End of training
Browse files- all_results.json +12 -0
- eval_results.json +8 -0
- runs/Aug08_01-22-12_dsai-server/events.out.tfevents.1691439464.dsai-server +3 -0
- train_results.json +7 -0
- trainer_state.json +1270 -0
all_results.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 0.43,
|
3 |
+
"eval_loss": 0.8638723492622375,
|
4 |
+
"eval_runtime": 557.3961,
|
5 |
+
"eval_samples_per_second": 4.397,
|
6 |
+
"eval_steps_per_second": 2.2,
|
7 |
+
"eval_wer": 54.70462356526814,
|
8 |
+
"train_loss": 0.996151273727417,
|
9 |
+
"train_runtime": 5789.5442,
|
10 |
+
"train_samples_per_second": 3.455,
|
11 |
+
"train_steps_per_second": 0.864
|
12 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 0.43,
|
3 |
+
"eval_loss": 0.8638723492622375,
|
4 |
+
"eval_runtime": 557.3961,
|
5 |
+
"eval_samples_per_second": 4.397,
|
6 |
+
"eval_steps_per_second": 2.2,
|
7 |
+
"eval_wer": 54.70462356526814
|
8 |
+
}
|
runs/Aug08_01-22-12_dsai-server/events.out.tfevents.1691439464.dsai-server
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:eb50f2dfb204649304c12b2a1ecf9043986b27a1deedee2e175a2bfade459481
|
3 |
+
size 40
|
train_results.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 0.43,
|
3 |
+
"train_loss": 0.996151273727417,
|
4 |
+
"train_runtime": 5789.5442,
|
5 |
+
"train_samples_per_second": 3.455,
|
6 |
+
"train_steps_per_second": 0.864
|
7 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,1270 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 54.70462356526814,
|
3 |
+
"best_model_checkpoint": "./output/small/yt-special-batch4-lr6-small/checkpoint-5000",
|
4 |
+
"epoch": 0.4294425835265825,
|
5 |
+
"global_step": 5000,
|
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": 4e-08,
|
13 |
+
"loss": 2.4854,
|
14 |
+
"step": 25
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.0,
|
18 |
+
"learning_rate": 9e-08,
|
19 |
+
"loss": 2.3311,
|
20 |
+
"step": 50
|
21 |
+
},
|
22 |
+
{
|
23 |
+
"epoch": 0.01,
|
24 |
+
"learning_rate": 1.4e-07,
|
25 |
+
"loss": 2.217,
|
26 |
+
"step": 75
|
27 |
+
},
|
28 |
+
{
|
29 |
+
"epoch": 0.01,
|
30 |
+
"learning_rate": 1.8999999999999998e-07,
|
31 |
+
"loss": 2.3064,
|
32 |
+
"step": 100
|
33 |
+
},
|
34 |
+
{
|
35 |
+
"epoch": 0.01,
|
36 |
+
"learning_rate": 2.4e-07,
|
37 |
+
"loss": 2.4084,
|
38 |
+
"step": 125
|
39 |
+
},
|
40 |
+
{
|
41 |
+
"epoch": 0.01,
|
42 |
+
"learning_rate": 2.9e-07,
|
43 |
+
"loss": 2.299,
|
44 |
+
"step": 150
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.02,
|
48 |
+
"learning_rate": 3.4000000000000003e-07,
|
49 |
+
"loss": 1.8911,
|
50 |
+
"step": 175
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.02,
|
54 |
+
"learning_rate": 3.8999999999999997e-07,
|
55 |
+
"loss": 2.0025,
|
56 |
+
"step": 200
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.02,
|
60 |
+
"learning_rate": 4.3999999999999997e-07,
|
61 |
+
"loss": 1.917,
|
62 |
+
"step": 225
|
63 |
+
},
|
64 |
+
{
|
65 |
+
"epoch": 0.02,
|
66 |
+
"learning_rate": 4.9e-07,
|
67 |
+
"loss": 1.8306,
|
68 |
+
"step": 250
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"epoch": 0.02,
|
72 |
+
"learning_rate": 5.4e-07,
|
73 |
+
"loss": 1.9323,
|
74 |
+
"step": 275
|
75 |
+
},
|
76 |
+
{
|
77 |
+
"epoch": 0.03,
|
78 |
+
