gary109 commited on
Commit
077c222
1 Parent(s): d4576b6

End of training

Browse files
all_results.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 99.99,
3
+ "eval_loss": 0.6712434887886047,
4
+ "eval_runtime": 9.0751,
5
+ "eval_samples": 63,
6
+ "eval_samples_per_second": 6.942,
7
+ "eval_steps_per_second": 1.763,
8
+ "eval_wer": 0.3762125685364825,
9
+ "train_loss": 0.9362126126289367,
10
+ "train_runtime": 11890.6212,
11
+ "train_samples": 563,
12
+ "train_samples_per_second": 4.735,
13
+ "train_steps_per_second": 0.294
14
+ }
eval_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 99.99,
3
+ "eval_loss": 0.6712434887886047,
4
+ "eval_runtime": 9.0751,
5
+ "eval_samples": 63,
6
+ "eval_samples_per_second": 6.942,
7
+ "eval_steps_per_second": 1.763,
8
+ "eval_wer": 0.3762125685364825
9
+ }
runs/Nov24_07-39-17_ab5393081461/events.out.tfevents.1669288127.ab5393081461.9913.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43b729d7d2f77d2a6b492a27d9592334dc4dd39fa348ebd343241b6be211c349
3
+ size 358
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 99.99,
3
+ "train_loss": 0.9362126126289367,
4
+ "train_runtime": 11890.6212,
5
+ "train_samples": 563,
6
+ "train_samples_per_second": 4.735,
7
+ "train_steps_per_second": 0.294
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,3025 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.6712434887886047,
3
+ "best_model_checkpoint": "ai-light-dance_drums_ft_pretrain_wav2vec2-base-new-v6-1/checkpoint-2590",
4
+ "epoch": 99.99290780141844,
5
+ "global_step": 3500,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.28,
12
+ "learning_rate": 6.666666666666667e-05,
13
+ "loss": 55.7655,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.57,
18
+ "learning_rate": 0.0001866666666666667,
19
+ "loss": 58.6473,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.85,
24
+ "learning_rate": 0.00032,
25
+ "loss": 17.2739,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.99,
30
+ "eval_loss": 2.9041128158569336,
31
+ "eval_runtime": 8.8924,
32
+ "eval_samples_per_second": 7.085,
33
+ "eval_steps_per_second": 1.799,
34
+ "eval_wer": 1.0,
35
+ "step": 35
36
+ },
37
+ {
38
+ "epoch": 1.14,
39
+ "learning_rate": 0.00039953890489913545,
40
+ "loss": 2.8972,
41
+ "step": 40
42
+ },
43
+ {
44
+ "epoch": 1.43,
45
+ "learning_rate": 0.0003983861671469741,
46
+ "loss": 2.082,
47
+ "step": 50
48
+ },
49
+ {
50
+ "epoch": 1.71,
51
+ "learning_rate": 0.0003972334293948127,
52
+ "loss": 2.1557,
53
+ "step": 60
54
+ },
55
+ {
56
+ "epoch": 1.99,
57
+ "learning_rate": 0.00039608069164265134,
58
+ "loss": 1.8424,
59
+ "step": 70
60
+ },
61
+ {
62
+ "epoch": 1.99,
63
+ "eval_loss": 3.505493402481079,
64
+ "eval_runtime": 9.0402,
65
+ "eval_samples_per_second": 6.969,
66
+ "eval_steps_per_second": 1.77,
67
+ "eval_wer": 1.0,
68
+ "step": 70
69
+ },
70
+ {
71
+ "epoch": 2.28,
72
+ "learning_rate": 0.0003949279538904899,
73
+ "loss": 1.8014,
74
+ "step": 80
75
+ },
76
+ {
77
+ "epoch": 2.57,
78
+ "learning_rate": 0.00039377521613832855,
79
+ "loss": 1.6721,
80
+ "step": 90
81
+ },
82
+ {
83
+ "epoch": 2.85,
84
+ "learning_rate": 0.0003926224783861672,
85
+ "loss": 1.7092,
86
+ "step": 100
87
+ },
88
+ {
89
+ "epoch": 2.99,
90
+ "eval_loss": 2.0045998096466064,
91
+ "eval_runtime": 9.0436,
92
+ "eval_samples_per_second": 6.966,
93
+ "eval_steps_per_second": 1.769,
94
+ "eval_wer": 1.0,
95
+ "step": 105
96
+ },
97
+ {
98
+ "epoch": 3.14,
99
+ "learning_rate": 0.00039146974063400576,
100
+ "loss": 1.6035,
101
+ "step": 110
102
+ },
103
+ {
104
+ "epoch": 3.43,
105
+ "learning_rate": 0.0003903170028818444,
106
+ "loss": 1.6196,
107
+ "step": 120
108
+ },
109
+ {
110
+ "epoch": 3.71,
111
+ "learning_rate": 0.000389164265129683,
112
+ "loss": 1.587,
113
+ "step": 130
114
+ },
115
+ {
116
+ "epoch": 3.99,
117
+ "learning_rate": 0.00038801152737752165,
118
+ "loss": 1.5022,
119
+ "step": 140
120
+ },
121
+ {
122
+ "epoch": 3.99,
123
+ "eval_loss": 1.9661929607391357,
124
+ "eval_runtime": 9.0602,
125
+ "eval_samples_per_second": 6.954,
126
+ "eval_steps_per_second": 1.766,
127
+ "eval_wer": 0.9675242513707296,
128
+ "step": 140
129
+ },
130
+ {
131
+ "epoch": 4.28,
132
+ "learning_rate": 0.00038685878962536023,
133
+ "loss": 1.486,
134
+ "step": 150
135
+ },
136
+ {
137
+ "epoch": 4.57,
138
+ "learning_rate": 0.00038570605187319886,
139
+ "loss": 1.382,
140
+ "step": 160
141
+ },
142
+ {
143
+ "epoch": 4.85,
144
+ "learning_rate": 0.0003845533141210375,
145
+ "loss": 1.2964,
146
+ "step": 170
147
+ },
148
+ {
149
+ "epoch": 4.99,
150
+ "eval_loss": 1.9017366170883179,
151
+ "eval_runtime": 8.7445,
152
+ "eval_samples_per_second": 7.204,
153
+ "eval_steps_per_second": 1.83,
154
+ "eval_wer": 0.5504006748207507,
155
+ "step": 175
156
+ },
157
+ {
158
+ "epoch": 5.14,
159
+ "learning_rate": 0.0003834005763688761,
160
+ "loss": 1.3537,
161
+ "step": 180
162
+ },
163
+ {
164
+ "epoch": 5.43,
165
+ "learning_rate": 0.0003822478386167147,
166
+ "loss": 1.1033,
167
+ "step": 190
168
+ },
169
+ {
170
+ "epoch": 5.71,
171
+ "learning_rate": 0.00038109510086455333,
172
+ "loss": 1.2447,
173
+ "step": 200
174
+ },
175
+ {
176
+ "epoch": 5.99,
177
+ "learning_rate": 0.00037994236311239196,
178
+ "loss": 1.1235,
179
+ "step": 210
180
+ },
181
+ {
182
+ "epoch": 5.99,
183
+ "eval_loss": 1.9874823093414307,
184
+ "eval_runtime": 8.6787,
185
+ "eval_samples_per_second": 7.259,
186
+ "eval_steps_per_second": 1.844,
187
+ "eval_wer": 0.4643610291016449,
188
+ "step": 210
189
+ },
190
+ {
191
+ "epoch": 6.28,
192
+ "learning_rate": 0.00037878962536023054,
193
+ "loss": 1.1092,
194
+ "step": 220
195
+ },
196
+ {
197
+ "epoch": 6.57,
198
+ "learning_rate": 0.00037763688760806917,
199
+ "loss": 1.0707,
200
+ "step": 230
201
+ },
202
+ {
203
+ "epoch": 6.85,
204
+ "learning_rate": 0.0003764841498559078,
205
+ "loss": 1.1056,
206
+ "step": 240
207
+ },
208
+ {
209
+ "epoch": 6.99,
210
+ "eval_loss": 1.8791475296020508,
211
+ "eval_runtime": 8.7777,
212
+ "eval_samples_per_second": 7.177,
213
+ "eval_steps_per_second": 1.823,
214
+ "eval_wer": 0.47617039223956137,
215
+ "step": 245
216
+ },
217
+ {
218
+ "epoch": 7.14,
219
+ "learning_rate": 0.00037533141210374643,
220
+ "loss": 1.0885,
221
+ "step": 250
222
+ },
223
+ {
224
+ "epoch": 7.43,
225
+ "learning_rate": 0.000374178674351585,
226
+ "loss": 0.9235,
227
+ "step": 260
228
+ },
229
+ {
230
+ "epoch": 7.71,
231
+ "learning_rate": 0.00037302593659942364,
232
+ "loss": 1.0039,
233
+ "step": 270
234
+ },
235
+ {
236
+ "epoch": 7.99,
237
+ "learning_rate": 0.0003718731988472623,
238
+ "loss": 0.8907,
239
+ "step": 280
240
+ },
241
+ {
242
+ "epoch": 7.99,
243
+ "eval_loss": 1.4811413288116455,
244
+ "eval_runtime": 8.7992,
245
+ "eval_samples_per_second": 7.16,
246
+ "eval_steps_per_second": 1.818,
247
+ "eval_wer": 0.467313369886124,
248
+ "step": 280
249
+ },
250
+ {
251
+ "epoch": 8.28,
252
+ "learning_rate": 0.0003707204610951009,
253
+ "loss": 1.0557,
254
+ "step": 290
255
+ },
256
+ {
257
+ "epoch": 8.57,
258
+ "learning_rate": 0.0003695677233429395,
259
+ "loss": 0.9808,
260
+ "step": 300
261
+ },
262
+ {
263
+ "epoch": 8.85,
264
+ "learning_rate": 0.0003684149855907781,
265
+ "loss": 0.8605,
266
+ "step": 310
267
+ },
268
+ {
269
+ "epoch": 8.99,
270
+ "eval_loss": 1.9113553762435913,
271
+ "eval_runtime": 8.737,
272
+ "eval_samples_per_second": 7.211,
273
+ "eval_steps_per_second": 1.831,
274
+ "eval_wer": 0.4479122733024041,
275
+ "step": 315
276
+ },
277
+ {
278
+ "epoch": 9.14,
279
+ "learning_rate": 0.00036726224783861674,
280
+ "loss": 0.8891,
281
+ "step": 320
282
+ },
283
+ {
284
+ "epoch": 9.43,
285
+ "learning_rate": 0.0003661095100864553,
286
+ "loss": 0.9639,
287
+ "step": 330
288
+ },
289
+ {
290
+ "epoch": 9.71,
291
+ "learning_rate": 0.00036495677233429395,
292
+ "loss": 0.8316,
293
+ "step": 340
294
+ },
295
+ {
296
+ "epoch": 9.99,
297
+ "learning_rate": 0.0003638040345821326,
298
+ "loss": 0.8498,
299
+ "step": 350
300
+ },
301
+ {
302
+ "epoch": 9.99,
303
+ "eval_loss": 1.210748553276062,
304
+ "eval_runtime": 8.7375,
305
+ "eval_samples_per_second": 7.21,
306
+ "eval_steps_per_second": 1.831,
307
+ "eval_wer": 0.47279628848587096,
308
+ "step": 350
309
+ },
310
+ {
311
+ "epoch": 10.28,
312
+ "learning_rate": 0.0003626512968299712,
313
+ "loss": 0.9453,
314
+ "step": 360
315
+ },
316
+ {
317
+ "epoch": 10.57,
318
+ "learning_rate": 0.0003614985590778098,
319
+ "loss": 0.8499,
320
+ "step": 370
321
+ },
322
+ {
323
+ "epoch": 10.85,
324
+ "learning_rate": 0.0003603458213256484,
325
+ "loss": 0.7205,
326
+ "step": 380
327
+ },
328
+ {
329
+ "epoch": 10.99,
330
+ "eval_loss": 1.574381947517395,
331
+ "eval_runtime": 8.7195,
332
+ "eval_samples_per_second": 7.225,
333
+ "eval_steps_per_second": 1.835,
334
+ "eval_wer": 0.45255166596372837,
335
+ "step": 385
336
+ },
337
+ {
338
+ "epoch": 11.14,
339
+ "learning_rate": 0.00035919308357348705,
340
+ "loss": 1.0183,
341
+ "step": 390
342
+ },
343
+ {
344
+ "epoch": 11.43,
345
+ "learning_rate": 0.0003580403458213257,
346
