patrickvonplaten commited on
Commit
e85edf8
1 Parent(s): c821fd5

End of training

Browse files
all_results.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "eval_loss": 2.9303946495056152,
4
+ "eval_runtime": 66.2786,
5
+ "eval_samples": 3394,
6
+ "eval_samples_per_second": 51.208,
7
+ "eval_steps_per_second": 0.815,
8
+ "eval_wer": 1.0,
9
+ "train_loss": 3.4216162567138673,
10
+ "train_runtime": 5111.6283,
11
+ "train_samples": 2194,
12
+ "train_samples_per_second": 42.922,
13
+ "train_steps_per_second": 0.685
14
+ }
emissions.csv ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ timestamp,experiment_id,project_name,duration,emissions,energy_consumed,country_name,country_iso_code,region,on_cloud,cloud_provider,cloud_region
2
+ 2021-11-15T23:59:22,7f6b74a8-b645-4658-a2bf-0934267850b3,codecarbon,5105.845121622086,0.8052671576957905,2.1894158719298273,USA,USA,nan,Y,aws,us-east-1
eval_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "eval_loss": 2.9303946495056152,
4
+ "eval_runtime": 66.2786,
5
+ "eval_samples": 3394,
6
+ "eval_samples_per_second": 51.208,
7
+ "eval_steps_per_second": 0.815,
8
+ "eval_wer": 1.0
9
+ }
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c573cc19e75c2a2dea706633694d6fcba9317d61578aade59843cb30799ce963
3
  size 377688471
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b1ba793990819fff59677345261468f9a5728a44d259a040d578b116b05b740
3
  size 377688471
runs/Nov15_22-29-19_ip-172-31-1-149/events.out.tfevents.1637015654.ip-172-31-1-149.39465.0 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:db0edf3ce20678b1f9135d0e6b84a2fe538f7a2e1ace9bfc69881571e0ea8bb8
3
- size 53160
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4591d1dca9c947e64583ac76002329f40f77606e595cafb7337eff8505abbaa
3
+ size 61682
runs/Nov15_22-29-19_ip-172-31-1-149/events.out.tfevents.1637020830.ip-172-31-1-149.39465.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b372e2c960014b839d00cf5bf4d3f5eaf43ddc243a0f848d0bad695c5f9c9f8
3
+ size 358
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "train_loss": 3.4216162567138673,
4
+ "train_runtime": 5111.6283,
5
+ "train_samples": 2194,
6
+ "train_samples_per_second": 42.922,
7
+ "train_steps_per_second": 0.685
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,2188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 100.0,
5
+ "global_step": 3500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.29,
12
+ "learning_rate": 4e-06,
13
+ "loss": 12.4261,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.57,
18
+ "learning_rate": 8.500000000000002e-06,
19
+ "loss": 14.8659,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.86,
24
+ "learning_rate": 1.35e-05,
25
+ "loss": 12.4176,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 1.14,
30
+ "learning_rate": 1.8e-05,
31
+ "loss": 13.4075,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 1.43,
36
+ "learning_rate": 2.3e-05,
37
+ "loss": 13.113,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 1.71,
42
+ "learning_rate": 2.8e-05,
43
+ "loss": 12.0587,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 2.0,
48
+ "learning_rate": 3.3e-05,
49
+ "loss": 11.7634,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 2.29,
54
+ "learning_rate": 3.8e-05,
55
+ "loss": 10.2845,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 2.57,
60
+ "learning_rate": 4.2999999999999995e-05,
61
+ "loss": 11.5741,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 2.86,
66
+ "learning_rate": 4.8e-05,
67
+ "loss": 9.9064,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 3.14,
72
+ "learning_rate": 5.3e-05,
73
+ "loss": 10.4647,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 3.43,
78
+ "learning_rate": 5.800000000000001e-05,
79
+ "loss": 9.851,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 3.71,
84
+ "learning_rate": 6.3e-05,
85
+ "loss": 10.0611,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 4.0,
90
+ "learning_rate": 6.800000000000001e-05,
91
+ "loss": 9.3195,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 4.29,
96
+ "learning_rate": 7.3e-05,
97
+ "loss": 8.3377,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 4.57,
102
+ "learning_rate": 7.8e-05,
103
+ "loss": 8.9758,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 4.86,
108
+ "learning_rate": 8.300000000000001e-05,
109
+ "loss": 7.912,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 5.14,
114
+ "learning_rate": 8.8e-05,
115
+ "loss": 7.9751,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 5.43,
120
+ "learning_rate": 9.3e-05,
121
+ "loss": 7.2092,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 5.71,
126
+ "learning_rate": 9.800000000000001e-05,
127
+ "loss": 7.0987,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 6.0,
132
+ "learning_rate": 0.000103,
133
+ "loss": 6.2911,
134
+ "step": 210
135
+ },
136
+ {
137
+ "epoch": 6.29,
138
+ "learning_rate": 0.000108,
139
+ "loss": 5.555,
140
+ "step": 220
141
+ },
142
+ {
143
+ "epoch": 6.57,
144
+ "learning_rate": 0.00011300000000000001,
145
+ "loss": 5.5446,
146
+ "step": 230
147
+ },
148
+ {
149
+ "epoch": 6.86,
150
+ "learning_rate": 0.000118,
151
+ "loss": 4.5362,
152
+ "step": 240
153
+ },
154
+ {
155
+ "epoch": 7.14,
156
+ "learning_rate": 0.000123,
157
+ "loss": 4.2736,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 7.43,
162
+ "learning_rate": 0.000128,
163
+ "loss": 3.8207,
164
+ "step": 260
165
+ },
166
+ {
167
+ "epoch": 7.71,
168
+ "learning_rate": 0.000133,
169
+ "loss": 3.7274,
170
+ "step": 270
171
+ },
172
+ {
173
+ "epoch": 8.0,
174
+ "learning_rate": 0.00013800000000000002,
175
+ "loss": 3.3005,
176
+ "step": 280
177
+ },
178
+ {
179
+ "epoch": 8.29,
180
+ "learning_rate": 0.00014299999999999998,
181
+ "loss": 3.1548,
182
+ "step": 290
183
+ },
184
+ {
185
+ "epoch": 8.57,
186
+ "learning_rate": 0.000148,
