AndrewMcDowell commited on
Commit
c950bb2
1 Parent(s): 97f1966

End of training

Browse files
Files changed (4) hide show
  1. all_results.json +12 -12
  2. eval_results.json +7 -7
  3. train_results.json +6 -6
  4. trainer_state.json +2371 -0
all_results.json CHANGED
@@ -1,14 +1,14 @@
1
  {
2
- "epoch": 2.0,
3
- "eval_loss": 0.14759068191051483,
4
- "eval_runtime": 1022.7659,
5
- "eval_samples": 16007,
6
- "eval_samples_per_second": 15.651,
7
- "eval_steps_per_second": 1.956,
8
- "eval_wer": 0.16119714847590955,
9
- "train_loss": 0.32739020716330625,
10
- "train_runtime": 49115.8494,
11
- "train_samples": 436168,
12
- "train_samples_per_second": 17.761,
13
- "train_steps_per_second": 0.555
14
  }
 
1
  {
2
+ "epoch": 2.5,
3
+ "eval_loss": 0.13549980521202087,
4
+ "eval_runtime": 1012.9633,
5
+ "eval_samples": 15995,
6
+ "eval_samples_per_second": 15.79,
7
+ "eval_steps_per_second": 1.974,
8
+ "eval_wer": 0.15316115109878853,
9
+ "train_loss": 1.0034960915109994,
10
+ "train_runtime": 150795.9745,
11
+ "train_samples": 436158,
12
+ "train_samples_per_second": 7.231,
13
+ "train_steps_per_second": 0.226
14
  }
eval_results.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
- "epoch": 2.0,
3
- "eval_loss": 0.14759068191051483,
4
- "eval_runtime": 1022.7659,
5
- "eval_samples": 16007,
6
- "eval_samples_per_second": 15.651,
7
- "eval_steps_per_second": 1.956,
8
- "eval_wer": 0.16119714847590955
9
  }
 
1
  {
2
+ "epoch": 2.5,
3
+ "eval_loss": 0.13549980521202087,
4
+ "eval_runtime": 1012.9633,
5
+ "eval_samples": 15995,
6
+ "eval_samples_per_second": 15.79,
7
+ "eval_steps_per_second": 1.974,
8
+ "eval_wer": 0.15316115109878853
9
  }
train_results.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
- "epoch": 2.0,
3
- "train_loss": 0.32739020716330625,
4
- "train_runtime": 49115.8494,
5
- "train_samples": 436168,
6
- "train_samples_per_second": 17.761,
7
- "train_steps_per_second": 0.555
8
  }
 
1
  {
2
+ "epoch": 2.5,
3
+ "train_loss": 1.0034960915109994,
4
+ "train_runtime": 150795.9745,
5
+ "train_samples": 436158,
6
+ "train_samples_per_second": 7.231,
7
+ "train_steps_per_second": 0.226
8
  }
trainer_state.json ADDED
@@ -0,0 +1,2371 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.5,
5
+ "global_step": 34075,
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": 3.675e-06,
13
+ "loss": 8.685,
14
+ "step": 100
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 7.425e-06,
19
+ "loss": 3.0395,
20
+ "step": 200
21
+ },
22
+ {
23
+ "epoch": 0.02,
24
+ "learning_rate": 1.1174999999999999e-05,
25
+ "loss": 2.9522,
26
+ "step": 300
27
+ },
28
+ {
29
+ "epoch": 0.03,
30
+ "learning_rate": 1.4925e-05,
31
+ "loss": 2.4455,
32
+ "step": 400
33
+ },
34
+ {
35
+ "epoch": 0.04,
36
+ "learning_rate": 1.8675e-05,
37
+ "loss": 1.7543,
38
+ "step": 500
39
+ },
40
+ {
41
+ "epoch": 0.04,
42
+ "learning_rate": 2.2424999999999996e-05,
43
+ "loss": 1.4833,
44
+ "step": 600
45
+ },
46
+ {
47
+ "epoch": 0.05,
48
+ "learning_rate": 2.6174999999999996e-05,
49
+ "loss": 1.2163,
50
+ "step": 700
51
+ },
52
+ {
53
+ "epoch": 0.06,
54
+ "learning_rate": 2.9925e-05,
55
+ "loss": 1.1407,
56
+ "step": 800
57
+ },
58
+ {
59
+ "epoch": 0.07,
60
+ "learning_rate": 3.3675e-05,
61
+ "loss": 1.1119,
62
+ "step": 900
63
+ },
64
+ {
65
+ "epoch": 0.07,
66
+ "learning_rate": 3.7424999999999995e-05,
67
+ "loss": 1.0826,
68
+ "step": 1000
69
+ },
70
+ {
71
+ "epoch": 0.07,
72
+ "eval_loss": 0.46367982029914856,
73
+ "eval_runtime": 1042.4054,
74
+ "eval_samples_per_second": 15.344,
75
+ "eval_steps_per_second": 1.919,
76
+ "eval_wer": 0.4653905513746807,
77
+ "step": 1000
78
+ },
79
+ {
80
+ "epoch": 0.08,
81
+ "learning_rate": 4.1175e-05,
82
+ "loss": 1.0901,
83
+ "step": 1100
84
+ },
85
+ {
86
+ "epoch": 0.09,
87
+ "learning_rate": 4.4924999999999994e-05,
88
+ "loss": 1.0794,
89
+ "step": 1200
90
+ },
91
+ {
92
+ "epoch": 0.1,
93
+ "learning_rate": 4.8675e-05,
94
+ "loss": 1.0774,
95
+ "step": 1300
96
+ },
97
+ {
98
+ "epoch": 0.1,
99
+ "learning_rate": 5.2424999999999994e-05,
100
+ "loss": 1.0604,
101
+ "step": 1400
102
+ },
103
+ {
104
+ "epoch": 0.11,
105
+ "learning_rate": 5.6175e-05,
106
+ "loss": 1.084,
107
+ "step": 1500
108
+ },
109
+ {
110
+ "epoch": 0.12,
111
+ "learning_rate": 5.9925e-05,
112
+ "loss": 1.0959,
113
+ "step": 1600
114
+ },
115
+ {
116
+ "epoch": 0.12,
117
+ "learning_rate": 6.367499999999999e-05,
118
+ "loss": 1.0873,
119
+ "step": 1700
120
+ },
121
+ {
122
+ "epoch": 0.13,
123
+ "learning_rate": 6.7425e-05,
124
+ "loss": 1.0938,
125
+ "step": 1800
126
+ },
127
+ {
128
+ "epoch": 0.14,
129
+ "learning_rate": 7.1175e-05,
130
+ "loss": 1.1218,
131
+ "step": 1900
132
+ },
133
+ {
134
+ "epoch": 0.15,
135
+ "learning_rate": 7.492499999999999e-05,
136
+ "loss": 1.118,
137
+ "step": 2000
138
+ },
139
+ {
140
+ "epoch": 0.15,
141
+ "eval_loss": 0.25947731733322144,
142
+ "eval_runtime": 1033.4206,
143
+ "eval_samples_per_second": 15.478,
144
+ "eval_steps_per_second": 1.935,
145
+ "eval_wer": 0.2686807708592267,
146
+ "step": 2000
147
+ },
148
+ {
149
+ "epoch": 0.15,
150
+ "learning_rate": 7.477084957131722e-05,
151
+ "loss": 1.1204,
152
+ "step": 2100
153
+ },
154
+ {
155
+ "epoch": 0.16,
156
+ "learning_rate": 7.453702260327356e-05,
157
+ "loss": 1.1267,
158
+ "step": 2200
159
+ },
160
+ {
161
+ "epoch": 0.17,
162
+ "learning_rate": 7.430319563522992e-05,
163
+ "loss": 1.1174,
164
+ "step": 2300
165
+ },
166
+ {
167
+ "epoch": 0.18,
168
+ "learning_rate": 7.406936866718628e-05,
169
+ "loss": 1.1229,
170
+ "step": 2400
171
+ },
172
+ {
173
+ "epoch": 0.18,
174
+ "learning_rate": 7.383554169914263e-05,
175
+ "loss": 1.1171,
176
+ "step": 2500
177
+ },
178
+ {
179
+ "epoch": 0.19,
180
+ "learning_rate": 7.360171473109897e-05,
181
+ "loss": 1.1209,
182
+ "step": 2600
183
+ },
184
+ {
185
+ "epoch": 0.2,
186
+ "learning_rate": 7.336788776305533e-05,
187
+ "loss": 1.1378,
188
+ "step": 2700
189
+ },
190
+ {
191
+ "epoch": 0.21,
192
+ "learning_rate": 7.313406079501168e-05,
193
+ "loss": 1.1187,
194
+ "step": 2800
195
+ },
196
+ {
197
+ "epoch": 0.21,
198
+ "learning_rate": 7.290023382696804e-05,
199
+ "loss": 1.1289,
200
+ "step": 2900
201
+ },
202
+ {
203
+ "epoch": 0.22,
204
+ "learning_rate": 7.26664068589244e-05,
205
+ "loss": 1.1268,
206
+ "step": 3000
207
+ },
208
+ {
209
+ "epoch": 0.22,
210
+ "eval_loss": 0.26353907585144043,
211
+ "eval_runtime": 1021.5774,
212
+ "eval_samples_per_second": 15.657,
213
+ "eval_steps_per_second": 1.958,
214
+ "eval_wer": 0.26611306117431743,
215
+ "step": 3000
216
+ },
217
+ {
218
+ "epoch": 0.23,
219
+ "learning_rate": 7.243257989088074e-05,
220
+ "loss": 1.1198,
221
+ "step": 3100
222
+ },
223
+ {
224
+ "epoch": 0.23,
225
+ "learning_rate": 7.219875292283709e-05,
226
+ "loss": 1.1013,
227
+ "step": 3200
228
+ },
229
+ {
230
+ "epoch": 0.24,
231
+ "learning_rate": 7.196492595479345e-05,
232
+ "loss": 1.1215,
233