+ "loss": 0.7708,
347
+ "step": 400
348
+ },
349
+ {
350
+ "epoch": 11.71,
351
+ "learning_rate": 0.00035688760806916426,
352
+ "loss": 0.8779,
353
+ "step": 410
354
+ },
355
+ {
356
+ "epoch": 11.99,
357
+ "learning_rate": 0.0003557348703170029,
358
+ "loss": 0.8417,
359
+ "step": 420
360
+ },
361
+ {
362
+ "epoch": 11.99,
363
+ "eval_loss": 1.468916654586792,
364
+ "eval_runtime": 8.7774,
365
+ "eval_samples_per_second": 7.178,
366
+ "eval_steps_per_second": 1.823,
367
+ "eval_wer": 0.453395191902151,
368
+ "step": 420
369
+ },
370
+ {
371
+ "epoch": 12.28,
372
+ "learning_rate": 0.0003545821325648415,
373
+ "loss": 0.8197,
374
+ "step": 430
375
+ },
376
+ {
377
+ "epoch": 12.57,
378
+ "learning_rate": 0.0003534293948126801,
379
+ "loss": 0.7148,
380
+ "step": 440
381
+ },
382
+ {
383
+ "epoch": 12.85,
384
+ "learning_rate": 0.00035227665706051873,
385
+ "loss": 0.7734,
386
+ "step": 450
387
+ },
388
+ {
389
+ "epoch": 12.99,
390
+ "eval_loss": 1.3531099557876587,
391
+ "eval_runtime": 8.7009,
392
+ "eval_samples_per_second": 7.241,
393
+ "eval_steps_per_second": 1.839,
394
+ "eval_wer": 0.4550822437789962,
395
+ "step": 455
396
+ },
397
+ {
398
+ "epoch": 13.14,
399
+ "learning_rate": 0.00035112391930835736,
400
+ "loss": 0.7908,
401
+ "step": 460
402
+ },
403
+ {
404
+ "epoch": 13.43,
405
+ "learning_rate": 0.000349971181556196,
406
+ "loss": 0.7157,
407
+ "step": 470
408
+ },
409
+ {
410
+ "epoch": 13.71,
411
+ "learning_rate": 0.00034881844380403457,
412
+ "loss": 0.7873,
413
+ "step": 480
414
+ },
415
+ {
416
+ "epoch": 13.99,
417
+ "learning_rate": 0.00034766570605187326,
418
+ "loss": 0.7762,
419
+ "step": 490
420
+ },
421
+ {
422
+ "epoch": 13.99,
423
+ "eval_loss": 1.2924444675445557,
424
+ "eval_runtime": 8.7984,
425
+ "eval_samples_per_second": 7.16,
426
+ "eval_steps_per_second": 1.819,
427
+ "eval_wer": 0.4664698439477014,
428
+ "step": 490
429
+ },
430
+ {
431
+ "epoch": 14.28,
432
+ "learning_rate": 0.00034651296829971183,
433
+ "loss": 0.8195,
434
+ "step": 500
435
+ },
436
+ {
437
+ "epoch": 14.57,
438
+ "learning_rate": 0.00034536023054755047,
439
+ "loss": 0.7457,
440
+ "step": 510
441
+ },
442
+ {
443
+ "epoch": 14.85,
444
+ "learning_rate": 0.00034420749279538904,
445
+ "loss": 0.6812,
446
+ "step": 520
447
+ },
448
+ {
449
+ "epoch": 14.99,
450
+ "eval_loss": 1.0827070474624634,
451
+ "eval_runtime": 8.7923,
452
+ "eval_samples_per_second": 7.165,
453
+ "eval_steps_per_second": 1.82,
454
+ "eval_wer": 0.40995360607338677,
455
+ "step": 525
456
+ },
457
+ {
458
+ "epoch": 15.14,
459
+ "learning_rate": 0.0003430547550432277,
460
+ "loss": 0.7591,
461
+ "step": 530
462
+ },
463
+ {
464
+ "epoch": 15.43,
465
+ "learning_rate": 0.0003419020172910663,
466
+ "loss": 0.7642,
467
+ "step": 540
468
+ },
469
+ {
470
+ "epoch": 15.71,
471
+ "learning_rate": 0.0003407492795389049,
472
+ "loss": 1.27,
473
+ "step": 550
474
+ },
475
+ {
476
+ "epoch": 15.99,
477
+ "learning_rate": 0.0003395965417867435,
478
+ "loss": 0.7245,
479
+ "step": 560
480
+ },
481
+ {
482
+ "epoch": 15.99,
483
+ "eval_loss": 1.4069534540176392,
484
+ "eval_runtime": 8.8121,
485
+ "eval_samples_per_second": 7.149,
486
+ "eval_steps_per_second": 1.816,
487
+ "eval_wer": 0.43525938422606497,
488
+ "step": 560
489
+ },
490
+ {
491
+ "epoch": 16.28,
492
+ "learning_rate": 0.00033844380403458214,
493
+ "loss": 0.7721,
494
+ "step": 570
495
+ },
496
+ {
497
+ "epoch": 16.57,
498
+ "learning_rate": 0.0003372910662824208,
499
+ "loss": 0.6945,
500
+ "step": 580
501
+ },
502
+ {
503
+ "epoch": 16.85,
504
+ "learning_rate": 0.00033613832853025935,
505
+ "loss": 0.6508,
506
+ "step": 590
507
+ },
508
+ {
509
+ "epoch": 16.99,
510
+ "eval_loss": 1.052022099494934,
511
+ "eval_runtime": 8.7064,
512
+ "eval_samples_per_second": 7.236,
513
+ "eval_steps_per_second": 1.838,
514
+ "eval_wer": 0.4086883171657528,
515
+ "step": 595
516
+ },
517
+ {
518
+ "epoch": 17.14,
519
+ "learning_rate": 0.00033498559077809804,
520
+ "loss": 0.735,
521
+ "step": 600
522
+ },
523
+ {
524
+ "epoch": 17.43,
525
+ "learning_rate": 0.0003338328530259366,
526
+ "loss": 0.6526,
527
+ "step": 610
528
+ },
529
+ {
530
+ "epoch": 17.71,
531
+ "learning_rate": 0.00033268011527377525,
532
+ "loss": 0.6955,
533
+ "step": 620
534
+ },
535
+ {
536
+ "epoch": 17.99,
537
+ "learning_rate": 0.0003315273775216138,
538
+ "loss": 0.7144,
539
+ "step": 630
540
+ },
541
+ {
542
+ "epoch": 17.99,
543
+ "eval_loss": 1.0728662014007568,
544
+ "eval_runtime": 8.8004,
545
+ "eval_samples_per_second": 7.159,
546
+ "eval_steps_per_second": 1.818,
547
+ "eval_wer": 0.42091944327288067,
548
+ "step": 630
549
+ },
550
+ {
551
+ "epoch": 18.28,
552
+ "learning_rate": 0.0003303746397694525,
553
+ "loss": 0.6136,
554
+ "step": 640
555
+ },
556
+ {
557
+ "epoch": 18.57,
558
+ "learning_rate": 0.0003292219020172911,
559
+ "loss": 0.6545,
560
+ "step": 650
561
+ },
562
+ {
563
+ "epoch": 18.85,
564
+ "learning_rate": 0.00032806916426512966,
565
+ "loss": 0.6566,
566
+ "step": 660
567
+ },
568
+ {
569
+ "epoch": 18.99,
570
+ "eval_loss": 1.1671946048736572,
571
+ "eval_runtime": 8.7903,
572
+ "eval_samples_per_second": 7.167,
573
+ "eval_steps_per_second": 1.82,
574
+ "eval_wer": 0.4053142134120624,
575
+ "step": 665
576
+ },
577
+ {
578
+ "epoch": 19.14,
579
+ "learning_rate": 0.00032691642651296835,
580
+ "loss": 0.7112,
581
+ "step": 670
582
+ },
583
+ {
584
+ "epoch": 19.43,
585
+ "learning_rate": 0.0003257636887608069,
586
+ "loss": 0.6853,
587
+ "step": 680
588
+ },
589
+ {
590
+ "epoch": 19.71,
591
+ "learning_rate": 0.00032461095100864556,
592
+ "loss": 0.7222,
593
+ "step": 690
594
+ },
595
+ {
596
+ "epoch": 19.99,
597
+ "learning_rate": 0.00032345821325648413,
598
+ "loss": 0.5802,
599
+ "step": 700
600
+ },
601
+ {
602
+ "epoch": 19.99,
603
+ "eval_loss": 1.0128839015960693,
604
+ "eval_runtime": 8.823,
605
+ "eval_samples_per_second": 7.14,
606
+ "eval_steps_per_second": 1.813,
607
+ "eval_wer": 0.4015183466891607,
608
+ "step": 700
609
+ },
610
+ {
611
+ "epoch": 20.28,
612
+ "learning_rate": 0.0003223054755043228,
613
+ "loss": 0.6989,
614
+ "step": 710
615
+ },
616
+ {
617
+ "epoch": 20.57,
618
+ "learning_rate": 0.0003211527377521614,
619
+ "loss": 0.6103,
620
+ "step": 720
621
+ },
622
+ {
623
+ "epoch": 20.85,
624
+ "learning_rate": 0.00032,
625
+ "loss": 0.5924,
626
+ "step": 730
627
+ },
628
+ {
629
+ "epoch": 20.99,
630
+ "eval_loss": 1.0761842727661133,
631
+ "eval_runtime": 8.6827,
632
+ "eval_samples_per_second": 7.256,
633
+ "eval_steps_per_second": 1.843,
634
+ "eval_wer": 0.40067482075073807,
635
+ "step": 735
636
+ },
637
+ {
638
+ "epoch": 21.14,
639
+ "learning_rate": 0.0003188472622478386,
640
+ "loss": 0.641,
641
+ "step": 740
642
+ },
643
+ {
644
+ "epoch": 21.43,
645
+ "learning_rate": 0.0003176945244956773,
646
+ "loss": 0.7465,
647
+ "step": 750
648
+ },
649
+ {
650
+ "epoch": 21.71,
651
+ "learning_rate": 0.00031654178674351587,
652
+ "loss": 0.6215,
653
+ "step": 760
654
+ },
655
+ {
656
+ "epoch": 21.99,
657
+ "learning_rate": 0.00031538904899135444,
658
+ "loss": 0.7051,
659
+ "step": 770
660
+ },
661
+ {
662
+ "epoch": 21.99,
663
+ "eval_loss": 1.025338053703308,
664
+ "eval_runtime": 8.7155,
665
+ "eval_samples_per_second": 7.229,
666
+ "eval_steps_per_second": 1.836,
667
+ "eval_wer": 0.4027836355967946,
668
+ "step": 770
669
+ },
670
+ {
671
+ "epoch": 22.28,
672
+ "learning_rate": 0.00031423631123919313,
673
+ "loss": 0.6773,
674
+ "step": 780
675
+ },
676
+ {
677
+ "epoch": 22.57,
678
+ "learning_rate": 0.0003130835734870317,
679
+ "loss": 0.6137,
680
+ "step": 790
681
+ },
682
+ {
683
+ "epoch": 22.85,
684
+ "learning_rate": 0.00031193083573487034,
685
+ "loss": 0.5669,
686
+ "step": 800
687
+ },
688
+ {
689
+ "epoch": 22.99,
690
+ "eval_loss": 1.0526078939437866,
691
+ "eval_runtime": 8.675,
692
+ "eval_samples_per_second": 7.262,
693
+ "eval_steps_per_second": 1.844,
694
+ "eval_wer": 0.41881062842682415,
695
+ "step": 805
696
+ },
697
+ {
698
+ "epoch": 23.14,
699
+ "learning_rate": 0.0003107780979827089,
700
+ "loss": 0.5976,
701
+ "step": 810
702
+ },
703
+ {
704
+ "epoch": 23.43,
705
+ "learning_rate": 0.0003096253602305476,
706
+ "loss": 0.6745,
707
+ "step": 820
708
+ },
709
+ {
710
+ "epoch": 23.71,
711
+ "learning_rate": 0.0003084726224783862,
712
+ "loss": 0.7289,
713
+ "step": 830
714
+ },
715
+ {
716
+ "epoch": 23.99,
717
+ "learning_rate": 0.0003073198847262248,
718
+ "loss": 0.6209,
719
+ "step": 840
720
+ },
721
+ {
722
+ "epoch": 23.99,
723
+ "eval_loss": 1.0177326202392578,
724
+ "eval_runtime": 8.832,
725
+ "eval_samples_per_second": 7.133,
726
+ "eval_steps_per_second": 1.812,
727
+ "eval_wer": 0.4213412062420919,
728
+ "step": 840
729
+ },
730
+ {
731
+ "epoch": 24.28,
732
+ "learning_rate": 0.00030616714697406344,
733
+ "loss": 0.6383,
734
+ "step": 850
735
+ },
736
+ {
737
+ "epoch": 24.57,
738
+ "learning_rate": 0.00030501440922190207,
739
+ "loss": 0.5804,
740
+ "step": 860
741
+ },
742
+ {
743
+ "epoch": 24.85,
744
+ "learning_rate": 0.00030386167146974065,
745
+ "loss": 0.635,