187
+ "loss": 3.0679,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 8.86,
192
+ "learning_rate": 0.000153,
193
+ "loss": 3.0221,
194
+ "step": 310
195
+ },
196
+ {
197
+ "epoch": 9.14,
198
+ "learning_rate": 0.000158,
199
+ "loss": 3.0757,
200
+ "step": 320
201
+ },
202
+ {
203
+ "epoch": 9.43,
204
+ "learning_rate": 0.000163,
205
+ "loss": 2.9748,
206
+ "step": 330
207
+ },
208
+ {
209
+ "epoch": 9.71,
210
+ "learning_rate": 0.00016800000000000002,
211
+ "loss": 2.9839,
212
+ "step": 340
213
+ },
214
+ {
215
+ "epoch": 10.0,
216
+ "learning_rate": 0.000173,
217
+ "loss": 2.9899,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 10.29,
222
+ "learning_rate": 0.000178,
223
+ "loss": 2.9415,
224
+ "step": 360
225
+ },
226
+ {
227
+ "epoch": 10.57,
228
+ "learning_rate": 0.000183,
229
+ "loss": 3.0035,
230
+ "step": 370
231
+ },
232
+ {
233
+ "epoch": 10.86,
234
+ "learning_rate": 0.00018800000000000002,
235
+ "loss": 2.9654,
236
+ "step": 380
237
+ },
238
+ {
239
+ "epoch": 11.14,
240
+ "learning_rate": 0.000193,
241
+ "loss": 2.9752,
242
+ "step": 390
243
+ },
244
+ {
245
+ "epoch": 11.43,
246
+ "learning_rate": 0.00019800000000000002,
247
+ "loss": 2.945,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 11.71,
252
+ "learning_rate": 0.00020300000000000003,
253
+ "loss": 3.0024,
254
+ "step": 410
255
+ },
256
+ {
257
+ "epoch": 12.0,
258
+ "learning_rate": 0.000208,
259
+ "loss": 2.9515,
260
+ "step": 420
261
+ },
262
+ {
263
+ "epoch": 12.29,
264
+ "learning_rate": 0.000213,
265
+ "loss": 2.9324,
266
+ "step": 430
267
+ },
268
+ {
269
+ "epoch": 12.57,
270
+ "learning_rate": 0.000218,
271
+ "loss": 2.9882,
272
+ "step": 440
273
+ },
274
+ {
275
+ "epoch": 12.86,
276
+ "learning_rate": 0.000223,
277
+ "loss": 2.9414,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 13.14,
282
+ "learning_rate": 0.000228,
283
+ "loss": 2.9913,
284
+ "step": 460
285
+ },
286
+ {
287
+ "epoch": 13.43,
288
+ "learning_rate": 0.00023300000000000003,
289
+ "loss": 2.9644,
290
+ "step": 470
291
+ },
292
+ {
293
+ "epoch": 13.71,
294
+ "learning_rate": 0.00023799999999999998,
295
+ "loss": 2.9708,
296
+ "step": 480
297
+ },
298
+ {
299
+ "epoch": 14.0,
300
+ "learning_rate": 0.000243,
301
+ "loss": 2.9498,
302
+ "step": 490
303
+ },
304
+ {
305
+ "epoch": 14.29,
306
+ "learning_rate": 0.000248,
307
+ "loss": 2.9545,
308
+ "step": 500
309
+ },
310
+ {
311
+ "epoch": 14.29,
312
+ "eval_loss": 2.9353771209716797,
313
+ "eval_runtime": 66.3897,
314
+ "eval_samples_per_second": 51.122,
315
+ "eval_steps_per_second": 0.813,
316
+ "eval_wer": 1.0,
317
+ "step": 500
318
+ },
319
+ {
320
+ "epoch": 14.57,
321
+ "learning_rate": 0.000253,
322
+ "loss": 2.9787,
323
+ "step": 510
324
+ },
325
+ {
326
+ "epoch": 14.86,
327
+ "learning_rate": 0.00025800000000000004,
328
+ "loss": 2.9257,
329
+ "step": 520
330
+ },
331
+ {
332
+ "epoch": 15.14,
333
+ "learning_rate": 0.000263,
334
+ "loss": 2.9613,
335
+ "step": 530
336
+ },
337
+ {
338
+ "epoch": 15.43,
339
+ "learning_rate": 0.000268,
340
+ "loss": 2.9755,
341
+ "step": 540
342
+ },
343
+ {
344
+ "epoch": 15.71,
345
+ "learning_rate": 0.000273,
346
+ "loss": 2.958,
347
+ "step": 550
348
+ },
349
+ {
350
+ "epoch": 16.0,
351
+ "learning_rate": 0.00027800000000000004,
352
+ "loss": 2.9364,
353
+ "step": 560
354
+ },
355
+ {
356
+ "epoch": 16.29,
357
+ "learning_rate": 0.000283,
358
+ "loss": 2.9692,
359
+ "step": 570
360
+ },
361
+ {
362
+ "epoch": 16.57,
363
+ "learning_rate": 0.000288,
364
+ "loss": 2.9823,
365
+ "step": 580
366
+ },
367
+ {
368
+ "epoch": 16.86,
369
+ "learning_rate": 0.00029299999999999997,
370
+ "loss": 2.9235,
371
+ "step": 590
372
+ },
373
+ {
374
+ "epoch": 17.14,
375
+ "learning_rate": 0.000298,
376
+ "loss": 2.9359,
377
+ "step": 600
378
+ },
379
+ {
380
+ "epoch": 17.43,
381
+ "learning_rate": 0.000303,
382
+ "loss": 2.9438,
383
+ "step": 610
384
+ },
385
+ {
386
+ "epoch": 17.71,
387
+ "learning_rate": 0.000308,
388
+ "loss": 2.966,
389
+ "step": 620
390
+ },
391
+ {
392
+ "epoch": 18.0,
393
+ "learning_rate": 0.000313,
394
+ "loss": 2.9659,
395
+ "step": 630
396
+ },
397
+ {
398
+ "epoch": 18.29,
399
+ "learning_rate": 0.00031800000000000003,
400
+ "loss": 2.9329,
401
+ "step": 640
402
+ },
403
+ {
404
+ "epoch": 18.57,
405
+ "learning_rate": 0.000323,
406
+ "loss": 2.9605,
407
+ "step": 650
408
+ },
409
+ {
410
+ "epoch": 18.86,
411
+ "learning_rate": 0.000328,
412
+ "loss": 2.9415,
413
+ "step": 660
414
+ },
415
+ {
416
+ "epoch": 19.14,
417
+ "learning_rate": 0.000333,
418
+ "loss": 2.9615,
419
+ "step": 670
420
+ },
421
+ {
422
+ "epoch": 19.43,
423
+ "learning_rate": 0.00033800000000000003,
424
+ "loss": 2.9254,
425
+ "step": 680
426
+ },
427
+ {
428
+ "epoch": 19.71,
429
+ "learning_rate": 0.00034300000000000004,
430
+ "loss": 2.9979,
431
+ "step": 690
432
+ },
433
+ {
434
+ "epoch": 20.0,
435
+ "learning_rate": 0.000348,
436
+ "loss": 2.9817,
437
+ "step": 700
438
+ },
439
+ {
440
+ "epoch": 20.29,
441
+ "learning_rate": 0.00035299999999999996,
442
+ "loss": 2.9375,
443
+ "step": 710
444
+ },
445
+ {
446
+ "epoch": 20.57,
447
+ "learning_rate": 0.000358,
448
+ "loss": 2.9355,
449
+ "step": 720
450
+ },
451
+ {
452
+ "epoch": 20.86,
453
+ "learning_rate": 0.000363,
454
+ "loss": 2.9937,
455
+ "step": 730
456
+ },
457
+ {
458
+ "epoch": 21.14,
459
+ "learning_rate": 0.000368,
460
+ "loss": 2.943,
461
+ "step": 740
462
+ },
463
+ {
464
+ "epoch": 21.43,
465
+ "learning_rate": 0.000373,
466
+ "loss": 2.9456,
467
+ "step": 750
468
+ },
469
+ {
470
+ "epoch": 21.71,
471
+ "learning_rate": 0.000378,
472
+ "loss": 2.9607,
473
+ "step": 760
474
+ },
475
+ {
476
+ "epoch": 22.0,
477
+ "learning_rate": 0.00038300000000000004,
478
+ "loss": 2.9646,
479
+ "step": 770
480
+ },
481
+ {
482
+ "epoch": 22.29,
483
+ "learning_rate": 0.000388,
