matthewpwilson-ibm commited on
Commit
d9bfa88
·
1 Parent(s): 40eefe0

Initial commit

Browse files
README.md CHANGED
@@ -1,3 +1,20 @@
1
  ---
2
  license: apache-2.0
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
+ library_name: transformers
4
+ pipeline_tag: feature-extraction
5
+ tags:
6
+ - chemistry
7
+ - transformers
8
  ---
9
+
10
+ # selfies-ted2m
11
+
12
+ selfies-ted is an transformer based encoder decoder model for molecular representations using SELFIES. This is a 2.2M parameter version of the model. For the full-sized version and more informatin on architecture, see [selfies-ted](https://huggingface.co/ibm-research/materials.selfies-ted).
13
+
14
+ This version also includes a projection layer to convert the last hidden state of the BART model (256-dimensional vector per token) to a single 128-dimension vector for the whole SELFIES sequence.
15
+
16
+
17
+
18
+
19
+
20
+
config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/dccstor/ipd/small-bart/test_trainer/checkpoint-50050",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "adapters": {
6
+ "adapters": {},
7
+ "config_map": {},
8
+ "fusion_config_map": {},
9
+ "fusions": {}
10
+ },
11
+ "architectures": [
12
+ "BartForConditionalGeneration"
13
+ ],
14
+ "attention_dropout": 0.0,
15
+ "bos_token_id": 1,
16
+ "classifier_dropout": 0.0,
17
+ "d_model": 256,
18
+ "decoder_attention_heads": 4,
19
+ "decoder_ffn_dim": 256,
20
+ "decoder_layerdrop": 0.0,
21
+ "decoder_layers": 2,
22
+ "decoder_start_token_id": 2,
23
+ "dmodel": 256,
24
+ "dropout": 0.1,
25
+ "encoder_attention_heads": 4,
26
+ "encoder_ffn_dim": 256,
27
+ "encoder_layerdrop": 0.0,
28
+ "encoder_layers": 2,
29
+ "eos_token_id": 2,
30
+ "forced_eos_token_id": 2,
31
+ "id2label": {
32
+ "0": "LABEL_0",
33
+ "1": "LABEL_1",
34
+ "2": "LABEL_2"
35
+ },
36
+ "init_std": 0.02,
37
+ "is_encoder_decoder": true,
38
+ "label2id": {
39
+ "LABEL_0": 0,
40
+ "LABEL_1": 1,
41
+ "LABEL_2": 2
42
+ },
43
+ "max_position_embeddings": 128,
44
+ "model_type": "bart",
45
+ "num_hidden_layers": 6,
46
+ "pad_token_id": 3,
47
+ "scale_embedding": false,
48
+ "torch_dtype": "float32",
49
+ "transformers_version": "4.21.3",
50
+ "use_cache": true,
51
+ "vocab_size": 173
52
+ }
invprojector_2023-05-25.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc61f71e93f6990f753d3b58275cde1cb7e3f593f49657dfb3a9e05efecc0a04
3
+ size 35988864
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:194a00dd868d2861c1ba80e94cdf387b07d5580369addce87e36e9f199c32b77
3
+ size 17831061
projector_2023-05-25.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0939886f08074af8b8a723efc308e9e24d4252ae388b71cf729fbdfaacfd8882
3
+ size 35923740
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c56beb1afc07c44598564b3ff802d63bb2c3d364c068b0f27b126ff9f9df62bc
3
+ size 8920081
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b9b9179f51efc8caa7a2c631cffc6bb347b285863b3d50ff6bfc6101bad2c0c
3
+ size 17499
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e12f8d8befe78fabb83cfbf6b51ed6e72b79737572a3699bf9d838ed33a0d84
3
+ size 623
special_tokens_map.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "mask_token": "[MASK]",
4
+ "pad_token": "[PAD]",
5
+ "sep_token": "[SEP]",
6
+ "unk_token": "[UNK]"
7
+ }
tokenizer.json ADDED
@@ -0,0 +1,293 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 0,
8
+ "content": "[UNK]",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 1,
17
+ "content": "[CLS]",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": false,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 2,
26
+ "content": "[SEP]",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": false,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 3,
35
+ "content": "[PAD]",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": false,
40
+ "special": true
41
+ },
42
+ {
43
+ "id": 4,
44
+ "content": "[MASK]",
45
+ "single_word": false,
46
+ "lstrip": false,
47
+ "rstrip": false,
48
+ "normalized": false,
49
+ "special": true
50
+ }
51
+ ],
52
+ "normalizer": null,
53
+ "pre_tokenizer": {
54
+ "type": "WhitespaceSplit"
55
+ },
56
+ "post_processor": {
57
+ "type": "TemplateProcessing",
58
+ "single": [
59
+ {
60
+ "SpecialToken": {
61
+ "id": "[CLS]",
62
+ "type_id": 0
63
+ }
64
+ },
65
+ {
66
+ "Sequence": {
67
+ "id": "A",
68
+ "type_id": 0
69
+ }
70
+ },
71
+ {
72
+ "SpecialToken": {
73
+ "id": "[SEP]",
74
+ "type_id": 0
75
+ }
76
+ }
77
+ ],
78
+ "pair": [
79
+ {
80
+ "Sequence": {
81
+ "id": "A",
82
+ "type_id": 0
83
+ }
84
+ },
85
+ {
86
+ "Sequence": {
87
+ "id": "B",
88
+ "type_id": 1
89
+ }
90
+ }
91
+ ],
92
+ "special_tokens": {
93
+ "[CLS]": {
94
+ "id": "[CLS]",
95
+ "ids": [
96
+ 1
97
+ ],
98
+ "tokens": [
99
+ "[CLS]"
100
+ ]
101
+ },
102
+ "[SEP]": {
103
+ "id": "[SEP]",
104
+ "ids": [
105
+ 2
106
+ ],
107
+ "tokens": [
108
+ "[SEP]"
109
+ ]
110
+ }
111
+ }
112
+ },
113
+ "decoder": null,
114
+ "model": {
115
+ "type": "WordLevel",
116
+ "vocab": {
117
+ "[UNK]": 0,
118
+ "[CLS]": 1,
119
+ "[SEP]": 2,
120
+ "[PAD]": 3,
121
+ "[MASK]": 4,
122
+ "[17O]": 5,
123
+ "[/S@@]": 6,
124
+ "[/N]": 7,
125
+ "[B@-1]": 8,
126
+ "[18OH1]": 9,
127
+ "[-/Ring2]": 10,
128
+ "[=17O]": 11,
129
+ "[/Br]": 12,
130
+ "[C@]": 13,
131
+ "[N@]": 14,
132
+ "[Branch1]": 15,
133
+ "[=Si]": 16,
134
+ "[=Branch2]": 17,
135
+ "[Cl].[Cl-1]": 18,
136
+ "[/C]": 19,
137
+ "[S].[Cl-1]": 20,
138
+ "[SH1]": 21,
139
+ "[Sn@]": 22,
140
+ "[#S]": 23,
141
+ "[NH1+1]": 24,
142
+ "[#Branch1]": 25,
143
+ "[#Branch1].[Cl-1]": 26,
144
+ "[BH2-1]": 27,
145
+ "[/NH1+1]": 28,
146
+ "[=P@]": 29,
147
+ "[I]": 30,
148
+ "[/S+1]": 31,
149
+ "[S-1]": 32,
150
+ "[=S@+1]": 33,
151
+ "[NH0]": 34,
152
+ "[SnH1]": 35,
153
+ "[F]": 36,
154
+ "[BH1-1]": 37,
155
+ "[=Ring2]": 38,
156
+ "[CH1]": 39,
157
+ "[=Ring1]": 40,
158
+ "[/S@]": 41,
159
+ "[N@+1]": 42,
160
+ "[Sn+2]": 43,
161
+ "[Sn+3]": 44,
162
+ "[#N+1]": 45,
163
+ "[/F]": 46,
164
+ "[Si]": 47,
165
+ "[/O]": 48,
166
+ "[Sn]": 49,
167
+ "[P]": 50,
168
+ "[Ring2]": 51,
169
+ "[#Branch2].[Cl-1]": 52,
170
+ "[=P@@]": 53,
171
+ "[PH1+1]": 54,
172
+ "[/Cl]": 55,
173
+ "[C@@]": 56,
174
+ "[Sn@@]": 57,
175
+ "[=S@]": 58,
176
+ "[=S]": 59,
177
+ "[C@H1]": 60,
178
+ "[Ring2].[Cl-1]": 61,
179
+ "[O-1].[Cl-1]": 62,
180
+ "[N-1]": 63,
181
+ "[S@@H1]": 64,
182
+ "[/O-1]": 65,
183
+ "[/NH1]": 66,
184
+ "[/C@@H1]": 67,
185
+ "[B-1]": 68,
186
+ "[OH0]": 69,
187
+ "[Ring1]": 70,
188
+ "[=O+1]": 71,
189
+ "[O+1]": 72,
190
+ "[=S@@+1]": 73,
191
+ "[=S+1]": 74,
192
+ "[/I]": 75,
193
+ "[Si@@]": 76,
194
+ "[=C]": 77,
195
+ "[P+1]": 78,
196
+ "[I].[Cl-1]": 79,
197
+ "[Branch1].[Cl-1]": 80,
198
+ "[P@@]": 81,
199
+ "[Branch1].[C]": 82,
200
+ "[Br+1]": 83,
201
+ "[=B]": 84,
202
+ "[CH1-1]": 85,
203
+ "[Sn+1]": 86,
204
+ "[Si-1]": 87,
205
+ "[/N+1]": 88,
206
+ "[N+1]": 89,
207
+ "[CH2]": 90,
208
+ "[SiH2]": 91,
209
+ "[/Sn]": 92,
210
+ "[=PH1]": 93,
211
+ "[=P+1]": 94,
212
+ "[S@@]": 95,
213
+ "[/C@H1]": 96,
214
+ "[N@@+1]": 97,
215
+ "[=S@@]": 98,
216
+ "[/NH2+1]": 99,
217
+ "[=N+1]": 100,
218
+ "[=C].[Cl-1]": 101,
219
+ "[=NH2+1]": 102,
220
+ "[=SH1]": 103,
221
+ "[/Si]": 104,
222
+ "[#N]": 105,
223
+ "[=N]": 106,
224
+ "[/OH0]": 107,
225
+ "[S]": 108,
226
+ "[Branch2].[Cl-1]": 109,
227
+ "[N@H1+1]": 110,
228
+ "[NH2+1]": 111,
229
+ "[Br]": 112,
230
+ "[O-1]": 113,
231
+ "[=Branch1]": 114,
232
+ "[CH0]": 115,
233
+ "[=Se]": 116,
234
+ "[#Branch2]": 117,
235
+ "[#C]": 118,
236
+ "[=Branch1].[Cl-1]": 119,
237
+ "[S@]": 120,
238
+ "[SiH1]": 121,
239
+ "[SiH3]": 122,
240
+ "[#C].[Cl-1]": 123,
241
+ "[#C-1]": 124,
242
+ "[S+1]": 125,
243
+ "[Cl]": 126,
244
+ "[#N].[Cl-1]": 127,
245
+ "[/C@@]": 128,
246
+ "[B@@-1]": 129,
247
+ "[F].[Cl-1]": 130,
248
+ "[Br].[Cl-1]": 131,
249
+ "[=Branch2].[Cl-1]": 132,
250
+ "[=P]": 133,
251
+ "[C].[Cl-1]": 134,
252
+ "[/B]": 135,
253
+ "[PH1]": 136,
254
+ "[C-1]": 137,
255
+ "[Si@@H1]": 138,
256
+ "[/P]": 139,
257
+ "[=O].[Cl-1]": 140,
258
+ "[O].[Cl-1]": 141,
259
+ "[N@@]": 142,
260
+ "[P].[Cl-1]": 143,
261
+ "[S@+1]": 144,
262
+ "[Branch2]": 145,
263
+ "[I+1]": 146,
264
+ "[-/Ring1]": 147,
265
+ "[/S]": 148,
266
+ "[P@]": 149,
267
+ "[P@@+1]": 150,
268
+ "[=Sn]": 151,
269
+ "[NH3+1]": 152,
270
+ "[=N-1]": 153,
271
+ "[=NH1+1]": 154,
272
+ "[NH1]": 155,
273
+ "[Si@]": 156,
274
+ "[N@@H1+1]": 157,
275
+ "[N]": 158,
276
+ "[Ring1].[Cl-1]": 159,
277
+ "[BH3-1]": 160,
278
+ "[B]": 161,
279
+ "[/C@]": 162,
280
+ "[C]": 163,
281
+ "[Si@H1]": 164,
282
+ "[O]": 165,
283
+ "[=NH0]": 166,
284
+ "[P@+1]": 167,
285
+ "[=N].[Cl-1]": 168,
286
+ "[/N-1]": 169,
287
+ "[S@@+1]": 170,
288
+ "[=O]": 171,
289
+ "[C@@H1]": 172
290
+ },
291
+ "unk_token": "[UNK]"
292
+ }
293
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": "[CLS]",
3
+ "mask_token": "[MASK]",
4
+ "model_max_length": 512,
5
+ "name_or_path": "zinc22-16B-tokenizer",
6
+ "pad_token": "[PAD]",
7
+ "sep_token": "[SEP]",
8
+ "special_tokens_map_file": "/u/indra/.cache/huggingface/transformers/410b9713b9ff338491c4439bb9c3426a17c79e66c5a02f0a729b5455a334037a.7da70648c6cb9951e284c9685f9ba7ae083dd59ed1d6d84bdfc0584a4ea94b6d",
9
+ "tokenizer_class": "PreTrainedTokenizerFast",
10
+ "unk_token": "[UNK]"
11
+ }
trainer_state.json ADDED
@@ -0,0 +1,3100 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "global_step": 51465,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 4.990284659477315e-05,
13