"learning_rate": 5.9e-07,
|
79 |
+
"loss": 1.5776,
|
80 |
+
"step": 300
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 0.03,
|
84 |
+
"learning_rate": 6.4e-07,
|
85 |
+
"loss": 1.4313,
|
86 |
+
"step": 325
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.03,
|
90 |
+
"learning_rate": 6.9e-07,
|
91 |
+
"loss": 1.3474,
|
92 |
+
"step": 350
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.03,
|
96 |
+
"learning_rate": 7.4e-07,
|
97 |
+
"loss": 1.29,
|
98 |
+
"step": 375
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.03,
|
102 |
+
"learning_rate": 7.9e-07,
|
103 |
+
"loss": 1.1974,
|
104 |
+
"step": 400
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 0.04,
|
108 |
+
"learning_rate": 8.399999999999999e-07,
|
109 |
+
"loss": 1.299,
|
110 |
+
"step": 425
|
111 |
+
},
|
112 |
+
{
|
113 |
+
"epoch": 0.04,
|
114 |
+
"learning_rate": 8.9e-07,
|
115 |
+
"loss": 1.1418,
|
116 |
+
"step": 450
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 0.04,
|
120 |
+
"learning_rate": 9.399999999999999e-07,
|
121 |
+
"loss": 1.205,
|
122 |
+
"step": 475
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 0.04,
|
126 |
+
"learning_rate": 9.9e-07,
|
127 |
+
"loss": 0.9866,
|
128 |
+
"step": 500
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.05,
|
132 |
+
"learning_rate": 9.955555555555554e-07,
|
133 |
+
"loss": 0.9506,
|
134 |
+
"step": 525
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 0.05,
|
138 |
+
"learning_rate": 9.9e-07,
|
139 |
+
"loss": 1.2328,
|
140 |
+
"step": 550
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 0.05,
|
144 |
+
"learning_rate": 9.844444444444445e-07,
|
145 |
+
"loss": 1.1429,
|
146 |
+
"step": 575
|
147 |
+
},
|
148 |
+
{
|
149 |
+
"epoch": 0.05,
|
150 |
+
"learning_rate": 9.788888888888889e-07,
|
151 |
+
"loss": 1.005,
|
152 |
+
"step": 600
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 0.05,
|
156 |
+
"learning_rate": 9.733333333333333e-07,
|
157 |
+
"loss": 0.9108,
|
158 |
+
"step": 625
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 0.06,
|
162 |
+
"learning_rate": 9.677777777777777e-07,
|
163 |
+
"loss": 0.9505,
|
164 |
+
"step": 650
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 0.06,
|
168 |
+
"learning_rate": 9.62222222222222e-07,
|
169 |
+
"loss": 1.1187,
|
170 |
+
"step": 675
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.06,
|
174 |
+
"learning_rate": 9.566666666666667e-07,
|
175 |
+
"loss": 0.9913,
|
176 |
+
"step": 700
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 0.06,
|
180 |
+
"learning_rate": 9.511111111111111e-07,
|
181 |
+
"loss": 1.132,
|
182 |
+
"step": 725
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 0.06,
|
186 |
+
"learning_rate": 9.455555555555556e-07,
|
187 |
+
"loss": 1.1477,
|
188 |
+
"step": 750
|
189 |
+
},
|
190 |
+
{
|
191 |
+
"epoch": 0.07,
|
192 |
+
"learning_rate": 9.399999999999999e-07,
|
193 |
+
"loss": 0.9871,
|
194 |
+
"step": 775
|
195 |
+
},
|
196 |
+
{
|
197 |
+
"epoch": 0.07,
|
198 |
+
"learning_rate": 9.344444444444444e-07,
|
199 |
+
"loss": 1.024,
|
200 |
+
"step": 800
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 0.07,
|
204 |
+
"learning_rate": 9.288888888888888e-07,
|
205 |
+
"loss": 0.8419,
|
206 |
+
"step": 825
|
207 |
+