746
+ "step": 870
747
+ },
748
+ {
749
+ "epoch": 24.99,
750
+ "eval_loss": 0.929938554763794,
751
+ "eval_runtime": 8.8255,
752
+ "eval_samples_per_second": 7.138,
753
+ "eval_steps_per_second": 1.813,
754
+ "eval_wer": 0.401940109658372,
755
+ "step": 875
756
+ },
757
+ {
758
+ "epoch": 25.14,
759
+ "learning_rate": 0.0003027089337175792,
760
+ "loss": 0.74,
761
+ "step": 880
762
+ },
763
+ {
764
+ "epoch": 25.43,
765
+ "learning_rate": 0.0003015561959654179,
766
+ "loss": 0.6235,
767
+ "step": 890
768
+ },
769
+ {
770
+ "epoch": 25.71,
771
+ "learning_rate": 0.0003004034582132565,
772
+ "loss": 0.6219,
773
+ "step": 900
774
+ },
775
+ {
776
+ "epoch": 25.99,
777
+ "learning_rate": 0.0002992507204610951,
778
+ "loss": 0.5914,
779
+ "step": 910
780
+ },
781
+ {
782
+ "epoch": 25.99,
783
+ "eval_loss": 1.0058447122573853,
784
+ "eval_runtime": 8.6702,
785
+ "eval_samples_per_second": 7.266,
786
+ "eval_steps_per_second": 1.845,
787
+ "eval_wer": 0.4141712357654998,
788
+ "step": 910
789
+ },
790
+ {
791
+ "epoch": 26.28,
792
+ "learning_rate": 0.00029809798270893375,
793
+ "loss": 0.6127,
794
+ "step": 920
795
+ },
796
+ {
797
+ "epoch": 26.57,
798
+ "learning_rate": 0.0002969452449567724,
799
+ "loss": 0.5335,
800
+ "step": 930
801
+ },
802
+ {
803
+ "epoch": 26.85,
804
+ "learning_rate": 0.00029579250720461096,
805
+ "loss": 0.5983,
806
+ "step": 940
807
+ },
808
+ {
809
+ "epoch": 26.99,
810
+ "eval_loss": 0.9719786643981934,
811
+ "eval_runtime": 8.6673,
812
+ "eval_samples_per_second": 7.269,
813
+ "eval_steps_per_second": 1.846,
814
+ "eval_wer": 0.4141712357654998,
815
+ "step": 945
816
+ },
817
+ {
818
+ "epoch": 27.14,
819
+ "learning_rate": 0.0002946397694524496,
820
+ "loss": 0.559,
821
+ "step": 950
822
+ },
823
+ {
824
+ "epoch": 27.43,
825
+ "learning_rate": 0.0002934870317002882,
826
+ "loss": 0.6208,
827
+ "step": 960
828
+ },
829
+ {
830
+ "epoch": 27.71,
831
+ "learning_rate": 0.00029233429394812685,
832
+ "loss": 0.5678,
833
+ "step": 970
834
+ },
835
+ {
836
+ "epoch": 27.99,
837
+ "learning_rate": 0.0002911815561959654,
838
+ "loss": 0.5631,
839
+ "step": 980
840
+ },
841
+ {
842
+ "epoch": 27.99,
843
+ "eval_loss": 0.8982506990432739,
844
+ "eval_runtime": 8.7362,
845
+ "eval_samples_per_second": 7.211,
846
+ "eval_steps_per_second": 1.831,
847
+ "eval_wer": 0.4027836355967946,
848
+ "step": 980
849
+ },
850
+ {
851
+ "epoch": 28.28,
852
+ "learning_rate": 0.000290028818443804,
853
+ "loss": 0.5503,
854
+ "step": 990
855
+ },
856
+ {
857
+ "epoch": 28.57,
858
+ "learning_rate": 0.0002888760806916427,
859
+ "loss": 0.5238,
860
+ "step": 1000
861
+ },
862
+ {
863
+ "epoch": 28.85,
864
+ "learning_rate": 0.00028772334293948127,
865
+ "loss": 0.552,
866
+ "step": 1010
867
+ },
868
+ {
869
+ "epoch": 28.99,
870
+ "eval_loss": 0.914803683757782,
871
+ "eval_runtime": 8.7166,
872
+ "eval_samples_per_second": 7.228,
873
+ "eval_steps_per_second": 1.836,
874
+ "eval_wer": 0.41838886545761284,
875
+ "step": 1015
876
+ },
877
+ {
878
+ "epoch": 29.14,
879
+ "learning_rate": 0.0002865706051873199,
880
+ "loss": 0.5816,
881
+ "step": 1020
882
+ },
883
+ {
884
+ "epoch": 29.43,
885
+ "learning_rate": 0.00028541786743515853,
886
+ "loss": 0.5295,
887
+ "step": 1030
888
+ },
889
+ {
890
+ "epoch": 29.71,
891
+ "learning_rate": 0.00028426512968299716,
892
+ "loss": 0.5742,
893
+ "step": 1040
894
+ },
895
+ {
896
+ "epoch": 29.99,
897
+ "learning_rate": 0.00028311239193083574,
898
+ "loss": 0.5213,
899
+ "step": 1050
900
+ },
901
+ {
902
+ "epoch": 29.99,
903
+ "eval_loss": 1.0816500186920166,
904
+ "eval_runtime": 8.9248,
905
+ "eval_samples_per_second": 7.059,
906
+ "eval_steps_per_second": 1.793,
907
+ "eval_wer": 0.4141712357654998,
908
+ "step": 1050
909
+ },
910
+ {
911
+ "epoch": 30.28,
912
+ "learning_rate": 0.00028195965417867437,
913
+ "loss": 0.681,
914
+ "step": 1060
915
+ },
916
+ {
917
+ "epoch": 30.57,
918
+ "learning_rate": 0.000280806916426513,
919
+ "loss": 0.5139,
920
+ "step": 1070
921
+ },
922
+ {
923
+ "epoch": 30.85,
924
+ "learning_rate": 0.00027965417867435163,
925
+ "loss": 0.5387,
926
+ "step": 1080
927
+ },
928
+ {
929
+ "epoch": 30.99,
930
+ "eval_loss": 0.9432059526443481,
931
+ "eval_runtime": 8.8063,
932
+ "eval_samples_per_second": 7.154,
933
+ "eval_steps_per_second": 1.817,
934
+ "eval_wer": 0.41881062842682415,
935
+ "step": 1085
936
+ },
937
+ {
938
+ "epoch": 31.14,
939
+ "learning_rate": 0.0002785014409221902,
940
+ "loss": 0.5541,
941
+ "step": 1090
942
+ },
943
+ {
944
+ "epoch": 31.43,
945
+ "learning_rate": 0.00027734870317002884,
946
+ "loss": 0.6006,
947
+ "step": 1100
948
+ },
949
+ {
950
+ "epoch": 31.71,
951
+ "learning_rate": 0.00027619596541786747,
952
+ "loss": 0.6104,
953
+ "step": 1110
954
+ },
955
+ {
956
+ "epoch": 31.99,
957
+ "learning_rate": 0.00027504322766570605,
958
+ "loss": 0.5276,
959
+ "step": 1120
960
+ },
961
+ {
962
+ "epoch": 31.99,
963
+ "eval_loss": 1.1206682920455933,
964
+ "eval_runtime": 8.7507,
965
+ "eval_samples_per_second": 7.199,
966
+ "eval_steps_per_second": 1.828,
967
+ "eval_wer": 0.40067482075073807,
968
+ "step": 1120
969
+ },
970
+ {
971
+ "epoch": 32.28,
972
+ "learning_rate": 0.0002738904899135447,
973
+ "loss": 0.511,
974
+ "step": 1130
975
+ },
976
+ {
977
+ "epoch": 32.57,
978
+ "learning_rate": 0.0002727377521613833,
979
+ "loss": 0.4948,
980
+ "step": 1140
981
+ },
982
+ {
983
+ "epoch": 32.85,
984
+ "learning_rate": 0.00027158501440922194,
985
+ "loss": 0.5778,
986
+ "step": 1150
987
+ },
988
+ {
989
+ "epoch": 32.99,
990
+ "eval_loss": 0.9253532886505127,
991
+ "eval_runtime": 8.8188,
992
+ "eval_samples_per_second": 7.144,
993
+ "eval_steps_per_second": 1.814,
994
+ "eval_wer": 0.4150147617039224,
995
+ "step": 1155
996
+ },
997
+ {
998
+ "epoch": 33.14,
999
+ "learning_rate": 0.0002704322766570605,
1000
+ "loss": 0.551,
1001
+ "step": 1160
1002
+ },
1003
+ {
1004
+ "epoch": 33.43,
1005
+ "learning_rate": 0.00026927953890489915,
1006
+ "loss": 0.5115,
1007
+ "step": 1170
1008
+ },
1009
+ {
1010
+ "epoch": 33.71,
1011
+ "learning_rate": 0.0002681268011527378,
1012
+ "loss": 0.6795,
1013
+ "step": 1180
1014
+ },
1015
+ {
1016
+ "epoch": 33.99,
1017
+ "learning_rate": 0.0002669740634005764,
1018
+ "loss": 0.5001,
1019
+ "step": 1190
1020
+ },
1021
+ {
1022
+ "epoch": 33.99,
1023
+ "eval_loss": 1.0393145084381104,
1024
+ "eval_runtime": 8.783,
1025
+ "eval_samples_per_second": 7.173,
1026
+ "eval_steps_per_second": 1.822,
1027
+ "eval_wer": 0.4192323913960354,
1028
+ "step": 1190
1029
+ },
1030
+ {
1031
+ "epoch": 34.28,
1032
+ "learning_rate": 0.000265821325648415,
1033
+ "loss": 0.5146,
1034
+ "step": 1200
1035
+ },
1036
+ {
1037
+ "epoch": 34.57,
1038
+ "learning_rate": 0.0002646685878962536,
1039
+ "loss": 0.5165,
1040
+ "step": 1210
1041
+ },
1042
+ {
1043
+ "epoch": 34.85,
1044
+ "learning_rate": 0.00026351585014409225,
1045
+ "loss": 0.5329,
1046
+ "step": 1220
1047
+ },
1048
+ {
1049
+ "epoch": 34.99,
1050
+ "eval_loss": 0.9108927249908447,
1051
+ "eval_runtime": 8.8479,
1052
+ "eval_samples_per_second": 7.12,
1053
+ "eval_steps_per_second": 1.808,
1054
+ "eval_wer": 0.39645719105862504,
1055
+ "step": 1225
1056
+ },
1057
+ {
1058
+ "epoch": 35.14,
1059
+ "learning_rate": 0.0002623631123919308,
1060
+ "loss": 0.5703,
1061
+ "step": 1230
1062
+ },
1063
+ {
1064
+ "epoch": 35.43,
1065
+ "learning_rate": 0.00026121037463976946,
1066
+ "loss": 0.5296,
1067
+ "step": 1240
1068
+ },
1069
+ {
1070
+ "epoch": 35.71,
1071
+ "learning_rate": 0.0002600576368876081,
1072
+ "loss": 0.5212,
1073
+ "step": 1250
1074
+ },
1075
+ {
1076
+ "epoch": 35.99,
1077
+ "learning_rate": 0.0002589048991354467,
1078
+ "loss": 0.5168,
1079
+ "step": 1260
1080
+ },
1081
+ {
1082
+ "epoch": 35.99,
1083
+ "eval_loss": 0.8983320593833923,
1084
+ "eval_runtime": 8.8493,
1085
+ "eval_samples_per_second": 7.119,
1086
+ "eval_steps_per_second": 1.808,
1087
+ "eval_wer": 0.429776465626318,
1088
+ "step": 1260
1089
+ },
1090
+ {
1091
+ "epoch": 36.28,
1092
+ "learning_rate": 0.0002577521613832853,
1093
+ "loss": 0.528,
1094
+ "step": 1270
1095
+ },
1096
+ {
1097
+ "epoch": 36.57,
1098
+ "learning_rate": 0.00025659942363112393,
1099
+ "loss": 0.4487,
1100
+ "step": 1280
1101
+ },
1102
+ {
1103
+ "epoch": 36.85,
1104
+ "learning_rate": 0.00025544668587896256,
1105
+ "loss": 0.4918,
1106
+ "step": 1290
1107
+ },
1108
+ {
1109
+ "epoch": 36.99,
1110
+ "eval_loss": 0.8411929607391357,
1111
+ "eval_runtime": 8.8153,
1112
+ "eval_samples_per_second": 7.147,
1113
+ "eval_steps_per_second": 1.815,
1114
+ "eval_wer": 0.4086883171657528,
1115
+ "step": 1295
1116
+ },
1117
+ {
1118
+ "epoch": 37.14,
1119
+ "learning_rate": 0.0002542939481268012,
1120
+ "loss": 0.5363,
1121
+ "step": 1300
1122
+ },
1123
+ {
1124
+ "epoch": 37.43,
1125
+ "learning_rate": 0.00025314121037463977,
1126
+ "loss": 0.4954,
1127
+ "step": 1310
1128
+ },
1129
+ {
1130
+ "epoch": 37.71,
1131
+ "learning_rate": 0.0002519884726224784,
1132
+ "loss": 0.4878,
1133
+ "step": 1320
1134
+ },
1135
+ {
1136
+ "epoch": 37.99,
1137