484
+ "loss": 2.9157,
485
+ "step": 780
486
+ },
487
+ {
488
+ "epoch": 22.57,
489
+ "learning_rate": 0.000393,
490
+ "loss": 2.9803,
491
+ "step": 790
492
+ },
493
+ {
494
+ "epoch": 22.86,
495
+ "learning_rate": 0.000398,
496
+ "loss": 2.9481,
497
+ "step": 800
498
+ },
499
+ {
500
+ "epoch": 23.14,
501
+ "learning_rate": 0.00040300000000000004,
502
+ "loss": 3.007,
503
+ "step": 810
504
+ },
505
+ {
506
+ "epoch": 23.43,
507
+ "learning_rate": 0.000408,
508
+ "loss": 2.9522,
509
+ "step": 820
510
+ },
511
+ {
512
+ "epoch": 23.71,
513
+ "learning_rate": 0.000413,
514
+ "loss": 2.9632,
515
+ "step": 830
516
+ },
517
+ {
518
+ "epoch": 24.0,
519
+ "learning_rate": 0.00041799999999999997,
520
+ "loss": 2.947,
521
+ "step": 840
522
+ },
523
+ {
524
+ "epoch": 24.29,
525
+ "learning_rate": 0.000423,
526
+ "loss": 2.929,
527
+ "step": 850
528
+ },
529
+ {
530
+ "epoch": 24.57,
531
+ "learning_rate": 0.000428,
532
+ "loss": 2.9513,
533
+ "step": 860
534
+ },
535
+ {
536
+ "epoch": 24.86,
537
+ "learning_rate": 0.000433,
538
+ "loss": 2.9646,
539
+ "step": 870
540
+ },
541
+ {
542
+ "epoch": 25.14,
543
+ "learning_rate": 0.000438,
544
+ "loss": 2.9381,
545
+ "step": 880
546
+ },
547
+ {
548
+ "epoch": 25.43,
549
+ "learning_rate": 0.00044300000000000003,
550
+ "loss": 2.931,
551
+ "step": 890
552
+ },
553
+ {
554
+ "epoch": 25.71,
555
+ "learning_rate": 0.000448,
556
+ "loss": 2.9765,
557
+ "step": 900
558
+ },
559
+ {
560
+ "epoch": 26.0,
561
+ "learning_rate": 0.000453,
562
+ "loss": 2.9666,
563
+ "step": 910
564
+ },
565
+ {
566
+ "epoch": 26.29,
567
+ "learning_rate": 0.000458,
568
+ "loss": 2.9378,
569
+ "step": 920
570
+ },
571
+ {
572
+ "epoch": 26.57,
573
+ "learning_rate": 0.00046300000000000003,
574
+ "loss": 2.9663,
575
+ "step": 930
576
+ },
577
+ {
578
+ "epoch": 26.86,
579
+ "learning_rate": 0.00046800000000000005,
580
+ "loss": 2.9334,
581
+ "step": 940
582
+ },
583
+ {
584
+ "epoch": 27.14,
585
+ "learning_rate": 0.000473,
586
+ "loss": 2.9717,
587
+ "step": 950
588
+ },
589
+ {
590
+ "epoch": 27.43,
591
+ "learning_rate": 0.00047799999999999996,
592
+ "loss": 2.9325,
593
+ "step": 960
594
+ },
595
+ {
596
+ "epoch": 27.71,
597
+ "learning_rate": 0.000483,
598
+ "loss": 2.9502,
599
+ "step": 970
600
+ },
601
+ {
602
+ "epoch": 28.0,
603
+ "learning_rate": 0.000488,
604
+ "loss": 2.9664,
605
+ "step": 980
606
+ },
607
+ {
608
+ "epoch": 28.29,
609
+ "learning_rate": 0.0004930000000000001,
610
+ "loss": 2.9713,
611
+ "step": 990
612
+ },
613
+ {
614
+ "epoch": 28.57,
615
+ "learning_rate": 0.000498,
616
+ "loss": 2.9537,
617
+ "step": 1000
618
+ },
619
+ {
620
+ "epoch": 28.57,
621
+ "eval_loss": 2.9358763694763184,
622
+ "eval_runtime": 65.8509,
623
+ "eval_samples_per_second": 51.541,
624
+ "eval_steps_per_second": 0.82,
625
+ "eval_wer": 1.0,
626
+ "step": 1000
627
+ },
628
+ {
629
+ "epoch": 28.86,
630
+ "learning_rate": 0.0004988,
631
+ "loss": 2.9211,
632
+ "step": 1010
633
+ },
634
+ {
635
+ "epoch": 29.14,
636
+ "learning_rate": 0.0004968,
637
+ "loss": 2.9524,
638
+ "step": 1020
639
+ },
640
+ {
641
+ "epoch": 29.43,
642
+ "learning_rate": 0.0004948,
643
+ "loss": 2.9605,
644
+ "step": 1030
645
+ },
646
+ {
647
+ "epoch": 29.71,
648
+ "learning_rate": 0.0004928,
649
+ "loss": 2.9451,
650
+ "step": 1040
651
+ },
652
+ {
653
+ "epoch": 30.0,
654
+ "learning_rate": 0.0004908,
655
+ "loss": 2.9759,
656
+ "step": 1050
657
+ },
658
+ {
659
+ "epoch": 30.29,
660
+ "learning_rate": 0.0004888000000000001,
661
+ "loss": 2.9576,
662
+ "step": 1060
663
+ },
664
+ {
665
+ "epoch": 30.57,
666
+ "learning_rate": 0.0004868,
667
+ "loss": 2.9476,
668
+ "step": 1070
669
+ },
670
+ {
671
+ "epoch": 30.86,
672
+ "learning_rate": 0.0004848,
673
+ "loss": 2.9421,
674
+ "step": 1080
675
+ },
676
+ {
677
+ "epoch": 31.14,
678
+ "learning_rate": 0.0004828,
679
+ "loss": 2.9508,
680
+ "step": 1090
681
+ },
682
+ {
683
+ "epoch": 31.43,
684
+ "learning_rate": 0.00048080000000000003,
685
+ "loss": 2.9596,
686
+ "step": 1100
687
+ },
688
+ {
689
+ "epoch": 31.71,
690
+ "learning_rate": 0.00047880000000000004,
691
+ "loss": 2.9904,
692
+ "step": 1110
693
+ },
694
+ {
695
+ "epoch": 32.0,
696
+ "learning_rate": 0.0004768,
697
+ "loss": 2.9336,
698
+ "step": 1120
699
+ },
700
+ {
701
+ "epoch": 32.29,
702
+ "learning_rate": 0.0004748,
703
+ "loss": 2.9416,
704
+ "step": 1130
705
+ },
706
+ {
707
+ "epoch": 32.57,
708
+ "learning_rate": 0.0004728,
709
+ "loss": 2.9831,
710
+ "step": 1140
711
+ },
712
+ {
713
+ "epoch": 32.86,
714
+ "learning_rate": 0.0004708,
715
+ "loss": 2.927,
716
+ "step": 1150
717
+ },
718
+ {
719
+ "epoch": 33.14,
720
+ "learning_rate": 0.0004688,
721
+ "loss": 2.9632,
722
+ "step": 1160
723
+ },
724
+ {
725
+ "epoch": 33.43,
726
+ "learning_rate": 0.0004668,
727
+ "loss": 2.9518,
728
+ "step": 1170
729
+ },
730
+ {
731
+ "epoch": 33.71,
732
+ "learning_rate": 0.0004648,
733
+ "loss": 2.9453,
734
+ "step": 1180
735
+ },
736
+ {
737
+ "epoch": 34.0,
738
+ "learning_rate": 0.0004628,
739
+ "loss": 2.9684,
740
+ "step": 1190
741
+ },
742
+ {
743
+ "epoch": 34.29,
744
+ "learning_rate": 0.0004608,
745
+ "loss": 2.9251,
746
+ "step": 1200
747
+ },
748
+ {
749
+ "epoch": 34.57,
750
+ "learning_rate": 0.0004588,
751
+ "loss": 2.9602,
752
+ "step": 1210
753
+ },
754
+ {
755
+ "epoch": 34.86,
756
+ "learning_rate": 0.0004568,
757
+ "loss": 2.9548,
758
+ "step": 1220
759
+ },
760
+ {
761
+ "epoch": 35.14,
762
+ "learning_rate": 0.0004548,
763
+ "loss": 2.9687,
764
+ "step": 1230
765
+ },
766
+ {
767
+ "epoch": 35.43,
768
+ "learning_rate": 0.0004528,
769
+ "loss": 2.9727,
770
+ "step": 1240
771
+ },
772
+ {
773
+ "epoch": 35.71,
774
+ "learning_rate": 0.0004508,
775
+ "loss": 2.9759,
776
+ "step": 1250
777
+ },
778
+ {
779
+ "epoch": 36.0,
780
+ "learning_rate": 0.0004488,
781
+ "loss": 2.9311,
