Hans-Christian Bøge Pedersen commited on
Commit
37c1337
1 Parent(s): 542e8a8

Training in progress, step 13600

Browse files
.gitignore DELETED
@@ -1 +0,0 @@
1
- checkpoint-*/
 
 
checkpoint-13200/config.json ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": "b05d55b99887e2a2713474fd1590c11ff0a45a94",
3
+ "_name_or_path": "Tech-Hans/detr-resnet-50_finetuned_fashion",
4
+ "activation_dropout": 0.0,
5
+ "activation_function": "relu",
6
+ "architectures": [
7
+ "DetrForObjectDetection"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "auxiliary_loss": false,
11
+ "backbone": "resnet50",
12
+ "backbone_config": null,
13
+ "bbox_cost": 5,
14
+ "bbox_loss_coefficient": 5,
15
+ "class_cost": 1,
16
+ "classifier_dropout": 0.0,
17
+ "d_model": 256,
18
+ "decoder_attention_heads": 8,
19
+ "decoder_ffn_dim": 2048,
20
+ "decoder_layerdrop": 0.0,
21
+ "decoder_layers": 6,
22
+ "dice_loss_coefficient": 1,
23
+ "dilation": false,
24
+ "dropout": 0.1,
25
+ "encoder_attention_heads": 8,
26
+ "encoder_ffn_dim": 2048,
27
+ "encoder_layerdrop": 0.0,
28
+ "encoder_layers": 6,
29
+ "eos_coefficient": 0.1,
30
+ "giou_cost": 2,
31
+ "giou_loss_coefficient": 2,
32
+ "id2label": {
33
+ "0": "bags",
34
+ "1": "belt",
35
+ "2": "bottom",
36
+ "3": "fullbody",
37
+ "4": "glasses",
38
+ "5": "glove",
39
+ "6": "hat",
40
+ "7": "scarf",
41
+ "8": "shoes",
42
+ "9": "sock",
43
+ "10": "tie",
44
+ "11": "tights",
45
+ "12": "top"
46
+ },
47
+ "init_std": 0.02,
48
+ "init_xavier_std": 1.0,
49
+ "is_encoder_decoder": true,
50
+ "label2id": {
51
+ "bags": 0,
52
+ "belt": 1,
53
+ "bottom": 2,
54
+ "fullbody": 3,
55
+ "glasses": 4,
56
+ "glove": 5,
57
+ "hat": 6,
58
+ "scarf": 7,
59
+ "shoes": 8,
60
+ "sock": 9,
61
+ "tie": 10,
62
+ "tights": 11,
63
+ "top": 12
64
+ },
65
+ "mask_loss_coefficient": 1,
66
+ "max_position_embeddings": 1024,
67
+ "model_type": "detr",
68
+ "num_channels": 3,
69
+ "num_hidden_layers": 6,
70
+ "num_queries": 100,
71
+ "position_embedding_type": "sine",
72
+ "scale_embedding": false,
73
+ "torch_dtype": "float32",
74
+ "transformers_version": "4.7.0.dev0",
75
+ "use_pretrained_backbone": true,
76
+ "use_timm_backbone": true
77
+ }
checkpoint-13200/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef36ca8c8c70233abc7a8cee6a5d1a0555740766ca04707bae03e3f0c580c5f7
3
+ size 330501178
checkpoint-13200/preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_pad": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "format": "coco_detection",
7
+ "image_mean": [
8
+ 0.485,
9
+ 0.456,
10
+ 0.406
11
+ ],
12
+ "image_processor_type": "DetrImageProcessor",
13
+ "image_std": [
14
+ 0.229,
15
+ 0.224,
16
+ 0.225
17
+ ],
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "longest_edge": 1333,
22
+ "shortest_edge": 800
23
+ }
24
+ }
checkpoint-13200/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a603da01c74d9504c35845be6ee4365c2df4b2bd1c14a68b036545f8b9bfe0cf
3
+ size 166628834
checkpoint-13200/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:baeb722c7eb9df4195b8f19894482c730fd38333b5ab545d6c278f67dba1e795
3
+ size 14244
checkpoint-13200/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9a14aac3f1a42a5643173abb85e00e68e78f46b63a124973b1fb9b5013022b7a
3
+ size 988
checkpoint-13200/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:811bdd1945539ce0d69d353e2dedff42b36763d7e95bd7d48ace3a36632e96c8
3
+ size 1064
checkpoint-13200/trainer_state.json ADDED
@@ -0,0 +1,1600 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.508075242257268,
5
+ "global_step": 13200,
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": 9.992019760592819e-06,
13
+ "loss": 1.1564,
14
+ "step": 50
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 9.982709481284439e-06,
19
+ "loss": 1.132,
20
+ "step": 100
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 9.97320919627589e-06,
25
+ "loss": 1.1507,
26
+ "step": 150
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 9.96370891126734e-06,
31
+ "loss": 1.0846,
32
+ "step": 200
33
+ },
34
+ {
35
+ "epoch": 0.05,
36
+ "learning_rate": 9.954208626258789e-06,
37
+ "loss": 1.0877,
38
+ "step": 250
39
+ },
40
+ {
41
+ "epoch": 0.06,
42
+ "learning_rate": 9.944708341250238e-06,
43
+ "loss": 1.1153,
44
+ "step": 300
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "learning_rate": 9.935208056241688e-06,
49
+ "loss": 1.0876,
50
+ "step": 350
51
+ },
52
+ {
53
+ "epoch": 0.08,
54
+ "learning_rate": 9.925707771233137e-06,
55
+ "loss": 1.0729,
56
+ "step": 400
57
+ },
58
+ {
59
+ "epoch": 0.09,
60
+ "learning_rate": 9.916207486224587e-06,
61
+ "loss": 1.126,
62
+ "step": 450
63
+ },
64
+ {
65
+ "epoch": 0.1,
66
+ "learning_rate": 9.906707201216036e-06,
67
+ "loss": 1.1262,
68
+ "step": 500
69
+ },
70
+ {
71
+ "epoch": 0.1,
72
+ "learning_rate": 9.897206916207487e-06,
73
+ "loss": 1.0883,
74
+ "step": 550
75
+ },
76
+ {
77
+ "epoch": 0.11,
78
+ "learning_rate": 9.887706631198937e-06,
79
+ "loss": 1.0926,
80
+ "step": 600
81
+ },
82
+ {
83
+ "epoch": 0.12,
84
+ "learning_rate": 9.878206346190386e-06,
85
+ "loss": 1.0916,
86
+ "step": 650
87
+ },
88
+ {
89
+ "epoch": 0.13,
90
+ "learning_rate": 9.868706061181836e-06,
91
+ "loss": 1.0892,
92
+ "step": 700
93
+ },
94
+ {
95
+ "epoch": 0.14,
96
+ "learning_rate": 9.859205776173287e-06,
97
+ "loss": 1.0653,
98
+ "step": 750
99
+ },
100
+ {
101
+ "epoch": 0.15,
102
+ "learning_rate": 9.849705491164736e-06,
103
+ "loss": 1.0391,
104
+ "step": 800
105
+ },
106
+ {
107
+ "epoch": 0.16,
108
+ "learning_rate": 9.840205206156186e-06,
109
+ "loss": 1.0866,
110
+ "step": 850
111
+ },
112
+ {
113
+ "epoch": 0.17,
114
+ "learning_rate": 9.830704921147635e-06,
115
+ "loss": 1.1192,
116
+ "step": 900
117
+ },
118
+ {
119
+ "epoch": 0.18,
120
+ "learning_rate": 9.821204636139085e-06,
121
+ "loss": 1.0898,
122
+ "step": 950
123
+ },
124
+ {
125
+ "epoch": 0.19,
126
+ "learning_rate": 9.811704351130534e-06,
127
+ "loss": 1.0693,
128
+ "step": 1000
129
+ },
130
+ {
131
+ "epoch": 0.2,
132
+ "learning_rate": 9.802204066121984e-06,
133
+ "loss": 1.1243,
134
+ "step": 1050
135
+ },
136
+ {
137
+ "epoch": 0.21,
138
+ "learning_rate": 9.792703781113433e-06,
139
+ "loss": 1.071,
140
+ "step": 1100
141
+ },
142
+ {
143
+ "epoch": 0.22,
144
+ "learning_rate": 9.783203496104884e-06,
145
+ "loss": 1.0569,
146
+ "step": 1150
147
+ },
148
+ {
149
+ "epoch": 0.23,
150
+ "learning_rate": 9.773703211096334e-06,
151
+ "loss": 1.0612,
152
+ "step": 1200
153
+ },
154
+ {
155
+ "epoch": 0.24,
156
+ "learning_rate": 9.764202926087783e-06,
157
+ "loss": 1.0432,
158
+ "step": 1250
159
+ },
160
+ {
161
+ "epoch": 0.25,
162
+ "learning_rate": 9.754702641079233e-06,
163
+ "loss": 1.092,
164
+ "step": 1300
165
+ },
166
+ {
167
+ "epoch": 0.26,
168
+ "learning_rate": 9.745202356070684e-06,
169
+ "loss": 1.0602,
170
+ "step": 1350
171
+ },
172
+ {
173
+ "epoch": 0.27,
174
+ "learning_rate": 9.735702071062133e-06,
175
+ "loss": 1.0689,
176
+ "step": 1400
177
+ },
178
+ {
179
+ "epoch": 0.28,
180
+ "learning_rate": 9.726201786053583e-06,
181
+ "loss": 1.0897,
182
+ "step": 1450
183
+ },
184
+ {
185
+ "epoch": 0.29,
186
+ "learning_rate": 9.716701501045032e-06,
187
+ "loss": 1.0482,
188
+ "step": 1500
189
+ },
190
+ {
191
+ "epoch": 0.29,
192
+ "learning_rate": 9.707201216036482e-06,
193
+ "loss": 1.1167,
194
+ "step": 1550
195
+ },
196
+ {
197
+ "epoch": 0.3,
198
+ "learning_rate": 9.697700931027931e-06,
199
+ "loss": 1.0707,
200
+ "step": 1600
201
+ },
202
+ {
203
+ "epoch": 0.31,
204
+ "learning_rate": 9.68820064601938e-06,
205
+ "loss": 1.0613,
206
+ "step": 1650
207
+ },
208
+ {
209
+ "epoch": 0.32,
210
+ "learning_rate": 9.67870036101083e-06,
211
+ "loss": 1.0534,
212
+ "step": 1700
213
+ },
214
+ {
215
+ "epoch": 0.33,
216
+ "learning_rate": 9.669200076002281e-06,
217
+ "loss": 1.0719,
218
+ "step": 1750
219
+ },
220
+ {
221
+ "epoch": 0.34,
222
+ "learning_rate": 9.659699790993731e-06,
223
+ "loss": 1.0678,
224
+ "step": 1800
225
+ },
226
+ {
227
+ "epoch": 0.35,
228
+ "learning_rate": 9.65019950598518e-06,
229
+ "loss": 1.064,
230
+ "step": 1850
231
+ },
232
+ {
233
+ "epoch": 0.36,
234
+ "learning_rate": 9.64069922097663e-06,
235
+ "loss": 1.0846,
236
+ "step": 1900
237
+ },
238
+ {
239
+ "epoch": 0.37,
240
+ "learning_rate": 9.631198935968081e-06,
241
+ "loss": 1.0716,
242
+ "step": 1950
243
+ },
244
+ {
245
+ "epoch": 0.38,
246
+ "learning_rate": 9.62169865095953e-06,
247
+ "loss": 1.0847,
248
+ "step": 2000
249
+ },
250
+ {
251
+ "epoch": 0.39,
252
+ "learning_rate": 9.612198365950978e-06,
253
+ "loss": 1.0237,
254
+ "step": 2050
255
+ },
256
+ {
257
+ "epoch": 0.4,
258
+ "learning_rate": 9.60269808094243e-06,
259
+ "loss": 1.0382,
260
+ "step": 2100
261
+ },
262
+ {
263
+ "epoch": 0.41,
264
+ "learning_rate": 9.593197795933879e-06,
265
+ "loss": 1.055,
266
+ "step": 2150
267
+ },
268
+ {
269
+ "epoch": 0.42,
270
+ "learning_rate": 9.583697510925328e-06,
271
+ "loss": 1.0257,
272
+ "step": 2200
273
+ },
274
+ {
275
+ "epoch": 0.43,
276
+ "learning_rate": 9.57438723161695e-06,
277
+ "loss": 1.0425,
278
+ "step": 2250
279
+ },
280
+ {
281
+ "epoch": 0.44,
282
+ "learning_rate": 9.5648869466084e-06,
283
+ "loss": 1.0603,
284
+ "step": 2300
285
+ },
286
+ {
287
+ "epoch": 0.45,
288
+ "learning_rate": 9.555386661599849e-06,
289
+ "loss": 1.0195,
290
+ "step": 2350
291
+ },
292
+ {
293
+ "epoch": 0.46,
294
+ "learning_rate": 9.545886376591299e-06,
295
+ "loss": 1.0311,
296
+ "step": 2400
297
+ },
298
+ {
299
+ "epoch": 0.47,
300
+ "learning_rate": 9.536386091582748e-06,
301
+ "loss": 1.0307,
302
+ "step": 2450
303
+ },
304
+ {
305
+ "epoch": 0.48,
306
+ "learning_rate": 9.526885806574197e-06,
307
+ "loss": 1.0965,
308
+ "step": 2500
309
+ },
310
+ {
311
+ "epoch": 0.48,
312
+ "learning_rate": 9.517385521565647e-06,
313
+ "loss": 1.0271,
314
+ "step": 2550
315
+ },
316
+ {
317
+ "epoch": 0.49,
318
+ "learning_rate": 9.507885236557098e-06,
319
+ "loss": 1.0315,
320
+ "step": 2600
321
+ },
322
+ {
323
+ "epoch": 0.5,
324
+ "learning_rate": 9.498384951548548e-06,
325
+ "loss": 1.0245,
326
+ "step": 2650
327
+ },
328
+ {
329
+ "epoch": 0.51,
330
+ "learning_rate": 9.488884666539997e-06,
331
+ "loss": 1.0208,
332
+ "step": 2700
333
+ },
334
+ {
335
+ "epoch": 0.52,
336
+ "learning_rate": 9.479384381531446e-06,
337
+ "loss": 1.0621,
338
+ "step": 2750
339
+ },
340
+ {
341
+ "epoch": 0.53,
342
+ "learning_rate": 9.469884096522896e-06,
343
+ "loss": 1.0774,
344
+ "step": 2800
345
+ },
346
+ {
347
+ "epoch": 0.54,
348
+ "learning_rate": 9.460383811514347e-06,
349
+ "loss": 1.0727,
350
+ "step": 2850
351
+ },
352
+ {
353
+ "epoch": 0.55,
354
+ "learning_rate": 9.450883526505797e-06,
355
+ "loss": 1.0411,
356
+ "step": 2900
357
+ },
358
+ {
359
+ "epoch": 0.56,
360
+ "learning_rate": 9.441383241497246e-06,
361
+ "loss": 1.0088,
362
+ "step": 2950
363
+ },
364
+ {
365
+ "epoch": 0.57,
366
+ "learning_rate": 9.431882956488696e-06,
367
+ "loss": 1.0133,
368
+ "step": 3000
369
+ },
370
+ {
371
+ "epoch": 0.58,
372
+ "learning_rate": 9.422382671480145e-06,
373
+ "loss": 1.0655,
374
+ "step": 3050
375
+ },
376
+ {
377
+ "epoch": 0.59,
378
+ "learning_rate": 9.412882386471594e-06,
379
+ "loss": 1.0241,
380
+ "step": 3100
381
+ },
382
+ {
383
+ "epoch": 0.6,
384
+ "learning_rate": 9.403382101463044e-06,
385
+ "loss": 1.0708,
386
+ "step": 3150
387
+ },
388
+ {
389
+ "epoch": 0.61,
390
+ "learning_rate": 9.393881816454495e-06,
391
+ "loss": 1.0304,
392
+ "step": 3200
393
+ },
394
+ {
395
+ "epoch": 0.62,
396
+ "learning_rate": 9.384381531445945e-06,
397
+ "loss": 1.0155,
398
+ "step": 3250
399
+ },
400
+ {
401
+ "epoch": 0.63,
402
+ "learning_rate": 9.374881246437394e-06,
403
+ "loss": 1.0224,
404
+ "step": 3300
405
+ },
406
+ {
407
+ "epoch": 0.64,
408
+ "learning_rate": 9.365380961428843e-06,
409
+ "loss": 1.0665,
410
+ "step": 3350
411
+ },
412
+ {
413
+ "epoch": 0.65,
414
+ "learning_rate": 9.355880676420293e-06,
415
+ "loss": 1.0238,
416
+ "step": 3400
417
+ },
418
+ {
419
+ "epoch": 0.66,
420
+ "learning_rate": 9.346380391411744e-06,
421
+ "loss": 1.1006,
422
+ "step": 3450
423
+ },
424
+ {
425
+ "epoch": 0.67,
426
+ "learning_rate": 9.336880106403194e-06,
427
+ "loss": 0.9987,
428
+ "step": 3500
429
+ },
430
+ {
431
+ "epoch": 0.67,
432
+ "learning_rate": 9.327379821394643e-06,
433
+ "loss": 1.0001,
434
+ "step": 3550
435
+ },
436
+ {
437
+ "epoch": 0.68,
438
+ "learning_rate": 9.317879536386093e-06,
439
+ "loss": 1.0634,
440
+ "step": 3600
441
+ },
442
+ {
443
+ "epoch": 0.69,
444
+ "learning_rate": 9.308379251377542e-06,
445
+ "loss": 1.0277,
446
+ "step": 3650
447
+ },
448
+ {
449
+ "epoch": 0.7,
450
+ "learning_rate": 9.298878966368991e-06,
451
+ "loss": 1.0311,
452
+ "step": 3700
453
+ },
454
+ {
455
+ "epoch": 0.71,
456
+ "learning_rate": 9.289378681360441e-06,
457
+ "loss": 1.0317,
458
+ "step": 3750
459
+ },
460
+ {
461
+ "epoch": 0.72,
462
+ "learning_rate": 9.279878396351892e-06,
463
+ "loss": 1.0787,
464
+ "step": 3800
465
+ },
466
+ {
467
+ "epoch": 0.73,
468
+ "learning_rate": 9.270378111343342e-06,
469
+ "loss": 1.0583,
470
+ "step": 3850
471
+ },
472
+ {
473
+ "epoch": 0.74,
474
+ "learning_rate": 9.260877826334791e-06,
475
+ "loss": 1.0326,
476
+ "step": 3900
477
+ },
478
+ {
479
+ "epoch": 0.75,
480
+ "learning_rate": 9.25137754132624e-06,
481
+ "loss": 0.989,
482
+ "step": 3950
483
+ },
484
+ {
485
+ "epoch": 0.76,
486
+ "learning_rate": 9.24187725631769e-06,
487
+ "loss": 0.9879,
488
+ "step": 4000
489
+ },
490
+ {
491
+ "epoch": 0.77,
492
+ "learning_rate": 9.232376971309141e-06,
493
+ "loss": 1.0399,
494
+ "step": 4050
495
+ },
496
+ {
497
+ "epoch": 0.78,
498
+ "learning_rate": 9.22287668630059e-06,
499
+ "loss": 1.0045,
500
+ "step": 4100
501
+ },
502
+ {
503
+ "epoch": 0.79,
504
+ "learning_rate": 9.213376401292038e-06,
505
+ "loss": 1.0079,
506
+ "step": 4150
507
+ },
508
+ {
509
+ "epoch": 0.8,
510
+ "learning_rate": 9.20387611628349e-06,
511
+ "loss": 0.999,
512
+ "step": 4200
513
+ },
514
+ {
515
+ "epoch": 0.81,
516
+ "learning_rate": 9.194375831274939e-06,
517
+ "loss": 0.9867,
518
+ "step": 4250
519
+ },
520
+ {
521
+ "epoch": 0.82,
522
+ "learning_rate": 9.184875546266388e-06,
523
+ "loss": 0.9878,
524
+ "step": 4300
525
+ },
526
+ {
527
+ "epoch": 0.83,
528
+ "learning_rate": 9.175375261257838e-06,
529