+ "step": 3300
234
+ },
235
+ {
236
+ "epoch": 0.25,
237
+ "learning_rate": 7.17310989867498e-05,
238
+ "loss": 1.1323,
239
+ "step": 3400
240
+ },
241
+ {
242
+ "epoch": 0.26,
243
+ "learning_rate": 7.149727201870615e-05,
244
+ "loss": 1.1317,
245
+ "step": 3500
246
+ },
247
+ {
248
+ "epoch": 0.26,
249
+ "learning_rate": 7.126344505066251e-05,
250
+ "loss": 1.1058,
251
+ "step": 3600
252
+ },
253
+ {
254
+ "epoch": 0.27,
255
+ "learning_rate": 7.102961808261886e-05,
256
+ "loss": 1.0934,
257
+ "step": 3700
258
+ },
259
+ {
260
+ "epoch": 0.28,
261
+ "learning_rate": 7.07957911145752e-05,
262
+ "loss": 1.1386,
263
+ "step": 3800
264
+ },
265
+ {
266
+ "epoch": 0.29,
267
+ "learning_rate": 7.056196414653156e-05,
268
+ "loss": 1.0968,
269
+ "step": 3900
270
+ },
271
+ {
272
+ "epoch": 0.29,
273
+ "learning_rate": 7.032813717848791e-05,
274
+ "loss": 1.0919,
275
+ "step": 4000
276
+ },
277
+ {
278
+ "epoch": 0.29,
279
+ "eval_loss": 0.24173684418201447,
280
+ "eval_runtime": 1025.0912,
281
+ "eval_samples_per_second": 15.603,
282
+ "eval_steps_per_second": 1.951,
283
+ "eval_wer": 0.256627559173416,
284
+ "step": 4000
285
+ },
286
+ {
287
+ "epoch": 0.3,
288
+ "learning_rate": 7.009431021044427e-05,
289
+ "loss": 1.1112,
290
+ "step": 4100
291
+ },
292
+ {
293
+ "epoch": 0.31,
294
+ "learning_rate": 6.986048324240062e-05,
295
+ "loss": 1.1122,
296
+ "step": 4200
297
+ },
298
+ {
299
+ "epoch": 0.32,
300
+ "learning_rate": 6.962665627435696e-05,
301
+ "loss": 1.1185,
302
+ "step": 4300
303
+ },
304
+ {
305
+ "epoch": 0.32,
306
+ "learning_rate": 6.939282930631332e-05,
307
+ "loss": 1.1078,
308
+ "step": 4400
309
+ },
310
+ {
311
+ "epoch": 0.33,
312
+ "learning_rate": 6.915900233826968e-05,
313
+ "loss": 1.0842,
314
+ "step": 4500
315
+ },
316
+ {
317
+ "epoch": 0.34,
318
+ "learning_rate": 6.892517537022603e-05,
319
+ "loss": 1.0929,
320
+ "step": 4600
321
+ },
322
+ {
323
+ "epoch": 0.34,
324
+ "learning_rate": 6.869368667186282e-05,
325
+ "loss": 1.1068,
326
+ "step": 4700
327
+ },
328
+ {
329
+ "epoch": 0.35,
330
+ "learning_rate": 6.845985970381916e-05,
331
+ "loss": 1.1078,
332
+ "step": 4800
333
+ },
334
+ {
335
+ "epoch": 0.36,
336
+ "learning_rate": 6.822603273577552e-05,
337
+ "loss": 1.1208,
338
+ "step": 4900
339
+ },
340
+ {
341
+ "epoch": 0.37,
342
+ "learning_rate": 6.799220576773187e-05,
343
+ "loss": 1.1013,
344
+ "step": 5000
345
+ },
346
+ {
347
+ "epoch": 0.37,
348
+ "eval_loss": 0.24144533276557922,
349
+ "eval_runtime": 1024.2346,
350
+ "eval_samples_per_second": 15.617,
351
+ "eval_steps_per_second": 1.953,
352
+ "eval_wer": 0.2567436524304465,
353
+ "step": 5000
354
+ },
355
+ {
356
+ "epoch": 0.37,
357
+ "learning_rate": 6.775837879968823e-05,
358
+ "loss": 1.0729,
359
+ "step": 5100
360
+ },
361
+ {
362
+ "epoch": 0.38,
363
+ "learning_rate": 6.752455183164457e-05,
364
+ "loss": 1.0884,
365
+ "step": 5200
366
+ },
367
+ {
368
+ "epoch": 0.39,
369
+ "learning_rate": 6.729072486360093e-05,
370
+ "loss": 1.0744,
371
+ "step": 5300
372
+ },
373
+ {
374
+ "epoch": 0.4,
375
+ "learning_rate": 6.705689789555728e-05,
376
+ "loss": 1.0939,
377
+ "step": 5400
378
+ },
379
+ {
380
+ "epoch": 0.4,
381
+ "learning_rate": 6.682307092751364e-05,
382
+ "loss": 1.0989,
383
+ "step": 5500
384
+ },
385
+ {
386
+ "epoch": 0.41,
387
+ "learning_rate": 6.658924395946998e-05,
388
+ "loss": 1.0922,
389
+ "step": 5600
390
+ },
391
+ {
392
+ "epoch": 0.42,
393
+ "learning_rate": 6.635541699142633e-05,
394
+ "loss": 1.0885,
395
+ "step": 5700
396
+ },
397
+ {
398
+ "epoch": 0.43,
399
+ "learning_rate": 6.612159002338269e-05,
400
+ "loss": 1.0909,
401
+ "step": 5800
402
+ },
403
+ {
404
+ "epoch": 0.43,
405
+ "learning_rate": 6.588776305533905e-05,
406
+ "loss": 1.0773,
407
+ "step": 5900
408
+ },
409
+ {
410
+ "epoch": 0.44,
411
+ "learning_rate": 6.56539360872954e-05,
412
+ "loss": 1.0898,
413
+ "step": 6000
414
+ },
415
+ {
416
+ "epoch": 0.44,
417
+ "eval_loss": 0.25459304451942444,
418
+ "eval_runtime": 1029.5775,
419
+ "eval_samples_per_second": 15.535,
420
+ "eval_steps_per_second": 1.943,
421
+ "eval_wer": 0.27307182758102627,
422
+ "step": 6000
423
+ },
424
+ {
425
+ "epoch": 0.45,
426
+ "learning_rate": 6.542010911925175e-05,
427
+ "loss": 1.069,
428
+ "step": 6100
429
+ },
430
+ {
431
+ "epoch": 0.45,
432
+ "learning_rate": 6.51862821512081e-05,
433
+ "loss": 1.0899,
434
+ "step": 6200
435
+ },
436
+ {
437
+ "epoch": 0.46,
438
+ "learning_rate": 6.495245518316445e-05,
439
+ "loss": 1.0907,
440
+ "step": 6300
441
+ },
442
+ {
443
+ "epoch": 0.47,
444
+ "learning_rate": 6.47186282151208e-05,
445
+ "loss": 1.0875,
446
+ "step": 6400
447
+ },
448
+ {
449
+ "epoch": 0.48,
450
+ "learning_rate": 6.448480124707717e-05,
451
+ "loss": 1.0729,
452
+ "step": 6500
453
+ },
454
+ {
455
+ "epoch": 0.48,
456
+ "learning_rate": 6.425097427903351e-05,
457
+ "loss": 1.0823,
458
+ "step": 6600
459
+ },
460
+ {
461
+ "epoch": 0.49,
462
+ "learning_rate": 6.401714731098986e-05,
463
+ "loss": 1.0558,
464
+ "step": 6700
465
+ },
466
+ {
467
+ "epoch": 0.5,
468
+ "learning_rate": 6.378332034294622e-05,
469
+ "loss": 1.0742,
470
+ "step": 6800
471
+ },
472
+ {
473
+ "epoch": 0.51,
474
+ "learning_rate": 6.354949337490256e-05,
475
+ "loss": 1.0748,
476
+ "step": 6900
477
+ },
478
+ {
479
+ "epoch": 0.51,
480
+ "learning_rate": 6.331566640685892e-05,
481
+ "loss": 1.0808,
482
+ "step": 7000
483
+ },
484
+ {
485
+ "epoch": 0.51,
486
+ "eval_loss": 0.23990128934383392,
487
+ "eval_runtime": 1020.4541,
488
+ "eval_samples_per_second": 15.674,
489
+ "eval_steps_per_second": 1.96,
490
+ "eval_wer": 0.25348621221847384,
491
+ "step": 7000
492
+ },
493
+ {
494
+ "epoch": 0.52,
495
+ "learning_rate": 6.308183943881527e-05,
496
+ "loss": 1.0688,
497
+ "step": 7100
498
+ },
499
+ {
500
+ "epoch": 0.53,
501
+ "learning_rate": 6.284801247077163e-05,
502
+ "loss": 1.065,
503
+ "step": 7200
504
+ },
505
+ {
506
+ "epoch": 0.54,
507
+ "learning_rate": 6.261418550272797e-05,
508
+ "loss": 1.0603,
509
+ "step": 7300
510
+ },
511
+ {
512
+ "epoch": 0.54,
513
+ "learning_rate": 6.238035853468433e-05,
514
+ "loss": 1.072,
515
+ "step": 7400
516
+ },
517
+ {
518
+ "epoch": 0.55,
519
+ "learning_rate": 6.214653156664068e-05,
520
+ "loss": 1.0736,
521
+ "step": 7500
522
+ },
523
+ {
524
+ "epoch": 0.56,
525
+ "learning_rate": 6.191270459859704e-05,
526
+ "loss": 1.066,
527
+ "step": 7600
528
+ },
529
+ {
530
+ "epoch": 0.56,
531
+ "learning_rate": 6.167887763055338e-05,
532
+ "loss": 1.0761,
533
+ "step": 7700
534
+ },
535
+ {
536
+ "epoch": 0.57,
537
+ "learning_rate": 6.144505066250973e-05,
538
+ "loss": 1.0824,
539
+ "step": 7800
540
+ },
541
+ {
542
+ "epoch": 0.58,
543
+ "learning_rate": 6.121122369446609e-05,
544
+ "loss": 1.0571,
545
+ "step": 7900
546
+ },
547
+ {
548
+ "epoch": 0.59,
549
+ "learning_rate": 6.097973499610288e-05,
550
+ "loss": 1.0719,
551
+ "step": 8000
552
+ },
553
