Zaid commited on
Commit
c03b4be
1 Parent(s): 06c684e

End of training

Browse files
all_results.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_loss": 0.8518440127372742,
4
+ "eval_runtime": 5199.1699,
5
+ "eval_samples_per_second": 1.088,
6
+ "eval_steps_per_second": 0.068,
7
+ "eval_wer": 16.3839925843315,
8
+ "train_loss": 0.2656231692790985,
9
+ "train_runtime": 120187.1267,
10
+ "train_samples_per_second": 2.663,
11
+ "train_steps_per_second": 0.083
12
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_loss": 0.8518440127372742,
4
+ "eval_runtime": 5199.1699,
5
+ "eval_samples_per_second": 1.088,
6
+ "eval_steps_per_second": 0.068,
7
+ "eval_wer": 16.3839925843315
8
+ }
runs/Dec14_09-02-25_129-146-107-47/events.out.tfevents.1671134090.129-146-107-47.118226.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d70afc55eabb1d5d42e72173804c40769632f42c6562902b27bd385777f1eef
3
+ size 358
train_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "train_loss": 0.2656231692790985,
4
+ "train_runtime": 120187.1267,
5
+ "train_samples_per_second": 2.663,
6
+ "train_steps_per_second": 0.083
7
+ }
trainer_state.json ADDED
@@ -0,0 +1,2515 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 16.3839925843315,
3
+ "best_model_checkpoint": "./checkpoint-9000",
4
+ "epoch": 1.0,
5
+ "global_step": 10000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 4.6000000000000004e-07,
13
+ "loss": 0.6709,
14
+ "step": 25
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 9.600000000000001e-07,
19
+ "loss": 0.596,
20
+ "step": 50
21
+ },
22
+ {
23
+ "epoch": 0.01,
24
+ "learning_rate": 1.46e-06,
25
+ "loss": 0.4857,
26
+ "step": 75
27
+ },
28
+ {
29
+ "epoch": 0.01,
30
+ "learning_rate": 1.9600000000000003e-06,
31
+ "loss": 0.3658,
32
+ "step": 100
33
+ },
34
+ {
35
+ "epoch": 0.01,
36
+ "learning_rate": 2.46e-06,
37
+ "loss": 0.4504,
38
+ "step": 125
39
+ },
40
+ {
41
+ "epoch": 0.01,
42
+ "learning_rate": 2.96e-06,
43
+ "loss": 0.4356,
44
+ "step": 150
45
+ },
46
+ {
47
+ "epoch": 0.02,
48
+ "learning_rate": 3.46e-06,
49
+ "loss": 0.4217,
50
+ "step": 175
51
+ },
52
+ {
53
+ "epoch": 0.02,
54
+ "learning_rate": 3.96e-06,
55
+ "loss": 0.4439,
56
+ "step": 200
57
+ },
58
+ {
59
+ "epoch": 0.02,
60
+ "learning_rate": 4.4600000000000005e-06,
61
+ "loss": 0.3948,
62
+ "step": 225
63
+ },
64
+ {
65
+ "epoch": 0.03,
66
+ "learning_rate": 4.960000000000001e-06,
67
+ "loss": 0.3626,
68
+ "step": 250
69
+ },
70
+ {
71
+ "epoch": 0.03,
72
+ "learning_rate": 5.460000000000001e-06,
73
+ "loss": 0.3838,
74
+ "step": 275
75
+ },
76
+ {
77
+ "epoch": 0.03,
78
+ "learning_rate": 5.9600000000000005e-06,
79
+ "loss": 0.4015,
80
+ "step": 300
81
+ },
82
+ {
83
+ "epoch": 0.03,
84
+ "learning_rate": 6.460000000000001e-06,
85
+ "loss": 0.3427,
86
+ "step": 325
87
+ },
88
+ {
89
+ "epoch": 0.04,
90
+ "learning_rate": 6.96e-06,
91
+ "loss": 0.3582,
92
+ "step": 350
93
+ },
94
+ {
95
+ "epoch": 0.04,
96
+ "learning_rate": 7.4600000000000006e-06,
97
+ "loss": 0.3408,
98
+ "step": 375
99
+ },
100
+ {
101
+ "epoch": 0.04,
102
+ "learning_rate": 7.960000000000002e-06,
103
+ "loss": 0.3252,
104
+ "step": 400
105
+ },
106
+ {
107
+ "epoch": 0.04,
108
+ "learning_rate": 8.46e-06,
109
+ "loss": 0.3157,
110
+ "step": 425
111
+ },
112
+ {
113
+ "epoch": 0.04,
114
+ "learning_rate": 8.96e-06,
115
+ "loss": 0.3813,
116
+ "step": 450
117
+ },
118
+ {
119
+ "epoch": 0.05,
120
+ "learning_rate": 9.460000000000001e-06,
121
+ "loss": 0.3489,
122
+ "step": 475
123
+ },
124
+ {
125
+ "epoch": 0.05,
126
+ "learning_rate": 9.960000000000001e-06,
127
+ "loss": 0.3676,
128
+ "step": 500
129
+ },
130
+ {
131
+ "epoch": 0.05,
132
+ "learning_rate": 9.975789473684211e-06,
133
+ "loss": 0.3151,
134
+ "step": 525
135
+ },
136
+ {
137
+ "epoch": 0.06,
138
+ "learning_rate": 9.949473684210526e-06,
139
+ "loss": 0.3342,
140
+ "step": 550
141
+ },
142
+ {
143
+ "epoch": 0.06,
144
+ "learning_rate": 9.923157894736844e-06,
145
+ "loss": 0.3241,
146
+ "step": 575
147
+ },
148
+ {
149
+ "epoch": 0.06,
150
+ "learning_rate": 9.89684210526316e-06,
151
+ "loss": 0.2749,
152
+ "step": 600
153
+ },
154
+ {
155
+ "epoch": 0.06,
156
+ "learning_rate": 9.870526315789474e-06,
157
+ "loss": 0.3745,
158
+ "step": 625
159
+ },
160
+ {
161
+ "epoch": 0.07,
162
+ "learning_rate": 9.84421052631579e-06,
163
+ "loss": 0.3269,
164
+ "step": 650
165
+ },
166
+ {
167
+ "epoch": 0.07,
168
+ "learning_rate": 9.817894736842106e-06,
169
+ "loss": 0.3545,
170
+ "step": 675
171
+ },
172
+ {
173
+ "epoch": 0.07,
174
+ "learning_rate": 9.791578947368422e-06,
175
+ "loss": 0.3464,
176
+ "step": 700
177
+ },
178
+ {
179
+ "epoch": 0.07,
180
+ "learning_rate": 9.765263157894737e-06,
181
+ "loss": 0.3367,
182
+ "step": 725
183
+ },
184
+ {
185
+ "epoch": 0.07,
186
+ "learning_rate": 9.738947368421054e-06,
187
+ "loss": 0.3749,
188
+ "step": 750
189
+ },
190
+ {
191
+ "epoch": 0.08,
192
+ "learning_rate": 9.712631578947369e-06,
193
+ "loss": 0.3322,
194
+ "step": 775
195
+ },
196
+ {
197
+ "epoch": 0.08,
198
+ "learning_rate": 9.686315789473684e-06,
199
+ "loss": 0.3304,
200
+ "step": 800
201
+ },
202
+ {
203
+ "epoch": 0.08,
204
+ "learning_rate": 9.66e-06,
205
+ "loss": 0.3169,
206
+ "step": 825
207
+ },
208
+ {
209
+ "epoch": 0.09,
210
+ "learning_rate": 9.633684210526316e-06,
211
+ "loss": 0.3407,
212
+ "step": 850
213
+ },
214
+ {
215
+ "epoch": 0.09,
216
+ "learning_rate": 9.607368421052632e-06,
217
+ "loss": 0.363,
218
+ "step": 875
219
+ },
220
+ {
221
+ "epoch": 0.09,
222
+ "learning_rate": 9.581052631578947e-06,
223
+ "loss": 0.329,
224
+ "step": 900
225
+ },
226
+ {
227
+ "epoch": 0.09,
228
+ "learning_rate": 9.554736842105264e-06,
229
+ "loss": 0.3262,
230
+ "step": 925
231
+ },
232
+ {
233
+ "epoch": 0.1,
234
+ "learning_rate": 9.52842105263158e-06,
235
+ "loss": 0.3153,
236
+ "step": 950
237
+ },
238
+ {
239
+ "epoch": 0.1,
240
+ "learning_rate": 9.502105263157896e-06,
241
+ "loss": 0.3095,
242
+ "step": 975
243
+ },
244
+ {
245
+ "epoch": 0.1,
246
+ "learning_rate": 9.475789473684212e-06,
247
+ "loss": 0.2963,
248
+ "step": 1000
249
+ },
250
+ {
251
+ "epoch": 0.1,
252
+ "eval_loss": 0.9114714860916138,
253
+ "eval_runtime": 6612.5212,
254
+ "eval_samples_per_second": 0.855,
255
+ "eval_steps_per_second": 0.054,
256
+ "eval_wer": 27.364106426429906,
257
+ "step": 1000
258
+ },
259
+ {
260
+ "epoch": 0.1,
261
+ "learning_rate": 9.449473684210527e-06,
262
+ "loss": 0.3157,
263
+ "step": 1025
264
+ },
265
+ {
266
+ "epoch": 0.1,
267
+ "learning_rate": 9.423157894736842e-06,
268
+ "loss": 0.3128,
269
+ "step": 1050
270
+ },
271
+ {
272
+ "epoch": 0.11,
273
+ "learning_rate": 9.396842105263159e-06,
274
+ "loss": 0.3081,
275
+ "step": 1075
276
+ },
277
+ {
278
+ "epoch": 0.11,
279
+ "learning_rate": 9.370526315789474e-06,
280
+ "loss": 0.3011,
281
+ "step": 1100
282
+ },
283
+ {
284
+ "epoch": 0.11,
285
+ "learning_rate": 9.34421052631579e-06,
286
+ "loss": 0.3038,
287