+ "loss": 1.0098,
530
+ "step": 4350
531
+ },
532
+ {
533
+ "epoch": 0.84,
534
+ "learning_rate": 9.165874976249289e-06,
535
+ "loss": 1.0325,
536
+ "step": 4400
537
+ },
538
+ {
539
+ "epoch": 0.85,
540
+ "learning_rate": 9.156374691240739e-06,
541
+ "loss": 1.0108,
542
+ "step": 4450
543
+ },
544
+ {
545
+ "epoch": 0.86,
546
+ "learning_rate": 9.146874406232188e-06,
547
+ "loss": 1.0167,
548
+ "step": 4500
549
+ },
550
+ {
551
+ "epoch": 0.86,
552
+ "learning_rate": 9.137374121223637e-06,
553
+ "loss": 0.9871,
554
+ "step": 4550
555
+ },
556
+ {
557
+ "epoch": 0.87,
558
+ "learning_rate": 9.128063841915258e-06,
559
+ "loss": 0.983,
560
+ "step": 4600
561
+ },
562
+ {
563
+ "epoch": 0.88,
564
+ "learning_rate": 9.118563556906707e-06,
565
+ "loss": 0.9807,
566
+ "step": 4650
567
+ },
568
+ {
569
+ "epoch": 0.89,
570
+ "learning_rate": 9.109063271898158e-06,
571
+ "loss": 0.9794,
572
+ "step": 4700
573
+ },
574
+ {
575
+ "epoch": 0.9,
576
+ "learning_rate": 9.099562986889608e-06,
577
+ "loss": 1.0505,
578
+ "step": 4750
579
+ },
580
+ {
581
+ "epoch": 0.91,
582
+ "learning_rate": 9.090062701881057e-06,
583
+ "loss": 1.0033,
584
+ "step": 4800
585
+ },
586
+ {
587
+ "epoch": 0.92,
588
+ "learning_rate": 9.080562416872507e-06,
589
+ "loss": 0.9822,
590
+ "step": 4850
591
+ },
592
+ {
593
+ "epoch": 0.93,
594
+ "learning_rate": 9.071062131863958e-06,
595
+ "loss": 1.0046,
596
+ "step": 4900
597
+ },
598
+ {
599
+ "epoch": 0.94,
600
+ "learning_rate": 9.061561846855407e-06,
601
+ "loss": 0.9944,
602
+ "step": 4950
603
+ },
604
+ {
605
+ "epoch": 0.95,
606
+ "learning_rate": 9.052061561846857e-06,
607
+ "loss": 1.0161,
608
+ "step": 5000
609
+ },
610
+ {
611
+ "epoch": 0.96,
612
+ "learning_rate": 9.042561276838306e-06,
613
+ "loss": 1.0239,
614
+ "step": 5050
615
+ },
616
+ {
617
+ "epoch": 0.97,
618
+ "learning_rate": 9.033060991829756e-06,
619
+ "loss": 0.978,
620
+ "step": 5100
621
+ },
622
+ {
623
+ "epoch": 0.98,
624
+ "learning_rate": 9.023560706821205e-06,
625
+ "loss": 1.0055,
626
+ "step": 5150
627
+ },
628
+ {
629
+ "epoch": 0.99,
630
+ "learning_rate": 9.014060421812655e-06,
631
+ "loss": 0.9623,
632
+ "step": 5200
633
+ },
634
+ {
635
+ "epoch": 1.0,
636
+ "learning_rate": 9.004560136804104e-06,
637
+ "loss": 0.9563,
638
+ "step": 5250
639
+ },
640
+ {
641
+ "epoch": 1.01,
642
+ "learning_rate": 8.995059851795555e-06,
643
+ "loss": 0.9714,
644
+ "step": 5300
645
+ },
646
+ {
647
+ "epoch": 1.02,
648
+ "learning_rate": 8.985559566787005e-06,
649
+ "loss": 0.9661,
650
+ "step": 5350
651
+ },
652
+ {
653
+ "epoch": 1.03,
654
+ "learning_rate": 8.976059281778454e-06,
655
+ "loss": 0.9727,
656
+ "step": 5400
657
+ },
658
+ {
659
+ "epoch": 1.04,
660
+ "learning_rate": 8.966558996769904e-06,
661
+ "loss": 0.9874,
662
+ "step": 5450
663
+ },
664
+ {
665
+ "epoch": 1.05,
666
+ "learning_rate": 8.957058711761355e-06,
667
+ "loss": 0.9978,
668
+ "step": 5500
669
+ },
670
+ {
671
+ "epoch": 1.05,
672
+ "learning_rate": 8.947558426752804e-06,
673
+ "loss": 0.9971,
674
+ "step": 5550
675
+ },
676
+ {
677
+ "epoch": 1.06,
678
+ "learning_rate": 8.938058141744254e-06,
679
+ "loss": 1.017,
680
+ "step": 5600
681
+ },
682
+ {
683
+ "epoch": 1.07,
684
+ "learning_rate": 8.928557856735703e-06,
685
+ "loss": 0.9685,
686
+ "step": 5650
687
+ },
688
+ {
689
+ "epoch": 1.08,
690
+ "learning_rate": 8.919057571727153e-06,
691
+ "loss": 0.9842,
692
+ "step": 5700
693
+ },
694
+ {
695
+ "epoch": 1.09,
696
+ "learning_rate": 8.909557286718602e-06,
697
+ "loss": 0.9858,
698
+ "step": 5750
699
+ },
700
+ {
701
+ "epoch": 1.1,
702
+ "learning_rate": 8.900057001710052e-06,
703
+ "loss": 1.0461,
704
+ "step": 5800
705
+ },
706
+ {
707
+ "epoch": 1.11,
708
+ "learning_rate": 8.890556716701501e-06,
709
+ "loss": 0.9706,
710
+ "step": 5850
711
+ },
712
+ {
713
+ "epoch": 1.12,
714
+ "learning_rate": 8.881056431692952e-06,
715
+ "loss": 0.9727,
716
+ "step": 5900
717
+ },
718
+ {
719
+ "epoch": 1.13,
720
+ "learning_rate": 8.871556146684402e-06,
721
+ "loss": 1.0025,
722
+ "step": 5950
723
+ },
724
+ {
725
+ "epoch": 1.14,
726
+ "learning_rate": 8.862055861675851e-06,
727
+ "loss": 1.0136,
728
+ "step": 6000
729
+ },
730
+ {
731
+ "epoch": 1.15,
732
+ "learning_rate": 8.8525555766673e-06,
733
+ "loss": 0.953,
734
+ "step": 6050
735
+ },
736
+ {
737
+ "epoch": 1.16,
738
+ "learning_rate": 8.843055291658752e-06,
739
+ "loss": 0.986,
740
+ "step": 6100
741
+ },
742
+ {
743
+ "epoch": 1.17,
744
+ "learning_rate": 8.833555006650201e-06,
745
+ "loss": 0.9878,
746
+ "step": 6150
747
+ },
748
+ {
749
+ "epoch": 1.18,
750
+ "learning_rate": 8.824054721641649e-06,
751
+ "loss": 0.9738,
752
+ "step": 6200
753
+ },
754
+ {
755
+ "epoch": 1.19,
756
+ "learning_rate": 8.8145544366331e-06,
757
+ "loss": 0.9894,
758
+ "step": 6250
759
+ },
760
+ {
761
+ "epoch": 1.2,
762
+ "learning_rate": 8.80505415162455e-06,
763
+ "loss": 0.9732,
764
+ "step": 6300
765
+ },
766
+ {
767
+ "epoch": 1.21,
768
+ "learning_rate": 8.795553866615999e-06,
769
+ "loss": 0.9799,
770
+ "step": 6350
771
+ },
772
+ {
773
+ "epoch": 1.22,
774
+ "learning_rate": 8.786053581607449e-06,
775
+ "loss": 1.0178,
776
+ "step": 6400
777
+ },
778
+ {
779
+ "epoch": 1.23,
780
+ "learning_rate": 8.776553296598898e-06,
781
+ "loss": 0.9576,
782
+ "step": 6450
783
+ },
784
+ {
785
+ "epoch": 1.24,
786
+ "learning_rate": 8.767053011590349e-06,
787
+ "loss": 1.0088,
788
+ "step": 6500
789
+ },
790
+ {
791
+ "epoch": 1.24,
792
+ "learning_rate": 8.757552726581799e-06,
793
+ "loss": 0.9657,
794
+ "step": 6550
795
+ },
796
+ {
797
+ "epoch": 1.25,
798
+ "learning_rate": 8.748052441573248e-06,
799
+ "loss": 0.9517,
800
+ "step": 6600
801
+ },
802
+ {
803
+ "epoch": 1.26,
804
+ "learning_rate": 8.738552156564698e-06,
805
+ "loss": 1.0624,
806
+ "step": 6650
807
+ },
808
+ {
809
+ "epoch": 1.27,
810
+ "learning_rate": 8.729051871556149e-06,
811
+ "loss": 1.0197,
812
+ "step": 6700
813
+ },
814
+ {
815
+ "epoch": 1.28,
816
+ "learning_rate": 8.719551586547596e-06,
817
+ "loss": 0.9647,
818
+ "step": 6750
819
+ },
820
+ {
821
+ "epoch": 1.29,
822
+ "learning_rate": 8.710051301539046e-06,
823
+ "loss": 1.0109,
824
+ "step": 6800
825
+ },
826
+ {
827
+ "epoch": 1.3,
828
+ "learning_rate": 8.700741022230668e-06,
829
+ "loss": 0.9923,
830
+ "step": 6850
831
+ },
832
+ {
833
+ "epoch": 1.31,
834
+ "learning_rate": 8.691240737222117e-06,
835
+ "loss": 0.9976,
836
+ "step": 6900
837
+ },
838
+ {
839
+ "epoch": 1.32,
840
+ "learning_rate": 8.681740452213567e-06,
841
+ "loss": 0.9929,
842
+ "step": 6950
843
+ },
844
+ {
845
+ "epoch": 1.33,
846
+ "learning_rate": 8.672240167205018e-06,
847
+ "loss": 0.9783,
848
+ "step": 7000
849
+ },
850
+ {
851
+ "epoch": 1.34,
852
+ "learning_rate": 8.662739882196467e-06,
853
+ "loss": 0.9503,
854
+ "step": 7050
855
+ },
856
+ {
857
+ "epoch": 1.35,
858
+ "learning_rate": 8.653239597187917e-06,
859
+ "loss": 0.9915,
860
+ "step": 7100
861
+ },
862
+ {
863
+ "epoch": 1.36,
864
+ "learning_rate": 8.643739312179366e-06,
865
+ "loss": 0.9502,
866
+ "step": 7150
867
+ },
868
+ {
869
+ "epoch": 1.37,
870
+ "learning_rate": 8.634239027170816e-06,
871
+ "loss": 0.9867,
872
+ "step": 7200
873
+ },
874
+ {
875
+ "epoch": 1.38,
876
+ "learning_rate": 8.624738742162265e-06,
877
+ "loss": 0.9929,
878
+ "step": 7250
879
+ },
880
+ {
881
+ "epoch": 1.39,
882
+ "learning_rate": 8.615238457153715e-06,
883
+ "loss": 1.0304,
884
+ "step": 7300
885
+ },
886
+ {
887
+ "epoch": 1.4,
888
+ "learning_rate": 8.605738172145164e-06,
889
+ "loss": 0.9841,
890
+ "step": 7350
891
+ },
892
+ {
893
+ "epoch": 1.41,
894
+ "learning_rate": 8.596237887136615e-06,
895
+ "loss": 0.9915,
896
+ "step": 7400
897
+ },
898
+ {
899
+ "epoch": 1.42,
900
+ "learning_rate": 8.586737602128065e-06,
901
+ "loss": 0.9691,
902
+ "step": 7450
903
+ },
904
+ {
905
+ "epoch": 1.43,
906
+ "learning_rate": 8.577237317119514e-06,
907
+ "loss": 0.9655,
908
+ "step": 7500
909
+ },
910
+ {
911
+ "epoch": 1.43,
912
+ "learning_rate": 8.567737032110964e-06,
913
+ "loss": 0.9884,
914
+ "step": 7550
915
+ },
916
+ {
917
+ "epoch": 1.44,
918
+ "learning_rate": 8.558236747102415e-06,
919
+ "loss": 0.9736,
920
+ "step": 7600
921
+ },
922
+ {
923
+ "epoch": 1.45,
924
+ "learning_rate": 8.548736462093864e-06,
925
+ "loss": 0.9462,
926
+ "step": 7650
927
+ },
928
+ {
929
+ "epoch": 1.46,
930
+ "learning_rate": 8.539236177085312e-06,
931
+ "loss": 1.0023,
932
+ "step": 7700
933
+ },
934
+ {
935
+ "epoch": 1.47,
936
+ "learning_rate": 8.529735892076763e-06,
937
+ "loss": 0.9696,
938
+ "step": 7750
939
+ },
940
+ {
941
+ "epoch": 1.48,
942
+ "learning_rate": 8.520235607068213e-06,
943
+ "loss": 0.9498,
944
+ "step": 7800
945
+ },
946
+ {
947
+ "epoch": 1.49,
948
+ "learning_rate": 8.510735322059662e-06,
949
+ "loss": 0.9689,
950
+ "step": 7850
951
+ },
952
+ {
953
+ "epoch": 1.5,
954
+ "learning_rate": 8.501235037051112e-06,
955
+ "loss": 1.0175,
956
+ "step": 7900
957
+ },
958
+ {
959
+ "epoch": 1.51,
960
+ "learning_rate": 8.491734752042561e-06,
961
+ "loss": 0.9783,
962
+ "step": 7950
963
+ },
964
+ {
965
+ "epoch": 1.52,
966
+ "learning_rate": 8.482234467034012e-06,
967
+ "loss": 0.9942,
968
+ "step": 8000
969
+ },
970
+ {
971
+ "epoch": 1.53,
972
+ "learning_rate": 8.472734182025462e-06,
973
+ "loss": 0.9631,
974
+ "step": 8050
975
+ },
976
+ {
977
+ "epoch": 1.54,
978
+ "learning_rate": 8.463233897016911e-06,
979
+ "loss": 0.9417,
980
+ "step": 8100
981
+ },
982
+ {
983
+ "epoch": 1.55,
984
+ "learning_rate": 8.45373361200836e-06,
985
+ "loss": 0.9449,
986
+ "step": 8150
987
+ },
988
+ {
989
+ "epoch": 1.56,
990
+ "learning_rate": 8.444233326999812e-06,
991
+ "loss": 0.986,
992
+ "step": 8200
993
+ },
994
+ {
995
+ "epoch": 1.57,
996
+ "learning_rate": 8.434733041991261e-06,
997
+ "loss": 0.9353,
998
+ "step": 8250
999
+ },
1000
+ {
1001
+ "epoch": 1.58,
1002
+ "learning_rate": 8.425232756982709e-06,
1003
+ "loss": 0.9693,
1004
+ "step": 8300
1005
+ },
1006
+ {
1007
+ "epoch": 1.59,
1008
+ "learning_rate": 8.41573247197416e-06,
1009
+ "loss": 0.966,
1010
+ "step": 8350
1011
+ },
1012
+ {
1013
+ "epoch": 1.6,
1014
+ "learning_rate": 8.40623218696561e-06,
1015
+ "loss": 0.968,
1016
+ "step": 8400
1017
+ },
1018
+ {
1019
+ "epoch": 1.61,
1020
+ "learning_rate": 8.396731901957059e-06,
1021
+ "loss": 0.9226,
1022
+ "step": 8450
1023
+ },
1024
+ {
1025
+ "epoch": 1.62,
1026
+ "learning_rate": 8.387231616948509e-06,
1027
+ "loss": 0.9361,
1028
+ "step": 8500
1029
+ },
1030
+ {
1031
+ "epoch": 1.62,
1032
+ "learning_rate": 8.377731331939958e-06,
1033
+ "loss": 0.9552,
1034
+ "step": 8550
1035
+ },
1036
+ {
1037
+ "epoch": 1.63,
1038
+ "learning_rate": 8.36823104693141e-06,
1039
+ "loss": 0.9482,
1040
+ "step": 8600
1041
+ },
1042
+ {
1043
+ "epoch": 1.64,
1044
+ "learning_rate": 8.358730761922859e-06,
1045
+ "loss": 0.968,
1046
+ "step": 8650
1047
+ },
1048
+ {
1049
+ "epoch": 1.65,
1050
+ "learning_rate": 8.349230476914308e-06,
1051
+ "loss": 0.9786,
1052
+ "step": 8700
1053
+ },
1054
+ {
1055
+ "epoch": 1.66,
1056
+ "learning_rate": 8.339730191905758e-06,
1057
+ "loss": 0.9615,
1058
+ "step": 8750
1059
+ },
1060
+ {
1061
+ "epoch": 1.67,
1062
+ "learning_rate": 8.330229906897209e-06,
1063
+ "loss": 0.964,
1064
+ "step": 8800
1065
+ },
1066
+ {
1067
+ "epoch": 1.68,
1068
+ "learning_rate": 8.320729621888657e-06,
1069
+ "loss": 0.9676,
1070
+ "step": 8850
1071
+ },
1072
+ {
1073
+ "epoch": 1.69,
1074
+ "learning_rate": 8.311229336880106e-06,
1075
+ "loss": 0.9532,
1076
+ "step": 8900
1077
+ },
1078
+ {
1079
+ "epoch": 1.7,
1080
+ "learning_rate": 8.301729051871557e-06,
1081
+ "loss": 0.9506,
1082
+ "step": 8950
1083
+ },
1084
+ {
1085
+ "epoch": 1.71,
1086
+ "learning_rate": 8.292418772563177e-06,
1087
+ "loss": 0.9513,
1088
+ "step": 9000
1089
+ },
1090
+ {
1091
+ "epoch": 1.72,
1092
+ "learning_rate": 8.282918487554627e-06,
1093
+ "loss": 0.9185,
1094
+ "step": 9050
1095
+ },
1096
+ {
1097
+ "epoch": 1.73,
1098
+ "learning_rate": 8.273418202546078e-06,
1099
+ "loss": 0.9706,
1100
+ "step": 9100
1101
+ },
1102
+ {
1103
+ "epoch": 1.74,
1104
+ "learning_rate": 8.263917917537527e-06,
1105
+ "loss": 0.9235,
1106
+ "step": 9150
1107
+ },
1108
+ {
1109
+ "epoch": 1.75,
1110
+ "learning_rate": 8.254417632528977e-06,
1111
+ "loss": 0.958,
1112
+ "step": 9200
1113
+ },
1114
+ {
1115
+ "epoch": 1.76,
1116
+ "learning_rate": 8.244917347520426e-06,
1117
+ "loss": 0.9587,
1118
+ "step": 9250
1119
+ },
1120
+ {
1121
+ "epoch": 1.77,
1122
+ "learning_rate": 8.235417062511876e-06,
1123
+ "loss": 0.9674,
1124
+ "step": 9300
1125
+ },
1126
+ {
1127
+ "epoch": 1.78,
1128
+ "learning_rate": 8.225916777503325e-06,
1129
+ "loss": 0.9496,
1130
+ "step": 9350
1131
+ },
1132
+ {
1133
+ "epoch": 1.79,
1134
+ "learning_rate": 8.216416492494775e-06,
1135
+ "loss": 0.9579,
1136
+ "step": 9400
1137
+ },
1138
+ {
1139
+ "epoch": 1.8,
1140
+ "learning_rate": 8.206916207486226e-06,
1141
+ "loss": 0.9654,
1142
+ "step": 9450
1143
+ },
1144
+ {
1145
+ "epoch": 1.81,
1146
+ "learning_rate": 8.197415922477675e-06,
1147
+ "loss": 0.9548,
1148
+ "step": 9500
1149
+ },
1150
+ {
1151
+ "epoch": 1.81,
1152
+ "learning_rate": 8.187915637469125e-06,
1153
+ "loss": 0.9631,
1154
+ "step": 9550
1155
+ },
1156
+ {
1157
+ "epoch": 1.82,
1158
+ "learning_rate": 8.178415352460574e-06,
1159
+ "loss": 0.9395,
1160
+ "step": 9600
1161
+ },
1162
+ {
1163
+ "epoch": 1.83,
1164
+ "learning_rate": 8.168915067452024e-06,
1165
+ "loss": 0.9646,
1166
+ "step": 9650
1167
+ },
1168
+ {
1169
+ "epoch": 1.84,
1170
+ "learning_rate": 8.159414782443475e-06,
1171
+ "loss": 0.9239,
1172
+ "step": 9700
1173
+ },
1174
+ {
1175
+ "epoch": 1.85,
1176
+ "learning_rate": 8.149914497434924e-06,
1177
+ "loss": 0.9699,
1178
+ "step": 9750
1179
+ },
1180
+ {
1181
+ "epoch": 1.86,
1182
+ "learning_rate": 8.140414212426372e-06,
1183
+ "loss": 0.9407,
1184
+ "step": 9800
1185
+ },
1186
+ {
1187
+ "epoch": 1.87,
1188
+ "learning_rate": 8.130913927417823e-06,
1189
+ "loss": 0.9519,
1190
+ "step": 9850
1191
+ },
1192
+ {
1193
+ "epoch": 1.88,
1194
+ "learning_rate": 8.121413642409273e-06,
1195
+ "loss": 0.9786,
1196
+ "step": 9900
1197
+ },
1198
+ {
1199
+ "epoch": 1.89,
1200
+ "learning_rate": 8.111913357400722e-06,
1201
+ "loss": 0.9691,
1202
+ "step": 9950
1203
+ },
1204
+ {
1205
