hyungjoochae commited on
Commit
03cdc1d
1 Parent(s): e4dddaa

Upload folder using huggingface_hub

Browse files
config.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/opt-1.3b",
3
+ "_remove_final_layer_norm": false,
4
+ "activation_dropout": 0.0,
5
+ "activation_function": "relu",
6
+ "architectures": [
7
+ "OPTForCausalLM"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "bos_token_id": 2,
11
+ "do_layer_norm_before": true,
12
+ "dropout": 0.1,
13
+ "enable_bias": true,
14
+ "eos_token_id": 2,
15
+ "ffn_dim": 8192,
16
+ "hidden_size": 2048,
17
+ "init_std": 0.02,
18
+ "layer_norm_elementwise_affine": true,
19
+ "layerdrop": 0.0,
20
+ "max_position_embeddings": 2048,
21
+ "model_type": "opt",
22
+ "num_attention_heads": 32,
23
+ "num_hidden_layers": 24,
24
+ "pad_token_id": 1,
25
+ "prefix": "</s>",
26
+ "torch_dtype": "float32",
27
+ "transformers_version": "4.30.0.dev0",
28
+ "use_cache": false,
29
+ "vocab_size": 50272,
30
+ "word_embed_proj_dim": 2048
31
+ }
generation_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_new_tokens": 256,
3
+ "transformers_version": "4.30.0.dev0"
4
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e87abbebb796c73b0b81608ee1d649136834f65024f7c90b6c66e5848fed7e91
3
+ size 5263162257
special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "</s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<pad>",
18
+ "lstrip": false,
19
+ "normalized": true,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "</s>",
25
+ "lstrip": false,
26
+ "normalized": true,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_prefix_space": false,
4
+ "bos_token": {
5
+ "__type": "AddedToken",
6
+ "content": "</s>",
7
+ "lstrip": false,
8
+ "normalized": true,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "clean_up_tokenization_spaces": true,
13
+ "eos_token": {
14
+ "__type": "AddedToken",
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "errors": "replace",
22
+ "model_max_length": 1000000000000000019884624838656,
23
+ "pad_token": {
24
+ "__type": "AddedToken",
25
+ "content": "<pad>",
26
+ "lstrip": false,
27
+ "normalized": true,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ },
31
+ "padding_side": "right",
32
+ "tokenizer_class": "GPT2Tokenizer",
33
+ "unk_token": {
34
+ "__type": "AddedToken",
35
+ "content": "</s>",
36
+ "lstrip": false,
37
+ "normalized": true,
38
+ "rstrip": false,
39
+ "single_word": false
40
+ }
41
+ }
trainer_state.json ADDED
@@ -0,0 +1,2112 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 4.0,
5
+ "global_step": 3448,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 0.0002,
13
+ "loss": 4.4935,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 0.0002,
19
+ "loss": 1.576,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 0.0002,
25
+ "loss": 0.9832,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.05,
30
+ "learning_rate": 0.0002,
31
+ "loss": 0.8691,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.06,
36
+ "learning_rate": 0.0002,
37
+ "loss": 0.7926,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.07,
42
+ "learning_rate": 0.0002,
43
+ "loss": 0.795,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.08,
48
+ "learning_rate": 0.0002,
49
+ "loss": 0.7327,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.09,
54
+ "learning_rate": 0.0002,
55
+ "loss": 0.7195,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.1,
60
+ "learning_rate": 0.0002,
61
+ "loss": 0.7307,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.12,
66
+ "learning_rate": 0.0002,
67
+ "loss": 0.6724,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.13,
72
+ "learning_rate": 0.0002,
73
+ "loss": 0.6968,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.14,
78
+ "learning_rate": 0.0002,
79
+ "loss": 0.7335,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.15,
84
+ "learning_rate": 0.0002,
85
+ "loss": 0.7235,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.16,
90
+ "learning_rate": 0.0002,
91
+ "loss": 0.6764,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.17,
96
+ "learning_rate": 0.0002,
97
+ "loss": 0.6414,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.19,
102
+ "learning_rate": 0.0002,
103
+ "loss": 0.6739,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.2,
108
+ "learning_rate": 0.0002,
109
+ "loss": 0.6421,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.21,
114
+ "learning_rate": 0.0002,
115
+ "loss": 0.6436,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.22,
120
+ "learning_rate": 0.0002,
121
+ "loss": 0.635,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 0.23,
126
+ "learning_rate": 0.0002,
127
+ "loss": 0.6113,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 0.24,
132
+ "learning_rate": 0.0002,
133
+ "loss": 0.6381,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 0.26,
138
+ "learning_rate": 0.0002,
139
+ "loss": 0.6109,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 0.27,
144
+ "learning_rate": 0.0002,
145
+ "loss": 0.614,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 0.28,
150
+ "learning_rate": 0.0002,
151
+ "loss": 0.6144,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 0.29,
156
+ "learning_rate": 0.0002,
157
+ "loss": 0.5815,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 0.3,
162
+ "learning_rate": 0.0002,
163
