
       "step": 16100
     },
     {
-      "epoch": 7.08,
-      "learning_rate": 2.9226736566186107e-05,
-      "loss": 2.5726,
+      "epoch": 2.02,
+      "learning_rate": 7.97803295057414e-06,
+      "loss": 2.5352,
       "step": 16200
     },
     {
-      "epoch": 7.12,
-      "learning_rate": 2.8789864569681084e-05,
-      "loss": 2.5727,
+      "epoch": 2.03,
+      "learning_rate": 7.965551672491264e-06,
+      "loss": 2.5441,
       "step": 16300
     },
     {
-      "epoch": 7.16,
-      "learning_rate": 2.835299257317606e-05,
-      "loss": 2.5766,
+      "epoch": 2.05,
+      "learning_rate": 7.953070394408388e-06,
+      "loss": 2.5431,
       "step": 16400
     },
     {
-      "epoch": 7.21,
-      "learning_rate": 2.7916120576671035e-05,
-      "loss": 2.571,
+      "epoch": 2.06,
+      "learning_rate": 7.940589116325513e-06,
+      "loss": 2.5556,
       "step": 16500
     },
     {
-      "epoch": 7.25,
-      "learning_rate": 2.7479248580166012e-05,
-      "loss": 2.5984,
+      "epoch": 2.07,
+      "learning_rate": 7.928107838242636e-06,
+      "loss": 2.5459,
       "step": 16600
     },
     {
-      "epoch": 7.3,
-      "learning_rate": 2.704237658366099e-05,
-      "loss": 2.613,
+      "epoch": 2.08,
+      "learning_rate": 7.915626560159761e-06,
+      "loss": 2.6096,
       "step": 16700
     },
     {
-      "epoch": 7.34,
-      "learning_rate": 2.6605504587155967e-05,
-      "loss": 2.5634,
+      "epoch": 2.1,
+      "learning_rate": 7.903145282076885e-06,
+      "loss": 2.5705,
       "step": 16800
     },
     {
-      "epoch": 7.38,
-      "learning_rate": 2.6168632590650944e-05,
-      "loss": 2.5756,
+      "epoch": 2.11,
+      "learning_rate": 7.89066400399401e-06,
+      "loss": 2.6404,
       "step": 16900
     },
     {
-      "epoch": 7.43,
-      "learning_rate": 2.5731760594145915e-05,
-      "loss": 2.5711,
+      "epoch": 2.12,
+      "learning_rate": 7.878182725911135e-06,
+      "loss": 2.5122,
       "step": 17000
     },
     {
-      "epoch": 7.47,
-      "learning_rate": 2.5294888597640892e-05,
-      "loss": 2.5701,
+      "epoch": 2.13,
+      "learning_rate": 7.865701447828258e-06,
+      "loss": 2.5761,
       "step": 17100
     },
     {
-      "epoch": 7.51,
-      "learning_rate": 2.485801660113587e-05,
-      "loss": 2.5676,
+      "epoch": 2.15,
+      "learning_rate": 7.853220169745383e-06,
+      "loss": 2.5586,
       "step": 17200
     },
     {
-      "epoch": 7.56,
-      "learning_rate": 2.4421144604630843e-05,
-      "loss": 2.5826,
+      "epoch": 2.16,
+      "learning_rate": 7.840738891662507e-06,
+      "loss": 2.5417,
       "step": 17300
     },
     {
-      "epoch": 7.6,
-      "learning_rate": 2.398427260812582e-05,
-      "loss": 2.5952,
+      "epoch": 2.17,
+      "learning_rate": 7.82825761357963e-06,
+      "loss": 2.5312,
       "step": 17400
     },
     {
-      "epoch": 7.65,
-      "learning_rate": 2.3547400611620795e-05,
-      "loss": 2.5394,
+      "epoch": 2.18,
+      "learning_rate": 7.815776335496755e-06,
+      "loss": 2.6014,
       "step": 17500
     },
     {
-      "epoch": 7.69,
-      "learning_rate": 2.3110528615115772e-05,
-      "loss": 2.5672,
+      "epoch": 2.2,
+      "learning_rate": 7.80329505741388e-06,
+      "loss": 2.5719,
       "step": 17600
     },
     {
-      "epoch": 7.73,
-      "learning_rate": 2.2673656618610746e-05,
-      "loss": 2.5839,
+      "epoch": 2.21,
+      "learning_rate": 7.790813779331005e-06,
+      "loss": 2.6213,
       "step": 17700
     },
     {
-      "epoch": 7.78,
-      "learning_rate": 2.2236784622105723e-05,
-      "loss": 2.5842,
+      "epoch": 2.22,
+      "learning_rate": 7.778332501248129e-06,
+      "loss": 2.5363,
       "step": 17800
     },
     {
-      "epoch": 7.82,
-      "learning_rate": 2.17999126256007e-05,
-      "loss": 2.5702,
+      "epoch": 2.23,
+      "learning_rate": 7.765851223165252e-06,
+      "loss": 2.5023,
       "step": 17900
     },
     {
-      "epoch": 7.86,
-      "learning_rate": 2.1363040629095678e-05,
-      "loss": 2.5695,
+      "epoch": 2.25,
+      "learning_rate": 7.753369945082377e-06,
+      "loss": 2.5229,
       "step": 18000
     },
     {
-      "epoch": 7.91,
-      "learning_rate": 2.092616863259065e-05,
-      "loss": 2.5704,
+      "epoch": 2.26,
+      "learning_rate": 7.7408886669995e-06,
+      "loss": 2.5801,
       "step": 18100
     },
     {
-      "epoch": 7.95,
-      "learning_rate": 2.048929663608563e-05,
-      "loss": 2.5631,
+      "epoch": 2.27,
+      "learning_rate": 7.728407388916626e-06,
+      "loss": 2.5993,
       "step": 18200
     },
     {
-      "epoch": 7.99,
-      "learning_rate": 2.0052424639580603e-05,