+ "epoch": 1.9,
1206
+ "learning_rate": 8.102413072392172e-06,
1207
+ "loss": 0.9379,
1208
+ "step": 10000
1209
+ },
1210
+ {
1211
+ "epoch": 1.91,
1212
+ "learning_rate": 8.092912787383623e-06,
1213
+ "loss": 0.9676,
1214
+ "step": 10050
1215
+ },
1216
+ {
1217
+ "epoch": 1.92,
1218
+ "learning_rate": 8.083412502375072e-06,
1219
+ "loss": 0.9751,
1220
+ "step": 10100
1221
+ },
1222
+ {
1223
+ "epoch": 1.93,
1224
+ "learning_rate": 8.073912217366522e-06,
1225
+ "loss": 0.9705,
1226
+ "step": 10150
1227
+ },
1228
+ {
1229
+ "epoch": 1.94,
1230
+ "learning_rate": 8.064411932357971e-06,
1231
+ "loss": 0.9586,
1232
+ "step": 10200
1233
+ },
1234
+ {
1235
+ "epoch": 1.95,
1236
+ "learning_rate": 8.05491164734942e-06,
1237
+ "loss": 0.9573,
1238
+ "step": 10250
1239
+ },
1240
+ {
1241
+ "epoch": 1.96,
1242
+ "learning_rate": 8.045411362340872e-06,
1243
+ "loss": 0.9581,
1244
+ "step": 10300
1245
+ },
1246
+ {
1247
+ "epoch": 1.97,
1248
+ "learning_rate": 8.03591107733232e-06,
1249
+ "loss": 0.9454,
1250
+ "step": 10350
1251
+ },
1252
+ {
1253
+ "epoch": 1.98,
1254
+ "learning_rate": 8.026410792323769e-06,
1255
+ "loss": 0.948,
1256
+ "step": 10400
1257
+ },
1258
+ {
1259
+ "epoch": 1.99,
1260
+ "learning_rate": 8.01691050731522e-06,
1261
+ "loss": 0.9652,
1262
+ "step": 10450
1263
+ },
1264
+ {
1265
+ "epoch": 2.0,
1266
+ "learning_rate": 8.00741022230667e-06,
1267
+ "loss": 0.9722,
1268
+ "step": 10500
1269
+ },
1270
+ {
1271
+ "epoch": 2.0,
1272
+ "learning_rate": 7.99790993729812e-06,
1273
+ "loss": 0.9283,
1274
+ "step": 10550
1275
+ },
1276
+ {
1277
+ "epoch": 2.01,
1278
+ "learning_rate": 7.988409652289569e-06,
1279
+ "loss": 0.9244,
1280
+ "step": 10600
1281
+ },
1282
+ {
1283
+ "epoch": 2.02,
1284
+ "learning_rate": 7.97890936728102e-06,
1285
+ "loss": 0.9315,
1286
+ "step": 10650
1287
+ },
1288
+ {
1289
+ "epoch": 2.03,
1290
+ "learning_rate": 7.96940908227247e-06,
1291
+ "loss": 0.9376,
1292
+ "step": 10700
1293
+ },
1294
+ {
1295
+ "epoch": 2.04,
1296
+ "learning_rate": 7.959908797263919e-06,
1297
+ "loss": 0.9886,
1298
+ "step": 10750
1299
+ },
1300
+ {
1301
+ "epoch": 2.05,
1302
+ "learning_rate": 7.950408512255368e-06,
1303
+ "loss": 0.9419,
1304
+ "step": 10800
1305
+ },
1306
+ {
1307
+ "epoch": 2.06,
1308
+ "learning_rate": 7.940908227246818e-06,
1309
+ "loss": 0.9548,
1310
+ "step": 10850
1311
+ },
1312
+ {
1313
+ "epoch": 2.07,
1314
+ "learning_rate": 7.931407942238267e-06,
1315
+ "loss": 0.9429,
1316
+ "step": 10900
1317
+ },
1318
+ {
1319
+ "epoch": 2.08,
1320
+ "learning_rate": 7.921907657229717e-06,
1321
+ "loss": 0.9772,
1322
+ "step": 10950
1323
+ },
1324
+ {
1325
+ "epoch": 2.09,
1326
+ "learning_rate": 7.912597377921338e-06,
1327
+ "loss": 0.9443,
1328
+ "step": 11000
1329
+ },
1330
+ {
1331
+ "epoch": 2.1,
1332
+ "learning_rate": 7.903097092912788e-06,
1333
+ "loss": 0.9411,
1334
+ "step": 11050
1335
+ },
1336
+ {
1337
+ "epoch": 2.11,
1338
+ "learning_rate": 7.893596807904237e-06,
1339
+ "loss": 0.9222,
1340
+ "step": 11100
1341
+ },
1342
+ {
1343
+ "epoch": 2.12,
1344
+ "learning_rate": 7.884096522895689e-06,
1345
+ "loss": 0.9312,
1346
+ "step": 11150
1347
+ },
1348
+ {
1349
+ "epoch": 2.13,
1350
+ "learning_rate": 7.874596237887138e-06,
1351
+ "loss": 0.9408,
1352
+ "step": 11200
1353
+ },
1354
+ {
1355
+ "epoch": 2.14,
1356
+ "learning_rate": 7.865095952878587e-06,
1357
+ "loss": 0.9164,
1358
+ "step": 11250
1359
+ },
1360
+ {
1361
+ "epoch": 2.15,
1362
+ "learning_rate": 7.855595667870037e-06,
1363
+ "loss": 0.9823,
1364
+ "step": 11300
1365
+ },
1366
+ {
1367
+ "epoch": 2.16,
1368
+ "learning_rate": 7.846095382861486e-06,
1369
+ "loss": 0.9457,
1370
+ "step": 11350
1371
+ },
1372
+ {
1373
+ "epoch": 2.17,
1374
+ "learning_rate": 7.836595097852936e-06,
1375
+ "loss": 0.9605,
1376
+ "step": 11400
1377
+ },
1378
+ {
1379
+ "epoch": 2.18,
1380
+ "learning_rate": 7.827094812844385e-06,
1381
+ "loss": 0.9283,
1382
+ "step": 11450
1383
+ },
1384
+ {
1385
+ "epoch": 2.19,
1386
+ "learning_rate": 7.817594527835835e-06,
1387
+ "loss": 0.9115,
1388
+ "step": 11500
1389
+ },
1390
+ {
1391
+ "epoch": 2.19,
1392
+ "learning_rate": 7.808094242827286e-06,
1393
+ "loss": 0.9508,
1394
+ "step": 11550
1395
+ },
1396
+ {
1397
+ "epoch": 2.2,
1398
+ "learning_rate": 7.798593957818735e-06,
1399
+ "loss": 0.9419,
1400
+ "step": 11600
1401
+ },
1402
+ {
1403
+ "epoch": 2.21,
1404
+ "learning_rate": 7.789093672810185e-06,
1405
+ "loss": 0.9078,
1406
+ "step": 11650
1407
+ },
1408
+ {
1409
+ "epoch": 2.22,
1410
+ "learning_rate": 7.779593387801634e-06,
1411
+ "loss": 0.9313,
1412
+ "step": 11700
1413
+ },
1414
+ {
1415
+ "epoch": 2.23,
1416
+ "learning_rate": 7.770093102793086e-06,
1417
+ "loss": 0.9452,
1418
+ "step": 11750
1419
+ },
1420
+ {
1421
+ "epoch": 2.24,
1422
+ "learning_rate": 7.760592817784535e-06,
1423
+ "loss": 0.9401,
1424
+ "step": 11800
1425
+ },
1426
+ {
1427
+ "epoch": 2.25,
1428
+ "learning_rate": 7.751092532775983e-06,
1429
+ "loss": 0.9278,
1430
+ "step": 11850
1431
+ },
1432
+ {
1433
+ "epoch": 2.26,
1434
+ "learning_rate": 7.741592247767434e-06,
1435
+ "loss": 0.931,
1436
+ "step": 11900
1437
+ },
1438
+ {
1439
+ "epoch": 2.27,
1440
+ "learning_rate": 7.732091962758883e-06,
1441
+ "loss": 0.9436,
1442
+ "step": 11950
1443
+ },
1444
+ {
1445
+ "epoch": 2.28,
1446
+ "learning_rate": 7.722591677750333e-06,
1447
+ "loss": 0.9212,
1448
+ "step": 12000
1449
+ },
1450
+ {
1451
+ "epoch": 2.29,
1452
+ "learning_rate": 7.713091392741782e-06,
1453
+ "loss": 0.95,
1454
+ "step": 12050
1455
+ },
1456
+ {
1457
+ "epoch": 2.3,
1458
+ "learning_rate": 7.703591107733232e-06,
1459
+ "loss": 0.9701,
1460
+ "step": 12100
1461
+ },
1462
+ {
1463
+ "epoch": 2.31,
1464
+ "learning_rate": 7.694090822724683e-06,
1465
+ "loss": 0.9402,
1466
+ "step": 12150
1467
+ },
1468
+ {
1469
+ "epoch": 2.32,
1470
+ "learning_rate": 7.684590537716132e-06,
1471
+ "loss": 0.9419,
1472
+ "step": 12200
1473
+ },
1474
+ {
1475
+ "epoch": 2.33,
1476
+ "learning_rate": 7.675090252707582e-06,
1477
+ "loss": 0.9436,
1478
+ "step": 12250
1479
+ },
1480
+ {
1481
+ "epoch": 2.34,
1482
+ "learning_rate": 7.665589967699031e-06,
1483
+ "loss": 0.9464,
1484
+ "step": 12300
1485
+ },
1486
+ {
1487
+ "epoch": 2.35,
1488
+ "learning_rate": 7.656089682690483e-06,
1489
+ "loss": 0.939,
1490
+ "step": 12350
1491
+ },
1492
+ {
1493
+ "epoch": 2.36,
1494
+ "learning_rate": 7.64658939768193e-06,
1495
+ "loss": 0.9839,
1496
+ "step": 12400
1497
+ },
1498
+ {
1499
+ "epoch": 2.37,
1500
+ "learning_rate": 7.63708911267338e-06,
1501
+ "loss": 0.9176,
1502
+ "step": 12450
1503
+ },
1504
+ {
1505
+ "epoch": 2.38,
1506
+ "learning_rate": 7.62758882766483e-06,
1507
+ "loss": 0.8935,
1508
+ "step": 12500
1509
+ },
1510
+ {
1511
+ "epoch": 2.38,
1512
+ "learning_rate": 7.61808854265628e-06,
1513
+ "loss": 0.9215,
1514
+ "step": 12550
1515
+ },
1516
+ {
1517
+ "epoch": 2.39,
1518
+ "learning_rate": 7.60858825764773e-06,
1519
+ "loss": 0.9403,
1520
+ "step": 12600
1521
+ },
1522
+ {
1523
+ "epoch": 2.4,
1524
+ "learning_rate": 7.599087972639179e-06,
1525
+ "loss": 0.9272,
1526
+ "step": 12650
1527
+ },
1528
+ {
1529
+ "epoch": 2.41,
1530
+ "learning_rate": 7.58958768763063e-06,
1531
+ "loss": 0.9528,
1532
+ "step": 12700
1533
+ },
1534
+ {
1535
+ "epoch": 2.42,
1536
+ "learning_rate": 7.580087402622079e-06,
1537
+ "loss": 0.9039,
1538
+ "step": 12750
1539
+ },
1540
+ {
1541
+ "epoch": 2.43,
1542
+ "learning_rate": 7.570587117613529e-06,
1543
+ "loss": 0.9194,
1544
+ "step": 12800
1545
+ },
1546
+ {
1547
+ "epoch": 2.44,
1548
+ "learning_rate": 7.561086832604979e-06,
1549
+ "loss": 0.9306,
1550
+ "step": 12850
1551
+ },
1552
+ {
1553
+ "epoch": 2.45,
1554
+ "learning_rate": 7.551586547596429e-06,
1555
+ "loss": 0.9242,
1556
+ "step": 12900
1557
+ },
1558
+ {
1559
+ "epoch": 2.46,
1560
+ "learning_rate": 7.542086262587879e-06,
1561
+ "loss": 0.8961,
1562
+ "step": 12950
1563
+ },
1564
+ {
1565
+ "epoch": 2.47,
1566
+ "learning_rate": 7.532585977579327e-06,
1567
+ "loss": 0.936,
1568
+ "step": 13000
1569
+ },
1570
+ {
1571
+ "epoch": 2.48,
1572
+ "learning_rate": 7.523085692570778e-06,
1573
+ "loss": 0.9598,
1574
+ "step": 13050
1575
+ },
1576
+ {
1577
+ "epoch": 2.49,
1578
+ "learning_rate": 7.513585407562227e-06,
1579
+ "loss": 0.9114,
1580
+ "step": 13100
1581
+ },
1582
+ {
1583
+ "epoch": 2.5,
1584
+ "learning_rate": 7.504085122553677e-06,
1585
+ "loss": 0.9217,
1586
+ "step": 13150
1587
+ },
1588
+ {
1589
+ "epoch": 2.51,
1590
+ "learning_rate": 7.494584837545127e-06,
1591
+ "loss": 0.9041,
1592
+ "step": 13200
1593
+ }
1594
+ ],
1595
+ "max_steps": 52630,
1596
+ "num_train_epochs": 10,
1597
+ "total_flos": 5.046728076278784e+19,
1598
+ "trial_name": null,
1599
+ "trial_params": null
1600
+ }
checkpoint-13200/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5edc01452a06bceea0b7057733be6a8f12700dfc6f213c8585ae1fc40583b0b
3
+ size 4088
checkpoint-13400/config.json ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": "b05d55b99887e2a2713474fd1590c11ff0a45a94",
3
+ "_name_or_path": "Tech-Hans/detr-resnet-50_finetuned_fashion",
4
+ "activation_dropout": 0.0,
5
+ "activation_function": "relu",
6
+ "architectures": [
7
+ "DetrForObjectDetection"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "auxiliary_loss": false,
11
+ "backbone": "resnet50",
12
+ "backbone_config": null,
13
+ "bbox_cost": 5,
14
+ "bbox_loss_coefficient": 5,
15
+ "class_cost": 1,
16
+ "classifier_dropout": 0.0,
17
+ "d_model": 256,
18
+ "decoder_attention_heads": 8,
19
+ "decoder_ffn_dim": 2048,
20
+ "decoder_layerdrop": 0.0,
21
+ "decoder_layers": 6,
22
+ "dice_loss_coefficient": 1,
23
+ "dilation": false,
24
+ "dropout": 0.1,
25
+ "encoder_attention_heads": 8,
26
+ "encoder_ffn_dim": 2048,
27
+ "encoder_layerdrop": 0.0,
28
+ "encoder_layers": 6,
29
+ "eos_coefficient": 0.1,
30
+ "giou_cost": 2,
31
+ "giou_loss_coefficient": 2,
32
+ "id2label": {
33
+ "0": "bags",
34
+ "1": "belt",
35
+ "2": "bottom",
36
+ "3": "fullbody",
37
+ "4": "glasses",
38
+ "5": "glove",
39
+ "6": "hat",
40
+ "7": "scarf",
41
+ "8": "shoes",
42
+ "9": "sock",
43
+ "10": "tie",
44
+ "11": "tights",
45
+ "12": "top"
46
+ },
47
+ "init_std": 0.02,
48
+ "init_xavier_std": 1.0,
49
+ "is_encoder_decoder": true,
50
+ "label2id": {
51
+ "bags": 0,
52
+ "belt": 1,
53
+ "bottom": 2,
54
+ "fullbody": 3,
55
+ "glasses": 4,
56
+ "glove": 5,
57
+ "hat": 6,
58
+ "scarf": 7,
59
+ "shoes": 8,
60
+ "sock": 9,
61
+ "tie": 10,
62
+ "tights": 11,
63
+ "top": 12
64
+ },
65
+ "mask_loss_coefficient": 1,
66
+ "max_position_embeddings": 1024,
67
+ "model_type": "detr",
68
+ "num_channels": 3,
69
+ "num_hidden_layers": 6,
70
+ "num_queries": 100,
71
+ "position_embedding_type": "sine",
72
+ "scale_embedding": false,
73
+ "torch_dtype": "float32",
74
+ "transformers_version": "4.7.0.dev0",
75
+ "use_pretrained_backbone": true,
76
+ "use_timm_backbone": true
77
+ }
checkpoint-13400/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb11b50882dfbd499ee6661a52c36b4ec4b515016250631233beed00e7e5835c
3
+ size 330501178
checkpoint-13400/preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_pad": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "format": "coco_detection",
7
+ "image_mean": [
8
+ 0.485,
9
+ 0.456,
10
+ 0.406
11
+ ],
12
+ "image_processor_type": "DetrImageProcessor",
13
+ "image_std": [
14
+ 0.229,
15
+ 0.224,
16
+ 0.225
17
+ ],
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "longest_edge": 1333,
22
+ "shortest_edge": 800
23
+ }
24
+ }
checkpoint-13400/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0af4f81c1895814163d1a974dca6d7a50bea06a94b0e684a5bbba66fb9b630f5
3
+ size 166628834
checkpoint-13400/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d9aee7830494f703ea64bf951cd07ca9989aa7d23181f77f31aacd96b13d923
3
+ size 14180
checkpoint-13400/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:583e48613255236d90b71180d149fbf856200dc946a482051a06d61c5867436e
3
+ size 988
checkpoint-13400/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9efec96c2d632f1a394256a03f37ec7d28055381f3c8b7367c8b08ffd1d58e8
3
+ size 1064
checkpoint-13400/trainer_state.json ADDED
@@ -0,0 +1,1624 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.546076382291469,
5
+ "global_step": 13400,
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": 9.992019760592819e-06,
13
+ "loss": 1.1564,
14
+ "step": 50
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 9.982709481284439e-06,
19
+ "loss": 1.132,
20
+ "step": 100
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 9.97320919627589e-06,
25
+ "loss": 1.1507,
26
+ "step": 150
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 9.96370891126734e-06,
31
+ "loss": 1.0846,
32
+ "step": 200
33
+ },
34
+ {
35
+ "epoch": 0.05,
36
+ "learning_rate": 9.954208626258789e-06,
37
+ "loss": 1.0877,
38
+ "step": 250
39
+ },
40
+ {
41
+ "epoch": 0.06,
42
+ "learning_rate": 9.944708341250238e-06,
43
+ "loss": 1.1153,
44
+ "step": 300
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "learning_rate": 9.935208056241688e-06,
49
+ "loss": 1.0876,
50
+ "step": 350
51
+ },
52
+ {
53
+ "epoch": 0.08,
54
+ "learning_rate": 9.925707771233137e-06,
55
+ "loss": 1.0729,
56
+ "step": 400
57
+ },
58
+ {
59
+ "epoch": 0.09,
60
+ "learning_rate": 9.916207486224587e-06,
61
+ "loss": 1.126,
62
+ "step": 450
63
+ },
64
+ {
65
+ "epoch": 0.1,
66
+ "learning_rate": 9.906707201216036e-06,
67
+ "loss": 1.1262,
68
+ "step": 500
69
+ },
70
+ {
71
+ "epoch": 0.1,
72
+ "learning_rate": 9.897206916207487e-06,
73
+ "loss": 1.0883,
74
+ "step": 550
75
+ },
76
+ {
77
+ "epoch": 0.11,
78
+ "learning_rate": 9.887706631198937e-06,
79
+ "loss": 1.0926,
80
+ "step": 600
81
+ },
82
+ {
83
+ "epoch": 0.12,
84
+ "learning_rate": 9.878206346190386e-06,
85
+ "loss": 1.0916,
86
+ "step": 650
87
+ },
88
+ {
89
+ "epoch": 0.13,
90
+ "learning_rate": 9.868706061181836e-06,
91
+ "loss": 1.0892,
92
+ "step": 700
93
+ },
94
+ {
95
+ "epoch": 0.14,
96
+ "learning_rate": 9.859205776173287e-06,
97
+ "loss": 1.0653,
98
+ "step": 750
99
+ },
100
+ {
101
+ "epoch": 0.15,
102
+ "learning_rate": 9.849705491164736e-06,
103
+ "loss": 1.0391,
104
+ "step": 800
105
+ },
106
+ {
107
+ "epoch": 0.16,
108
+ "learning_rate": 9.840205206156186e-06,
109
+ "loss": 1.0866,
110
+ "step": 850
111
+ },
112
+ {
113