+ "learning_rate": 0.00025083573487031703,
1138
+ "loss": 0.5651,
1139
+ "step": 1330
1140
+ },
1141
+ {
1142
+ "epoch": 37.99,
1143
+ "eval_loss": 0.8560371398925781,
1144
+ "eval_runtime": 9.0193,
1145
+ "eval_samples_per_second": 6.985,
1146
+ "eval_steps_per_second": 1.774,
1147
+ "eval_wer": 0.42176296921130324,
1148
+ "step": 1330
1149
+ },
1150
+ {
1151
+ "epoch": 38.28,
1152
+ "learning_rate": 0.0002496829971181556,
1153
+ "loss": 0.5888,
1154
+ "step": 1340
1155
+ },
1156
+ {
1157
+ "epoch": 38.57,
1158
+ "learning_rate": 0.00024853025936599424,
1159
+ "loss": 0.485,
1160
+ "step": 1350
1161
+ },
1162
+ {
1163
+ "epoch": 38.85,
1164
+ "learning_rate": 0.00024737752161383287,
1165
+ "loss": 0.438,
1166
+ "step": 1360
1167
+ },
1168
+ {
1169
+ "epoch": 38.99,
1170
+ "eval_loss": 0.8555735349655151,
1171
+ "eval_runtime": 8.8578,
1172
+ "eval_samples_per_second": 7.112,
1173
+ "eval_steps_per_second": 1.806,
1174
+ "eval_wer": 0.4171235765499789,
1175
+ "step": 1365
1176
+ },
1177
+ {
1178
+ "epoch": 39.14,
1179
+ "learning_rate": 0.0002462247838616715,
1180
+ "loss": 0.5556,
1181
+ "step": 1370
1182
+ },
1183
+ {
1184
+ "epoch": 39.43,
1185
+ "learning_rate": 0.0002450720461095101,
1186
+ "loss": 0.4795,
1187
+ "step": 1380
1188
+ },
1189
+ {
1190
+ "epoch": 39.71,
1191
+ "learning_rate": 0.00024391930835734874,
1192
+ "loss": 0.4785,
1193
+ "step": 1390
1194
+ },
1195
+ {
1196
+ "epoch": 39.99,
1197
+ "learning_rate": 0.00024276657060518734,
1198
+ "loss": 0.4808,
1199
+ "step": 1400
1200
+ },
1201
+ {
1202
+ "epoch": 39.99,
1203
+ "eval_loss": 0.8319878578186035,
1204
+ "eval_runtime": 8.7692,
1205
+ "eval_samples_per_second": 7.184,
1206
+ "eval_steps_per_second": 1.825,
1207
+ "eval_wer": 0.4175453395191902,
1208
+ "step": 1400
1209
+ },
1210
+ {
1211
+ "epoch": 40.28,
1212
+ "learning_rate": 0.00024161383285302597,
1213
+ "loss": 0.5304,
1214
+ "step": 1410
1215
+ },
1216
+ {
1217
+ "epoch": 40.57,
1218
+ "learning_rate": 0.00024046109510086458,
1219
+ "loss": 0.4896,
1220
+ "step": 1420
1221
+ },
1222
+ {
1223
+ "epoch": 40.85,
1224
+ "learning_rate": 0.00023930835734870315,
1225
+ "loss": 0.5372,
1226
+ "step": 1430
1227
+ },
1228
+ {
1229
+ "epoch": 40.99,
1230
+ "eval_loss": 0.9745492339134216,
1231
+ "eval_runtime": 8.7927,
1232
+ "eval_samples_per_second": 7.165,
1233
+ "eval_steps_per_second": 1.82,
1234
+ "eval_wer": 0.39561366512020246,
1235
+ "step": 1435
1236
+ },
1237
+ {
1238
+ "epoch": 41.14,
1239
+ "learning_rate": 0.0002381556195965418,
1240
+ "loss": 0.4811,
1241
+ "step": 1440
1242
+ },
1243
+ {
1244
+ "epoch": 41.43,
1245
+ "learning_rate": 0.0002370028818443804,
1246
+ "loss": 0.4714,
1247
+ "step": 1450
1248
+ },
1249
+ {
1250
+ "epoch": 41.71,
1251
+ "learning_rate": 0.00023585014409221905,
1252
+ "loss": 0.5127,
1253
+ "step": 1460
1254
+ },
1255
+ {
1256
+ "epoch": 41.99,
1257
+ "learning_rate": 0.00023469740634005762,
1258
+ "loss": 0.4814,
1259
+ "step": 1470
1260
+ },
1261
+ {
1262
+ "epoch": 41.99,
1263
+ "eval_loss": 0.8032535910606384,
1264
+ "eval_runtime": 8.8256,
1265
+ "eval_samples_per_second": 7.138,
1266
+ "eval_steps_per_second": 1.813,
1267
+ "eval_wer": 0.4120624209194433,
1268
+ "step": 1470
1269
+ },
1270
+ {
1271
+ "epoch": 42.28,
1272
+ "learning_rate": 0.00023354466858789628,
1273
+ "loss": 0.4302,
1274
+ "step": 1480
1275
+ },
1276
+ {
1277
+ "epoch": 42.57,
1278
+ "learning_rate": 0.00023239193083573489,
1279
+ "loss": 0.4579,
1280
+ "step": 1490
1281
+ },
1282
+ {
1283
+ "epoch": 42.85,
1284
+ "learning_rate": 0.00023123919308357352,
1285
+ "loss": 0.4416,
1286
+ "step": 1500
1287
+ },
1288
+ {
1289
+ "epoch": 42.99,
1290
+ "eval_loss": 0.8194773197174072,
1291
+ "eval_runtime": 8.7558,
1292
+ "eval_samples_per_second": 7.195,
1293
+ "eval_steps_per_second": 1.827,
1294
+ "eval_wer": 0.39898776887389287,
1295
+ "step": 1505
1296
+ },
1297
+ {
1298
+ "epoch": 43.14,
1299
+ "learning_rate": 0.00023008645533141212,
1300
+ "loss": 0.4628,
1301
+ "step": 1510
1302
+ },
1303
+ {
1304
+ "epoch": 43.43,
1305
+ "learning_rate": 0.00022893371757925075,
1306
+ "loss": 0.4765,
1307
+ "step": 1520
1308
+ },
1309
+ {
1310
+ "epoch": 43.71,
1311
+ "learning_rate": 0.00022778097982708936,
1312
+ "loss": 0.4504,
1313
+ "step": 1530
1314
+ },
1315
+ {
1316
+ "epoch": 43.99,
1317
+ "learning_rate": 0.00022662824207492793,
1318
+ "loss": 0.4958,
1319
+ "step": 1540
1320
+ },
1321
+ {
1322
+ "epoch": 43.99,
1323
+ "eval_loss": 0.8263903856277466,
1324
+ "eval_runtime": 8.775,
1325
+ "eval_samples_per_second": 7.179,
1326
+ "eval_steps_per_second": 1.823,
1327
+ "eval_wer": 0.39561366512020246,
1328
+ "step": 1540
1329
+ },
1330
+ {
1331
+ "epoch": 44.28,
1332
+ "learning_rate": 0.0002254755043227666,
1333
+ "loss": 0.4835,
1334
+ "step": 1550
1335
+ },
1336
+ {
1337
+ "epoch": 44.57,
1338
+ "learning_rate": 0.00022432276657060517,
1339
+ "loss": 0.4188,
1340
+ "step": 1560
1341
+ },
1342
+ {
1343
+ "epoch": 44.85,
1344
+ "learning_rate": 0.00022317002881844383,
1345
+ "loss": 0.4665,
1346
+ "step": 1570
1347
+ },
1348
+ {
1349
+ "epoch": 44.99,
1350
+ "eval_loss": 0.8172122240066528,
1351
+ "eval_runtime": 8.7583,
1352
+ "eval_samples_per_second": 7.193,
1353
+ "eval_steps_per_second": 1.827,
1354
+ "eval_wer": 0.4070012652889076,
1355
+ "step": 1575
1356
+ },
1357
+ {
1358
+ "epoch": 45.14,
1359
+ "learning_rate": 0.00022201729106628243,
1360
+ "loss": 0.4708,
1361
+ "step": 1580
1362
+ },
1363
+ {
1364
+ "epoch": 45.43,
1365
+ "learning_rate": 0.00022086455331412106,
1366
+ "loss": 0.4321,
1367
+ "step": 1590
1368
+ },
1369
+ {
1370
+ "epoch": 45.71,
1371
+ "learning_rate": 0.00021971181556195967,
1372
+ "loss": 0.5132,
1373
+ "step": 1600
1374
+ },
1375
+ {
1376
+ "epoch": 45.99,
1377
+ "learning_rate": 0.0002185590778097983,
1378
+ "loss": 0.4196,
1379
+ "step": 1610
1380
+ },
1381
+ {
1382
+ "epoch": 45.99,
1383
+ "eval_loss": 0.7970888018608093,
1384
+ "eval_runtime": 8.8547,
1385
+ "eval_samples_per_second": 7.115,
1386
+ "eval_steps_per_second": 1.807,
1387
+ "eval_wer": 0.39519190215099115,
1388
+ "step": 1610
1389
+ },
1390
+ {
1391
+ "epoch": 46.28,
1392
+ "learning_rate": 0.0002174063400576369,
1393
+ "loss": 0.5437,
1394
+ "step": 1620
1395
+ },
1396
+ {
1397
+ "epoch": 46.57,
1398
+ "learning_rate": 0.00021625360230547553,
1399
+ "loss": 0.4058,
1400
+ "step": 1630
1401
+ },
1402
+ {
1403
+ "epoch": 46.85,
1404
+ "learning_rate": 0.00021510086455331414,
1405
+ "loss": 0.4088,
1406
+ "step": 1640
1407
+ },
1408
+ {
1409
+ "epoch": 46.99,
1410
+ "eval_loss": 0.7417133450508118,
1411
+ "eval_runtime": 8.7997,
1412
+ "eval_samples_per_second": 7.159,
1413
+ "eval_steps_per_second": 1.818,
1414
+ "eval_wer": 0.38802193167439897,
1415
+ "step": 1645
1416
+ },
1417
+ {
1418
+ "epoch": 47.14,
1419
+ "learning_rate": 0.0002139481268011527,
1420
+ "loss": 0.4692,
1421
+ "step": 1650
1422
+ },
1423
+ {
1424
+ "epoch": 47.43,
1425
+ "learning_rate": 0.00021279538904899137,
1426
+ "loss": 0.4082,
1427
+ "step": 1660
1428
+ },
1429
+ {
1430
+ "epoch": 47.71,
1431
+ "learning_rate": 0.00021164265129682998,
1432
+ "loss": 0.4934,
1433
+ "step": 1670
1434
+ },
1435
+ {
1436
+ "epoch": 47.99,
1437
+ "learning_rate": 0.0002104899135446686,
1438
+ "loss": 0.4308,
1439
+ "step": 1680
1440
+ },
1441
+ {
1442
+ "epoch": 47.99,
1443
+ "eval_loss": 0.7806001901626587,
1444
+ "eval_runtime": 8.8422,
1445
+ "eval_samples_per_second": 7.125,
1446
+ "eval_steps_per_second": 1.81,
1447
+ "eval_wer": 0.39308308730493463,
1448
+ "step": 1680
1449
+ },
1450
+ {
1451
+ "epoch": 48.28,
1452
+ "learning_rate": 0.0002093371757925072,
1453
+ "loss": 0.4835,
1454
+ "step": 1690
1455
+ },
1456
+ {
1457
+ "epoch": 48.57,
1458
+ "learning_rate": 0.00020818443804034584,
1459
+ "loss": 0.4237,
1460
+ "step": 1700
1461
+ },
1462
+ {
1463
+ "epoch": 48.85,
1464
+ "learning_rate": 0.00020703170028818445,
1465
+ "loss": 0.4173,
1466
+ "step": 1710
1467
+ },
1468
+ {
1469
+ "epoch": 48.99,
1470
+ "eval_loss": 0.7380473613739014,
1471
+ "eval_runtime": 8.9386,
1472
+ "eval_samples_per_second": 7.048,
1473
+ "eval_steps_per_second": 1.79,
1474
+ "eval_wer": 0.392239561366512,
1475
+ "step": 1715
1476
+ },
1477
+ {
1478
+ "epoch": 49.14,
1479
+ "learning_rate": 0.00020587896253602308,
1480
+ "loss": 0.6076,
1481
+ "step": 1720
1482
+ },
1483
+ {
1484
+ "epoch": 49.43,
1485
+ "learning_rate": 0.00020472622478386168,
1486
+ "loss": 0.4505,
1487
+ "step": 1730
1488
+ },
1489
+ {
1490
+ "epoch": 49.71,
1491
+ "learning_rate": 0.0002035734870317003,
1492
+ "loss": 0.4394,
1493
+ "step": 1740
1494
+ },
1495
+ {
1496
+ "epoch": 49.99,
1497
+ "learning_rate": 0.00020242074927953892,
1498
+ "loss": 0.4653,
1499
+ "step": 1750
1500
+ },
1501
+ {
1502
+ "epoch": 49.99,
1503
+ "eval_loss": 0.8961731195449829,
1504
+ "eval_runtime": 8.8706,
1505
+ "eval_samples_per_second": 7.102,
1506
+ "eval_steps_per_second": 1.804,
1507
+ "eval_wer": 0.4027836355967946,
1508
+ "step": 1750
1509
+ },
1510
+ {
1511
+ "epoch": 50.28,
1512
+ "learning_rate": 0.00020126801152737752,
1513