+ "loss": 0.0,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 0.0,
18
+ "learning_rate": 4.98056931895463e-05,
19
+ "loss": 0.0,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 4.970853978431944e-05,
25
+ "loss": 0.0,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 4.961138637909259e-05,
31
+ "loss": 0.0,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 4.951423297386574e-05,
37
+ "loss": 0.0,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 4.941707956863889e-05,
43
+ "loss": 0.0,
44
+ "step": 600
45
+ },
46
+ {
47
+ "epoch": 0.01,
48
+ "learning_rate": 4.931992616341203e-05,
49
+ "loss": 0.0,
50
+ "step": 700
51
+ },
52
+ {
53
+ "epoch": 0.02,
54
+ "learning_rate": 4.922277275818518e-05,
55
+ "loss": 0.0,
56
+ "step": 800
57
+ },
58
+ {
59
+ "epoch": 0.02,
60
+ "learning_rate": 4.9125619352958326e-05,
61
+ "loss": 0.0,
62
+ "step": 900
63
+ },
64
+ {
65
+ "epoch": 0.02,
66
+ "learning_rate": 4.9028465947731474e-05,
67
+ "loss": 0.0,
68
+ "step": 1000
69
+ },
70
+ {
71
+ "epoch": 0.02,
72
+ "learning_rate": 4.8931312542504616e-05,
73
+ "loss": 0.0,
74
+ "step": 1100
75
+ },
76
+ {
77
+ "epoch": 0.02,
78
+ "learning_rate": 4.8834159137277764e-05,
79
+ "loss": 0.0,
80
+ "step": 1200
81
+ },
82
+ {
83
+ "epoch": 0.03,
84
+ "learning_rate": 4.873700573205091e-05,
85
+ "loss": 0.0,
86
+ "step": 1300
87
+ },
88
+ {
89
+ "epoch": 0.03,
90
+ "learning_rate": 4.863985232682406e-05,
91
+ "loss": 0.0,
92
+ "step": 1400
93
+ },
94
+ {
95
+ "epoch": 0.03,
96
+ "learning_rate": 4.85426989215972e-05,
97
+ "loss": 0.0,
98
+ "step": 1500
99
+ },
100
+ {
101
+ "epoch": 0.03,
102
+ "learning_rate": 4.844554551637035e-05,
103
+ "loss": 0.0,
104
+ "step": 1600
105
+ },
106
+ {
107
+ "epoch": 0.03,
108
+ "learning_rate": 4.83483921111435e-05,
109
+ "loss": 0.0,
110
+ "step": 1700
111
+ },
112
+ {
113
+ "epoch": 0.03,
114
+ "learning_rate": 4.825123870591665e-05,
115
+ "loss": 0.0,
116
+ "step": 1800
117
+ },
118
+ {
119
+ "epoch": 0.04,
120
+ "learning_rate": 4.81540853006898e-05,
121
+ "loss": 0.0,
122
+ "step": 1900
123
+ },
124
+ {
125
+ "epoch": 0.04,
126
+ "learning_rate": 4.805693189546294e-05,
127
+ "loss": 0.0,
128
+ "step": 2000
129
+ },
130
+ {
131
+ "epoch": 0.04,
132
+ "learning_rate": 4.795977849023609e-05,
133
+ "loss": 0.0,
134
+ "step": 2100
135
+ },
136
+ {
137
+ "epoch": 0.04,
138
+ "learning_rate": 4.7862625085009236e-05,
139
+ "loss": 0.0,
140
+ "step": 2200
141
+ },
142
+ {
143
+ "epoch": 0.04,
144
+ "learning_rate": 4.776547167978238e-05,
145
+ "loss": 0.0,
146
+ "step": 2300
147
+ },
148
+ {
149
+ "epoch": 0.05,
150
+ "learning_rate": 4.7668318274555526e-05,
151
+ "loss": 0.0,
152
+ "step": 2400
153
+ },
154
+ {
155
+ "epoch": 0.05,
156
+ "learning_rate": 4.7571164869328675e-05,
157
+ "loss": 0.0,
158
+ "step": 2500
159
+ },
160
+ {
161
+ "epoch": 0.05,
162
+ "learning_rate": 4.747401146410182e-05,
163
+ "loss": 0.0,
164
+ "step": 2600
165
+ },
166
+ {
167
+ "epoch": 0.05,
168
+ "learning_rate": 4.7376858058874965e-05,
169
+ "loss": 0.0,
170
+ "step": 2700
171
+ },
172
+ {
173
+ "epoch": 0.05,
174
+ "learning_rate": 4.727970465364811e-05,
175
+ "loss": 0.0,
176
+ "step": 2800
177
+ },
178
+ {
179
+ "epoch": 0.06,
180
+ "learning_rate": 4.7182551248421255e-05,
181
+ "loss": 0.0,
182
+ "step": 2900
183
+ },
184
+ {
185
+ "epoch": 0.06,
186
+ "learning_rate": 4.7085397843194404e-05,
187
+ "loss": 0.0,
188
+ "step": 3000
189
+ },
190
+ {
191
+ "epoch": 0.06,
192
+ "learning_rate": 4.698824443796755e-05,
193
+ "loss": 0.0,
194
+ "step": 3100
195
+ },
196
+ {
197
+ "epoch": 0.06,
198
+ "learning_rate": 4.6891091032740694e-05,
199
+ "loss": 0.0,
200
+ "step": 3200
201
+ },
202
+ {
203
+ "epoch": 0.06,
204
+ "learning_rate": 4.679393762751384e-05,
205
+ "loss": 0.0,
206
+ "step": 3300
207
+ },
208
+ {
209
+ "epoch": 0.07,
210
+ "learning_rate": 4.669678422228699e-05,
211
+ "loss": 0.0,
212
+ "step": 3400
213
+ },
214
+ {
215
+ "epoch": 0.07,
216
+ "learning_rate": 4.659963081706014e-05,
217
+ "loss": 0.0,
218
+ "step": 3500
219
+ },
220
+ {
221
+ "epoch": 0.07,
222
+ "learning_rate": 4.650247741183329e-05,
223
+ "loss": 0.0,
224
+ "step": 3600
225
+ },
226
+ {
227
+ "epoch": 0.07,
228
+ "learning_rate": 4.640532400660643e-05,
229
+ "loss": 0.0,
230
+ "step": 3700
231
+ },
232
+ {
233
+ "epoch": 0.07,
234
+ "learning_rate": 4.630817060137958e-05,
235
+ "loss": 0.0,
236
+ "step": 3800
237
+ },
238
+ {
239
+ "epoch": 0.08,
240
+ "learning_rate": 4.621101719615273e-05,
241
+ "loss": 0.0,
242
+ "step": 3900
243
+ },
244
+ {
245
+ "epoch": 0.08,
246
+ "learning_rate": 4.6113863790925876e-05,
247
+ "loss": 0.0,
248
+ "step": 4000
249
+ },
250
+ {
251
+ "epoch": 0.08,
252
+ "learning_rate": 4.601671038569902e-05,
253
+ "loss": 0.0,
254
+ "step": 4100
255
+ },
256
+ {
257
+ "epoch": 0.08,
258
+ "learning_rate": 4.5919556980472166e-05,
259
+ "loss": 0.0,
260
+ "step": 4200
261
+ },
262
+ {
263
+ "epoch": 0.08,
264
+ "learning_rate": 4.5822403575245314e-05,
265
+ "loss": 0.0,
266
+ "step": 4300
267
+ },
268
+ {
269
+ "epoch": 0.09,
270
+ "learning_rate": 4.572525017001846e-05,
271
+ "loss": 0.0,
272
+ "step": 4400
273
+ },
274
+ {
275
+ "epoch": 0.09,
276
+ "learning_rate": 4.5628096764791605e-05,
277
+ "loss": 0.0,
278
+ "step": 4500
279
+ },
280
+ {
281
+ "epoch": 0.09,
282
+ "learning_rate": 4.553094335956475e-05,
283
+ "loss": 0.0,
284
+ "step": 4600
285
+ },
286
+ {
287
+ "epoch": 0.09,
288
+ "learning_rate": 4.54337899543379e-05,
289
+ "loss": 0.0,
290
+ "step": 4700
291
+ },
292
+ {
293
+ "epoch": 0.09,
294
+ "learning_rate": 4.533663654911105e-05,
295
+ "loss": 0.0,
296
+ "step": 4800
297
+ },
298
+ {
299
+ "epoch": 0.1,
300
+ "learning_rate": 4.523948314388419e-05,
301
+ "loss": 0.0,
302
+ "step": 4900
303
+ },
304
+ {
305
+ "epoch": 0.1,
306
+ "learning_rate": 4.514232973865734e-05,
307
+ "loss": 0.0,
308
+ "step": 5000
309
+ },
310
+ {
311
+ "epoch": 0.1,
312
+ "learning_rate": 4.504517633343049e-05,
313
+ "loss": 0.0,
314
+ "step": 5100
315
+ },
316
+ {
317
+ "epoch": 0.1,
318
+ "learning_rate": 4.494802292820364e-05,
319
+ "loss": 0.0,
320
+ "step": 5200
321
+ },
322
+ {
323
+ "epoch": 0.1,
324
+ "learning_rate": 4.4850869522976786e-05,
325
+ "loss": 0.0,
326
+ "step": 5300
327
+ },
328
+ {
329
+ "epoch": 0.1,
330
+ "learning_rate": 4.475371611774993e-05,
331
+ "loss": 0.0,
332
+ "step": 5400
333
+ },
334
+ {
335
+ "epoch": 0.11,
336
+ "learning_rate": 4.4656562712523076e-05,
337
+ "loss": 0.0,
338
+ "step": 5500
339
+ },
340
+ {
341
+ "epoch": 0.11,
342
+ "learning_rate": 4.4559409307296225e-05,
343
+ "loss": 0.0,
344
+ "step": 5600
345
+ },
346
+ {
347
+ "epoch": 0.11,
348
+ "learning_rate": 4.4462255902069374e-05,
349
+ "loss": 0.0,
350
+ "step": 5700
351
+ },
352
+ {
353
+ "epoch": 0.11,
354
+ "learning_rate": 4.4365102496842515e-05,
355
+ "loss": 0.0,
356
+ "step": 5800
357
+ },
358
+ {
359
+ "epoch": 0.11,
360
+ "learning_rate": 4.4267949091615664e-05,
361
+ "loss": 0.0,
362
+ "step": 5900
363
+ },
364
+ {
365
+ "epoch": 0.12,
366
+ "learning_rate": 4.417079568638881e-05,
367
+ "loss": 0.0,
368
+ "step": 6000
369
+ },
370
+ {
371
+ "epoch": 0.12,
372
+ "learning_rate": 4.407364228116196e-05,
373
+ "loss": 0.0,
374
+ "step": 6100
375
+ },
376
+ {
377
+ "epoch": 0.12,
378
+ "learning_rate": 4.39764888759351e-05,
379
+ "loss": 0.0,
380
+ "step": 6200
381
+ },
382
+ {
383
+ "epoch": 0.12,
384
+ "learning_rate": 4.387933547070825e-05,
385
+ "loss": 0.0,
386
+ "step": 6300
387
+ },
388
+ {
389
+ "epoch": 0.12,
390
+ "learning_rate": 4.37821820654814e-05,
391
+ "loss": 0.0,
392
+ "step": 6400
393
+ },
394
+ {
395
+ "epoch": 0.13,
396
+ "learning_rate": 4.368502866025455e-05,
397
+ "loss": 0.0,
398
+ "step": 6500
399
+ },
400
+ {
401
+ "epoch": 0.13,
402
+ "learning_rate": 4.358787525502769e-05,
403
+ "loss": 0.0,
404
+ "step": 6600
405
+ },
406
+ {
407
+ "epoch": 0.13,
408
+ "learning_rate": 4.349072184980084e-05,
409
+ "loss": 0.0,
410
+ "step": 6700
411
+ },
412
+ {
413
+ "epoch": 0.13,
414
+ "learning_rate": 4.339356844457399e-05,
415
+ "loss": 0.0,
416
+ "step": 6800
417
+ },
418
+ {
419
+ "epoch": 0.13,
420
+ "learning_rate": 4.3296415039347136e-05,
421
+ "loss": 0.0,
422
+ "step": 6900
423
+ },
424
+ {
425
+ "epoch": 0.14,
426
+ "learning_rate": 4.319926163412028e-05,
427
+ "loss": 0.0,
428
+ "step": 7000
429
+ },
430
+ {
431
+ "epoch": 0.14,
432
+ "learning_rate": 4.3102108228893426e-05,
433
+ "loss": 0.0,
434
+ "step": 7100
435
+ },
436
+ {
437
+ "epoch": 0.14,
438
+ "learning_rate": 4.3004954823666574e-05,
439
+ "loss": 0.0,
440
+ "step": 7200
441
+ },
442
+ {
443
+ "epoch": 0.14,
444
+ "learning_rate": 4.2907801418439716e-05,
445
+ "loss": 0.0,
446
+ "step": 7300
447
+ },
448
+ {
449
+ "epoch": 0.14,
450
+ "learning_rate": 4.2810648013212865e-05,
451
+ "loss": 0.0,
452
+ "step": 7400
453
+ },
454
+ {
455
+ "epoch": 0.15,
456
+ "learning_rate": 4.271349460798601e-05,
457
+ "loss": 0.0,
458
+ "step": 7500
459
+ },
460
+ {
461
+ "epoch": 0.15,
462
+ "learning_rate": 4.2616341202759155e-05,
463
+ "loss": 0.0,
464
+ "step": 7600
465
+ },
466
+ {
467
+ "epoch": 0.15,
468
+ "learning_rate": 4.2519187797532303e-05,
469
+ "loss": 0.0,
470
+ "step": 7700
471
+ },
472
+ {
473
+ "epoch": 0.15,
474
+ "learning_rate": 4.242203439230545e-05,
475
+ "loss": 0.0,
476
+ "step": 7800
477
+ },
478
+ {
479
+ "epoch": 0.15,
480
+ "learning_rate": 4.2324880987078594e-05,