+ "epoch": 0.17,
114
+ "learning_rate": 9.830704921147635e-06,
115
+ "loss": 1.1192,
116
+ "step": 900
117
+ },
118
+ {
119
+ "epoch": 0.18,
120
+ "learning_rate": 9.821204636139085e-06,
121
+ "loss": 1.0898,
122
+ "step": 950
123
+ },
124
+ {
125
+ "epoch": 0.19,
126
+ "learning_rate": 9.811704351130534e-06,
127
+ "loss": 1.0693,
128
+ "step": 1000
129
+ },
130
+ {
131
+ "epoch": 0.2,
132
+ "learning_rate": 9.802204066121984e-06,
133
+ "loss": 1.1243,
134
+ "step": 1050
135
+ },
136
+ {
137
+ "epoch": 0.21,
138
+ "learning_rate": 9.792703781113433e-06,
139
+ "loss": 1.071,
140
+ "step": 1100
141
+ },
142
+ {
143
+ "epoch": 0.22,
144
+ "learning_rate": 9.783203496104884e-06,
145
+ "loss": 1.0569,
146
+ "step": 1150
147
+ },
148
+ {
149
+ "epoch": 0.23,
150
+ "learning_rate": 9.773703211096334e-06,
151
+ "loss": 1.0612,
152
+ "step": 1200
153
+ },
154
+ {
155
+ "epoch": 0.24,
156
+ "learning_rate": 9.764202926087783e-06,
157
+ "loss": 1.0432,
158
+ "step": 1250
159
+ },
160
+ {
161
+ "epoch": 0.25,
162
+ "learning_rate": 9.754702641079233e-06,
163
+ "loss": 1.092,
164
+ "step": 1300
165
+ },
166
+ {
167
+ "epoch": 0.26,
168
+ "learning_rate": 9.745202356070684e-06,
169
+ "loss": 1.0602,
170
+ "step": 1350
171
+ },
172
+ {
173
+ "epoch": 0.27,
174
+ "learning_rate": 9.735702071062133e-06,
175
+ "loss": 1.0689,
176
+ "step": 1400
177
+ },
178
+ {
179
+ "epoch": 0.28,
180
+ "learning_rate": 9.726201786053583e-06,
181
+ "loss": 1.0897,
182
+ "step": 1450
183
+ },
184
+ {
185
+ "epoch": 0.29,
186
+ "learning_rate": 9.716701501045032e-06,
187
+ "loss": 1.0482,
188
+ "step": 1500
189
+ },
190
+ {
191
+ "epoch": 0.29,
192
+ "learning_rate": 9.707201216036482e-06,
193
+ "loss": 1.1167,
194
+ "step": 1550
195
+ },
196
+ {
197
+ "epoch": 0.3,
198
+ "learning_rate": 9.697700931027931e-06,
199
+ "loss": 1.0707,
200
+ "step": 1600
201
+ },
202
+ {
203
+ "epoch": 0.31,
204
+ "learning_rate": 9.68820064601938e-06,
205
+ "loss": 1.0613,
206
+ "step": 1650
207
+ },
208
+ {
209
+ "epoch": 0.32,
210
+ "learning_rate": 9.67870036101083e-06,
211
+ "loss": 1.0534,
212
+ "step": 1700
213
+ },
214
+ {
215
+ "epoch": 0.33,
216
+ "learning_rate": 9.669200076002281e-06,
217
+ "loss": 1.0719,
218
+ "step": 1750
219
+ },
220
+ {
221
+ "epoch": 0.34,
222
+ "learning_rate": 9.659699790993731e-06,
223
+ "loss": 1.0678,
224
+ "step": 1800
225
+ },
226
+ {
227
+ "epoch": 0.35,
228
+ "learning_rate": 9.65019950598518e-06,
229
+ "loss": 1.064,
230
+ "step": 1850
231
+ },
232
+ {
233
+ "epoch": 0.36,
234
+ "learning_rate": 9.64069922097663e-06,
235
+ "loss": 1.0846,
236
+ "step": 1900
237
+ },
238
+ {
239
+ "epoch": 0.37,
240
+ "learning_rate": 9.631198935968081e-06,
241
+ "loss": 1.0716,
242
+ "step": 1950
243
+ },
244
+ {
245
+ "epoch": 0.38,
246
+ "learning_rate": 9.62169865095953e-06,
247
+ "loss": 1.0847,
248
+ "step": 2000
249
+ },
250
+ {
251
+ "epoch": 0.39,
252
+ "learning_rate": 9.612198365950978e-06,
253
+ "loss": 1.0237,
254
+ "step": 2050
255
+ },
256
+ {
257
+ "epoch": 0.4,
258
+ "learning_rate": 9.60269808094243e-06,
259
+ "loss": 1.0382,
260
+ "step": 2100
261
+ },
262
+ {
263
+ "epoch": 0.41,
264
+ "learning_rate": 9.593197795933879e-06,
265
+ "loss": 1.055,
266
+ "step": 2150
267
+ },
268
+ {
269
+ "epoch": 0.42,
270
+ "learning_rate": 9.583697510925328e-06,
271
+ "loss": 1.0257,
272
+ "step": 2200
273
+ },
274
+ {
275
+ "epoch": 0.43,
276
+ "learning_rate": 9.57438723161695e-06,
277
+ "loss": 1.0425,
278
+ "step": 2250
279
+ },
280
+ {
281
+ "epoch": 0.44,
282
+ "learning_rate": 9.5648869466084e-06,
283
+ "loss": 1.0603,
284
+ "step": 2300
285
+ },
286
+ {
287
+ "epoch": 0.45,
288
+ "learning_rate": 9.555386661599849e-06,
289
+ "loss": 1.0195,
290
+ "step": 2350
291
+ },
292
+ {
293
+ "epoch": 0.46,
294
+ "learning_rate": 9.545886376591299e-06,
295
+ "loss": 1.0311,
296
+ "step": 2400
297
+ },
298
+ {
299
+ "epoch": 0.47,
300
+ "learning_rate": 9.536386091582748e-06,
301
+ "loss": 1.0307,
302
+ "step": 2450
303
+ },
304
+ {
305
+ "epoch": 0.48,
306
+ "learning_rate": 9.526885806574197e-06,
307
+ "loss": 1.0965,
308
+ "step": 2500
309
+ },
310
+ {
311
+ "epoch": 0.48,
312
+ "learning_rate": 9.517385521565647e-06,
313
+ "loss": 1.0271,
314
+ "step": 2550
315
+ },
316
+ {
317
+ "epoch": 0.49,
318
+ "learning_rate": 9.507885236557098e-06,
319
+ "loss": 1.0315,
320
+ "step": 2600
321
+ },
322
+ {
323
+ "epoch": 0.5,
324
+ "learning_rate": 9.498384951548548e-06,
325
+ "loss": 1.0245,
326
+ "step": 2650
327
+ },
328
+ {
329
+ "epoch": 0.51,
330
+ "learning_rate": 9.488884666539997e-06,
331
+ "loss": 1.0208,
332
+ "step": 2700
333
+ },
334
+ {
335
+ "epoch": 0.52,
336
+ "learning_rate": 9.479384381531446e-06,
337
+ "loss": 1.0621,
338
+ "step": 2750
339
+ },
340
+ {
341
+ "epoch": 0.53,
342
+ "learning_rate": 9.469884096522896e-06,
343
+ "loss": 1.0774,
344
+ "step": 2800
345
+ },
346
+ {
347
+ "epoch": 0.54,
348
+ "learning_rate": 9.460383811514347e-06,
349
+ "loss": 1.0727,
350
+ "step": 2850
351
+ },
352
+ {
353
+ "epoch": 0.55,
354
+ "learning_rate": 9.450883526505797e-06,
355
+ "loss": 1.0411,
356
+ "step": 2900
357
+ },
358
+ {
359
+ "epoch": 0.56,
360
+ "learning_rate": 9.441383241497246e-06,
361
+ "loss": 1.0088,
362
+ "step": 2950
363
+ },
364
+ {
365
+ "epoch": 0.57,
366
+ "learning_rate": 9.431882956488696e-06,
367
+ "loss": 1.0133,
368
+ "step": 3000
369
+ },
370
+ {
371
+ "epoch": 0.58,
372
+ "learning_rate": 9.422382671480145e-06,
373
+ "loss": 1.0655,
374
+ "step": 3050
375
+ },
376
+ {
377
+ "epoch": 0.59,
378
+ "learning_rate": 9.412882386471594e-06,
379
+ "loss": 1.0241,
380
+ "step": 3100
381
+ },
382
+ {
383
+ "epoch": 0.6,
384
+ "learning_rate": 9.403382101463044e-06,
385
+ "loss": 1.0708,
386
+ "step": 3150
387
+ },
388
+ {
389
+ "epoch": 0.61,
390
+ "learning_rate": 9.393881816454495e-06,
391
+ "loss": 1.0304,
392
+ "step": 3200
393
+ },
394
+ {
395
+ "epoch": 0.62,
396
+ "learning_rate": 9.384381531445945e-06,
397
+ "loss": 1.0155,
398
+ "step": 3250
399
+ },
400
+ {
401
+ "epoch": 0.63,
402
+ "learning_rate": 9.374881246437394e-06,
403
+ "loss": 1.0224,
404
+ "step": 3300
405
+ },
406
+ {
407
+ "epoch": 0.64,
408
+ "learning_rate": 9.365380961428843e-06,
409
+ "loss": 1.0665,
410
+ "step": 3350
411
+ },
412
+ {
413
+ "epoch": 0.65,
414
+ "learning_rate": 9.355880676420293e-06,
415
+ "loss": 1.0238,
416
+ "step": 3400
417
+ },
418
+ {
419
+ "epoch": 0.66,
420
+ "learning_rate": 9.346380391411744e-06,
421
+ "loss": 1.1006,
422
+ "step": 3450
423
+ },
424
+ {
425
+ "epoch": 0.67,
426
+ "learning_rate": 9.336880106403194e-06,
427
+ "loss": 0.9987,
428
+ "step": 3500
429
+ },
430
+ {
431
+ "epoch": 0.67,
432
+ "learning_rate": 9.327379821394643e-06,
433
+ "loss": 1.0001,
434
+ "step": 3550
435
+ },
436
+ {
437
+ "epoch": 0.68,
438
+ "learning_rate": 9.317879536386093e-06,
439
+ "loss": 1.0634,
440
+ "step": 3600
441
+ },
442
+ {
443
+ "epoch": 0.69,
444
+ "learning_rate": 9.308379251377542e-06,
445
+ "loss": 1.0277,
446
+ "step": 3650
447
+ },
448
+ {
449
+ "epoch": 0.7,
450
+ "learning_rate": 9.298878966368991e-06,
451
+ "loss": 1.0311,
452
+ "step": 3700
453
+ },
454
+ {
455
+ "epoch": 0.71,
456
+ "learning_rate": 9.289378681360441e-06,
457
+ "loss": 1.0317,
458
+ "step": 3750
459
+ },
460
+ {
461
+ "epoch": 0.72,
462
+ "learning_rate": 9.279878396351892e-06,
463
+ "loss": 1.0787,
464
+ "step": 3800
465
+ },
466
+ {
467
+ "epoch": 0.73,
468
+ "learning_rate": 9.270378111343342e-06,
469
+ "loss": 1.0583,
470
+ "step": 3850
471
+ },
472
+ {
473
+ "epoch": 0.74,
474
+ "learning_rate": 9.260877826334791e-06,
475
+ "loss": 1.0326,
476
+ "step": 3900
477
+ },
478
+ {
479
+ "epoch": 0.75,
480
+ "learning_rate": 9.25137754132624e-06,
481
+ "loss": 0.989,
482
+ "step": 3950
483
+ },
484
+ {
485
+ "epoch": 0.76,
486
+ "learning_rate": 9.24187725631769e-06,
487
+ "loss": 0.9879,
488
+ "step": 4000
489
+ },
490
+ {
491
+ "epoch": 0.77,
492
+ "learning_rate": 9.232376971309141e-06,
493
+ "loss": 1.0399,
494
+ "step": 4050
495
+ },
496
+ {
497
+ "epoch": 0.78,
498
+ "learning_rate": 9.22287668630059e-06,
499
+ "loss": 1.0045,
500
+ "step": 4100
501
+ },
502
+ {
503
+ "epoch": 0.79,
504
+ "learning_rate": 9.213376401292038e-06,
505
+ "loss": 1.0079,
506
+ "step": 4150
507
+ },
508
+ {
509
+ "epoch": 0.8,
510
+ "learning_rate": 9.20387611628349e-06,
511
+ "loss": 0.999,
512
+ "step": 4200
513
+ },
514
+ {
515
+ "epoch": 0.81,
516
+ "learning_rate": 9.194375831274939e-06,
517
+ "loss": 0.9867,
518
+ "step": 4250
519
+ },
520
+ {
521
+ "epoch": 0.82,
522
+ "learning_rate": 9.184875546266388e-06,
523
+ "loss": 0.9878,
524
+ "step": 4300
525
+ },
526
+ {
527
+ "epoch": 0.83,
528
+ "learning_rate": 9.175375261257838e-06,
529
+ "loss": 1.0098,
530
+ "step": 4350
531
+ },
532
+ {
533
+ "epoch": 0.84,
534
+ "learning_rate": 9.165874976249289e-06,
535
+ "loss": 1.0325,
536
+ "step": 4400
537
+ },
538
+ {
539
+ "epoch": 0.85,
540
+ "learning_rate": 9.156374691240739e-06,
541
+ "loss": 1.0108,
542
+ "step": 4450
543
+ },
544
+ {
545
+ "epoch": 0.86,
546
+ "learning_rate": 9.146874406232188e-06,
547
+ "loss": 1.0167,
548
+ "step": 4500
549
+ },
550
+ {
551
+ "epoch": 0.86,
552
+ "learning_rate": 9.137374121223637e-06,
553
+ "loss": 0.9871,
554
+ "step": 4550
555
+ },
556
+ {
557
+ "epoch": 0.87,
558
+ "learning_rate": 9.128063841915258e-06,
559
+ "loss": 0.983,
560
+ "step": 4600
561
+ },
562
+ {
563
+ "epoch": 0.88,
564
+ "learning_rate": 9.118563556906707e-06,
565
+ "loss": 0.9807,
566
+ "step": 4650
567
+ },
568
+ {
569
+ "epoch": 0.89,
570
+ "learning_rate": 9.109063271898158e-06,
571
+ "loss": 0.9794,
572
+ "step": 4700
573
+ },
574
+ {
575
+ "epoch": 0.9,
576
+ "learning_rate": 9.099562986889608e-06,
577
+ "loss": 1.0505,
578
+ "step": 4750
579
+ },
580
+ {
581
+ "epoch": 0.91,
582
+ "learning_rate": 9.090062701881057e-06,
583
+ "loss": 1.0033,
584
+ "step": 4800
585
+ },
586
+ {
587
+ "epoch": 0.92,
588
+ "learning_rate": 9.080562416872507e-06,
589
+ "loss": 0.9822,
590
+ "step": 4850
591
+ },
592
+ {
593
+ "epoch": 0.93,
594
+ "learning_rate": 9.071062131863958e-06,
595
+ "loss": 1.0046,
596
+ "step": 4900
597
+ },
598
+ {
599
+ "epoch": 0.94,
600
+ "learning_rate": 9.061561846855407e-06,
601
+ "loss": 0.9944,
602
+ "step": 4950
603
+ },
604
+ {
605
+ "epoch": 0.95,
606
+ "learning_rate": 9.052061561846857e-06,
607
+ "loss": 1.0161,
608
+ "step": 5000
609
+ },
610
+ {
611
+ "epoch": 0.96,
612
+ "learning_rate": 9.042561276838306e-06,
613
+ "loss": 1.0239,
614
+ "step": 5050
615
+ },
616
+ {
617
+ "epoch": 0.97,
618
+ "learning_rate": 9.033060991829756e-06,
619
+ "loss": 0.978,
620
+ "step": 5100
621
+ },
622
+ {
623
+ "epoch": 0.98,
624
+ "learning_rate": 9.023560706821205e-06,
625
+ "loss": 1.0055,
626
+ "step": 5150
627
+ },
628
+ {
629
+ "epoch": 0.99,
630
+ "learning_rate": 9.014060421812655e-06,
631
+ "loss": 0.9623,
632
+ "step": 5200
633
+ },
634
+ {
635
+ "epoch": 1.0,
636
+ "learning_rate": 9.004560136804104e-06,
637
+ "loss": 0.9563,
638
+ "step": 5250
639
+ },
640
+ {
641
+ "epoch": 1.01,
642
+ "learning_rate": 8.995059851795555e-06,
643
+ "loss": 0.9714,
644
+ "step": 5300
645
+ },
646
+ {
647
+ "epoch": 1.02,
648
+ "learning_rate": 8.985559566787005e-06,
649
+ "loss": 0.9661,
650
+ "step": 5350
651
+ },
652
+ {
653
+ "epoch": 1.03,
654
+ "learning_rate": 8.976059281778454e-06,
655
+ "loss": 0.9727,
656
+ "step": 5400
657
+ },
658
+ {
659
+ "epoch": 1.04,
660
+ "learning_rate": 8.966558996769904e-06,
661
+ "loss": 0.9874,
662
+ "step": 5450
663
+ },
664
+ {
665
+ "epoch": 1.05,
666
+ "learning_rate": 8.957058711761355e-06,
667
+ "loss": 0.9978,
668
+ "step": 5500
669
+ },
670
+ {
671
+ "epoch": 1.05,
672
+ "learning_rate": 8.947558426752804e-06,
673
+ "loss": 0.9971,
674
+ "step": 5550
675
+ },
676
+ {
677
+ "epoch": 1.06,
678
+ "learning_rate": 8.938058141744254e-06,
679
+ "loss": 1.017,
680
+ "step": 5600
681
+ },
682
+ {
683
+ "epoch": 1.07,
684
+ "learning_rate": 8.928557856735703e-06,
685
+ "loss": 0.9685,
686
+ "step": 5650
687
+ },
688
+ {
689
+ "epoch": 1.08,
690
+ "learning_rate": 8.919057571727153e-06,
691
+ "loss": 0.9842,
692
+ "step": 5700
693
+ },
694
+ {
695
+ "epoch": 1.09,
696
+ "learning_rate": 8.909557286718602e-06,
697
+ "loss": 0.9858,
698
+ "step": 5750
699
+ },
700
+ {
701
+ "epoch": 1.1,
702
+ "learning_rate": 8.900057001710052e-06,
703
+ "loss": 1.0461,
704
+ "step": 5800
705
+ },
706
+ {
707
+ "epoch": 1.11,
708
+ "learning_rate": 8.890556716701501e-06,
709
+ "loss": 0.9706,
710
+ "step": 5850
711
+ },
712
+ {
713
+ "epoch": 1.12,
714
+ "learning_rate": 8.881056431692952e-06,
715
+ "loss": 0.9727,
716
+ "step": 5900
717
+ },
718
+ {
719
+ "epoch": 1.13,
720
+ "learning_rate": 8.871556146684402e-06,
721
+ "loss": 1.0025,
722
+ "step": 5950
723
+ },
724
+ {
725
+ "epoch": 1.14,
726
+ "learning_rate": 8.862055861675851e-06,
727
+ "loss": 1.0136,
728
+ "step": 6000
729
+ },
730
+ {
731
+ "epoch": 1.15,
732
+ "learning_rate": 8.8525555766673e-06,
733
+ "loss": 0.953,
734
+ "step": 6050
735
+ },
736
+ {
737
+ "epoch": 1.16,
738
+ "learning_rate": 8.843055291658752e-06,
739
+ "loss": 0.986,
740
+ "step": 6100
741
+ },
742
+ {
743
+ "epoch": 1.17,
744
+ "learning_rate": 8.833555006650201e-06,
745
+ "loss": 0.9878,
746
+ "step": 6150
747
+ },
748
+ {
749
+ "epoch": 1.18,
750
+ "learning_rate": 8.824054721641649e-06,
751
+ "loss": 0.9738,
752
+ "step": 6200
753
+ },
754
+ {
755
+ "epoch": 1.19,
756
+ "learning_rate": 8.8145544366331e-06,
757
+ "loss": 0.9894,
758
+ "step": 6250
759
+ },
760
+ {
761
+ "epoch": 1.2,
762
+ "learning_rate": 8.80505415162455e-06,
763
+ "loss": 0.9732,
764
+ "step": 6300
765
+ },
766
+ {
767
+ "epoch": 1.21,
768
+ "learning_rate": 8.795553866615999e-06,
769
+ "loss": 0.9799,
770
+ "step": 6350
771
+ },
772
+ {
773
+ "epoch": 1.22,
774
+ "learning_rate": 8.786053581607449e-06,
775
+ "loss": 1.0178,
776
+ "step": 6400
777
+ },
778
+ {
779
+ "epoch": 1.23,
780
+ "learning_rate": 8.776553296598898e-06,
781
+ "loss": 0.9576,
782
+ "step": 6450
783
+ },
784
+ {
785
+ "epoch": 1.24,
786
+ "learning_rate": 8.767053011590349e-06,
787
+ "loss": 1.0088,
788
+ "step": 6500
789
+ },
790
+ {
791
+ "epoch": 1.24,
792
+ "learning_rate": 8.757552726581799e-06,
793
+ "loss": 0.9657,
794
+ "step": 6550
795
+ },
796
+ {
797
+ "epoch": 1.25,
798