+ "loss": 0.6123,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 0.31,
168
+ "learning_rate": 0.0002,
169
+ "loss": 0.5971,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 0.32,
174
+ "learning_rate": 0.0002,
175
+ "loss": 0.5952,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 0.34,
180
+ "learning_rate": 0.0002,
181
+ "loss": 0.6136,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 0.35,
186
+ "learning_rate": 0.0002,
187
+ "loss": 0.5892,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 0.36,
192
+ "learning_rate": 0.0002,
193
+ "loss": 0.6023,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 0.37,
198
+ "learning_rate": 0.0002,
199
+ "loss": 0.5737,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 0.38,
204
+ "learning_rate": 0.0002,
205
+ "loss": 0.578,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 0.39,
210
+ "learning_rate": 0.0002,
211
+ "loss": 0.5737,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 0.41,
216
+ "learning_rate": 0.0002,
217
+ "loss": 0.5561,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 0.42,
222
+ "learning_rate": 0.0002,
223
+ "loss": 0.5794,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 0.43,
228
+ "learning_rate": 0.0002,
229
+ "loss": 0.5677,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 0.44,
234
+ "learning_rate": 0.0002,
235
+ "loss": 0.5603,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 0.45,
240
+ "learning_rate": 0.0002,
241
+ "loss": 0.5622,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 0.46,
246
+ "learning_rate": 0.0002,
247
+ "loss": 0.544,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 0.48,
252
+ "learning_rate": 0.0002,
253
+ "loss": 0.5675,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 0.49,
258
+ "learning_rate": 0.0002,
259
+ "loss": 0.5656,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 0.5,
264
+ "learning_rate": 0.0002,
265
+ "loss": 0.5494,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 0.51,
270
+ "learning_rate": 0.0002,
271
+ "loss": 0.5525,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 0.52,
276
+ "learning_rate": 0.0002,
277
+ "loss": 0.5325,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 0.53,
282
+ "learning_rate": 0.0002,
283
+ "loss": 0.5571,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 0.55,
288
+ "learning_rate": 0.0002,
289
+ "loss": 0.5422,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 0.56,
294
+ "learning_rate": 0.0002,
295
+ "loss": 0.5415,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 0.57,
300
+ "learning_rate": 0.0002,
301
+ "loss": 0.5473,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 0.58,
306
+ "learning_rate": 0.0002,
307
+ "loss": 0.532,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 0.59,
312
+ "learning_rate": 0.0002,
313
+ "loss": 0.5466,
314
+ "step": 510
315
+ },
316
+ {
317
+ "epoch": 0.6,
318
+ "learning_rate": 0.0002,
319
+ "loss": 0.5431,
320
+ "step": 520
321
+ },
322
+ {
323
+ "epoch": 0.61,
324
+ "learning_rate": 0.0002,
325
+ "loss": 0.5358,
326
+ "step": 530
327
+ },
328
+ {
329
+ "epoch": 0.63,
330
+ "learning_rate": 0.0002,
331
+ "loss": 0.5304,
332
+ "step": 540
333
+ },
334
+ {
335
+ "epoch": 0.64,
336
+ "learning_rate": 0.0002,
337
+ "loss": 0.5142,
338
+ "step": 550
339
+ },
340
+ {
341
+ "epoch": 0.65,
342
+ "learning_rate": 0.0002,
343
+ "loss": 0.5426,
344
+ "step": 560
345
+ },
346
+ {
347
+ "epoch": 0.66,
348
+ "learning_rate": 0.0002,
349
+ "loss": 0.5303,
350
+ "step": 570
351
+ },
352
+ {
353
+ "epoch": 0.67,
354
+ "learning_rate": 0.0002,
355
+ "loss": 0.5285,
356
+ "step": 580
357
+ },
358
+ {
359
+ "epoch": 0.68,
360
+ "learning_rate": 0.0002,
361
+ "loss": 0.5239,
362
+ "step": 590
363
+ },
364
+ {
365
+ "epoch": 0.7,
366
+ "learning_rate": 0.0002,
367
+ "loss": 0.5144,
368
+ "step": 600
369
+ },
370
+ {
371
+ "epoch": 0.71,
372
+ "learning_rate": 0.0002,
373
+ "loss": 0.5326,
374
+ "step": 610
375
+ },
376
+ {
377
+ "epoch": 0.72,
378
+ "learning_rate": 0.0002,
379
+ "loss": 0.5147,
380
+ "step": 620
381
+ },
382
+ {
383
+ "epoch": 0.73,
384
+ "learning_rate": 0.0002,
385
+ "loss": 0.5238,
386
+ "step": 630
387
+ },
388
+ {
389
+ "epoch": 0.74,
390
+ "learning_rate": 0.0002,
391
+ "loss": 0.5328,
392
+ "step": 640
393
+ },
394
+ {
395
+ "epoch": 0.75,
396
+ "learning_rate": 0.0002,
397
+ "loss": 0.5057,
398
+ "step": 650
399
+ },
400
+ {
401
+ "epoch": 0.77,
402
+ "learning_rate": 0.0002,
403
+ "loss": 0.5276,
404
+ "step": 660
405
+ },
406
+ {
407
+ "epoch": 0.78,
408
+ "learning_rate": 0.0002,
409
+ "loss": 0.5204,
410
+ "step": 670
411
+ },
412
+ {
413
+ "epoch": 0.79,
414
+ "learning_rate": 0.0002,
415
+ "loss": 0.52,
416
+ "step": 680
417
+ },
418
+ {
419
+ "epoch": 0.8,
420
+ "learning_rate": 0.0002,
421
+ "loss": 0.5197,
422
+ "step": 690
423
+ },
424
+ {
425
+ "epoch": 0.81,
426
+ "learning_rate": 0.0002,
427
+ "loss": 0.5032,
428
+ "step": 700
429
+ },
430
+ {
431
+ "epoch": 0.82,
432
+ "learning_rate": 0.0002,
433
+ "loss": 0.5275,
434
+ "step": 710
435
+ },
436
+ {
437
+ "epoch": 0.84,
438
+ "learning_rate": 0.0002,
439
+ "loss": 0.5047,
440
+ "step": 720
441
+ },
442
+ {
443
+ "epoch": 0.85,
444
+ "learning_rate": 0.0002,
445
+ "loss": 0.5055,
446
+ "step": 730
447
+ },
448
+ {
449
+ "epoch": 0.86,
450
+ "learning_rate": 0.0002,
451
+ "loss": 0.5216,
452
+ "step": 740
453
+ },
454
+ {