+ "step": 1125
288
+ },
289
+ {
290
+ "epoch": 0.12,
291
+ "learning_rate": 9.317894736842105e-06,
292
+ "loss": 0.3061,
293
+ "step": 1150
294
+ },
295
+ {
296
+ "epoch": 0.12,
297
+ "learning_rate": 9.291578947368422e-06,
298
+ "loss": 0.3351,
299
+ "step": 1175
300
+ },
301
+ {
302
+ "epoch": 0.12,
303
+ "learning_rate": 9.265263157894737e-06,
304
+ "loss": 0.3328,
305
+ "step": 1200
306
+ },
307
+ {
308
+ "epoch": 0.12,
309
+ "learning_rate": 9.238947368421052e-06,
310
+ "loss": 0.3272,
311
+ "step": 1225
312
+ },
313
+ {
314
+ "epoch": 0.12,
315
+ "learning_rate": 9.21263157894737e-06,
316
+ "loss": 0.369,
317
+ "step": 1250
318
+ },
319
+ {
320
+ "epoch": 0.13,
321
+ "learning_rate": 9.186315789473685e-06,
322
+ "loss": 0.3143,
323
+ "step": 1275
324
+ },
325
+ {
326
+ "epoch": 0.13,
327
+ "learning_rate": 9.16e-06,
328
+ "loss": 0.2937,
329
+ "step": 1300
330
+ },
331
+ {
332
+ "epoch": 0.13,
333
+ "learning_rate": 9.133684210526317e-06,
334
+ "loss": 0.3536,
335
+ "step": 1325
336
+ },
337
+ {
338
+ "epoch": 0.14,
339
+ "learning_rate": 9.107368421052632e-06,
340
+ "loss": 0.3534,
341
+ "step": 1350
342
+ },
343
+ {
344
+ "epoch": 0.14,
345
+ "learning_rate": 9.081052631578949e-06,
346
+ "loss": 0.3063,
347
+ "step": 1375
348
+ },
349
+ {
350
+ "epoch": 0.14,
351
+ "learning_rate": 9.054736842105264e-06,
352
+ "loss": 0.3121,
353
+ "step": 1400
354
+ },
355
+ {
356
+ "epoch": 0.14,
357
+ "learning_rate": 9.02842105263158e-06,
358
+ "loss": 0.3115,
359
+ "step": 1425
360
+ },
361
+ {
362
+ "epoch": 0.14,
363
+ "learning_rate": 9.002105263157895e-06,
364
+ "loss": 0.3373,
365
+ "step": 1450
366
+ },
367
+ {
368
+ "epoch": 0.15,
369
+ "learning_rate": 8.97578947368421e-06,
370
+ "loss": 0.3083,
371
+ "step": 1475
372
+ },
373
+ {
374
+ "epoch": 0.15,
375
+ "learning_rate": 8.949473684210527e-06,
376
+ "loss": 0.2566,
377
+ "step": 1500
378
+ },
379
+ {
380
+ "epoch": 0.15,
381
+ "learning_rate": 8.923157894736842e-06,
382
+ "loss": 0.3019,
383
+ "step": 1525
384
+ },
385
+ {
386
+ "epoch": 0.15,
387
+ "learning_rate": 8.896842105263159e-06,
388
+ "loss": 0.2855,
389
+ "step": 1550
390
+ },
391
+ {
392
+ "epoch": 0.16,
393
+ "learning_rate": 8.870526315789474e-06,
394
+ "loss": 0.2961,
395
+ "step": 1575
396
+ },
397
+ {
398
+ "epoch": 0.16,
399
+ "learning_rate": 8.84421052631579e-06,
400
+ "loss": 0.2786,
401
+ "step": 1600
402
+ },
403
+ {
404
+ "epoch": 0.16,
405
+ "learning_rate": 8.817894736842107e-06,
406
+ "loss": 0.3422,
407
+ "step": 1625
408
+ },
409
+ {
410
+ "epoch": 0.17,
411
+ "learning_rate": 8.791578947368422e-06,
412
+ "loss": 0.2714,
413
+ "step": 1650
414
+ },
415
+ {
416
+ "epoch": 0.17,
417
+ "learning_rate": 8.765263157894739e-06,
418
+ "loss": 0.2986,
419
+ "step": 1675
420
+ },
421
+ {
422
+ "epoch": 0.17,
423
+ "learning_rate": 8.738947368421053e-06,
424
+ "loss": 0.264,
425
+ "step": 1700
426
+ },
427
+ {
428
+ "epoch": 0.17,
429
+ "learning_rate": 8.712631578947368e-06,
430
+ "loss": 0.3402,
431
+ "step": 1725
432
+ },
433
+ {
434
+ "epoch": 0.17,
435
+ "learning_rate": 8.686315789473685e-06,
436
+ "loss": 0.2863,
437
+ "step": 1750
438
+ },
439
+ {
440
+ "epoch": 0.18,
441
+ "learning_rate": 8.66e-06,
442
+ "loss": 0.258,
443
+ "step": 1775
444
+ },
445
+ {
446
+ "epoch": 0.18,
447
+ "learning_rate": 8.633684210526317e-06,
448
+ "loss": 0.3005,
449
+ "step": 1800
450
+ },
451
+ {
452
+ "epoch": 0.18,
453
+ "learning_rate": 8.607368421052632e-06,
454
+ "loss": 0.2873,
455
+ "step": 1825
456
+ },
457
+ {
458
+ "epoch": 0.18,
459
+ "learning_rate": 8.581052631578948e-06,
460
+ "loss": 0.3047,
461
+ "step": 1850
462
+ },
463
+ {
464
+ "epoch": 0.19,
465
+ "learning_rate": 8.554736842105263e-06,
466
+ "loss": 0.3019,
467
+ "step": 1875
468
+ },
469
+ {
470
+ "epoch": 0.19,
471
+ "learning_rate": 8.528421052631578e-06,
472
+ "loss": 0.3317,
473
+ "step": 1900
474
+ },
475
+ {
476
+ "epoch": 0.19,
477
+ "learning_rate": 8.502105263157897e-06,
478
+ "loss": 0.293,
479
+ "step": 1925
480
+ },
481
+ {
482
+ "epoch": 0.2,
483
+ "learning_rate": 8.475789473684212e-06,
484
+ "loss": 0.2969,
485
+ "step": 1950
486
+ },
487
+ {
488
+ "epoch": 0.2,
489
+ "learning_rate": 8.449473684210527e-06,
490
+ "loss": 0.2888,
491
+ "step": 1975
492
+ },
493
+ {
494
+ "epoch": 0.2,
495
+ "learning_rate": 8.423157894736843e-06,
496
+ "loss": 0.2676,
497
+ "step": 2000
498
+ },
499
+ {
500
+ "epoch": 0.2,
501
+ "eval_loss": 0.8795959949493408,
502
+ "eval_runtime": 6297.5185,
503
+ "eval_samples_per_second": 0.898,
504
+ "eval_steps_per_second": 0.056,
505
+ "eval_wer": 24.10237098619701,
506
+ "step": 2000
507
+ },
508
+ {
509
+ "epoch": 0.2,
510
+ "learning_rate": 8.396842105263158e-06,
511
+ "loss": 0.248,
512
+ "step": 2025
513
+ },
514
+ {
515
+ "epoch": 0.2,
516
+ "learning_rate": 8.370526315789475e-06,
517
+ "loss": 0.2437,
518
+ "step": 2050
519
+ },
520
+ {
521
+ "epoch": 0.21,
522
+ "learning_rate": 8.34421052631579e-06,
523
+ "loss": 0.2451,
524
+ "step": 2075
525
+ },
526
+ {
527
+ "epoch": 0.21,
528
+ "learning_rate": 8.317894736842107e-06,
529
+ "loss": 0.2911,
530
+ "step": 2100
531
+ },
532
+ {
533
+ "epoch": 0.21,
534
+ "learning_rate": 8.291578947368422e-06,
535
+ "loss": 0.3174,
536
+ "step": 2125
537
+ },
538
+ {
539
+ "epoch": 0.21,
540
+ "learning_rate": 8.265263157894737e-06,
541
+ "loss": 0.3139,
542
+ "step": 2150
543
+ },
544
+ {
545
+ "epoch": 0.22,
546
+ "learning_rate": 8.238947368421053e-06,
547
+ "loss": 0.2514,
548
+ "step": 2175
549
+ },
550
+ {
551
+ "epoch": 0.22,
552
+ "learning_rate": 8.212631578947368e-06,
553
+ "loss": 0.2619,
554
+ "step": 2200
555
+ },
556
+ {
557
+ "epoch": 0.22,
558
+ "learning_rate": 8.186315789473685e-06,
559
+ "loss": 0.3192,
560
+ "step": 2225
561
+ },
562
+ {
563
+ "epoch": 0.23,
564
+ "learning_rate": 8.16e-06,
565
+ "loss": 0.3024,
566
+ "step": 2250
567
+ },
568
+ {
569
+ "epoch": 0.23,
570
+ "learning_rate": 8.133684210526316e-06,
571
+ "loss": 0.2978,
572
+ "step": 2275
573
+ },
574
+ {
575
+ "epoch": 0.23,
576
+ "learning_rate": 8.107368421052633e-06,
577
+ "loss": 0.3104,
578
+ "step": 2300
579
+ },
580
+ {
581
+ "epoch": 0.23,
582
+ "learning_rate": 8.081052631578948e-06,
583
+ "loss": 0.2983,
584
+ "step": 2325
585
+ },
586
+ {
587
+ "epoch": 0.23,
588
+ "learning_rate": 8.054736842105265e-06,
589
+ "loss": 0.2777,
590
+ "step": 2350
591
+ },
592
+ {
593
+ "epoch": 0.24,
594
+ "learning_rate": 8.02842105263158e-06,
595
+ "loss": 0.2971,
596
+ "step": 2375
597
+ },
598
+ {
599
+ "epoch": 0.24,
600
+ "learning_rate": 8.002105263157895e-06,
601
+ "loss": 0.2638,
602
+ "step": 2400
603
+ },
604
+ {
605
+ "epoch": 0.24,
606
+ "learning_rate": 7.975789473684211e-06,
607
+ "loss": 0.24,
608
+ "step": 2425
609
+ },
610
+ {
611
+ "epoch": 0.24,
612
+ "learning_rate": 7.949473684210526e-06,
613
+ "loss": 0.2579,
614
+ "step": 2450
615
+ },
616
+ {
617
+ "epoch": 0.25,
618
+ "learning_rate": 7.923157894736843e-06,
619
+ "loss": 0.2687,