782
+ "step": 1260
783
+ },
784
+ {
785
+ "epoch": 36.29,
786
+ "learning_rate": 0.00044679999999999996,
787
+ "loss": 2.9056,
788
+ "step": 1270
789
+ },
790
+ {
791
+ "epoch": 36.57,
792
+ "learning_rate": 0.00044479999999999997,
793
+ "loss": 2.9899,
794
+ "step": 1280
795
+ },
796
+ {
797
+ "epoch": 36.86,
798
+ "learning_rate": 0.00044280000000000003,
799
+ "loss": 2.9501,
800
+ "step": 1290
801
+ },
802
+ {
803
+ "epoch": 37.14,
804
+ "learning_rate": 0.00044080000000000004,
805
+ "loss": 2.9453,
806
+ "step": 1300
807
+ },
808
+ {
809
+ "epoch": 37.43,
810
+ "learning_rate": 0.00043880000000000004,
811
+ "loss": 2.9338,
812
+ "step": 1310
813
+ },
814
+ {
815
+ "epoch": 37.71,
816
+ "learning_rate": 0.00043680000000000005,
817
+ "loss": 2.9629,
818
+ "step": 1320
819
+ },
820
+ {
821
+ "epoch": 38.0,
822
+ "learning_rate": 0.00043480000000000005,
823
+ "loss": 2.9897,
824
+ "step": 1330
825
+ },
826
+ {
827
+ "epoch": 38.29,
828
+ "learning_rate": 0.0004328,
829
+ "loss": 2.9602,
830
+ "step": 1340
831
+ },
832
+ {
833
+ "epoch": 38.57,
834
+ "learning_rate": 0.0004308,
835
+ "loss": 2.9804,
836
+ "step": 1350
837
+ },
838
+ {
839
+ "epoch": 38.86,
840
+ "learning_rate": 0.0004288,
841
+ "loss": 2.9159,
842
+ "step": 1360
843
+ },
844
+ {
845
+ "epoch": 39.14,
846
+ "learning_rate": 0.0004268,
847
+ "loss": 2.9451,
848
+ "step": 1370
849
+ },
850
+ {
851
+ "epoch": 39.43,
852
+ "learning_rate": 0.0004248,
853
+ "loss": 2.9379,
854
+ "step": 1380
855
+ },
856
+ {
857
+ "epoch": 39.71,
858
+ "learning_rate": 0.00042280000000000003,
859
+ "loss": 2.9911,
860
+ "step": 1390
861
+ },
862
+ {
863
+ "epoch": 40.0,
864
+ "learning_rate": 0.00042080000000000004,
865
+ "loss": 2.9358,
866
+ "step": 1400
867
+ },
868
+ {
869
+ "epoch": 40.29,
870
+ "learning_rate": 0.0004188,
871
+ "loss": 2.9187,
872
+ "step": 1410
873
+ },
874
+ {
875
+ "epoch": 40.57,
876
+ "learning_rate": 0.0004168,
877
+ "loss": 3.0139,
878
+ "step": 1420
879
+ },
880
+ {
881
+ "epoch": 40.86,
882
+ "learning_rate": 0.0004148,
883
+ "loss": 2.9355,
884
+ "step": 1430
885
+ },
886
+ {
887
+ "epoch": 41.14,
888
+ "learning_rate": 0.0004128,
889
+ "loss": 2.9418,
890
+ "step": 1440
891
+ },
892
+ {
893
+ "epoch": 41.43,
894
+ "learning_rate": 0.0004108,
895
+ "loss": 2.9447,
896
+ "step": 1450
897
+ },
898
+ {
899
+ "epoch": 41.71,
900
+ "learning_rate": 0.0004088,
901
+ "loss": 2.9398,
902
+ "step": 1460
903
+ },
904
+ {
905
+ "epoch": 42.0,
906
+ "learning_rate": 0.0004068,
907
+ "loss": 2.9802,
908
+ "step": 1470
909
+ },
910
+ {
911
+ "epoch": 42.29,
912
+ "learning_rate": 0.00040480000000000003,
913
+ "loss": 2.9597,
914
+ "step": 1480
915
+ },
916
+ {
917
+ "epoch": 42.57,
918
+ "learning_rate": 0.0004028,
919
+ "loss": 2.9553,
920
+ "step": 1490
921
+ },
922
+ {
923
+ "epoch": 42.86,
924
+ "learning_rate": 0.0004008,
925
+ "loss": 2.9602,
926
+ "step": 1500
927
+ },
928
+ {
929
+ "epoch": 42.86,
930
+ "eval_loss": 2.9302072525024414,
931
+ "eval_runtime": 66.1205,
932
+ "eval_samples_per_second": 51.331,
933
+ "eval_steps_per_second": 0.817,
934
+ "eval_wer": 1.0,
935
+ "step": 1500
936
+ },
937
+ {
938
+ "epoch": 43.14,
939
+ "learning_rate": 0.0003988,
940
+ "loss": 2.9544,
941
+ "step": 1510
942
+ },
943
+ {
944
+ "epoch": 43.43,
945
+ "learning_rate": 0.0003968,
946
+ "loss": 2.9437,
947
+ "step": 1520
948
+ },
949
+ {
950
+ "epoch": 43.71,
951
+ "learning_rate": 0.0003948,
952
+ "loss": 2.942,
953
+ "step": 1530
954
+ },
955
+ {
956
+ "epoch": 44.0,
957
+ "learning_rate": 0.0003928,
958
+ "loss": 2.9629,
959
+ "step": 1540
960
+ },
961
+ {
962
+ "epoch": 44.29,
963
+ "learning_rate": 0.0003908,
964
+ "loss": 2.9859,
965
+ "step": 1550
966
+ },
967
+ {
968
+ "epoch": 44.57,
969
+ "learning_rate": 0.00038879999999999996,
970
+ "loss": 2.9469,
971
+ "step": 1560
972
+ },
973
+ {
974
+ "epoch": 44.86,
975
+ "learning_rate": 0.00038679999999999997,
976
+ "loss": 2.9093,
977
+ "step": 1570
978
+ },
979
+ {
980
+ "epoch": 45.14,
981
+ "learning_rate": 0.0003848,
982
+ "loss": 2.9465,
983
+ "step": 1580
984
+ },
985
+ {
986
+ "epoch": 45.43,
987
+ "learning_rate": 0.0003828,
988
+ "loss": 2.9415,
989
+ "step": 1590
990
+ },
991
+ {
992
+ "epoch": 45.71,
993
+ "learning_rate": 0.00038080000000000004,
994
+ "loss": 2.984,
995
+ "step": 1600
996
+ },
997
+ {
998
+ "epoch": 46.0,
999
+ "learning_rate": 0.00037880000000000005,
1000
+ "loss": 2.9673,
1001
+ "step": 1610
1002
+ },
1003
+ {
1004
+ "epoch": 46.29,
1005
+ "learning_rate": 0.00037680000000000005,
1006
+ "loss": 2.9158,
1007
+ "step": 1620
1008
+ },
1009
+ {
1010
+ "epoch": 46.57,
1011
+ "learning_rate": 0.0003748,
1012
+ "loss": 2.9977,
1013
+ "step": 1630
1014
+ },
1015
+ {
1016
+ "epoch": 46.86,
1017
+ "learning_rate": 0.0003728,
1018
+ "loss": 2.9243,
1019
+ "step": 1640
1020
+ },
1021
+ {
1022
+ "epoch": 47.14,
1023
+ "learning_rate": 0.0003708,
1024
+ "loss": 2.9622,
1025
+ "step": 1650
1026
+ },
1027
+ {
1028
+ "epoch": 47.43,
1029
+ "learning_rate": 0.0003688,
1030
+ "loss": 2.9526,
1031
+ "step": 1660
1032
+ },
1033
+ {
1034
+ "epoch": 47.71,
1035
+ "learning_rate": 0.0003668,
1036
+ "loss": 2.97,
1037
+ "step": 1670
1038
+ },
1039
+ {
1040
+ "epoch": 48.0,
1041
+ "learning_rate": 0.00036480000000000003,
1042
+ "loss": 2.9454,
1043
+ "step": 1680
1044
+ },
1045
+ {
1046
+ "epoch": 48.29,
1047
+ "learning_rate": 0.00036280000000000004,
1048
+ "loss": 2.9616,
1049
+ "step": 1690
1050
+ },
1051
+ {
1052
+ "epoch": 48.57,
1053
+ "learning_rate": 0.00036080000000000004,
1054
+ "loss": 2.9774,
1055
+ "step": 1700
1056
+ },
1057
+ {
1058
+ "epoch": 48.86,
1059
+ "learning_rate": 0.0003588,
1060
+ "loss": 2.9154,
1061
+ "step": 1710
1062
+ },
1063
+ {
1064
+ "epoch": 49.14,
1065
+ "learning_rate": 0.0003568,
1066
+ "loss": 2.9377,
1067
+ "step": 1720
1068
+ },
1069
+ {