455
+ "epoch": 0.87,
456
+ "learning_rate": 0.0002,
457
+ "loss": 0.5019,
458
+ "step": 750
459
+ },
460
+ {
461
+ "epoch": 0.88,
462
+ "learning_rate": 0.0002,
463
+ "loss": 0.5134,
464
+ "step": 760
465
+ },
466
+ {
467
+ "epoch": 0.89,
468
+ "learning_rate": 0.0002,
469
+ "loss": 0.4997,
470
+ "step": 770
471
+ },
472
+ {
473
+ "epoch": 0.9,
474
+ "learning_rate": 0.0002,
475
+ "loss": 0.4985,
476
+ "step": 780
477
+ },
478
+ {
479
+ "epoch": 0.92,
480
+ "learning_rate": 0.0002,
481
+ "loss": 0.5033,
482
+ "step": 790
483
+ },
484
+ {
485
+ "epoch": 0.93,
486
+ "learning_rate": 0.0002,
487
+ "loss": 0.4919,
488
+ "step": 800
489
+ },
490
+ {
491
+ "epoch": 0.94,
492
+ "learning_rate": 0.0002,
493
+ "loss": 0.5104,
494
+ "step": 810
495
+ },
496
+ {
497
+ "epoch": 0.95,
498
+ "learning_rate": 0.0002,
499
+ "loss": 0.4918,
500
+ "step": 820
501
+ },
502
+ {
503
+ "epoch": 0.96,
504
+ "learning_rate": 0.0002,
505
+ "loss": 0.4877,
506
+ "step": 830
507
+ },
508
+ {
509
+ "epoch": 0.97,
510
+ "learning_rate": 0.0002,
511
+ "loss": 0.5057,
512
+ "step": 840
513
+ },
514
+ {
515
+ "epoch": 0.99,
516
+ "learning_rate": 0.0002,
517
+ "loss": 0.4728,
518
+ "step": 850
519
+ },
520
+ {
521
+ "epoch": 1.0,
522
+ "learning_rate": 0.0002,
523
+ "loss": 0.5035,
524
+ "step": 860
525
+ },
526
+ {
527
+ "epoch": 1.0,
528
+ "eval_loss": 0.4923032224178314,
529
+ "eval_runtime": 5.6257,
530
+ "eval_samples_per_second": 177.756,
531
+ "eval_steps_per_second": 2.844,
532
+ "step": 862
533
+ },
534
+ {
535
+ "epoch": 1.01,
536
+ "learning_rate": 0.0002,
537
+ "loss": 0.4254,
538
+ "step": 870
539
+ },
540
+ {
541
+ "epoch": 1.02,
542
+ "learning_rate": 0.0002,
543
+ "loss": 0.4024,
544
+ "step": 880
545
+ },
546
+ {
547
+ "epoch": 1.03,
548
+ "learning_rate": 0.0002,
549
+ "loss": 0.4108,
550
+ "step": 890
551
+ },
552
+ {
553
+ "epoch": 1.04,
554
+ "learning_rate": 0.0002,
555
+ "loss": 0.3998,
556
+ "step": 900
557
+ },
558
+ {
559
+ "epoch": 1.06,
560
+ "learning_rate": 0.0002,
561
+ "loss": 0.4134,
562
+ "step": 910
563
+ },
564
+ {
565
+ "epoch": 1.07,
566
+ "learning_rate": 0.0002,
567
+ "loss": 0.418,
568
+ "step": 920
569
+ },
570
+ {
571
+ "epoch": 1.08,
572
+ "learning_rate": 0.0002,
573
+ "loss": 0.4121,
574
+ "step": 930
575
+ },
576
+ {
577
+ "epoch": 1.09,
578
+ "learning_rate": 0.0002,
579
+ "loss": 0.4197,
580
+ "step": 940
581
+ },
582
+ {
583
+ "epoch": 1.1,
584
+ "learning_rate": 0.0002,
585
+ "loss": 0.4064,
586
+ "step": 950
587
+ },
588
+ {
589
+ "epoch": 1.11,
590
+ "learning_rate": 0.0002,
591
+ "loss": 0.4144,
592
+ "step": 960
593
+ },
594
+ {
595
+ "epoch": 1.13,
596
+ "learning_rate": 0.0002,
597
+ "loss": 0.4218,
598
+ "step": 970
599
+ },
600
+ {
601
+ "epoch": 1.14,
602
+ "learning_rate": 0.0002,
603
+ "loss": 0.4118,
604
+ "step": 980
605
+ },
606
+ {
607
+ "epoch": 1.15,
608
+ "learning_rate": 0.0002,
609
+ "loss": 0.4216,
610
+ "step": 990
611
+ },
612
+ {
613
+ "epoch": 1.16,
614
+ "learning_rate": 0.0002,
615
+ "loss": 0.4056,
616
+ "step": 1000
617
+ },
618
+ {
619
+ "epoch": 1.17,
620
+ "learning_rate": 0.0002,
621
+ "loss": 0.4157,
622
+ "step": 1010
623
+ },
624
+ {
625
+ "epoch": 1.18,
626
+ "learning_rate": 0.0002,
627
+ "loss": 0.4177,
628
+ "step": 1020
629
+ },
630
+ {
631
+ "epoch": 1.19,
632
+ "learning_rate": 0.0002,
633
+ "loss": 0.4059,
634
+ "step": 1030
635
+ },
636
+ {
637
+ "epoch": 1.21,
638
+ "learning_rate": 0.0002,
639
+ "loss": 0.4206,
640
+ "step": 1040
641
+ },
642
+ {
643
+ "epoch": 1.22,
644
+ "learning_rate": 0.0002,
645
+ "loss": 0.4175,
646
+ "step": 1050
647
+ },
648
+ {
649
+ "epoch": 1.23,
650
+ "learning_rate": 0.0002,
651
+ "loss": 0.4223,
652
+ "step": 1060
653
+ },
654
+ {
655
+ "epoch": 1.24,
656
+ "learning_rate": 0.0002,
657
+ "loss": 0.4179,
658
+ "step": 1070
659
+ },
660
+ {
661
+ "epoch": 1.25,
662
+ "learning_rate": 0.0002,
663
+ "loss": 0.4149,
664
+ "step": 1080
665
+ },
666
+ {
667
+ "epoch": 1.26,
668
+ "learning_rate": 0.0002,
669
+ "loss": 0.4166,
670
+ "step": 1090
671
+ },
672
+ {
673
+ "epoch": 1.28,
674
+ "learning_rate": 0.0002,
675
+ "loss": 0.4097,
676
+ "step": 1100
677
+ },
678
+ {
679
+ "epoch": 1.29,
680
+ "learning_rate": 0.0002,
681
+ "loss": 0.4266,
682
+ "step": 1110
683
+ },
684
+ {
685
+ "epoch": 1.3,
686
+ "learning_rate": 0.0002,
687
+ "loss": 0.429,
688
+ "step": 1120
689
+ },
690
+ {
691
+ "epoch": 1.31,
692
+ "learning_rate": 0.0002,
693
+ "loss": 0.4159,
694
+ "step": 1130
695
+ },
696
+ {
697
+ "epoch": 1.32,
698
+ "learning_rate": 0.0002,
699
+ "loss": 0.4239,
700
+ "step": 1140
701
+ },
702
+ {
703
+ "epoch": 1.33,
704
+ "learning_rate": 0.0002,
705
+ "loss": 0.4074,
706
+ "step": 1150
707
+ },
708
+ {
709
+ "epoch": 1.35,
710
+ "learning_rate": 0.0002,
711
+ "loss": 0.4212,
712
+ "step": 1160
713
+ },
714
+ {
715
+ "epoch": 1.36,
716
+ "learning_rate": 0.0002,
717
+ "loss": 0.4208,
718
+ "step": 1170
719
+ },
720
+ {
721
+ "epoch": 1.37,
722
+ "learning_rate": 0.0002,
723
+ "loss": 0.4156,
724
+ "step": 1180
725
+ },
726
+ {
727
+ "epoch": 1.38,
728
+ "learning_rate": 0.0002,
729
+ "loss": 0.4275,
730
+ "step": 1190
731
+ },
732
+ {
733
+ "epoch": 1.39,
734
+ "learning_rate": 0.0002,
735
+ "loss": 0.462,
736
+ "step": 1200
737
+ },
738
+ {
739
+ "epoch": 1.4,
740
+ "learning_rate": 0.0002,
741
+ "loss": 0.4299,
742
+ "step": 1210