+ {
554
+ "epoch": 0.59,
555
+ "eval_loss": 0.23534023761749268,
556
+ "eval_runtime": 1024.9025,
557
+ "eval_samples_per_second": 15.606,
558
+ "eval_steps_per_second": 1.951,
559
+ "eval_wer": 0.2527691656309327,
560
+ "step": 8000
561
+ },
562
+ {
563
+ "epoch": 0.59,
564
+ "learning_rate": 6.074590802805923e-05,
565
+ "loss": 1.0739,
566
+ "step": 8100
567
+ },
568
+ {
569
+ "epoch": 0.6,
570
+ "learning_rate": 6.051208106001558e-05,
571
+ "loss": 1.0628,
572
+ "step": 8200
573
+ },
574
+ {
575
+ "epoch": 0.61,
576
+ "learning_rate": 6.027825409197193e-05,
577
+ "loss": 1.0791,
578
+ "step": 8300
579
+ },
580
+ {
581
+ "epoch": 0.62,
582
+ "learning_rate": 6.004442712392829e-05,
583
+ "loss": 1.0769,
584
+ "step": 8400
585
+ },
586
+ {
587
+ "epoch": 0.62,
588
+ "learning_rate": 5.981060015588464e-05,
589
+ "loss": 1.0729,
590
+ "step": 8500
591
+ },
592
+ {
593
+ "epoch": 0.63,
594
+ "learning_rate": 5.9576773187840996e-05,
595
+ "loss": 1.0848,
596
+ "step": 8600
597
+ },
598
+ {
599
+ "epoch": 0.64,
600
+ "learning_rate": 5.934294621979734e-05,
601
+ "loss": 1.0636,
602
+ "step": 8700
603
+ },
604
+ {
605
+ "epoch": 0.65,
606
+ "learning_rate": 5.9109119251753694e-05,
607
+ "loss": 1.0537,
608
+ "step": 8800
609
+ },
610
+ {
611
+ "epoch": 0.65,
612
+ "learning_rate": 5.887529228371005e-05,
613
+ "loss": 1.0623,
614
+ "step": 8900
615
+ },
616
+ {
617
+ "epoch": 0.66,
618
+ "learning_rate": 5.86414653156664e-05,
619
+ "loss": 1.0446,
620
+ "step": 9000
621
+ },
622
+ {
623
+ "epoch": 0.66,
624
+ "eval_loss": 0.24269308149814606,
625
+ "eval_runtime": 1017.8128,
626
+ "eval_samples_per_second": 15.715,
627
+ "eval_steps_per_second": 1.965,
628
+ "eval_wer": 0.25453788054686755,
629
+ "step": 9000
630
+ },
631
+ {
632
+ "epoch": 0.67,
633
+ "learning_rate": 5.840763834762276e-05,
634
+ "loss": 1.0692,
635
+ "step": 9100
636
+ },
637
+ {
638
+ "epoch": 0.67,
639
+ "learning_rate": 5.817614964925954e-05,
640
+ "loss": 1.042,
641
+ "step": 9200
642
+ },
643
+ {
644
+ "epoch": 0.68,
645
+ "learning_rate": 5.7942322681215896e-05,
646
+ "loss": 1.059,
647
+ "step": 9300
648
+ },
649
+ {
650
+ "epoch": 0.69,
651
+ "learning_rate": 5.770849571317225e-05,
652
+ "loss": 1.0443,
653
+ "step": 9400
654
+ },
655
+ {
656
+ "epoch": 0.7,
657
+ "learning_rate": 5.74746687451286e-05,
658
+ "loss": 1.0586,
659
+ "step": 9500
660
+ },
661
+ {
662
+ "epoch": 0.7,
663
+ "learning_rate": 5.7240841777084954e-05,
664
+ "loss": 1.0584,
665
+ "step": 9600
666
+ },
667
+ {
668
+ "epoch": 0.71,
669
+ "learning_rate": 5.70070148090413e-05,
670
+ "loss": 1.0414,
671
+ "step": 9700
672
+ },
673
+ {
674
+ "epoch": 0.72,
675
+ "learning_rate": 5.677318784099765e-05,
676
+ "loss": 1.0428,
677
+ "step": 9800
678
+ },
679
+ {
680
+ "epoch": 0.73,
681
+ "learning_rate": 5.653936087295401e-05,
682
+ "loss": 1.0408,
683
+ "step": 9900
684
+ },
685
+ {
686
+ "epoch": 0.73,
687
+ "learning_rate": 5.6305533904910365e-05,
688
+ "loss": 1.0347,
689
+ "step": 10000
690
+ },
691
+ {
692
+ "epoch": 0.73,
693
+ "eval_loss": 0.22658555209636688,
694
+ "eval_runtime": 1019.9931,
695
+ "eval_samples_per_second": 15.681,
696
+ "eval_steps_per_second": 1.961,
697
+ "eval_wer": 0.24018329076580575,
698
+ "step": 10000
699
+ },
700
+ {
701
+ "epoch": 0.74,
702
+ "learning_rate": 5.607170693686672e-05,
703
+ "loss": 1.0356,
704
+ "step": 10100
705
+ },
706
+ {
707
+ "epoch": 0.75,
708
+ "learning_rate": 5.583787996882306e-05,
709
+ "loss": 1.0553,
710
+ "step": 10200
711
+ },
712
+ {
713
+ "epoch": 0.76,
714
+ "learning_rate": 5.5604053000779416e-05,
715
+ "loss": 1.0464,
716
+ "step": 10300
717
+ },
718
+ {
719
+ "epoch": 0.76,
720
+ "learning_rate": 5.537022603273577e-05,
721
+ "loss": 1.0545,
722
+ "step": 10400
723
+ },
724
+ {
725
+ "epoch": 0.77,
726
+ "learning_rate": 5.513639906469212e-05,
727
+ "loss": 1.0457,
728
+ "step": 10500
729
+ },
730
+ {
731
+ "epoch": 0.78,
732
+ "learning_rate": 5.490257209664848e-05,
733
+ "loss": 1.0446,
734
+ "step": 10600
735
+ },
736
+ {
737
+ "epoch": 0.79,
738
+ "learning_rate": 5.4668745128604826e-05,
739
+ "loss": 1.051,
740
+ "step": 10700
741
+ },
742
+ {
743
+ "epoch": 0.79,
744
+ "learning_rate": 5.443491816056118e-05,
745
+ "loss": 1.0533,
746
+ "step": 10800
747
+ },
748
+ {
749
+ "epoch": 0.8,
750
+ "learning_rate": 5.420109119251753e-05,
751
+ "loss": 1.0414,
752
+ "step": 10900
753
+ },
754
+ {
755
+ "epoch": 0.81,
756
+ "learning_rate": 5.3967264224473884e-05,
757
+ "loss": 1.0457,
758
+ "step": 11000
759
+ },
760
+ {
761
+ "epoch": 0.81,
762
+ "eval_loss": 0.22899799048900604,
763
+ "eval_runtime": 1038.9889,
764
+ "eval_samples_per_second": 15.395,
765
+ "eval_steps_per_second": 1.925,
766
+ "eval_wer": 0.24484067907726348,
767
+ "step": 11000
768
+ },
769
+ {
770
+ "epoch": 0.81,
771
+ "learning_rate": 5.373343725643024e-05,
772
+ "loss": 1.0395,
773
+ "step": 11100
774
+ },
775
+ {
776
+ "epoch": 0.82,
777
+ "learning_rate": 5.349961028838658e-05,
778
+ "loss": 1.0475,
779
+ "step": 11200
780
+ },
781
+ {
782
+ "epoch": 0.83,
783
+ "learning_rate": 5.326578332034294e-05,
784
+ "loss": 1.0339,
785
+ "step": 11300
786
+ },
787
+ {
788
+ "epoch": 0.84,
789
+ "learning_rate": 5.3031956352299295e-05,
790
+ "loss": 1.0444,
791
+ "step": 11400
792
+ },
793
+ {
794
+ "epoch": 0.84,
795
+ "learning_rate": 5.279812938425565e-05,
796
+ "loss": 1.0359,
797
+ "step": 11500
798
+ },
799
+ {
800
+ "epoch": 0.85,
801
+ "learning_rate": 5.256664068589244e-05,
802
+ "loss": 1.0287,
803
+ "step": 11600
804
+ },
805
+ {
806
+ "epoch": 0.86,
807
+ "learning_rate": 5.2332813717848785e-05,
808
+ "loss": 1.0094,
809
+ "step": 11700
810
+ },
811
+ {
812
+ "epoch": 0.87,
813
+ "learning_rate": 5.209898674980514e-05,
814
+ "loss": 1.0234,
815
+ "step": 11800
816
+ },
817
+ {
818
+ "epoch": 0.87,
819
+ "learning_rate": 5.186515978176149e-05,
820
+ "loss": 1.0261,
821
+ "step": 11900
822
+ },
823
+ {
824
+ "epoch": 0.88,
825
+ "learning_rate": 5.163133281371784e-05,
826
+ "loss": 1.0124,
827
+ "step": 12000
828
+ },
829
+ {
830
+ "epoch": 0.88,
831
+ "eval_loss": 0.22948846220970154,
832
+ "eval_runtime": 1027.4606,
833
+ "eval_samples_per_second": 15.568,
834
+ "eval_steps_per_second": 1.947,
835
+ "eval_wer": 0.2447928759714274,
836
+ "step": 12000
837
+ },
838
+ {
839
+ "epoch": 0.89,
840
+ "learning_rate": 5.13975058456742e-05,
841
+ "loss": 1.0341,
842
+ "step": 12100
843
+ },
844
+ {
845
+ "epoch": 0.9,
846
+ "learning_rate": 5.116367887763055e-05,
847
+ "loss": 1.0335,
848
+ "step": 12200
849
+ },
850
+ {
851
+ "epoch": 0.9,
852
+ "learning_rate": 5.09298519095869e-05,
853
+ "loss": 1.0088,
854
+ "step": 12300
855
+ },
856
+ {
857
+ "epoch": 0.91,
858
+ "learning_rate": 5.069602494154325e-05,
859
+ "loss": 1.0151,
860
+ "step": 12400
861
+ },
862
+ {
863
+ "epoch": 0.92,
864
+ "learning_rate": 5.0462197973499606e-05,
865
+ "loss": 1.0037,
866
+ "step": 12500