-      "loss": 2.5974,
+      "epoch": 2.28,
+      "learning_rate": 7.71592611083375e-06,
+      "loss": 2.5494,
       "step": 18300
     },
     {
-      "epoch": 8.04,
-      "learning_rate": 1.961555264307558e-05,
-      "loss": 2.5626,
+      "epoch": 2.3,
+      "learning_rate": 7.703444832750874e-06,
+      "loss": 2.5426,
       "step": 18400
     },
     {
-      "epoch": 8.08,
-      "learning_rate": 1.9178680646570557e-05,
-      "loss": 2.5612,
+      "epoch": 2.31,
+      "learning_rate": 7.690963554667998e-06,
+      "loss": 2.5575,
       "step": 18500
     },
     {
-      "epoch": 8.13,
-      "learning_rate": 1.874180865006553e-05,
-      "loss": 2.5727,
+      "epoch": 2.32,
+      "learning_rate": 7.678482276585123e-06,
+      "loss": 2.5636,
       "step": 18600
     },
     {
-      "epoch": 8.17,
-      "learning_rate": 1.830493665356051e-05,
-      "loss": 2.5643,
+      "epoch": 2.33,
+      "learning_rate": 7.666000998502248e-06,
+      "loss": 2.6409,
       "step": 18700
     },
     {
-      "epoch": 8.21,
-      "learning_rate": 1.7868064657055482e-05,
-      "loss": 2.5721,
+      "epoch": 2.35,
+      "learning_rate": 7.653519720419371e-06,
+      "loss": 2.6014,
       "step": 18800
     },
     {
-      "epoch": 8.26,
-      "learning_rate": 1.743119266055046e-05,
-      "loss": 2.5551,
+      "epoch": 2.36,
+      "learning_rate": 7.641038442336496e-06,
+      "loss": 2.5964,
       "step": 18900
     },
     {
-      "epoch": 8.3,
-      "learning_rate": 1.6994320664045434e-05,
-      "loss": 2.5685,
+      "epoch": 2.37,
+      "learning_rate": 7.6285571642536205e-06,
+      "loss": 2.5992,
       "step": 19000
     },
     {
-      "epoch": 8.34,
-      "learning_rate": 1.655744866754041e-05,
-      "loss": 2.5237,
+      "epoch": 2.38,
+      "learning_rate": 7.616075886170744e-06,
+      "loss": 2.5852,
       "step": 19100
     },
     {
-      "epoch": 8.39,
-      "learning_rate": 1.6120576671035388e-05,
-      "loss": 2.5657,
+      "epoch": 2.4,
+      "learning_rate": 7.603594608087869e-06,
+      "loss": 2.6166,
       "step": 19200
     },
     {
-      "epoch": 8.43,
-      "learning_rate": 1.5683704674530366e-05,
-      "loss": 2.548,
+      "epoch": 2.41,
+      "learning_rate": 7.591113330004993e-06,
+      "loss": 2.5835,
       "step": 19300
     },
     {
-      "epoch": 8.48,
-      "learning_rate": 1.5246832678025338e-05,
-      "loss": 2.5407,
+      "epoch": 2.42,
+      "learning_rate": 7.578632051922118e-06,
+      "loss": 2.5523,
       "step": 19400
     },
     {
-      "epoch": 8.52,
-      "learning_rate": 1.4809960681520315e-05,
-      "loss": 2.5206,
+      "epoch": 2.43,
+      "learning_rate": 7.566150773839242e-06,
+      "loss": 2.5704,
       "step": 19500
     },
     {
-      "epoch": 8.56,
-      "learning_rate": 1.437308868501529e-05,
-      "loss": 2.5629,
+      "epoch": 2.45,
+      "learning_rate": 7.553669495756366e-06,
+      "loss": 2.6,
       "step": 19600
     },
     {
-      "epoch": 8.61,
-      "learning_rate": 1.3936216688510268e-05,
-      "loss": 2.5647,
+      "epoch": 2.46,
+      "learning_rate": 7.541188217673491e-06,
+      "loss": 2.5654,
       "step": 19700
     },
     {
-      "epoch": 8.65,
-      "learning_rate": 1.3499344692005242e-05,
-      "loss": 2.5473,
+      "epoch": 2.47,
+      "learning_rate": 7.528706939590614e-06,
+      "loss": 2.5707,
       "step": 19800
     },
     {
-      "epoch": 8.69,
-      "learning_rate": 1.3062472695500219e-05,
-      "loss": 2.5505,
+      "epoch": 2.48,
+      "learning_rate": 7.516225661507739e-06,
+      "loss": 2.5366,
       "step": 19900
     },
     {
-      "epoch": 8.74,
-      "learning_rate": 1.2625600698995196e-05,
-      "loss": 2.5513,
+      "epoch": 2.5,
+      "learning_rate": 7.503744383424864e-06,
+      "loss": 2.5468,
       "step": 20000
     },
     {
-      "epoch": 8.78,
-      "learning_rate": 1.218872870249017e-05,
-      "loss": 2.5401,
+      "epoch": 2.51,
+      "learning_rate": 7.491263105341987e-06,
+      "loss": 2.5059,
       "step": 20100
     },
     {
-      "epoch": 8.82,
-      "learning_rate": 1.1751856705985146e-05,
-      "loss": 2.5493,
+      "epoch": 2.52,
+      "learning_rate": 7.478781827259111e-06,
+      "loss": 2.5545,
       "step": 20200
     },
     {
-      "epoch": 8.87,
-      "learning_rate": 1.1314984709480123e-05,
-      "loss": 2.5554,
+      "epoch": 2.53,
+      "learning_rate": 7.466300549176236e-06,
+      "loss": 2.6233,
       "step": 20300
     },
     {
-      "epoch": 8.91,
-      "learning_rate": 1.0878112712975099e-05,
-      "loss": 2.5414,
+      "epoch": 2.55,
+      "learning_rate": 7.453819271093361e-06,
+      "loss": 2.5522,