743
+ },
744
+ {
745
+ "epoch": 1.42,
746
+ "learning_rate": 0.0002,
747
+ "loss": 0.4216,
748
+ "step": 1220
749
+ },
750
+ {
751
+ "epoch": 1.43,
752
+ "learning_rate": 0.0002,
753
+ "loss": 0.4112,
754
+ "step": 1230
755
+ },
756
+ {
757
+ "epoch": 1.44,
758
+ "learning_rate": 0.0002,
759
+ "loss": 0.4219,
760
+ "step": 1240
761
+ },
762
+ {
763
+ "epoch": 1.45,
764
+ "learning_rate": 0.0002,
765
+ "loss": 0.4081,
766
+ "step": 1250
767
+ },
768
+ {
769
+ "epoch": 1.46,
770
+ "learning_rate": 0.0002,
771
+ "loss": 0.4218,
772
+ "step": 1260
773
+ },
774
+ {
775
+ "epoch": 1.47,
776
+ "learning_rate": 0.0002,
777
+ "loss": 0.4182,
778
+ "step": 1270
779
+ },
780
+ {
781
+ "epoch": 1.48,
782
+ "learning_rate": 0.0002,
783
+ "loss": 0.4104,
784
+ "step": 1280
785
+ },
786
+ {
787
+ "epoch": 1.5,
788
+ "learning_rate": 0.0002,
789
+ "loss": 0.4216,
790
+ "step": 1290
791
+ },
792
+ {
793
+ "epoch": 1.51,
794
+ "learning_rate": 0.0002,
795
+ "loss": 0.4098,
796
+ "step": 1300
797
+ },
798
+ {
799
+ "epoch": 1.52,
800
+ "learning_rate": 0.0002,
801
+ "loss": 0.4132,
802
+ "step": 1310
803
+ },
804
+ {
805
+ "epoch": 1.53,
806
+ "learning_rate": 0.0002,
807
+ "loss": 0.4118,
808
+ "step": 1320
809
+ },
810
+ {
811
+ "epoch": 1.54,
812
+ "learning_rate": 0.0002,
813
+ "loss": 0.4194,
814
+ "step": 1330
815
+ },
816
+ {
817
+ "epoch": 1.55,
818
+ "learning_rate": 0.0002,
819
+ "loss": 0.4206,
820
+ "step": 1340
821
+ },
822
+ {
823
+ "epoch": 1.57,
824
+ "learning_rate": 0.0002,
825
+ "loss": 0.4073,
826
+ "step": 1350
827
+ },
828
+ {
829
+ "epoch": 1.58,
830
+ "learning_rate": 0.0002,
831
+ "loss": 0.4167,
832
+ "step": 1360
833
+ },
834
+ {
835
+ "epoch": 1.59,
836
+ "learning_rate": 0.0002,
837
+ "loss": 0.406,
838
+ "step": 1370
839
+ },
840
+ {
841
+ "epoch": 1.6,
842
+ "learning_rate": 0.0002,
843
+ "loss": 0.4093,
844
+ "step": 1380
845
+ },
846
+ {
847
+ "epoch": 1.61,
848
+ "learning_rate": 0.0002,
849
+ "loss": 0.418,
850
+ "step": 1390
851
+ },
852
+ {
853
+ "epoch": 1.62,
854
+ "learning_rate": 0.0002,
855
+ "loss": 0.4035,
856
+ "step": 1400
857
+ },
858
+ {
859
+ "epoch": 1.64,
860
+ "learning_rate": 0.0002,
861
+ "loss": 0.4139,
862
+ "step": 1410
863
+ },
864
+ {
865
+ "epoch": 1.65,
866
+ "learning_rate": 0.0002,
867
+ "loss": 0.4058,
868
+ "step": 1420
869
+ },
870
+ {
871
+ "epoch": 1.66,
872
+ "learning_rate": 0.0002,
873
+ "loss": 0.4063,
874
+ "step": 1430
875
+ },
876
+ {
877
+ "epoch": 1.67,
878
+ "learning_rate": 0.0002,
879
+ "loss": 0.4155,
880
+ "step": 1440
881
+ },
882
+ {
883
+ "epoch": 1.68,
884
+ "learning_rate": 0.0002,
885
+ "loss": 0.4307,
886
+ "step": 1450
887
+ },
888
+ {
889
+ "epoch": 1.69,
890
+ "learning_rate": 0.0002,
891
+ "loss": 0.4259,
892
+ "step": 1460
893
+ },
894
+ {
895
+ "epoch": 1.71,
896
+ "learning_rate": 0.0002,
897
+ "loss": 0.4049,
898
+ "step": 1470
899
+ },
900
+ {
901
+ "epoch": 1.72,
902
+ "learning_rate": 0.0002,
903
+ "loss": 0.4101,
904
+ "step": 1480
905
+ },
906
+ {
907
+ "epoch": 1.73,
908
+ "learning_rate": 0.0002,
909
+ "loss": 0.4553,
910
+ "step": 1490
911
+ },
912
+ {
913
+ "epoch": 1.74,
914
+ "learning_rate": 0.0002,
915
+ "loss": 0.4012,
916
+ "step": 1500
917
+ },
918
+ {
919
+ "epoch": 1.75,
920
+ "learning_rate": 0.0002,
921
+ "loss": 0.4094,
922
+ "step": 1510
923
+ },
924
+ {
925
+ "epoch": 1.76,
926
+ "learning_rate": 0.0002,
927
+ "loss": 0.4003,
928
+ "step": 1520
929
+ },
930
+ {
931
+ "epoch": 1.77,
932
+ "learning_rate": 0.0002,
933
+ "loss": 0.4019,
934
+ "step": 1530
935
+ },
936
+ {
937
+ "epoch": 1.79,
938
+ "learning_rate": 0.0002,
939
+ "loss": 0.4107,
940
+ "step": 1540
941
+ },
942
+ {
943
+ "epoch": 1.8,
944
+ "learning_rate": 0.0002,
945
+ "loss": 0.4054,
946
+ "step": 1550
947
+ },
948
+ {
949
+ "epoch": 1.81,
950
+ "learning_rate": 0.0002,
951
+ "loss": 0.4154,
952
+ "step": 1560
953
+ },
954
+ {
955
+ "epoch": 1.82,
956
+ "learning_rate": 0.0002,
957
+ "loss": 0.407,
958
+ "step": 1570
959
+ },
960
+ {
961
+ "epoch": 1.83,
962
+ "learning_rate": 0.0002,
963
+ "loss": 0.4067,
964
+ "step": 1580
965
+ },
966
+ {
967
+ "epoch": 1.84,
968
+ "learning_rate": 0.0002,
969
+ "loss": 0.4109,
970
+ "step": 1590
971
+ },
972
+ {
973
+ "epoch": 1.86,
974
+ "learning_rate": 0.0002,
975
+ "loss": 0.4056,
976
+ "step": 1600
977
+ },
978
+ {
979
+ "epoch": 1.87,
980
+ "learning_rate": 0.0002,
981
+ "loss": 0.4159,
982
+ "step": 1610
983
+ },
984
+ {
985
+ "epoch": 1.88,
986
+ "learning_rate": 0.0002,
987
+ "loss": 0.3987,
988
+ "step": 1620
989
+ },
990
+ {
991
+ "epoch": 1.89,
992
+ "learning_rate": 0.0002,
993
+ "loss": 0.3993,
994
+ "step": 1630
995
+ },
996
+ {
997
+ "epoch": 1.9,
998
+ "learning_rate": 0.0002,
999
+ "loss": 0.4081,
1000
+ "step": 1640
1001
+ },
1002
+ {
1003
+ "epoch": 1.91,
1004
+ "learning_rate": 0.0002,
1005
+ "loss": 0.3993,
1006
+ "step": 1650
1007
+ },
1008
+ {
1009
+ "epoch": 1.93,
1010
+ "learning_rate": 0.0002,
1011
+ "loss": 0.4112,
1012
+ "step": 1660
1013
+ },
1014
+ {
1015
+ "epoch": 1.94,
1016
+ "learning_rate": 0.0002,
1017
+ "loss": 0.4052,
1018
+ "step": 1670
1019
+ },
1020
+ {
1021
+ "epoch": 1.95,
1022
+ "learning_rate": 0.0002,
1023
+ "loss": 0.4028,
1024
+ "step": 1680
1025
+ },
1026
+ {
1027
+ "epoch": 1.96,
1028
+ "learning_rate": 0.0002,
1029
+ "loss": 0.4099,
1030
+ "step": 1690
1031
+ },