867
+ },
868
+ {
869
+ "epoch": 0.92,
870
+ "learning_rate": 5.023070927513639e-05,
871
+ "loss": 1.0189,
872
+ "step": 12600
873
+ },
874
+ {
875
+ "epoch": 0.93,
876
+ "learning_rate": 4.9996882307092743e-05,
877
+ "loss": 0.9913,
878
+ "step": 12700
879
+ },
880
+ {
881
+ "epoch": 0.94,
882
+ "learning_rate": 4.9763055339049096e-05,
883
+ "loss": 0.9999,
884
+ "step": 12800
885
+ },
886
+ {
887
+ "epoch": 0.95,
888
+ "learning_rate": 4.9529228371005455e-05,
889
+ "loss": 1.0152,
890
+ "step": 12900
891
+ },
892
+ {
893
+ "epoch": 0.95,
894
+ "learning_rate": 4.929540140296181e-05,
895
+ "loss": 1.025,
896
+ "step": 13000
897
+ },
898
+ {
899
+ "epoch": 0.95,
900
+ "eval_loss": 0.21379277110099792,
901
+ "eval_runtime": 1021.0623,
902
+ "eval_samples_per_second": 15.665,
903
+ "eval_steps_per_second": 1.959,
904
+ "eval_wer": 0.2345493532922682,
905
+ "step": 13000
906
+ },
907
+ {
908
+ "epoch": 0.96,
909
+ "learning_rate": 4.9061574434918154e-05,
910
+ "loss": 0.9976,
911
+ "step": 13100
912
+ },
913
+ {
914
+ "epoch": 0.97,
915
+ "learning_rate": 4.8827747466874507e-05,
916
+ "loss": 1.0144,
917
+ "step": 13200
918
+ },
919
+ {
920
+ "epoch": 0.98,
921
+ "learning_rate": 4.859392049883086e-05,
922
+ "loss": 1.0086,
923
+ "step": 13300
924
+ },
925
+ {
926
+ "epoch": 0.98,
927
+ "learning_rate": 4.836009353078721e-05,
928
+ "loss": 0.9987,
929
+ "step": 13400
930
+ },
931
+ {
932
+ "epoch": 0.99,
933
+ "learning_rate": 4.812626656274357e-05,
934
+ "loss": 1.0206,
935
+ "step": 13500
936
+ },
937
+ {
938
+ "epoch": 1.0,
939
+ "learning_rate": 4.7892439594699924e-05,
940
+ "loss": 1.01,
941
+ "step": 13600
942
+ },
943
+ {
944
+ "epoch": 1.01,
945
+ "learning_rate": 4.765861262665627e-05,
946
+ "loss": 0.9801,
947
+ "step": 13700
948
+ },
949
+ {
950
+ "epoch": 1.01,
951
+ "learning_rate": 4.742478565861262e-05,
952
+ "loss": 1.0058,
953
+ "step": 13800
954
+ },
955
+ {
956
+ "epoch": 1.02,
957
+ "learning_rate": 4.7190958690568975e-05,
958
+ "loss": 0.999,
959
+ "step": 13900
960
+ },
961
+ {
962
+ "epoch": 1.03,
963
+ "learning_rate": 4.695713172252533e-05,
964
+ "loss": 1.0107,
965
+ "step": 14000
966
+ },
967
+ {
968
+ "epoch": 1.03,
969
+ "eval_loss": 0.21082927286624908,
970
+ "eval_runtime": 1032.6325,
971
+ "eval_samples_per_second": 15.49,
972
+ "eval_steps_per_second": 1.937,
973
+ "eval_wer": 0.22944124998292748,
974
+ "step": 14000
975
+ },
976
+ {
977
+ "epoch": 1.03,
978
+ "learning_rate": 4.672330475448168e-05,
979
+ "loss": 0.9762,
980
+ "step": 14100
981
+ },
982
+ {
983
+ "epoch": 1.04,
984
+ "learning_rate": 4.6489477786438026e-05,
985
+ "loss": 1.0014,
986
+ "step": 14200
987
+ },
988
+ {
989
+ "epoch": 1.05,
990
+ "learning_rate": 4.6255650818394385e-05,
991
+ "loss": 1.0038,
992
+ "step": 14300
993
+ },
994
+ {
995
+ "epoch": 1.06,
996
+ "learning_rate": 4.602182385035074e-05,
997
+ "loss": 0.9838,
998
+ "step": 14400
999
+ },
1000
+ {
1001
+ "epoch": 1.06,
1002
+ "learning_rate": 4.578799688230709e-05,
1003
+ "loss": 0.9931,
1004
+ "step": 14500
1005
+ },
1006
+ {
1007
+ "epoch": 1.07,
1008
+ "learning_rate": 4.555416991426344e-05,
1009
+ "loss": 0.9834,
1010
+ "step": 14600
1011
+ },
1012
+ {
1013
+ "epoch": 1.08,
1014
+ "learning_rate": 4.532034294621979e-05,
1015
+ "loss": 0.9867,
1016
+ "step": 14700
1017
+ },
1018
+ {
1019
+ "epoch": 1.09,
1020
+ "learning_rate": 4.508651597817614e-05,
1021
+ "loss": 1.0056,
1022
+ "step": 14800
1023
+ },
1024
+ {
1025
+ "epoch": 1.09,
1026
+ "learning_rate": 4.4852689010132494e-05,
1027
+ "loss": 0.9834,
1028
+ "step": 14900
1029
+ },
1030
+ {
1031
+ "epoch": 1.1,
1032
+ "learning_rate": 4.4618862042088854e-05,
1033
+ "loss": 0.9758,
1034
+ "step": 15000
1035
+ },
1036
+ {
1037
+ "epoch": 1.1,
1038
+ "eval_loss": 0.20192867517471313,
1039
+ "eval_runtime": 1027.9664,
1040
+ "eval_samples_per_second": 15.56,
1041
+ "eval_steps_per_second": 1.946,
1042
+ "eval_wer": 0.2203996339647896,
1043
+ "step": 15000
1044
+ },
1045
+ {
1046
+ "epoch": 1.11,
1047
+ "learning_rate": 4.4385035074045206e-05,
1048
+ "loss": 0.9737,
1049
+ "step": 15100
1050
+ },
1051
+ {
1052
+ "epoch": 1.12,
1053
+ "learning_rate": 4.415120810600155e-05,
1054
+ "loss": 0.9755,
1055
+ "step": 15200
1056
+ },
1057
+ {
1058
+ "epoch": 1.12,
1059
+ "learning_rate": 4.3917381137957905e-05,
1060
+ "loss": 0.9793,
1061
+ "step": 15300
1062
+ },
1063
+ {
1064
+ "epoch": 1.13,
1065
+ "learning_rate": 4.368355416991426e-05,
1066
+ "loss": 0.9818,
1067
+ "step": 15400
1068
+ },
1069
+ {
1070
+ "epoch": 1.14,
1071
+ "learning_rate": 4.344972720187061e-05,
1072
+ "loss": 0.9867,
1073
+ "step": 15500
1074
+ },
1075
+ {
1076
+ "epoch": 1.14,
1077
+ "learning_rate": 4.321590023382697e-05,
1078
+ "loss": 0.9802,
1079
+ "step": 15600
1080
+ },
1081
+ {
1082
+ "epoch": 1.15,
1083
+ "learning_rate": 4.298207326578332e-05,
1084
+ "loss": 0.9823,
1085
+ "step": 15700
1086
+ },
1087
+ {
1088
+ "epoch": 1.16,
1089
+ "learning_rate": 4.274824629773967e-05,
1090
+ "loss": 0.9669,
1091
+ "step": 15800
1092
+ },
1093
+ {
1094
+ "epoch": 1.17,
1095
+ "learning_rate": 4.251441932969602e-05,
1096
+ "loss": 0.9626,
1097
+ "step": 15900
1098
+ },
1099
+ {
1100
+ "epoch": 1.17,
1101
+ "learning_rate": 4.228059236165237e-05,
1102
+ "loss": 0.9547,
1103
+ "step": 16000
1104
+ },
1105
+ {
1106
+ "epoch": 1.17,
1107
+ "eval_loss": 0.19999034702777863,
1108
+ "eval_runtime": 1020.7566,
1109
+ "eval_samples_per_second": 15.67,
1110
+ "eval_steps_per_second": 1.959,
1111
+ "eval_wer": 0.2178250952647609,
1112
+ "step": 16000
1113
+ },
1114
+ {
1115
+ "epoch": 1.18,
1116
+ "learning_rate": 4.2046765393608726e-05,
1117
+ "loss": 0.9711,
1118
+ "step": 16100
1119
+ },
1120
+ {
1121
+ "epoch": 1.19,
1122
+ "learning_rate": 4.181293842556508e-05,
1123
+ "loss": 0.9871,
1124
+ "step": 16200
1125
+ },
1126
+ {
1127
+ "epoch": 1.2,
1128
+ "learning_rate": 4.1579111457521424e-05,
1129
+ "loss": 0.9638,
1130
+ "step": 16300
1131
+ },
1132
+ {
1133
+ "epoch": 1.2,
1134
+ "learning_rate": 4.1345284489477784e-05,
1135
+ "loss": 0.9632,
1136
+ "step": 16400
1137
+ },
1138
+ {
1139
+ "epoch": 1.21,
1140
+ "learning_rate": 4.1111457521434136e-05,
1141
+ "loss": 0.9732,
1142
+ "step": 16500
1143
+ },
1144
+ {
1145
+ "epoch": 1.22,
1146
+ "learning_rate": 4.087763055339049e-05,
1147
+ "loss": 0.9506,
1148
+ "step": 16600
1149
+ },
1150
+ {
1151
+ "epoch": 1.23,
1152
+ "learning_rate": 4.064380358534684e-05,
1153
+ "loss": 0.9718,
1154
+ "step": 16700
1155
+ },
1156
+ {
1157
+ "epoch": 1.23,
1158
+ "learning_rate": 4.040997661730319e-05,
1159
+ "loss": 0.9656,
1160
+ "step": 16800
1161
+ },
1162
+ {
1163
+ "epoch": 1.24,
1164
+ "learning_rate": 4.017614964925954e-05,
1165
+ "loss": 0.9722,
1166
+ "step": 16900
1167
+ },
1168
+ {
1169
+ "epoch": 1.25,
1170
+ "learning_rate": 3.99423226812159e-05,
1171
+ "loss": 0.986,
1172
+ "step": 17000
1173
+ },
1174
+ {
1175
+ "epoch": 1.25,