       "step": 20400
     },
     {
-      "epoch": 8.96,
-      "learning_rate": 1.0441240716470074e-05,
-      "loss": 2.5356,
+      "epoch": 2.56,
+      "learning_rate": 7.441337993010485e-06,
+      "loss": 2.5458,
       "step": 20500
     },
     {
-      "epoch": 9.0,
-      "learning_rate": 1.000436871996505e-05,
-      "loss": 2.5573,
+      "epoch": 2.57,
+      "learning_rate": 7.428856714927609e-06,
+      "loss": 2.5973,
       "step": 20600
     },
     {
-      "epoch": 9.04,
-      "learning_rate": 9.567496723460026e-06,
-      "loss": 2.5378,
+      "epoch": 2.58,
+      "learning_rate": 7.416375436844734e-06,
+      "loss": 2.5794,
       "step": 20700
     },
     {
-      "epoch": 9.09,
-      "learning_rate": 9.130624726955003e-06,
-      "loss": 2.5419,
+      "epoch": 2.6,
+      "learning_rate": 7.403894158761858e-06,
+      "loss": 2.5503,
       "step": 20800
     },
     {
-      "epoch": 9.13,
-      "learning_rate": 8.693752730449979e-06,
-      "loss": 2.5643,
+      "epoch": 2.61,
+      "learning_rate": 7.391412880678982e-06,
+      "loss": 2.5904,
       "step": 20900
     },
     {
-      "epoch": 9.17,
-      "learning_rate": 8.256880733944954e-06,
-      "loss": 2.5396,
+      "epoch": 2.62,
+      "learning_rate": 7.378931602596107e-06,
+      "loss": 2.5643,
       "step": 21000
     },
     {
-      "epoch": 9.22,
-      "learning_rate": 7.820008737439931e-06,
-      "loss": 2.5592,
+      "epoch": 2.63,
+      "learning_rate": 7.366450324513231e-06,
+      "loss": 2.6286,
       "step": 21100
     },
     {
-      "epoch": 9.26,
-      "learning_rate": 7.383136740934907e-06,
-      "loss": 2.5343,
+      "epoch": 2.65,
+      "learning_rate": 7.3539690464303545e-06,
+      "loss": 2.4896,
       "step": 21200
     },
     {
-      "epoch": 9.31,
-      "learning_rate": 6.946264744429883e-06,
-      "loss": 2.5287,
+      "epoch": 2.66,
+      "learning_rate": 7.34148776834748e-06,
+      "loss": 2.5652,
       "step": 21300
     },
     {
-      "epoch": 9.35,
-      "learning_rate": 6.509392747924858e-06,
-      "loss": 2.5486,
+      "epoch": 2.67,
+      "learning_rate": 7.329006490264604e-06,
+      "loss": 2.5344,
       "step": 21400
     },
     {
-      "epoch": 9.39,
-      "learning_rate": 6.072520751419834e-06,
-      "loss": 2.5298,
+      "epoch": 2.68,
+      "learning_rate": 7.316525212181727e-06,
+      "loss": 2.5794,
       "step": 21500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 7.304043934098852e-06,
+      "loss": 2.5678,
+      "step": 21600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 7.2915626560159766e-06,
+      "loss": 2.5789,
+      "step": 21700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 7.279081377933101e-06,
+      "loss": 2.5663,
+      "step": 21800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 7.266600099850225e-06,
+      "loss": 2.5707,
+      "step": 21900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 7.25411882176735e-06,
+      "loss": 2.552,
+      "step": 22000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.241637543684474e-06,
+      "loss": 2.5694,
+      "step": 22100
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.229156265601598e-06,
+      "loss": 2.5828,
+      "step": 22200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.216674987518723e-06,
+      "loss": 2.5308,
+      "step": 22300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.204193709435847e-06,
+      "loss": 2.4647,
+      "step": 22400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.1917124313529705e-06,
+      "loss": 2.5888,
+      "step": 22500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.1792311532700956e-06,
+      "loss": 2.5999,
+      "step": 22600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 7.16674987518722e-06,
+      "loss": 2.6143,
+      "step": 22700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 7.154268597104343e-06,
+      "loss": 2.6138,
+      "step": 22800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 7.141787319021468e-06,
+      "loss": 2.5534,
+      "step": 22900
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 7.1293060409385925e-06,
+      "loss": 2.5856,
+      "step": 23000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.116824762855718e-06,
+      "loss": 2.5619,
+      "step": 23100
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.104343484772841e-06,
+      "loss": 2.5742,
+      "step": 23200
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.091862206689965e-06,
+      "loss": 2.5515,
+      "step": 23300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.07938092860709e-06,
+      "loss": 2.543,
+      "step": 23400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.066899650524214e-06,