1032
+ {
1033
+ "epoch": 1.97,
1034
+ "learning_rate": 0.0002,
1035
+ "loss": 0.4026,
1036
+ "step": 1700
1037
+ },
1038
+ {
1039
+ "epoch": 1.98,
1040
+ "learning_rate": 0.0002,
1041
+ "loss": 0.4015,
1042
+ "step": 1710
1043
+ },
1044
+ {
1045
+ "epoch": 2.0,
1046
+ "learning_rate": 0.0002,
1047
+ "loss": 0.4046,
1048
+ "step": 1720
1049
+ },
1050
+ {
1051
+ "epoch": 2.0,
1052
+ "eval_loss": 0.4530680179595947,
1053
+ "eval_runtime": 5.6216,
1054
+ "eval_samples_per_second": 177.885,
1055
+ "eval_steps_per_second": 2.846,
1056
+ "step": 1724
1057
+ },
1058
+ {
1059
+ "epoch": 2.01,
1060
+ "learning_rate": 0.0002,
1061
+ "loss": 0.3456,
1062
+ "step": 1730
1063
+ },
1064
+ {
1065
+ "epoch": 2.02,
1066
+ "learning_rate": 0.0002,
1067
+ "loss": 0.3203,
1068
+ "step": 1740
1069
+ },
1070
+ {
1071
+ "epoch": 2.03,
1072
+ "learning_rate": 0.0002,
1073
+ "loss": 0.3087,
1074
+ "step": 1750
1075
+ },
1076
+ {
1077
+ "epoch": 2.04,
1078
+ "learning_rate": 0.0002,
1079
+ "loss": 0.3051,
1080
+ "step": 1760
1081
+ },
1082
+ {
1083
+ "epoch": 2.05,
1084
+ "learning_rate": 0.0002,
1085
+ "loss": 0.3183,
1086
+ "step": 1770
1087
+ },
1088
+ {
1089
+ "epoch": 2.06,
1090
+ "learning_rate": 0.0002,
1091
+ "loss": 0.3098,
1092
+ "step": 1780
1093
+ },
1094
+ {
1095
+ "epoch": 2.08,
1096
+ "learning_rate": 0.0002,
1097
+ "loss": 0.3256,
1098
+ "step": 1790
1099
+ },
1100
+ {
1101
+ "epoch": 2.09,
1102
+ "learning_rate": 0.0002,
1103
+ "loss": 0.3189,
1104
+ "step": 1800
1105
+ },
1106
+ {
1107
+ "epoch": 2.1,
1108
+ "learning_rate": 0.0002,
1109
+ "loss": 0.3162,
1110
+ "step": 1810
1111
+ },
1112
+ {
1113
+ "epoch": 2.11,
1114
+ "learning_rate": 0.0002,
1115
+ "loss": 0.326,
1116
+ "step": 1820
1117
+ },
1118
+ {
1119
+ "epoch": 2.12,
1120
+ "learning_rate": 0.0002,
1121
+ "loss": 0.3127,
1122
+ "step": 1830
1123
+ },
1124
+ {
1125
+ "epoch": 2.13,
1126
+ "learning_rate": 0.0002,
1127
+ "loss": 0.3306,
1128
+ "step": 1840
1129
+ },
1130
+ {
1131
+ "epoch": 2.15,
1132
+ "learning_rate": 0.0002,
1133
+ "loss": 0.3255,
1134
+ "step": 1850
1135
+ },
1136
+ {
1137
+ "epoch": 2.16,
1138
+ "learning_rate": 0.0002,
1139
+ "loss": 0.3199,
1140
+ "step": 1860
1141
+ },
1142
+ {
1143
+ "epoch": 2.17,
1144
+ "learning_rate": 0.0002,
1145
+ "loss": 0.3265,
1146
+ "step": 1870
1147
+ },
1148
+ {
1149
+ "epoch": 2.18,
1150
+ "learning_rate": 0.0002,
1151
+ "loss": 0.3185,
1152
+ "step": 1880
1153
+ },
1154
+ {
1155
+ "epoch": 2.19,
1156
+ "learning_rate": 0.0002,
1157
+ "loss": 0.3306,
1158
+ "step": 1890
1159
+ },
1160
+ {
1161
+ "epoch": 2.2,
1162
+ "learning_rate": 0.0002,
1163
+ "loss": 0.3238,
1164
+ "step": 1900
1165
+ },
1166
+ {
1167
+ "epoch": 2.22,
1168
+ "learning_rate": 0.0002,
1169
+ "loss": 0.3297,
1170
+ "step": 1910
1171
+ },
1172
+ {
1173
+ "epoch": 2.23,
1174
+ "learning_rate": 0.0002,
1175
+ "loss": 0.3348,
1176
+ "step": 1920
1177
+ },
1178
+ {
1179
+ "epoch": 2.24,
1180
+ "learning_rate": 0.0002,
1181
+ "loss": 0.329,
1182
+ "step": 1930
1183
+ },
1184
+ {
1185
+ "epoch": 2.25,
1186
+ "learning_rate": 0.0002,
1187
+ "loss": 0.3422,
1188
+ "step": 1940
1189
+ },
1190
+ {
1191
+ "epoch": 2.26,
1192
+ "learning_rate": 0.0002,
1193
+ "loss": 0.3325,
1194
+ "step": 1950
1195
+ },
1196
+ {
1197
+ "epoch": 2.27,
1198
+ "learning_rate": 0.0002,
1199
+ "loss": 0.3327,
1200
+ "step": 1960
1201
+ },
1202
+ {
1203
+ "epoch": 2.29,
1204
+ "learning_rate": 0.0002,
1205
+ "loss": 0.3398,
1206
+ "step": 1970
1207
+ },
1208
+ {
1209
+ "epoch": 2.3,
1210
+ "learning_rate": 0.0002,
1211
+ "loss": 0.3279,
1212
+ "step": 1980
1213
+ },
1214
+ {
1215
+ "epoch": 2.31,
1216
+ "learning_rate": 0.0002,
1217
+ "loss": 0.3463,
1218
+ "step": 1990
1219
+ },
1220
+ {
1221
+ "epoch": 2.32,
1222
+ "learning_rate": 0.0002,
1223
+ "loss": 0.3295,
1224
+ "step": 2000
1225
+ },
1226
+ {
1227
+ "epoch": 2.33,
1228
+ "learning_rate": 0.0002,
1229
+ "loss": 0.3354,
1230
+ "step": 2010
1231
+ },
1232
+ {
1233
+ "epoch": 2.34,
1234
+ "learning_rate": 0.0002,
1235
+ "loss": 0.3411,
1236
+ "step": 2020
1237
+ },
1238
+ {
1239
+ "epoch": 2.35,
1240
+ "learning_rate": 0.0002,
1241
+ "loss": 0.3263,
1242
+ "step": 2030
1243
+ },
1244
+ {
1245
+ "epoch": 2.37,
1246
+ "learning_rate": 0.0002,
1247
+ "loss": 0.3379,
1248
+ "step": 2040
1249
+ },
1250
+ {
1251
+ "epoch": 2.38,
1252
+ "learning_rate": 0.0002,
1253
+ "loss": 0.3305,
1254
+ "step": 2050
1255
+ },
1256
+ {
1257
+ "epoch": 2.39,
1258
+ "learning_rate": 0.0002,
1259
+ "loss": 0.3296,
1260
+ "step": 2060
1261
+ },
1262
+ {
1263
+ "epoch": 2.4,
1264
+ "learning_rate": 0.0002,
1265
+ "loss": 0.3489,
1266
+ "step": 2070
1267
+ },
1268
+ {
1269
+ "epoch": 2.41,
1270
+ "learning_rate": 0.0002,
1271
+ "loss": 0.3325,
1272
+ "step": 2080
1273
+ },
1274
+ {
1275
+ "epoch": 2.42,
1276
+ "learning_rate": 0.0002,
1277
+ "loss": 0.34,
1278
+ "step": 2090
1279
+ },
1280
+ {
1281
+ "epoch": 2.44,
1282
+ "learning_rate": 0.0002,
1283
+ "loss": 0.34,
1284
+ "step": 2100
1285
+ },
1286
+ {
1287
+ "epoch": 2.45,
1288
+ "learning_rate": 0.0002,
1289
+ "loss": 0.3424,
1290
+ "step": 2110
1291
+ },
1292
+ {
1293
+ "epoch": 2.46,
1294
+ "learning_rate": 0.0002,
1295
+ "loss": 0.3421,
1296
+ "step": 2120
1297
+ },
1298
+ {
1299
+ "epoch": 2.47,
1300
+ "learning_rate": 0.0002,
1301
+ "loss": 0.336,
1302
+ "step": 2130
1303
+ },
1304
+ {
1305