1176
+ "eval_loss": 0.20177510380744934,
1177
+ "eval_runtime": 1018.9329,
1178
+ "eval_samples_per_second": 15.698,
1179
+ "eval_steps_per_second": 1.963,
1180
+ "eval_wer": 0.21997623502738434,
1181
+ "step": 17000
1182
+ },
1183
+ {
1184
+ "epoch": 1.25,
1185
+ "learning_rate": 3.970849571317225e-05,
1186
+ "loss": 0.9662,
1187
+ "step": 17100
1188
+ },
1189
+ {
1190
+ "epoch": 1.26,
1191
+ "learning_rate": 3.9474668745128605e-05,
1192
+ "loss": 0.969,
1193
+ "step": 17200
1194
+ },
1195
+ {
1196
+ "epoch": 1.27,
1197
+ "learning_rate": 3.924084177708495e-05,
1198
+ "loss": 0.9641,
1199
+ "step": 17300
1200
+ },
1201
+ {
1202
+ "epoch": 1.28,
1203
+ "learning_rate": 3.90070148090413e-05,
1204
+ "loss": 0.9647,
1205
+ "step": 17400
1206
+ },
1207
+ {
1208
+ "epoch": 1.28,
1209
+ "learning_rate": 3.8773187840997656e-05,
1210
+ "loss": 0.9658,
1211
+ "step": 17500
1212
+ },
1213
+ {
1214
+ "epoch": 1.29,
1215
+ "learning_rate": 3.853936087295401e-05,
1216
+ "loss": 0.9615,
1217
+ "step": 17600
1218
+ },
1219
+ {
1220
+ "epoch": 1.3,
1221
+ "learning_rate": 3.830553390491037e-05,
1222
+ "loss": 0.9639,
1223
+ "step": 17700
1224
+ },
1225
+ {
1226
+ "epoch": 1.31,
1227
+ "learning_rate": 3.807170693686672e-05,
1228
+ "loss": 0.9543,
1229
+ "step": 17800
1230
+ },
1231
+ {
1232
+ "epoch": 1.31,
1233
+ "learning_rate": 3.7837879968823066e-05,
1234
+ "loss": 0.9446,
1235
+ "step": 17900
1236
+ },
1237
+ {
1238
+ "epoch": 1.32,
1239
+ "learning_rate": 3.760405300077942e-05,
1240
+ "loss": 0.9588,
1241
+ "step": 18000
1242
+ },
1243
+ {
1244
+ "epoch": 1.32,
1245
+ "eval_loss": 0.1992081105709076,
1246
+ "eval_runtime": 1010.377,
1247
+ "eval_samples_per_second": 15.831,
1248
+ "eval_steps_per_second": 1.979,
1249
+ "eval_wer": 0.21378914732917217,
1250
+ "step": 18000
1251
+ },
1252
+ {
1253
+ "epoch": 1.33,
1254
+ "learning_rate": 3.737256430241621e-05,
1255
+ "loss": 0.9683,
1256
+ "step": 18100
1257
+ },
1258
+ {
1259
+ "epoch": 1.34,
1260
+ "learning_rate": 3.713873733437256e-05,
1261
+ "loss": 0.958,
1262
+ "step": 18200
1263
+ },
1264
+ {
1265
+ "epoch": 1.34,
1266
+ "learning_rate": 3.6904910366328916e-05,
1267
+ "loss": 0.961,
1268
+ "step": 18300
1269
+ },
1270
+ {
1271
+ "epoch": 1.35,
1272
+ "learning_rate": 3.667108339828526e-05,
1273
+ "loss": 0.9441,
1274
+ "step": 18400
1275
+ },
1276
+ {
1277
+ "epoch": 1.36,
1278
+ "learning_rate": 3.643725643024162e-05,
1279
+ "loss": 0.9401,
1280
+ "step": 18500
1281
+ },
1282
+ {
1283
+ "epoch": 1.36,
1284
+ "learning_rate": 3.6203429462197974e-05,
1285
+ "loss": 0.939,
1286
+ "step": 18600
1287
+ },
1288
+ {
1289
+ "epoch": 1.37,
1290
+ "learning_rate": 3.596960249415432e-05,
1291
+ "loss": 0.9637,
1292
+ "step": 18700
1293
+ },
1294
+ {
1295
+ "epoch": 1.38,
1296
+ "learning_rate": 3.573577552611068e-05,
1297
+ "loss": 0.9412,
1298
+ "step": 18800
1299
+ },
1300
+ {
1301
+ "epoch": 1.39,
1302
+ "learning_rate": 3.5501948558067025e-05,
1303
+ "loss": 0.9399,
1304
+ "step": 18900
1305
+ },
1306
+ {
1307
+ "epoch": 1.39,
1308
+ "learning_rate": 3.526812159002338e-05,
1309
+ "loss": 0.9413,
1310
+ "step": 19000
1311
+ },
1312
+ {
1313
+ "epoch": 1.39,
1314
+ "eval_loss": 0.18979620933532715,
1315
+ "eval_runtime": 1016.8182,
1316
+ "eval_samples_per_second": 15.73,
1317
+ "eval_steps_per_second": 1.967,
1318
+ "eval_wer": 0.20486362456806478,
1319
+ "step": 19000
1320
+ },
1321
+ {
1322
+ "epoch": 1.4,
1323
+ "learning_rate": 3.503429462197973e-05,
1324
+ "loss": 0.9497,
1325
+ "step": 19100
1326
+ },
1327
+ {
1328
+ "epoch": 1.41,
1329
+ "learning_rate": 3.480280592361652e-05,
1330
+ "loss": 0.9462,
1331
+ "step": 19200
1332
+ },
1333
+ {
1334
+ "epoch": 1.42,
1335
+ "learning_rate": 3.4568978955572874e-05,
1336
+ "loss": 0.9427,
1337
+ "step": 19300
1338
+ },
1339
+ {
1340
+ "epoch": 1.42,
1341
+ "learning_rate": 3.433515198752923e-05,
1342
+ "loss": 0.9466,
1343
+ "step": 19400
1344
+ },
1345
+ {
1346
+ "epoch": 1.43,
1347
+ "learning_rate": 3.410132501948558e-05,
1348
+ "loss": 0.9284,
1349
+ "step": 19500
1350
+ },
1351
+ {
1352
+ "epoch": 1.44,
1353
+ "learning_rate": 3.3867498051441925e-05,
1354
+ "loss": 0.9438,
1355
+ "step": 19600
1356
+ },
1357
+ {
1358
+ "epoch": 1.45,
1359
+ "learning_rate": 3.3633671083398285e-05,
1360
+ "loss": 0.9281,
1361
+ "step": 19700
1362
+ },
1363
+ {
1364
+ "epoch": 1.45,
1365
+ "learning_rate": 3.339984411535464e-05,
1366
+ "loss": 0.9291,
1367
+ "step": 19800
1368
+ },
1369
+ {
1370
+ "epoch": 1.46,
1371
+ "learning_rate": 3.316601714731098e-05,
1372
+ "loss": 0.9467,
1373
+ "step": 19900
1374
+ },
1375
+ {
1376
+ "epoch": 1.47,
1377
+ "learning_rate": 3.293219017926734e-05,
1378
+ "loss": 0.9339,
1379
+ "step": 20000
1380
+ },
1381
+ {
1382
+ "epoch": 1.47,
1383
+ "eval_loss": 0.1874116212129593,
1384
+ "eval_runtime": 1013.6201,
1385
+ "eval_samples_per_second": 15.78,
1386
+ "eval_steps_per_second": 1.973,
1387
+ "eval_wer": 0.2056421322916809,
1388
+ "step": 20000
1389
+ },
1390
+ {
1391
+ "epoch": 1.47,
1392
+ "learning_rate": 3.269836321122369e-05,
1393
+ "loss": 0.9356,
1394
+ "step": 20100
1395
+ },
1396
+ {
1397
+ "epoch": 1.48,
1398
+ "learning_rate": 3.246453624318004e-05,
1399
+ "loss": 0.9206,
1400
+ "step": 20200
1401
+ },
1402
+ {
1403
+ "epoch": 1.49,
1404
+ "learning_rate": 3.22307092751364e-05,
1405
+ "loss": 0.9282,
1406
+ "step": 20300
1407
+ },
1408
+ {
1409
+ "epoch": 1.5,
1410
+ "learning_rate": 3.1996882307092746e-05,
1411
+ "loss": 0.927,
1412
+ "step": 20400
1413
+ },
1414
+ {
1415
+ "epoch": 1.5,
1416
+ "learning_rate": 3.17630553390491e-05,
1417
+ "loss": 0.9318,
1418
+ "step": 20500
1419
+ },
1420
+ {
1421
+ "epoch": 1.51,
1422
+ "learning_rate": 3.152922837100545e-05,
1423
+ "loss": 0.9394,
1424
+ "step": 20600
1425
+ },
1426
+ {
1427
+ "epoch": 1.52,
1428
+ "learning_rate": 3.1295401402961804e-05,
1429
+ "loss": 0.9227,
1430
+ "step": 20700
1431
+ },
1432
+ {
1433
+ "epoch": 1.53,
1434
+ "learning_rate": 3.106157443491816e-05,
1435
+ "loss": 0.9197,
1436
+ "step": 20800
1437
+ },
1438
+ {
1439
+ "epoch": 1.53,
1440
+ "learning_rate": 3.082774746687451e-05,
1441
+ "loss": 0.9176,
1442
+ "step": 20900
1443
+ },
1444
+ {
1445
+ "epoch": 1.54,
1446
+ "learning_rate": 3.059392049883086e-05,
1447
+ "loss": 0.9268,
1448
+ "step": 21000
1449
+ },
1450
+ {
1451
+ "epoch": 1.54,
1452
+ "eval_loss": 0.1797132045030594,
1453
+ "eval_runtime": 1006.5438,
1454
+ "eval_samples_per_second": 15.891,
1455
+ "eval_steps_per_second": 1.987,
1456
+ "eval_wer": 0.19759072346586176,
1457
+ "step": 21000
1458
+ },
1459
+ {
1460
+ "epoch": 1.55,
1461
+ "learning_rate": 3.0360093530787215e-05,
1462
+ "loss": 0.93,
1463
+ "step": 21100
1464
+ },
1465
+ {
1466
+ "epoch": 1.56,
1467
+ "learning_rate": 3.0126266562743567e-05,
1468
+ "loss": 0.9069,
1469
+ "step": 21200
1470
+ },
1471
+ {
1472
+ "epoch": 1.56,
1473
+ "learning_rate": 2.989243959469992e-05,