1070
+ "epoch": 49.43,
1071
+ "learning_rate": 0.0003548,
1072
+ "loss": 2.9443,
1073
+ "step": 1730
1074
+ },
1075
+ {
1076
+ "epoch": 49.71,
1077
+ "learning_rate": 0.0003528,
1078
+ "loss": 2.9345,
1079
+ "step": 1740
1080
+ },
1081
+ {
1082
+ "epoch": 50.0,
1083
+ "learning_rate": 0.0003508,
1084
+ "loss": 2.9977,
1085
+ "step": 1750
1086
+ },
1087
+ {
1088
+ "epoch": 50.29,
1089
+ "learning_rate": 0.0003488,
1090
+ "loss": 2.9369,
1091
+ "step": 1760
1092
+ },
1093
+ {
1094
+ "epoch": 50.57,
1095
+ "learning_rate": 0.0003468,
1096
+ "loss": 2.9816,
1097
+ "step": 1770
1098
+ },
1099
+ {
1100
+ "epoch": 50.86,
1101
+ "learning_rate": 0.0003448,
1102
+ "loss": 2.9333,
1103
+ "step": 1780
1104
+ },
1105
+ {
1106
+ "epoch": 51.14,
1107
+ "learning_rate": 0.0003428,
1108
+ "loss": 2.9837,
1109
+ "step": 1790
1110
+ },
1111
+ {
1112
+ "epoch": 51.43,
1113
+ "learning_rate": 0.0003408,
1114
+ "loss": 2.9518,
1115
+ "step": 1800
1116
+ },
1117
+ {
1118
+ "epoch": 51.71,
1119
+ "learning_rate": 0.0003388,
1120
+ "loss": 2.9514,
1121
+ "step": 1810
1122
+ },
1123
+ {
1124
+ "epoch": 52.0,
1125
+ "learning_rate": 0.0003368,
1126
+ "loss": 2.9507,
1127
+ "step": 1820
1128
+ },
1129
+ {
1130
+ "epoch": 52.29,
1131
+ "learning_rate": 0.0003348,
1132
+ "loss": 2.9371,
1133
+ "step": 1830
1134
+ },
1135
+ {
1136
+ "epoch": 52.57,
1137
+ "learning_rate": 0.0003328,
1138
+ "loss": 2.9755,
1139
+ "step": 1840
1140
+ },
1141
+ {
1142
+ "epoch": 52.86,
1143
+ "learning_rate": 0.00033079999999999996,
1144
+ "loss": 2.9404,
1145
+ "step": 1850
1146
+ },
1147
+ {
1148
+ "epoch": 53.14,
1149
+ "learning_rate": 0.00032879999999999997,
1150
+ "loss": 2.9284,
1151
+ "step": 1860
1152
+ },
1153
+ {
1154
+ "epoch": 53.43,
1155
+ "learning_rate": 0.0003268,
1156
+ "loss": 2.9372,
1157
+ "step": 1870
1158
+ },
1159
+ {
1160
+ "epoch": 53.71,
1161
+ "learning_rate": 0.0003248,
1162
+ "loss": 2.9748,
1163
+ "step": 1880
1164
+ },
1165
+ {
1166
+ "epoch": 54.0,
1167
+ "learning_rate": 0.0003228,
1168
+ "loss": 2.9814,
1169
+ "step": 1890
1170
+ },
1171
+ {
1172
+ "epoch": 54.29,
1173
+ "learning_rate": 0.0003208,
1174
+ "loss": 2.9361,
1175
+ "step": 1900
1176
+ },
1177
+ {
1178
+ "epoch": 54.57,
1179
+ "learning_rate": 0.0003188,
1180
+ "loss": 2.9629,
1181
+ "step": 1910
1182
+ },
1183
+ {
1184
+ "epoch": 54.86,
1185
+ "learning_rate": 0.00031680000000000006,
1186
+ "loss": 2.9547,
1187
+ "step": 1920
1188
+ },
1189
+ {
1190
+ "epoch": 55.14,
1191
+ "learning_rate": 0.0003148,
1192
+ "loss": 2.914,
1193
+ "step": 1930
1194
+ },
1195
+ {
1196
+ "epoch": 55.43,
1197
+ "learning_rate": 0.0003128,
1198
+ "loss": 2.9773,
1199
+ "step": 1940
1200
+ },
1201
+ {
1202
+ "epoch": 55.71,
1203
+ "learning_rate": 0.0003108,
1204
+ "loss": 2.9639,
1205
+ "step": 1950
1206
+ },
1207
+ {
1208
+ "epoch": 56.0,
1209
+ "learning_rate": 0.0003088,
1210
+ "loss": 2.9579,
1211
+ "step": 1960
1212
+ },
1213
+ {
1214
+ "epoch": 56.29,
1215
+ "learning_rate": 0.00030680000000000003,
1216
+ "loss": 2.9494,
1217
+ "step": 1970
1218
+ },
1219
+ {
1220
+ "epoch": 56.57,
1221
+ "learning_rate": 0.00030480000000000004,
1222
+ "loss": 2.9539,
1223
+ "step": 1980
1224
+ },
1225
+ {
1226
+ "epoch": 56.86,
1227
+ "learning_rate": 0.00030280000000000004,
1228
+ "loss": 2.9248,
1229
+ "step": 1990
1230
+ },
1231
+ {
1232
+ "epoch": 57.14,
1233
+ "learning_rate": 0.0003008,
1234
+ "loss": 2.9586,
1235
+ "step": 2000
1236
+ },
1237
+ {
1238
+ "epoch": 57.14,
1239
+ "eval_loss": 2.9298441410064697,
1240
+ "eval_runtime": 65.7809,
1241
+ "eval_samples_per_second": 51.596,
1242
+ "eval_steps_per_second": 0.821,
1243
+ "eval_wer": 1.0,
1244
+ "step": 2000
1245
+ },
1246
+ {
1247
+ "epoch": 57.43,
1248
+ "learning_rate": 0.0002988,
1249
+ "loss": 2.9489,
1250
+ "step": 2010
1251
+ },
1252
+ {
1253
+ "epoch": 57.71,
1254
+ "learning_rate": 0.0002968,
1255
+ "loss": 2.9719,
1256
+ "step": 2020
1257
+ },
1258
+ {
1259
+ "epoch": 58.0,
1260
+ "learning_rate": 0.0002948,
1261
+ "loss": 2.9337,
1262
+ "step": 2030
1263
+ },
1264
+ {
1265
+ "epoch": 58.29,
1266
+ "learning_rate": 0.0002928,
1267
+ "loss": 2.9317,
1268
+ "step": 2040
1269
+ },
1270
+ {
1271
+ "epoch": 58.57,
1272
+ "learning_rate": 0.0002908,
1273
+ "loss": 2.9499,
1274
+ "step": 2050
1275
+ },
1276
+ {
1277
+ "epoch": 58.86,
1278
+ "learning_rate": 0.0002888,
1279
+ "loss": 2.9416,
1280
+ "step": 2060
1281
+ },
1282
+ {
1283
+ "epoch": 59.14,
1284
+ "learning_rate": 0.00028680000000000003,
1285
+ "loss": 2.9647,
1286
+ "step": 2070
1287
+ },
1288
+ {
1289
+ "epoch": 59.43,
1290
+ "learning_rate": 0.0002848,
1291
+ "loss": 2.971,
1292
+ "step": 2080
1293
+ },
1294
+ {
1295
+ "epoch": 59.71,
1296
+ "learning_rate": 0.0002828,
1297
+ "loss": 2.9469,
1298
+ "step": 2090
1299
+ },
1300
+ {
1301
+ "epoch": 60.0,
1302
+ "learning_rate": 0.0002808,
1303
+ "loss": 2.9519,
1304
+ "step": 2100
1305
+ },
1306
+ {
1307
+ "epoch": 60.29,
1308
+ "learning_rate": 0.0002788,
1309
+ "loss": 2.923,
1310
+ "step": 2110
1311
+ },
1312
+ {
1313
+ "epoch": 60.57,
1314
+ "learning_rate": 0.0002768,
1315
+ "loss": 2.9823,
1316
+ "step": 2120
1317
+ },
1318
+ {
1319
+ "epoch": 60.86,
1320
+ "learning_rate": 0.0002748,
1321
+ "loss": 2.9494,
1322
+ "step": 2130
1323
+ },
1324
+ {
1325
+ "epoch": 61.14,
1326
+ "learning_rate": 0.0002728,
1327
+ "loss": 2.9558,
1328
+ "step": 2140
1329
+ },
1330
+ {
1331
+ "epoch": 61.43,
1332
+ "learning_rate": 0.00027079999999999997,
1333
+ "loss": 2.9288,
1334
+ "step": 2150
1335
+ },
1336
+ {
1337
+ "epoch": 61.71,
1338
+ "learning_rate": 0.0002688,
1339
+ "loss": 2.9741,
1340
+ "step": 2160
1341
+ },
1342
+ {
1343
+ "epoch": 62.0,
1344
+ "learning_rate": 0.0002668,
1345
+ "loss": 2.9881,
1346
+ "step": 2170
1347
+ },
1348
+ {
1349
+ "epoch": 62.29,
1350
+ "learning_rate": 0.0002648,
1351