+ "epoch": 2.48,
1306
+ "learning_rate": 0.0002,
1307
+ "loss": 0.3481,
1308
+ "step": 2140
1309
+ },
1310
+ {
1311
+ "epoch": 2.49,
1312
+ "learning_rate": 0.0002,
1313
+ "loss": 0.3426,
1314
+ "step": 2150
1315
+ },
1316
+ {
1317
+ "epoch": 2.51,
1318
+ "learning_rate": 0.0002,
1319
+ "loss": 0.3425,
1320
+ "step": 2160
1321
+ },
1322
+ {
1323
+ "epoch": 2.52,
1324
+ "learning_rate": 0.0002,
1325
+ "loss": 0.3442,
1326
+ "step": 2170
1327
+ },
1328
+ {
1329
+ "epoch": 2.53,
1330
+ "learning_rate": 0.0002,
1331
+ "loss": 0.3351,
1332
+ "step": 2180
1333
+ },
1334
+ {
1335
+ "epoch": 2.54,
1336
+ "learning_rate": 0.0002,
1337
+ "loss": 0.3582,
1338
+ "step": 2190
1339
+ },
1340
+ {
1341
+ "epoch": 2.55,
1342
+ "learning_rate": 0.0002,
1343
+ "loss": 0.3429,
1344
+ "step": 2200
1345
+ },
1346
+ {
1347
+ "epoch": 2.56,
1348
+ "learning_rate": 0.0002,
1349
+ "loss": 0.3421,
1350
+ "step": 2210
1351
+ },
1352
+ {
1353
+ "epoch": 2.58,
1354
+ "learning_rate": 0.0002,
1355
+ "loss": 0.3444,
1356
+ "step": 2220
1357
+ },
1358
+ {
1359
+ "epoch": 2.59,
1360
+ "learning_rate": 0.0002,
1361
+ "loss": 0.3345,
1362
+ "step": 2230
1363
+ },
1364
+ {
1365
+ "epoch": 2.6,
1366
+ "learning_rate": 0.0002,
1367
+ "loss": 0.3484,
1368
+ "step": 2240
1369
+ },
1370
+ {
1371
+ "epoch": 2.61,
1372
+ "learning_rate": 0.0002,
1373
+ "loss": 0.3451,
1374
+ "step": 2250
1375
+ },
1376
+ {
1377
+ "epoch": 2.62,
1378
+ "learning_rate": 0.0002,
1379
+ "loss": 0.341,
1380
+ "step": 2260
1381
+ },
1382
+ {
1383
+ "epoch": 2.63,
1384
+ "learning_rate": 0.0002,
1385
+ "loss": 0.3447,
1386
+ "step": 2270
1387
+ },
1388
+ {
1389
+ "epoch": 2.65,
1390
+ "learning_rate": 0.0002,
1391
+ "loss": 0.3354,
1392
+ "step": 2280
1393
+ },
1394
+ {
1395
+ "epoch": 2.66,
1396
+ "learning_rate": 0.0002,
1397
+ "loss": 0.3476,
1398
+ "step": 2290
1399
+ },
1400
+ {
1401
+ "epoch": 2.67,
1402
+ "learning_rate": 0.0002,
1403
+ "loss": 0.3377,
1404
+ "step": 2300
1405
+ },
1406
+ {
1407
+ "epoch": 2.68,
1408
+ "learning_rate": 0.0002,
1409
+ "loss": 0.3349,
1410
+ "step": 2310
1411
+ },
1412
+ {
1413
+ "epoch": 2.69,
1414
+ "learning_rate": 0.0002,
1415
+ "loss": 0.3432,
1416
+ "step": 2320
1417
+ },
1418
+ {
1419
+ "epoch": 2.7,
1420
+ "learning_rate": 0.0002,
1421
+ "loss": 0.3321,
1422
+ "step": 2330
1423
+ },
1424
+ {
1425
+ "epoch": 2.71,
1426
+ "learning_rate": 0.0002,
1427
+ "loss": 0.3475,
1428
+ "step": 2340
1429
+ },
1430
+ {
1431
+ "epoch": 2.73,
1432
+ "learning_rate": 0.0002,
1433
+ "loss": 0.3367,
1434
+ "step": 2350
1435
+ },
1436
+ {
1437
+ "epoch": 2.74,
1438
+ "learning_rate": 0.0002,
1439
+ "loss": 0.3408,
1440
+ "step": 2360
1441
+ },
1442
+ {
1443
+ "epoch": 2.75,
1444
+ "learning_rate": 0.0002,
1445
+ "loss": 0.3465,
1446
+ "step": 2370
1447
+ },
1448
+ {
1449
+ "epoch": 2.76,
1450
+ "learning_rate": 0.0002,
1451
+ "loss": 0.3331,
1452
+ "step": 2380
1453
+ },
1454
+ {
1455
+ "epoch": 2.77,
1456
+ "learning_rate": 0.0002,
1457
+ "loss": 0.3532,
1458
+ "step": 2390
1459
+ },
1460
+ {
1461
+ "epoch": 2.78,
1462
+ "learning_rate": 0.0002,
1463
+ "loss": 0.3425,
1464
+ "step": 2400
1465
+ },
1466
+ {
1467
+ "epoch": 2.8,
1468
+ "learning_rate": 0.0002,
1469
+ "loss": 0.3412,
1470
+ "step": 2410
1471
+ },
1472
+ {
1473
+ "epoch": 2.81,
1474
+ "learning_rate": 0.0002,
1475
+ "loss": 0.344,
1476
+ "step": 2420
1477
+ },
1478
+ {
1479
+ "epoch": 2.82,
1480
+ "learning_rate": 0.0002,
1481
+ "loss": 0.3409,
1482
+ "step": 2430
1483
+ },
1484
+ {
1485
+ "epoch": 2.83,
1486
+ "learning_rate": 0.0002,
1487
+ "loss": 0.3455,
1488
+ "step": 2440
1489
+ },
1490
+ {
1491
+ "epoch": 2.84,
1492
+ "learning_rate": 0.0002,
1493
+ "loss": 0.344,
1494
+ "step": 2450
1495
+ },
1496
+ {
1497
+ "epoch": 2.85,
1498
+ "learning_rate": 0.0002,
1499
+ "loss": 0.3386,
1500
+ "step": 2460
1501
+ },
1502
+ {
1503
+ "epoch": 2.87,
1504
+ "learning_rate": 0.0002,
1505
+ "loss": 0.3443,
1506
+ "step": 2470
1507
+ },
1508
+ {
1509
+ "epoch": 2.88,
1510
+ "learning_rate": 0.0002,
1511
+ "loss": 0.3364,
1512
+ "step": 2480
1513
+ },
1514
+ {
1515
+ "epoch": 2.89,
1516
+ "learning_rate": 0.0002,
1517
+ "loss": 0.3584,
1518
+ "step": 2490
1519
+ },
1520
+ {
1521
+ "epoch": 2.9,
1522
+ "learning_rate": 0.0002,
1523
+ "loss": 0.3469,
1524
+ "step": 2500
1525
+ },
1526
+ {
1527
+ "epoch": 2.91,
1528
+ "learning_rate": 0.0002,
1529
+ "loss": 0.3476,
1530
+ "step": 2510
1531
+ },
1532
+ {
1533
+ "epoch": 2.92,
1534
+ "learning_rate": 0.0002,
1535
+ "loss": 0.3544,
1536
+ "step": 2520
1537
+ },
1538
+ {
1539
+ "epoch": 2.94,
1540
+ "learning_rate": 0.0002,
1541
+ "loss": 0.3384,
1542
+ "step": 2530
1543
+ },
1544
+ {
1545
+ "epoch": 2.95,
1546
+ "learning_rate": 0.0002,
1547
+ "loss": 0.351,
1548
+ "step": 2540
1549
+ },
1550
+ {
1551
+ "epoch": 2.96,
1552
+ "learning_rate": 0.0002,
1553
+ "loss": 0.3383,
1554
+ "step": 2550
1555
+ },
1556
+ {
1557
+ "epoch": 2.97,
1558
+ "learning_rate": 0.0002,
1559
+ "loss": 0.3435,
1560
+ "step": 2560
1561
+ },
1562
+ {
1563
+ "epoch": 2.98,
1564
+ "learning_rate": 0.0002,
1565
+ "loss": 0.3542,
1566
+ "step": 2570
1567
+ },
1568
+ {
1569
+ "epoch": 2.99,
1570
+ "learning_rate": 0.0002,
1571
+ "loss": 0.3395,
1572
+ "step": 2580
1573
+ },
1574
+ {
1575
+ "epoch": 3.0,
1576
+ "eval_loss": 0.43056488037109375,
1577
+ "eval_runtime": 5.6316,
1578