620
+ "step": 2475
621
+ },
622
+ {
623
+ "epoch": 0.25,
624
+ "learning_rate": 7.896842105263158e-06,
625
+ "loss": 0.2591,
626
+ "step": 2500
627
+ },
628
+ {
629
+ "epoch": 0.25,
630
+ "learning_rate": 7.870526315789475e-06,
631
+ "loss": 0.3053,
632
+ "step": 2525
633
+ },
634
+ {
635
+ "epoch": 0.26,
636
+ "learning_rate": 7.84421052631579e-06,
637
+ "loss": 0.2626,
638
+ "step": 2550
639
+ },
640
+ {
641
+ "epoch": 0.26,
642
+ "learning_rate": 7.817894736842105e-06,
643
+ "loss": 0.2799,
644
+ "step": 2575
645
+ },
646
+ {
647
+ "epoch": 0.26,
648
+ "learning_rate": 7.791578947368423e-06,
649
+ "loss": 0.2293,
650
+ "step": 2600
651
+ },
652
+ {
653
+ "epoch": 0.26,
654
+ "learning_rate": 7.765263157894738e-06,
655
+ "loss": 0.251,
656
+ "step": 2625
657
+ },
658
+ {
659
+ "epoch": 0.27,
660
+ "learning_rate": 7.738947368421053e-06,
661
+ "loss": 0.3127,
662
+ "step": 2650
663
+ },
664
+ {
665
+ "epoch": 0.27,
666
+ "learning_rate": 7.71263157894737e-06,
667
+ "loss": 0.2738,
668
+ "step": 2675
669
+ },
670
+ {
671
+ "epoch": 0.27,
672
+ "learning_rate": 7.686315789473685e-06,
673
+ "loss": 0.2327,
674
+ "step": 2700
675
+ },
676
+ {
677
+ "epoch": 0.27,
678
+ "learning_rate": 7.660000000000001e-06,
679
+ "loss": 0.2591,
680
+ "step": 2725
681
+ },
682
+ {
683
+ "epoch": 0.28,
684
+ "learning_rate": 7.633684210526316e-06,
685
+ "loss": 0.2121,
686
+ "step": 2750
687
+ },
688
+ {
689
+ "epoch": 0.28,
690
+ "learning_rate": 7.607368421052632e-06,
691
+ "loss": 0.3053,
692
+ "step": 2775
693
+ },
694
+ {
695
+ "epoch": 0.28,
696
+ "learning_rate": 7.581052631578948e-06,
697
+ "loss": 0.2791,
698
+ "step": 2800
699
+ },
700
+ {
701
+ "epoch": 0.28,
702
+ "learning_rate": 7.554736842105264e-06,
703
+ "loss": 0.2519,
704
+ "step": 2825
705
+ },
706
+ {
707
+ "epoch": 0.28,
708
+ "learning_rate": 7.5284210526315794e-06,
709
+ "loss": 0.2564,
710
+ "step": 2850
711
+ },
712
+ {
713
+ "epoch": 0.29,
714
+ "learning_rate": 7.502105263157895e-06,
715
+ "loss": 0.2464,
716
+ "step": 2875
717
+ },
718
+ {
719
+ "epoch": 0.29,
720
+ "learning_rate": 7.475789473684211e-06,
721
+ "loss": 0.3187,
722
+ "step": 2900
723
+ },
724
+ {
725
+ "epoch": 0.29,
726
+ "learning_rate": 7.449473684210526e-06,
727
+ "loss": 0.3153,
728
+ "step": 2925
729
+ },
730
+ {
731
+ "epoch": 0.29,
732
+ "learning_rate": 7.4231578947368436e-06,
733
+ "loss": 0.2961,
734
+ "step": 2950
735
+ },
736
+ {
737
+ "epoch": 0.3,
738
+ "learning_rate": 7.3968421052631585e-06,
739
+ "loss": 0.2807,
740
+ "step": 2975
741
+ },
742
+ {
743
+ "epoch": 0.3,
744
+ "learning_rate": 7.370526315789474e-06,
745
+ "loss": 0.3166,
746
+ "step": 3000
747
+ },
748
+ {
749
+ "epoch": 0.3,
750
+ "eval_loss": 0.8467275500297546,
751
+ "eval_runtime": 5805.1114,
752
+ "eval_samples_per_second": 0.974,
753
+ "eval_steps_per_second": 0.061,
754
+ "eval_wer": 20.17003896122706,
755
+ "step": 3000
756
+ },
757
+ {
758
+ "epoch": 0.3,
759
+ "learning_rate": 7.34421052631579e-06,
760
+ "loss": 0.2834,
761
+ "step": 3025
762
+ },
763
+ {
764
+ "epoch": 0.3,
765
+ "learning_rate": 7.317894736842106e-06,
766
+ "loss": 0.2622,
767
+ "step": 3050
768
+ },
769
+ {
770
+ "epoch": 0.31,
771
+ "learning_rate": 7.291578947368422e-06,
772
+ "loss": 0.2616,
773
+ "step": 3075
774
+ },
775
+ {
776
+ "epoch": 0.31,
777
+ "learning_rate": 7.265263157894738e-06,
778
+ "loss": 0.2829,
779
+ "step": 3100
780
+ },
781
+ {
782
+ "epoch": 0.31,
783
+ "learning_rate": 7.2389473684210534e-06,
784
+ "loss": 0.2634,
785
+ "step": 3125
786
+ },
787
+ {
788
+ "epoch": 0.32,
789
+ "learning_rate": 7.212631578947369e-06,
790
+ "loss": 0.2936,
791
+ "step": 3150
792
+ },
793
+ {
794
+ "epoch": 0.32,
795
+ "learning_rate": 7.186315789473684e-06,
796
+ "loss": 0.2626,
797
+ "step": 3175
798
+ },
799
+ {
800
+ "epoch": 0.32,
801
+ "learning_rate": 7.16e-06,
802
+ "loss": 0.3031,
803
+ "step": 3200
804
+ },
805
+ {
806
+ "epoch": 0.32,
807
+ "learning_rate": 7.133684210526316e-06,
808
+ "loss": 0.2826,
809
+ "step": 3225
810
+ },
811
+ {
812
+ "epoch": 0.33,
813
+ "learning_rate": 7.107368421052632e-06,
814
+ "loss": 0.2545,
815
+ "step": 3250
816
+ },
817
+ {
818
+ "epoch": 0.33,
819
+ "learning_rate": 7.0810526315789475e-06,
820
+ "loss": 0.248,
821
+ "step": 3275
822
+ },
823
+ {
824
+ "epoch": 0.33,
825
+ "learning_rate": 7.054736842105264e-06,
826
+ "loss": 0.2606,
827
+ "step": 3300
828
+ },
829
+ {
830
+ "epoch": 0.33,
831
+ "learning_rate": 7.02842105263158e-06,
832
+ "loss": 0.2495,
833
+ "step": 3325
834
+ },
835
+ {
836
+ "epoch": 0.34,
837
+ "learning_rate": 7.002105263157896e-06,
838
+ "loss": 0.2657,
839
+ "step": 3350
840
+ },
841
+ {
842
+ "epoch": 0.34,
843
+ "learning_rate": 6.975789473684212e-06,
844
+ "loss": 0.3054,
845
+ "step": 3375
846
+ },
847
+ {
848
+ "epoch": 0.34,
849
+ "learning_rate": 6.9494736842105275e-06,
850
+ "loss": 0.2985,
851
+ "step": 3400
852
+ },
853
+ {
854
+ "epoch": 0.34,
855
+ "learning_rate": 6.9231578947368424e-06,
856
+ "loss": 0.2544,
857
+ "step": 3425
858
+ },
859
+ {
860
+ "epoch": 0.34,
861
+ "learning_rate": 6.896842105263158e-06,
862
+ "loss": 0.2464,
863
+ "step": 3450
864
+ },
865
+ {
866
+ "epoch": 0.35,
867
+ "learning_rate": 6.870526315789474e-06,
868
+ "loss": 0.256,
869
+ "step": 3475
870
+ },
871
+ {
872
+ "epoch": 0.35,
873
+ "learning_rate": 6.84421052631579e-06,
874
+ "loss": 0.2706,
875
+ "step": 3500
876
+ },
877
+ {
878
+ "epoch": 0.35,
879
+ "learning_rate": 6.817894736842106e-06,
880
+ "loss": 0.2725,
881
+ "step": 3525
882
+ },
883
+ {
884
+ "epoch": 0.35,
885
+ "learning_rate": 6.7915789473684215e-06,
886
+ "loss": 0.2693,
887
+ "step": 3550
888
+ },
889
+ {
890
+ "epoch": 0.36,
891
+ "learning_rate": 6.765263157894737e-06,
892
+ "loss": 0.2447,
893
+ "step": 3575
894
+ },
895
+ {
896
+ "epoch": 0.36,
897
+ "learning_rate": 6.738947368421052e-06,
898
+ "loss": 0.2581,
899
+ "step": 3600
900
+ },
901
+ {
902
+ "epoch": 0.36,
903
+ "learning_rate": 6.71263157894737e-06,
904
+ "loss": 0.2797,
905
+ "step": 3625
906
+ },
907
+ {
908
+ "epoch": 0.36,
909
+ "learning_rate": 6.686315789473685e-06,
910
+ "loss": 0.252,
911
+ "step": 3650
912
+ },
913
+ {
914
+ "epoch": 0.37,
915
+ "learning_rate": 6.660000000000001e-06,
916
+ "loss": 0.2335,
917
+ "step": 3675
918
+ },
919
+ {
920
+ "epoch": 0.37,
921
+ "learning_rate": 6.6336842105263164e-06,
922
+ "loss": 0.2673,
923
+ "step": 3700
924
+ },
925
+ {
926
+ "epoch": 0.37,
927
+ "learning_rate": 6.607368421052632e-06,
928
+ "loss": 0.2552,
929
+ "step": 3725
930
+ },
931
+ {
932
+ "epoch": 0.38,
933
+ "learning_rate": 6.581052631578948e-06,
934
+ "loss": 0.2655,
935
+ "step": 3750
936
+ },
937
+ {
938
+ "epoch": 0.38,
939
+ "learning_rate": 6.554736842105264e-06,
940
+ "loss": 0.2437,
941
+ "step": 3775
942
+ },
943
+ {
944
+ "epoch": 0.38,
945
+ "learning_rate": 6.52842105263158e-06,
946
+ "loss": 0.225,
947
+ "step": 3800
948
+ },
949
+ {
950
+ "epoch": 0.38,
951