481
+ "loss": 0.0,
482
+ "step": 7900
483
+ },
484
+ {
485
+ "epoch": 0.16,
486
+ "learning_rate": 4.222772758185174e-05,
487
+ "loss": 0.0,
488
+ "step": 8000
489
+ },
490
+ {
491
+ "epoch": 0.16,
492
+ "learning_rate": 4.213057417662489e-05,
493
+ "loss": 0.0,
494
+ "step": 8100
495
+ },
496
+ {
497
+ "epoch": 0.16,
498
+ "learning_rate": 4.203342077139804e-05,
499
+ "loss": 0.0,
500
+ "step": 8200
501
+ },
502
+ {
503
+ "epoch": 0.16,
504
+ "learning_rate": 4.193626736617118e-05,
505
+ "loss": 0.0,
506
+ "step": 8300
507
+ },
508
+ {
509
+ "epoch": 0.16,
510
+ "learning_rate": 4.183911396094433e-05,
511
+ "loss": 0.0,
512
+ "step": 8400
513
+ },
514
+ {
515
+ "epoch": 0.17,
516
+ "learning_rate": 4.174196055571748e-05,
517
+ "loss": 0.0,
518
+ "step": 8500
519
+ },
520
+ {
521
+ "epoch": 0.17,
522
+ "learning_rate": 4.1644807150490627e-05,
523
+ "loss": 0.0,
524
+ "step": 8600
525
+ },
526
+ {
527
+ "epoch": 0.17,
528
+ "learning_rate": 4.1547653745263775e-05,
529
+ "loss": 0.0,
530
+ "step": 8700
531
+ },
532
+ {
533
+ "epoch": 0.17,
534
+ "learning_rate": 4.145050034003692e-05,
535
+ "loss": 0.0,
536
+ "step": 8800
537
+ },
538
+ {
539
+ "epoch": 0.17,
540
+ "learning_rate": 4.1353346934810065e-05,
541
+ "loss": 0.0,
542
+ "step": 8900
543
+ },
544
+ {
545
+ "epoch": 0.17,
546
+ "learning_rate": 4.1256193529583214e-05,
547
+ "loss": 0.0,
548
+ "step": 9000
549
+ },
550
+ {
551
+ "epoch": 0.18,
552
+ "learning_rate": 4.115904012435636e-05,
553
+ "loss": 0.0,
554
+ "step": 9100
555
+ },
556
+ {
557
+ "epoch": 0.18,
558
+ "learning_rate": 4.1061886719129504e-05,
559
+ "loss": 0.0,
560
+ "step": 9200
561
+ },
562
+ {
563
+ "epoch": 0.18,
564
+ "learning_rate": 4.096473331390265e-05,
565
+ "loss": 0.0,
566
+ "step": 9300
567
+ },
568
+ {
569
+ "epoch": 0.18,
570
+ "learning_rate": 4.08675799086758e-05,
571
+ "loss": 0.0,
572
+ "step": 9400
573
+ },
574
+ {
575
+ "epoch": 0.18,
576
+ "learning_rate": 4.077042650344895e-05,
577
+ "loss": 0.0,
578
+ "step": 9500
579
+ },
580
+ {
581
+ "epoch": 0.19,
582
+ "learning_rate": 4.067327309822209e-05,
583
+ "loss": 0.0,
584
+ "step": 9600
585
+ },
586
+ {
587
+ "epoch": 0.19,
588
+ "learning_rate": 4.057611969299524e-05,
589
+ "loss": 0.0,
590
+ "step": 9700
591
+ },
592
+ {
593
+ "epoch": 0.19,
594
+ "learning_rate": 4.047896628776839e-05,
595
+ "loss": 0.0,
596
+ "step": 9800
597
+ },
598
+ {
599
+ "epoch": 0.19,
600
+ "learning_rate": 4.038181288254154e-05,
601
+ "loss": 0.0,
602
+ "step": 9900
603
+ },
604
+ {
605
+ "epoch": 0.19,
606
+ "learning_rate": 4.028465947731468e-05,
607
+ "loss": 0.0,
608
+ "step": 10000
609
+ },
610
+ {
611
+ "epoch": 0.2,
612
+ "learning_rate": 4.018750607208783e-05,
613
+ "loss": 0.0,
614
+ "step": 10100
615
+ },
616
+ {
617
+ "epoch": 0.2,
618
+ "learning_rate": 4.0090352666860976e-05,
619
+ "loss": 0.0,
620
+ "step": 10200
621
+ },
622
+ {
623
+ "epoch": 0.2,
624
+ "learning_rate": 3.9993199261634124e-05,
625
+ "loss": 0.0,
626
+ "step": 10300
627
+ },
628
+ {
629
+ "epoch": 0.2,
630
+ "learning_rate": 3.989604585640727e-05,
631
+ "loss": 0.0,
632
+ "step": 10400
633
+ },
634
+ {
635
+ "epoch": 0.2,
636
+ "learning_rate": 3.9798892451180415e-05,
637
+ "loss": 0.0,
638
+ "step": 10500
639
+ },
640
+ {
641
+ "epoch": 0.21,
642
+ "learning_rate": 3.970173904595356e-05,
643
+ "loss": 0.0,
644
+ "step": 10600
645
+ },
646
+ {
647
+ "epoch": 0.21,
648
+ "learning_rate": 3.960458564072671e-05,
649
+ "loss": 0.0,
650
+ "step": 10700
651
+ },
652
+ {
653
+ "epoch": 0.21,
654
+ "learning_rate": 3.950743223549986e-05,
655
+ "loss": 0.0,
656
+ "step": 10800
657
+ },
658
+ {
659
+ "epoch": 0.21,
660
+ "learning_rate": 3.9410278830273e-05,
661
+ "loss": 0.0,
662
+ "step": 10900
663
+ },
664
+ {
665
+ "epoch": 0.21,
666
+ "learning_rate": 3.931312542504615e-05,
667
+ "loss": 0.0,
668
+ "step": 11000
669
+ },
670
+ {
671
+ "epoch": 0.22,
672
+ "learning_rate": 3.92159720198193e-05,
673
+ "loss": 0.0,
674
+ "step": 11100
675
+ },
676
+ {
677
+ "epoch": 0.22,
678
+ "learning_rate": 3.911881861459245e-05,
679
+ "loss": 0.0,
680
+ "step": 11200
681
+ },
682
+ {
683
+ "epoch": 0.22,
684
+ "learning_rate": 3.902166520936559e-05,
685
+ "loss": 0.0,
686
+ "step": 11300
687
+ },
688
+ {
689
+ "epoch": 0.22,
690
+ "learning_rate": 3.892451180413874e-05,
691
+ "loss": 0.0,
692
+ "step": 11400
693
+ },
694
+ {
695
+ "epoch": 0.22,
696
+ "learning_rate": 3.8827358398911886e-05,
697
+ "loss": 0.0,
698
+ "step": 11500
699
+ },
700
+ {
701
+ "epoch": 0.23,
702
+ "learning_rate": 3.8730204993685035e-05,
703
+ "loss": 0.0,
704
+ "step": 11600
705
+ },
706
+ {
707
+ "epoch": 0.23,
708
+ "learning_rate": 3.8633051588458183e-05,
709
+ "loss": 0.0,
710
+ "step": 11700
711
+ },
712
+ {
713
+ "epoch": 0.23,
714
+ "learning_rate": 3.8535898183231325e-05,
715
+ "loss": 0.0,
716
+ "step": 11800
717
+ },
718
+ {
719
+ "epoch": 0.23,
720
+ "learning_rate": 3.8438744778004474e-05,
721
+ "loss": 0.0,
722
+ "step": 11900
723
+ },
724
+ {
725
+ "epoch": 0.23,
726
+ "learning_rate": 3.8341591372777615e-05,
727
+ "loss": 0.0,
728
+ "step": 12000
729
+ },
730
+ {
731
+ "epoch": 0.24,
732
+ "learning_rate": 3.8244437967550764e-05,
733
+ "loss": 0.0,
734
+ "step": 12100
735
+ },
736
+ {
737
+ "epoch": 0.24,
738
+ "learning_rate": 3.814728456232391e-05,
739
+ "loss": 0.0,
740
+ "step": 12200
741
+ },
742
+ {
743
+ "epoch": 0.24,
744
+ "learning_rate": 3.8050131157097054e-05,
745
+ "loss": 0.0,
746
+ "step": 12300
747
+ },
748
+ {
749
+ "epoch": 0.24,
750
+ "learning_rate": 3.79529777518702e-05,
751
+ "loss": 0.0,
752
+ "step": 12400
753
+ },
754
+ {
755
+ "epoch": 0.24,
756
+ "learning_rate": 3.785582434664335e-05,
757
+ "loss": 0.0,
758
+ "step": 12500
759
+ },
760
+ {
761
+ "epoch": 0.24,
762
+ "learning_rate": 3.775867094141649e-05,
763
+ "loss": 0.0,
764
+ "step": 12600
765
+ },
766
+ {
767
+ "epoch": 0.25,
768
+ "learning_rate": 3.766151753618964e-05,
769
+ "loss": 0.0,
770
+ "step": 12700
771
+ },
772
+ {
773
+ "epoch": 0.25,
774
+ "learning_rate": 3.756436413096279e-05,
775
+ "loss": 0.0,
776
+ "step": 12800
777
+ },
778
+ {
779
+ "epoch": 0.25,
780
+ "learning_rate": 3.746721072573594e-05,
781
+ "loss": 0.0,
782
+ "step": 12900
783
+ },
784
+ {
785
+ "epoch": 0.25,
786
+ "learning_rate": 3.737005732050908e-05,
787
+ "loss": 0.0,
788
+ "step": 13000
789
+ },
790
+ {
791
+ "epoch": 0.25,
792
+ "learning_rate": 3.727290391528223e-05,
793
+ "loss": 0.0,
794
+ "step": 13100
795
+ },
796
+ {
797
+ "epoch": 0.26,
798
+ "learning_rate": 3.717575051005538e-05,
799
+ "loss": 0.0,
800
+ "step": 13200
801
+ },
802
+ {
803
+ "epoch": 0.26,
804
+ "learning_rate": 3.7078597104828526e-05,
805
+ "loss": 0.0,
806
+ "step": 13300
807
+ },
808
+ {
809
+ "epoch": 0.26,
810
+ "learning_rate": 3.6981443699601675e-05,
811
+ "loss": 0.0,
812
+ "step": 13400
813
+ },
814
+ {
815
+ "epoch": 0.26,
816
+ "learning_rate": 3.6884290294374816e-05,
817
+ "loss": 0.0,
818
+ "step": 13500
819
+ },
820
+ {
821
+ "epoch": 0.26,
822
+ "learning_rate": 3.6787136889147965e-05,
823
+ "loss": 0.0,
824
+ "step": 13600
825
+ },
826
+ {
827
+ "epoch": 0.27,
828
+ "learning_rate": 3.668998348392111e-05,
829
+ "loss": 0.0,
830
+ "step": 13700
831
+ },
832
+ {
833
+ "epoch": 0.27,
834
+ "learning_rate": 3.659283007869426e-05,
835
+ "loss": 0.0,
836
+ "step": 13800
837
+ },
838
+ {
839
+ "epoch": 0.27,
840
+ "learning_rate": 3.6495676673467404e-05,
841
+ "loss": 0.0,
842
+ "step": 13900
843
+ },
844
+ {
845
+ "epoch": 0.27,
846
+ "learning_rate": 3.639852326824055e-05,
847
+ "loss": 0.0,
848
+ "step": 14000
849
+ },
850
+ {
851
+ "epoch": 0.27,
852
+ "learning_rate": 3.63013698630137e-05,
853
+ "loss": 0.0,
854
+ "step": 14100
855
+ },
856
+ {
857
+ "epoch": 0.28,
858
+ "learning_rate": 3.620421645778685e-05,
859
+ "loss": 0.0,
860
+ "step": 14200
861
+ },
862
+ {
863
+ "epoch": 0.28,
864
+ "learning_rate": 3.610706305255999e-05,
865
+ "loss": 0.0,
866
+ "step": 14300
867
+ },
868
+ {
869
+ "epoch": 0.28,
870
+ "learning_rate": 3.600990964733314e-05,
871
+ "loss": 0.0,
872
+ "step": 14400
873
+ },
874
+ {
875
+ "epoch": 0.28,
876
+ "learning_rate": 3.591275624210629e-05,
877
+ "loss": 0.0,
878
+ "step": 14500
879
+ },
880
+ {
881
+ "epoch": 0.28,
882
+ "learning_rate": 3.5815602836879437e-05,
883
+ "loss": 0.0,
884
+ "step": 14600
885
+ },
886
+ {
887
+ "epoch": 0.29,
888
+ "learning_rate": 3.571844943165258e-05,
889
+ "loss": 0.0,
890
+ "step": 14700
891
+ },
892
+ {
893
+ "epoch": 0.29,
894
+ "learning_rate": 3.562129602642573e-05,
895
+ "loss": 0.0,
896
+ "step": 14800
897
+ },
898
+ {
899
+ "epoch": 0.29,
900
+ "learning_rate": 3.5524142621198875e-05,
901
+ "loss": 0.0,
902
+ "step": 14900
903
+ },
904
+ {
905
+ "epoch": 0.29,
906
+ "learning_rate": 3.5426989215972024e-05,
907
+ "loss": 0.0,
908
+ "step": 15000
909
+ },
910
+ {
911
+ "epoch": 0.29,
912
+ "learning_rate": 3.532983581074517e-05,
913
+ "loss": 0.0,
914
+ "step": 15100
915
+ },
916
+ {
917
+ "epoch": 0.3,
918
+ "learning_rate": 3.5232682405518314e-05,
919
+ "loss": 0.0,
920
+ "step": 15200
921
+ },
922
+ {
923
+ "epoch": 0.3,
924
+ "learning_rate": 3.513552900029146e-05,
925
+ "loss": 0.0,
926
+ "step": 15300
927
+ },
928
+ {
929
+ "epoch": 0.3,
930
+ "learning_rate": 3.503837559506461e-05,
931
+ "loss": 0.0,
932
+ "step": 15400
933
+ },
934
+ {
935
+ "epoch": 0.3,
936
+ "learning_rate": 3.494122218983776e-05,
937