1474
+ "loss": 0.9359,
1475
+ "step": 21300
1476
+ },
1477
+ {
1478
+ "epoch": 1.57,
1479
+ "learning_rate": 2.966095089633671e-05,
1480
+ "loss": 0.9289,
1481
+ "step": 21400
1482
+ },
1483
+ {
1484
+ "epoch": 1.58,
1485
+ "learning_rate": 2.942712392829306e-05,
1486
+ "loss": 0.9092,
1487
+ "step": 21500
1488
+ },
1489
+ {
1490
+ "epoch": 1.58,
1491
+ "learning_rate": 2.919329696024941e-05,
1492
+ "loss": 0.9282,
1493
+ "step": 21600
1494
+ },
1495
+ {
1496
+ "epoch": 1.59,
1497
+ "learning_rate": 2.8959469992205766e-05,
1498
+ "loss": 0.9108,
1499
+ "step": 21700
1500
+ },
1501
+ {
1502
+ "epoch": 1.6,
1503
+ "learning_rate": 2.872564302416212e-05,
1504
+ "loss": 0.9193,
1505
+ "step": 21800
1506
+ },
1507
+ {
1508
+ "epoch": 1.61,
1509
+ "learning_rate": 2.8491816056118468e-05,
1510
+ "loss": 0.898,
1511
+ "step": 21900
1512
+ },
1513
+ {
1514
+ "epoch": 1.61,
1515
+ "learning_rate": 2.8257989088074824e-05,
1516
+ "loss": 0.9194,
1517
+ "step": 22000
1518
+ },
1519
+ {
1520
+ "epoch": 1.61,
1521
+ "eval_loss": 0.17433622479438782,
1522
+ "eval_runtime": 1020.4225,
1523
+ "eval_samples_per_second": 15.675,
1524
+ "eval_steps_per_second": 1.96,
1525
+ "eval_wer": 0.19052952183236133,
1526
+ "step": 22000
1527
+ },
1528
+ {
1529
+ "epoch": 1.62,
1530
+ "learning_rate": 2.8024162120031173e-05,
1531
+ "loss": 0.9215,
1532
+ "step": 22100
1533
+ },
1534
+ {
1535
+ "epoch": 1.63,
1536
+ "learning_rate": 2.7790335151987526e-05,
1537
+ "loss": 0.915,
1538
+ "step": 22200
1539
+ },
1540
+ {
1541
+ "epoch": 1.64,
1542
+ "learning_rate": 2.7556508183943882e-05,
1543
+ "loss": 0.901,
1544
+ "step": 22300
1545
+ },
1546
+ {
1547
+ "epoch": 1.64,
1548
+ "learning_rate": 2.732268121590023e-05,
1549
+ "loss": 0.913,
1550
+ "step": 22400
1551
+ },
1552
+ {
1553
+ "epoch": 1.65,
1554
+ "learning_rate": 2.7088854247856584e-05,
1555
+ "loss": 0.9155,
1556
+ "step": 22500
1557
+ },
1558
+ {
1559
+ "epoch": 1.66,
1560
+ "learning_rate": 2.6855027279812936e-05,
1561
+ "loss": 0.9,
1562
+ "step": 22600
1563
+ },
1564
+ {
1565
+ "epoch": 1.67,
1566
+ "learning_rate": 2.662120031176929e-05,
1567
+ "loss": 0.8909,
1568
+ "step": 22700
1569
+ },
1570
+ {
1571
+ "epoch": 1.67,
1572
+ "learning_rate": 2.638737334372564e-05,
1573
+ "loss": 0.9014,
1574
+ "step": 22800
1575
+ },
1576
+ {
1577
+ "epoch": 1.68,
1578
+ "learning_rate": 2.615354637568199e-05,
1579
+ "loss": 0.9036,
1580
+ "step": 22900
1581
+ },
1582
+ {
1583
+ "epoch": 1.69,
1584
+ "learning_rate": 2.5919719407638347e-05,
1585
+ "loss": 0.8987,
1586
+ "step": 23000
1587
+ },
1588
+ {
1589
+ "epoch": 1.69,
1590
+ "eval_loss": 0.1737690269947052,
1591
+ "eval_runtime": 1004.9747,
1592
+ "eval_samples_per_second": 15.916,
1593
+ "eval_steps_per_second": 1.99,
1594
+ "eval_wer": 0.19324064083477882,
1595
+ "step": 23000
1596
+ },
1597
+ {
1598
+ "epoch": 1.69,
1599
+ "learning_rate": 2.56858924395947e-05,
1600
+ "loss": 0.9074,
1601
+ "step": 23100
1602
+ },
1603
+ {
1604
+ "epoch": 1.7,
1605
+ "learning_rate": 2.545206547155105e-05,
1606
+ "loss": 0.8877,
1607
+ "step": 23200
1608
+ },
1609
+ {
1610
+ "epoch": 1.71,
1611
+ "learning_rate": 2.52182385035074e-05,
1612
+ "loss": 0.8995,
1613
+ "step": 23300
1614
+ },
1615
+ {
1616
+ "epoch": 1.72,
1617
+ "learning_rate": 2.4984411535463757e-05,
1618
+ "loss": 0.9006,
1619
+ "step": 23400
1620
+ },
1621
+ {
1622
+ "epoch": 1.72,
1623
+ "learning_rate": 2.4750584567420107e-05,
1624
+ "loss": 0.9072,
1625
+ "step": 23500
1626
+ },
1627
+ {
1628
+ "epoch": 1.73,
1629
+ "learning_rate": 2.451675759937646e-05,
1630
+ "loss": 0.891,
1631
+ "step": 23600
1632
+ },
1633
+ {
1634
+ "epoch": 1.74,
1635
+ "learning_rate": 2.428293063133281e-05,
1636
+ "loss": 0.886,
1637
+ "step": 23700
1638
+ },
1639
+ {
1640
+ "epoch": 1.75,
1641
+ "learning_rate": 2.4049103663289164e-05,
1642
+ "loss": 0.9053,
1643
+ "step": 23800
1644
+ },
1645
+ {
1646
+ "epoch": 1.75,
1647
+ "learning_rate": 2.3815276695245517e-05,
1648
+ "loss": 0.9086,
1649
+ "step": 23900
1650
+ },
1651
+ {
1652
+ "epoch": 1.76,
1653
+ "learning_rate": 2.3581449727201866e-05,
1654
+ "loss": 0.8884,
1655
+ "step": 24000
1656
+ },
1657
+ {
1658
+ "epoch": 1.76,
1659
+ "eval_loss": 0.17026115953922272,
1660
+ "eval_runtime": 1004.0446,
1661
+ "eval_samples_per_second": 15.931,
1662
+ "eval_steps_per_second": 1.992,
1663
+ "eval_wer": 0.18726525260526927,
1664
+ "step": 24000
1665
+ },
1666
+ {
1667
+ "epoch": 1.77,
1668
+ "learning_rate": 2.3347622759158222e-05,
1669
+ "loss": 0.8968,
1670
+ "step": 24100
1671
+ },
1672
+ {
1673
+ "epoch": 1.78,
1674
+ "learning_rate": 2.311613406079501e-05,
1675
+ "loss": 0.8838,
1676
+ "step": 24200
1677
+ },
1678
+ {
1679
+ "epoch": 1.78,
1680
+ "learning_rate": 2.2882307092751363e-05,
1681
+ "loss": 0.8787,
1682
+ "step": 24300
1683
+ },
1684
+ {
1685
+ "epoch": 1.79,
1686
+ "learning_rate": 2.2648480124707713e-05,
1687
+ "loss": 0.8904,
1688
+ "step": 24400
1689
+ },
1690
+ {
1691
+ "epoch": 1.8,
1692
+ "learning_rate": 2.241465315666407e-05,
1693
+ "loss": 0.8854,
1694
+ "step": 24500
1695
+ },
1696
+ {
1697
+ "epoch": 1.8,
1698
+ "learning_rate": 2.218082618862042e-05,
1699
+ "loss": 0.8795,
1700
+ "step": 24600
1701
+ },
1702
+ {
1703
+ "epoch": 1.81,
1704
+ "learning_rate": 2.194699922057677e-05,
1705
+ "loss": 0.8897,
1706
+ "step": 24700
1707
+ },
1708
+ {
1709
+ "epoch": 1.82,
1710
+ "learning_rate": 2.1713172252533123e-05,
1711
+ "loss": 0.8888,
1712
+ "step": 24800
1713
+ },
1714
+ {
1715
+ "epoch": 1.83,
1716
+ "learning_rate": 2.1479345284489476e-05,
1717
+ "loss": 0.8817,
1718
+ "step": 24900
1719
+ },
1720
+ {
1721
+ "epoch": 1.83,
1722
+ "learning_rate": 2.1245518316445828e-05,
1723
+ "loss": 0.8939,
1724
+ "step": 25000
1725
+ },
1726
+ {
1727
+ "epoch": 1.83,
1728
+ "eval_loss": 0.16330334544181824,
1729
+ "eval_runtime": 1011.258,
1730
+ "eval_samples_per_second": 15.817,
1731
+ "eval_steps_per_second": 1.978,
1732
+ "eval_wer": 0.18310638239753063,
1733
+ "step": 25000
1734
+ },
1735
+ {
1736
+ "epoch": 1.84,
1737
+ "learning_rate": 2.101169134840218e-05,
1738
+ "loss": 0.8837,
1739
+ "step": 25100
1740
+ },
1741
+ {
1742
+ "epoch": 1.85,
1743
+ "learning_rate": 2.077786438035853e-05,
1744
+ "loss": 0.8832,
1745
+ "step": 25200
1746
+ },
1747
+ {
1748
+ "epoch": 1.86,
1749
+ "learning_rate": 2.0544037412314886e-05,
1750
+ "loss": 0.869,
1751
+ "step": 25300
1752
+ },
1753
+ {
1754
+ "epoch": 1.86,
1755
+ "learning_rate": 2.0310210444271235e-05,
1756
+ "loss": 0.8638,
1757
+ "step": 25400
1758
+ },
1759
+ {
1760
+ "epoch": 1.87,
1761
+ "learning_rate": 2.0076383476227588e-05,
1762
+ "loss": 0.8785,
1763
+ "step": 25500
1764
+ },
1765
+ {
1766
+ "epoch": 1.88,
1767
+ "learning_rate": 1.9842556508183944e-05,
1768
+ "loss": 0.8666,
1769
+ "step": 25600
1770
+ },
1771
+ {
1772
+ "epoch": 1.89,
1773
+ "learning_rate": 1.9608729540140293e-05,
1774
+ "loss": 0.8725,
1775