+ "learning_rate": 6.5021052631578955e-06,
952
+ "loss": 0.2483,
953
+ "step": 3825
954
+ },
955
+ {
956
+ "epoch": 0.39,
957
+ "learning_rate": 6.4757894736842105e-06,
958
+ "loss": 0.2731,
959
+ "step": 3850
960
+ },
961
+ {
962
+ "epoch": 0.39,
963
+ "learning_rate": 6.449473684210526e-06,
964
+ "loss": 0.2181,
965
+ "step": 3875
966
+ },
967
+ {
968
+ "epoch": 0.39,
969
+ "learning_rate": 6.423157894736842e-06,
970
+ "loss": 0.231,
971
+ "step": 3900
972
+ },
973
+ {
974
+ "epoch": 0.39,
975
+ "learning_rate": 6.396842105263158e-06,
976
+ "loss": 0.2596,
977
+ "step": 3925
978
+ },
979
+ {
980
+ "epoch": 0.4,
981
+ "learning_rate": 6.370526315789474e-06,
982
+ "loss": 0.2612,
983
+ "step": 3950
984
+ },
985
+ {
986
+ "epoch": 0.4,
987
+ "learning_rate": 6.3442105263157904e-06,
988
+ "loss": 0.22,
989
+ "step": 3975
990
+ },
991
+ {
992
+ "epoch": 0.4,
993
+ "learning_rate": 6.317894736842106e-06,
994
+ "loss": 0.2797,
995
+ "step": 4000
996
+ },
997
+ {
998
+ "epoch": 0.4,
999
+ "eval_loss": 0.8756310343742371,
1000
+ "eval_runtime": 6384.9348,
1001
+ "eval_samples_per_second": 0.886,
1002
+ "eval_steps_per_second": 0.055,
1003
+ "eval_wer": 29.48886925539157,
1004
+ "step": 4000
1005
+ },
1006
+ {
1007
+ "epoch": 0.4,
1008
+ "learning_rate": 6.292631578947369e-06,
1009
+ "loss": 0.298,
1010
+ "step": 4025
1011
+ },
1012
+ {
1013
+ "epoch": 0.41,
1014
+ "learning_rate": 6.266315789473685e-06,
1015
+ "loss": 0.2248,
1016
+ "step": 4050
1017
+ },
1018
+ {
1019
+ "epoch": 0.41,
1020
+ "learning_rate": 6.24e-06,
1021
+ "loss": 0.2592,
1022
+ "step": 4075
1023
+ },
1024
+ {
1025
+ "epoch": 0.41,
1026
+ "learning_rate": 6.213684210526316e-06,
1027
+ "loss": 0.2575,
1028
+ "step": 4100
1029
+ },
1030
+ {
1031
+ "epoch": 0.41,
1032
+ "learning_rate": 6.187368421052632e-06,
1033
+ "loss": 0.2678,
1034
+ "step": 4125
1035
+ },
1036
+ {
1037
+ "epoch": 0.41,
1038
+ "learning_rate": 6.161052631578948e-06,
1039
+ "loss": 0.2751,
1040
+ "step": 4150
1041
+ },
1042
+ {
1043
+ "epoch": 0.42,
1044
+ "learning_rate": 6.134736842105264e-06,
1045
+ "loss": 0.2259,
1046
+ "step": 4175
1047
+ },
1048
+ {
1049
+ "epoch": 0.42,
1050
+ "learning_rate": 6.108421052631579e-06,
1051
+ "loss": 0.2301,
1052
+ "step": 4200
1053
+ },
1054
+ {
1055
+ "epoch": 0.42,
1056
+ "learning_rate": 6.0821052631578945e-06,
1057
+ "loss": 0.2767,
1058
+ "step": 4225
1059
+ },
1060
+ {
1061
+ "epoch": 0.42,
1062
+ "learning_rate": 6.055789473684212e-06,
1063
+ "loss": 0.2747,
1064
+ "step": 4250
1065
+ },
1066
+ {
1067
+ "epoch": 0.43,
1068
+ "learning_rate": 6.029473684210527e-06,
1069
+ "loss": 0.2657,
1070
+ "step": 4275
1071
+ },
1072
+ {
1073
+ "epoch": 0.43,
1074
+ "learning_rate": 6.003157894736843e-06,
1075
+ "loss": 0.2437,
1076
+ "step": 4300
1077
+ },
1078
+ {
1079
+ "epoch": 0.43,
1080
+ "learning_rate": 5.976842105263159e-06,
1081
+ "loss": 0.2511,
1082
+ "step": 4325
1083
+ },
1084
+ {
1085
+ "epoch": 0.43,
1086
+ "learning_rate": 5.950526315789474e-06,
1087
+ "loss": 0.2957,
1088
+ "step": 4350
1089
+ },
1090
+ {
1091
+ "epoch": 0.44,
1092
+ "learning_rate": 5.92421052631579e-06,
1093
+ "loss": 0.2479,
1094
+ "step": 4375
1095
+ },
1096
+ {
1097
+ "epoch": 0.44,
1098
+ "learning_rate": 5.897894736842106e-06,
1099
+ "loss": 0.2582,
1100
+ "step": 4400
1101
+ },
1102
+ {
1103
+ "epoch": 0.44,
1104
+ "learning_rate": 5.871578947368422e-06,
1105
+ "loss": 0.2621,
1106
+ "step": 4425
1107
+ },
1108
+ {
1109
+ "epoch": 0.45,
1110
+ "learning_rate": 5.845263157894737e-06,
1111
+ "loss": 0.2489,
1112
+ "step": 4450
1113
+ },
1114
+ {
1115
+ "epoch": 0.45,
1116
+ "learning_rate": 5.818947368421053e-06,
1117
+ "loss": 0.2428,
1118
+ "step": 4475
1119
+ },
1120
+ {
1121
+ "epoch": 0.45,
1122
+ "learning_rate": 5.7926315789473685e-06,
1123
+ "loss": 0.2415,
1124
+ "step": 4500
1125
+ },
1126
+ {
1127
+ "epoch": 0.45,
1128
+ "learning_rate": 5.766315789473684e-06,
1129
+ "loss": 0.2493,
1130
+ "step": 4525
1131
+ },
1132
+ {
1133
+ "epoch": 0.46,
1134
+ "learning_rate": 5.74e-06,
1135
+ "loss": 0.2741,
1136
+ "step": 4550
1137
+ },
1138
+ {
1139
+ "epoch": 0.46,
1140
+ "learning_rate": 5.713684210526316e-06,
1141
+ "loss": 0.2756,
1142
+ "step": 4575
1143
+ },
1144
+ {
1145
+ "epoch": 0.46,
1146
+ "learning_rate": 5.687368421052633e-06,
1147
+ "loss": 0.2737,
1148
+ "step": 4600
1149
+ },
1150
+ {
1151
+ "epoch": 0.46,
1152
+ "learning_rate": 5.661052631578948e-06,
1153
+ "loss": 0.2446,
1154
+ "step": 4625
1155
+ },
1156
+ {
1157
+ "epoch": 0.47,
1158
+ "learning_rate": 5.634736842105264e-06,
1159
+ "loss": 0.2724,
1160
+ "step": 4650
1161
+ },
1162
+ {
1163
+ "epoch": 0.47,
1164
+ "learning_rate": 5.60842105263158e-06,
1165
+ "loss": 0.2267,
1166
+ "step": 4675
1167
+ },
1168
+ {
1169
+ "epoch": 0.47,
1170
+ "learning_rate": 5.582105263157895e-06,
1171
+ "loss": 0.2626,
1172
+ "step": 4700
1173
+ },
1174
+ {
1175
+ "epoch": 0.47,
1176
+ "learning_rate": 5.555789473684211e-06,
1177
+ "loss": 0.2432,
1178
+ "step": 4725
1179
+ },
1180
+ {
1181
+ "epoch": 0.47,
1182
+ "learning_rate": 5.529473684210527e-06,
1183
+ "loss": 0.2654,
1184
+ "step": 4750
1185
+ },
1186
+ {
1187
+ "epoch": 0.48,
1188
+ "learning_rate": 5.5031578947368425e-06,
1189
+ "loss": 0.2832,
1190
+ "step": 4775
1191
+ },
1192
+ {
1193
+ "epoch": 0.48,
1194
+ "learning_rate": 5.476842105263158e-06,
1195
+ "loss": 0.263,
1196
+ "step": 4800
1197
+ },
1198
+ {
1199
+ "epoch": 0.48,
1200
+ "learning_rate": 5.450526315789474e-06,
1201
+ "loss": 0.2561,
1202
+ "step": 4825
1203
+ },
1204
+ {
1205
+ "epoch": 0.48,
1206
+ "learning_rate": 5.42421052631579e-06,
1207
+ "loss": 0.2831,
1208
+ "step": 4850
1209
+ },
1210
+ {
1211
+ "epoch": 0.49,
1212
+ "learning_rate": 5.397894736842105e-06,
1213
+ "loss": 0.2593,
1214
+ "step": 4875
1215
+ },
1216
+ {
1217
+ "epoch": 0.49,
1218
+ "learning_rate": 5.371578947368421e-06,
1219
+ "loss": 0.235,
1220
+ "step": 4900
1221
+ },
1222
+ {
1223
+ "epoch": 0.49,
1224
+ "learning_rate": 5.345263157894738e-06,
1225
+ "loss": 0.2342,
1226
+ "step": 4925
1227
+ },
1228
+ {
1229
+ "epoch": 0.49,
1230
+ "learning_rate": 5.318947368421053e-06,
1231
+ "loss": 0.2304,
1232
+ "step": 4950
1233
+ },
1234
+ {
1235
+ "epoch": 0.5,
1236
+ "learning_rate": 5.292631578947369e-06,
1237
+ "loss": 0.2422,
1238
+ "step": 4975
1239
+ },
1240
+ {
1241
+ "epoch": 0.5,
1242
+ "learning_rate": 5.266315789473685e-06,
1243
+ "loss": 0.2302,
1244
+ "step": 5000
1245
+ },
1246
+ {
1247
+ "epoch": 0.5,
1248
+ "eval_loss": 0.8522536754608154,
1249
+ "eval_runtime": 5802.9054,
1250
+ "eval_samples_per_second": 0.975,
1251
+ "eval_steps_per_second": 0.061,
1252
+ "eval_wer": 19.641382906304766,
1253
+ "step": 5000
1254
+ },
1255
+ {
1256
+ "epoch": 0.5,
1257
+ "learning_rate": 5.240000000000001e-06,
1258
+ "loss": 0.272,
1259
+ "step": 5025
1260
+ },
1261
+ {
1262
+ "epoch": 0.51,
1263
+ "learning_rate": 5.2136842105263165e-06,
1264
+ "loss": 0.2574,
1265
+ "step": 5050