+ "loss": 0.0,
938
+ "step": 15500
939
+ },
940
+ {
941
+ "epoch": 0.3,
942
+ "learning_rate": 3.48440687846109e-05,
943
+ "loss": 0.0,
944
+ "step": 15600
945
+ },
946
+ {
947
+ "epoch": 0.31,
948
+ "learning_rate": 3.474691537938405e-05,
949
+ "loss": 0.0,
950
+ "step": 15700
951
+ },
952
+ {
953
+ "epoch": 0.31,
954
+ "learning_rate": 3.46497619741572e-05,
955
+ "loss": 0.0,
956
+ "step": 15800
957
+ },
958
+ {
959
+ "epoch": 0.31,
960
+ "learning_rate": 3.455260856893035e-05,
961
+ "loss": 0.0,
962
+ "step": 15900
963
+ },
964
+ {
965
+ "epoch": 0.31,
966
+ "learning_rate": 3.445545516370349e-05,
967
+ "loss": 0.0,
968
+ "step": 16000
969
+ },
970
+ {
971
+ "epoch": 0.31,
972
+ "learning_rate": 3.435830175847664e-05,
973
+ "loss": 0.0,
974
+ "step": 16100
975
+ },
976
+ {
977
+ "epoch": 0.31,
978
+ "learning_rate": 3.4261148353249786e-05,
979
+ "loss": 0.0,
980
+ "step": 16200
981
+ },
982
+ {
983
+ "epoch": 0.32,
984
+ "learning_rate": 3.4163994948022934e-05,
985
+ "loss": 0.0,
986
+ "step": 16300
987
+ },
988
+ {
989
+ "epoch": 0.32,
990
+ "learning_rate": 3.4066841542796076e-05,
991
+ "loss": 0.0,
992
+ "step": 16400
993
+ },
994
+ {
995
+ "epoch": 0.32,
996
+ "learning_rate": 3.3969688137569225e-05,
997
+ "loss": 0.0,
998
+ "step": 16500
999
+ },
1000
+ {
1001
+ "epoch": 0.32,
1002
+ "learning_rate": 3.387253473234237e-05,
1003
+ "loss": 0.0,
1004
+ "step": 16600
1005
+ },
1006
+ {
1007
+ "epoch": 0.32,
1008
+ "learning_rate": 3.377538132711552e-05,
1009
+ "loss": 0.0,
1010
+ "step": 16700
1011
+ },
1012
+ {
1013
+ "epoch": 0.33,
1014
+ "learning_rate": 3.3678227921888663e-05,
1015
+ "loss": 0.0,
1016
+ "step": 16800
1017
+ },
1018
+ {
1019
+ "epoch": 0.33,
1020
+ "learning_rate": 3.358107451666181e-05,
1021
+ "loss": 0.0,
1022
+ "step": 16900
1023
+ },
1024
+ {
1025
+ "epoch": 0.33,
1026
+ "learning_rate": 3.348392111143496e-05,
1027
+ "loss": 0.0,
1028
+ "step": 17000
1029
+ },
1030
+ {
1031
+ "epoch": 0.33,
1032
+ "learning_rate": 3.33867677062081e-05,
1033
+ "loss": 0.0,
1034
+ "step": 17100
1035
+ },
1036
+ {
1037
+ "epoch": 0.33,
1038
+ "learning_rate": 3.328961430098125e-05,
1039
+ "loss": 0.0,
1040
+ "step": 17200
1041
+ },
1042
+ {
1043
+ "epoch": 0.34,
1044
+ "learning_rate": 3.319246089575439e-05,
1045
+ "loss": 0.0,
1046
+ "step": 17300
1047
+ },
1048
+ {
1049
+ "epoch": 0.34,
1050
+ "learning_rate": 3.309530749052754e-05,
1051
+ "loss": 0.0,
1052
+ "step": 17400
1053
+ },
1054
+ {
1055
+ "epoch": 0.34,
1056
+ "learning_rate": 3.299815408530069e-05,
1057
+ "loss": 0.0,
1058
+ "step": 17500
1059
+ },
1060
+ {
1061
+ "epoch": 0.34,
1062
+ "learning_rate": 3.290100068007384e-05,
1063
+ "loss": 0.0,
1064
+ "step": 17600
1065
+ },
1066
+ {
1067
+ "epoch": 0.34,
1068
+ "learning_rate": 3.280384727484698e-05,
1069
+ "loss": 0.0,
1070
+ "step": 17700
1071
+ },
1072
+ {
1073
+ "epoch": 0.35,
1074
+ "learning_rate": 3.270669386962013e-05,
1075
+ "loss": 0.0,
1076
+ "step": 17800
1077
+ },
1078
+ {
1079
+ "epoch": 0.35,
1080
+ "learning_rate": 3.260954046439328e-05,
1081
+ "loss": 0.0,
1082
+ "step": 17900
1083
+ },
1084
+ {
1085
+ "epoch": 0.35,
1086
+ "learning_rate": 3.2512387059166425e-05,
1087
+ "loss": 0.0,
1088
+ "step": 18000
1089
+ },
1090
+ {
1091
+ "epoch": 0.35,
1092
+ "learning_rate": 3.241523365393957e-05,
1093
+ "loss": 0.0,
1094
+ "step": 18100
1095
+ },
1096
+ {
1097
+ "epoch": 0.35,
1098
+ "learning_rate": 3.2318080248712716e-05,
1099
+ "loss": 0.0,
1100
+ "step": 18200
1101
+ },
1102
+ {
1103
+ "epoch": 0.36,
1104
+ "learning_rate": 3.2220926843485864e-05,
1105
+ "loss": 0.0,
1106
+ "step": 18300
1107
+ },
1108
+ {
1109
+ "epoch": 0.36,
1110
+ "learning_rate": 3.212377343825901e-05,
1111
+ "loss": 0.0,
1112
+ "step": 18400
1113
+ },
1114
+ {
1115
+ "epoch": 0.36,
1116
+ "learning_rate": 3.202662003303216e-05,
1117
+ "loss": 0.0,
1118
+ "step": 18500
1119
+ },
1120
+ {
1121
+ "epoch": 0.36,
1122
+ "learning_rate": 3.19294666278053e-05,
1123
+ "loss": 0.0,
1124
+ "step": 18600
1125
+ },
1126
+ {
1127
+ "epoch": 0.36,
1128
+ "learning_rate": 3.183231322257845e-05,
1129
+ "loss": 0.0,
1130
+ "step": 18700
1131
+ },
1132
+ {
1133
+ "epoch": 0.37,
1134
+ "learning_rate": 3.17351598173516e-05,
1135
+ "loss": 0.0,
1136
+ "step": 18800
1137
+ },
1138
+ {
1139
+ "epoch": 0.37,
1140
+ "learning_rate": 3.163800641212475e-05,
1141
+ "loss": 0.0,
1142
+ "step": 18900
1143
+ },
1144
+ {
1145
+ "epoch": 0.37,
1146
+ "learning_rate": 3.154085300689789e-05,
1147
+ "loss": 0.0,
1148
+ "step": 19000
1149
+ },
1150
+ {
1151
+ "epoch": 0.37,
1152
+ "learning_rate": 3.144369960167104e-05,
1153
+ "loss": 0.0,
1154
+ "step": 19100
1155
+ },
1156
+ {
1157
+ "epoch": 0.37,
1158
+ "learning_rate": 3.134654619644419e-05,
1159
+ "loss": 0.0,
1160
+ "step": 19200
1161
+ },
1162
+ {
1163
+ "epoch": 0.38,
1164
+ "learning_rate": 3.1249392791217336e-05,
1165
+ "loss": 0.0,
1166
+ "step": 19300
1167
+ },
1168
+ {
1169
+ "epoch": 0.38,
1170
+ "learning_rate": 3.115223938599048e-05,
1171
+ "loss": 0.0,
1172
+ "step": 19400
1173
+ },
1174
+ {
1175
+ "epoch": 0.38,
1176
+ "learning_rate": 3.1055085980763626e-05,
1177
+ "loss": 0.0,
1178
+ "step": 19500
1179
+ },
1180
+ {
1181
+ "epoch": 0.38,
1182
+ "learning_rate": 3.0957932575536775e-05,
1183
+ "loss": 0.0,
1184
+ "step": 19600
1185
+ },
1186
+ {
1187
+ "epoch": 0.38,
1188
+ "learning_rate": 3.086077917030992e-05,
1189
+ "loss": 0.0,
1190
+ "step": 19700
1191
+ },
1192
+ {
1193
+ "epoch": 0.38,
1194
+ "learning_rate": 3.0763625765083065e-05,
1195
+ "loss": 0.0,
1196
+ "step": 19800
1197
+ },
1198
+ {
1199
+ "epoch": 0.39,
1200
+ "learning_rate": 3.0666472359856214e-05,
1201
+ "loss": 0.0,
1202
+ "step": 19900
1203
+ },
1204
+ {
1205
+ "epoch": 0.39,
1206
+ "learning_rate": 3.056931895462936e-05,
1207
+ "loss": 0.0,
1208
+ "step": 20000
1209
+ },
1210
+ {
1211
+ "epoch": 0.39,
1212
+ "learning_rate": 3.047216554940251e-05,
1213
+ "loss": 0.0,
1214
+ "step": 20100
1215
+ },
1216
+ {
1217
+ "epoch": 0.39,
1218
+ "learning_rate": 3.037501214417566e-05,
1219
+ "loss": 0.0,
1220
+ "step": 20200
1221
+ },
1222
+ {
1223
+ "epoch": 0.39,
1224
+ "learning_rate": 3.02778587389488e-05,
1225
+ "loss": 0.0,
1226
+ "step": 20300
1227
+ },
1228
+ {
1229
+ "epoch": 0.4,
1230
+ "learning_rate": 3.018070533372195e-05,
1231
+ "loss": 0.0,
1232
+ "step": 20400
1233
+ },
1234
+ {
1235
+ "epoch": 0.4,
1236
+ "learning_rate": 3.0083551928495095e-05,
1237
+ "loss": 0.0,
1238
+ "step": 20500
1239
+ },
1240
+ {
1241
+ "epoch": 0.4,
1242
+ "learning_rate": 2.9986398523268243e-05,
1243
+ "loss": 0.0,
1244
+ "step": 20600
1245
+ },
1246
+ {
1247
+ "epoch": 0.4,
1248
+ "learning_rate": 2.9889245118041388e-05,
1249
+ "loss": 0.0,
1250
+ "step": 20700
1251
+ },
1252
+ {
1253
+ "epoch": 0.4,
1254
+ "learning_rate": 2.9792091712814533e-05,
1255
+ "loss": 0.0,
1256
+ "step": 20800
1257
+ },
1258
+ {
1259
+ "epoch": 0.41,
1260
+ "learning_rate": 2.9694938307587682e-05,
1261
+ "loss": 0.0,
1262
+ "step": 20900
1263
+ },
1264
+ {
1265
+ "epoch": 0.41,
1266
+ "learning_rate": 2.959778490236083e-05,
1267
+ "loss": 0.0,
1268
+ "step": 21000
1269
+ },
1270
+ {
1271
+ "epoch": 0.41,
1272
+ "learning_rate": 2.9500631497133972e-05,
1273
+ "loss": 0.0,
1274
+ "step": 21100
1275
+ },
1276
+ {
1277
+ "epoch": 0.41,
1278
+ "learning_rate": 2.940347809190712e-05,
1279
+ "loss": 0.0,
1280
+ "step": 21200
1281
+ },
1282
+ {
1283
+ "epoch": 0.41,
1284
+ "learning_rate": 2.930632468668027e-05,
1285
+ "loss": 0.0,
1286
+ "step": 21300
1287
+ },
1288
+ {
1289
+ "epoch": 0.42,
1290
+ "learning_rate": 2.9209171281453418e-05,
1291
+ "loss": 0.0,
1292
+ "step": 21400
1293
+ },
1294
+ {
1295
+ "epoch": 0.42,
1296
+ "learning_rate": 2.911201787622656e-05,
1297
+ "loss": 0.0,
1298
+ "step": 21500
1299
+ },
1300
+ {
1301
+ "epoch": 0.42,
1302
+ "learning_rate": 2.9014864470999708e-05,
1303
+ "loss": 0.0,
1304
+ "step": 21600
1305
+ },
1306
+ {
1307
+ "epoch": 0.42,
1308
+ "learning_rate": 2.8917711065772857e-05,
1309
+ "loss": 0.0,
1310
+ "step": 21700
1311
+ },
1312
+ {
1313
+ "epoch": 0.42,
1314
+ "learning_rate": 2.8820557660546005e-05,
1315
+ "loss": 0.0,
1316
+ "step": 21800
1317
+ },
1318
+ {
1319
+ "epoch": 0.43,
1320
+ "learning_rate": 2.8723404255319154e-05,
1321
+ "loss": 0.0,
1322
+ "step": 21900
1323
+ },
1324
+ {
1325
+ "epoch": 0.43,
1326
+ "learning_rate": 2.8626250850092295e-05,
1327
+ "loss": 0.0,
1328
+ "step": 22000
1329
+ },
1330
+ {
1331
+ "epoch": 0.43,
1332
+ "learning_rate": 2.8529097444865444e-05,
1333
+ "loss": 0.0,
1334
+ "step": 22100
1335
+ },
1336
+ {
1337
+ "epoch": 0.43,
1338
+ "learning_rate": 2.8431944039638592e-05,
1339
+ "loss": 0.0,
1340
+ "step": 22200
1341
+ },
1342
+ {
1343
+ "epoch": 0.43,
1344
+ "learning_rate": 2.833479063441174e-05,
1345
+ "loss": 0.0,
1346
+ "step": 22300
1347
+ },
1348
+ {
1349
+ "epoch": 0.44,
1350
+ "learning_rate": 2.8237637229184883e-05,
1351
+ "loss": 0.0,
1352
+ "step": 22400
1353
+ },
1354
+ {
1355
+ "epoch": 0.44,
1356
+ "learning_rate": 2.814048382395803e-05,
1357
+ "loss": 0.0,
1358
+ "step": 22500
1359
+ },
1360
+ {
1361
+ "epoch": 0.44,
1362
+ "learning_rate": 2.804333041873118e-05,
1363
+ "loss": 0.0,
1364
+ "step": 22600
1365
+ },
1366
+ {
1367
+ "epoch": 0.44,
1368
+ "learning_rate": 2.7946177013504328e-05,
1369
+ "loss": 0.0,
1370
+ "step": 22700
1371
+ },
1372
+ {
1373
+ "epoch": 0.44,
1374