+      "loss": 2.5166,
+      "step": 23500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.054418372441339e-06,
+      "loss": 2.5233,
+      "step": 23600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.041937094358463e-06,
+      "loss": 2.571,
+      "step": 23700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.029455816275588e-06,
+      "loss": 2.6225,
+      "step": 23800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.0169745381927115e-06,
+      "loss": 2.5945,
+      "step": 23900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.004493260109836e-06,
+      "loss": 2.5936,
+      "step": 24000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.992011982026961e-06,
+      "loss": 2.6008,
+      "step": 24100
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.979530703944084e-06,
+      "loss": 2.5778,
+      "step": 24200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.9670494258612085e-06,
+      "loss": 2.5153,
+      "step": 24300
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.9545681477783335e-06,
+      "loss": 2.5507,
+      "step": 24400
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.942086869695457e-06,
+      "loss": 2.5331,
+      "step": 24500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.929605591612581e-06,
+      "loss": 2.4988,
+      "step": 24600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.917124313529706e-06,
+      "loss": 2.6262,
+      "step": 24700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.9046430354468305e-06,
+      "loss": 2.5813,
+      "step": 24800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.892161757363955e-06,
+      "loss": 2.475,
+      "step": 24900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.879680479281079e-06,
+      "loss": 2.5655,
+      "step": 25000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.867199201198203e-06,
+      "loss": 2.5762,
+      "step": 25100
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.8547179231153274e-06,
+      "loss": 2.5742,
+      "step": 25200
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.842236645032452e-06,
+      "loss": 2.6108,
+      "step": 25300
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.829755366949577e-06,
+      "loss": 2.576,
+      "step": 25400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.817274088866701e-06,
+      "loss": 2.5012,
+      "step": 25500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.804792810783824e-06,
+      "loss": 2.528,
+      "step": 25600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.7923115327009495e-06,
+      "loss": 2.5362,
+      "step": 25700
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 6.779830254618074e-06,
+      "loss": 2.5012,
+      "step": 25800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 6.767348976535197e-06,
+      "loss": 2.5318,
+      "step": 25900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.754867698452322e-06,
+      "loss": 2.5424,
+      "step": 26000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.7423864203694464e-06,
+      "loss": 2.4849,
+      "step": 26100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.72990514228657e-06,
+      "loss": 2.5104,
+      "step": 26200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.717423864203695e-06,
+      "loss": 2.6342,
+      "step": 26300
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.704942586120819e-06,
+      "loss": 2.5654,
+      "step": 26400
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.692461308037944e-06,
+      "loss": 2.5578,
+      "step": 26500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.679980029955068e-06,
+      "loss": 2.561,
+      "step": 26600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.667498751872193e-06,
+      "loss": 2.5678,
+      "step": 26700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.655017473789317e-06,
+      "loss": 2.5214,
+      "step": 26800
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.64253619570644e-06,
+      "loss": 2.6021,
+      "step": 26900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.6300549176235654e-06,
+      "loss": 2.524,
+      "step": 27000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.61757363954069e-06,
+      "loss": 2.5641,
+      "step": 27100
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.605092361457813e-06,
+      "loss": 2.59,
+      "step": 27200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.592611083374938e-06,
+      "loss": 2.5593,
+      "step": 27300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.580129805292062e-06,
+      "loss": 2.5457,
+      "step": 27400