+ "step": 25700
1776
+ },
1777
+ {
1778
+ "epoch": 1.89,
1779
+ "learning_rate": 1.9374902572096646e-05,
1780
+ "loss": 0.8757,
1781
+ "step": 25800
1782
+ },
1783
+ {
1784
+ "epoch": 1.9,
1785
+ "learning_rate": 1.9141075604053002e-05,
1786
+ "loss": 0.8617,
1787
+ "step": 25900
1788
+ },
1789
+ {
1790
+ "epoch": 1.91,
1791
+ "learning_rate": 1.890724863600935e-05,
1792
+ "loss": 0.8629,
1793
+ "step": 26000
1794
+ },
1795
+ {
1796
+ "epoch": 1.91,
1797
+ "eval_loss": 0.154932901263237,
1798
+ "eval_runtime": 1005.7394,
1799
+ "eval_samples_per_second": 15.904,
1800
+ "eval_steps_per_second": 1.989,
1801
+ "eval_wer": 0.17495936736003934,
1802
+ "step": 26000
1803
+ },
1804
+ {
1805
+ "epoch": 1.91,
1806
+ "learning_rate": 1.8673421667965704e-05,
1807
+ "loss": 0.8748,
1808
+ "step": 26100
1809
+ },
1810
+ {
1811
+ "epoch": 1.92,
1812
+ "learning_rate": 1.8439594699922056e-05,
1813
+ "loss": 0.8659,
1814
+ "step": 26200
1815
+ },
1816
+ {
1817
+ "epoch": 1.93,
1818
+ "learning_rate": 1.820576773187841e-05,
1819
+ "loss": 0.8594,
1820
+ "step": 26300
1821
+ },
1822
+ {
1823
+ "epoch": 1.94,
1824
+ "learning_rate": 1.7974279033515197e-05,
1825
+ "loss": 0.8568,
1826
+ "step": 26400
1827
+ },
1828
+ {
1829
+ "epoch": 1.94,
1830
+ "learning_rate": 1.774045206547155e-05,
1831
+ "loss": 0.8694,
1832
+ "step": 26500
1833
+ },
1834
+ {
1835
+ "epoch": 1.95,
1836
+ "learning_rate": 1.7506625097427903e-05,
1837
+ "loss": 0.8652,
1838
+ "step": 26600
1839
+ },
1840
+ {
1841
+ "epoch": 1.96,
1842
+ "learning_rate": 1.7272798129384255e-05,
1843
+ "loss": 0.8565,
1844
+ "step": 26700
1845
+ },
1846
+ {
1847
+ "epoch": 1.97,
1848
+ "learning_rate": 1.7038971161340608e-05,
1849
+ "loss": 0.8608,
1850
+ "step": 26800
1851
+ },
1852
+ {
1853
+ "epoch": 1.97,
1854
+ "learning_rate": 1.680514419329696e-05,
1855
+ "loss": 0.8554,
1856
+ "step": 26900
1857
+ },
1858
+ {
1859
+ "epoch": 1.98,
1860
+ "learning_rate": 1.657131722525331e-05,
1861
+ "loss": 0.8607,
1862
+ "step": 27000
1863
+ },
1864
+ {
1865
+ "epoch": 1.98,
1866
+ "eval_loss": 0.15500280261039734,
1867
+ "eval_runtime": 1016.9779,
1868
+ "eval_samples_per_second": 15.728,
1869
+ "eval_steps_per_second": 1.967,
1870
+ "eval_wer": 0.17375746069901798,
1871
+ "step": 27000
1872
+ },
1873
+ {
1874
+ "epoch": 1.99,
1875
+ "learning_rate": 1.6339828526890098e-05,
1876
+ "loss": 0.8551,
1877
+ "step": 27100
1878
+ },
1879
+ {
1880
+ "epoch": 2.0,
1881
+ "learning_rate": 1.6106001558846454e-05,
1882
+ "loss": 0.8588,
1883
+ "step": 27200
1884
+ },
1885
+ {
1886
+ "epoch": 2.0,
1887
+ "learning_rate": 1.5872174590802803e-05,
1888
+ "loss": 0.8518,
1889
+ "step": 27300
1890
+ },
1891
+ {
1892
+ "epoch": 2.01,
1893
+ "learning_rate": 1.5638347622759156e-05,
1894
+ "loss": 0.8359,
1895
+ "step": 27400
1896
+ },
1897
+ {
1898
+ "epoch": 2.02,
1899
+ "learning_rate": 1.540452065471551e-05,
1900
+ "loss": 0.8404,
1901
+ "step": 27500
1902
+ },
1903
+ {
1904
+ "epoch": 2.02,
1905
+ "learning_rate": 1.5170693686671863e-05,
1906
+ "loss": 0.8596,
1907
+ "step": 27600
1908
+ },
1909
+ {
1910
+ "epoch": 2.03,
1911
+ "learning_rate": 1.4936866718628214e-05,
1912
+ "loss": 0.8499,
1913
+ "step": 27700
1914
+ },
1915
+ {
1916
+ "epoch": 2.04,
1917
+ "learning_rate": 1.4703039750584566e-05,
1918
+ "loss": 0.8425,
1919
+ "step": 27800
1920
+ },
1921
+ {
1922
+ "epoch": 2.05,
1923
+ "learning_rate": 1.4469212782540917e-05,
1924
+ "loss": 0.8391,
1925
+ "step": 27900
1926
+ },
1927
+ {
1928
+ "epoch": 2.05,
1929
+ "learning_rate": 1.4235385814497272e-05,
1930
+ "loss": 0.8316,
1931
+ "step": 28000
1932
+ },
1933
+ {
1934
+ "epoch": 2.05,
1935
+ "eval_loss": 0.15122003853321075,
1936
+ "eval_runtime": 1007.5325,
1937
+ "eval_samples_per_second": 15.875,
1938
+ "eval_steps_per_second": 1.985,
1939
+ "eval_wer": 0.17086878730349508,
1940
+ "step": 28000
1941
+ },
1942
+ {
1943
+ "epoch": 2.06,
1944
+ "learning_rate": 1.4001558846453624e-05,
1945
+ "loss": 0.8325,
1946
+ "step": 28100
1947
+ },
1948
+ {
1949
+ "epoch": 2.07,
1950
+ "learning_rate": 1.3767731878409975e-05,
1951
+ "loss": 0.8311,
1952
+ "step": 28200
1953
+ },
1954
+ {
1955
+ "epoch": 2.08,
1956
+ "learning_rate": 1.3533904910366328e-05,
1957
+ "loss": 0.8336,
1958
+ "step": 28300
1959
+ },
1960
+ {
1961
+ "epoch": 2.08,
1962
+ "learning_rate": 1.330007794232268e-05,
1963
+ "loss": 0.8248,
1964
+ "step": 28400
1965
+ },
1966
+ {
1967
+ "epoch": 2.09,
1968
+ "learning_rate": 1.3066250974279033e-05,
1969
+ "loss": 0.8316,
1970
+ "step": 28500
1971
+ },
1972
+ {
1973
+ "epoch": 2.1,
1974
+ "learning_rate": 1.2832424006235384e-05,
1975
+ "loss": 0.8313,
1976
+ "step": 28600
1977
+ },
1978
+ {
1979
+ "epoch": 2.11,
1980
+ "learning_rate": 1.2598597038191737e-05,
1981
+ "loss": 0.8284,
1982
+ "step": 28700
1983
+ },
1984
+ {
1985
+ "epoch": 2.11,
1986
+ "learning_rate": 1.236477007014809e-05,
1987
+ "loss": 0.8309,
1988
+ "step": 28800
1989
+ },
1990
+ {
1991
+ "epoch": 2.12,
1992
+ "learning_rate": 1.2130943102104442e-05,
1993
+ "loss": 0.8287,
1994
+ "step": 28900
1995
+ },
1996
+ {
1997
+ "epoch": 2.13,
1998
+ "learning_rate": 1.189945440374123e-05,
1999
+ "loss": 0.8321,
2000
+ "step": 29000
2001
+ },
2002
+ {
2003
+ "epoch": 2.13,
2004
+ "eval_loss": 0.14810478687286377,
2005
+ "eval_runtime": 999.3435,
2006
+ "eval_samples_per_second": 16.006,
2007
+ "eval_steps_per_second": 2.001,
2008
+ "eval_wer": 0.16567873581272108,
2009
+ "step": 29000
2010
+ },
2011
+ {
2012
+ "epoch": 2.13,
2013
+ "learning_rate": 1.1665627435697581e-05,
2014
+ "loss": 0.8264,
2015
+ "step": 29100
2016
+ },
2017
+ {
2018
+ "epoch": 2.14,
2019
+ "learning_rate": 1.1431800467653935e-05,
2020
+ "loss": 0.8186,
2021
+ "step": 29200
2022
+ },
2023
+ {
2024
+ "epoch": 2.15,
2025
+ "learning_rate": 1.1197973499610288e-05,
2026
+ "loss": 0.8264,
2027
+ "step": 29300
2028
+ },
2029
+ {
2030
+ "epoch": 2.16,
2031
+ "learning_rate": 1.0964146531566639e-05,
2032
+ "loss": 0.8285,
2033
+ "step": 29400
2034
+ },
2035
+ {
2036
+ "epoch": 2.16,
2037
+ "learning_rate": 1.0730319563522992e-05,
2038
+ "loss": 0.8321,
2039
+ "step": 29500
2040
+ },
2041
+ {
2042
+ "epoch": 2.17,
2043
+ "learning_rate": 1.0496492595479346e-05,
2044
+ "loss": 0.8261,
2045
+ "step": 29600
2046
+ },
2047
+ {
2048
+ "epoch": 2.18,
2049
+ "learning_rate": 1.0262665627435697e-05,
2050
+ "loss": 0.836,
2051
+ "step": 29700
2052
+ },
2053
+ {
2054
+ "epoch": 2.19,
2055
+ "learning_rate": 1.002883865939205e-05,
2056
+ "loss": 0.8178,
2057
+ "step": 29800
2058
+ },
2059
+ {
2060
+ "epoch": 2.19,
2061
+ "learning_rate": 9.7950116913484e-06,
2062
+ "loss": 0.8258,
2063
+ "step": 29900
2064
+ },
2065
+ {
2066
+ "epoch": 2.2,
2067
+ "learning_rate": 9.561184723304755e-06,
2068
+ "loss": 0.825,
2069
+ "step": 30000
2070
+ },
2071
+ {
2072
+ "epoch": 2.2,
2073
+ "eval_loss": 0.14461292326450348,
2074
+ "eval_runtime": 1002.0806,
2075