+ "loss": 2.9711,
1352
+ "step": 2180
1353
+ },
1354
+ {
1355
+ "epoch": 62.57,
1356
+ "learning_rate": 0.0002628,
1357
+ "loss": 2.9422,
1358
+ "step": 2190
1359
+ },
1360
+ {
1361
+ "epoch": 62.86,
1362
+ "learning_rate": 0.0002608,
1363
+ "loss": 2.9263,
1364
+ "step": 2200
1365
+ },
1366
+ {
1367
+ "epoch": 63.14,
1368
+ "learning_rate": 0.0002588,
1369
+ "loss": 2.9625,
1370
+ "step": 2210
1371
+ },
1372
+ {
1373
+ "epoch": 63.43,
1374
+ "learning_rate": 0.00025679999999999995,
1375
+ "loss": 2.9551,
1376
+ "step": 2220
1377
+ },
1378
+ {
1379
+ "epoch": 63.71,
1380
+ "learning_rate": 0.0002548,
1381
+ "loss": 2.9793,
1382
+ "step": 2230
1383
+ },
1384
+ {
1385
+ "epoch": 64.0,
1386
+ "learning_rate": 0.0002528,
1387
+ "loss": 2.9382,
1388
+ "step": 2240
1389
+ },
1390
+ {
1391
+ "epoch": 64.29,
1392
+ "learning_rate": 0.0002508,
1393
+ "loss": 2.924,
1394
+ "step": 2250
1395
+ },
1396
+ {
1397
+ "epoch": 64.57,
1398
+ "learning_rate": 0.0002488,
1399
+ "loss": 2.9732,
1400
+ "step": 2260
1401
+ },
1402
+ {
1403
+ "epoch": 64.86,
1404
+ "learning_rate": 0.0002468,
1405
+ "loss": 2.9312,
1406
+ "step": 2270
1407
+ },
1408
+ {
1409
+ "epoch": 65.14,
1410
+ "learning_rate": 0.0002448,
1411
+ "loss": 2.9755,
1412
+ "step": 2280
1413
+ },
1414
+ {
1415
+ "epoch": 65.43,
1416
+ "learning_rate": 0.0002428,
1417
+ "loss": 2.9621,
1418
+ "step": 2290
1419
+ },
1420
+ {
1421
+ "epoch": 65.71,
1422
+ "learning_rate": 0.0002408,
1423
+ "loss": 2.9642,
1424
+ "step": 2300
1425
+ },
1426
+ {
1427
+ "epoch": 66.0,
1428
+ "learning_rate": 0.0002388,
1429
+ "loss": 2.9313,
1430
+ "step": 2310
1431
+ },
1432
+ {
1433
+ "epoch": 66.29,
1434
+ "learning_rate": 0.0002368,
1435
+ "loss": 2.9545,
1436
+ "step": 2320
1437
+ },
1438
+ {
1439
+ "epoch": 66.57,
1440
+ "learning_rate": 0.00023480000000000002,
1441
+ "loss": 2.9405,
1442
+ "step": 2330
1443
+ },
1444
+ {
1445
+ "epoch": 66.86,
1446
+ "learning_rate": 0.00023280000000000002,
1447
+ "loss": 2.944,
1448
+ "step": 2340
1449
+ },
1450
+ {
1451
+ "epoch": 67.14,
1452
+ "learning_rate": 0.0002308,
1453
+ "loss": 2.9489,
1454
+ "step": 2350
1455
+ },
1456
+ {
1457
+ "epoch": 67.43,
1458
+ "learning_rate": 0.0002288,
1459
+ "loss": 2.9503,
1460
+ "step": 2360
1461
+ },
1462
+ {
1463
+ "epoch": 67.71,
1464
+ "learning_rate": 0.0002268,
1465
+ "loss": 2.9919,
1466
+ "step": 2370
1467
+ },
1468
+ {
1469
+ "epoch": 68.0,
1470
+ "learning_rate": 0.00022480000000000002,
1471
+ "loss": 2.9345,
1472
+ "step": 2380
1473
+ },
1474
+ {
1475
+ "epoch": 68.29,
1476
+ "learning_rate": 0.0002228,
1477
+ "loss": 2.9068,
1478
+ "step": 2390
1479
+ },
1480
+ {
1481
+ "epoch": 68.57,
1482
+ "learning_rate": 0.0002208,
1483
+ "loss": 2.98,
1484
+ "step": 2400
1485
+ },
1486
+ {
1487
+ "epoch": 68.86,
1488
+ "learning_rate": 0.0002188,
1489
+ "loss": 2.9278,
1490
+ "step": 2410
1491
+ },
1492
+ {
1493
+ "epoch": 69.14,
1494
+ "learning_rate": 0.00021679999999999998,
1495
+ "loss": 2.992,
1496
+ "step": 2420
1497
+ },
1498
+ {
1499
+ "epoch": 69.43,
1500
+ "learning_rate": 0.0002148,
1501
+ "loss": 2.9384,
1502
+ "step": 2430
1503
+ },
1504
+ {
1505
+ "epoch": 69.71,
1506
+ "learning_rate": 0.0002128,
1507
+ "loss": 2.9699,
1508
+ "step": 2440
1509
+ },
1510
+ {
1511
+ "epoch": 70.0,
1512
+ "learning_rate": 0.0002108,
1513
+ "loss": 2.9611,
1514
+ "step": 2450
1515
+ },
1516
+ {
1517
+ "epoch": 70.29,
1518
+ "learning_rate": 0.0002088,
1519
+ "loss": 2.9521,
1520
+ "step": 2460
1521
+ },
1522
+ {
1523
+ "epoch": 70.57,
1524
+ "learning_rate": 0.0002068,
1525
+ "loss": 2.9557,
1526
+ "step": 2470
1527
+ },
1528
+ {
1529
+ "epoch": 70.86,
1530
+ "learning_rate": 0.00020480000000000002,
1531
+ "loss": 2.9253,
1532
+ "step": 2480
1533
+ },
1534
+ {
1535
+ "epoch": 71.14,
1536
+ "learning_rate": 0.00020280000000000002,
1537
+ "loss": 2.9362,
1538
+ "step": 2490
1539
+ },
1540
+ {
1541
+ "epoch": 71.43,
1542
+ "learning_rate": 0.0002008,
1543
+ "loss": 2.9331,
1544
+ "step": 2500
1545
+ },
1546
+ {
1547
+ "epoch": 71.43,
1548
+ "eval_loss": 2.9313948154449463,
1549
+ "eval_runtime": 65.7925,
1550
+ "eval_samples_per_second": 51.586,
1551
+ "eval_steps_per_second": 0.821,
1552
+ "eval_wer": 1.0,
1553
+ "step": 2500
1554
+ },
1555
+ {
1556
+ "epoch": 71.71,
1557
+ "learning_rate": 0.0001988,
1558
+ "loss": 3.0128,
1559
+ "step": 2510
1560
+ },
1561
+ {
1562
+ "epoch": 72.0,
1563
+ "learning_rate": 0.0001968,
1564
+ "loss": 2.9246,
1565
+ "step": 2520
1566
+ },
1567
+ {
1568
+ "epoch": 72.29,
1569
+ "learning_rate": 0.0001948,
1570
+ "loss": 2.94,
1571
+ "step": 2530
1572
+ },
1573
+ {
1574
+ "epoch": 72.57,
1575
+ "learning_rate": 0.0001928,
1576
+ "loss": 2.9585,
1577
+ "step": 2540
1578
+ },
1579
+ {
1580
+ "epoch": 72.86,
1581
+ "learning_rate": 0.0001908,
1582
+ "loss": 2.9438,
1583
+ "step": 2550
1584
+ },
1585
+ {
1586
+ "epoch": 73.14,
1587
+ "learning_rate": 0.0001888,
1588
+ "loss": 2.9463,
1589
+ "step": 2560
1590
+ },
1591
+ {
1592
+ "epoch": 73.43,
1593
+ "learning_rate": 0.0001868,
1594
+ "loss": 2.9362,
1595
+ "step": 2570
1596
+ },
1597
+ {
1598
+ "epoch": 73.71,
1599
+ "learning_rate": 0.0001848,
1600
+ "loss": 2.9395,
1601
+ "step": 2580
1602
+ },
1603
+ {
1604
+ "epoch": 74.0,
1605
+ "learning_rate": 0.0001828,
1606
+ "loss": 2.9878,
1607
+ "step": 2590
1608
+ },
1609
+ {
1610
+ "epoch": 74.29,
1611
+ "learning_rate": 0.0001808,
1612
+ "loss": 2.9226,
1613
+ "step": 2600
1614
+ },
1615
+ {
1616
+ "epoch": 74.57,
1617
+ "learning_rate": 0.00017879999999999998,
1618
+ "loss": 2.9537,
1619
+ "step": 2610
1620
+ },
1621
+ {
1622
+ "epoch": 74.86,
1623
+ "learning_rate": 0.00017680000000000001,
1624
+ "loss": 2.9453,
1625
+ "step": 2620
1626
+ },
1627
+ {
1628
+ "epoch": 75.14,
1629
+ "learning_rate": 0.00017480000000000002,
1630
+ "loss": 2.9661,
1631
+ "step": 2630
1632
+ },
1633