+ "loss": 0.4578,
1514
+ "step": 1760
1515
+ },
1516
+ {
1517
+ "epoch": 50.57,
1518
+ "learning_rate": 0.00020011527377521615,
1519
+ "loss": 1.0308,
1520
+ "step": 1770
1521
+ },
1522
+ {
1523
+ "epoch": 50.85,
1524
+ "learning_rate": 0.00019896253602305476,
1525
+ "loss": 0.4406,
1526
+ "step": 1780
1527
+ },
1528
+ {
1529
+ "epoch": 50.99,
1530
+ "eval_loss": 0.7790281176567078,
1531
+ "eval_runtime": 8.8636,
1532
+ "eval_samples_per_second": 7.108,
1533
+ "eval_steps_per_second": 1.805,
1534
+ "eval_wer": 0.39350485027414595,
1535
+ "step": 1785
1536
+ },
1537
+ {
1538
+ "epoch": 51.14,
1539
+ "learning_rate": 0.0001978097982708934,
1540
+ "loss": 0.5174,
1541
+ "step": 1790
1542
+ },
1543
+ {
1544
+ "epoch": 51.43,
1545
+ "learning_rate": 0.00019665706051873202,
1546
+ "loss": 0.489,
1547
+ "step": 1800
1548
+ },
1549
+ {
1550
+ "epoch": 51.71,
1551
+ "learning_rate": 0.00019550432276657062,
1552
+ "loss": 0.6726,
1553
+ "step": 1810
1554
+ },
1555
+ {
1556
+ "epoch": 51.99,
1557
+ "learning_rate": 0.00019435158501440923,
1558
+ "loss": 0.4664,
1559
+ "step": 1820
1560
+ },
1561
+ {
1562
+ "epoch": 51.99,
1563
+ "eval_loss": 0.917263925075531,
1564
+ "eval_runtime": 8.8897,
1565
+ "eval_samples_per_second": 7.087,
1566
+ "eval_steps_per_second": 1.8,
1567
+ "eval_wer": 0.3892872205820329,
1568
+ "step": 1820
1569
+ },
1570
+ {
1571
+ "epoch": 52.28,
1572
+ "learning_rate": 0.00019319884726224783,
1573
+ "loss": 0.4466,
1574
+ "step": 1830
1575
+ },
1576
+ {
1577
+ "epoch": 52.57,
1578
+ "learning_rate": 0.00019204610951008646,
1579
+ "loss": 0.4465,
1580
+ "step": 1840
1581
+ },
1582
+ {
1583
+ "epoch": 52.85,
1584
+ "learning_rate": 0.00019089337175792507,
1585
+ "loss": 0.4486,
1586
+ "step": 1850
1587
+ },
1588
+ {
1589
+ "epoch": 52.99,
1590
+ "eval_loss": 0.8234918713569641,
1591
+ "eval_runtime": 8.8605,
1592
+ "eval_samples_per_second": 7.11,
1593
+ "eval_steps_per_second": 1.806,
1594
+ "eval_wer": 0.392239561366512,
1595
+ "step": 1855
1596
+ },
1597
+ {
1598
+ "epoch": 53.14,
1599
+ "learning_rate": 0.0001897406340057637,
1600
+ "loss": 0.4336,
1601
+ "step": 1860
1602
+ },
1603
+ {
1604
+ "epoch": 53.43,
1605
+ "learning_rate": 0.0001885878962536023,
1606
+ "loss": 0.4376,
1607
+ "step": 1870
1608
+ },
1609
+ {
1610
+ "epoch": 53.71,
1611
+ "learning_rate": 0.00018743515850144093,
1612
+ "loss": 0.4706,
1613
+ "step": 1880
1614
+ },
1615
+ {
1616
+ "epoch": 53.99,
1617
+ "learning_rate": 0.00018628242074927956,
1618
+ "loss": 0.4137,
1619
+ "step": 1890
1620
+ },
1621
+ {
1622
+ "epoch": 53.99,
1623
+ "eval_loss": 0.8031628727912903,
1624
+ "eval_runtime": 8.8634,
1625
+ "eval_samples_per_second": 7.108,
1626
+ "eval_steps_per_second": 1.805,
1627
+ "eval_wer": 0.3926613243357233,
1628
+ "step": 1890
1629
+ },
1630
+ {
1631
+ "epoch": 54.28,
1632
+ "learning_rate": 0.00018512968299711817,
1633
+ "loss": 0.4758,
1634
+ "step": 1900
1635
+ },
1636
+ {
1637
+ "epoch": 54.57,
1638
+ "learning_rate": 0.0001839769452449568,
1639
+ "loss": 0.3621,
1640
+ "step": 1910
1641
+ },
1642
+ {
1643
+ "epoch": 54.85,
1644
+ "learning_rate": 0.0001828242074927954,
1645
+ "loss": 0.4402,
1646
+ "step": 1920
1647
+ },
1648
+ {
1649
+ "epoch": 54.99,
1650
+ "eval_loss": 0.7658433318138123,
1651
+ "eval_runtime": 8.8656,
1652
+ "eval_samples_per_second": 7.106,
1653
+ "eval_steps_per_second": 1.805,
1654
+ "eval_wer": 0.38296077604386336,
1655
+ "step": 1925
1656
+ },
1657
+ {
1658
+ "epoch": 55.14,
1659
+ "learning_rate": 0.000181671469740634,
1660
+ "loss": 0.4614,
1661
+ "step": 1930
1662
+ },
1663
+ {
1664
+ "epoch": 55.43,
1665
+ "learning_rate": 0.0001805187319884726,
1666
+ "loss": 0.4183,
1667
+ "step": 1940
1668
+ },
1669
+ {
1670
+ "epoch": 55.71,
1671
+ "learning_rate": 0.00017936599423631124,
1672
+ "loss": 0.4065,
1673
+ "step": 1950
1674
+ },
1675
+ {
1676
+ "epoch": 55.99,
1677
+ "learning_rate": 0.00017821325648414985,
1678
+ "loss": 0.4101,
1679
+ "step": 1960
1680
+ },
1681
+ {
1682
+ "epoch": 55.99,
1683
+ "eval_loss": 0.8620884418487549,
1684
+ "eval_runtime": 8.9342,
1685
+ "eval_samples_per_second": 7.052,
1686
+ "eval_steps_per_second": 1.791,
1687
+ "eval_wer": 0.3994095318431042,
1688
+ "step": 1960
1689
+ },
1690
+ {
1691
+ "epoch": 56.28,
1692
+ "learning_rate": 0.00017706051873198848,
1693
+ "loss": 0.5536,
1694
+ "step": 1970
1695
+ },
1696
+ {
1697
+ "epoch": 56.57,
1698
+ "learning_rate": 0.0001759077809798271,
1699
+ "loss": 0.4167,
1700
+ "step": 1980
1701
+ },
1702
+ {
1703
+ "epoch": 56.85,
1704
+ "learning_rate": 0.0001747550432276657,
1705
+ "loss": 0.5239,
1706
+ "step": 1990
1707
+ },
1708
+ {
1709
+ "epoch": 56.99,
1710
+ "eval_loss": 0.7903460264205933,
1711
+ "eval_runtime": 8.918,
1712
+ "eval_samples_per_second": 7.064,
1713
+ "eval_steps_per_second": 1.794,
1714
+ "eval_wer": 0.39561366512020246,
1715
+ "step": 1995
1716
+ },
1717
+ {
1718
+ "epoch": 57.14,
1719
+ "learning_rate": 0.00017360230547550434,
1720
+ "loss": 0.6093,
1721
+ "step": 2000
1722
+ },
1723
+ {
1724
+ "epoch": 57.43,
1725
+ "learning_rate": 0.00017244956772334295,
1726
+ "loss": 0.4394,
1727
+ "step": 2010
1728
+ },
1729
+ {
1730
+ "epoch": 57.71,
1731
+ "learning_rate": 0.00017129682997118158,
1732
+ "loss": 0.4504,
1733
+ "step": 2020
1734
+ },
1735
+ {
1736
+ "epoch": 57.99,
1737
+ "learning_rate": 0.00017014409221902018,
1738
+ "loss": 0.4151,
1739
+ "step": 2030
1740
+ },
1741
+ {
1742
+ "epoch": 57.99,
1743
+ "eval_loss": 0.7849175333976746,
1744
+ "eval_runtime": 8.9241,
1745
+ "eval_samples_per_second": 7.06,
1746
+ "eval_steps_per_second": 1.793,
1747
+ "eval_wer": 0.3871784057359764,
1748
+ "step": 2030
1749
+ },
1750
+ {
1751
+ "epoch": 58.28,
1752
+ "learning_rate": 0.0001689913544668588,
1753
+ "loss": 0.463,
1754
+ "step": 2040
1755
+ },
1756
+ {
1757
+ "epoch": 58.57,
1758
+ "learning_rate": 0.00016783861671469742,
1759
+ "loss": 0.4195,
1760
+ "step": 2050
1761
+ },
1762
+ {
1763
+ "epoch": 58.85,
1764
+ "learning_rate": 0.00016668587896253602,
1765
+ "loss": 0.4766,
1766
+ "step": 2060
1767
+ },
1768
+ {
1769
+ "epoch": 58.99,
1770
+ "eval_loss": 0.8306421041488647,
1771
+ "eval_runtime": 9.0122,
1772
+ "eval_samples_per_second": 6.99,
1773
+ "eval_steps_per_second": 1.775,
1774
+ "eval_wer": 0.39181779839730074,
1775
+ "step": 2065
1776
+ },
1777
+ {
1778
+ "epoch": 59.14,
1779
+ "learning_rate": 0.00016553314121037465,
1780
+ "loss": 0.4292,
1781
+ "step": 2070
1782
+ },
1783
+ {
1784
+ "epoch": 59.43,
1785
+ "learning_rate": 0.00016438040345821326,
1786
+ "loss": 0.4278,
1787
+ "step": 2080
1788
+ },
1789
+ {
1790
+ "epoch": 59.71,
1791
+ "learning_rate": 0.0001632276657060519,
1792
+ "loss": 0.4586,
1793
+ "step": 2090
1794
+ },
1795
+ {
1796
+ "epoch": 59.99,
1797
+ "learning_rate": 0.0001620749279538905,
1798
+ "loss": 0.4882,
1799
+ "step": 2100
1800
+ },
1801
+ {
1802
+ "epoch": 59.99,
1803
+ "eval_loss": 0.8134064674377441,
1804
+ "eval_runtime": 8.8415,
1805
+ "eval_samples_per_second": 7.125,
1806
+ "eval_steps_per_second": 1.81,
1807
+ "eval_wer": 0.3926613243357233,
1808
+ "step": 2100
1809
+ },
1810
+ {
1811
+ "epoch": 60.28,
1812
+ "learning_rate": 0.00016092219020172912,
1813
+ "loss": 0.3925,
1814
+ "step": 2110
1815
+ },
1816
+ {
1817
+ "epoch": 60.57,
1818
+ "learning_rate": 0.00015976945244956773,
1819
+ "loss": 0.4318,
1820
+ "step": 2120
1821
+ },
1822
+ {
1823
+ "epoch": 60.85,
1824
+ "learning_rate": 0.00015861671469740636,
1825
+ "loss": 0.4583,
1826
+ "step": 2130
1827
+ },
1828
+ {
1829
+ "epoch": 60.99,
1830
+ "eval_loss": 0.9527338743209839,
1831
+ "eval_runtime": 8.8501,
1832
+ "eval_samples_per_second": 7.119,
1833
+ "eval_steps_per_second": 1.808,
1834
+ "eval_wer": 0.3850695908899199,
1835
+ "step": 2135
1836
+ },
1837
+ {
1838
+ "epoch": 61.14,
1839
+ "learning_rate": 0.00015746397694524496,
1840
+ "loss": 0.6257,
1841
+ "step": 2140
1842
+ },
1843
+ {
1844
+ "epoch": 61.43,
1845
+ "learning_rate": 0.0001563112391930836,
1846
+ "loss": 0.3906,
1847
+ "step": 2150
1848
+ },
1849
+ {
1850
+ "epoch": 61.71,
1851
+ "learning_rate": 0.0001551585014409222,
1852
+ "loss": 0.468,
1853
+ "step": 2160
1854
+ },
1855
+ {
1856
+ "epoch": 61.99,
1857
+ "learning_rate": 0.0001540057636887608,
1858
+ "loss": 0.4284,
1859
+ "step": 2170
1860
+ },
1861
+ {
1862
+ "epoch": 61.99,
1863
+ "eval_loss": 0.9743154048919678,
1864
+ "eval_runtime": 8.9267,
1865
+ "eval_samples_per_second": 7.057,
1866
+ "eval_steps_per_second": 1.792,
1867
+ "eval_wer": 0.3998312948123155,
1868
+ "step": 2170
1869
+ },
1870
+ {
1871
+ "epoch": 62.28,
1872
+ "learning_rate": 0.00015285302593659943,
1873
+ "loss": 0.3827,
1874
+ "step": 2180
1875
+ },
1876
+ {
1877
+ "epoch": 62.57,
1878
+ "learning_rate": 0.00015170028818443804,
1879
+ "loss": 0.3974,
1880
+ "step": 2190
1881
+ },
1882
+ {
1883
+ "epoch": 62.85,
1884
+ "learning_rate": 0.00015054755043227667,
1885
+ "loss": 0.46,
1886
+ "step": 2200
1887
+ },
1888
+ {
1889
+ "epoch": 62.99,
1890
+ "eval_loss": 0.7807419300079346,
1891
+ "eval_runtime": 8.9409,
1892
+ "eval_samples_per_second": 7.046,