+ "learning_rate": 2.784902360827747e-05,
1375
+ "loss": 0.0,
1376
+ "step": 22800
1377
+ },
1378
+ {
1379
+ "epoch": 0.44,
1380
+ "learning_rate": 2.775187020305062e-05,
1381
+ "loss": 0.0,
1382
+ "step": 22900
1383
+ },
1384
+ {
1385
+ "epoch": 0.45,
1386
+ "learning_rate": 2.7654716797823767e-05,
1387
+ "loss": 0.0,
1388
+ "step": 23000
1389
+ },
1390
+ {
1391
+ "epoch": 0.45,
1392
+ "learning_rate": 2.7557563392596912e-05,
1393
+ "loss": 0.0,
1394
+ "step": 23100
1395
+ },
1396
+ {
1397
+ "epoch": 0.45,
1398
+ "learning_rate": 2.7460409987370057e-05,
1399
+ "loss": 0.0,
1400
+ "step": 23200
1401
+ },
1402
+ {
1403
+ "epoch": 0.45,
1404
+ "learning_rate": 2.7363256582143202e-05,
1405
+ "loss": 0.0,
1406
+ "step": 23300
1407
+ },
1408
+ {
1409
+ "epoch": 0.45,
1410
+ "learning_rate": 2.726610317691635e-05,
1411
+ "loss": 0.0,
1412
+ "step": 23400
1413
+ },
1414
+ {
1415
+ "epoch": 0.46,
1416
+ "learning_rate": 2.71689497716895e-05,
1417
+ "loss": 0.0,
1418
+ "step": 23500
1419
+ },
1420
+ {
1421
+ "epoch": 0.46,
1422
+ "learning_rate": 2.7071796366462648e-05,
1423
+ "loss": 0.0,
1424
+ "step": 23600
1425
+ },
1426
+ {
1427
+ "epoch": 0.46,
1428
+ "learning_rate": 2.697464296123579e-05,
1429
+ "loss": 0.0,
1430
+ "step": 23700
1431
+ },
1432
+ {
1433
+ "epoch": 0.46,
1434
+ "learning_rate": 2.6877489556008938e-05,
1435
+ "loss": 0.0,
1436
+ "step": 23800
1437
+ },
1438
+ {
1439
+ "epoch": 0.46,
1440
+ "learning_rate": 2.6780336150782087e-05,
1441
+ "loss": 0.0,
1442
+ "step": 23900
1443
+ },
1444
+ {
1445
+ "epoch": 0.47,
1446
+ "learning_rate": 2.6683182745555235e-05,
1447
+ "loss": 0.0,
1448
+ "step": 24000
1449
+ },
1450
+ {
1451
+ "epoch": 0.47,
1452
+ "learning_rate": 2.6586029340328377e-05,
1453
+ "loss": 0.0,
1454
+ "step": 24100
1455
+ },
1456
+ {
1457
+ "epoch": 0.47,
1458
+ "learning_rate": 2.6488875935101526e-05,
1459
+ "loss": 0.0,
1460
+ "step": 24200
1461
+ },
1462
+ {
1463
+ "epoch": 0.47,
1464
+ "learning_rate": 2.6391722529874674e-05,
1465
+ "loss": 0.0,
1466
+ "step": 24300
1467
+ },
1468
+ {
1469
+ "epoch": 0.47,
1470
+ "learning_rate": 2.6294569124647823e-05,
1471
+ "loss": 0.0,
1472
+ "step": 24400
1473
+ },
1474
+ {
1475
+ "epoch": 0.48,
1476
+ "learning_rate": 2.6197415719420964e-05,
1477
+ "loss": 0.0,
1478
+ "step": 24500
1479
+ },
1480
+ {
1481
+ "epoch": 0.48,
1482
+ "learning_rate": 2.6100262314194113e-05,
1483
+ "loss": 0.0,
1484
+ "step": 24600
1485
+ },
1486
+ {
1487
+ "epoch": 0.48,
1488
+ "learning_rate": 2.600310890896726e-05,
1489
+ "loss": 0.0,
1490
+ "step": 24700
1491
+ },
1492
+ {
1493
+ "epoch": 0.48,
1494
+ "learning_rate": 2.590595550374041e-05,
1495
+ "loss": 0.0,
1496
+ "step": 24800
1497
+ },
1498
+ {
1499
+ "epoch": 0.48,
1500
+ "learning_rate": 2.5808802098513552e-05,
1501
+ "loss": 0.0,
1502
+ "step": 24900
1503
+ },
1504
+ {
1505
+ "epoch": 0.49,
1506
+ "learning_rate": 2.57116486932867e-05,
1507
+ "loss": 0.0,
1508
+ "step": 25000
1509
+ },
1510
+ {
1511
+ "epoch": 0.49,
1512
+ "learning_rate": 2.561449528805985e-05,
1513
+ "loss": 0.0,
1514
+ "step": 25100
1515
+ },
1516
+ {
1517
+ "epoch": 0.49,
1518
+ "learning_rate": 2.5517341882832997e-05,
1519
+ "loss": 0.0,
1520
+ "step": 25200
1521
+ },
1522
+ {
1523
+ "epoch": 0.49,
1524
+ "learning_rate": 2.5420188477606142e-05,
1525
+ "loss": 0.0,
1526
+ "step": 25300
1527
+ },
1528
+ {
1529
+ "epoch": 0.49,
1530
+ "learning_rate": 2.5323035072379288e-05,
1531
+ "loss": 0.0,
1532
+ "step": 25400
1533
+ },
1534
+ {
1535
+ "epoch": 0.5,
1536
+ "learning_rate": 2.5225881667152436e-05,
1537
+ "loss": 0.0,
1538
+ "step": 25500
1539
+ },
1540
+ {
1541
+ "epoch": 0.5,
1542
+ "learning_rate": 2.512872826192558e-05,
1543
+ "loss": 0.0,
1544
+ "step": 25600
1545
+ },
1546
+ {
1547
+ "epoch": 0.5,
1548
+ "learning_rate": 2.503157485669873e-05,
1549
+ "loss": 0.0,
1550
+ "step": 25700
1551
+ },
1552
+ {
1553
+ "epoch": 0.5,
1554
+ "learning_rate": 2.4934421451471875e-05,
1555
+ "loss": 0.0,
1556
+ "step": 25800
1557
+ },
1558
+ {
1559
+ "epoch": 0.5,
1560
+ "learning_rate": 2.483726804624502e-05,
1561
+ "loss": 0.0,
1562
+ "step": 25900
1563
+ },
1564
+ {
1565
+ "epoch": 0.51,
1566
+ "learning_rate": 2.474011464101817e-05,
1567
+ "loss": 0.0,
1568
+ "step": 26000
1569
+ },
1570
+ {
1571
+ "epoch": 0.51,
1572
+ "learning_rate": 2.4642961235791314e-05,
1573
+ "loss": 0.0,
1574
+ "step": 26100
1575
+ },
1576
+ {
1577
+ "epoch": 0.51,
1578
+ "learning_rate": 2.4545807830564462e-05,
1579
+ "loss": 0.0,
1580
+ "step": 26200
1581
+ },
1582
+ {
1583
+ "epoch": 0.51,
1584
+ "learning_rate": 2.4448654425337607e-05,
1585
+ "loss": 0.0,
1586
+ "step": 26300
1587
+ },
1588
+ {
1589
+ "epoch": 0.51,
1590
+ "learning_rate": 2.4351501020110756e-05,
1591
+ "loss": 0.0,
1592
+ "step": 26400
1593
+ },
1594
+ {
1595
+ "epoch": 0.51,
1596
+ "learning_rate": 2.42543476148839e-05,
1597
+ "loss": 0.0,
1598
+ "step": 26500
1599
+ },
1600
+ {
1601
+ "epoch": 0.52,
1602
+ "learning_rate": 2.415719420965705e-05,
1603
+ "loss": 0.0,
1604
+ "step": 26600
1605
+ },
1606
+ {
1607
+ "epoch": 0.52,
1608
+ "learning_rate": 2.4060040804430198e-05,
1609
+ "loss": 0.0,
1610
+ "step": 26700
1611
+ },
1612
+ {
1613
+ "epoch": 0.52,
1614
+ "learning_rate": 2.3962887399203343e-05,
1615
+ "loss": 0.0,
1616
+ "step": 26800
1617
+ },
1618
+ {
1619
+ "epoch": 0.52,
1620
+ "learning_rate": 2.3865733993976492e-05,
1621
+ "loss": 0.0,
1622
+ "step": 26900
1623
+ },
1624
+ {
1625
+ "epoch": 0.52,
1626
+ "learning_rate": 2.3768580588749637e-05,
1627
+ "loss": 0.0,
1628
+ "step": 27000
1629
+ },
1630
+ {
1631
+ "epoch": 0.53,
1632
+ "learning_rate": 2.3671427183522785e-05,
1633
+ "loss": 0.0,
1634
+ "step": 27100
1635
+ },
1636
+ {
1637
+ "epoch": 0.53,
1638
+ "learning_rate": 2.357427377829593e-05,
1639
+ "loss": 0.0,
1640
+ "step": 27200
1641
+ },
1642
+ {
1643
+ "epoch": 0.53,
1644
+ "learning_rate": 2.347712037306908e-05,
1645
+ "loss": 0.0,
1646
+ "step": 27300
1647
+ },
1648
+ {
1649
+ "epoch": 0.53,
1650
+ "learning_rate": 2.3379966967842224e-05,
1651
+ "loss": 0.0,
1652
+ "step": 27400
1653
+ },
1654
+ {
1655
+ "epoch": 0.53,
1656
+ "learning_rate": 2.3282813562615373e-05,
1657
+ "loss": 0.0,
1658
+ "step": 27500
1659
+ },
1660
+ {
1661
+ "epoch": 0.54,
1662
+ "learning_rate": 2.3185660157388518e-05,
1663
+ "loss": 0.0,
1664
+ "step": 27600
1665
+ },
1666
+ {
1667
+ "epoch": 0.54,
1668
+ "learning_rate": 2.3088506752161666e-05,
1669
+ "loss": 0.0,
1670
+ "step": 27700
1671
+ },
1672
+ {
1673
+ "epoch": 0.54,
1674
+ "learning_rate": 2.299135334693481e-05,
1675
+ "loss": 0.0,
1676
+ "step": 27800
1677
+ },
1678
+ {
1679
+ "epoch": 0.54,
1680
+ "learning_rate": 2.2894199941707957e-05,
1681
+ "loss": 0.0,
1682
+ "step": 27900
1683
+ },
1684
+ {
1685
+ "epoch": 0.54,
1686
+ "learning_rate": 2.2797046536481105e-05,
1687
+ "loss": 0.0,
1688
+ "step": 28000
1689
+ },
1690
+ {
1691
+ "epoch": 0.55,
1692
+ "learning_rate": 2.269989313125425e-05,
1693
+ "loss": 0.0,
1694
+ "step": 28100
1695
+ },
1696
+ {
1697
+ "epoch": 0.55,
1698
+ "learning_rate": 2.2602739726027396e-05,
1699
+ "loss": 0.0,
1700
+ "step": 28200
1701
+ },
1702
+ {
1703
+ "epoch": 0.55,
1704
+ "learning_rate": 2.2505586320800544e-05,
1705
+ "loss": 0.0,
1706
+ "step": 28300
1707
+ },
1708
+ {
1709
+ "epoch": 0.55,
1710
+ "learning_rate": 2.2408432915573693e-05,
1711
+ "loss": 0.0,
1712
+ "step": 28400
1713
+ },
1714
+ {
1715
+ "epoch": 0.55,
1716
+ "learning_rate": 2.2311279510346838e-05,
1717
+ "loss": 0.0,
1718
+ "step": 28500
1719
+ },
1720
+ {
1721
+ "epoch": 0.56,
1722
+ "learning_rate": 2.2214126105119986e-05,
1723
+ "loss": 0.0,
1724
+ "step": 28600
1725
+ },
1726
+ {
1727
+ "epoch": 0.56,
1728
+ "learning_rate": 2.211697269989313e-05,
1729
+ "loss": 0.0,
1730
+ "step": 28700
1731
+ },
1732
+ {
1733
+ "epoch": 0.56,
1734
+ "learning_rate": 2.201981929466628e-05,
1735
+ "loss": 0.0,
1736
+ "step": 28800
1737
+ },
1738
+ {
1739
+ "epoch": 0.56,
1740
+ "learning_rate": 2.1922665889439425e-05,
1741
+ "loss": 0.0,
1742
+ "step": 28900
1743
+ },
1744
+ {
1745
+ "epoch": 0.56,
1746
+ "learning_rate": 2.1825512484212574e-05,
1747
+ "loss": 0.0,
1748
+ "step": 29000
1749
+ },
1750
+ {
1751
+ "epoch": 0.57,
1752
+ "learning_rate": 2.172835907898572e-05,
1753
+ "loss": 0.0,
1754
+ "step": 29100
1755
+ },
1756
+ {
1757
+ "epoch": 0.57,
1758
+ "learning_rate": 2.1631205673758867e-05,
1759
+ "loss": 0.0,
1760
+ "step": 29200
1761
+ },
1762
+ {
1763
+ "epoch": 0.57,
1764
+ "learning_rate": 2.1534052268532012e-05,
1765
+ "loss": 0.0,
1766
+ "step": 29300
1767
+ },
1768
+ {
1769
+ "epoch": 0.57,
1770
+ "learning_rate": 2.143689886330516e-05,
1771
+ "loss": 0.0,
1772
+ "step": 29400
1773
+ },
1774
+ {
1775
+ "epoch": 0.57,
1776
+ "learning_rate": 2.1339745458078306e-05,
1777
+ "loss": 0.0,
1778
+ "step": 29500
1779
+ },
1780
+ {
1781
+ "epoch": 0.58,
1782
+ "learning_rate": 2.1242592052851455e-05,
1783
+ "loss": 0.0,
1784
+ "step": 29600
1785
+ },
1786
+ {
1787
+ "epoch": 0.58,
1788
+ "learning_rate": 2.11454386476246e-05,
1789
+ "loss": 0.0,
1790
+ "step": 29700
1791
+ },
1792
+ {
1793
+ "epoch": 0.58,
1794
+ "learning_rate": 2.1048285242397748e-05,
1795
+ "loss": 0.0,
1796
+ "step": 29800
1797
+ },
1798
+ {
1799
+ "epoch": 0.58,
1800
+ "learning_rate": 2.0951131837170893e-05,
1801
+ "loss": 0.0,
1802
+ "step": 29900
1803
+ },
1804
+ {
1805
+ "epoch": 0.58,
1806
+ "learning_rate": 2.0853978431944042e-05,