+ "eval_samples_per_second": 177.57,
1579
+ "eval_steps_per_second": 2.841,
1580
+ "step": 2586
1581
+ },
1582
+ {
1583
+ "epoch": 3.0,
1584
+ "learning_rate": 0.0002,
1585
+ "loss": 0.3188,
1586
+ "step": 2590
1587
+ },
1588
+ {
1589
+ "epoch": 3.02,
1590
+ "learning_rate": 0.0002,
1591
+ "loss": 0.2607,
1592
+ "step": 2600
1593
+ },
1594
+ {
1595
+ "epoch": 3.03,
1596
+ "learning_rate": 0.0002,
1597
+ "loss": 0.2477,
1598
+ "step": 2610
1599
+ },
1600
+ {
1601
+ "epoch": 3.04,
1602
+ "learning_rate": 0.0002,
1603
+ "loss": 0.2591,
1604
+ "step": 2620
1605
+ },
1606
+ {
1607
+ "epoch": 3.05,
1608
+ "learning_rate": 0.0002,
1609
+ "loss": 0.2514,
1610
+ "step": 2630
1611
+ },
1612
+ {
1613
+ "epoch": 3.06,
1614
+ "learning_rate": 0.0002,
1615
+ "loss": 0.2539,
1616
+ "step": 2640
1617
+ },
1618
+ {
1619
+ "epoch": 3.07,
1620
+ "learning_rate": 0.0002,
1621
+ "loss": 0.2631,
1622
+ "step": 2650
1623
+ },
1624
+ {
1625
+ "epoch": 3.09,
1626
+ "learning_rate": 0.0002,
1627
+ "loss": 0.2571,
1628
+ "step": 2660
1629
+ },
1630
+ {
1631
+ "epoch": 3.1,
1632
+ "learning_rate": 0.0002,
1633
+ "loss": 0.2645,
1634
+ "step": 2670
1635
+ },
1636
+ {
1637
+ "epoch": 3.11,
1638
+ "learning_rate": 0.0002,
1639
+ "loss": 0.2571,
1640
+ "step": 2680
1641
+ },
1642
+ {
1643
+ "epoch": 3.12,
1644
+ "learning_rate": 0.0002,
1645
+ "loss": 0.2715,
1646
+ "step": 2690
1647
+ },
1648
+ {
1649
+ "epoch": 3.13,
1650
+ "learning_rate": 0.0002,
1651
+ "loss": 0.2692,
1652
+ "step": 2700
1653
+ },
1654
+ {
1655
+ "epoch": 3.14,
1656
+ "learning_rate": 0.0002,
1657
+ "loss": 0.2582,
1658
+ "step": 2710
1659
+ },
1660
+ {
1661
+ "epoch": 3.16,
1662
+ "learning_rate": 0.0002,
1663
+ "loss": 0.2683,
1664
+ "step": 2720
1665
+ },
1666
+ {
1667
+ "epoch": 3.17,
1668
+ "learning_rate": 0.0002,
1669
+ "loss": 0.2612,
1670
+ "step": 2730
1671
+ },
1672
+ {
1673
+ "epoch": 3.18,
1674
+ "learning_rate": 0.0002,
1675
+ "loss": 0.2736,
1676
+ "step": 2740
1677
+ },
1678
+ {
1679
+ "epoch": 3.19,
1680
+ "learning_rate": 0.0002,
1681
+ "loss": 0.2678,
1682
+ "step": 2750
1683
+ },
1684
+ {
1685
+ "epoch": 3.2,
1686
+ "learning_rate": 0.0002,
1687
+ "loss": 0.2623,
1688
+ "step": 2760
1689
+ },
1690
+ {
1691
+ "epoch": 3.21,
1692
+ "learning_rate": 0.0002,
1693
+ "loss": 0.2768,
1694
+ "step": 2770
1695
+ },
1696
+ {
1697
+ "epoch": 3.23,
1698
+ "learning_rate": 0.0002,
1699
+ "loss": 0.2651,
1700
+ "step": 2780
1701
+ },
1702
+ {
1703
+ "epoch": 3.24,
1704
+ "learning_rate": 0.0002,
1705
+ "loss": 0.2702,
1706
+ "step": 2790
1707
+ },
1708
+ {
1709
+ "epoch": 3.25,
1710
+ "learning_rate": 0.0002,
1711
+ "loss": 0.2744,
1712
+ "step": 2800
1713
+ },
1714
+ {
1715
+ "epoch": 3.26,
1716
+ "learning_rate": 0.0002,
1717
+ "loss": 0.2686,
1718
+ "step": 2810
1719
+ },
1720
+ {
1721
+ "epoch": 3.27,
1722
+ "learning_rate": 0.0002,
1723
+ "loss": 0.278,
1724
+ "step": 2820
1725
+ },
1726
+ {
1727
+ "epoch": 3.28,
1728
+ "learning_rate": 0.0002,
1729
+ "loss": 0.2724,
1730
+ "step": 2830
1731
+ },
1732
+ {
1733
+ "epoch": 3.29,
1734
+ "learning_rate": 0.0002,
1735
+ "loss": 0.284,
1736
+ "step": 2840
1737
+ },
1738
+ {
1739
+ "epoch": 3.31,
1740
+ "learning_rate": 0.0002,
1741
+ "loss": 0.2831,
1742
+ "step": 2850
1743
+ },
1744
+ {
1745
+ "epoch": 3.32,
1746
+ "learning_rate": 0.0002,
1747
+ "loss": 0.2736,
1748
+ "step": 2860
1749
+ },
1750
+ {
1751
+ "epoch": 3.33,
1752
+ "learning_rate": 0.0002,
1753
+ "loss": 0.288,
1754
+ "step": 2870
1755
+ },
1756
+ {
1757
+ "epoch": 3.34,
1758
+ "learning_rate": 0.0002,
1759
+ "loss": 0.2779,
1760
+ "step": 2880
1761
+ },
1762
+ {
1763
+ "epoch": 3.35,
1764
+ "learning_rate": 0.0002,
1765
+ "loss": 0.2841,
1766
+ "step": 2890
1767
+ },
1768
+ {
1769
+ "epoch": 3.36,
1770
+ "learning_rate": 0.0002,
1771
+ "loss": 0.2822,
1772
+ "step": 2900
1773
+ },
1774
+ {
1775
+ "epoch": 3.38,
1776
+ "learning_rate": 0.0002,
1777
+ "loss": 0.2751,
1778
+ "step": 2910
1779
+ },
1780
+ {
1781
+ "epoch": 3.39,
1782
+ "learning_rate": 0.0002,
1783
+ "loss": 0.2851,
1784
+ "step": 2920
1785
+ },
1786
+ {
1787
+ "epoch": 3.4,
1788
+ "learning_rate": 0.0002,
1789
+ "loss": 0.2723,
1790
+ "step": 2930
1791
+ },
1792
+ {
1793
+ "epoch": 3.41,
1794
+ "learning_rate": 0.0002,
1795
+ "loss": 0.2852,
1796
+ "step": 2940
1797
+ },
1798
+ {
1799
+ "epoch": 3.42,
1800
+ "learning_rate": 0.0002,
1801
+ "loss": 0.2871,
1802
+ "step": 2950
1803
+ },
1804
+ {
1805
+ "epoch": 3.43,
1806
+ "learning_rate": 0.0002,
1807
+ "loss": 0.2774,
1808
+ "step": 2960
1809
+ },
1810
+ {
1811
+ "epoch": 3.45,
1812
+ "learning_rate": 0.0002,
1813
+ "loss": 0.2857,
1814
+ "step": 2970
1815
+ },
1816
+ {
1817
+ "epoch": 3.46,
1818
+ "learning_rate": 0.0002,
1819
+ "loss": 0.2768,
1820
+ "step": 2980
1821
+ },
1822
+ {
1823
+ "epoch": 3.47,
1824
+ "learning_rate": 0.0002,
1825
+ "loss": 0.2888,
1826
+ "step": 2990
1827
+ },
1828
+ {
1829
+ "epoch": 3.48,
1830
+ "learning_rate": 0.0002,
1831
+ "loss": 0.2879,
1832
+ "step": 3000
1833
+ },
1834
+ {
1835
+ "epoch": 3.49,
1836
+ "learning_rate": 0.0002,
1837
+ "loss": 0.2798,
1838
+ "step": 3010
1839
+ },
1840
+ {
1841
+ "epoch": 3.5,
1842
+ "learning_rate": 0.0002,
1843
+ "loss": 0.2908,
1844
+ "step": 3020
1845
+ },
1846
+ {
1847
+ "epoch": 3.52,
1848
+ "learning_rate": 0.0002,
1849
+ "loss": 0.2756,
1850
+ "step": 3030
1851
+ },
1852
+ {
1853