1893
+ "eval_steps_per_second": 1.79,
1894
+ "eval_wer": 0.38296077604386336,
1895
+ "step": 2205
1896
+ },
1897
+ {
1898
+ "epoch": 63.14,
1899
+ "learning_rate": 0.00014939481268011527,
1900
+ "loss": 0.4637,
1901
+ "step": 2210
1902
+ },
1903
+ {
1904
+ "epoch": 63.43,
1905
+ "learning_rate": 0.0001482420749279539,
1906
+ "loss": 0.3948,
1907
+ "step": 2220
1908
+ },
1909
+ {
1910
+ "epoch": 63.71,
1911
+ "learning_rate": 0.0001470893371757925,
1912
+ "loss": 0.4112,
1913
+ "step": 2230
1914
+ },
1915
+ {
1916
+ "epoch": 63.99,
1917
+ "learning_rate": 0.00014593659942363114,
1918
+ "loss": 0.4039,
1919
+ "step": 2240
1920
+ },
1921
+ {
1922
+ "epoch": 63.99,
1923
+ "eval_loss": 0.8864094614982605,
1924
+ "eval_runtime": 8.7598,
1925
+ "eval_samples_per_second": 7.192,
1926
+ "eval_steps_per_second": 1.827,
1927
+ "eval_wer": 0.3884436946436103,
1928
+ "step": 2240
1929
+ },
1930
+ {
1931
+ "epoch": 64.28,
1932
+ "learning_rate": 0.00014478386167146974,
1933
+ "loss": 0.3935,
1934
+ "step": 2250
1935
+ },
1936
+ {
1937
+ "epoch": 64.57,
1938
+ "learning_rate": 0.00014363112391930838,
1939
+ "loss": 0.3677,
1940
+ "step": 2260
1941
+ },
1942
+ {
1943
+ "epoch": 64.85,
1944
+ "learning_rate": 0.00014247838616714698,
1945
+ "loss": 0.3868,
1946
+ "step": 2270
1947
+ },
1948
+ {
1949
+ "epoch": 64.99,
1950
+ "eval_loss": 0.7304245233535767,
1951
+ "eval_runtime": 8.9023,
1952
+ "eval_samples_per_second": 7.077,
1953
+ "eval_steps_per_second": 1.797,
1954
+ "eval_wer": 0.3816954871362294,
1955
+ "step": 2275
1956
+ },
1957
+ {
1958
+ "epoch": 65.14,
1959
+ "learning_rate": 0.00014132564841498558,
1960
+ "loss": 0.4321,
1961
+ "step": 2280
1962
+ },
1963
+ {
1964
+ "epoch": 65.43,
1965
+ "learning_rate": 0.00014017291066282421,
1966
+ "loss": 0.3726,
1967
+ "step": 2290
1968
+ },
1969
+ {
1970
+ "epoch": 65.71,
1971
+ "learning_rate": 0.00013902017291066282,
1972
+ "loss": 0.428,
1973
+ "step": 2300
1974
+ },
1975
+ {
1976
+ "epoch": 65.99,
1977
+ "learning_rate": 0.00013786743515850145,
1978
+ "loss": 0.3934,
1979
+ "step": 2310
1980
+ },
1981
+ {
1982
+ "epoch": 65.99,
1983
+ "eval_loss": 0.875798761844635,
1984
+ "eval_runtime": 8.8052,
1985
+ "eval_samples_per_second": 7.155,
1986
+ "eval_steps_per_second": 1.817,
1987
+ "eval_wer": 0.38464782792070856,
1988
+ "step": 2310
1989
+ },
1990
+ {
1991
+ "epoch": 66.28,
1992
+ "learning_rate": 0.00013671469740634005,
1993
+ "loss": 0.3963,
1994
+ "step": 2320
1995
+ },
1996
+ {
1997
+ "epoch": 66.57,
1998
+ "learning_rate": 0.00013556195965417869,
1999
+ "loss": 0.3708,
2000
+ "step": 2330
2001
+ },
2002
+ {
2003
+ "epoch": 66.85,
2004
+ "learning_rate": 0.0001344092219020173,
2005
+ "loss": 0.3776,
2006
+ "step": 2340
2007
+ },
2008
+ {
2009
+ "epoch": 66.99,
2010
+ "eval_loss": 0.8156087398529053,
2011
+ "eval_runtime": 8.9243,
2012
+ "eval_samples_per_second": 7.059,
2013
+ "eval_steps_per_second": 1.793,
2014
+ "eval_wer": 0.3762125685364825,
2015
+ "step": 2345
2016
+ },
2017
+ {
2018
+ "epoch": 67.14,
2019
+ "learning_rate": 0.00013325648414985592,
2020
+ "loss": 0.4032,
2021
+ "step": 2350
2022
+ },
2023
+ {
2024
+ "epoch": 67.43,
2025
+ "learning_rate": 0.00013210374639769455,
2026
+ "loss": 0.4047,
2027
+ "step": 2360
2028
+ },
2029
+ {
2030
+ "epoch": 67.71,
2031
+ "learning_rate": 0.00013095100864553316,
2032
+ "loss": 0.4063,
2033
+ "step": 2370
2034
+ },
2035
+ {
2036
+ "epoch": 67.99,
2037
+ "learning_rate": 0.00012979827089337176,
2038
+ "loss": 0.3499,
2039
+ "step": 2380
2040
+ },
2041
+ {
2042
+ "epoch": 67.99,
2043
+ "eval_loss": 0.814339816570282,
2044
+ "eval_runtime": 8.9099,
2045
+ "eval_samples_per_second": 7.071,
2046
+ "eval_steps_per_second": 1.796,
2047
+ "eval_wer": 0.3888654576128216,
2048
+ "step": 2380
2049
+ },
2050
+ {
2051
+ "epoch": 68.28,
2052
+ "learning_rate": 0.00012864553314121036,
2053
+ "loss": 0.4262,
2054
+ "step": 2390
2055
+ },
2056
+ {
2057
+ "epoch": 68.57,
2058
+ "learning_rate": 0.000127492795389049,
2059
+ "loss": 0.4213,
2060
+ "step": 2400
2061
+ },
2062
+ {
2063
+ "epoch": 68.85,
2064
+ "learning_rate": 0.0001263400576368876,
2065
+ "loss": 0.4055,
2066
+ "step": 2410
2067
+ },
2068
+ {
2069
+ "epoch": 68.99,
2070
+ "eval_loss": 0.7502649426460266,
2071
+ "eval_runtime": 8.8806,
2072
+ "eval_samples_per_second": 7.094,
2073
+ "eval_steps_per_second": 1.802,
2074
+ "eval_wer": 0.3795866722901729,
2075
+ "step": 2415
2076
+ },
2077
+ {
2078
+ "epoch": 69.14,
2079
+ "learning_rate": 0.00012518731988472623,
2080
+ "loss": 0.4393,
2081
+ "step": 2420
2082
+ },
2083
+ {
2084
+ "epoch": 69.43,
2085
+ "learning_rate": 0.00012403458213256483,
2086
+ "loss": 0.3695,
2087
+ "step": 2430
2088
+ },
2089
+ {
2090
+ "epoch": 69.71,
2091
+ "learning_rate": 0.00012288184438040347,
2092
+ "loss": 0.4452,
2093
+ "step": 2440
2094
+ },
2095
+ {
2096
+ "epoch": 69.99,
2097
+ "learning_rate": 0.00012172910662824208,
2098
+ "loss": 0.3505,
2099
+ "step": 2450
2100
+ },
2101
+ {
2102
+ "epoch": 69.99,
2103
+ "eval_loss": 0.7138158082962036,
2104
+ "eval_runtime": 8.8451,
2105
+ "eval_samples_per_second": 7.123,
2106
+ "eval_steps_per_second": 1.809,
2107
+ "eval_wer": 0.38043019822859553,
2108
+ "step": 2450
2109
+ },
2110
+ {
2111
+ "epoch": 70.28,
2112
+ "learning_rate": 0.0001205763688760807,
2113
+ "loss": 0.4222,
2114
+ "step": 2460
2115
+ },
2116
+ {
2117
+ "epoch": 70.57,
2118
+ "learning_rate": 0.00011942363112391932,
2119
+ "loss": 0.3144,
2120
+ "step": 2470
2121
+ },
2122
+ {
2123
+ "epoch": 70.85,
2124
+ "learning_rate": 0.00011827089337175794,
2125
+ "loss": 0.3755,
2126
+ "step": 2480
2127
+ },
2128
+ {
2129
+ "epoch": 70.99,
2130
+ "eval_loss": 0.8071521520614624,
2131
+ "eval_runtime": 8.8022,
2132
+ "eval_samples_per_second": 7.157,
2133
+ "eval_steps_per_second": 1.818,
2134
+ "eval_wer": 0.3800084352593842,
2135
+ "step": 2485
2136
+ },
2137
+ {
2138
+ "epoch": 71.14,
2139
+ "learning_rate": 0.00011711815561959654,
2140
+ "loss": 0.3778,
2141
+ "step": 2490
2142
+ },
2143
+ {
2144
+ "epoch": 71.43,
2145
+ "learning_rate": 0.00011596541786743516,
2146
+ "loss": 0.3831,
2147
+ "step": 2500
2148
+ },
2149
+ {
2150
+ "epoch": 71.71,
2151
+ "learning_rate": 0.00011481268011527378,
2152
+ "loss": 0.3552,
2153
+ "step": 2510
2154
+ },
2155
+ {
2156
+ "epoch": 71.99,
2157
+ "learning_rate": 0.0001136599423631124,
2158
+ "loss": 0.3594,
2159
+ "step": 2520
2160
+ },
2161
+ {
2162
+ "epoch": 71.99,
2163
+ "eval_loss": 0.7692469358444214,
2164
+ "eval_runtime": 8.7672,
2165
+ "eval_samples_per_second": 7.186,
2166
+ "eval_steps_per_second": 1.825,
2167
+ "eval_wer": 0.3850695908899199,
2168
+ "step": 2520
2169
+ },
2170
+ {
2171
+ "epoch": 72.28,
2172
+ "learning_rate": 0.00011250720461095101,
2173
+ "loss": 0.4549,
2174
+ "step": 2530
2175
+ },
2176
+ {
2177
+ "epoch": 72.57,
2178
+ "learning_rate": 0.00011135446685878963,
2179
+ "loss": 0.3636,
2180
+ "step": 2540
2181
+ },
2182
+ {
2183
+ "epoch": 72.85,
2184
+ "learning_rate": 0.00011020172910662825,
2185
+ "loss": 0.3167,
2186
+ "step": 2550
2187
+ },
2188
+ {
2189
+ "epoch": 72.99,
2190
+ "eval_loss": 0.6994994282722473,
2191
+ "eval_runtime": 8.766,
2192
+ "eval_samples_per_second": 7.187,
2193
+ "eval_steps_per_second": 1.825,
2194
+ "eval_wer": 0.3745255166596373,
2195
+ "step": 2555
2196
+ },
2197
+ {
2198
+ "epoch": 73.14,
2199
+ "learning_rate": 0.00010904899135446686,
2200
+ "loss": 0.3651,
2201
+ "step": 2560
2202
+ },
2203
+ {
2204
+ "epoch": 73.43,
2205
+ "learning_rate": 0.00010789625360230548,
2206
+ "loss": 0.3751,
2207
+ "step": 2570
2208
+ },
2209
+ {
2210
+ "epoch": 73.71,
2211
+ "learning_rate": 0.0001067435158501441,
2212
+ "loss": 0.4196,
2213
+ "step": 2580
2214
+ },
2215
+ {
2216
+ "epoch": 73.99,
2217
+ "learning_rate": 0.00010559077809798272,
2218
+ "loss": 0.3915,
2219
+ "step": 2590
2220
+ },
2221
+ {
2222
+ "epoch": 73.99,
2223
+ "eval_loss": 0.6712434887886047,
2224
+ "eval_runtime": 8.8181,
2225
+ "eval_samples_per_second": 7.144,
2226
+ "eval_steps_per_second": 1.814,
2227
+ "eval_wer": 0.3762125685364825,
2228
+ "step": 2590
2229
+ },
2230
+ {
2231
+ "epoch": 74.28,
2232
+ "learning_rate": 0.00010443804034582132,
2233
+ "loss": 0.3952,
2234
+ "step": 2600
2235
+ },
2236
+ {
2237
+ "epoch": 74.57,
2238
+ "learning_rate": 0.00010328530259365994,
2239
+ "loss": 0.3115,
2240
+ "step": 2610
2241
+ },
2242
+ {
2243
+ "epoch": 74.85,
2244
+ "learning_rate": 0.00010213256484149856,
2245
+ "loss": 0.3741,
2246
+ "step": 2620
2247
+ },
2248
+ {
2249
+ "epoch": 74.99,
2250
+ "eval_loss": 0.7139262557029724,
2251
+ "eval_runtime": 8.8329,
2252
+ "eval_samples_per_second": 7.132,
2253
+ "eval_steps_per_second": 1.811,
2254
+ "eval_wer": 0.3800084352593842,
2255
+ "step": 2625
2256
+ },
2257
+ {
2258
+ "epoch": 75.14,
2259
+ "learning_rate": 0.00010097982708933717,
2260
+ "loss": 0.3542,
2261
+ "step": 2630
2262
+ },
2263
+ {
2264
+ "epoch": 75.43,
2265
+ "learning_rate": 9.982708933717579e-05,
2266
+ "loss": 0.4682,
2267
+ "step": 2640
2268
+ },
2269
+ {
2270
+ "epoch": 75.71,
2271