+ "learning_rate": 8.748052441573248e-06,
799
+ "loss": 0.9517,
800
+ "step": 6600
801
+ },
802
+ {
803
+ "epoch": 1.26,
804
+ "learning_rate": 8.738552156564698e-06,
805
+ "loss": 1.0624,
806
+ "step": 6650
807
+ },
808
+ {
809
+ "epoch": 1.27,
810
+ "learning_rate": 8.729051871556149e-06,
811
+ "loss": 1.0197,
812
+ "step": 6700
813
+ },
814
+ {
815
+ "epoch": 1.28,
816
+ "learning_rate": 8.719551586547596e-06,
817
+ "loss": 0.9647,
818
+ "step": 6750
819
+ },
820
+ {
821
+ "epoch": 1.29,
822
+ "learning_rate": 8.710051301539046e-06,
823
+ "loss": 1.0109,
824
+ "step": 6800
825
+ },
826
+ {
827
+ "epoch": 1.3,
828
+ "learning_rate": 8.700741022230668e-06,
829
+ "loss": 0.9923,
830
+ "step": 6850
831
+ },
832
+ {
833
+ "epoch": 1.31,
834
+ "learning_rate": 8.691240737222117e-06,
835
+ "loss": 0.9976,
836
+ "step": 6900
837
+ },
838
+ {
839
+ "epoch": 1.32,
840
+ "learning_rate": 8.681740452213567e-06,
841
+ "loss": 0.9929,
842
+ "step": 6950
843
+ },
844
+ {
845
+ "epoch": 1.33,
846
+ "learning_rate": 8.672240167205018e-06,
847
+ "loss": 0.9783,
848
+ "step": 7000
849
+ },
850
+ {
851
+ "epoch": 1.34,
852
+ "learning_rate": 8.662739882196467e-06,
853
+ "loss": 0.9503,
854
+ "step": 7050
855
+ },
856
+ {
857
+ "epoch": 1.35,
858
+ "learning_rate": 8.653239597187917e-06,
859
+ "loss": 0.9915,
860
+ "step": 7100
861
+ },
862
+ {
863
+ "epoch": 1.36,
864
+ "learning_rate": 8.643739312179366e-06,
865
+ "loss": 0.9502,
866
+ "step": 7150
867
+ },
868
+ {
869
+ "epoch": 1.37,
870
+ "learning_rate": 8.634239027170816e-06,
871
+ "loss": 0.9867,
872
+ "step": 7200
873
+ },
874
+ {
875
+ "epoch": 1.38,
876
+ "learning_rate": 8.624738742162265e-06,
877
+ "loss": 0.9929,
878
+ "step": 7250
879
+ },
880
+ {
881
+ "epoch": 1.39,
882
+ "learning_rate": 8.615238457153715e-06,
883
+ "loss": 1.0304,
884
+ "step": 7300
885
+ },
886
+ {
887
+ "epoch": 1.4,
888
+ "learning_rate": 8.605738172145164e-06,
889
+ "loss": 0.9841,
890
+ "step": 7350
891
+ },
892
+ {
893
+ "epoch": 1.41,
894
+ "learning_rate": 8.596237887136615e-06,
895
+ "loss": 0.9915,
896
+ "step": 7400
897
+ },
898
+ {
899
+ "epoch": 1.42,
900
+ "learning_rate": 8.586737602128065e-06,
901
+ "loss": 0.9691,
902
+ "step": 7450
903
+ },
904
+ {
905
+ "epoch": 1.43,
906
+ "learning_rate": 8.577237317119514e-06,
907
+ "loss": 0.9655,
908
+ "step": 7500
909
+ },
910
+ {
911
+ "epoch": 1.43,
912
+ "learning_rate": 8.567737032110964e-06,
913
+ "loss": 0.9884,
914
+ "step": 7550
915
+ },
916
+ {
917
+ "epoch": 1.44,
918
+ "learning_rate": 8.558236747102415e-06,
919
+ "loss": 0.9736,
920
+ "step": 7600
921
+ },
922
+ {
923
+ "epoch": 1.45,
924
+ "learning_rate": 8.548736462093864e-06,
925
+ "loss": 0.9462,
926
+ "step": 7650
927
+ },
928
+ {
929
+ "epoch": 1.46,
930
+ "learning_rate": 8.539236177085312e-06,
931
+ "loss": 1.0023,
932
+ "step": 7700
933
+ },
934
+ {
935
+ "epoch": 1.47,
936
+ "learning_rate": 8.529735892076763e-06,
937
+ "loss": 0.9696,
938
+ "step": 7750
939
+ },
940
+ {
941
+ "epoch": 1.48,
942
+ "learning_rate": 8.520235607068213e-06,
943
+ "loss": 0.9498,
944
+ "step": 7800
945
+ },
946
+ {
947
+ "epoch": 1.49,
948
+ "learning_rate": 8.510735322059662e-06,
949
+ "loss": 0.9689,
950
+ "step": 7850
951
+ },
952
+ {
953
+ "epoch": 1.5,
954
+ "learning_rate": 8.501235037051112e-06,
955
+ "loss": 1.0175,
956
+ "step": 7900
957
+ },
958
+ {
959
+ "epoch": 1.51,
960
+ "learning_rate": 8.491734752042561e-06,
961
+ "loss": 0.9783,
962
+ "step": 7950
963
+ },
964
+ {
965
+ "epoch": 1.52,
966
+ "learning_rate": 8.482234467034012e-06,
967
+ "loss": 0.9942,
968
+ "step": 8000
969
+ },
970
+ {
971
+ "epoch": 1.53,
972
+ "learning_rate": 8.472734182025462e-06,
973
+ "loss": 0.9631,
974
+ "step": 8050
975
+ },
976
+ {
977
+ "epoch": 1.54,
978
+ "learning_rate": 8.463233897016911e-06,
979
+ "loss": 0.9417,
980
+ "step": 8100
981
+ },
982
+ {
983
+ "epoch": 1.55,
984
+ "learning_rate": 8.45373361200836e-06,
985
+ "loss": 0.9449,
986
+ "step": 8150
987
+ },
988
+ {
989
+ "epoch": 1.56,
990
+ "learning_rate": 8.444233326999812e-06,
991
+ "loss": 0.986,
992
+ "step": 8200
993
+ },
994
+ {
995
+ "epoch": 1.57,
996
+ "learning_rate": 8.434733041991261e-06,
997
+ "loss": 0.9353,
998
+ "step": 8250
999
+ },
1000
+ {
1001
+ "epoch": 1.58,
1002
+ "learning_rate": 8.425232756982709e-06,
1003
+ "loss": 0.9693,
1004
+ "step": 8300
1005
+ },
1006
+ {
1007
+ "epoch": 1.59,
1008
+ "learning_rate": 8.41573247197416e-06,
1009
+ "loss": 0.966,
1010
+ "step": 8350
1011
+ },
1012
+ {
1013
+ "epoch": 1.6,
1014
+ "learning_rate": 8.40623218696561e-06,
1015
+ "loss": 0.968,
1016
+ "step": 8400
1017
+ },
1018
+ {
1019
+ "epoch": 1.61,
1020
+ "learning_rate": 8.396731901957059e-06,
1021
+ "loss": 0.9226,
1022
+ "step": 8450
1023
+ },
1024
+ {
1025
+ "epoch": 1.62,
1026
+ "learning_rate": 8.387231616948509e-06,
1027
+ "loss": 0.9361,
1028
+ "step": 8500
1029
+ },
1030
+ {
1031
+ "epoch": 1.62,
1032
+ "learning_rate": 8.377731331939958e-06,
1033
+ "loss": 0.9552,
1034
+ "step": 8550
1035
+ },
1036
+ {
1037
+ "epoch": 1.63,
1038
+ "learning_rate": 8.36823104693141e-06,
1039
+ "loss": 0.9482,
1040
+ "step": 8600
1041
+ },
1042
+ {
1043
+ "epoch": 1.64,
1044
+ "learning_rate": 8.358730761922859e-06,
1045
+ "loss": 0.968,
1046
+ "step": 8650
1047
+ },
1048
+ {
1049
+ "epoch": 1.65,
1050
+ "learning_rate": 8.349230476914308e-06,
1051
+ "loss": 0.9786,
1052
+ "step": 8700
1053
+ },
1054
+ {
1055
+ "epoch": 1.66,
1056
+ "learning_rate": 8.339730191905758e-06,
1057
+ "loss": 0.9615,
1058
+ "step": 8750
1059
+ },
1060
+ {
1061
+ "epoch": 1.67,
1062
+ "learning_rate": 8.330229906897209e-06,
1063
+ "loss": 0.964,
1064
+ "step": 8800
1065
+ },
1066
+ {
1067
+ "epoch": 1.68,
1068
+ "learning_rate": 8.320729621888657e-06,
1069
+ "loss": 0.9676,
1070
+ "step": 8850
1071
+ },
1072
+ {
1073
+ "epoch": 1.69,
1074
+ "learning_rate": 8.311229336880106e-06,
1075
+ "loss": 0.9532,
1076
+ "step": 8900
1077
+ },
1078
+ {
1079
+ "epoch": 1.7,
1080
+ "learning_rate": 8.301729051871557e-06,
1081
+ "loss": 0.9506,
1082
+ "step": 8950
1083
+ },
1084
+ {
1085
+ "epoch": 1.71,
1086
+ "learning_rate": 8.292418772563177e-06,
1087
+ "loss": 0.9513,
1088
+ "step": 9000
1089
+ },
1090
+ {
1091
+ "epoch": 1.72,
1092
+ "learning_rate": 8.282918487554627e-06,
1093
+ "loss": 0.9185,
1094
+ "step": 9050
1095
+ },
1096
+ {
1097
+ "epoch": 1.73,
1098
+ "learning_rate": 8.273418202546078e-06,
1099
+ "loss": 0.9706,
1100
+ "step": 9100
1101
+ },
1102
+ {
1103
+ "epoch": 1.74,
1104
+ "learning_rate": 8.263917917537527e-06,
1105
+ "loss": 0.9235,
1106
+ "step": 9150
1107
+ },
1108
+ {
1109
+ "epoch": 1.75,
1110
+ "learning_rate": 8.254417632528977e-06,
1111
+ "loss": 0.958,
1112
+ "step": 9200
1113
+ },
1114
+ {
1115
+ "epoch": 1.76,
1116
+ "learning_rate": 8.244917347520426e-06,
1117
+ "loss": 0.9587,
1118
+ "step": 9250
1119
+ },
1120
+ {
1121
+ "epoch": 1.77,
1122
+ "learning_rate": 8.235417062511876e-06,
1123
+ "loss": 0.9674,
1124
+ "step": 9300
1125
+ },
1126
+ {
1127
+ "epoch": 1.78,
1128
+ "learning_rate": 8.225916777503325e-06,
1129
+ "loss": 0.9496,
1130
+ "step": 9350
1131
+ },
1132
+ {
1133
+ "epoch": 1.79,
1134
+ "learning_rate": 8.216416492494775e-06,
1135
+ "loss": 0.9579,
1136
+ "step": 9400
1137
+ },
1138
+ {
1139
+ "epoch": 1.8,
1140
+ "learning_rate": 8.206916207486226e-06,
1141
+ "loss": 0.9654,
1142
+ "step": 9450
1143
+ },
1144
+ {
1145
+ "epoch": 1.81,
1146
+ "learning_rate": 8.197415922477675e-06,
1147
+ "loss": 0.9548,
1148
+ "step": 9500
1149
+ },
1150
+ {
1151
+ "epoch": 1.81,
1152
+ "learning_rate": 8.187915637469125e-06,
1153
+ "loss": 0.9631,
1154
+ "step": 9550
1155
+ },
1156
+ {
1157
+ "epoch": 1.82,
1158
+ "learning_rate": 8.178415352460574e-06,
1159
+ "loss": 0.9395,
1160
+ "step": 9600
1161
+ },
1162
+ {
1163
+ "epoch": 1.83,
1164
+ "learning_rate": 8.168915067452024e-06,
1165
+ "loss": 0.9646,
1166
+ "step": 9650
1167
+ },
1168
+ {
1169
+ "epoch": 1.84,
1170
+ "learning_rate": 8.159414782443475e-06,
1171
+ "loss": 0.9239,
1172
+ "step": 9700
1173
+ },
1174
+ {
1175
+ "epoch": 1.85,
1176
+ "learning_rate": 8.149914497434924e-06,
1177
+ "loss": 0.9699,
1178
+ "step": 9750
1179
+ },
1180
+ {
1181
+ "epoch": 1.86,
1182
+ "learning_rate": 8.140414212426372e-06,
1183
+ "loss": 0.9407,
1184
+ "step": 9800
1185
+ },
1186
+ {
1187
+ "epoch": 1.87,
1188
+ "learning_rate": 8.130913927417823e-06,
1189
+ "loss": 0.9519,
1190
+ "step": 9850
1191
+ },
1192
+ {
1193
+ "epoch": 1.88,
1194
+ "learning_rate": 8.121413642409273e-06,
1195
+ "loss": 0.9786,
1196
+ "step": 9900
1197
+ },
1198
+ {
1199
+ "epoch": 1.89,
1200
+ "learning_rate": 8.111913357400722e-06,
1201
+ "loss": 0.9691,
1202
+ "step": 9950
1203
+ },
1204
+ {
1205
+ "epoch": 1.9,
1206
+ "learning_rate": 8.102413072392172e-06,
1207
+ "loss": 0.9379,
1208
+ "step": 10000
1209
+ },
1210
+ {
1211
+ "epoch": 1.91,
1212
+ "learning_rate": 8.092912787383623e-06,
1213
+ "loss": 0.9676,
1214
+ "step": 10050
1215
+ },
1216
+ {
1217
+ "epoch": 1.92,
1218
+ "learning_rate": 8.083412502375072e-06,
1219
+ "loss": 0.9751,
1220
+ "step": 10100
1221
+ },
1222
+ {
1223
+ "epoch": 1.93,
1224
+ "learning_rate": 8.073912217366522e-06,
1225
+ "loss": 0.9705,
1226
+ "step": 10150
1227
+ },
1228
+ {
1229
+ "epoch": 1.94,
1230
+ "learning_rate": 8.064411932357971e-06,
1231
+ "loss": 0.9586,
1232
+ "step": 10200
1233
+ },
1234
+ {
1235
+ "epoch": 1.95,
1236
+ "learning_rate": 8.05491164734942e-06,
1237
+ "loss": 0.9573,
1238
+ "step": 10250
1239
+ },
1240
+ {
1241
+ "epoch": 1.96,
1242
+ "learning_rate": 8.045411362340872e-06,
1243
+ "loss": 0.9581,
1244
+ "step": 10300
1245
+ },
1246
+ {
1247
+ "epoch": 1.97,
1248
+ "learning_rate": 8.03591107733232e-06,
1249
+ "loss": 0.9454,
1250
+ "step": 10350
1251
+ },
1252
+ {
1253
+ "epoch": 1.98,
1254
+ "learning_rate": 8.026410792323769e-06,
1255
+ "loss": 0.948,
1256
+ "step": 10400
1257
+ },
1258
+ {
1259
+ "epoch": 1.99,
1260
+ "learning_rate": 8.01691050731522e-06,
1261
+ "loss": 0.9652,
1262
+ "step": 10450
1263
+ },
1264
+ {
1265
+ "epoch": 2.0,
1266
+ "learning_rate": 8.00741022230667e-06,
1267
+ "loss": 0.9722,
1268
+ "step": 10500
1269
+ },
1270
+ {
1271
+ "epoch": 2.0,
1272
+ "learning_rate": 7.99790993729812e-06,
1273
+ "loss": 0.9283,
1274
+ "step": 10550
1275
+ },
1276
+ {
1277
+ "epoch": 2.01,
1278
+ "learning_rate": 7.988409652289569e-06,
1279
+ "loss": 0.9244,
1280
+ "step": 10600
1281
+ },
1282
+ {
1283
+ "epoch": 2.02,
1284
+ "learning_rate": 7.97890936728102e-06,
1285
+ "loss": 0.9315,
1286
+ "step": 10650
1287
+ },
1288
+ {
1289
+ "epoch": 2.03,
1290
+ "learning_rate": 7.96940908227247e-06,
1291
+ "loss": 0.9376,
1292
+ "step": 10700
1293
+ },
1294
+ {
1295
+ "epoch": 2.04,
1296
+ "learning_rate": 7.959908797263919e-06,
1297
+ "loss": 0.9886,
1298
+ "step": 10750
1299
+ },
1300
+ {
1301
+ "epoch": 2.05,
1302
+ "learning_rate": 7.950408512255368e-06,
1303
+ "loss": 0.9419,
1304
+ "step": 10800
1305
+ },
1306
+ {
1307
+ "epoch": 2.06,
1308
+ "learning_rate": 7.940908227246818e-06,
1309
+ "loss": 0.9548,
1310
+ "step": 10850
1311
+ },
1312
+ {
1313
+ "epoch": 2.07,
1314
+ "learning_rate": 7.931407942238267e-06,
1315
+ "loss": 0.9429,
1316
+ "step": 10900
1317
+ },
1318
+ {
1319
+ "epoch": 2.08,
1320
+ "learning_rate": 7.921907657229717e-06,
1321
+ "loss": 0.9772,
1322
+ "step": 10950
1323
+ },
1324
+ {
1325
+ "epoch": 2.09,
1326
+ "learning_rate": 7.912597377921338e-06,
1327
+ "loss": 0.9443,
1328
+ "step": 11000
1329
+ },
1330
+ {
1331
+ "epoch": 2.1,
1332
+ "learning_rate": 7.903097092912788e-06,
1333
+ "loss": 0.9411,
1334
+ "step": 11050
1335
+ },
1336
+ {
1337
+ "epoch": 2.11,
1338
+ "learning_rate": 7.893596807904237e-06,
1339
+ "loss": 0.9222,
1340
+ "step": 11100
1341
+ },
1342
+ {
1343
+ "epoch": 2.12,
1344
+ "learning_rate": 7.884096522895689e-06,
1345
+ "loss": 0.9312,
1346
+ "step": 11150
1347
+ },
1348
+ {
1349
+ "epoch": 2.13,
1350
+ "learning_rate": 7.874596237887138e-06,
1351
+ "loss": 0.9408,
1352
+ "step": 11200
1353
+ },
1354
+ {
1355
+ "epoch": 2.14,
1356
+ "learning_rate": 7.865095952878587e-06,
1357
+ "loss": 0.9164,
1358
+ "step": 11250
1359
+ },
1360
+ {
1361
+ "epoch": 2.15,
1362
+ "learning_rate": 7.855595667870037e-06,
1363
+ "loss": 0.9823,
1364
+ "step": 11300
1365
+ },
1366
+ {
1367
+ "epoch": 2.16,
1368
+ "learning_rate": 7.846095382861486e-06,
1369
+ "loss": 0.9457,
1370
+ "step": 11350
1371
+ },
1372
+ {
1373
+ "epoch": 2.17,
1374
+ "learning_rate": 7.836595097852936e-06,
1375
+ "loss": 0.9605,
1376
+ "step": 11400
1377
+ },
1378
+ {
1379
+ "epoch": 2.18,
1380
+ "learning_rate": 7.827094812844385e-06,
1381
+ "loss": 0.9283,
1382
+ "step": 11450
1383
+ },
1384
+ {
1385
+ "epoch": 2.19,
1386
+ "learning_rate": 7.817594527835835e-06,
1387
+ "loss": 0.9115,
1388
+ "step": 11500
1389
+ },
1390
+ {
1391
+ "epoch": 2.19,
1392
+ "learning_rate": 7.808094242827286e-06,
1393
+ "loss": 0.9508,
1394
+ "step": 11550
1395
+ },
1396
+ {
1397
+ "epoch": 2.2,
1398
+ "learning_rate": 7.798593957818735e-06,
1399
+ "loss": 0.9419,
1400
+ "step": 11600
1401
+ },
1402
+ {
1403
+ "epoch": 2.21,
1404
+ "learning_rate": 7.789093672810185e-06,
1405
+ "loss": 0.9078,
1406
+ "step": 11650
1407
+ },
1408
+ {
1409
+ "epoch": 2.22,
1410
+ "learning_rate": 7.779593387801634e-06,
1411
+ "loss": 0.9313,
1412
+ "step": 11700
1413
+ },
1414
+ {
1415
+ "epoch": 2.23,
1416
+ "learning_rate": 7.770093102793086e-06,
1417
+ "loss": 0.9452,
1418
+ "step": 11750
1419
+ },
1420
+ {
1421
+ "epoch": 2.24,
1422
+ "learning_rate": 7.760592817784535e-06,
1423
+ "loss": 0.9401,
1424
+ "step": 11800
1425
+ },
1426
+ {
1427
+ "epoch": 2.25,
1428
+ "learning_rate": 7.751092532775983e-06,
1429
+ "loss": 0.9278,
1430
+ "step": 11850
1431
+ },
1432
+ {
1433
+ "epoch": 2.26,
1434
+ "learning_rate": 7.741592247767434e-06,
1435
+ "loss": 0.931,
1436
+ "step": 11900
1437
+ },
1438
+ {
1439
+ "epoch": 2.27,
1440
+ "learning_rate": 7.732091962758883e-06,
1441
+ "loss": 0.9436,
1442
+ "step": 11950
1443
+ },
1444
+ {
1445
+ "epoch": 2.28,
1446
+ "learning_rate": 7.722591677750333e-06,
1447
+ "loss": 0.9212,
1448
+ "step": 12000
1449
+ },
1450
+ {
1451
+ "epoch": 2.29,
1452
+ "learning_rate": 7.713091392741782e-06,
1453
+ "loss": 0.95,
1454
+ "step": 12050
1455
+ },
1456
+ {
1457
+ "epoch": 2.3,
1458
+ "learning_rate": 7.703591107733232e-06,