+ {
1634
+ "epoch": 75.43,
1635
+ "learning_rate": 0.00017280000000000003,
1636
+ "loss": 2.9578,
1637
+ "step": 2640
1638
+ },
1639
+ {
1640
+ "epoch": 75.71,
1641
+ "learning_rate": 0.0001708,
1642
+ "loss": 2.9374,
1643
+ "step": 2650
1644
+ },
1645
+ {
1646
+ "epoch": 76.0,
1647
+ "learning_rate": 0.0001688,
1648
+ "loss": 2.9715,
1649
+ "step": 2660
1650
+ },
1651
+ {
1652
+ "epoch": 76.29,
1653
+ "learning_rate": 0.00016680000000000002,
1654
+ "loss": 2.9757,
1655
+ "step": 2670
1656
+ },
1657
+ {
1658
+ "epoch": 76.57,
1659
+ "learning_rate": 0.0001648,
1660
+ "loss": 2.9317,
1661
+ "step": 2680
1662
+ },
1663
+ {
1664
+ "epoch": 76.86,
1665
+ "learning_rate": 0.0001628,
1666
+ "loss": 2.9299,
1667
+ "step": 2690
1668
+ },
1669
+ {
1670
+ "epoch": 77.14,
1671
+ "learning_rate": 0.0001608,
1672
+ "loss": 2.9397,
1673
+ "step": 2700
1674
+ },
1675
+ {
1676
+ "epoch": 77.43,
1677
+ "learning_rate": 0.0001588,
1678
+ "loss": 2.9614,
1679
+ "step": 2710
1680
+ },
1681
+ {
1682
+ "epoch": 77.71,
1683
+ "learning_rate": 0.0001568,
1684
+ "loss": 2.9265,
1685
+ "step": 2720
1686
+ },
1687
+ {
1688
+ "epoch": 78.0,
1689
+ "learning_rate": 0.0001548,
1690
+ "loss": 2.9905,
1691
+ "step": 2730
1692
+ },
1693
+ {
1694
+ "epoch": 78.29,
1695
+ "learning_rate": 0.0001528,
1696
+ "loss": 2.9141,
1697
+ "step": 2740
1698
+ },
1699
+ {
1700
+ "epoch": 78.57,
1701
+ "learning_rate": 0.00015079999999999998,
1702
+ "loss": 2.9771,
1703
+ "step": 2750
1704
+ },
1705
+ {
1706
+ "epoch": 78.86,
1707
+ "learning_rate": 0.00014879999999999998,
1708
+ "loss": 2.9364,
1709
+ "step": 2760
1710
+ },
1711
+ {
1712
+ "epoch": 79.14,
1713
+ "learning_rate": 0.00014680000000000002,
1714
+ "loss": 2.9487,
1715
+ "step": 2770
1716
+ },
1717
+ {
1718
+ "epoch": 79.43,
1719
+ "learning_rate": 0.00014480000000000002,
1720
+ "loss": 2.9566,
1721
+ "step": 2780
1722
+ },
1723
+ {
1724
+ "epoch": 79.71,
1725
+ "learning_rate": 0.0001428,
1726
+ "loss": 2.9481,
1727
+ "step": 2790
1728
+ },
1729
+ {
1730
+ "epoch": 80.0,
1731
+ "learning_rate": 0.0001408,
1732
+ "loss": 2.9693,
1733
+ "step": 2800
1734
+ },
1735
+ {
1736
+ "epoch": 80.29,
1737
+ "learning_rate": 0.0001388,
1738
+ "loss": 2.9414,
1739
+ "step": 2810
1740
+ },
1741
+ {
1742
+ "epoch": 80.57,
1743
+ "learning_rate": 0.00013680000000000002,
1744
+ "loss": 2.9957,
1745
+ "step": 2820
1746
+ },
1747
+ {
1748
+ "epoch": 80.86,
1749
+ "learning_rate": 0.0001348,
1750
+ "loss": 2.934,
1751
+ "step": 2830
1752
+ },
1753
+ {
1754
+ "epoch": 81.14,
1755
+ "learning_rate": 0.0001328,
1756
+ "loss": 2.9112,
1757
+ "step": 2840
1758
+ },
1759
+ {
1760
+ "epoch": 81.43,
1761
+ "learning_rate": 0.0001308,
1762
+ "loss": 2.9581,
1763
+ "step": 2850
1764
+ },
1765
+ {
1766
+ "epoch": 81.71,
1767
+ "learning_rate": 0.00012880000000000001,
1768
+ "loss": 2.9418,
1769
+ "step": 2860
1770
+ },
1771
+ {
1772
+ "epoch": 82.0,
1773
+ "learning_rate": 0.0001268,
1774
+ "loss": 2.977,
1775
+ "step": 2870
1776
+ },
1777
+ {
1778
+ "epoch": 82.29,
1779
+ "learning_rate": 0.0001248,
1780
+ "loss": 2.9398,
1781
+ "step": 2880
1782
+ },
1783
+ {
1784
+ "epoch": 82.57,
1785
+ "learning_rate": 0.0001228,
1786
+ "loss": 2.9491,
1787
+ "step": 2890
1788
+ },
1789
+ {
1790
+ "epoch": 82.86,
1791
+ "learning_rate": 0.00012080000000000001,
1792
+ "loss": 2.9556,
1793
+ "step": 2900
1794
+ },
1795
+ {
1796
+ "epoch": 83.14,
1797
+ "learning_rate": 0.0001188,
1798
+ "loss": 2.9354,
1799
+ "step": 2910
1800
+ },
1801
+ {
1802
+ "epoch": 83.43,
1803
+ "learning_rate": 0.0001168,
1804
+ "loss": 2.9571,
1805
+ "step": 2920
1806
+ },
1807
+ {
1808
+ "epoch": 83.71,
1809
+ "learning_rate": 0.0001148,
1810
+ "loss": 2.934,
1811
+ "step": 2930
1812
+ },
1813
+ {
1814
+ "epoch": 84.0,
1815
+ "learning_rate": 0.0001128,
1816
+ "loss": 2.9884,
1817
+ "step": 2940
1818
+ },
1819
+ {
1820
+ "epoch": 84.29,
1821
+ "learning_rate": 0.0001108,
1822
+ "loss": 2.9358,
1823
+ "step": 2950
1824
+ },
1825
+ {
1826
+ "epoch": 84.57,
1827
+ "learning_rate": 0.0001088,
1828
+ "loss": 2.9325,
1829
+ "step": 2960
1830
+ },
1831
+ {
1832
+ "epoch": 84.86,
1833
+ "learning_rate": 0.00010680000000000001,
1834
+ "loss": 2.9796,
1835
+ "step": 2970
1836
+ },
1837
+ {
1838
+ "epoch": 85.14,
1839
+ "learning_rate": 0.00010480000000000001,
1840
+ "loss": 2.9281,
1841
+ "step": 2980
1842
+ },
1843
+ {
1844
+ "epoch": 85.43,
1845
+ "learning_rate": 0.0001028,
1846
+ "loss": 2.9646,
1847
+ "step": 2990
1848
+ },
1849
+ {
1850
+ "epoch": 85.71,
1851
+ "learning_rate": 0.0001008,
1852
+ "loss": 2.9321,
1853
+ "step": 3000
1854
+ },
1855
+ {
1856
+ "epoch": 85.71,
1857
+ "eval_loss": 2.9303746223449707,
1858
+ "eval_runtime": 66.1157,
1859
+ "eval_samples_per_second": 51.334,
1860
+ "eval_steps_per_second": 0.817,
1861
+ "eval_wer": 1.0,
1862
+ "step": 3000
1863
+ },
1864
+ {
1865
+ "epoch": 86.0,
1866
+ "learning_rate": 9.88e-05,
1867
+ "loss": 2.9703,
1868
+ "step": 3010
1869
+ },
1870
+ {
1871
+ "epoch": 86.29,
1872
+ "learning_rate": 9.68e-05,
1873
+ "loss": 2.9676,
1874
+ "step": 3020
1875
+ },
1876
+ {
1877
+ "epoch": 86.57,
1878
+ "learning_rate": 9.48e-05,
1879
+ "loss": 2.9489,
1880
+ "step": 3030
1881
+ },
1882
+ {
1883
+ "epoch": 86.86,
1884
+ "learning_rate": 9.279999999999999e-05,
1885
+ "loss": 2.927,
1886
+ "step": 3040
1887
+ },
1888
+ {
1889
+ "epoch": 87.14,
1890
+ "learning_rate": 9.080000000000001e-05,
1891
+ "loss": 2.9381,
1892
+ "step": 3050
1893
+ },
1894
+ {
1895
+ "epoch": 87.43,
1896
+ "learning_rate": 8.88e-05,
1897
+ "loss": 2.9631,
1898
+ "step": 3060
1899
+ },
1900
+ {
1901
+ "epoch": 87.71,
1902
+ "learning_rate": 8.680000000000001e-05,
1903
+ "loss": 2.9556,
1904
+ "step": 3070
1905
+ },
1906
+ {
1907
+ "epoch": 88.0,
1908
+ "learning_rate": 8.48e-05,
1909
+ "loss": 2.9472,
1910
+ "step": 3080
1911
+ },
1912
+ {