+ "learning_rate": 9.867435158501441e-05,
2272
+ "loss": 0.3378,
2273
+ "step": 2650
2274
+ },
2275
+ {
2276
+ "epoch": 75.99,
2277
+ "learning_rate": 9.752161383285303e-05,
2278
+ "loss": 0.3708,
2279
+ "step": 2660
2280
+ },
2281
+ {
2282
+ "epoch": 75.99,
2283
+ "eval_loss": 0.7064934968948364,
2284
+ "eval_runtime": 8.8273,
2285
+ "eval_samples_per_second": 7.137,
2286
+ "eval_steps_per_second": 1.813,
2287
+ "eval_wer": 0.3833825390130747,
2288
+ "step": 2660
2289
+ },
2290
+ {
2291
+ "epoch": 76.28,
2292
+ "learning_rate": 9.636887608069164e-05,
2293
+ "loss": 0.4179,
2294
+ "step": 2670
2295
+ },
2296
+ {
2297
+ "epoch": 76.57,
2298
+ "learning_rate": 9.521613832853028e-05,
2299
+ "loss": 0.3543,
2300
+ "step": 2680
2301
+ },
2302
+ {
2303
+ "epoch": 76.85,
2304
+ "learning_rate": 9.406340057636888e-05,
2305
+ "loss": 0.3731,
2306
+ "step": 2690
2307
+ },
2308
+ {
2309
+ "epoch": 76.99,
2310
+ "eval_loss": 0.7315667867660522,
2311
+ "eval_runtime": 8.8516,
2312
+ "eval_samples_per_second": 7.117,
2313
+ "eval_steps_per_second": 1.808,
2314
+ "eval_wer": 0.37536904259805987,
2315
+ "step": 2695
2316
+ },
2317
+ {
2318
+ "epoch": 77.14,
2319
+ "learning_rate": 9.29106628242075e-05,
2320
+ "loss": 0.4854,
2321
+ "step": 2700
2322
+ },
2323
+ {
2324
+ "epoch": 77.43,
2325
+ "learning_rate": 9.175792507204611e-05,
2326
+ "loss": 0.3493,
2327
+ "step": 2710
2328
+ },
2329
+ {
2330
+ "epoch": 77.71,
2331
+ "learning_rate": 9.060518731988473e-05,
2332
+ "loss": 0.3559,
2333
+ "step": 2720
2334
+ },
2335
+ {
2336
+ "epoch": 77.99,
2337
+ "learning_rate": 8.945244956772335e-05,
2338
+ "loss": 0.3785,
2339
+ "step": 2730
2340
+ },
2341
+ {
2342
+ "epoch": 77.99,
2343
+ "eval_loss": 0.7071495056152344,
2344
+ "eval_runtime": 8.8836,
2345
+ "eval_samples_per_second": 7.092,
2346
+ "eval_steps_per_second": 1.801,
2347
+ "eval_wer": 0.3757908055672712,
2348
+ "step": 2730
2349
+ },
2350
+ {
2351
+ "epoch": 78.28,
2352
+ "learning_rate": 8.829971181556195e-05,
2353
+ "loss": 0.3751,
2354
+ "step": 2740
2355
+ },
2356
+ {
2357
+ "epoch": 78.57,
2358
+ "learning_rate": 8.714697406340057e-05,
2359
+ "loss": 0.3622,
2360
+ "step": 2750
2361
+ },
2362
+ {
2363
+ "epoch": 78.85,
2364
+ "learning_rate": 8.599423631123919e-05,
2365
+ "loss": 0.3466,
2366
+ "step": 2760
2367
+ },
2368
+ {
2369
+ "epoch": 78.99,
2370
+ "eval_loss": 0.7361500263214111,
2371
+ "eval_runtime": 9.0867,
2372
+ "eval_samples_per_second": 6.933,
2373
+ "eval_steps_per_second": 1.761,
2374
+ "eval_wer": 0.3833825390130747,
2375
+ "step": 2765
2376
+ },
2377
+ {
2378
+ "epoch": 79.14,
2379
+ "learning_rate": 8.484149855907782e-05,
2380
+ "loss": 0.3678,
2381
+ "step": 2770
2382
+ },
2383
+ {
2384
+ "epoch": 79.43,
2385
+ "learning_rate": 8.368876080691644e-05,
2386
+ "loss": 0.3829,
2387
+ "step": 2780
2388
+ },
2389
+ {
2390
+ "epoch": 79.71,
2391
+ "learning_rate": 8.253602305475506e-05,
2392
+ "loss": 0.4725,
2393
+ "step": 2790
2394
+ },
2395
+ {
2396
+ "epoch": 79.99,
2397
+ "learning_rate": 8.138328530259366e-05,
2398
+ "loss": 0.3505,
2399
+ "step": 2800
2400
+ },
2401
+ {
2402
+ "epoch": 79.99,
2403
+ "eval_loss": 0.6964561343193054,
2404
+ "eval_runtime": 8.8201,
2405
+ "eval_samples_per_second": 7.143,
2406
+ "eval_steps_per_second": 1.814,
2407
+ "eval_wer": 0.3800084352593842,
2408
+ "step": 2800
2409
+ },
2410
+ {
2411
+ "epoch": 80.28,
2412
+ "learning_rate": 8.023054755043228e-05,
2413
+ "loss": 0.3439,
2414
+ "step": 2810
2415
+ },
2416
+ {
2417
+ "epoch": 80.57,
2418
+ "learning_rate": 7.90778097982709e-05,
2419
+ "loss": 0.3645,
2420
+ "step": 2820
2421
+ },
2422
+ {
2423
+ "epoch": 80.85,
2424
+ "learning_rate": 7.792507204610951e-05,
2425
+ "loss": 0.4003,
2426
+ "step": 2830
2427
+ },
2428
+ {
2429
+ "epoch": 80.99,
2430
+ "eval_loss": 0.752064049243927,
2431
+ "eval_runtime": 8.7801,
2432
+ "eval_samples_per_second": 7.175,
2433
+ "eval_steps_per_second": 1.822,
2434
+ "eval_wer": 0.3766343315056938,
2435
+ "step": 2835
2436
+ },
2437
+ {
2438
+ "epoch": 81.14,
2439
+ "learning_rate": 7.677233429394813e-05,
2440
+ "loss": 0.4152,
2441
+ "step": 2840
2442
+ },
2443
+ {
2444
+ "epoch": 81.43,
2445
+ "learning_rate": 7.561959654178675e-05,
2446
+ "loss": 0.3652,
2447
+ "step": 2850
2448
+ },
2449
+ {
2450
+ "epoch": 81.71,
2451
+ "learning_rate": 7.446685878962537e-05,
2452
+ "loss": 0.3254,
2453
+ "step": 2860
2454
+ },
2455
+ {
2456
+ "epoch": 81.99,
2457
+ "learning_rate": 7.331412103746398e-05,
2458
+ "loss": 0.3723,
2459
+ "step": 2870
2460
+ },
2461
+ {
2462
+ "epoch": 81.99,
2463
+ "eval_loss": 0.7616933584213257,
2464
+ "eval_runtime": 8.8771,
2465
+ "eval_samples_per_second": 7.097,
2466
+ "eval_steps_per_second": 1.802,
2467
+ "eval_wer": 0.3749472796288486,
2468
+ "step": 2870
2469
+ },
2470
+ {
2471
+ "epoch": 82.28,
2472
+ "learning_rate": 7.21613832853026e-05,
2473
+ "loss": 0.3791,
2474
+ "step": 2880
2475
+ },
2476
+ {
2477
+ "epoch": 82.57,
2478
+ "learning_rate": 7.100864553314122e-05,
2479
+ "loss": 0.3127,
2480
+ "step": 2890
2481
+ },
2482
+ {
2483
+ "epoch": 82.85,
2484
+ "learning_rate": 6.985590778097984e-05,
2485
+ "loss": 0.4029,
2486
+ "step": 2900
2487
+ },
2488
+ {
2489
+ "epoch": 82.99,
2490
+ "eval_loss": 0.7659004926681519,
2491
+ "eval_runtime": 8.8596,
2492
+ "eval_samples_per_second": 7.111,
2493
+ "eval_steps_per_second": 1.806,
2494
+ "eval_wer": 0.38127372416701816,
2495
+ "step": 2905
2496
+ },
2497
+ {
2498
+ "epoch": 83.14,
2499
+ "learning_rate": 6.870317002881844e-05,
2500
+ "loss": 0.3742,
2501
+ "step": 2910
2502
+ },
2503
+ {
2504
+ "epoch": 83.43,
2505
+ "learning_rate": 6.755043227665706e-05,
2506
+ "loss": 0.3072,
2507
+ "step": 2920
2508
+ },
2509
+ {
2510
+ "epoch": 83.71,
2511
+ "learning_rate": 6.639769452449568e-05,
2512
+ "loss": 0.3836,
2513
+ "step": 2930
2514
+ },
2515
+ {
2516
+ "epoch": 83.99,
2517
+ "learning_rate": 6.52449567723343e-05,
2518
+ "loss": 0.3478,
2519
+ "step": 2940
2520
+ },
2521
+ {
2522
+ "epoch": 83.99,
2523
+ "eval_loss": 0.707669198513031,
2524
+ "eval_runtime": 8.9849,
2525
+ "eval_samples_per_second": 7.012,
2526
+ "eval_steps_per_second": 1.781,
2527
+ "eval_wer": 0.3833825390130747,
2528
+ "step": 2940
2529
+ },
2530
+ {
2531
+ "epoch": 84.28,
2532
+ "learning_rate": 6.409221902017291e-05,
2533
+ "loss": 0.3467,
2534
+ "step": 2950
2535
+ },
2536
+ {
2537
+ "epoch": 84.57,
2538
+ "learning_rate": 6.293948126801153e-05,
2539
+ "loss": 0.324,
2540
+ "step": 2960
2541
+ },
2542
+ {
2543
+ "epoch": 84.85,
2544
+ "learning_rate": 6.178674351585015e-05,
2545
+ "loss": 0.3363,
2546
+ "step": 2970
2547
+ },
2548
+ {
2549
+ "epoch": 84.99,
2550
+ "eval_loss": 0.7333377599716187,
2551
+ "eval_runtime": 8.8448,
2552
+ "eval_samples_per_second": 7.123,
2553
+ "eval_steps_per_second": 1.809,
2554
+ "eval_wer": 0.3787431463517503,
2555
+ "step": 2975
2556
+ },
2557
+ {
2558
+ "epoch": 85.14,
2559
+ "learning_rate": 6.0634005763688764e-05,
2560
+ "loss": 0.343,
2561
+ "step": 2980
2562
+ },
2563
+ {
2564
+ "epoch": 85.43,
2565
+ "learning_rate": 5.948126801152738e-05,
2566
+ "loss": 0.3371,
2567
+ "step": 2990
2568
+ },
2569
+ {
2570
+ "epoch": 85.71,
2571
+ "learning_rate": 5.8328530259366e-05,
2572
+ "loss": 0.3989,
2573
+ "step": 3000
2574
+ },
2575
+ {
2576
+ "epoch": 85.99,
2577
+ "learning_rate": 5.717579250720462e-05,
2578
+ "loss": 0.4228,
2579
+ "step": 3010
2580
+ },
2581
+ {
2582
+ "epoch": 85.99,
2583
+ "eval_loss": 0.7196153998374939,
2584
+ "eval_runtime": 8.9305,
2585
+ "eval_samples_per_second": 7.054,
2586
+ "eval_steps_per_second": 1.792,
2587
+ "eval_wer": 0.3745255166596373,
2588
+ "step": 3010
2589
+ },
2590
+ {
2591
+ "epoch": 86.28,
2592
+ "learning_rate": 5.602305475504323e-05,
2593
+ "loss": 0.3931,
2594
+ "step": 3020
2595
+ },
2596
+ {
2597
+ "epoch": 86.57,
2598
+ "learning_rate": 5.4870317002881845e-05,
2599
+ "loss": 0.3359,
2600
+ "step": 3030
2601
+ },
2602
+ {
2603
+ "epoch": 86.85,
2604
+ "learning_rate": 5.371757925072046e-05,
2605
+ "loss": 0.3823,
2606
+ "step": 3040
2607
+ },
2608
+ {
2609
+ "epoch": 86.99,
2610
+ "eval_loss": 0.7194885015487671,
2611
+ "eval_runtime": 8.8773,
2612
+ "eval_samples_per_second": 7.097,
2613
+ "eval_steps_per_second": 1.802,
2614
+ "eval_wer": 0.37536904259805987,
2615
+ "step": 3045
2616
+ },
2617
+ {
2618
+ "epoch": 87.14,
2619
+ "learning_rate": 5.256484149855908e-05,
2620
+ "loss": 0.3635,
2621
+ "step": 3050
2622
+ },
2623
+ {
2624
+ "epoch": 87.43,
2625
+ "learning_rate": 5.14121037463977e-05,
2626
+ "loss": 0.3576,
2627
+ "step": 3060
2628
+ },
2629
+ {
2630
+ "epoch": 87.71,
2631
+ "learning_rate": 5.0259365994236316e-05,
2632
+ "loss": 0.4654,
2633
+ "step": 3070
2634
+ },
2635
+ {
2636
+ "epoch": 87.99,
2637
+ "learning_rate": 4.9106628242074933e-05,
2638
+ "loss": 0.3574,
2639
+ "step": 3080
2640
+ },
2641
+ {
2642
+ "epoch": 87.99,
2643
+ "eval_loss": 0.713660478591919,
2644
+ "eval_runtime": 8.8845,
2645
+ "eval_samples_per_second": 7.091,
2646
+ "eval_steps_per_second": 1.801,
2647
+ "eval_wer": 0.3795866722901729,
2648
+ "step": 3080