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.5676485272091875e-06,
+      "loss": 2.5725,
+      "step": 27500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.555167249126311e-06,
+      "loss": 2.5082,
+      "step": 27600
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.542685971043435e-06,
+      "loss": 2.5807,
+      "step": 27700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.53020469296056e-06,
+      "loss": 2.6353,
+      "step": 27800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.5177234148776836e-06,
+      "loss": 2.6223,
+      "step": 27900
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.505242136794808e-06,
+      "loss": 2.5538,
+      "step": 28000
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.492760858711933e-06,
+      "loss": 2.5544,
+      "step": 28100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.480279580629057e-06,
+      "loss": 2.6314,
+      "step": 28200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.467798302546181e-06,
+      "loss": 2.5278,
+      "step": 28300
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.455317024463306e-06,
+      "loss": 2.585,
+      "step": 28400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.442835746380431e-06,
+      "loss": 2.5468,
+      "step": 28500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.430354468297554e-06,
+      "loss": 2.5496,
+      "step": 28600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.417873190214678e-06,
+      "loss": 2.5182,
+      "step": 28700
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 6.405391912131803e-06,
+      "loss": 2.6226,
+      "step": 28800
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 6.392910634048927e-06,
+      "loss": 2.5062,
+      "step": 28900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 6.380429355966051e-06,
+      "loss": 2.5962,
+      "step": 29000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 6.367948077883176e-06,
+      "loss": 2.5113,
+      "step": 29100
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 6.3554667998003e-06,
+      "loss": 2.5236,
+      "step": 29200
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.342985521717424e-06,
+      "loss": 2.4738,
+      "step": 29300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.330504243634549e-06,
+      "loss": 2.5689,
+      "step": 29400
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.318022965551673e-06,
+      "loss": 2.6064,
+      "step": 29500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.305541687468797e-06,
+      "loss": 2.5285,
+      "step": 29600
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.2930604093859216e-06,
+      "loss": 2.6252,
+      "step": 29700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.280579131303046e-06,
+      "loss": 2.5797,
+      "step": 29800
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.26809785322017e-06,
+      "loss": 2.526,
+      "step": 29900
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.255616575137294e-06,
+      "loss": 2.5698,
+      "step": 30000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.243135297054419e-06,
+      "loss": 2.587,
+      "step": 30100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.230654018971544e-06,
+      "loss": 2.5236,
+      "step": 30200
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.218172740888667e-06,
+      "loss": 2.539,
+      "step": 30300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.205691462805792e-06,
+      "loss": 2.5915,
+      "step": 30400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.193210184722916e-06,
+      "loss": 2.5222,
+      "step": 30500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.18072890664004e-06,
+      "loss": 2.5455,
+      "step": 30600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.168247628557165e-06,
+      "loss": 2.5418,
+      "step": 30700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.155766350474289e-06,
+      "loss": 2.5469,
+      "step": 30800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.143285072391414e-06,
+      "loss": 2.5557,
+      "step": 30900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.1308037943085375e-06,
+      "loss": 2.534,
+      "step": 31000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.118322516225662e-06,
+      "loss": 2.5967,
+      "step": 31100
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.105841238142787e-06,
+      "loss": 2.493,
+      "step": 31200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.09335996005991e-06,
+      "loss": 2.4888,
+      "step": 31300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 6.080878681977035e-06,
+      "loss": 2.4598,
+      "step": 31400