1266
+ },
1267
+ {
1268
+ "epoch": 0.51,
1269
+ "learning_rate": 5.187368421052632e-06,
1270
+ "loss": 0.3026,
1271
+ "step": 5075
1272
+ },
1273
+ {
1274
+ "epoch": 0.51,
1275
+ "learning_rate": 5.161052631578948e-06,
1276
+ "loss": 0.2287,
1277
+ "step": 5100
1278
+ },
1279
+ {
1280
+ "epoch": 0.51,
1281
+ "learning_rate": 5.134736842105263e-06,
1282
+ "loss": 0.2492,
1283
+ "step": 5125
1284
+ },
1285
+ {
1286
+ "epoch": 0.52,
1287
+ "learning_rate": 5.108421052631579e-06,
1288
+ "loss": 0.2267,
1289
+ "step": 5150
1290
+ },
1291
+ {
1292
+ "epoch": 0.52,
1293
+ "learning_rate": 5.082105263157895e-06,
1294
+ "loss": 0.2427,
1295
+ "step": 5175
1296
+ },
1297
+ {
1298
+ "epoch": 0.52,
1299
+ "learning_rate": 5.0557894736842105e-06,
1300
+ "loss": 0.2528,
1301
+ "step": 5200
1302
+ },
1303
+ {
1304
+ "epoch": 0.52,
1305
+ "learning_rate": 5.029473684210526e-06,
1306
+ "loss": 0.2463,
1307
+ "step": 5225
1308
+ },
1309
+ {
1310
+ "epoch": 0.53,
1311
+ "learning_rate": 5.003157894736842e-06,
1312
+ "loss": 0.2235,
1313
+ "step": 5250
1314
+ },
1315
+ {
1316
+ "epoch": 0.53,
1317
+ "learning_rate": 4.976842105263158e-06,
1318
+ "loss": 0.2199,
1319
+ "step": 5275
1320
+ },
1321
+ {
1322
+ "epoch": 0.53,
1323
+ "learning_rate": 4.950526315789474e-06,
1324
+ "loss": 0.2966,
1325
+ "step": 5300
1326
+ },
1327
+ {
1328
+ "epoch": 0.53,
1329
+ "learning_rate": 4.92421052631579e-06,
1330
+ "loss": 0.2853,
1331
+ "step": 5325
1332
+ },
1333
+ {
1334
+ "epoch": 0.54,
1335
+ "learning_rate": 4.897894736842106e-06,
1336
+ "loss": 0.2454,
1337
+ "step": 5350
1338
+ },
1339
+ {
1340
+ "epoch": 0.54,
1341
+ "learning_rate": 4.871578947368421e-06,
1342
+ "loss": 0.2306,
1343
+ "step": 5375
1344
+ },
1345
+ {
1346
+ "epoch": 0.54,
1347
+ "learning_rate": 4.845263157894737e-06,
1348
+ "loss": 0.2181,
1349
+ "step": 5400
1350
+ },
1351
+ {
1352
+ "epoch": 0.54,
1353
+ "learning_rate": 4.818947368421053e-06,
1354
+ "loss": 0.2565,
1355
+ "step": 5425
1356
+ },
1357
+ {
1358
+ "epoch": 0.55,
1359
+ "learning_rate": 4.792631578947369e-06,
1360
+ "loss": 0.2506,
1361
+ "step": 5450
1362
+ },
1363
+ {
1364
+ "epoch": 0.55,
1365
+ "learning_rate": 4.7663157894736845e-06,
1366
+ "loss": 0.2515,
1367
+ "step": 5475
1368
+ },
1369
+ {
1370
+ "epoch": 0.55,
1371
+ "learning_rate": 4.74e-06,
1372
+ "loss": 0.2233,
1373
+ "step": 5500
1374
+ },
1375
+ {
1376
+ "epoch": 0.55,
1377
+ "learning_rate": 4.713684210526316e-06,
1378
+ "loss": 0.2675,
1379
+ "step": 5525
1380
+ },
1381
+ {
1382
+ "epoch": 0.56,
1383
+ "learning_rate": 4.687368421052632e-06,
1384
+ "loss": 0.2312,
1385
+ "step": 5550
1386
+ },
1387
+ {
1388
+ "epoch": 0.56,
1389
+ "learning_rate": 4.661052631578948e-06,
1390
+ "loss": 0.2872,
1391
+ "step": 5575
1392
+ },
1393
+ {
1394
+ "epoch": 0.56,
1395
+ "learning_rate": 4.634736842105264e-06,
1396
+ "loss": 0.298,
1397
+ "step": 5600
1398
+ },
1399
+ {
1400
+ "epoch": 0.56,
1401
+ "learning_rate": 4.6084210526315794e-06,
1402
+ "loss": 0.2632,
1403
+ "step": 5625
1404
+ },
1405
+ {
1406
+ "epoch": 0.56,
1407
+ "learning_rate": 4.582105263157895e-06,
1408
+ "loss": 0.2398,
1409
+ "step": 5650
1410
+ },
1411
+ {
1412
+ "epoch": 0.57,
1413
+ "learning_rate": 4.55578947368421e-06,
1414
+ "loss": 0.2713,
1415
+ "step": 5675
1416
+ },
1417
+ {
1418
+ "epoch": 0.57,
1419
+ "learning_rate": 4.529473684210527e-06,
1420
+ "loss": 0.2764,
1421
+ "step": 5700
1422
+ },
1423
+ {
1424
+ "epoch": 0.57,
1425
+ "learning_rate": 4.503157894736843e-06,
1426
+ "loss": 0.2476,
1427
+ "step": 5725
1428
+ },
1429
+ {
1430
+ "epoch": 0.57,
1431
+ "learning_rate": 4.4768421052631585e-06,
1432
+ "loss": 0.2391,
1433
+ "step": 5750
1434
+ },
1435
+ {
1436
+ "epoch": 0.58,
1437
+ "learning_rate": 4.450526315789474e-06,
1438
+ "loss": 0.2045,
1439
+ "step": 5775
1440
+ },
1441
+ {
1442
+ "epoch": 0.58,
1443
+ "learning_rate": 4.424210526315789e-06,
1444
+ "loss": 0.2232,
1445
+ "step": 5800
1446
+ },
1447
+ {
1448
+ "epoch": 0.58,
1449
+ "learning_rate": 4.397894736842105e-06,
1450
+ "loss": 0.2835,
1451
+ "step": 5825
1452
+ },
1453
+ {
1454
+ "epoch": 0.58,
1455
+ "learning_rate": 4.371578947368421e-06,
1456
+ "loss": 0.2671,
1457
+ "step": 5850
1458
+ },
1459
+ {
1460
+ "epoch": 0.59,
1461
+ "learning_rate": 4.345263157894738e-06,
1462
+ "loss": 0.2324,
1463
+ "step": 5875
1464
+ },
1465
+ {
1466
+ "epoch": 0.59,
1467
+ "learning_rate": 4.3189473684210535e-06,
1468
+ "loss": 0.2082,
1469
+ "step": 5900
1470
+ },
1471
+ {
1472
+ "epoch": 0.59,
1473
+ "learning_rate": 4.2926315789473684e-06,
1474
+ "loss": 0.2369,
1475
+ "step": 5925
1476
+ },
1477
+ {
1478
+ "epoch": 0.59,
1479
+ "learning_rate": 4.266315789473684e-06,
1480
+ "loss": 0.2593,
1481
+ "step": 5950
1482
+ },
1483
+ {
1484
+ "epoch": 0.6,
1485
+ "learning_rate": 4.24e-06,
1486
+ "loss": 0.3048,
1487
+ "step": 5975
1488
+ },
1489
+ {
1490
+ "epoch": 0.6,
1491
+ "learning_rate": 4.213684210526316e-06,
1492
+ "loss": 0.2803,
1493
+ "step": 6000
1494
+ },
1495
+ {
1496
+ "epoch": 0.6,
1497
+ "eval_loss": 0.8714997172355652,
1498
+ "eval_runtime": 5666.5233,
1499
+ "eval_samples_per_second": 0.998,
1500
+ "eval_steps_per_second": 0.062,
1501
+ "eval_wer": 19.74132062627638,
1502
+ "step": 6000
1503
+ },
1504
+ {
1505
+ "epoch": 0.6,
1506
+ "learning_rate": 4.1873684210526325e-06,
1507
+ "loss": 0.236,
1508
+ "step": 6025
1509
+ },
1510
+ {
1511
+ "epoch": 0.6,
1512
+ "learning_rate": 4.1610526315789475e-06,
1513
+ "loss": 0.2652,
1514
+ "step": 6050
1515
+ },
1516
+ {
1517
+ "epoch": 0.61,
1518
+ "learning_rate": 4.134736842105263e-06,
1519
+ "loss": 0.2346,
1520
+ "step": 6075
1521
+ },
1522
+ {
1523
+ "epoch": 0.61,
1524
+ "learning_rate": 4.108421052631579e-06,
1525
+ "loss": 0.2464,
1526
+ "step": 6100
1527
+ },
1528
+ {
1529
+ "epoch": 0.61,
1530
+ "learning_rate": 4.082105263157895e-06,
1531
+ "loss": 0.2253,
1532
+ "step": 6125
1533
+ },
1534
+ {
1535
+ "epoch": 0.61,
1536
+ "learning_rate": 4.055789473684211e-06,
1537
+ "loss": 0.2244,
1538
+ "step": 6150
1539
+ },
1540
+ {
1541
+ "epoch": 0.62,
1542
+ "learning_rate": 4.029473684210527e-06,
1543
+ "loss": 0.2151,
1544
+ "step": 6175
1545
+ },
1546
+ {
1547
+ "epoch": 0.62,
1548
+ "learning_rate": 4.0031578947368424e-06,
1549
+ "loss": 0.2573,
1550
+ "step": 6200
1551
+ },
1552
+ {
1553
+ "epoch": 0.62,
1554
+ "learning_rate": 3.977894736842106e-06,
1555
+ "loss": 0.241,
1556
+ "step": 6225
1557
+ },
1558
+ {
1559
+ "epoch": 0.62,
1560
+ "learning_rate": 3.951578947368422e-06,
1561
+ "loss": 0.2797,
1562
+ "step": 6250
1563
+ },
1564
+ {
1565
+ "epoch": 0.63,
1566
+ "learning_rate": 3.9252631578947366e-06,
1567
+ "loss": 0.2502,
1568
+ "step": 6275
1569
+ },
1570
+ {
1571
+ "epoch": 0.63,
1572
+ "learning_rate": 3.898947368421052e-06,
1573
+ "loss": 0.2128,
1574
+ "step": 6300
1575
+ },
1576
+ {
1577
+ "epoch": 0.63,
1578
+ "learning_rate": 3.872631578947369e-06,
1579
+ "loss": 0.2127,
1580
+ "step": 6325
1581