1459
+ "loss": 0.9701,
1460
+ "step": 12100
1461
+ },
1462
+ {
1463
+ "epoch": 2.31,
1464
+ "learning_rate": 7.694090822724683e-06,
1465
+ "loss": 0.9402,
1466
+ "step": 12150
1467
+ },
1468
+ {
1469
+ "epoch": 2.32,
1470
+ "learning_rate": 7.684590537716132e-06,
1471
+ "loss": 0.9419,
1472
+ "step": 12200
1473
+ },
1474
+ {
1475
+ "epoch": 2.33,
1476
+ "learning_rate": 7.675090252707582e-06,
1477
+ "loss": 0.9436,
1478
+ "step": 12250
1479
+ },
1480
+ {
1481
+ "epoch": 2.34,
1482
+ "learning_rate": 7.665589967699031e-06,
1483
+ "loss": 0.9464,
1484
+ "step": 12300
1485
+ },
1486
+ {
1487
+ "epoch": 2.35,
1488
+ "learning_rate": 7.656089682690483e-06,
1489
+ "loss": 0.939,
1490
+ "step": 12350
1491
+ },
1492
+ {
1493
+ "epoch": 2.36,
1494
+ "learning_rate": 7.64658939768193e-06,
1495
+ "loss": 0.9839,
1496
+ "step": 12400
1497
+ },
1498
+ {
1499
+ "epoch": 2.37,
1500
+ "learning_rate": 7.63708911267338e-06,
1501
+ "loss": 0.9176,
1502
+ "step": 12450
1503
+ },
1504
+ {
1505
+ "epoch": 2.38,
1506
+ "learning_rate": 7.62758882766483e-06,
1507
+ "loss": 0.8935,
1508
+ "step": 12500
1509
+ },
1510
+ {
1511
+ "epoch": 2.38,
1512
+ "learning_rate": 7.61808854265628e-06,
1513
+ "loss": 0.9215,
1514
+ "step": 12550
1515
+ },
1516
+ {
1517
+ "epoch": 2.39,
1518
+ "learning_rate": 7.60858825764773e-06,
1519
+ "loss": 0.9403,
1520
+ "step": 12600
1521
+ },
1522
+ {
1523
+ "epoch": 2.4,
1524
+ "learning_rate": 7.599087972639179e-06,
1525
+ "loss": 0.9272,
1526
+ "step": 12650
1527
+ },
1528
+ {
1529
+ "epoch": 2.41,
1530
+ "learning_rate": 7.58958768763063e-06,
1531
+ "loss": 0.9528,
1532
+ "step": 12700
1533
+ },
1534
+ {
1535
+ "epoch": 2.42,
1536
+ "learning_rate": 7.580087402622079e-06,
1537
+ "loss": 0.9039,
1538
+ "step": 12750
1539
+ },
1540
+ {
1541
+ "epoch": 2.43,
1542
+ "learning_rate": 7.570587117613529e-06,
1543
+ "loss": 0.9194,
1544
+ "step": 12800
1545
+ },
1546
+ {
1547
+ "epoch": 2.44,
1548
+ "learning_rate": 7.561086832604979e-06,
1549
+ "loss": 0.9306,
1550
+ "step": 12850
1551
+ },
1552
+ {
1553
+ "epoch": 2.45,
1554
+ "learning_rate": 7.551586547596429e-06,
1555
+ "loss": 0.9242,
1556
+ "step": 12900
1557
+ },
1558
+ {
1559
+ "epoch": 2.46,
1560
+ "learning_rate": 7.542086262587879e-06,
1561
+ "loss": 0.8961,
1562
+ "step": 12950
1563
+ },
1564
+ {
1565
+ "epoch": 2.47,
1566
+ "learning_rate": 7.532585977579327e-06,
1567
+ "loss": 0.936,
1568
+ "step": 13000
1569
+ },
1570
+ {
1571
+ "epoch": 2.48,
1572
+ "learning_rate": 7.523085692570778e-06,
1573
+ "loss": 0.9598,
1574
+ "step": 13050
1575
+ },
1576
+ {
1577
+ "epoch": 2.49,
1578
+ "learning_rate": 7.513585407562227e-06,
1579
+ "loss": 0.9114,
1580
+ "step": 13100
1581
+ },
1582
+ {
1583
+ "epoch": 2.5,
1584
+ "learning_rate": 7.504085122553677e-06,
1585
+ "loss": 0.9217,
1586
+ "step": 13150
1587
+ },
1588
+ {
1589
+ "epoch": 2.51,
1590
+ "learning_rate": 7.494584837545127e-06,
1591
+ "loss": 0.9041,
1592
+ "step": 13200
1593
+ },
1594
+ {
1595
+ "epoch": 2.52,
1596
+ "learning_rate": 7.485274558236748e-06,
1597
+ "loss": 0.9399,
1598
+ "step": 13250
1599
+ },
1600
+ {
1601
+ "epoch": 2.53,
1602
+ "learning_rate": 7.475774273228198e-06,
1603
+ "loss": 0.9176,
1604
+ "step": 13300
1605
+ },
1606
+ {
1607
+ "epoch": 2.54,
1608
+ "learning_rate": 7.466273988219647e-06,
1609
+ "loss": 0.9235,
1610
+ "step": 13350
1611
+ },
1612
+ {
1613
+ "epoch": 2.55,
1614
+ "learning_rate": 7.456773703211096e-06,
1615
+ "loss": 0.8947,
1616
+ "step": 13400
1617
+ }
1618
+ ],
1619
+ "max_steps": 52630,
1620
+ "num_train_epochs": 10,
1621
+ "total_flos": 5.123196549725184e+19,
1622
+ "trial_name": null,
1623
+ "trial_params": null
1624
+ }
checkpoint-13400/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5edc01452a06bceea0b7057733be6a8f12700dfc6f213c8585ae1fc40583b0b
3
+ size 4088
checkpoint-13600/config.json ADDED
@@ -0,0 +1,77 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_commit_hash": "b05d55b99887e2a2713474fd1590c11ff0a45a94",
3
+ "_name_or_path": "Tech-Hans/detr-resnet-50_finetuned_fashion",
4
+ "activation_dropout": 0.0,
5
+ "activation_function": "relu",
6
+ "architectures": [
7
+ "DetrForObjectDetection"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "auxiliary_loss": false,
11
+ "backbone": "resnet50",
12
+ "backbone_config": null,
13
+ "bbox_cost": 5,
14
+ "bbox_loss_coefficient": 5,
15
+ "class_cost": 1,
16
+ "classifier_dropout": 0.0,
17
+ "d_model": 256,
18
+ "decoder_attention_heads": 8,
19
+ "decoder_ffn_dim": 2048,
20
+ "decoder_layerdrop": 0.0,
21
+ "decoder_layers": 6,
22
+ "dice_loss_coefficient": 1,
23
+ "dilation": false,
24
+ "dropout": 0.1,
25
+ "encoder_attention_heads": 8,
26
+ "encoder_ffn_dim": 2048,
27
+ "encoder_layerdrop": 0.0,
28
+ "encoder_layers": 6,
29
+ "eos_coefficient": 0.1,
30
+ "giou_cost": 2,
31
+ "giou_loss_coefficient": 2,
32
+ "id2label": {
33
+ "0": "bags",
34
+ "1": "belt",
35
+ "2": "bottom",
36
+ "3": "fullbody",
37
+ "4": "glasses",
38
+ "5": "glove",
39
+ "6": "hat",
40
+ "7": "scarf",
41
+ "8": "shoes",
42
+ "9": "sock",
43
+ "10": "tie",
44
+ "11": "tights",
45
+ "12": "top"
46
+ },
47
+ "init_std": 0.02,
48
+ "init_xavier_std": 1.0,
49
+ "is_encoder_decoder": true,
50
+ "label2id": {
51
+ "bags": 0,
52
+ "belt": 1,
53
+ "bottom": 2,
54
+ "fullbody": 3,
55
+ "glasses": 4,
56
+ "glove": 5,
57
+ "hat": 6,
58
+ "scarf": 7,
59
+ "shoes": 8,
60
+ "sock": 9,
61
+ "tie": 10,
62
+ "tights": 11,
63
+ "top": 12
64
+ },
65
+ "mask_loss_coefficient": 1,
66
+ "max_position_embeddings": 1024,
67
+ "model_type": "detr",
68
+ "num_channels": 3,
69
+ "num_hidden_layers": 6,
70
+ "num_queries": 100,
71
+ "position_embedding_type": "sine",
72
+ "scale_embedding": false,
73
+ "torch_dtype": "float32",
74
+ "transformers_version": "4.7.0.dev0",
75
+ "use_pretrained_backbone": true,
76
+ "use_timm_backbone": true
77
+ }
checkpoint-13600/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2ef09549a78125d1de375d2095ccf6196b177421fc89c97d84799588f3fa5a7c
3
+ size 330501178
checkpoint-13600/preprocessor_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "do_pad": true,
4
+ "do_rescale": true,
5
+ "do_resize": true,
6
+ "format": "coco_detection",
7
+ "image_mean": [
8
+ 0.485,
9
+ 0.456,
10
+ 0.406
11
+ ],
12
+ "image_processor_type": "DetrImageProcessor",
13
+ "image_std": [
14
+ 0.229,
15
+ 0.224,
16
+ 0.225
17
+ ],
18
+ "resample": 2,
19
+ "rescale_factor": 0.00392156862745098,
20
+ "size": {
21
+ "longest_edge": 1333,
22
+ "shortest_edge": 800
23
+ }
24
+ }
checkpoint-13600/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d4989b2d7f4cae57b5a3aa1cf86e415ff5b1965edd161d195369354348da688
3
+ size 166628834
checkpoint-13600/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:537b572ce49c7fb1c6fd03aafb446241aa726fc8f5b60652b0e98a5dbb1d2e11
3
+ size 14180
checkpoint-13600/scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0966922245d3c6dcafc3e3da4ae01f193f0ff913d9252f11da8684e4ef0bac2e
3
+ size 988
checkpoint-13600/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d47ce7fd4eab69b12315aaacde513de620180503352b93c9d2b97de65db0e8a4
3
+ size 1064
checkpoint-13600/trainer_state.json ADDED
@@ -0,0 +1,1648 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.5840775223256696,
5
+ "global_step": 13600,
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": 9.992019760592819e-06,
13
+ "loss": 1.1564,
14
+ "step": 50
15
+ },
16
+ {
17
+ "epoch": 0.02,
18
+ "learning_rate": 9.982709481284439e-06,
19
+ "loss": 1.132,
20
+ "step": 100
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 9.97320919627589e-06,
25
+ "loss": 1.1507,
26
+ "step": 150
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 9.96370891126734e-06,
31
+ "loss": 1.0846,
32
+ "step": 200
33
+ },
34
+ {
35
+ "epoch": 0.05,
36
+ "learning_rate": 9.954208626258789e-06,
37
+ "loss": 1.0877,
38
+ "step": 250
39
+ },
40
+ {
41
+ "epoch": 0.06,
42
+ "learning_rate": 9.944708341250238e-06,
43
+ "loss": 1.1153,
44
+ "step": 300
45
+ },
46
+ {
47
+ "epoch": 0.07,
48
+ "learning_rate": 9.935208056241688e-06,
49
+ "loss": 1.0876,
50
+ "step": 350
51
+ },
52
+ {
53
+ "epoch": 0.08,
54
+ "learning_rate": 9.925707771233137e-06,
55
+ "loss": 1.0729,
56
+ "step": 400
57
+ },
58
+ {
59
+ "epoch": 0.09,
60
+ "learning_rate": 9.916207486224587e-06,
61
+ "loss": 1.126,
62
+ "step": 450
63
+ },
64
+ {
65
+ "epoch": 0.1,
66
+ "learning_rate": 9.906707201216036e-06,
67
+ "loss": 1.1262,
68
+ "step": 500
69
+ },
70
+ {
71
+ "epoch": 0.1,
72
+ "learning_rate": 9.897206916207487e-06,
73
+ "loss": 1.0883,
74
+ "step": 550
75
+ },
76
+ {
77
+ "epoch": 0.11,
78
+ "learning_rate": 9.887706631198937e-06,
79
+ "loss": 1.0926,
80
+ "step": 600
81
+ },
82
+ {
83
+ "epoch": 0.12,
84
+ "learning_rate": 9.878206346190386e-06,
85
+ "loss": 1.0916,
86
+ "step": 650
87
+ },
88
+ {
89
+ "epoch": 0.13,
90
+ "learning_rate": 9.868706061181836e-06,
91
+ "loss": 1.0892,
92
+ "step": 700
93
+ },
94
+ {
95
+ "epoch": 0.14,
96
+ "learning_rate": 9.859205776173287e-06,
97
+ "loss": 1.0653,
98
+ "step": 750
99
+ },
100
+ {
101
+ "epoch": 0.15,
102
+ "learning_rate": 9.849705491164736e-06,
103
+ "loss": 1.0391,
104
+ "step": 800
105
+ },
106
+ {
107
+ "epoch": 0.16,
108
+ "learning_rate": 9.840205206156186e-06,
109
+ "loss": 1.0866,
110
+ "step": 850
111
+ },
112
+ {
113
+ "epoch": 0.17,
114
+ "learning_rate": 9.830704921147635e-06,
115
+ "loss": 1.1192,
116
+ "step": 900
117
+ },
118
+ {
119
+ "epoch": 0.18,
120
+ "learning_rate": 9.821204636139085e-06,
121
+ "loss": 1.0898,
122
+ "step": 950
123
+ },
124
+ {
125
+ "epoch": 0.19,
126
+ "learning_rate": 9.811704351130534e-06,
127
+ "loss": 1.0693,
128
+ "step": 1000
129
+ },
130
+ {
131
+ "epoch": 0.2,
132
+ "learning_rate": 9.802204066121984e-06,
133
+ "loss": 1.1243,
134
+ "step": 1050
135
+ },
136
+ {
137
+ "epoch": 0.21,
138
+ "learning_rate": 9.792703781113433e-06,
139
+ "loss": 1.071,
140
+ "step": 1100
141
+ },
142
+ {
143
+ "epoch": 0.22,
144
+ "learning_rate": 9.783203496104884e-06,
145
+ "loss": 1.0569,
146
+ "step": 1150
147
+ },
148
+ {
149
+ "epoch": 0.23,
150
+ "learning_rate": 9.773703211096334e-06,
151
+ "loss": 1.0612,
152
+ "step": 1200
153
+ },
154
+ {
155
+ "epoch": 0.24,
156
+ "learning_rate": 9.764202926087783e-06,
157
+ "loss": 1.0432,
158
+ "step": 1250
159
+ },
160
+ {
161
+ "epoch": 0.25,
162
+ "learning_rate": 9.754702641079233e-06,
163
+ "loss": 1.092,
164
+ "step": 1300
165
+ },
166
+ {
167
+ "epoch": 0.26,
168
+ "learning_rate": 9.745202356070684e-06,
169
+ "loss": 1.0602,
170
+ "step": 1350
171
+ },
172
+ {
173
+ "epoch": 0.27,
174
+ "learning_rate": 9.735702071062133e-06,
175
+ "loss": 1.0689,
176
+ "step": 1400
177
+ },
178
+ {
179
+ "epoch": 0.28,
180
+ "learning_rate": 9.726201786053583e-06,
181
+ "loss": 1.0897,
182
+ "step": 1450
183
+ },
184
+ {
185
+ "epoch": 0.29,
186
+ "learning_rate": 9.716701501045032e-06,
187
+ "loss": 1.0482,
188
+ "step": 1500
189
+ },
190
+ {
191
+ "epoch": 0.29,
192
+ "learning_rate": 9.707201216036482e-06,
193
+ "loss": 1.1167,
194
+ "step": 1550
195
+ },
196
+ {
197
+ "epoch": 0.3,
198
+ "learning_rate": 9.697700931027931e-06,
199
+ "loss": 1.0707,
200
+ "step": 1600
201
+ },
202
+ {
203
+ "epoch": 0.31,
204
+ "learning_rate": 9.68820064601938e-06,
205
+ "loss": 1.0613,
206
+ "step": 1650
207
+ },
208
+ {
209
+ "epoch": 0.32,
210
+ "learning_rate": 9.67870036101083e-06,
211
+ "loss": 1.0534,
212
+ "step": 1700
213
+ },
214
+ {
215
+ "epoch": 0.33,
216
+ "learning_rate": 9.669200076002281e-06,
217
+ "loss": 1.0719,
218
+ "step": 1750
219
+ },
220
+ {
221
+ "epoch": 0.34,
222
+ "learning_rate": 9.659699790993731e-06,
223
+ "loss": 1.0678,
224
+ "step": 1800
225
+ },
226
+ {
227
+ "epoch": 0.35,
228
+ "learning_rate": 9.65019950598518e-06,
229
+ "loss": 1.064,
230
+ "step": 1850
231
+ },
232
+ {
233
+ "epoch": 0.36,
234
+ "learning_rate": 9.64069922097663e-06,
235
+ "loss": 1.0846,
236
+ "step": 1900
237
+ },
238
+ {
239
+ "epoch": 0.37,
240
+ "learning_rate": 9.631198935968081e-06,
241
+ "loss": 1.0716,
242
+ "step": 1950
243
+ },
244
+ {
245
+ "epoch": 0.38,
246
+ "learning_rate": 9.62169865095953e-06,
247
+ "loss": 1.0847,
248
+ "step": 2000
249
+ },
250
+ {
251
+ "epoch": 0.39,
252
+ "learning_rate": 9.612198365950978e-06,
253
+ "loss": 1.0237,
254
+ "step": 2050
255
+ },
256
+ {
257
+ "epoch": 0.4,
258
+ "learning_rate": 9.60269808094243e-06,
259
+ "loss": 1.0382,
260
+ "step": 2100
261
+ },
262
+ {
263
+ "epoch": 0.41,
264
+ "learning_rate": 9.593197795933879e-06,
265
+ "loss": 1.055,
266
+ "step": 2150
267
+ },
268
+ {
269
+ "epoch": 0.42,
270
+ "learning_rate": 9.583697510925328e-06,
271
+ "loss": 1.0257,
272
+ "step": 2200
273
+ },
274
+ {
275
+ "epoch": 0.43,
276
+ "learning_rate": 9.57438723161695e-06,
277
+ "loss": 1.0425,
278
+ "step": 2250
279
+ },
280
+ {
281
+ "epoch": 0.44,
282
+ "learning_rate": 9.5648869466084e-06,
283
+ "loss": 1.0603,
284
+ "step": 2300
285
+ },
286
+ {
287
+ "epoch": 0.45,
288
+ "learning_rate": 9.555386661599849e-06,
289
+ "loss": 1.0195,
290
+ "step": 2350
291
+ },
292
+ {
293
+ "epoch": 0.46,
294
+ "learning_rate": 9.545886376591299e-06,
295
+ "loss": 1.0311,
296
+ "step": 2400
297
+ },
298
+ {
299
+ "epoch": 0.47,
300
+ "learning_rate": 9.536386091582748e-06,
301
+ "loss": 1.0307,
302
+ "step": 2450
303
+ },
304
+ {
305
+ "epoch": 0.48,
306
+ "learning_rate": 9.526885806574197e-06,
307
+ "loss": 1.0965,
308
+ "step": 2500
309
+ },
310
+ {
311
+ "epoch": 0.48,
312
+ "learning_rate": 9.517385521565647e-06,
313
+ "loss": 1.0271,
314
+ "step": 2550
315
+ },
316
+ {
317
+ "epoch": 0.49,
318
+ "learning_rate": 9.507885236557098e-06,
319
+ "loss": 1.0315,
320
+ "step": 2600
321
+ },
322
+ {
323
+ "epoch": 0.5,
324
+ "learning_rate": 9.498384951548548e-06,
325
+ "loss": 1.0245,
326
+ "step": 2650
327
+ },
328
+ {
329
+ "epoch": 0.51,
330
+ "learning_rate": 9.488884666539997e-06,
331
+ "loss": 1.0208,
332
+ "step": 2700
333
+ },
334
+ {
335
+ "epoch": 0.52,
336
+ "learning_rate": 9.479384381531446e-06,
337
+ "loss": 1.0621,
338
+ "step": 2750
339
+ },
340
+ {
341
+ "epoch": 0.53,
342
+ "learning_rate": 9.469884096522896e-06,
343
+ "loss": 1.0774,
344
+ "step": 2800
345
+ },
346
+ {
347
+ "epoch": 0.54,
348
+ "learning_rate": 9.460383811514347e-06,
349
+ "loss": 1.0727,
350
+ "step": 2850