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 6.0683974038941595e-06,
+      "loss": 2.5489,
+      "step": 31500
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 6.055916125811283e-06,
+      "loss": 2.6207,
+      "step": 31600
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.043434847728408e-06,
+      "loss": 2.4955,
+      "step": 31700
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 6.030953569645532e-06,
+      "loss": 2.5457,
+      "step": 31800
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 6.018472291562657e-06,
+      "loss": 2.5991,
+      "step": 31900
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.005991013479781e-06,
+      "loss": 2.6215,
+      "step": 32000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.993509735396905e-06,
+      "loss": 2.483,
+      "step": 32100
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.98102845731403e-06,
+      "loss": 2.554,
+      "step": 32200
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.9685471792311534e-06,
+      "loss": 2.5586,
+      "step": 32300
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.956065901148278e-06,
+      "loss": 2.5568,
+      "step": 32400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.943584623065403e-06,
+      "loss": 2.5593,
+      "step": 32500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 5.931103344982526e-06,
+      "loss": 2.5854,
+      "step": 32600
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 5.91862206689965e-06,
+      "loss": 2.5227,
+      "step": 32700
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 5.9061407888167755e-06,
+      "loss": 2.5099,
+      "step": 32800
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 5.8936595107339e-06,
+      "loss": 2.4718,
+      "step": 32900
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 5.881178232651024e-06,
+      "loss": 2.5361,
+      "step": 33000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 5.868696954568148e-06,
+      "loss": 2.5505,
+      "step": 33100
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 5.856215676485273e-06,
+      "loss": 2.5501,
+      "step": 33200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 5.843734398402397e-06,
+      "loss": 2.5911,
+      "step": 33300
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 5.831253120319521e-06,
+      "loss": 2.5891,
+      "step": 33400
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 5.818771842236646e-06,
+      "loss": 2.5229,
+      "step": 33500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 5.80629056415377e-06,
+      "loss": 2.518,
+      "step": 33600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 5.793809286070894e-06,
+      "loss": 2.5585,
+      "step": 33700
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.781328007988019e-06,
+      "loss": 2.5518,
+      "step": 33800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.768846729905143e-06,
+      "loss": 2.5389,
+      "step": 33900
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.756365451822266e-06,
+      "loss": 2.5181,
+      "step": 34000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.743884173739391e-06,
+      "loss": 2.4888,
+      "step": 34100
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.731402895656516e-06,
+      "loss": 2.5526,
+      "step": 34200
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.71892161757364e-06,
+      "loss": 2.5189,
+      "step": 34300
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.706440339490764e-06,
+      "loss": 2.4954,
+      "step": 34400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.693959061407888e-06,
+      "loss": 2.5498,
+      "step": 34500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.6814777833250135e-06,
+      "loss": 2.5855,
+      "step": 34600
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.668996505242137e-06,
+      "loss": 2.5233,
+      "step": 34700
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.656515227159262e-06,
+      "loss": 2.4986,
+      "step": 34800
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.644033949076386e-06,
+      "loss": 2.4563,
+      "step": 34900
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 5.6315526709935096e-06,
+      "loss": 2.5183,
+      "step": 35000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.619071392910635e-06,
+      "loss": 2.5314,
+      "step": 35100
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.606590114827759e-06,
+      "loss": 2.5084,
+      "step": 35200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.594108836744882e-06,
+      "loss": 2.5564,
+      "step": 35300
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.581627558662007e-06,
+      "loss": 2.4687,