+ },
1582
+ {
1583
+ "epoch": 0.64,
1584
+ "learning_rate": 3.846315789473685e-06,
1585
+ "loss": 0.2245,
1586
+ "step": 6350
1587
+ },
1588
+ {
1589
+ "epoch": 0.64,
1590
+ "learning_rate": 3.820000000000001e-06,
1591
+ "loss": 0.2224,
1592
+ "step": 6375
1593
+ },
1594
+ {
1595
+ "epoch": 0.64,
1596
+ "learning_rate": 3.793684210526316e-06,
1597
+ "loss": 0.1967,
1598
+ "step": 6400
1599
+ },
1600
+ {
1601
+ "epoch": 0.64,
1602
+ "learning_rate": 3.767368421052632e-06,
1603
+ "loss": 0.2379,
1604
+ "step": 6425
1605
+ },
1606
+ {
1607
+ "epoch": 0.65,
1608
+ "learning_rate": 3.7410526315789473e-06,
1609
+ "loss": 0.2202,
1610
+ "step": 6450
1611
+ },
1612
+ {
1613
+ "epoch": 0.65,
1614
+ "learning_rate": 3.714736842105263e-06,
1615
+ "loss": 0.2221,
1616
+ "step": 6475
1617
+ },
1618
+ {
1619
+ "epoch": 0.65,
1620
+ "learning_rate": 3.6884210526315794e-06,
1621
+ "loss": 0.2227,
1622
+ "step": 6500
1623
+ },
1624
+ {
1625
+ "epoch": 0.65,
1626
+ "learning_rate": 3.662105263157895e-06,
1627
+ "loss": 0.2374,
1628
+ "step": 6525
1629
+ },
1630
+ {
1631
+ "epoch": 0.66,
1632
+ "learning_rate": 3.635789473684211e-06,
1633
+ "loss": 0.25,
1634
+ "step": 6550
1635
+ },
1636
+ {
1637
+ "epoch": 0.66,
1638
+ "learning_rate": 3.6094736842105264e-06,
1639
+ "loss": 0.244,
1640
+ "step": 6575
1641
+ },
1642
+ {
1643
+ "epoch": 0.66,
1644
+ "learning_rate": 3.5831578947368422e-06,
1645
+ "loss": 0.2068,
1646
+ "step": 6600
1647
+ },
1648
+ {
1649
+ "epoch": 0.66,
1650
+ "learning_rate": 3.556842105263158e-06,
1651
+ "loss": 0.2312,
1652
+ "step": 6625
1653
+ },
1654
+ {
1655
+ "epoch": 0.67,
1656
+ "learning_rate": 3.5305263157894743e-06,
1657
+ "loss": 0.2171,
1658
+ "step": 6650
1659
+ },
1660
+ {
1661
+ "epoch": 0.67,
1662
+ "learning_rate": 3.50421052631579e-06,
1663
+ "loss": 0.2417,
1664
+ "step": 6675
1665
+ },
1666
+ {
1667
+ "epoch": 0.67,
1668
+ "learning_rate": 3.4778947368421055e-06,
1669
+ "loss": 0.2452,
1670
+ "step": 6700
1671
+ },
1672
+ {
1673
+ "epoch": 0.67,
1674
+ "learning_rate": 3.4515789473684213e-06,
1675
+ "loss": 0.2459,
1676
+ "step": 6725
1677
+ },
1678
+ {
1679
+ "epoch": 0.68,
1680
+ "learning_rate": 3.425263157894737e-06,
1681
+ "loss": 0.2265,
1682
+ "step": 6750
1683
+ },
1684
+ {
1685
+ "epoch": 0.68,
1686
+ "learning_rate": 3.398947368421053e-06,
1687
+ "loss": 0.2276,
1688
+ "step": 6775
1689
+ },
1690
+ {
1691
+ "epoch": 0.68,
1692
+ "learning_rate": 3.3726315789473683e-06,
1693
+ "loss": 0.258,
1694
+ "step": 6800
1695
+ },
1696
+ {
1697
+ "epoch": 0.68,
1698
+ "learning_rate": 3.3463157894736846e-06,
1699
+ "loss": 0.2292,
1700
+ "step": 6825
1701
+ },
1702
+ {
1703
+ "epoch": 0.69,
1704
+ "learning_rate": 3.3200000000000004e-06,
1705
+ "loss": 0.3045,
1706
+ "step": 6850
1707
+ },
1708
+ {
1709
+ "epoch": 0.69,
1710
+ "learning_rate": 3.2936842105263162e-06,
1711
+ "loss": 0.2752,
1712
+ "step": 6875
1713
+ },
1714
+ {
1715
+ "epoch": 0.69,
1716
+ "learning_rate": 3.267368421052632e-06,
1717
+ "loss": 0.2702,
1718
+ "step": 6900
1719
+ },
1720
+ {
1721
+ "epoch": 0.69,
1722
+ "learning_rate": 3.2410526315789474e-06,
1723
+ "loss": 0.2365,
1724
+ "step": 6925
1725
+ },
1726
+ {
1727
+ "epoch": 0.69,
1728
+ "learning_rate": 3.2147368421052633e-06,
1729
+ "loss": 0.2123,
1730
+ "step": 6950
1731
+ },
1732
+ {
1733
+ "epoch": 0.7,
1734
+ "learning_rate": 3.188421052631579e-06,
1735
+ "loss": 0.2672,
1736
+ "step": 6975
1737
+ },
1738
+ {
1739
+ "epoch": 0.7,
1740
+ "learning_rate": 3.1621052631578953e-06,
1741
+ "loss": 0.2794,
1742
+ "step": 7000
1743
+ },
1744
+ {
1745
+ "epoch": 0.7,
1746
+ "eval_loss": 0.8547870516777039,
1747
+ "eval_runtime": 5789.6547,
1748
+ "eval_samples_per_second": 0.977,
1749
+ "eval_steps_per_second": 0.061,
1750
+ "eval_wer": 18.68400851643179,
1751
+ "step": 7000
1752
+ },
1753
+ {
1754
+ "epoch": 0.7,
1755
+ "learning_rate": 3.135789473684211e-06,
1756
+ "loss": 0.2769,
1757
+ "step": 7025
1758
+ },
1759
+ {
1760
+ "epoch": 0.7,
1761
+ "learning_rate": 3.1094736842105265e-06,
1762
+ "loss": 0.2691,
1763
+ "step": 7050
1764
+ },
1765
+ {
1766
+ "epoch": 0.71,
1767
+ "learning_rate": 3.0831578947368423e-06,
1768
+ "loss": 0.2285,
1769
+ "step": 7075
1770
+ },
1771
+ {
1772
+ "epoch": 0.71,
1773
+ "learning_rate": 3.056842105263158e-06,
1774
+ "loss": 0.218,
1775
+ "step": 7100
1776
+ },
1777
+ {
1778
+ "epoch": 0.71,
1779
+ "learning_rate": 3.0305263157894736e-06,
1780
+ "loss": 0.2302,
1781
+ "step": 7125
1782
+ },
1783
+ {
1784
+ "epoch": 0.71,
1785
+ "learning_rate": 3.0042105263157894e-06,
1786
+ "loss": 0.2808,
1787
+ "step": 7150
1788
+ },
1789
+ {
1790
+ "epoch": 0.72,
1791
+ "learning_rate": 2.9778947368421056e-06,
1792
+ "loss": 0.2532,
1793
+ "step": 7175
1794
+ },
1795
+ {
1796
+ "epoch": 0.72,
1797
+ "learning_rate": 2.9515789473684214e-06,
1798
+ "loss": 0.2556,
1799
+ "step": 7200
1800
+ },
1801
+ {
1802
+ "epoch": 0.72,
1803
+ "learning_rate": 2.9252631578947373e-06,
1804
+ "loss": 0.2608,
1805
+ "step": 7225
1806
+ },
1807
+ {
1808
+ "epoch": 0.72,
1809
+ "learning_rate": 2.8989473684210526e-06,
1810
+ "loss": 0.2286,
1811
+ "step": 7250
1812
+ },
1813
+ {
1814
+ "epoch": 0.73,
1815
+ "learning_rate": 2.8726315789473685e-06,
1816
+ "loss": 0.2501,
1817
+ "step": 7275
1818
+ },
1819
+ {
1820
+ "epoch": 0.73,
1821
+ "learning_rate": 2.8463157894736843e-06,
1822
+ "loss": 0.2358,
1823
+ "step": 7300
1824
+ },
1825
+ {
1826
+ "epoch": 0.73,
1827
+ "learning_rate": 2.82e-06,
1828
+ "loss": 0.2021,
1829
+ "step": 7325
1830
+ },
1831
+ {
1832
+ "epoch": 0.73,
1833
+ "learning_rate": 2.7936842105263163e-06,
1834
+ "loss": 0.2187,
1835
+ "step": 7350
1836
+ },
1837
+ {
1838
+ "epoch": 0.74,
1839
+ "learning_rate": 2.7673684210526317e-06,
1840
+ "loss": 0.2468,
1841
+ "step": 7375
1842
+ },
1843
+ {
1844
+ "epoch": 0.74,
1845
+ "learning_rate": 2.7410526315789476e-06,
1846
+ "loss": 0.2527,
1847
+ "step": 7400
1848
+ },
1849
+ {
1850
+ "epoch": 0.74,
1851
+ "learning_rate": 2.7147368421052634e-06,
1852
+ "loss": 0.2422,
1853
+ "step": 7425
1854
+ },
1855
+ {
1856
+ "epoch": 0.74,
1857
+ "learning_rate": 2.688421052631579e-06,
1858
+ "loss": 0.2499,
1859
+ "step": 7450
1860
+ },
1861
+ {
1862
+ "epoch": 0.75,
1863
+ "learning_rate": 2.6621052631578946e-06,
1864
+ "loss": 0.2997,
1865
+ "step": 7475
1866
+ },
1867
+ {
1868
+ "epoch": 0.75,
1869
+ "learning_rate": 2.635789473684211e-06,
1870
+ "loss": 0.2786,
1871
+ "step": 7500
1872
+ },
1873
+ {
1874
+ "epoch": 0.75,
1875
+ "learning_rate": 2.6094736842105267e-06,
1876
+ "loss": 0.2804,
1877
+ "step": 7525
1878
+ },
1879
+ {
1880
+ "epoch": 0.76,
1881
+ "learning_rate": 2.5831578947368425e-06,
1882
+ "loss": 0.2601,
1883
+ "step": 7550
1884
+ },
1885
+ {
1886
+ "epoch": 0.76,
1887
+ "learning_rate": 2.5568421052631583e-06,
1888
+ "loss": 0.2857,
1889
+ "step": 7575
1890
+ },
1891
+ {
1892
+ "epoch": 0.76,
1893
+ "learning_rate": 2.5305263157894737e-06,
1894
+ "loss": 0.2386,
1895