2649
+ },
2650
+ {
2651
+ "epoch": 88.28,
2652
+ "learning_rate": 4.7953890489913544e-05,
2653
+ "loss": 0.3677,
2654
+ "step": 3090
2655
+ },
2656
+ {
2657
+ "epoch": 88.57,
2658
+ "learning_rate": 4.680115273775216e-05,
2659
+ "loss": 0.3065,
2660
+ "step": 3100
2661
+ },
2662
+ {
2663
+ "epoch": 88.85,
2664
+ "learning_rate": 4.564841498559078e-05,
2665
+ "loss": 0.3371,
2666
+ "step": 3110
2667
+ },
2668
+ {
2669
+ "epoch": 88.99,
2670
+ "eval_loss": 0.7164372801780701,
2671
+ "eval_runtime": 8.865,
2672
+ "eval_samples_per_second": 7.107,
2673
+ "eval_steps_per_second": 1.805,
2674
+ "eval_wer": 0.3762125685364825,
2675
+ "step": 3115
2676
+ },
2677
+ {
2678
+ "epoch": 89.14,
2679
+ "learning_rate": 4.44956772334294e-05,
2680
+ "loss": 0.6018,
2681
+ "step": 3120
2682
+ },
2683
+ {
2684
+ "epoch": 89.43,
2685
+ "learning_rate": 4.3342939481268015e-05,
2686
+ "loss": 0.3404,
2687
+ "step": 3130
2688
+ },
2689
+ {
2690
+ "epoch": 89.71,
2691
+ "learning_rate": 4.219020172910663e-05,
2692
+ "loss": 0.3659,
2693
+ "step": 3140
2694
+ },
2695
+ {
2696
+ "epoch": 89.99,
2697
+ "learning_rate": 4.103746397694524e-05,
2698
+ "loss": 0.3548,
2699
+ "step": 3150
2700
+ },
2701
+ {
2702
+ "epoch": 89.99,
2703
+ "eval_loss": 0.7766362428665161,
2704
+ "eval_runtime": 8.874,
2705
+ "eval_samples_per_second": 7.099,
2706
+ "eval_steps_per_second": 1.803,
2707
+ "eval_wer": 0.37916490932096164,
2708
+ "step": 3150
2709
+ },
2710
+ {
2711
+ "epoch": 90.28,
2712
+ "learning_rate": 3.988472622478386e-05,
2713
+ "loss": 0.3393,
2714
+ "step": 3160
2715
+ },
2716
+ {
2717
+ "epoch": 90.57,
2718
+ "learning_rate": 3.8731988472622485e-05,
2719
+ "loss": 0.2986,
2720
+ "step": 3170
2721
+ },
2722
+ {
2723
+ "epoch": 90.85,
2724
+ "learning_rate": 3.7579250720461096e-05,
2725
+ "loss": 0.4042,
2726
+ "step": 3180
2727
+ },
2728
+ {
2729
+ "epoch": 90.99,
2730
+ "eval_loss": 0.7587747573852539,
2731
+ "eval_runtime": 8.7904,
2732
+ "eval_samples_per_second": 7.167,
2733
+ "eval_steps_per_second": 1.82,
2734
+ "eval_wer": 0.3766343315056938,
2735
+ "step": 3185
2736
+ },
2737
+ {
2738
+ "epoch": 91.14,
2739
+ "learning_rate": 3.6426512968299714e-05,
2740
+ "loss": 0.3223,
2741
+ "step": 3190
2742
+ },
2743
+ {
2744
+ "epoch": 91.43,
2745
+ "learning_rate": 3.527377521613833e-05,
2746
+ "loss": 0.5178,
2747
+ "step": 3200
2748
+ },
2749
+ {
2750
+ "epoch": 91.71,
2751
+ "learning_rate": 3.412103746397695e-05,
2752
+ "loss": 0.3521,
2753
+ "step": 3210
2754
+ },
2755
+ {
2756
+ "epoch": 91.99,
2757
+ "learning_rate": 3.296829971181557e-05,
2758
+ "loss": 0.3989,
2759
+ "step": 3220
2760
+ },
2761
+ {
2762
+ "epoch": 91.99,
2763
+ "eval_loss": 0.7310720086097717,
2764
+ "eval_runtime": 8.8614,
2765
+ "eval_samples_per_second": 7.109,
2766
+ "eval_steps_per_second": 1.806,
2767
+ "eval_wer": 0.3774778574441164,
2768
+ "step": 3220
2769
+ },
2770
+ {
2771
+ "epoch": 92.28,
2772
+ "learning_rate": 3.181556195965418e-05,
2773
+ "loss": 0.329,
2774
+ "step": 3230
2775
+ },
2776
+ {
2777
+ "epoch": 92.57,
2778
+ "learning_rate": 3.0662824207492795e-05,
2779
+ "loss": 0.3594,
2780
+ "step": 3240
2781
+ },
2782
+ {
2783
+ "epoch": 92.85,
2784
+ "learning_rate": 2.9510086455331416e-05,
2785
+ "loss": 0.3625,
2786
+ "step": 3250
2787
+ },
2788
+ {
2789
+ "epoch": 92.99,
2790
+ "eval_loss": 0.7475365400314331,
2791
+ "eval_runtime": 8.8406,
2792
+ "eval_samples_per_second": 7.126,
2793
+ "eval_steps_per_second": 1.81,
2794
+ "eval_wer": 0.3745255166596373,
2795
+ "step": 3255
2796
+ },
2797
+ {
2798
+ "epoch": 93.14,
2799
+ "learning_rate": 2.8357348703170027e-05,
2800
+ "loss": 0.3361,
2801
+ "step": 3260
2802
+ },
2803
+ {
2804
+ "epoch": 93.43,
2805
+ "learning_rate": 2.7204610951008648e-05,
2806
+ "loss": 0.3521,
2807
+ "step": 3270
2808
+ },
2809
+ {
2810
+ "epoch": 93.71,
2811
+ "learning_rate": 2.6051873198847266e-05,
2812
+ "loss": 0.3261,
2813
+ "step": 3280
2814
+ },
2815
+ {
2816
+ "epoch": 93.99,
2817
+ "learning_rate": 2.489913544668588e-05,
2818
+ "loss": 0.3036,
2819
+ "step": 3290
2820
+ },
2821
+ {
2822
+ "epoch": 93.99,
2823
+ "eval_loss": 0.7138022184371948,
2824
+ "eval_runtime": 8.8395,
2825
+ "eval_samples_per_second": 7.127,
2826
+ "eval_steps_per_second": 1.81,
2827
+ "eval_wer": 0.37157317587515815,
2828
+ "step": 3290
2829
+ },
2830
+ {
2831
+ "epoch": 94.28,
2832
+ "learning_rate": 2.3746397694524498e-05,
2833
+ "loss": 0.3412,
2834
+ "step": 3300
2835
+ },
2836
+ {
2837
+ "epoch": 94.57,
2838
+ "learning_rate": 2.2593659942363112e-05,
2839
+ "loss": 0.3451,
2840
+ "step": 3310
2841
+ },
2842
+ {
2843
+ "epoch": 94.85,
2844
+ "learning_rate": 2.144092219020173e-05,
2845
+ "loss": 0.5157,
2846
+ "step": 3320
2847
+ },
2848
+ {
2849
+ "epoch": 94.99,
2850
+ "eval_loss": 0.7245524525642395,
2851
+ "eval_runtime": 8.8889,
2852
+ "eval_samples_per_second": 7.087,
2853
+ "eval_steps_per_second": 1.8,
2854
+ "eval_wer": 0.3787431463517503,
2855
+ "step": 3325
2856
+ },
2857
+ {
2858
+ "epoch": 95.14,
2859
+ "learning_rate": 2.0288184438040347e-05,
2860
+ "loss": 0.3089,
2861
+ "step": 3330
2862
+ },
2863
+ {
2864
+ "epoch": 95.43,
2865
+ "learning_rate": 1.9135446685878965e-05,
2866
+ "loss": 0.3115,
2867
+ "step": 3340
2868
+ },
2869
+ {
2870
+ "epoch": 95.71,
2871
+ "learning_rate": 1.798270893371758e-05,
2872
+ "loss": 0.3742,
2873
+ "step": 3350
2874
+ },
2875
+ {
2876
+ "epoch": 95.99,
2877
+ "learning_rate": 1.6829971181556197e-05,
2878
+ "loss": 0.4072,
2879
+ "step": 3360
2880
+ },
2881
+ {
2882
+ "epoch": 95.99,
2883
+ "eval_loss": 0.7321768999099731,
2884
+ "eval_runtime": 8.9645,
2885
+ "eval_samples_per_second": 7.028,
2886
+ "eval_steps_per_second": 1.785,
2887
+ "eval_wer": 0.3762125685364825,
2888
+ "step": 3360
2889
+ },
2890
+ {
2891
+ "epoch": 96.28,
2892
+ "learning_rate": 1.5677233429394814e-05,
2893
+ "loss": 0.3496,
2894
+ "step": 3370
2895
+ },
2896
+ {
2897
+ "epoch": 96.57,
2898
+ "learning_rate": 1.452449567723343e-05,
2899
+ "loss": 0.4089,
2900
+ "step": 3380
2901
+ },
2902
+ {
2903
+ "epoch": 96.85,
2904
+ "learning_rate": 1.3371757925072045e-05,
2905
+ "loss": 0.3406,
2906
+ "step": 3390
2907
+ },
2908
+ {
2909
+ "epoch": 96.99,
2910
+ "eval_loss": 0.7133801579475403,
2911
+ "eval_runtime": 8.9109,
2912
+ "eval_samples_per_second": 7.07,
2913
+ "eval_steps_per_second": 1.796,
2914
+ "eval_wer": 0.3770560944749051,
2915
+ "step": 3395
2916
+ },
2917
+ {
2918
+ "epoch": 97.14,
2919
+ "learning_rate": 1.2219020172910664e-05,
2920
+ "loss": 0.3305,
2921
+ "step": 3400
2922
+ },
2923
+ {
2924
+ "epoch": 97.43,
2925
+ "learning_rate": 1.106628242074928e-05,
2926
+ "loss": 0.3521,
2927
+ "step": 3410
2928
+ },
2929
+ {
2930
+ "epoch": 97.71,
2931
+ "learning_rate": 9.913544668587897e-06,
2932
+ "loss": 0.3405,
2933
+ "step": 3420
2934
+ },
2935
+ {
2936
+ "epoch": 97.99,
2937
+ "learning_rate": 8.760806916426513e-06,
2938
+ "loss": 0.2987,
2939
+ "step": 3430
2940
+ },
2941
+ {
2942
+ "epoch": 97.99,
2943
+ "eval_loss": 0.69514000415802,
2944
+ "eval_runtime": 8.9709,
2945
+ "eval_samples_per_second": 7.023,
2946
+ "eval_steps_per_second": 1.784,
2947
+ "eval_wer": 0.37536904259805987,
2948
+ "step": 3430
2949
+ },
2950
+ {
2951
+ "epoch": 98.28,
2952
+ "learning_rate": 7.60806916426513e-06,
2953
+ "loss": 0.3394,
2954
+ "step": 3440
2955
+ },
2956
+ {
2957
+ "epoch": 98.57,
2958
+ "learning_rate": 6.455331412103746e-06,
2959
+ "loss": 0.3566,
2960
+ "step": 3450
2961
+ },
2962
+ {
2963
+ "epoch": 98.85,
2964
+ "learning_rate": 5.302593659942364e-06,
2965
+ "loss": 0.3355,
2966
+ "step": 3460
2967
+ },
2968
+ {
2969
+ "epoch": 98.99,
2970
+ "eval_loss": 0.7004749178886414,
2971
+ "eval_runtime": 8.9802,
2972
+ "eval_samples_per_second": 7.015,
2973
+ "eval_steps_per_second": 1.782,
2974
+ "eval_wer": 0.3766343315056938,
2975
+ "step": 3465
2976
+ },
2977
+ {
2978
+ "epoch": 99.14,
2979
+ "learning_rate": 4.14985590778098e-06,
2980
+ "loss": 0.3434,
2981
+ "step": 3470
2982
+ },
2983
+ {
2984
+ "epoch": 99.43,
2985
+ "learning_rate": 2.9971181556195965e-06,
2986
+ "loss": 0.3309,
2987
+ "step": 3480
2988
+ },
2989
+ {
2990
+ "epoch": 99.71,
2991
+ "learning_rate": 1.8443804034582135e-06,
2992
+ "loss": 0.369,
2993
+ "step": 3490
2994
+ },
2995
+ {
2996
+ "epoch": 99.99,
2997
+ "learning_rate": 6.9164265129683e-07,
2998
+ "loss": 0.341,
2999
+ "step": 3500
3000
+ },
3001
+ {
3002
+ "epoch": 99.99,
3003
+ "eval_loss": 0.6989557147026062,
3004
+ "eval_runtime": 8.9073,
3005
+ "eval_samples_per_second": 7.073,
3006
+ "eval_steps_per_second": 1.796,
3007
+ "eval_wer": 0.3770560944749051,
3008
+ "step": 3500
3009
+ },
3010
+ {
3011
+ "epoch": 99.99,
3012
+ "step": 3500,
3013
+ "total_flos": 5.739635222597828e+18,
3014
+ "train_loss": 0.9362126126289367,
3015
+ "train_runtime": 11890.6212,
3016
+ "train_samples_per_second": 4.735,
3017
+ "train_steps_per_second": 0.294
3018
+ }
3019
+ ],
3020
+ "max_steps": 3500,
3021
+ "num_train_epochs": 100,
3022
+ "total_flos": 5.739635222597828e+18,
3023
+ "trial_name": null,
3024
+ "trial_params": null
3025
+ }