351
+ },
352
+ {
353
+ "epoch": 0.55,
354
+ "learning_rate": 9.450883526505797e-06,
355
+ "loss": 1.0411,
356
+ "step": 2900
357
+ },
358
+ {
359
+ "epoch": 0.56,
360
+ "learning_rate": 9.441383241497246e-06,
361
+ "loss": 1.0088,
362
+ "step": 2950
363
+ },
364
+ {
365
+ "epoch": 0.57,
366
+ "learning_rate": 9.431882956488696e-06,
367
+ "loss": 1.0133,
368
+ "step": 3000
369
+ },
370
+ {
371
+ "epoch": 0.58,
372
+ "learning_rate": 9.422382671480145e-06,
373
+ "loss": 1.0655,
374
+ "step": 3050
375
+ },
376
+ {
377
+ "epoch": 0.59,
378
+ "learning_rate": 9.412882386471594e-06,
379
+ "loss": 1.0241,
380
+ "step": 3100
381
+ },
382
+ {
383
+ "epoch": 0.6,
384
+ "learning_rate": 9.403382101463044e-06,
385
+ "loss": 1.0708,
386
+ "step": 3150
387
+ },
388
+ {
389
+ "epoch": 0.61,
390
+ "learning_rate": 9.393881816454495e-06,
391
+ "loss": 1.0304,
392
+ "step": 3200
393
+ },
394
+ {
395
+ "epoch": 0.62,
396
+ "learning_rate": 9.384381531445945e-06,
397
+ "loss": 1.0155,
398
+ "step": 3250
399
+ },
400
+ {
401
+ "epoch": 0.63,
402
+ "learning_rate": 9.374881246437394e-06,
403
+ "loss": 1.0224,
404
+ "step": 3300
405
+ },
406
+ {
407
+ "epoch": 0.64,
408
+ "learning_rate": 9.365380961428843e-06,
409
+ "loss": 1.0665,
410
+ "step": 3350
411
+ },
412
+ {
413
+ "epoch": 0.65,
414
+ "learning_rate": 9.355880676420293e-06,
415
+ "loss": 1.0238,
416
+ "step": 3400
417
+ },
418
+ {
419
+ "epoch": 0.66,
420
+ "learning_rate": 9.346380391411744e-06,
421
+ "loss": 1.1006,
422
+ "step": 3450
423
+ },
424
+ {
425
+ "epoch": 0.67,
426
+ "learning_rate": 9.336880106403194e-06,
427
+ "loss": 0.9987,
428
+ "step": 3500
429
+ },
430
+ {
431
+ "epoch": 0.67,
432
+ "learning_rate": 9.327379821394643e-06,
433
+ "loss": 1.0001,
434
+ "step": 3550
435
+ },
436
+ {
437
+ "epoch": 0.68,
438
+ "learning_rate": 9.317879536386093e-06,
439
+ "loss": 1.0634,
440
+ "step": 3600
441
+ },
442
+ {
443
+ "epoch": 0.69,
444
+ "learning_rate": 9.308379251377542e-06,
445
+ "loss": 1.0277,
446
+ "step": 3650
447
+ },
448
+ {
449
+ "epoch": 0.7,
450
+ "learning_rate": 9.298878966368991e-06,
451
+ "loss": 1.0311,
452
+ "step": 3700
453
+ },
454
+ {
455
+ "epoch": 0.71,
456
+ "learning_rate": 9.289378681360441e-06,
457
+ "loss": 1.0317,
458
+ "step": 3750
459
+ },
460
+ {
461
+ "epoch": 0.72,
462
+ "learning_rate": 9.279878396351892e-06,
463
+ "loss": 1.0787,
464
+ "step": 3800
465
+ },
466
+ {
467
+ "epoch": 0.73,
468
+ "learning_rate": 9.270378111343342e-06,
469
+ "loss": 1.0583,
470
+ "step": 3850
471
+ },
472
+ {
473
+ "epoch": 0.74,
474
+ "learning_rate": 9.260877826334791e-06,
475
+ "loss": 1.0326,
476
+ "step": 3900
477
+ },
478
+ {
479
+ "epoch": 0.75,
480
+ "learning_rate": 9.25137754132624e-06,
481
+ "loss": 0.989,
482
+ "step": 3950
483
+ },
484
+ {
485
+ "epoch": 0.76,
486
+ "learning_rate": 9.24187725631769e-06,
487
+ "loss": 0.9879,
488
+ "step": 4000
489
+ },
490
+ {
491
+ "epoch": 0.77,
492
+ "learning_rate": 9.232376971309141e-06,
493
+ "loss": 1.0399,
494
+ "step": 4050
495
+ },
496
+ {
497
+ "epoch": 0.78,
498
+ "learning_rate": 9.22287668630059e-06,
499
+ "loss": 1.0045,
500
+ "step": 4100
501
+ },
502
+ {
503
+ "epoch": 0.79,
504
+ "learning_rate": 9.213376401292038e-06,
505
+ "loss": 1.0079,
506
+ "step": 4150
507
+ },
508
+ {
509
+ "epoch": 0.8,
510
+ "learning_rate": 9.20387611628349e-06,
511
+ "loss": 0.999,
512
+ "step": 4200
513
+ },
514
+ {
515
+ "epoch": 0.81,
516
+ "learning_rate": 9.194375831274939e-06,
517
+ "loss": 0.9867,
518
+ "step": 4250
519
+ },
520
+ {
521
+ "epoch": 0.82,
522
+ "learning_rate": 9.184875546266388e-06,
523
+ "loss": 0.9878,
524
+ "step": 4300
525
+ },
526
+ {
527
+ "epoch": 0.83,
528
+ "learning_rate": 9.175375261257838e-06,
529
+ "loss": 1.0098,
530
+ "step": 4350
531
+ },
532
+ {
533
+ "epoch": 0.84,
534
+ "learning_rate": 9.165874976249289e-06,
535
+ "loss": 1.0325,
536
+ "step": 4400
537
+ },
538
+ {
539
+ "epoch": 0.85,
540
+ "learning_rate": 9.156374691240739e-06,
541
+ "loss": 1.0108,
542
+ "step": 4450
543
+ },
544
+ {
545
+ "epoch": 0.86,
546
+ "learning_rate": 9.146874406232188e-06,
547
+ "loss": 1.0167,
548
+ "step": 4500
549
+ },
550
+ {
551
+ "epoch": 0.86,
552
+ "learning_rate": 9.137374121223637e-06,
553
+ "loss": 0.9871,
554
+ "step": 4550
555
+ },
556
+ {
557
+ "epoch": 0.87,
558
+ "learning_rate": 9.128063841915258e-06,
559
+ "loss": 0.983,
560
+ "step": 4600
561
+ },
562
+ {
563
+ "epoch": 0.88,
564
+ "learning_rate": 9.118563556906707e-06,
565
+ "loss": 0.9807,
566
+ "step": 4650
567
+ },
568
+ {
569
+ "epoch": 0.89,
570
+ "learning_rate": 9.109063271898158e-06,
571
+ "loss": 0.9794,
572
+ "step": 4700
573
+ },
574
+ {
575
+ "epoch": 0.9,
576
+ "learning_rate": 9.099562986889608e-06,
577
+ "loss": 1.0505,
578
+ "step": 4750
579
+ },
580
+ {
581
+ "epoch": 0.91,
582
+ "learning_rate": 9.090062701881057e-06,
583
+ "loss": 1.0033,
584
+ "step": 4800
585
+ },
586
+ {
587
+ "epoch": 0.92,
588
+ "learning_rate": 9.080562416872507e-06,
589
+ "loss": 0.9822,
590
+ "step": 4850
591
+ },
592
+ {
593
+ "epoch": 0.93,
594
+ "learning_rate": 9.071062131863958e-06,
595
+ "loss": 1.0046,
596
+ "step": 4900
597
+ },
598
+ {
599
+ "epoch": 0.94,
600
+ "learning_rate": 9.061561846855407e-06,
601
+ "loss": 0.9944,
602
+ "step": 4950
603
+ },
604
+ {
605
+ "epoch": 0.95,
606
+ "learning_rate": 9.052061561846857e-06,
607
+ "loss": 1.0161,
608
+ "step": 5000
609
+ },
610
+ {
611
+ "epoch": 0.96,
612
+ "learning_rate": 9.042561276838306e-06,
613
+ "loss": 1.0239,
614
+ "step": 5050
615
+ },
616
+ {
617
+ "epoch": 0.97,
618
+ "learning_rate": 9.033060991829756e-06,
619
+ "loss": 0.978,
620
+ "step": 5100
621
+ },
622
+ {
623
+ "epoch": 0.98,
624
+ "learning_rate": 9.023560706821205e-06,
625
+ "loss": 1.0055,
626
+ "step": 5150
627
+ },
628
+ {
629
+ "epoch": 0.99,
630
+ "learning_rate": 9.014060421812655e-06,
631
+ "loss": 0.9623,
632
+ "step": 5200
633
+ },
634
+ {
635
+ "epoch": 1.0,
636
+ "learning_rate": 9.004560136804104e-06,
637
+ "loss": 0.9563,
638
+ "step": 5250
639
+ },
640
+ {
641
+ "epoch": 1.01,
642
+ "learning_rate": 8.995059851795555e-06,
643
+ "loss": 0.9714,
644
+ "step": 5300
645
+ },
646
+ {
647
+ "epoch": 1.02,
648
+ "learning_rate": 8.985559566787005e-06,
649
+ "loss": 0.9661,
650
+ "step": 5350
651
+ },
652
+ {
653
+ "epoch": 1.03,
654
+ "learning_rate": 8.976059281778454e-06,
655
+ "loss": 0.9727,
656
+ "step": 5400
657
+ },
658
+ {
659
+ "epoch": 1.04,
660
+ "learning_rate": 8.966558996769904e-06,
661
+ "loss": 0.9874,
662
+ "step": 5450
663
+ },
664
+ {
665
+ "epoch": 1.05,
666
+ "learning_rate": 8.957058711761355e-06,
667
+ "loss": 0.9978,
668
+ "step": 5500
669
+ },
670
+ {
671
+ "epoch": 1.05,
672
+ "learning_rate": 8.947558426752804e-06,
673
+ "loss": 0.9971,
674
+ "step": 5550
675
+ },
676
+ {
677
+ "epoch": 1.06,
678
+ "learning_rate": 8.938058141744254e-06,
679
+ "loss": 1.017,
680
+ "step": 5600
681
+ },
682
+ {
683
+ "epoch": 1.07,
684
+ "learning_rate": 8.928557856735703e-06,
685
+ "loss": 0.9685,
686
+ "step": 5650
687
+ },
688
+ {
689
+ "epoch": 1.08,
690
+ "learning_rate": 8.919057571727153e-06,
691
+ "loss": 0.9842,
692
+ "step": 5700
693
+ },
694
+ {
695
+ "epoch": 1.09,
696
+ "learning_rate": 8.909557286718602e-06,
697
+ "loss": 0.9858,
698
+ "step": 5750
699
+ },
700
+ {
701
+ "epoch": 1.1,
702
+ "learning_rate": 8.900057001710052e-06,
703
+ "loss": 1.0461,
704
+ "step": 5800
705
+ },
706
+ {
707
+ "epoch": 1.11,
708
+ "learning_rate": 8.890556716701501e-06,
709
+ "loss": 0.9706,
710
+ "step": 5850
711
+ },
712
+ {
713
+ "epoch": 1.12,
714
+ "learning_rate": 8.881056431692952e-06,
715
+ "loss": 0.9727,
716
+ "step": 5900
717
+ },
718
+ {
719
+ "epoch": 1.13,
720
+ "learning_rate": 8.871556146684402e-06,
721
+ "loss": 1.0025,
722
+ "step": 5950
723
+ },
724
+ {
725
+ "epoch": 1.14,
726
+ "learning_rate": 8.862055861675851e-06,
727
+ "loss": 1.0136,
728
+ "step": 6000
729
+ },
730
+ {
731
+ "epoch": 1.15,
732
+ "learning_rate": 8.8525555766673e-06,
733
+ "loss": 0.953,
734
+ "step": 6050
735
+ },
736
+ {
737
+ "epoch": 1.16,
738
+ "learning_rate": 8.843055291658752e-06,
739
+ "loss": 0.986,
740
+ "step": 6100
741
+ },
742
+ {
743
+ "epoch": 1.17,
744
+ "learning_rate": 8.833555006650201e-06,
745
+ "loss": 0.9878,
746
+ "step": 6150
747
+ },
748
+ {
749
+ "epoch": 1.18,
750
+ "learning_rate": 8.824054721641649e-06,
751
+ "loss": 0.9738,
752
+ "step": 6200
753
+ },
754
+ {
755
+ "epoch": 1.19,
756
+ "learning_rate": 8.8145544366331e-06,
757
+ "loss": 0.9894,
758
+ "step": 6250
759
+ },
760
+ {
761
+ "epoch": 1.2,
762
+ "learning_rate": 8.80505415162455e-06,
763
+ "loss": 0.9732,
764
+ "step": 6300
765
+ },
766
+ {
767
+ "epoch": 1.21,
768
+ "learning_rate": 8.795553866615999e-06,
769
+ "loss": 0.9799,
770
+ "step": 6350
771
+ },
772
+ {
773
+ "epoch": 1.22,
774
+ "learning_rate": 8.786053581607449e-06,
775
+ "loss": 1.0178,
776
+ "step": 6400
777
+ },
778
+ {
779
+ "epoch": 1.23,
780
+ "learning_rate": 8.776553296598898e-06,
781
+ "loss": 0.9576,
782
+ "step": 6450
783
+ },
784
+ {
785
+ "epoch": 1.24,
786
+ "learning_rate": 8.767053011590349e-06,
787
+ "loss": 1.0088,
788
+ "step": 6500
789
+ },
790
+ {
791
+ "epoch": 1.24,
792
+ "learning_rate": 8.757552726581799e-06,
793
+ "loss": 0.9657,
794
+ "step": 6550
795
+ },
796
+ {
797
+ "epoch": 1.25,
798
+ "learning_rate": 8.748052441573248e-06,
799
+ "loss": 0.9517,
800
+ "step": 6600
801
+ },
802
+ {
803
+ "epoch": 1.26,
804
+ "learning_rate": 8.738552156564698e-06,
805
+ "loss": 1.0624,
806
+ "step": 6650
807
+ },
808
+ {
809
+ "epoch": 1.27,
810
+ "learning_rate": 8.729051871556149e-06,
811
+ "loss": 1.0197,
812
+ "step": 6700
813
+ },
814
+ {
815
+ "epoch": 1.28,
816
+ "learning_rate": 8.719551586547596e-06,
817
+ "loss": 0.9647,
818
+ "step": 6750
819
+ },
820
+ {
821
+ "epoch": 1.29,
822
+ "learning_rate": 8.710051301539046e-06,
823
+ "loss": 1.0109,
824
+ "step": 6800
825
+ },
826
+ {
827
+ "epoch": 1.3,
828
+ "learning_rate": 8.700741022230668e-06,
829
+ "loss": 0.9923,
830
+ "step": 6850
831
+ },
832
+ {
833
+ "epoch": 1.31,
834
+ "learning_rate": 8.691240737222117e-06,
835
+ "loss": 0.9976,
836
+ "step": 6900
837
+ },
838
+ {
839
+ "epoch": 1.32,
840
+ "learning_rate": 8.681740452213567e-06,
841
+ "loss": 0.9929,
842
+ "step": 6950
843
+ },
844
+ {
845
+ "epoch": 1.33,
846
+ "learning_rate": 8.672240167205018e-06,
847
+ "loss": 0.9783,
848
+ "step": 7000
849
+ },
850
+ {
851
+ "epoch": 1.34,
852
+ "learning_rate": 8.662739882196467e-06,
853
+ "loss": 0.9503,
854
+ "step": 7050
855
+ },
856
+ {
857
+ "epoch": 1.35,
858
+ "learning_rate": 8.653239597187917e-06,
859
+ "loss": 0.9915,
860
+ "step": 7100
861
+ },
862
+ {
863
+ "epoch": 1.36,
864
+ "learning_rate": 8.643739312179366e-06,
865
+ "loss": 0.9502,
866
+ "step": 7150
867
+ },
868
+ {
869
+ "epoch": 1.37,
870
+ "learning_rate": 8.634239027170816e-06,
871
+ "loss": 0.9867,
872
+ "step": 7200
873
+ },
874
+ {
875
+ "epoch": 1.38,
876
+ "learning_rate": 8.624738742162265e-06,
877
+ "loss": 0.9929,
878
+ "step": 7250
879
+ },
880
+ {
881
+ "epoch": 1.39,
882
+ "learning_rate": 8.615238457153715e-06,
883
+ "loss": 1.0304,
884
+ "step": 7300
885
+ },
886
+ {
887
+ "epoch": 1.4,
888
+ "learning_rate": 8.605738172145164e-06,
889
+ "loss": 0.9841,
890
+ "step": 7350
891
+ },
892
+ {
893
+ "epoch": 1.41,
894
+ "learning_rate": 8.596237887136615e-06,
895
+ "loss": 0.9915,
896
+ "step": 7400
897
+ },
898
+ {
899
+ "epoch": 1.42,
900
+ "learning_rate": 8.586737602128065e-06,
901
+ "loss": 0.9691,
902
+ "step": 7450
903
+ },
904
+ {
905
+ "epoch": 1.43,
906
+ "learning_rate": 8.577237317119514e-06,
907
+ "loss": 0.9655,
908
+ "step": 7500
909
+ },
910
+ {
911
+ "epoch": 1.43,
912
+ "learning_rate": 8.567737032110964e-06,
913
+ "loss": 0.9884,
914
+ "step": 7550
915
+ },
916
+ {
917
+ "epoch": 1.44,
918
+ "learning_rate": 8.558236747102415e-06,
919
+ "loss": 0.9736,
920
+ "step": 7600
921
+ },
922
+ {
923
+ "epoch": 1.45,
924
+ "learning_rate": 8.548736462093864e-06,
925
+ "loss": 0.9462,
926
+ "step": 7650
927
+ },
928
+ {
929
+ "epoch": 1.46,
930
+ "learning_rate": 8.539236177085312e-06,
931
+ "loss": 1.0023,
932
+ "step": 7700
933
+ },
934
+ {
935
+ "epoch": 1.47,
936
+ "learning_rate": 8.529735892076763e-06,
937
+ "loss": 0.9696,
938
+ "step": 7750
939
+ },
940
+ {
941
+ "epoch": 1.48,
942
+ "learning_rate": 8.520235607068213e-06,
943
+ "loss": 0.9498,
944
+ "step": 7800
945
+ },
946
+ {
947
+ "epoch": 1.49,
948
+ "learning_rate": 8.510735322059662e-06,
949
+ "loss": 0.9689,
950
+ "step": 7850
951
+ },
952
+ {
953
+ "epoch": 1.5,
954
+ "learning_rate": 8.501235037051112e-06,
955
+ "loss": 1.0175,
956
+ "step": 7900
957
+ },
958
+ {
959
+ "epoch": 1.51,
960
+ "learning_rate": 8.491734752042561e-06,
961
+ "loss": 0.9783,
962
+ "step": 7950
963
+ },
964
+ {
965
+ "epoch": 1.52,
966
+ "learning_rate": 8.482234467034012e-06,
967
+ "loss": 0.9942,
968
+ "step": 8000
969
+ },
970
+ {
971
+ "epoch": 1.53,
972
+ "learning_rate": 8.472734182025462e-06,
973
+ "loss": 0.9631,
974
+ "step": 8050
975
+ },
976
+ {
977
+ "epoch": 1.54,
978
+ "learning_rate": 8.463233897016911e-06,
979
+ "loss": 0.9417,
980
+ "step": 8100
981
+ },
982
+ {
983
+ "epoch": 1.55,
984
+ "learning_rate": 8.45373361200836e-06,
985
+ "loss": 0.9449,
986
+ "step": 8150
987
+ },
988
+ {
989
+ "epoch": 1.56,
990
+ "learning_rate": 8.444233326999812e-06,
991
+ "loss": 0.986,
992
+ "step": 8200
993
+ },
994
+ {
995
+ "epoch": 1.57,
996
+ "learning_rate": 8.434733041991261e-06,
997
+ "loss": 0.9353,
998
+ "step": 8250
999
+ },
1000
+ {
1001
+ "epoch": 1.58,
1002
+ "learning_rate": 8.425232756982709e-06,
1003
+ "loss": 0.9693,
1004
+ "step": 8300
1005
+ },
1006
+ {
1007
+ "epoch": 1.59,
1008
+ "learning_rate": 8.41573247197416e-06,
1009
+ "loss": 0.966,
1010
+ "step": 8350
1011
+ },
1012
+ {
1013
+ "epoch": 1.6,
1014
+ "learning_rate": 8.40623218696561e-06,
1015
+ "loss": 0.968,
1016
+ "step": 8400
1017
+ },
1018
+ {
1019
+ "epoch": 1.61,
1020
+ "learning_rate": 8.396731901957059e-06,
1021
+ "loss": 0.9226,
1022
+ "step": 8450
1023
+ },
1024
+ {
1025
+ "epoch": 1.62,
1026
+ "learning_rate": 8.387231616948509e-06,
1027
+ "loss": 0.9361,
1028
+ "step": 8500
1029
+ },
1030
+ {
1031
+ "epoch": 1.62,
1032
+ "learning_rate": 8.377731331939958e-06,
1033
+ "loss": 0.9552,
1034