1807
+ "loss": 0.0,
1808
+ "step": 30000
1809
+ },
1810
+ {
1811
+ "epoch": 0.58,
1812
+ "learning_rate": 2.075682502671719e-05,
1813
+ "loss": 0.0,
1814
+ "step": 30100
1815
+ },
1816
+ {
1817
+ "epoch": 0.59,
1818
+ "learning_rate": 2.0659671621490336e-05,
1819
+ "loss": 0.0,
1820
+ "step": 30200
1821
+ },
1822
+ {
1823
+ "epoch": 0.59,
1824
+ "learning_rate": 2.056251821626348e-05,
1825
+ "loss": 0.0,
1826
+ "step": 30300
1827
+ },
1828
+ {
1829
+ "epoch": 0.59,
1830
+ "learning_rate": 2.0465364811036626e-05,
1831
+ "loss": 0.0,
1832
+ "step": 30400
1833
+ },
1834
+ {
1835
+ "epoch": 0.59,
1836
+ "learning_rate": 2.0368211405809774e-05,
1837
+ "loss": 0.0,
1838
+ "step": 30500
1839
+ },
1840
+ {
1841
+ "epoch": 0.59,
1842
+ "learning_rate": 2.027105800058292e-05,
1843
+ "loss": 0.0,
1844
+ "step": 30600
1845
+ },
1846
+ {
1847
+ "epoch": 0.6,
1848
+ "learning_rate": 2.0173904595356068e-05,
1849
+ "loss": 0.0,
1850
+ "step": 30700
1851
+ },
1852
+ {
1853
+ "epoch": 0.6,
1854
+ "learning_rate": 2.0076751190129213e-05,
1855
+ "loss": 0.0,
1856
+ "step": 30800
1857
+ },
1858
+ {
1859
+ "epoch": 0.6,
1860
+ "learning_rate": 1.997959778490236e-05,
1861
+ "loss": 0.0,
1862
+ "step": 30900
1863
+ },
1864
+ {
1865
+ "epoch": 0.6,
1866
+ "learning_rate": 1.9882444379675507e-05,
1867
+ "loss": 0.0,
1868
+ "step": 31000
1869
+ },
1870
+ {
1871
+ "epoch": 0.6,
1872
+ "learning_rate": 1.9785290974448655e-05,
1873
+ "loss": 0.0,
1874
+ "step": 31100
1875
+ },
1876
+ {
1877
+ "epoch": 0.61,
1878
+ "learning_rate": 1.96881375692218e-05,
1879
+ "loss": 0.0,
1880
+ "step": 31200
1881
+ },
1882
+ {
1883
+ "epoch": 0.61,
1884
+ "learning_rate": 1.959098416399495e-05,
1885
+ "loss": 0.0,
1886
+ "step": 31300
1887
+ },
1888
+ {
1889
+ "epoch": 0.61,
1890
+ "learning_rate": 1.9493830758768094e-05,
1891
+ "loss": 0.0,
1892
+ "step": 31400
1893
+ },
1894
+ {
1895
+ "epoch": 0.61,
1896
+ "learning_rate": 1.9396677353541243e-05,
1897
+ "loss": 0.0,
1898
+ "step": 31500
1899
+ },
1900
+ {
1901
+ "epoch": 0.61,
1902
+ "learning_rate": 1.9299523948314388e-05,
1903
+ "loss": 0.0,
1904
+ "step": 31600
1905
+ },
1906
+ {
1907
+ "epoch": 0.62,
1908
+ "learning_rate": 1.9202370543087536e-05,
1909
+ "loss": 0.0,
1910
+ "step": 31700
1911
+ },
1912
+ {
1913
+ "epoch": 0.62,
1914
+ "learning_rate": 1.9105217137860685e-05,
1915
+ "loss": 0.0,
1916
+ "step": 31800
1917
+ },
1918
+ {
1919
+ "epoch": 0.62,
1920
+ "learning_rate": 1.900806373263383e-05,
1921
+ "loss": 0.0,
1922
+ "step": 31900
1923
+ },
1924
+ {
1925
+ "epoch": 0.62,
1926
+ "learning_rate": 1.891091032740698e-05,
1927
+ "loss": 0.0,
1928
+ "step": 32000
1929
+ },
1930
+ {
1931
+ "epoch": 0.62,
1932
+ "learning_rate": 1.8813756922180124e-05,
1933
+ "loss": 0.0,
1934
+ "step": 32100
1935
+ },
1936
+ {
1937
+ "epoch": 0.63,
1938
+ "learning_rate": 1.8716603516953272e-05,
1939
+ "loss": 0.0,
1940
+ "step": 32200
1941
+ },
1942
+ {
1943
+ "epoch": 0.63,
1944
+ "learning_rate": 1.8619450111726417e-05,
1945
+ "loss": 0.0,
1946
+ "step": 32300
1947
+ },
1948
+ {
1949
+ "epoch": 0.63,
1950
+ "learning_rate": 1.8522296706499566e-05,
1951
+ "loss": 0.0,
1952
+ "step": 32400
1953
+ },
1954
+ {
1955
+ "epoch": 0.63,
1956
+ "learning_rate": 1.842514330127271e-05,
1957
+ "loss": 0.0,
1958
+ "step": 32500
1959
+ },
1960
+ {
1961
+ "epoch": 0.63,
1962
+ "learning_rate": 1.832798989604586e-05,
1963
+ "loss": 0.0,
1964
+ "step": 32600
1965
+ },
1966
+ {
1967
+ "epoch": 0.64,
1968
+ "learning_rate": 1.8230836490819005e-05,
1969
+ "loss": 0.0,
1970
+ "step": 32700
1971
+ },
1972
+ {
1973
+ "epoch": 0.64,
1974
+ "learning_rate": 1.813368308559215e-05,
1975
+ "loss": 0.0,
1976
+ "step": 32800
1977
+ },
1978
+ {
1979
+ "epoch": 0.64,
1980
+ "learning_rate": 1.80365296803653e-05,
1981
+ "loss": 0.0,
1982
+ "step": 32900
1983
+ },
1984
+ {
1985
+ "epoch": 0.64,
1986
+ "learning_rate": 1.7939376275138443e-05,
1987
+ "loss": 0.0,
1988
+ "step": 33000
1989
+ },
1990
+ {
1991
+ "epoch": 0.64,
1992
+ "learning_rate": 1.784222286991159e-05,
1993
+ "loss": 0.0,
1994
+ "step": 33100
1995
+ },
1996
+ {
1997
+ "epoch": 0.65,
1998
+ "learning_rate": 1.7745069464684737e-05,
1999
+ "loss": 0.0,
2000
+ "step": 33200
2001
+ },
2002
+ {
2003
+ "epoch": 0.65,
2004
+ "learning_rate": 1.7647916059457882e-05,
2005
+ "loss": 0.0,
2006
+ "step": 33300
2007
+ },
2008
+ {
2009
+ "epoch": 0.65,
2010
+ "learning_rate": 1.755076265423103e-05,
2011
+ "loss": 0.0,
2012
+ "step": 33400
2013
+ },
2014
+ {
2015
+ "epoch": 0.65,
2016
+ "learning_rate": 1.745360924900418e-05,
2017
+ "loss": 0.0,
2018
+ "step": 33500
2019
+ },
2020
+ {
2021
+ "epoch": 0.65,
2022
+ "learning_rate": 1.7356455843777324e-05,
2023
+ "loss": 0.0,
2024
+ "step": 33600
2025
+ },
2026
+ {
2027
+ "epoch": 0.65,
2028
+ "learning_rate": 1.7259302438550473e-05,
2029
+ "loss": 0.0,
2030
+ "step": 33700
2031
+ },
2032
+ {
2033
+ "epoch": 0.66,
2034
+ "learning_rate": 1.7162149033323618e-05,
2035
+ "loss": 0.0,
2036
+ "step": 33800
2037
+ },
2038
+ {
2039
+ "epoch": 0.66,
2040
+ "learning_rate": 1.7064995628096767e-05,
2041
+ "loss": 0.0,
2042
+ "step": 33900
2043
+ },
2044
+ {
2045
+ "epoch": 0.66,
2046
+ "learning_rate": 1.6967842222869912e-05,
2047
+ "loss": 0.0,
2048
+ "step": 34000
2049
+ },
2050
+ {
2051
+ "epoch": 0.66,
2052
+ "learning_rate": 1.687068881764306e-05,
2053
+ "loss": 0.0,
2054
+ "step": 34100
2055
+ },
2056
+ {
2057
+ "epoch": 0.66,
2058
+ "learning_rate": 1.6773535412416205e-05,
2059
+ "loss": 0.0,
2060
+ "step": 34200
2061
+ },
2062
+ {
2063
+ "epoch": 0.67,
2064
+ "learning_rate": 1.6676382007189354e-05,
2065
+ "loss": 0.0,
2066
+ "step": 34300
2067
+ },
2068
+ {
2069
+ "epoch": 0.67,
2070
+ "learning_rate": 1.65792286019625e-05,
2071
+ "loss": 0.0,
2072
+ "step": 34400
2073
+ },
2074
+ {
2075
+ "epoch": 0.67,
2076
+ "learning_rate": 1.6482075196735648e-05,
2077
+ "loss": 0.0,
2078
+ "step": 34500
2079
+ },
2080
+ {
2081
+ "epoch": 0.67,
2082
+ "learning_rate": 1.6384921791508793e-05,
2083
+ "loss": 0.0,
2084
+ "step": 34600
2085
+ },
2086
+ {
2087
+ "epoch": 0.67,
2088
+ "learning_rate": 1.628776838628194e-05,
2089
+ "loss": 0.0,
2090
+ "step": 34700
2091
+ },
2092
+ {
2093
+ "epoch": 0.68,
2094
+ "learning_rate": 1.6190614981055086e-05,
2095
+ "loss": 0.0,
2096
+ "step": 34800
2097
+ },
2098
+ {
2099
+ "epoch": 0.68,
2100
+ "learning_rate": 1.6093461575828235e-05,
2101
+ "loss": 0.0,
2102
+ "step": 34900
2103
+ },
2104
+ {
2105
+ "epoch": 0.68,
2106
+ "learning_rate": 1.599630817060138e-05,
2107
+ "loss": 0.0,
2108
+ "step": 35000
2109
+ },
2110
+ {
2111
+ "epoch": 0.68,
2112
+ "learning_rate": 1.589915476537453e-05,
2113
+ "loss": 0.0,
2114
+ "step": 35100
2115
+ },
2116
+ {
2117
+ "epoch": 0.68,
2118
+ "learning_rate": 1.5802001360147674e-05,
2119
+ "loss": 0.0,
2120
+ "step": 35200
2121
+ },
2122
+ {
2123
+ "epoch": 0.69,
2124
+ "learning_rate": 1.570484795492082e-05,
2125
+ "loss": 0.0,
2126
+ "step": 35300
2127
+ },
2128
+ {
2129
+ "epoch": 0.69,
2130
+ "learning_rate": 1.5607694549693967e-05,
2131
+ "loss": 0.0,
2132
+ "step": 35400
2133
+ },
2134
+ {
2135
+ "epoch": 0.69,
2136
+ "learning_rate": 1.5510541144467113e-05,
2137
+ "loss": 0.0,
2138
+ "step": 35500
2139
+ },
2140
+ {
2141
+ "epoch": 0.69,
2142
+ "learning_rate": 1.541338773924026e-05,
2143
+ "loss": 0.0,
2144
+ "step": 35600
2145
+ },
2146
+ {
2147
+ "epoch": 0.69,
2148
+ "learning_rate": 1.5316234334013406e-05,
2149
+ "loss": 0.0,
2150
+ "step": 35700
2151
+ },
2152
+ {
2153
+ "epoch": 0.7,
2154
+ "learning_rate": 1.5219080928786555e-05,
2155
+ "loss": 0.0,
2156
+ "step": 35800
2157
+ },
2158
+ {
2159
+ "epoch": 0.7,
2160
+ "learning_rate": 1.5121927523559702e-05,
2161
+ "loss": 0.0,
2162
+ "step": 35900
2163
+ },
2164
+ {
2165
+ "epoch": 0.7,
2166
+ "learning_rate": 1.5024774118332848e-05,
2167
+ "loss": 0.0,
2168
+ "step": 36000
2169
+ },
2170
+ {
2171
+ "epoch": 0.7,
2172
+ "learning_rate": 1.4927620713105994e-05,
2173
+ "loss": 0.0,
2174
+ "step": 36100
2175
+ },
2176
+ {
2177
+ "epoch": 0.7,
2178
+ "learning_rate": 1.4830467307879142e-05,
2179
+ "loss": 0.0,
2180
+ "step": 36200
2181
+ },
2182
+ {
2183
+ "epoch": 0.71,
2184
+ "learning_rate": 1.4733313902652287e-05,
2185
+ "loss": 0.0,
2186
+ "step": 36300
2187
+ },
2188
+ {
2189
+ "epoch": 0.71,
2190
+ "learning_rate": 1.4636160497425436e-05,
2191
+ "loss": 0.0,
2192
+ "step": 36400
2193
+ },
2194
+ {
2195
+ "epoch": 0.71,
2196
+ "learning_rate": 1.4539007092198581e-05,
2197
+ "loss": 0.0,
2198
+ "step": 36500
2199
+ },
2200
+ {
2201
+ "epoch": 0.71,
2202
+ "learning_rate": 1.444185368697173e-05,
2203
+ "loss": 0.0,
2204
+ "step": 36600
2205
+ },
2206
+ {
2207
+ "epoch": 0.71,
2208
+ "learning_rate": 1.4344700281744875e-05,
2209
+ "loss": 0.0,
2210
+ "step": 36700
2211
+ },
2212
+ {
2213
+ "epoch": 0.72,
2214
+ "learning_rate": 1.4247546876518023e-05,
2215
+ "loss": 0.0,
2216
+ "step": 36800
2217
+ },
2218
+ {
2219
+ "epoch": 0.72,
2220
+ "learning_rate": 1.415039347129117e-05,
2221
+ "loss": 0.0,
2222
+ "step": 36900
2223
+ },
2224
+ {
2225
+ "epoch": 0.72,
2226
+ "learning_rate": 1.4053240066064317e-05,
2227
+ "loss": 0.0,
2228
+ "step": 37000
2229
+ },
2230
+ {
2231
+ "epoch": 0.72,
2232
+ "learning_rate": 1.3956086660837464e-05,
2233
+ "loss": 0.0,
2234
+ "step": 37100
2235
+ },
2236
+ {
2237
+ "epoch": 0.72,
2238
+ "learning_rate": 1.3858933255610609e-05,
2239
+ "loss": 0.0,