+ "eval_samples_per_second": 15.962,
2076
+ "eval_steps_per_second": 1.996,
2077
+ "eval_wer": 0.16271494325088437,
2078
+ "step": 30000
2079
+ },
2080
+ {
2081
+ "epoch": 2.21,
2082
+ "learning_rate": 9.327357755261106e-06,
2083
+ "loss": 0.8247,
2084
+ "step": 30100
2085
+ },
2086
+ {
2087
+ "epoch": 2.22,
2088
+ "learning_rate": 9.093530787217458e-06,
2089
+ "loss": 0.8223,
2090
+ "step": 30200
2091
+ },
2092
+ {
2093
+ "epoch": 2.22,
2094
+ "learning_rate": 8.85970381917381e-06,
2095
+ "loss": 0.8144,
2096
+ "step": 30300
2097
+ },
2098
+ {
2099
+ "epoch": 2.23,
2100
+ "learning_rate": 8.625876851130163e-06,
2101
+ "loss": 0.821,
2102
+ "step": 30400
2103
+ },
2104
+ {
2105
+ "epoch": 2.24,
2106
+ "learning_rate": 8.392049883086516e-06,
2107
+ "loss": 0.8069,
2108
+ "step": 30500
2109
+ },
2110
+ {
2111
+ "epoch": 2.25,
2112
+ "learning_rate": 8.158222915042867e-06,
2113
+ "loss": 0.8122,
2114
+ "step": 30600
2115
+ },
2116
+ {
2117
+ "epoch": 2.25,
2118
+ "learning_rate": 7.92439594699922e-06,
2119
+ "loss": 0.8042,
2120
+ "step": 30700
2121
+ },
2122
+ {
2123
+ "epoch": 2.26,
2124
+ "learning_rate": 7.690568978955572e-06,
2125
+ "loss": 0.8159,
2126
+ "step": 30800
2127
+ },
2128
+ {
2129
+ "epoch": 2.27,
2130
+ "learning_rate": 7.456742010911924e-06,
2131
+ "loss": 0.8125,
2132
+ "step": 30900
2133
+ },
2134
+ {
2135
+ "epoch": 2.27,
2136
+ "learning_rate": 7.222915042868277e-06,
2137
+ "loss": 0.8115,
2138
+ "step": 31000
2139
+ },
2140
+ {
2141
+ "epoch": 2.27,
2142
+ "eval_loss": 0.13957646489143372,
2143
+ "eval_runtime": 1001.1222,
2144
+ "eval_samples_per_second": 15.977,
2145
+ "eval_steps_per_second": 1.998,
2146
+ "eval_wer": 0.15831705751396533,
2147
+ "step": 31000
2148
+ },
2149
+ {
2150
+ "epoch": 2.28,
2151
+ "learning_rate": 6.989088074824628e-06,
2152
+ "loss": 0.7972,
2153
+ "step": 31100
2154
+ },
2155
+ {
2156
+ "epoch": 2.29,
2157
+ "learning_rate": 6.755261106780982e-06,
2158
+ "loss": 0.8031,
2159
+ "step": 31200
2160
+ },
2161
+ {
2162
+ "epoch": 2.3,
2163
+ "learning_rate": 6.521434138737334e-06,
2164
+ "loss": 0.8017,
2165
+ "step": 31300
2166
+ },
2167
+ {
2168
+ "epoch": 2.3,
2169
+ "learning_rate": 6.287607170693686e-06,
2170
+ "loss": 0.8036,
2171
+ "step": 31400
2172
+ },
2173
+ {
2174
+ "epoch": 2.31,
2175
+ "learning_rate": 6.053780202650038e-06,
2176
+ "loss": 0.8001,
2177
+ "step": 31500
2178
+ },
2179
+ {
2180
+ "epoch": 2.32,
2181
+ "learning_rate": 5.819953234606391e-06,
2182
+ "loss": 0.8133,
2183
+ "step": 31600
2184
+ },
2185
+ {
2186
+ "epoch": 2.33,
2187
+ "learning_rate": 5.586126266562743e-06,
2188
+ "loss": 0.8105,
2189
+ "step": 31700
2190
+ },
2191
+ {
2192
+ "epoch": 2.33,
2193
+ "learning_rate": 5.352299298519096e-06,
2194
+ "loss": 0.8104,
2195
+ "step": 31800
2196
+ },
2197
+ {
2198
+ "epoch": 2.34,
2199
+ "learning_rate": 5.118472330475448e-06,
2200
+ "loss": 0.7934,
2201
+ "step": 31900
2202
+ },
2203
+ {
2204
+ "epoch": 2.35,
2205
+ "learning_rate": 4.8846453624318e-06,
2206
+ "loss": 0.7959,
2207
+ "step": 32000
2208
+ },
2209
+ {
2210
+ "epoch": 2.35,
2211
+ "eval_loss": 0.1389056146144867,
2212
+ "eval_runtime": 1015.5862,
2213
+ "eval_samples_per_second": 15.75,
2214
+ "eval_steps_per_second": 1.969,
2215
+ "eval_wer": 0.15688296433888305,
2216
+ "step": 32000
2217
+ },
2218
+ {
2219
+ "epoch": 2.36,
2220
+ "learning_rate": 4.650818394388152e-06,
2221
+ "loss": 0.8036,
2222
+ "step": 32100
2223
+ },
2224
+ {
2225
+ "epoch": 2.36,
2226
+ "learning_rate": 4.416991426344505e-06,
2227
+ "loss": 0.7954,
2228
+ "step": 32200
2229
+ },
2230
+ {
2231
+ "epoch": 2.37,
2232
+ "learning_rate": 4.183164458300857e-06,
2233
+ "loss": 0.7958,
2234
+ "step": 32300
2235
+ },
2236
+ {
2237
+ "epoch": 2.38,
2238
+ "learning_rate": 3.949337490257209e-06,
2239
+ "loss": 0.7973,
2240
+ "step": 32400
2241
+ },
2242
+ {
2243
+ "epoch": 2.38,
2244
+ "learning_rate": 3.7155105222135617e-06,
2245
+ "loss": 0.7989,
2246
+ "step": 32500
2247
+ },
2248
+ {
2249
+ "epoch": 2.39,
2250
+ "learning_rate": 3.481683554169914e-06,
2251
+ "loss": 0.7959,
2252
+ "step": 32600
2253
+ },
2254
+ {
2255
+ "epoch": 2.4,
2256
+ "learning_rate": 3.2478565861262665e-06,
2257
+ "loss": 0.7945,
2258
+ "step": 32700
2259
+ },
2260
+ {
2261
+ "epoch": 2.41,
2262
+ "learning_rate": 3.0140296180826187e-06,
2263
+ "loss": 0.8044,
2264
+ "step": 32800
2265
+ },
2266
+ {
2267
+ "epoch": 2.41,
2268
+ "learning_rate": 2.780202650038971e-06,
2269
+ "loss": 0.7979,
2270
+ "step": 32900
2271
+ },
2272
+ {
2273
+ "epoch": 2.42,
2274
+ "learning_rate": 2.5463756819953235e-06,
2275
+ "loss": 0.7835,
2276
+ "step": 33000
2277
+ },
2278
+ {
2279
+ "epoch": 2.42,
2280
+ "eval_loss": 0.1362341344356537,
2281
+ "eval_runtime": 1006.8998,
2282
+ "eval_samples_per_second": 15.885,
2283
+ "eval_steps_per_second": 1.986,
2284
+ "eval_wer": 0.15451329609243755,
2285
+ "step": 33000
2286
+ },
2287
+ {
2288
+ "epoch": 2.43,
2289
+ "learning_rate": 2.3125487139516753e-06,
2290
+ "loss": 0.7871,
2291
+ "step": 33100
2292
+ },
2293
+ {
2294
+ "epoch": 2.44,
2295
+ "learning_rate": 2.078721745908028e-06,
2296
+ "loss": 0.7973,
2297
+ "step": 33200
2298
+ },
2299
+ {
2300
+ "epoch": 2.44,
2301
+ "learning_rate": 1.84489477786438e-06,
2302
+ "loss": 0.7855,
2303
+ "step": 33300
2304
+ },
2305
+ {
2306
+ "epoch": 2.45,
2307
+ "learning_rate": 1.6110678098207325e-06,
2308
+ "loss": 0.7884,
2309
+ "step": 33400
2310
+ },
2311
+ {
2312
+ "epoch": 2.46,
2313
+ "learning_rate": 1.377240841777085e-06,
2314
+ "loss": 0.7848,
2315
+ "step": 33500
2316
+ },
2317
+ {
2318
+ "epoch": 2.47,
2319
+ "learning_rate": 1.1434138737334373e-06,
2320
+ "loss": 0.7843,
2321
+ "step": 33600
2322
+ },
2323
+ {
2324
+ "epoch": 2.47,
2325
+ "learning_rate": 9.095869056897894e-07,
2326
+ "loss": 0.7996,
2327
+ "step": 33700
2328
+ },
2329
+ {
2330
+ "epoch": 2.48,
2331
+ "learning_rate": 6.757599376461418e-07,
2332
+ "loss": 0.7862,
2333
+ "step": 33800
2334
+ },
2335
+ {
2336
+ "epoch": 2.49,
2337
+ "learning_rate": 4.419329696024941e-07,
2338
+ "loss": 0.7888,
2339
+ "step": 33900
2340
+ },
2341
+ {
2342
+ "epoch": 2.49,
2343
+ "learning_rate": 2.0810600155884645e-07,
2344
+ "loss": 0.7959,
2345
+ "step": 34000
2346
+ },
2347
+ {
2348
+ "epoch": 2.49,
2349
+ "eval_loss": 0.13550546765327454,
2350
+ "eval_runtime": 1005.5212,
2351
+ "eval_samples_per_second": 15.907,
2352
+ "eval_steps_per_second": 1.989,
2353
+ "eval_wer": 0.15314749306854966,
2354
+ "step": 34000
2355
+ },
2356
+ {
2357
+ "epoch": 2.5,
2358
+ "step": 34075,
2359
+ "total_flos": 5.050048908581704e+20,
2360
+ "train_loss": 1.0034960915109994,
2361
+ "train_runtime": 150795.9745,
2362
+ "train_samples_per_second": 7.231,
2363
+ "train_steps_per_second": 0.226
2364
+ }
2365
+ ],
2366
+ "max_steps": 34075,
2367
+ "num_train_epochs": 3,
2368
+ "total_flos": 5.050048908581704e+20,
2369
+ "trial_name": null,
2370
+ "trial_params": null
2371
+ }