+      "step": 35400
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 5.569146280579132e-06,
+      "loss": 2.5122,
+      "step": 35500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.556665002496257e-06,
+      "loss": 2.5395,
+      "step": 35600
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.54418372441338e-06,
+      "loss": 2.6325,
+      "step": 35700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.531702446330504e-06,
+      "loss": 2.5592,
+      "step": 35800
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.519221168247629e-06,
+      "loss": 2.5724,
+      "step": 35900
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.506739890164753e-06,
+      "loss": 2.5779,
+      "step": 36000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.494258612081878e-06,
+      "loss": 2.556,
+      "step": 36100
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 5.481777333999002e-06,
+      "loss": 2.4931,
+      "step": 36200
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 5.469296055916127e-06,
+      "loss": 2.5409,
+      "step": 36300
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 5.456814777833251e-06,
+      "loss": 2.5513,
+      "step": 36400
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.444333499750375e-06,
+      "loss": 2.5558,
+      "step": 36500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.4318522216675e-06,
+      "loss": 2.5406,
+      "step": 36600
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.419370943584623e-06,
+      "loss": 2.5845,
+      "step": 36700
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.4068896655017475e-06,
+      "loss": 2.5388,
+      "step": 36800
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 5.394408387418873e-06,
+      "loss": 2.5498,
+      "step": 36900
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 5.381927109335996e-06,
+      "loss": 2.4883,
+      "step": 37000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 5.36944583125312e-06,
+      "loss": 2.5466,
+      "step": 37100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 5.356964553170245e-06,
+      "loss": 2.5449,
+      "step": 37200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.3444832750873696e-06,
+      "loss": 2.5573,
+      "step": 37300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.332001997004494e-06,
+      "loss": 2.5178,
+      "step": 37400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.319520718921618e-06,
+      "loss": 2.5307,
+      "step": 37500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.307039440838742e-06,
+      "loss": 2.5952,
+      "step": 37600
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 5.2945581627558665e-06,
+      "loss": 2.6333,
+      "step": 37700
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 5.282076884672991e-06,
+      "loss": 2.5498,
+      "step": 37800
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 5.269595606590116e-06,
+      "loss": 2.5644,
+      "step": 37900
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 5.257114328507239e-06,
+      "loss": 2.5154,
+      "step": 38000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.2446330504243635e-06,
+      "loss": 2.5258,
+      "step": 38100
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.2321517723414886e-06,
+      "loss": 2.6026,
+      "step": 38200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.219670494258613e-06,
+      "loss": 2.5448,
+      "step": 38300
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 5.207189216175736e-06,
+      "loss": 2.5479,
+      "step": 38400
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 5.194707938092861e-06,
+      "loss": 2.6038,
+      "step": 38500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 5.1822266600099855e-06,
+      "loss": 2.5809,
+      "step": 38600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.169745381927109e-06,
+      "loss": 2.5219,
+      "step": 38700
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.157264103844234e-06,
+      "loss": 2.5623,
+      "step": 38800
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.144782825761358e-06,
+      "loss": 2.5276,
+      "step": 38900
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.132301547678483e-06,
+      "loss": 2.5995,
+      "step": 39000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.119820269595607e-06,
+      "loss": 2.5641,
+      "step": 39100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.107338991512732e-06,
+      "loss": 2.5158,
+      "step": 39200
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 5.094857713429856e-06,
+      "loss": 2.4968,
+      "step": 39300
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 5.082376435346979e-06,