2240
+ "step": 37200
2241
+ },
2242
+ {
2243
+ "epoch": 0.72,
2244
+ "learning_rate": 1.3761779850383757e-05,
2245
+ "loss": 0.0,
2246
+ "step": 37300
2247
+ },
2248
+ {
2249
+ "epoch": 0.73,
2250
+ "learning_rate": 1.3664626445156902e-05,
2251
+ "loss": 0.0,
2252
+ "step": 37400
2253
+ },
2254
+ {
2255
+ "epoch": 0.73,
2256
+ "learning_rate": 1.3567473039930051e-05,
2257
+ "loss": 0.0,
2258
+ "step": 37500
2259
+ },
2260
+ {
2261
+ "epoch": 0.73,
2262
+ "learning_rate": 1.3470319634703196e-05,
2263
+ "loss": 0.0,
2264
+ "step": 37600
2265
+ },
2266
+ {
2267
+ "epoch": 0.73,
2268
+ "learning_rate": 1.3373166229476345e-05,
2269
+ "loss": 0.0,
2270
+ "step": 37700
2271
+ },
2272
+ {
2273
+ "epoch": 0.73,
2274
+ "learning_rate": 1.327601282424949e-05,
2275
+ "loss": 0.0,
2276
+ "step": 37800
2277
+ },
2278
+ {
2279
+ "epoch": 0.74,
2280
+ "learning_rate": 1.3178859419022638e-05,
2281
+ "loss": 0.0,
2282
+ "step": 37900
2283
+ },
2284
+ {
2285
+ "epoch": 0.74,
2286
+ "learning_rate": 1.3081706013795783e-05,
2287
+ "loss": 0.0,
2288
+ "step": 38000
2289
+ },
2290
+ {
2291
+ "epoch": 0.74,
2292
+ "learning_rate": 1.2984552608568932e-05,
2293
+ "loss": 0.0,
2294
+ "step": 38100
2295
+ },
2296
+ {
2297
+ "epoch": 0.74,
2298
+ "learning_rate": 1.2887399203342077e-05,
2299
+ "loss": 0.0,
2300
+ "step": 38200
2301
+ },
2302
+ {
2303
+ "epoch": 0.74,
2304
+ "learning_rate": 1.2790245798115224e-05,
2305
+ "loss": 0.0,
2306
+ "step": 38300
2307
+ },
2308
+ {
2309
+ "epoch": 0.75,
2310
+ "learning_rate": 1.2693092392888372e-05,
2311
+ "loss": 0.0,
2312
+ "step": 38400
2313
+ },
2314
+ {
2315
+ "epoch": 0.75,
2316
+ "learning_rate": 1.2595938987661518e-05,
2317
+ "loss": 0.0,
2318
+ "step": 38500
2319
+ },
2320
+ {
2321
+ "epoch": 0.75,
2322
+ "learning_rate": 1.2498785582434664e-05,
2323
+ "loss": 0.0,
2324
+ "step": 38600
2325
+ },
2326
+ {
2327
+ "epoch": 0.75,
2328
+ "learning_rate": 1.2401632177207811e-05,
2329
+ "loss": 0.0,
2330
+ "step": 38700
2331
+ },
2332
+ {
2333
+ "epoch": 0.75,
2334
+ "learning_rate": 1.2304478771980958e-05,
2335
+ "loss": 0.0,
2336
+ "step": 38800
2337
+ },
2338
+ {
2339
+ "epoch": 0.76,
2340
+ "learning_rate": 1.2207325366754105e-05,
2341
+ "loss": 0.0,
2342
+ "step": 38900
2343
+ },
2344
+ {
2345
+ "epoch": 0.76,
2346
+ "learning_rate": 1.2110171961527252e-05,
2347
+ "loss": 0.0,
2348
+ "step": 39000
2349
+ },
2350
+ {
2351
+ "epoch": 0.76,
2352
+ "learning_rate": 1.2013018556300399e-05,
2353
+ "loss": 0.0,
2354
+ "step": 39100
2355
+ },
2356
+ {
2357
+ "epoch": 0.76,
2358
+ "learning_rate": 1.1915865151073545e-05,
2359
+ "loss": 0.0,
2360
+ "step": 39200
2361
+ },
2362
+ {
2363
+ "epoch": 0.76,
2364
+ "learning_rate": 1.1818711745846694e-05,
2365
+ "loss": 0.0,
2366
+ "step": 39300
2367
+ },
2368
+ {
2369
+ "epoch": 0.77,
2370
+ "learning_rate": 1.1721558340619839e-05,
2371
+ "loss": 0.0,
2372
+ "step": 39400
2373
+ },
2374
+ {
2375
+ "epoch": 0.77,
2376
+ "learning_rate": 1.1624404935392986e-05,
2377
+ "loss": 0.0,
2378
+ "step": 39500
2379
+ },
2380
+ {
2381
+ "epoch": 0.77,
2382
+ "learning_rate": 1.1527251530166133e-05,
2383
+ "loss": 0.0,
2384
+ "step": 39600
2385
+ },
2386
+ {
2387
+ "epoch": 0.77,
2388
+ "learning_rate": 1.143009812493928e-05,
2389
+ "loss": 0.0,
2390
+ "step": 39700
2391
+ },
2392
+ {
2393
+ "epoch": 0.77,
2394
+ "learning_rate": 1.1332944719712426e-05,
2395
+ "loss": 0.0,
2396
+ "step": 39800
2397
+ },
2398
+ {
2399
+ "epoch": 0.78,
2400
+ "learning_rate": 1.1235791314485573e-05,
2401
+ "loss": 0.0,
2402
+ "step": 39900
2403
+ },
2404
+ {
2405
+ "epoch": 0.78,
2406
+ "learning_rate": 1.113863790925872e-05,
2407
+ "loss": 0.0,
2408
+ "step": 40000
2409
+ },
2410
+ {
2411
+ "epoch": 0.78,
2412
+ "learning_rate": 1.1041484504031867e-05,
2413
+ "loss": 0.0,
2414
+ "step": 40100
2415
+ },
2416
+ {
2417
+ "epoch": 0.78,
2418
+ "learning_rate": 1.0944331098805014e-05,
2419
+ "loss": 0.0,
2420
+ "step": 40200
2421
+ },
2422
+ {
2423
+ "epoch": 0.78,
2424
+ "learning_rate": 1.084717769357816e-05,
2425
+ "loss": 0.0,
2426
+ "step": 40300
2427
+ },
2428
+ {
2429
+ "epoch": 0.78,
2430
+ "learning_rate": 1.0750024288351307e-05,
2431
+ "loss": 0.0,
2432
+ "step": 40400
2433
+ },
2434
+ {
2435
+ "epoch": 0.79,
2436
+ "learning_rate": 1.0652870883124454e-05,
2437
+ "loss": 0.0,
2438
+ "step": 40500
2439
+ },
2440
+ {
2441
+ "epoch": 0.79,
2442
+ "learning_rate": 1.0555717477897601e-05,
2443
+ "loss": 0.0,
2444
+ "step": 40600
2445
+ },
2446
+ {
2447
+ "epoch": 0.79,
2448
+ "learning_rate": 1.0458564072670748e-05,
2449
+ "loss": 0.0,
2450
+ "step": 40700
2451
+ },
2452
+ {
2453
+ "epoch": 0.79,
2454
+ "learning_rate": 1.0361410667443893e-05,
2455
+ "loss": 0.0,
2456
+ "step": 40800
2457
+ },
2458
+ {
2459
+ "epoch": 0.79,
2460
+ "learning_rate": 1.0264257262217042e-05,
2461
+ "loss": 0.0,
2462
+ "step": 40900
2463
+ },
2464
+ {
2465
+ "epoch": 0.8,
2466
+ "learning_rate": 1.0167103856990188e-05,
2467
+ "loss": 0.0,
2468
+ "step": 41000
2469
+ },
2470
+ {
2471
+ "epoch": 0.8,
2472
+ "learning_rate": 1.0069950451763335e-05,
2473
+ "loss": 0.0,
2474
+ "step": 41100
2475
+ },
2476
+ {
2477
+ "epoch": 0.8,
2478
+ "learning_rate": 9.972797046536482e-06,
2479
+ "loss": 0.0,
2480
+ "step": 41200
2481
+ },
2482
+ {
2483
+ "epoch": 0.8,
2484
+ "learning_rate": 9.875643641309629e-06,
2485
+ "loss": 0.0,
2486
+ "step": 41300
2487
+ },
2488
+ {
2489
+ "epoch": 0.8,
2490
+ "learning_rate": 9.778490236082776e-06,
2491
+ "loss": 0.0,
2492
+ "step": 41400
2493
+ },
2494
+ {
2495
+ "epoch": 0.81,
2496
+ "learning_rate": 9.681336830855922e-06,
2497
+ "loss": 0.0,
2498
+ "step": 41500
2499
+ },
2500
+ {
2501
+ "epoch": 0.81,
2502
+ "learning_rate": 9.58418342562907e-06,
2503
+ "loss": 0.0,
2504
+ "step": 41600
2505
+ },
2506
+ {
2507
+ "epoch": 0.81,
2508
+ "learning_rate": 9.487030020402216e-06,
2509
+ "loss": 0.0,
2510
+ "step": 41700
2511
+ },
2512
+ {
2513
+ "epoch": 0.81,
2514
+ "learning_rate": 9.389876615175363e-06,
2515
+ "loss": 0.0,
2516
+ "step": 41800
2517
+ },
2518
+ {
2519
+ "epoch": 0.81,
2520
+ "learning_rate": 9.29272320994851e-06,
2521
+ "loss": 0.0,
2522
+ "step": 41900
2523
+ },
2524
+ {
2525
+ "epoch": 0.82,
2526
+ "learning_rate": 9.195569804721655e-06,
2527
+ "loss": 0.0,
2528
+ "step": 42000
2529
+ },
2530
+ {
2531
+ "epoch": 0.82,
2532
+ "learning_rate": 9.098416399494802e-06,
2533
+ "loss": 0.0,
2534
+ "step": 42100
2535
+ },
2536
+ {
2537
+ "epoch": 0.82,
2538
+ "learning_rate": 9.001262994267949e-06,
2539
+ "loss": 0.0,
2540
+ "step": 42200
2541
+ },
2542
+ {
2543
+ "epoch": 0.82,
2544
+ "learning_rate": 8.904109589041095e-06,
2545
+ "loss": 0.0,
2546
+ "step": 42300
2547
+ },
2548
+ {
2549
+ "epoch": 0.82,
2550
+ "learning_rate": 8.806956183814242e-06,
2551
+ "loss": 0.0,
2552
+ "step": 42400
2553
+ },
2554
+ {
2555
+ "epoch": 0.83,
2556
+ "learning_rate": 8.709802778587389e-06,
2557
+ "loss": 0.0,
2558
+ "step": 42500
2559
+ },
2560
+ {
2561
+ "epoch": 0.83,
2562
+ "learning_rate": 8.612649373360538e-06,
2563
+ "loss": 0.0,
2564
+ "step": 42600
2565
+ },
2566
+ {
2567
+ "epoch": 0.83,
2568
+ "learning_rate": 8.515495968133684e-06,
2569
+ "loss": 0.0,
2570
+ "step": 42700
2571
+ },
2572
+ {
2573
+ "epoch": 0.83,
2574
+ "learning_rate": 8.418342562906831e-06,
2575
+ "loss": 0.0,
2576
+ "step": 42800
2577
+ },
2578
+ {
2579
+ "epoch": 0.83,
2580
+ "learning_rate": 8.321189157679978e-06,
2581
+ "loss": 0.0,
2582
+ "step": 42900
2583
+ },
2584
+ {
2585
+ "epoch": 0.84,
2586
+ "learning_rate": 8.224035752453125e-06,
2587
+ "loss": 0.0,
2588
+ "step": 43000
2589
+ },
2590
+ {
2591
+ "epoch": 0.84,
2592
+ "learning_rate": 8.12688234722627e-06,
2593
+ "loss": 0.0,
2594
+ "step": 43100
2595
+ },
2596
+ {
2597
+ "epoch": 0.84,
2598
+ "learning_rate": 8.029728941999417e-06,
2599
+ "loss": 0.0,
2600
+ "step": 43200
2601
+ },
2602
+ {
2603
+ "epoch": 0.84,
2604
+ "learning_rate": 7.932575536772564e-06,
2605
+ "loss": 0.0,
2606
+ "step": 43300
2607
+ },
2608
+ {
2609
+ "epoch": 0.84,
2610
+ "learning_rate": 7.83542213154571e-06,
2611
+ "loss": 0.0,
2612
+ "step": 43400
2613
+ },
2614
+ {
2615
+ "epoch": 0.85,
2616
+ "learning_rate": 7.738268726318857e-06,
2617
+ "loss": 0.0,
2618
+ "step": 43500
2619
+ },
2620
+ {
2621
+ "epoch": 0.85,
2622
+ "learning_rate": 7.641115321092004e-06,
2623
+ "loss": 0.0,
2624
+ "step": 43600
2625
+ },
2626
+ {
2627
+ "epoch": 0.85,
2628
+ "learning_rate": 7.543961915865151e-06,
2629
+ "loss": 0.0,
2630
+ "step": 43700
2631
+ },
2632
+ {
2633
+ "epoch": 0.85,
2634
+ "learning_rate": 7.446808510638298e-06,
2635
+ "loss": 0.0,
2636
+ "step": 43800
2637
+ },
2638
+ {
2639
+ "epoch": 0.85,
2640
+ "learning_rate": 7.349655105411445e-06,
2641
+ "loss": 0.0,
2642
+ "step": 43900
2643
+ },
2644
+ {
2645
+ "epoch": 0.85,
2646
+ "learning_rate": 7.252501700184592e-06,
2647
+ "loss": 0.0,
2648
+ "step": 44000
2649
+ },
2650
+ {
2651
+ "epoch": 0.86,
2652
+ "learning_rate": 7.155348294957738e-06,
2653
+ "loss": 0.0,
2654
+ "step": 44100
2655
+ },
2656
+ {
2657
+ "epoch": 0.86,
2658
+ "learning_rate": 7.058194889730884e-06,
2659
+ "loss": 0.0,
2660
+ "step": 44200
2661
+ },
2662
+ {
2663
+ "epoch": 0.86,
2664
+ "learning_rate": 6.961041484504033e-06,
2665
+ "loss": 0.0,
2666
+ "step": 44300
2667
+ },
2668
+ {
2669
+ "epoch": 0.86,
2670
+ "learning_rate": 6.86388807927718e-06,
2671
+ "loss": 0.0,
2672
+ "step": 44400
2673