+ "epoch": 3.53,
1854
+ "learning_rate": 0.0002,
1855
+ "loss": 0.2941,
1856
+ "step": 3040
1857
+ },
1858
+ {
1859
+ "epoch": 3.54,
1860
+ "learning_rate": 0.0002,
1861
+ "loss": 0.2893,
1862
+ "step": 3050
1863
+ },
1864
+ {
1865
+ "epoch": 3.55,
1866
+ "learning_rate": 0.0002,
1867
+ "loss": 0.2803,
1868
+ "step": 3060
1869
+ },
1870
+ {
1871
+ "epoch": 3.56,
1872
+ "learning_rate": 0.0002,
1873
+ "loss": 0.2899,
1874
+ "step": 3070
1875
+ },
1876
+ {
1877
+ "epoch": 3.57,
1878
+ "learning_rate": 0.0002,
1879
+ "loss": 0.2833,
1880
+ "step": 3080
1881
+ },
1882
+ {
1883
+ "epoch": 3.58,
1884
+ "learning_rate": 0.0002,
1885
+ "loss": 0.2908,
1886
+ "step": 3090
1887
+ },
1888
+ {
1889
+ "epoch": 3.6,
1890
+ "learning_rate": 0.0002,
1891
+ "loss": 0.2919,
1892
+ "step": 3100
1893
+ },
1894
+ {
1895
+ "epoch": 3.61,
1896
+ "learning_rate": 0.0002,
1897
+ "loss": 0.2865,
1898
+ "step": 3110
1899
+ },
1900
+ {
1901
+ "epoch": 3.62,
1902
+ "learning_rate": 0.0002,
1903
+ "loss": 0.3033,
1904
+ "step": 3120
1905
+ },
1906
+ {
1907
+ "epoch": 3.63,
1908
+ "learning_rate": 0.0002,
1909
+ "loss": 0.2802,
1910
+ "step": 3130
1911
+ },
1912
+ {
1913
+ "epoch": 3.64,
1914
+ "learning_rate": 0.0002,
1915
+ "loss": 0.2918,
1916
+ "step": 3140
1917
+ },
1918
+ {
1919
+ "epoch": 3.65,
1920
+ "learning_rate": 0.0002,
1921
+ "loss": 0.2911,
1922
+ "step": 3150
1923
+ },
1924
+ {
1925
+ "epoch": 3.67,
1926
+ "learning_rate": 0.0002,
1927
+ "loss": 0.2882,
1928
+ "step": 3160
1929
+ },
1930
+ {
1931
+ "epoch": 3.68,
1932
+ "learning_rate": 0.0002,
1933
+ "loss": 0.2873,
1934
+ "step": 3170
1935
+ },
1936
+ {
1937
+ "epoch": 3.69,
1938
+ "learning_rate": 0.0002,
1939
+ "loss": 0.286,
1940
+ "step": 3180
1941
+ },
1942
+ {
1943
+ "epoch": 3.7,
1944
+ "learning_rate": 0.0002,
1945
+ "loss": 0.294,
1946
+ "step": 3190
1947
+ },
1948
+ {
1949
+ "epoch": 3.71,
1950
+ "learning_rate": 0.0002,
1951
+ "loss": 0.2903,
1952
+ "step": 3200
1953
+ },
1954
+ {
1955
+ "epoch": 3.72,
1956
+ "learning_rate": 0.0002,
1957
+ "loss": 0.2818,
1958
+ "step": 3210
1959
+ },
1960
+ {
1961
+ "epoch": 3.74,
1962
+ "learning_rate": 0.0002,
1963
+ "loss": 0.2925,
1964
+ "step": 3220
1965
+ },
1966
+ {
1967
+ "epoch": 3.75,
1968
+ "learning_rate": 0.0002,
1969
+ "loss": 0.2834,
1970
+ "step": 3230
1971
+ },
1972
+ {
1973
+ "epoch": 3.76,
1974
+ "learning_rate": 0.0002,
1975
+ "loss": 0.2932,
1976
+ "step": 3240
1977
+ },
1978
+ {
1979
+ "epoch": 3.77,
1980
+ "learning_rate": 0.0002,
1981
+ "loss": 0.3,
1982
+ "step": 3250
1983
+ },
1984
+ {
1985
+ "epoch": 3.78,
1986
+ "learning_rate": 0.0002,
1987
+ "loss": 0.2875,
1988
+ "step": 3260
1989
+ },
1990
+ {
1991
+ "epoch": 3.79,
1992
+ "learning_rate": 0.0002,
1993
+ "loss": 0.2922,
1994
+ "step": 3270
1995
+ },
1996
+ {
1997
+ "epoch": 3.81,
1998
+ "learning_rate": 0.0002,
1999
+ "loss": 0.2855,
2000
+ "step": 3280
2001
+ },
2002
+ {
2003
+ "epoch": 3.82,
2004
+ "learning_rate": 0.0002,
2005
+ "loss": 0.2977,
2006
+ "step": 3290
2007
+ },
2008
+ {
2009
+ "epoch": 3.83,
2010
+ "learning_rate": 0.0002,
2011
+ "loss": 0.2987,
2012
+ "step": 3300
2013
+ },
2014
+ {
2015
+ "epoch": 3.84,
2016
+ "learning_rate": 0.0002,
2017
+ "loss": 0.2929,
2018
+ "step": 3310
2019
+ },
2020
+ {
2021
+ "epoch": 3.85,
2022
+ "learning_rate": 0.0002,
2023
+ "loss": 0.2979,
2024
+ "step": 3320
2025
+ },
2026
+ {
2027
+ "epoch": 3.86,
2028
+ "learning_rate": 0.0002,
2029
+ "loss": 0.2904,
2030
+ "step": 3330
2031
+ },
2032
+ {
2033
+ "epoch": 3.87,
2034
+ "learning_rate": 0.0002,
2035
+ "loss": 0.2977,
2036
+ "step": 3340
2037
+ },
2038
+ {
2039
+ "epoch": 3.89,
2040
+ "learning_rate": 0.0002,
2041
+ "loss": 0.2936,
2042
+ "step": 3350
2043
+ },
2044
+ {
2045
+ "epoch": 3.9,
2046
+ "learning_rate": 0.0002,
2047
+ "loss": 0.2924,
2048
+ "step": 3360
2049
+ },
2050
+ {
2051
+ "epoch": 3.91,
2052
+ "learning_rate": 0.0002,
2053
+ "loss": 0.2982,
2054
+ "step": 3370
2055
+ },
2056
+ {
2057
+ "epoch": 3.92,
2058
+ "learning_rate": 0.0002,
2059
+ "loss": 0.2897,
2060
+ "step": 3380
2061
+ },
2062
+ {
2063
+ "epoch": 3.93,
2064
+ "learning_rate": 0.0002,
2065
+ "loss": 0.3037,
2066
+ "step": 3390
2067
+ },
2068
+ {
2069
+ "epoch": 3.94,
2070
+ "learning_rate": 0.0002,
2071
+ "loss": 0.3005,
2072
+ "step": 3400
2073
+ },
2074
+ {
2075
+ "epoch": 3.96,
2076
+ "learning_rate": 0.0002,
2077
+ "loss": 0.294,
2078
+ "step": 3410
2079
+ },
2080
+ {
2081
+ "epoch": 3.97,
2082
+ "learning_rate": 0.0002,
2083
+ "loss": 0.3017,
2084
+ "step": 3420
2085
+ },
2086
+ {
2087
+ "epoch": 3.98,
2088
+ "learning_rate": 0.0002,
2089
+ "loss": 0.2917,
2090
+ "step": 3430
2091
+ },
2092
+ {
2093
+ "epoch": 3.99,
2094
+ "learning_rate": 0.0002,
2095
+ "loss": 0.3012,
2096
+ "step": 3440
2097
+ },
2098
+ {
2099
+ "epoch": 4.0,
2100
+ "eval_loss": 0.4269254803657532,
2101
+ "eval_runtime": 5.6372,
2102
+ "eval_samples_per_second": 177.394,
2103
+ "eval_steps_per_second": 2.838,
2104
+ "step": 3448
2105
+ }
2106
+ ],
2107
+ "max_steps": 10000,
2108
+ "num_train_epochs": 12,
2109
+ "total_flos": 1.6741838366462444e+18,
2110
+ "trial_name": null,
2111
+ "trial_params": null
2112
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b44edf850481de8b79394a9ce0c71577a9b864ae7d091ca4e903fe1cc87d84dd
3
+ size 5615
vocab.json ADDED
The diff for this file is too large to render. See raw diff