+ "step": 7600
1896
+ },
1897
+ {
1898
+ "epoch": 0.76,
1899
+ "learning_rate": 2.5042105263157895e-06,
1900
+ "loss": 0.2035,
1901
+ "step": 7625
1902
+ },
1903
+ {
1904
+ "epoch": 0.77,
1905
+ "learning_rate": 2.4778947368421053e-06,
1906
+ "loss": 0.2247,
1907
+ "step": 7650
1908
+ },
1909
+ {
1910
+ "epoch": 0.77,
1911
+ "learning_rate": 2.451578947368421e-06,
1912
+ "loss": 0.2178,
1913
+ "step": 7675
1914
+ },
1915
+ {
1916
+ "epoch": 0.77,
1917
+ "learning_rate": 2.425263157894737e-06,
1918
+ "loss": 0.2212,
1919
+ "step": 7700
1920
+ },
1921
+ {
1922
+ "epoch": 0.77,
1923
+ "learning_rate": 2.3989473684210528e-06,
1924
+ "loss": 0.2123,
1925
+ "step": 7725
1926
+ },
1927
+ {
1928
+ "epoch": 0.78,
1929
+ "learning_rate": 2.3726315789473686e-06,
1930
+ "loss": 0.2143,
1931
+ "step": 7750
1932
+ },
1933
+ {
1934
+ "epoch": 0.78,
1935
+ "learning_rate": 2.3463157894736844e-06,
1936
+ "loss": 0.2002,
1937
+ "step": 7775
1938
+ },
1939
+ {
1940
+ "epoch": 0.78,
1941
+ "learning_rate": 2.3200000000000002e-06,
1942
+ "loss": 0.2219,
1943
+ "step": 7800
1944
+ },
1945
+ {
1946
+ "epoch": 0.78,
1947
+ "learning_rate": 2.293684210526316e-06,
1948
+ "loss": 0.2597,
1949
+ "step": 7825
1950
+ },
1951
+ {
1952
+ "epoch": 0.79,
1953
+ "learning_rate": 2.267368421052632e-06,
1954
+ "loss": 0.2168,
1955
+ "step": 7850
1956
+ },
1957
+ {
1958
+ "epoch": 0.79,
1959
+ "learning_rate": 2.2410526315789473e-06,
1960
+ "loss": 0.2489,
1961
+ "step": 7875
1962
+ },
1963
+ {
1964
+ "epoch": 0.79,
1965
+ "learning_rate": 2.2147368421052635e-06,
1966
+ "loss": 0.1965,
1967
+ "step": 7900
1968
+ },
1969
+ {
1970
+ "epoch": 0.79,
1971
+ "learning_rate": 2.188421052631579e-06,
1972
+ "loss": 0.2449,
1973
+ "step": 7925
1974
+ },
1975
+ {
1976
+ "epoch": 0.8,
1977
+ "learning_rate": 2.1621052631578947e-06,
1978
+ "loss": 0.2284,
1979
+ "step": 7950
1980
+ },
1981
+ {
1982
+ "epoch": 0.8,
1983
+ "learning_rate": 2.135789473684211e-06,
1984
+ "loss": 0.2127,
1985
+ "step": 7975
1986
+ },
1987
+ {
1988
+ "epoch": 0.8,
1989
+ "learning_rate": 2.1094736842105264e-06,
1990
+ "loss": 0.2173,
1991
+ "step": 8000
1992
+ },
1993
+ {
1994
+ "epoch": 0.8,
1995
+ "eval_loss": 0.8542773723602295,
1996
+ "eval_runtime": 5521.1614,
1997
+ "eval_samples_per_second": 1.024,
1998
+ "eval_steps_per_second": 0.064,
1999
+ "eval_wer": 17.901887229697437,
2000
+ "step": 8000
2001
+ },
2002
+ {
2003
+ "epoch": 0.8,
2004
+ "learning_rate": 2.083157894736842e-06,
2005
+ "loss": 0.2428,
2006
+ "step": 8025
2007
+ },
2008
+ {
2009
+ "epoch": 0.81,
2010
+ "learning_rate": 2.056842105263158e-06,
2011
+ "loss": 0.2218,
2012
+ "step": 8050
2013
+ },
2014
+ {
2015
+ "epoch": 0.81,
2016
+ "learning_rate": 2.030526315789474e-06,
2017
+ "loss": 0.231,
2018
+ "step": 8075
2019
+ },
2020
+ {
2021
+ "epoch": 0.81,
2022
+ "learning_rate": 2.0042105263157896e-06,
2023
+ "loss": 0.2091,
2024
+ "step": 8100
2025
+ },
2026
+ {
2027
+ "epoch": 0.81,
2028
+ "learning_rate": 1.9778947368421055e-06,
2029
+ "loss": 0.2813,
2030
+ "step": 8125
2031
+ },
2032
+ {
2033
+ "epoch": 0.81,
2034
+ "learning_rate": 1.9515789473684213e-06,
2035
+ "loss": 0.2378,
2036
+ "step": 8150
2037
+ },
2038
+ {
2039
+ "epoch": 0.82,
2040
+ "learning_rate": 1.925263157894737e-06,
2041
+ "loss": 0.221,
2042
+ "step": 8175
2043
+ },
2044
+ {
2045
+ "epoch": 0.82,
2046
+ "learning_rate": 1.8989473684210527e-06,
2047
+ "loss": 0.2133,
2048
+ "step": 8200
2049
+ },
2050
+ {
2051
+ "epoch": 0.82,
2052
+ "learning_rate": 1.8726315789473687e-06,
2053
+ "loss": 0.2172,
2054
+ "step": 8225
2055
+ },
2056
+ {
2057
+ "epoch": 0.82,
2058
+ "learning_rate": 1.8463157894736843e-06,
2059
+ "loss": 0.2301,
2060
+ "step": 8250
2061
+ },
2062
+ {
2063
+ "epoch": 0.83,
2064
+ "learning_rate": 1.8200000000000002e-06,
2065
+ "loss": 0.2039,
2066
+ "step": 8275
2067
+ },
2068
+ {
2069
+ "epoch": 0.83,
2070
+ "learning_rate": 1.7936842105263158e-06,
2071
+ "loss": 0.2103,
2072
+ "step": 8300
2073
+ },
2074
+ {
2075
+ "epoch": 0.83,
2076
+ "learning_rate": 1.7673684210526318e-06,
2077
+ "loss": 0.2143,
2078
+ "step": 8325
2079
+ },
2080
+ {
2081
+ "epoch": 0.83,
2082
+ "learning_rate": 1.742105263157895e-06,
2083
+ "loss": 0.2607,
2084
+ "step": 8350
2085
+ },
2086
+ {
2087
+ "epoch": 0.84,
2088
+ "learning_rate": 1.7157894736842107e-06,
2089
+ "loss": 0.2785,
2090
+ "step": 8375
2091
+ },
2092
+ {
2093
+ "epoch": 0.84,
2094
+ "learning_rate": 1.6894736842105264e-06,
2095
+ "loss": 0.2195,
2096
+ "step": 8400
2097
+ },
2098
+ {
2099
+ "epoch": 0.84,
2100
+ "learning_rate": 1.6631578947368424e-06,
2101
+ "loss": 0.1813,
2102
+ "step": 8425
2103
+ },
2104
+ {
2105
+ "epoch": 0.84,
2106
+ "learning_rate": 1.636842105263158e-06,
2107
+ "loss": 0.2221,
2108
+ "step": 8450
2109
+ },
2110
+ {
2111
+ "epoch": 0.85,
2112
+ "learning_rate": 1.6105263157894738e-06,
2113
+ "loss": 0.1919,
2114
+ "step": 8475
2115
+ },
2116
+ {
2117
+ "epoch": 0.85,
2118
+ "learning_rate": 1.5842105263157894e-06,
2119
+ "loss": 0.2215,
2120
+ "step": 8500
2121
+ },
2122
+ {
2123
+ "epoch": 0.85,
2124
+ "learning_rate": 1.5578947368421054e-06,
2125
+ "loss": 0.2439,
2126
+ "step": 8525
2127
+ },
2128
+ {
2129
+ "epoch": 0.85,
2130
+ "learning_rate": 1.5315789473684213e-06,
2131
+ "loss": 0.2338,
2132
+ "step": 8550
2133
+ },
2134
+ {
2135
+ "epoch": 0.86,
2136
+ "learning_rate": 1.5052631578947369e-06,
2137
+ "loss": 0.2049,
2138
+ "step": 8575
2139
+ },
2140
+ {
2141
+ "epoch": 0.86,
2142
+ "learning_rate": 1.478947368421053e-06,
2143
+ "loss": 0.1977,
2144
+ "step": 8600
2145
+ },
2146
+ {
2147
+ "epoch": 0.86,
2148
+ "learning_rate": 1.4526315789473685e-06,
2149
+ "loss": 0.2129,
2150
+ "step": 8625
2151
+ },
2152
+ {
2153
+ "epoch": 0.86,
2154
+ "learning_rate": 1.4263157894736843e-06,
2155
+ "loss": 0.2319,
2156
+ "step": 8650
2157
+ },
2158
+ {
2159
+ "epoch": 0.87,
2160
+ "learning_rate": 1.4000000000000001e-06,
2161
+ "loss": 0.2209,
2162
+ "step": 8675
2163
+ },
2164
+ {
2165
+ "epoch": 0.87,
2166
+ "learning_rate": 1.373684210526316e-06,
2167
+ "loss": 0.23,
2168
+ "step": 8700
2169
+ },
2170
+ {
2171
+ "epoch": 0.87,
2172
+ "learning_rate": 1.3473684210526316e-06,
2173
+ "loss": 0.2418,
2174
+ "step": 8725
2175
+ },
2176
+ {
2177
+ "epoch": 0.88,
2178
+ "learning_rate": 1.3210526315789474e-06,
2179
+ "loss": 0.2623,
2180
+ "step": 8750
2181
+ },
2182
+ {
2183
+ "epoch": 0.88,
2184
+ "learning_rate": 1.2947368421052634e-06,
2185
+ "loss": 0.2338,
2186
+ "step": 8775
2187
+ },
2188
+ {
2189
+ "epoch": 0.88,
2190
+ "learning_rate": 1.268421052631579e-06,
2191
+ "loss": 0.2409,
2192
+ "step": 8800
2193
+ },
2194
+ {
2195
+ "epoch": 0.88,
2196
+ "learning_rate": 1.2421052631578948e-06,
2197
+ "loss": 0.2761,
2198
+ "step": 8825
2199
+ },
2200
+ {
2201
+ "epoch": 0.89,
2202
+ "learning_rate": 1.2157894736842107e-06,
2203
+ "loss": 0.2146,
2204
+ "step": 8850
2205
+ },
2206
+ {
2207
+ "epoch": 0.89,
2208
+ "learning_rate": 1.1894736842105265e-06,