+ },
2674
+ {
2675
+ "epoch": 0.86,
2676
+ "learning_rate": 6.766734674050327e-06,
2677
+ "loss": 0.0,
2678
+ "step": 44500
2679
+ },
2680
+ {
2681
+ "epoch": 0.87,
2682
+ "learning_rate": 6.669581268823473e-06,
2683
+ "loss": 0.0,
2684
+ "step": 44600
2685
+ },
2686
+ {
2687
+ "epoch": 0.87,
2688
+ "learning_rate": 6.5724278635966194e-06,
2689
+ "loss": 0.0,
2690
+ "step": 44700
2691
+ },
2692
+ {
2693
+ "epoch": 0.87,
2694
+ "learning_rate": 6.475274458369766e-06,
2695
+ "loss": 0.0,
2696
+ "step": 44800
2697
+ },
2698
+ {
2699
+ "epoch": 0.87,
2700
+ "learning_rate": 6.378121053142913e-06,
2701
+ "loss": 0.0,
2702
+ "step": 44900
2703
+ },
2704
+ {
2705
+ "epoch": 0.87,
2706
+ "learning_rate": 6.28096764791606e-06,
2707
+ "loss": 0.0,
2708
+ "step": 45000
2709
+ },
2710
+ {
2711
+ "epoch": 0.88,
2712
+ "learning_rate": 6.183814242689207e-06,
2713
+ "loss": 0.0,
2714
+ "step": 45100
2715
+ },
2716
+ {
2717
+ "epoch": 0.88,
2718
+ "learning_rate": 6.086660837462354e-06,
2719
+ "loss": 0.0,
2720
+ "step": 45200
2721
+ },
2722
+ {
2723
+ "epoch": 0.88,
2724
+ "learning_rate": 5.9895074322354996e-06,
2725
+ "loss": 0.0,
2726
+ "step": 45300
2727
+ },
2728
+ {
2729
+ "epoch": 0.88,
2730
+ "learning_rate": 5.892354027008646e-06,
2731
+ "loss": 0.0,
2732
+ "step": 45400
2733
+ },
2734
+ {
2735
+ "epoch": 0.88,
2736
+ "learning_rate": 5.795200621781794e-06,
2737
+ "loss": 0.0,
2738
+ "step": 45500
2739
+ },
2740
+ {
2741
+ "epoch": 0.89,
2742
+ "learning_rate": 5.698047216554941e-06,
2743
+ "loss": 0.0,
2744
+ "step": 45600
2745
+ },
2746
+ {
2747
+ "epoch": 0.89,
2748
+ "learning_rate": 5.600893811328088e-06,
2749
+ "loss": 0.0,
2750
+ "step": 45700
2751
+ },
2752
+ {
2753
+ "epoch": 0.89,
2754
+ "learning_rate": 5.503740406101235e-06,
2755
+ "loss": 0.0,
2756
+ "step": 45800
2757
+ },
2758
+ {
2759
+ "epoch": 0.89,
2760
+ "learning_rate": 5.4065870008743806e-06,
2761
+ "loss": 0.0,
2762
+ "step": 45900
2763
+ },
2764
+ {
2765
+ "epoch": 0.89,
2766
+ "learning_rate": 5.309433595647527e-06,
2767
+ "loss": 0.0,
2768
+ "step": 46000
2769
+ },
2770
+ {
2771
+ "epoch": 0.9,
2772
+ "learning_rate": 5.212280190420674e-06,
2773
+ "loss": 0.0,
2774
+ "step": 46100
2775
+ },
2776
+ {
2777
+ "epoch": 0.9,
2778
+ "learning_rate": 5.115126785193821e-06,
2779
+ "loss": 0.0,
2780
+ "step": 46200
2781
+ },
2782
+ {
2783
+ "epoch": 0.9,
2784
+ "learning_rate": 5.017973379966969e-06,
2785
+ "loss": 0.0,
2786
+ "step": 46300
2787
+ },
2788
+ {
2789
+ "epoch": 0.9,
2790
+ "learning_rate": 4.920819974740115e-06,
2791
+ "loss": 0.0,
2792
+ "step": 46400
2793
+ },
2794
+ {
2795
+ "epoch": 0.9,
2796
+ "learning_rate": 4.8236665695132616e-06,
2797
+ "loss": 0.0,
2798
+ "step": 46500
2799
+ },
2800
+ {
2801
+ "epoch": 0.91,
2802
+ "learning_rate": 4.726513164286408e-06,
2803
+ "loss": 0.0,
2804
+ "step": 46600
2805
+ },
2806
+ {
2807
+ "epoch": 0.91,
2808
+ "learning_rate": 4.629359759059555e-06,
2809
+ "loss": 0.0,
2810
+ "step": 46700
2811
+ },
2812
+ {
2813
+ "epoch": 0.91,
2814
+ "learning_rate": 4.532206353832702e-06,
2815
+ "loss": 0.0,
2816
+ "step": 46800
2817
+ },
2818
+ {
2819
+ "epoch": 0.91,
2820
+ "learning_rate": 4.435052948605849e-06,
2821
+ "loss": 0.0,
2822
+ "step": 46900
2823
+ },
2824
+ {
2825
+ "epoch": 0.91,
2826
+ "learning_rate": 4.337899543378996e-06,
2827
+ "loss": 0.0,
2828
+ "step": 47000
2829
+ },
2830
+ {
2831
+ "epoch": 0.92,
2832
+ "learning_rate": 4.240746138152142e-06,
2833
+ "loss": 0.0,
2834
+ "step": 47100
2835
+ },
2836
+ {
2837
+ "epoch": 0.92,
2838
+ "learning_rate": 4.143592732925289e-06,
2839
+ "loss": 0.0,
2840
+ "step": 47200
2841
+ },
2842
+ {
2843
+ "epoch": 0.92,
2844
+ "learning_rate": 4.046439327698436e-06,
2845
+ "loss": 0.0,
2846
+ "step": 47300
2847
+ },
2848
+ {
2849
+ "epoch": 0.92,
2850
+ "learning_rate": 3.949285922471583e-06,
2851
+ "loss": 0.0,
2852
+ "step": 47400
2853
+ },
2854
+ {
2855
+ "epoch": 0.92,
2856
+ "learning_rate": 3.85213251724473e-06,
2857
+ "loss": 0.0,
2858
+ "step": 47500
2859
+ },
2860
+ {
2861
+ "epoch": 0.92,
2862
+ "learning_rate": 3.7549791120178763e-06,
2863
+ "loss": 0.0,
2864
+ "step": 47600
2865
+ },
2866
+ {
2867
+ "epoch": 0.93,
2868
+ "learning_rate": 3.657825706791023e-06,
2869
+ "loss": 0.0,
2870
+ "step": 47700
2871
+ },
2872
+ {
2873
+ "epoch": 0.93,
2874
+ "learning_rate": 3.56067230156417e-06,
2875
+ "loss": 0.0,
2876
+ "step": 47800
2877
+ },
2878
+ {
2879
+ "epoch": 0.93,
2880
+ "learning_rate": 3.4635188963373164e-06,
2881
+ "loss": 0.0,
2882
+ "step": 47900
2883
+ },
2884
+ {
2885
+ "epoch": 0.93,
2886
+ "learning_rate": 3.366365491110464e-06,
2887
+ "loss": 0.0,
2888
+ "step": 48000
2889
+ },
2890
+ {
2891
+ "epoch": 0.93,
2892
+ "learning_rate": 3.2692120858836105e-06,
2893
+ "loss": 0.0,
2894
+ "step": 48100
2895
+ },
2896
+ {
2897
+ "epoch": 0.94,
2898
+ "learning_rate": 3.1720586806567573e-06,
2899
+ "loss": 0.0,
2900
+ "step": 48200
2901
+ },
2902
+ {
2903
+ "epoch": 0.94,
2904
+ "learning_rate": 3.074905275429904e-06,
2905
+ "loss": 0.0,
2906
+ "step": 48300
2907
+ },
2908
+ {
2909
+ "epoch": 0.94,
2910
+ "learning_rate": 2.977751870203051e-06,
2911
+ "loss": 0.0,
2912
+ "step": 48400
2913
+ },
2914
+ {
2915
+ "epoch": 0.94,
2916
+ "learning_rate": 2.8805984649761974e-06,
2917
+ "loss": 0.0,
2918
+ "step": 48500
2919
+ },
2920
+ {
2921
+ "epoch": 0.94,
2922
+ "learning_rate": 2.7834450597493446e-06,
2923
+ "loss": 0.0,
2924
+ "step": 48600
2925
+ },
2926
+ {
2927
+ "epoch": 0.95,
2928
+ "learning_rate": 2.686291654522491e-06,
2929
+ "loss": 0.0,
2930
+ "step": 48700
2931
+ },
2932
+ {
2933
+ "epoch": 0.95,
2934
+ "learning_rate": 2.589138249295638e-06,
2935
+ "loss": 0.0,
2936
+ "step": 48800
2937
+ },
2938
+ {
2939
+ "epoch": 0.95,
2940
+ "learning_rate": 2.4919848440687847e-06,
2941
+ "loss": 0.0,
2942
+ "step": 48900
2943
+ },
2944
+ {
2945
+ "epoch": 0.95,
2946
+ "learning_rate": 2.3948314388419315e-06,
2947
+ "loss": 0.0,
2948
+ "step": 49000
2949
+ },
2950
+ {
2951
+ "epoch": 0.95,
2952
+ "learning_rate": 2.2976780336150784e-06,
2953
+ "loss": 0.0,
2954
+ "step": 49100
2955
+ },
2956
+ {
2957
+ "epoch": 0.96,
2958
+ "learning_rate": 2.200524628388225e-06,
2959
+ "loss": 0.0,
2960
+ "step": 49200
2961
+ },
2962
+ {
2963
+ "epoch": 0.96,
2964
+ "learning_rate": 2.103371223161372e-06,
2965
+ "loss": 0.0,
2966
+ "step": 49300
2967
+ },
2968
+ {
2969
+ "epoch": 0.96,
2970
+ "learning_rate": 2.0062178179345184e-06,
2971
+ "loss": 0.0,
2972
+ "step": 49400
2973
+ },
2974
+ {
2975
+ "epoch": 0.96,
2976
+ "learning_rate": 1.9090644127076657e-06,
2977
+ "loss": 0.0,
2978
+ "step": 49500
2979
+ },
2980
+ {
2981
+ "epoch": 0.96,
2982
+ "learning_rate": 1.8119110074808123e-06,
2983
+ "loss": 0.0,
2984
+ "step": 49600
2985
+ },
2986
+ {
2987
+ "epoch": 0.97,
2988
+ "learning_rate": 1.714757602253959e-06,
2989
+ "loss": 0.0,
2990
+ "step": 49700
2991
+ },
2992
+ {
2993
+ "epoch": 0.97,
2994
+ "learning_rate": 1.6176041970271058e-06,
2995
+ "loss": 0.0,
2996
+ "step": 49800
2997
+ },
2998
+ {
2999
+ "epoch": 0.97,
3000
+ "learning_rate": 1.5204507918002526e-06,
3001
+ "loss": 0.0,
3002
+ "step": 49900
3003
+ },
3004
+ {
3005
+ "epoch": 0.97,
3006
+ "learning_rate": 1.4232973865733994e-06,
3007
+ "loss": 0.0,
3008
+ "step": 50000
3009
+ },
3010
+ {
3011
+ "epoch": 0.97,
3012
+ "learning_rate": 1.3261439813465462e-06,
3013
+ "loss": 0.0,
3014
+ "step": 50100
3015
+ },
3016
+ {
3017
+ "epoch": 0.98,
3018
+ "learning_rate": 1.228990576119693e-06,
3019
+ "loss": 0.0,
3020
+ "step": 50200
3021
+ },
3022
+ {
3023
+ "epoch": 0.98,
3024
+ "learning_rate": 1.13183717089284e-06,
3025
+ "loss": 0.0,
3026
+ "step": 50300
3027
+ },
3028
+ {
3029
+ "epoch": 0.98,
3030
+ "learning_rate": 1.0346837656659867e-06,
3031
+ "loss": 0.0,
3032
+ "step": 50400
3033
+ },
3034
+ {
3035
+ "epoch": 0.98,
3036
+ "learning_rate": 9.375303604391334e-07,
3037
+ "loss": 0.0,
3038
+ "step": 50500
3039
+ },
3040
+ {
3041
+ "epoch": 0.98,
3042
+ "learning_rate": 8.403769552122803e-07,
3043
+ "loss": 0.0,
3044
+ "step": 50600
3045
+ },
3046
+ {
3047
+ "epoch": 0.99,
3048
+ "learning_rate": 7.43223549985427e-07,
3049
+ "loss": 0.0,
3050
+ "step": 50700
3051
+ },
3052
+ {
3053
+ "epoch": 0.99,
3054
+ "learning_rate": 6.460701447585738e-07,
3055
+ "loss": 0.0,
3056
+ "step": 50800
3057
+ },
3058
+ {
3059
+ "epoch": 0.99,
3060
+ "learning_rate": 5.489167395317206e-07,
3061
+ "loss": 0.0,
3062
+ "step": 50900
3063
+ },
3064
+ {
3065
+ "epoch": 0.99,
3066
+ "learning_rate": 4.517633343048674e-07,
3067
+ "loss": 0.0,
3068
+ "step": 51000
3069
+ },
3070
+ {
3071
+ "epoch": 0.99,
3072
+ "learning_rate": 3.546099290780142e-07,
3073
+ "loss": 0.0,
3074
+ "step": 51100
3075
+ },
3076
+ {
3077
+ "epoch": 0.99,
3078
+ "learning_rate": 2.57456523851161e-07,
3079
+ "loss": 0.0,
3080
+ "step": 51200
3081
+ },
3082
+ {
3083
+ "epoch": 1.0,
3084
+ "learning_rate": 1.6030311862430779e-07,
3085
+ "loss": 0.0,
3086
+ "step": 51300
3087
+ },
3088
+ {
3089
+ "epoch": 1.0,
3090
+ "learning_rate": 6.314971339745459e-08,
3091
+ "loss": 0.0,
3092
+ "step": 51400
3093
+ }
3094
+ ],
3095
+ "max_steps": 51465,
3096
+ "num_train_epochs": 1,
3097
+ "total_flos": 8.545904720294707e+16,
3098
+ "trial_name": null,
3099
+ "trial_params": null
3100
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d071a74a7be06063e9051e3b4b76ca2e88f3ebea1cd21081ec84f7f9789dae42
3
+ size 3311