1913
+ "epoch": 88.29,
1914
+ "learning_rate": 8.280000000000001e-05,
1915
+ "loss": 2.9214,
1916
+ "step": 3090
1917
+ },
1918
+ {
1919
+ "epoch": 88.57,
1920
+ "learning_rate": 8.08e-05,
1921
+ "loss": 2.9877,
1922
+ "step": 3100
1923
+ },
1924
+ {
1925
+ "epoch": 88.86,
1926
+ "learning_rate": 7.879999999999999e-05,
1927
+ "loss": 2.9664,
1928
+ "step": 3110
1929
+ },
1930
+ {
1931
+ "epoch": 89.14,
1932
+ "learning_rate": 7.68e-05,
1933
+ "loss": 2.9161,
1934
+ "step": 3120
1935
+ },
1936
+ {
1937
+ "epoch": 89.43,
1938
+ "learning_rate": 7.48e-05,
1939
+ "loss": 2.9676,
1940
+ "step": 3130
1941
+ },
1942
+ {
1943
+ "epoch": 89.71,
1944
+ "learning_rate": 7.280000000000001e-05,
1945
+ "loss": 2.9451,
1946
+ "step": 3140
1947
+ },
1948
+ {
1949
+ "epoch": 90.0,
1950
+ "learning_rate": 7.08e-05,
1951
+ "loss": 2.956,
1952
+ "step": 3150
1953
+ },
1954
+ {
1955
+ "epoch": 90.29,
1956
+ "learning_rate": 6.88e-05,
1957
+ "loss": 2.9261,
1958
+ "step": 3160
1959
+ },
1960
+ {
1961
+ "epoch": 90.57,
1962
+ "learning_rate": 6.68e-05,
1963
+ "loss": 2.9622,
1964
+ "step": 3170
1965
+ },
1966
+ {
1967
+ "epoch": 90.86,
1968
+ "learning_rate": 6.48e-05,
1969
+ "loss": 2.9435,
1970
+ "step": 3180
1971
+ },
1972
+ {
1973
+ "epoch": 91.14,
1974
+ "learning_rate": 6.28e-05,
1975
+ "loss": 2.9906,
1976
+ "step": 3190
1977
+ },
1978
+ {
1979
+ "epoch": 91.43,
1980
+ "learning_rate": 6.08e-05,
1981
+ "loss": 2.925,
1982
+ "step": 3200
1983
+ },
1984
+ {
1985
+ "epoch": 91.71,
1986
+ "learning_rate": 5.88e-05,
1987
+ "loss": 2.9326,
1988
+ "step": 3210
1989
+ },
1990
+ {
1991
+ "epoch": 92.0,
1992
+ "learning_rate": 5.6800000000000005e-05,
1993
+ "loss": 2.9782,
1994
+ "step": 3220
1995
+ },
1996
+ {
1997
+ "epoch": 92.29,
1998
+ "learning_rate": 5.4800000000000004e-05,
1999
+ "loss": 2.9054,
2000
+ "step": 3230
2001
+ },
2002
+ {
2003
+ "epoch": 92.57,
2004
+ "learning_rate": 5.28e-05,
2005
+ "loss": 2.9512,
2006
+ "step": 3240
2007
+ },
2008
+ {
2009
+ "epoch": 92.86,
2010
+ "learning_rate": 5.08e-05,
2011
+ "loss": 2.9623,
2012
+ "step": 3250
2013
+ },
2014
+ {
2015
+ "epoch": 93.14,
2016
+ "learning_rate": 4.880000000000001e-05,
2017
+ "loss": 2.9675,
2018
+ "step": 3260
2019
+ },
2020
+ {
2021
+ "epoch": 93.43,
2022
+ "learning_rate": 4.68e-05,
2023
+ "loss": 2.9418,
2024
+ "step": 3270
2025
+ },
2026
+ {
2027
+ "epoch": 93.71,
2028
+ "learning_rate": 4.48e-05,
2029
+ "loss": 2.9781,
2030
+ "step": 3280
2031
+ },
2032
+ {
2033
+ "epoch": 94.0,
2034
+ "learning_rate": 4.28e-05,
2035
+ "loss": 2.929,
2036
+ "step": 3290
2037
+ },
2038
+ {
2039
+ "epoch": 94.29,
2040
+ "learning_rate": 4.08e-05,
2041
+ "loss": 2.9389,
2042
+ "step": 3300
2043
+ },
2044
+ {
2045
+ "epoch": 94.57,
2046
+ "learning_rate": 3.88e-05,
2047
+ "loss": 2.9574,
2048
+ "step": 3310
2049
+ },
2050
+ {
2051
+ "epoch": 94.86,
2052
+ "learning_rate": 3.68e-05,
2053
+ "loss": 2.9591,
2054
+ "step": 3320
2055
+ },
2056
+ {
2057
+ "epoch": 95.14,
2058
+ "learning_rate": 3.48e-05,
2059
+ "loss": 2.9468,
2060
+ "step": 3330
2061
+ },
2062
+ {
2063
+ "epoch": 95.43,
2064
+ "learning_rate": 3.2800000000000004e-05,
2065
+ "loss": 2.9147,
2066
+ "step": 3340
2067
+ },
2068
+ {
2069
+ "epoch": 95.71,
2070
+ "learning_rate": 3.08e-05,
2071
+ "loss": 2.9729,
2072
+ "step": 3350
2073
+ },
2074
+ {
2075
+ "epoch": 96.0,
2076
+ "learning_rate": 2.88e-05,
2077
+ "loss": 2.9574,
2078
+ "step": 3360
2079
+ },
2080
+ {
2081
+ "epoch": 96.29,
2082
+ "learning_rate": 2.68e-05,
2083
+ "loss": 2.9449,
2084
+ "step": 3370
2085
+ },
2086
+ {
2087
+ "epoch": 96.57,
2088
+ "learning_rate": 2.48e-05,
2089
+ "loss": 2.9767,
2090
+ "step": 3380
2091
+ },
2092
+ {
2093
+ "epoch": 96.86,
2094
+ "learning_rate": 2.2800000000000002e-05,
2095
+ "loss": 2.9346,
2096
+ "step": 3390
2097
+ },
2098
+ {
2099
+ "epoch": 97.14,
2100
+ "learning_rate": 2.08e-05,
2101
+ "loss": 2.9423,
2102
+ "step": 3400
2103
+ },
2104
+ {
2105
+ "epoch": 97.43,
2106
+ "learning_rate": 1.88e-05,
2107
+ "loss": 2.9399,
2108
+ "step": 3410
2109
+ },
2110
+ {
2111
+ "epoch": 97.71,
2112
+ "learning_rate": 1.68e-05,
2113
+ "loss": 2.9729,
2114
+ "step": 3420
2115
+ },
2116
+ {
2117
+ "epoch": 98.0,
2118
+ "learning_rate": 1.48e-05,
2119
+ "loss": 2.9432,
2120
+ "step": 3430
2121
+ },
2122
+ {
2123
+ "epoch": 98.29,
2124
+ "learning_rate": 1.2800000000000001e-05,
2125
+ "loss": 2.9322,
2126
+ "step": 3440
2127
+ },
2128
+ {
2129
+ "epoch": 98.57,
2130
+ "learning_rate": 1.0800000000000002e-05,
2131
+ "loss": 2.989,
2132
+ "step": 3450
2133
+ },
2134
+ {
2135
+ "epoch": 98.86,
2136
+ "learning_rate": 8.8e-06,
2137
+ "loss": 2.9351,
2138
+ "step": 3460
2139
+ },
2140
+ {
2141
+ "epoch": 99.14,
2142
+ "learning_rate": 6.8e-06,
2143
+ "loss": 2.9286,
2144
+ "step": 3470
2145
+ },
2146
+ {
2147
+ "epoch": 99.43,
2148
+ "learning_rate": 4.8e-06,
2149
+ "loss": 2.9326,
2150
+ "step": 3480
2151
+ },
2152
+ {
2153
+ "epoch": 99.71,
2154
+ "learning_rate": 2.8e-06,
2155
+ "loss": 2.962,
2156
+ "step": 3490
2157
+ },
2158
+ {
2159
+ "epoch": 100.0,
2160
+ "learning_rate": 8.000000000000001e-07,
2161
+ "loss": 2.9652,
2162
+ "step": 3500
2163
+ },
2164
+ {
2165
+ "epoch": 100.0,
2166
+ "eval_loss": 2.9303946495056152,
2167
+ "eval_runtime": 66.0696,
2168
+ "eval_samples_per_second": 51.37,
2169
+ "eval_steps_per_second": 0.817,
2170
+ "eval_wer": 1.0,
2171
+ "step": 3500
2172
+ },
2173
+ {
2174
+ "epoch": 100.0,
2175
+ "step": 3500,
2176
+ "total_flos": 3.191123523691086e+19,
2177
+ "train_loss": 3.4216162567138673,
2178
+ "train_runtime": 5111.6283,
2179
+ "train_samples_per_second": 42.922,
2180
+ "train_steps_per_second": 0.685
2181
+ }
2182
+ ],
2183
+ "max_steps": 3500,
2184
+ "num_train_epochs": 100,
2185
+ "total_flos": 3.191123523691086e+19,
2186
+ "trial_name": null,
2187
+ "trial_params": null
2188
+ }