2209
+ "loss": 0.2272,
2210
+ "step": 8875
2211
+ },
2212
+ {
2213
+ "epoch": 0.89,
2214
+ "learning_rate": 1.163157894736842e-06,
2215
+ "loss": 0.2265,
2216
+ "step": 8900
2217
+ },
2218
+ {
2219
+ "epoch": 0.89,
2220
+ "learning_rate": 1.136842105263158e-06,
2221
+ "loss": 0.2196,
2222
+ "step": 8925
2223
+ },
2224
+ {
2225
+ "epoch": 0.9,
2226
+ "learning_rate": 1.110526315789474e-06,
2227
+ "loss": 0.2143,
2228
+ "step": 8950
2229
+ },
2230
+ {
2231
+ "epoch": 0.9,
2232
+ "learning_rate": 1.0842105263157895e-06,
2233
+ "loss": 0.2129,
2234
+ "step": 8975
2235
+ },
2236
+ {
2237
+ "epoch": 0.9,
2238
+ "learning_rate": 1.0578947368421054e-06,
2239
+ "loss": 0.217,
2240
+ "step": 9000
2241
+ },
2242
+ {
2243
+ "epoch": 0.9,
2244
+ "eval_loss": 0.8518440127372742,
2245
+ "eval_runtime": 5194.633,
2246
+ "eval_samples_per_second": 1.089,
2247
+ "eval_steps_per_second": 0.068,
2248
+ "eval_wer": 16.3839925843315,
2249
+ "step": 9000
2250
+ },
2251
+ {
2252
+ "epoch": 0.9,
2253
+ "learning_rate": 1.0315789473684212e-06,
2254
+ "loss": 0.2194,
2255
+ "step": 9025
2256
+ },
2257
+ {
2258
+ "epoch": 0.91,
2259
+ "learning_rate": 1.005263157894737e-06,
2260
+ "loss": 0.1901,
2261
+ "step": 9050
2262
+ },
2263
+ {
2264
+ "epoch": 0.91,
2265
+ "learning_rate": 9.789473684210526e-07,
2266
+ "loss": 0.1898,
2267
+ "step": 9075
2268
+ },
2269
+ {
2270
+ "epoch": 0.91,
2271
+ "learning_rate": 9.526315789473685e-07,
2272
+ "loss": 0.224,
2273
+ "step": 9100
2274
+ },
2275
+ {
2276
+ "epoch": 0.91,
2277
+ "learning_rate": 9.263157894736844e-07,
2278
+ "loss": 0.2209,
2279
+ "step": 9125
2280
+ },
2281
+ {
2282
+ "epoch": 0.92,
2283
+ "learning_rate": 9.000000000000001e-07,
2284
+ "loss": 0.29,
2285
+ "step": 9150
2286
+ },
2287
+ {
2288
+ "epoch": 0.92,
2289
+ "learning_rate": 8.736842105263159e-07,
2290
+ "loss": 0.1973,
2291
+ "step": 9175
2292
+ },
2293
+ {
2294
+ "epoch": 0.92,
2295
+ "learning_rate": 8.473684210526316e-07,
2296
+ "loss": 0.182,
2297
+ "step": 9200
2298
+ },
2299
+ {
2300
+ "epoch": 0.92,
2301
+ "learning_rate": 8.210526315789474e-07,
2302
+ "loss": 0.2654,
2303
+ "step": 9225
2304
+ },
2305
+ {
2306
+ "epoch": 0.93,
2307
+ "learning_rate": 7.947368421052632e-07,
2308
+ "loss": 0.2342,
2309
+ "step": 9250
2310
+ },
2311
+ {
2312
+ "epoch": 0.93,
2313
+ "learning_rate": 7.684210526315789e-07,
2314
+ "loss": 0.2254,
2315
+ "step": 9275
2316
+ },
2317
+ {
2318
+ "epoch": 0.93,
2319
+ "learning_rate": 7.421052631578949e-07,
2320
+ "loss": 0.2037,
2321
+ "step": 9300
2322
+ },
2323
+ {
2324
+ "epoch": 0.93,
2325
+ "learning_rate": 7.157894736842106e-07,
2326
+ "loss": 0.2146,
2327
+ "step": 9325
2328
+ },
2329
+ {
2330
+ "epoch": 0.94,
2331
+ "learning_rate": 6.894736842105264e-07,
2332
+ "loss": 0.218,
2333
+ "step": 9350
2334
+ },
2335
+ {
2336
+ "epoch": 0.94,
2337
+ "learning_rate": 6.631578947368422e-07,
2338
+ "loss": 0.2405,
2339
+ "step": 9375
2340
+ },
2341
+ {
2342
+ "epoch": 0.94,
2343
+ "learning_rate": 6.368421052631579e-07,
2344
+ "loss": 0.2306,
2345
+ "step": 9400
2346
+ },
2347
+ {
2348
+ "epoch": 0.94,
2349
+ "learning_rate": 6.105263157894738e-07,
2350
+ "loss": 0.2291,
2351
+ "step": 9425
2352
+ },
2353
+ {
2354
+ "epoch": 0.94,
2355
+ "learning_rate": 5.842105263157896e-07,
2356
+ "loss": 0.2411,
2357
+ "step": 9450
2358
+ },
2359
+ {
2360
+ "epoch": 0.95,
2361
+ "learning_rate": 5.578947368421053e-07,
2362
+ "loss": 0.2503,
2363
+ "step": 9475
2364
+ },
2365
+ {
2366
+ "epoch": 0.95,
2367
+ "learning_rate": 5.315789473684211e-07,
2368
+ "loss": 0.1941,
2369
+ "step": 9500
2370
+ },
2371
+ {
2372
+ "epoch": 0.95,
2373
+ "learning_rate": 5.052631578947369e-07,
2374
+ "loss": 0.1873,
2375
+ "step": 9525
2376
+ },
2377
+ {
2378
+ "epoch": 0.95,
2379
+ "learning_rate": 4.789473684210526e-07,
2380
+ "loss": 0.2644,
2381
+ "step": 9550
2382
+ },
2383
+ {
2384
+ "epoch": 0.96,
2385
+ "learning_rate": 4.526315789473685e-07,
2386
+ "loss": 0.2611,
2387
+ "step": 9575
2388
+ },
2389
+ {
2390
+ "epoch": 0.96,
2391
+ "learning_rate": 4.2631578947368427e-07,
2392
+ "loss": 0.2303,
2393
+ "step": 9600
2394
+ },
2395
+ {
2396
+ "epoch": 0.96,
2397
+ "learning_rate": 4.0000000000000003e-07,
2398
+ "loss": 0.2403,
2399
+ "step": 9625
2400
+ },
2401
+ {
2402
+ "epoch": 0.96,
2403
+ "learning_rate": 3.736842105263158e-07,
2404
+ "loss": 0.2652,
2405
+ "step": 9650
2406
+ },
2407
+ {
2408
+ "epoch": 0.97,
2409
+ "learning_rate": 3.4736842105263157e-07,
2410
+ "loss": 0.2225,
2411
+ "step": 9675
2412
+ },
2413
+ {
2414
+ "epoch": 0.97,
2415
+ "learning_rate": 3.210526315789474e-07,
2416
+ "loss": 0.2045,
2417
+ "step": 9700
2418
+ },
2419
+ {
2420
+ "epoch": 0.97,
2421
+ "learning_rate": 2.9473684210526315e-07,
2422
+ "loss": 0.1908,
2423
+ "step": 9725
2424
+ },
2425
+ {
2426
+ "epoch": 0.97,
2427
+ "learning_rate": 2.6842105263157897e-07,
2428
+ "loss": 0.2981,
2429
+ "step": 9750
2430
+ },
2431
+ {
2432
+ "epoch": 0.98,
2433
+ "learning_rate": 2.4210526315789473e-07,
2434
+ "loss": 0.2498,
2435
+ "step": 9775
2436
+ },
2437
+ {
2438
+ "epoch": 0.98,
2439
+ "learning_rate": 2.1578947368421053e-07,
2440
+ "loss": 0.2385,
2441
+ "step": 9800
2442
+ },
2443
+ {
2444
+ "epoch": 0.98,
2445
+ "learning_rate": 1.8947368421052634e-07,
2446
+ "loss": 0.2189,
2447
+ "step": 9825
2448
+ },
2449
+ {
2450
+ "epoch": 0.98,
2451
+ "learning_rate": 1.631578947368421e-07,
2452
+ "loss": 0.2586,
2453
+ "step": 9850
2454
+ },
2455
+ {
2456
+ "epoch": 0.99,
2457
+ "learning_rate": 1.368421052631579e-07,
2458
+ "loss": 0.2696,
2459
+ "step": 9875
2460
+ },
2461
+ {
2462
+ "epoch": 0.99,
2463
+ "learning_rate": 1.1052631578947368e-07,
2464
+ "loss": 0.2125,
2465
+ "step": 9900
2466
+ },
2467
+ {
2468
+ "epoch": 0.99,
2469
+ "learning_rate": 8.421052631578947e-08,
2470
+ "loss": 0.2214,
2471
+ "step": 9925
2472
+ },
2473
+ {
2474
+ "epoch": 0.99,
2475
+ "learning_rate": 5.789473684210527e-08,
2476
+ "loss": 0.2398,
2477
+ "step": 9950
2478
+ },
2479
+ {
2480
+ "epoch": 1.0,
2481
+ "learning_rate": 3.157894736842106e-08,
2482
+ "loss": 0.2005,
2483
+ "step": 9975
2484
+ },
2485
+ {
2486
+ "epoch": 1.0,
2487
+ "learning_rate": 5.263157894736842e-09,
2488
+ "loss": 0.1718,
2489
+ "step": 10000
2490
+ },
2491
+ {
2492
+ "epoch": 1.0,
2493
+ "eval_loss": 0.8487818241119385,
2494
+ "eval_runtime": 5428.8291,
2495
+ "eval_samples_per_second": 1.042,
2496
+ "eval_steps_per_second": 0.065,
2497
+ "eval_wer": 16.588213142534364,
2498
+ "step": 10000
2499
+ },
2500
+ {
2501
+ "epoch": 1.0,
2502
+ "step": 10000,
2503
+ "total_flos": 3.265935704064e+20,
2504
+ "train_loss": 0.2656231692790985,
2505
+ "train_runtime": 120187.1267,
2506
+ "train_samples_per_second": 2.663,
2507
+ "train_steps_per_second": 0.083
2508
+ }
2509
+ ],
2510
+ "max_steps": 10000,
2511
+ "num_train_epochs": 9223372036854775807,
2512
+ "total_flos": 3.265935704064e+20,
2513
+ "trial_name": null,
2514
+ "trial_params": null
2515
+ }