+ "step": 8550
1035
+ },
1036
+ {
1037
+ "epoch": 1.63,
1038
+ "learning_rate": 8.36823104693141e-06,
1039
+ "loss": 0.9482,
1040
+ "step": 8600
1041
+ },
1042
+ {
1043
+ "epoch": 1.64,
1044
+ "learning_rate": 8.358730761922859e-06,
1045
+ "loss": 0.968,
1046
+ "step": 8650
1047
+ },
1048
+ {
1049
+ "epoch": 1.65,
1050
+ "learning_rate": 8.349230476914308e-06,
1051
+ "loss": 0.9786,
1052
+ "step": 8700
1053
+ },
1054
+ {
1055
+ "epoch": 1.66,
1056
+ "learning_rate": 8.339730191905758e-06,
1057
+ "loss": 0.9615,
1058
+ "step": 8750
1059
+ },
1060
+ {
1061
+ "epoch": 1.67,
1062
+ "learning_rate": 8.330229906897209e-06,
1063
+ "loss": 0.964,
1064
+ "step": 8800
1065
+ },
1066
+ {
1067
+ "epoch": 1.68,
1068
+ "learning_rate": 8.320729621888657e-06,
1069
+ "loss": 0.9676,
1070
+ "step": 8850
1071
+ },
1072
+ {
1073
+ "epoch": 1.69,
1074
+ "learning_rate": 8.311229336880106e-06,
1075
+ "loss": 0.9532,
1076
+ "step": 8900
1077
+ },
1078
+ {
1079
+ "epoch": 1.7,
1080
+ "learning_rate": 8.301729051871557e-06,
1081
+ "loss": 0.9506,
1082
+ "step": 8950
1083
+ },
1084
+ {
1085
+ "epoch": 1.71,
1086
+ "learning_rate": 8.292418772563177e-06,
1087
+ "loss": 0.9513,
1088
+ "step": 9000
1089
+ },
1090
+ {
1091
+ "epoch": 1.72,
1092
+ "learning_rate": 8.282918487554627e-06,
1093
+ "loss": 0.9185,
1094
+ "step": 9050
1095
+ },
1096
+ {
1097
+ "epoch": 1.73,
1098
+ "learning_rate": 8.273418202546078e-06,
1099
+ "loss": 0.9706,
1100
+ "step": 9100
1101
+ },
1102
+ {
1103
+ "epoch": 1.74,
1104
+ "learning_rate": 8.263917917537527e-06,
1105
+ "loss": 0.9235,
1106
+ "step": 9150
1107
+ },
1108
+ {
1109
+ "epoch": 1.75,
1110
+ "learning_rate": 8.254417632528977e-06,
1111
+ "loss": 0.958,
1112
+ "step": 9200
1113
+ },
1114
+ {
1115
+ "epoch": 1.76,
1116
+ "learning_rate": 8.244917347520426e-06,
1117
+ "loss": 0.9587,
1118
+ "step": 9250
1119
+ },
1120
+ {
1121
+ "epoch": 1.77,
1122
+ "learning_rate": 8.235417062511876e-06,
1123
+ "loss": 0.9674,
1124
+ "step": 9300
1125
+ },
1126
+ {
1127
+ "epoch": 1.78,
1128
+ "learning_rate": 8.225916777503325e-06,
1129
+ "loss": 0.9496,
1130
+ "step": 9350
1131
+ },
1132
+ {
1133
+ "epoch": 1.79,
1134
+ "learning_rate": 8.216416492494775e-06,
1135
+ "loss": 0.9579,
1136
+ "step": 9400
1137
+ },
1138
+ {
1139
+ "epoch": 1.8,
1140
+ "learning_rate": 8.206916207486226e-06,
1141
+ "loss": 0.9654,
1142
+ "step": 9450
1143
+ },
1144
+ {
1145
+ "epoch": 1.81,
1146
+ "learning_rate": 8.197415922477675e-06,
1147
+ "loss": 0.9548,
1148
+ "step": 9500
1149
+ },
1150
+ {
1151
+ "epoch": 1.81,
1152
+ "learning_rate": 8.187915637469125e-06,
1153
+ "loss": 0.9631,
1154
+ "step": 9550
1155
+ },
1156
+ {
1157
+ "epoch": 1.82,
1158
+ "learning_rate": 8.178415352460574e-06,
1159
+ "loss": 0.9395,
1160
+ "step": 9600
1161
+ },
1162
+ {
1163
+ "epoch": 1.83,
1164
+ "learning_rate": 8.168915067452024e-06,
1165
+ "loss": 0.9646,
1166
+ "step": 9650
1167
+ },
1168
+ {
1169
+ "epoch": 1.84,
1170
+ "learning_rate": 8.159414782443475e-06,
1171
+ "loss": 0.9239,
1172
+ "step": 9700
1173
+ },
1174
+ {
1175
+ "epoch": 1.85,
1176
+ "learning_rate": 8.149914497434924e-06,
1177
+ "loss": 0.9699,
1178
+ "step": 9750
1179
+ },
1180
+ {
1181
+ "epoch": 1.86,
1182
+ "learning_rate": 8.140414212426372e-06,
1183
+ "loss": 0.9407,
1184
+ "step": 9800
1185
+ },
1186
+ {
1187
+ "epoch": 1.87,
1188
+ "learning_rate": 8.130913927417823e-06,
1189
+ "loss": 0.9519,
1190
+ "step": 9850
1191
+ },
1192
+ {
1193
+ "epoch": 1.88,
1194
+ "learning_rate": 8.121413642409273e-06,
1195
+ "loss": 0.9786,
1196
+ "step": 9900
1197
+ },
1198
+ {
1199
+ "epoch": 1.89,
1200
+ "learning_rate": 8.111913357400722e-06,
1201
+ "loss": 0.9691,
1202
+ "step": 9950
1203
+ },
1204
+ {
1205
+ "epoch": 1.9,
1206
+ "learning_rate": 8.102413072392172e-06,
1207
+ "loss": 0.9379,
1208
+ "step": 10000
1209
+ },
1210
+ {
1211
+ "epoch": 1.91,
1212
+ "learning_rate": 8.092912787383623e-06,
1213
+ "loss": 0.9676,
1214
+ "step": 10050
1215
+ },
1216
+ {
1217
+ "epoch": 1.92,
1218
+ "learning_rate": 8.083412502375072e-06,
1219
+ "loss": 0.9751,
1220
+ "step": 10100
1221
+ },
1222
+ {
1223
+ "epoch": 1.93,
1224
+ "learning_rate": 8.073912217366522e-06,
1225
+ "loss": 0.9705,
1226
+ "step": 10150
1227
+ },
1228
+ {
1229
+ "epoch": 1.94,
1230
+ "learning_rate": 8.064411932357971e-06,
1231
+ "loss": 0.9586,
1232
+ "step": 10200
1233
+ },
1234
+ {
1235
+ "epoch": 1.95,
1236
+ "learning_rate": 8.05491164734942e-06,
1237
+ "loss": 0.9573,
1238
+ "step": 10250
1239
+ },
1240
+ {
1241
+ "epoch": 1.96,
1242
+ "learning_rate": 8.045411362340872e-06,
1243
+ "loss": 0.9581,
1244
+ "step": 10300
1245
+ },
1246
+ {
1247
+ "epoch": 1.97,
1248
+ "learning_rate": 8.03591107733232e-06,
1249
+ "loss": 0.9454,
1250
+ "step": 10350
1251
+ },
1252
+ {
1253
+ "epoch": 1.98,
1254
+ "learning_rate": 8.026410792323769e-06,
1255
+ "loss": 0.948,
1256
+ "step": 10400
1257
+ },
1258
+ {
1259
+ "epoch": 1.99,
1260
+ "learning_rate": 8.01691050731522e-06,
1261
+ "loss": 0.9652,
1262
+ "step": 10450
1263
+ },
1264
+ {
1265
+ "epoch": 2.0,
1266
+ "learning_rate": 8.00741022230667e-06,
1267
+ "loss": 0.9722,
1268
+ "step": 10500
1269
+ },
1270
+ {
1271
+ "epoch": 2.0,
1272
+ "learning_rate": 7.99790993729812e-06,
1273
+ "loss": 0.9283,
1274
+ "step": 10550
1275
+ },
1276
+ {
1277
+ "epoch": 2.01,
1278
+ "learning_rate": 7.988409652289569e-06,
1279
+ "loss": 0.9244,
1280
+ "step": 10600
1281
+ },
1282
+ {
1283
+ "epoch": 2.02,
1284
+ "learning_rate": 7.97890936728102e-06,
1285
+ "loss": 0.9315,
1286
+ "step": 10650
1287
+ },
1288
+ {
1289
+ "epoch": 2.03,
1290
+ "learning_rate": 7.96940908227247e-06,
1291
+ "loss": 0.9376,
1292
+ "step": 10700
1293
+ },
1294
+ {
1295
+ "epoch": 2.04,
1296
+ "learning_rate": 7.959908797263919e-06,
1297
+ "loss": 0.9886,
1298
+ "step": 10750
1299
+ },
1300
+ {
1301
+ "epoch": 2.05,
1302
+ "learning_rate": 7.950408512255368e-06,
1303
+ "loss": 0.9419,
1304
+ "step": 10800
1305
+ },
1306
+ {
1307
+ "epoch": 2.06,
1308
+ "learning_rate": 7.940908227246818e-06,
1309
+ "loss": 0.9548,
1310
+ "step": 10850
1311
+ },
1312
+ {
1313
+ "epoch": 2.07,
1314
+ "learning_rate": 7.931407942238267e-06,
1315
+ "loss": 0.9429,
1316
+ "step": 10900
1317
+ },
1318
+ {
1319
+ "epoch": 2.08,
1320
+ "learning_rate": 7.921907657229717e-06,
1321
+ "loss": 0.9772,
1322
+ "step": 10950
1323
+ },
1324
+ {
1325
+ "epoch": 2.09,
1326
+ "learning_rate": 7.912597377921338e-06,
1327
+ "loss": 0.9443,
1328
+ "step": 11000
1329
+ },
1330
+ {
1331
+ "epoch": 2.1,
1332
+ "learning_rate": 7.903097092912788e-06,
1333
+ "loss": 0.9411,
1334
+ "step": 11050
1335
+ },
1336
+ {
1337
+ "epoch": 2.11,
1338
+ "learning_rate": 7.893596807904237e-06,
1339
+ "loss": 0.9222,
1340
+ "step": 11100
1341
+ },
1342
+ {
1343
+ "epoch": 2.12,
1344
+ "learning_rate": 7.884096522895689e-06,
1345
+ "loss": 0.9312,
1346
+ "step": 11150
1347
+ },
1348
+ {
1349
+ "epoch": 2.13,
1350
+ "learning_rate": 7.874596237887138e-06,
1351
+ "loss": 0.9408,
1352
+ "step": 11200
1353
+ },
1354
+ {
1355
+ "epoch": 2.14,
1356
+ "learning_rate": 7.865095952878587e-06,
1357
+ "loss": 0.9164,
1358
+ "step": 11250
1359
+ },
1360
+ {
1361
+ "epoch": 2.15,
1362
+ "learning_rate": 7.855595667870037e-06,
1363
+ "loss": 0.9823,
1364
+ "step": 11300
1365
+ },
1366
+ {
1367
+ "epoch": 2.16,
1368
+ "learning_rate": 7.846095382861486e-06,
1369
+ "loss": 0.9457,
1370
+ "step": 11350
1371
+ },
1372
+ {
1373
+ "epoch": 2.17,
1374
+ "learning_rate": 7.836595097852936e-06,
1375
+ "loss": 0.9605,
1376
+ "step": 11400
1377
+ },
1378
+ {
1379
+ "epoch": 2.18,
1380
+ "learning_rate": 7.827094812844385e-06,
1381
+ "loss": 0.9283,
1382
+ "step": 11450
1383
+ },
1384
+ {
1385
+ "epoch": 2.19,
1386
+ "learning_rate": 7.817594527835835e-06,
1387
+ "loss": 0.9115,
1388
+ "step": 11500
1389
+ },
1390
+ {
1391
+ "epoch": 2.19,
1392
+ "learning_rate": 7.808094242827286e-06,
1393
+ "loss": 0.9508,
1394
+ "step": 11550
1395
+ },
1396
+ {
1397
+ "epoch": 2.2,
1398
+ "learning_rate": 7.798593957818735e-06,
1399
+ "loss": 0.9419,
1400
+ "step": 11600
1401
+ },
1402
+ {
1403
+ "epoch": 2.21,
1404
+ "learning_rate": 7.789093672810185e-06,
1405
+ "loss": 0.9078,
1406
+ "step": 11650
1407
+ },
1408
+ {
1409
+ "epoch": 2.22,
1410
+ "learning_rate": 7.779593387801634e-06,
1411
+ "loss": 0.9313,
1412
+ "step": 11700
1413
+ },
1414
+ {
1415
+ "epoch": 2.23,
1416
+ "learning_rate": 7.770093102793086e-06,
1417
+ "loss": 0.9452,
1418
+ "step": 11750
1419
+ },
1420
+ {
1421
+ "epoch": 2.24,
1422
+ "learning_rate": 7.760592817784535e-06,
1423
+ "loss": 0.9401,
1424
+ "step": 11800
1425
+ },
1426
+ {
1427
+ "epoch": 2.25,
1428
+ "learning_rate": 7.751092532775983e-06,
1429
+ "loss": 0.9278,
1430
+ "step": 11850
1431
+ },
1432
+ {
1433
+ "epoch": 2.26,
1434
+ "learning_rate": 7.741592247767434e-06,
1435
+ "loss": 0.931,
1436
+ "step": 11900
1437
+ },
1438
+ {
1439
+ "epoch": 2.27,
1440
+ "learning_rate": 7.732091962758883e-06,
1441
+ "loss": 0.9436,
1442
+ "step": 11950
1443
+ },
1444
+ {
1445
+ "epoch": 2.28,
1446
+ "learning_rate": 7.722591677750333e-06,
1447
+ "loss": 0.9212,
1448
+ "step": 12000
1449
+ },
1450
+ {
1451
+ "epoch": 2.29,
1452
+ "learning_rate": 7.713091392741782e-06,
1453
+ "loss": 0.95,
1454
+ "step": 12050
1455
+ },
1456
+ {
1457
+ "epoch": 2.3,
1458
+ "learning_rate": 7.703591107733232e-06,
1459
+ "loss": 0.9701,
1460
+ "step": 12100
1461
+ },
1462
+ {
1463
+ "epoch": 2.31,
1464
+ "learning_rate": 7.694090822724683e-06,
1465
+ "loss": 0.9402,
1466
+ "step": 12150
1467
+ },
1468
+ {
1469
+ "epoch": 2.32,
1470
+ "learning_rate": 7.684590537716132e-06,
1471
+ "loss": 0.9419,
1472
+ "step": 12200
1473
+ },
1474
+ {
1475
+ "epoch": 2.33,
1476
+ "learning_rate": 7.675090252707582e-06,
1477
+ "loss": 0.9436,
1478
+ "step": 12250
1479
+ },
1480
+ {
1481
+ "epoch": 2.34,
1482
+ "learning_rate": 7.665589967699031e-06,
1483
+ "loss": 0.9464,
1484
+ "step": 12300
1485
+ },
1486
+ {
1487
+ "epoch": 2.35,
1488
+ "learning_rate": 7.656089682690483e-06,
1489
+ "loss": 0.939,
1490
+ "step": 12350
1491
+ },
1492
+ {
1493
+ "epoch": 2.36,
1494
+ "learning_rate": 7.64658939768193e-06,
1495
+ "loss": 0.9839,
1496
+ "step": 12400
1497
+ },
1498
+ {
1499
+ "epoch": 2.37,
1500
+ "learning_rate": 7.63708911267338e-06,
1501
+ "loss": 0.9176,
1502
+ "step": 12450
1503
+ },
1504
+ {
1505
+ "epoch": 2.38,
1506
+ "learning_rate": 7.62758882766483e-06,
1507
+ "loss": 0.8935,
1508
+ "step": 12500
1509
+ },
1510
+ {
1511
+ "epoch": 2.38,
1512
+ "learning_rate": 7.61808854265628e-06,
1513
+ "loss": 0.9215,
1514
+ "step": 12550
1515
+ },
1516
+ {
1517
+ "epoch": 2.39,
1518
+ "learning_rate": 7.60858825764773e-06,
1519
+ "loss": 0.9403,
1520
+ "step": 12600
1521
+ },
1522
+ {
1523
+ "epoch": 2.4,
1524
+ "learning_rate": 7.599087972639179e-06,
1525
+ "loss": 0.9272,
1526
+ "step": 12650
1527
+ },
1528
+ {
1529
+ "epoch": 2.41,
1530
+ "learning_rate": 7.58958768763063e-06,
1531
+ "loss": 0.9528,
1532
+ "step": 12700
1533
+ },
1534
+ {
1535
+ "epoch": 2.42,
1536
+ "learning_rate": 7.580087402622079e-06,
1537
+ "loss": 0.9039,
1538
+ "step": 12750
1539
+ },
1540
+ {
1541
+ "epoch": 2.43,
1542
+ "learning_rate": 7.570587117613529e-06,
1543
+ "loss": 0.9194,
1544
+ "step": 12800
1545
+ },
1546
+ {
1547
+ "epoch": 2.44,
1548
+ "learning_rate": 7.561086832604979e-06,
1549
+ "loss": 0.9306,
1550
+ "step": 12850
1551
+ },
1552
+ {
1553
+ "epoch": 2.45,
1554
+ "learning_rate": 7.551586547596429e-06,
1555
+ "loss": 0.9242,
1556
+ "step": 12900
1557
+ },
1558
+ {
1559
+ "epoch": 2.46,
1560
+ "learning_rate": 7.542086262587879e-06,
1561
+ "loss": 0.8961,
1562
+ "step": 12950
1563
+ },
1564
+ {
1565
+ "epoch": 2.47,
1566
+ "learning_rate": 7.532585977579327e-06,
1567
+ "loss": 0.936,
1568
+ "step": 13000
1569
+ },
1570
+ {
1571
+ "epoch": 2.48,
1572
+ "learning_rate": 7.523085692570778e-06,
1573
+ "loss": 0.9598,
1574
+ "step": 13050
1575
+ },
1576
+ {
1577
+ "epoch": 2.49,
1578
+ "learning_rate": 7.513585407562227e-06,
1579
+ "loss": 0.9114,
1580
+ "step": 13100
1581
+ },
1582
+ {
1583
+ "epoch": 2.5,
1584
+ "learning_rate": 7.504085122553677e-06,
1585
+ "loss": 0.9217,
1586
+ "step": 13150
1587
+ },
1588
+ {
1589
+ "epoch": 2.51,
1590
+ "learning_rate": 7.494584837545127e-06,
1591
+ "loss": 0.9041,
1592
+ "step": 13200
1593
+ },
1594
+ {
1595
+ "epoch": 2.52,
1596
+ "learning_rate": 7.485274558236748e-06,
1597
+ "loss": 0.9399,
1598
+ "step": 13250
1599
+ },
1600
+ {
1601
+ "epoch": 2.53,
1602
+ "learning_rate": 7.475774273228198e-06,
1603
+ "loss": 0.9176,
1604
+ "step": 13300
1605
+ },
1606
+ {
1607
+ "epoch": 2.54,
1608
+ "learning_rate": 7.466273988219647e-06,
1609
+ "loss": 0.9235,
1610
+ "step": 13350
1611
+ },
1612
+ {
1613
+ "epoch": 2.55,
1614
+ "learning_rate": 7.456773703211096e-06,
1615
+ "loss": 0.8947,
1616
+ "step": 13400
1617
+ },
1618
+ {
1619
+ "epoch": 2.56,
1620
+ "learning_rate": 7.4472734182025465e-06,
1621
+ "loss": 0.9205,
1622
+ "step": 13450
1623
+ },
1624
+ {
1625
+ "epoch": 2.57,
1626
+ "learning_rate": 7.437773133193996e-06,
1627
+ "loss": 0.9188,
1628
+ "step": 13500
1629
+ },
1630
+ {
1631
+ "epoch": 2.57,
1632
+ "learning_rate": 7.428272848185446e-06,
1633
+ "loss": 0.9285,
1634
+ "step": 13550
1635
+ },
1636
+ {
1637
+ "epoch": 2.58,
1638
+ "learning_rate": 7.418772563176896e-06,
1639
+ "loss": 0.93,
1640
+ "step": 13600
1641
+ }
1642
+ ],
1643
+ "max_steps": 52630,
1644
+ "num_train_epochs": 10,
1645
+ "total_flos": 5.199665023171584e+19,
1646
+ "trial_name": null,
1647
+ "trial_params": null
1648
+ }
checkpoint-13600/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5edc01452a06bceea0b7057733be6a8f12700dfc6f213c8585ae1fc40583b0b
3
+ size 4088
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0af4f81c1895814163d1a974dca6d7a50bea06a94b0e684a5bbba66fb9b630f5
3
  size 166628834
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d4989b2d7f4cae57b5a3aa1cf86e415ff5b1965edd161d195369354348da688
3
  size 166628834
runs/Feb29_13-56-06_3a0664c7c2dd/events.out.tfevents.1709215109.3a0664c7c2dd.2321.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b612e3fde255f5804d04bc8b0d011ce1cf45687816ce07d1627c14bd61321798
3
- size 8909
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1d5debafc9304f7db453dcd62a7183edc61c47e61bd1ddedd70eef86531d738
3
+ size 9537