+      "loss": 2.5531,
+      "step": 39400
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.0698951572641045e-06,
+      "loss": 2.5547,
+      "step": 39500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.057413879181229e-06,
+      "loss": 2.5746,
+      "step": 39600
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.044932601098352e-06,
+      "loss": 2.461,
+      "step": 39700
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.032451323015477e-06,
+      "loss": 2.5375,
+      "step": 39800
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.0199700449326015e-06,
+      "loss": 2.5461,
+      "step": 39900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.0074887668497265e-06,
+      "loss": 2.5562,
+      "step": 40000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.995007488766851e-06,
+      "loss": 2.5117,
+      "step": 40100
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 4.982526210683974e-06,
+      "loss": 2.5144,
+      "step": 40200
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.970044932601098e-06,
+      "loss": 2.4917,
+      "step": 40300
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 4.9575636545182235e-06,
+      "loss": 2.5242,
+      "step": 40400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.945082376435347e-06,
+      "loss": 2.5614,
+      "step": 40500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 4.932601098352472e-06,
+      "loss": 2.5885,
+      "step": 40600
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.920119820269596e-06,
+      "loss": 2.5738,
+      "step": 40700
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 4.9076385421867204e-06,
+      "loss": 2.5221,
+      "step": 40800
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.895157264103845e-06,
+      "loss": 2.579,
+      "step": 40900
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.882675986020969e-06,
+      "loss": 2.5234,
+      "step": 41000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 4.870194707938093e-06,
+      "loss": 2.5614,
+      "step": 41100
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.857713429855217e-06,
+      "loss": 2.5128,
+      "step": 41200
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.845232151772342e-06,
+      "loss": 2.4828,
+      "step": 41300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.832750873689466e-06,
+      "loss": 2.5557,
+      "step": 41400
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 4.82026959560659e-06,
+      "loss": 2.556,
+      "step": 41500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.807788317523715e-06,
+      "loss": 2.5401,
+      "step": 41600
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 4.7953070394408394e-06,
+      "loss": 2.501,
+      "step": 41700
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 4.782825761357963e-06,
+      "loss": 2.4316,
+      "step": 41800
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.770344483275088e-06,
+      "loss": 2.5133,
+      "step": 41900
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.757863205192212e-06,
+      "loss": 2.574,
+      "step": 42000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 4.745381927109336e-06,
+      "loss": 2.6097,
+      "step": 42100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 4.732900649026461e-06,
+      "loss": 2.5314,
+      "step": 42200
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.720419370943585e-06,
+      "loss": 2.5095,
+      "step": 42300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 4.707938092860709e-06,
+      "loss": 2.5508,
+      "step": 42400
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.695456814777833e-06,
+      "loss": 2.4751,
+      "step": 42500
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.6829755366949584e-06,
+      "loss": 2.5443,
+      "step": 42600
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 4.670494258612082e-06,
+      "loss": 2.5164,
+      "step": 42700
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.658012980529207e-06,
+      "loss": 2.5643,
+      "step": 42800
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.645531702446331e-06,
+      "loss": 2.5964,
+      "step": 42900
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.633050424363455e-06,
+      "loss": 2.5625,
+      "step": 43000
     }
   ],
-  "max_steps": 22890,
+  "max_steps": 80120,
   "num_train_epochs": 10,
-  "total_flos": 6.59804997251202e+17,
+  "total_flos": 3.2295236369381376e+17,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:92aaf2c6d5b51dbf0b8baf747263cd46a12b554eee631ce3143468a81e1cc8ee
+oid sha256:737e26502c6090960eb553ee6bb362b7c43cd9705f8a721737f7bf7282910dcd
 size 2607