},
|
208 |
+
{
|
209 |
+
"epoch": 0.07,
|
210 |
+
"learning_rate": 9.233333333333333e-07,
|
211 |
+
"loss": 0.896,
|
212 |
+
"step": 850
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.08,
|
216 |
+
"learning_rate": 9.177777777777777e-07,
|
217 |
+
"loss": 1.0737,
|
218 |
+
"step": 875
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 0.08,
|
222 |
+
"learning_rate": 9.122222222222222e-07,
|
223 |
+
"loss": 1.057,
|
224 |
+
"step": 900
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 0.08,
|
228 |
+
"learning_rate": 9.066666666666665e-07,
|
229 |
+
"loss": 0.9556,
|
230 |
+
"step": 925
|
231 |
+
},
|
232 |
+
{
|
233 |
+
"epoch": 0.08,
|
234 |
+
"learning_rate": 9.01111111111111e-07,
|
235 |
+
"loss": 1.0581,
|
236 |
+
"step": 950
|
237 |
+
},
|
238 |
+
{
|
239 |
+
"epoch": 0.08,
|
240 |
+
"learning_rate": 8.955555555555555e-07,
|
241 |
+
"loss": 0.8865,
|
242 |
+
"step": 975
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 0.09,
|
246 |
+
"learning_rate": 8.9e-07,
|
247 |
+
"loss": 1.1374,
|
248 |
+
"step": 1000
|
249 |
+
},
|
250 |
+
{
|
251 |
+
"epoch": 0.09,
|
252 |
+
"eval_loss": 0.9853608012199402,
|
253 |
+
"eval_runtime": 596.2346,
|
254 |
+
"eval_samples_per_second": 4.111,
|
255 |
+
"eval_steps_per_second": 2.056,
|
256 |
+
"eval_wer": 64.96336164969846,
|
257 |
+
"step": 1000
|
258 |
+
},
|
259 |
+
{
|
260 |
+
"epoch": 0.09,
|
261 |
+
"learning_rate": 8.844444444444445e-07,
|
262 |
+
"loss": 0.9606,
|
263 |
+
"step": 1025
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 0.09,
|
267 |
+
"learning_rate": 8.788888888888889e-07,
|
268 |
+
"loss": 0.9526,
|
269 |
+
"step": 1050
|
270 |
+
},
|
271 |
+
{
|
272 |
+
"epoch": 0.09,
|
273 |
+
"learning_rate": 8.733333333333333e-07,
|
274 |
+
"loss": 1.0831,
|
275 |
+
"step": 1075
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.09,
|
279 |
+
"learning_rate": 8.677777777777777e-07,
|
280 |
+
"loss": 0.8729,
|
281 |
+
"step": 1100
|
282 |
+
},
|
283 |
+
{
|
284 |
+
"epoch": 0.1,
|
285 |
+
"learning_rate": 8.622222222222222e-07,
|
286 |
+
"loss": 0.977,
|
287 |
+
"step": 1125
|
288 |
+
},
|
289 |
+
{
|
290 |
+
"epoch": 0.1,
|
291 |
+
"learning_rate": 8.566666666666667e-07,
|
292 |
+
"loss": 0.9041,
|
293 |
+
"step": 1150
|
294 |
+
},
|
295 |
+
{
|
296 |
+
"epoch": 0.1,
|
297 |
+
"learning_rate": 8.511111111111111e-07,
|
298 |
+
"loss": 1.0707,
|
299 |
+
"step": 1175
|
300 |
+
},
|
301 |
+
{
|
302 |
+
"epoch": 0.1,
|
303 |
+
"learning_rate": 8.455555555555555e-07,
|
304 |
+
"loss": 1.093,
|
305 |
+
"step": 1200
|
306 |
+
},
|
307 |
+
{
|
308 |
+
"epoch": 0.11,
|
309 |
+
"learning_rate": 8.399999999999999e-07,
|
310 |
+
"loss": 1.102,
|
311 |
+
"step": 1225
|
312 |
+
},
|
313 |
+
{
|
314 |
+
"epoch": 0.11,
|
315 |
+
"learning_rate": 8.344444444444444e-07,
|
316 |
+
"loss": 0.8571,
|
317 |
+
"step": 1250
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.11,
|
321 |
+
"learning_rate": 8.288888888888888e-07,
|
322 |
+
"loss": 1.0215,
|
323 |
+
"step": 1275
|
324 |
+
},
|
325 |
+
{
|
326 |
+
"epoch": 0.11,
|
327 |
+
"learning_rate": 8.233333333333333e-07,
|
328 |
+
"loss": 0.9254,
|
329 |
+
"step": 1300
|
330 |
+
},
|
331 |
+
{
|
332 |
+
"epoch": 0.11,
|
333 |
+
"learning_rate": 8.177777777777778e-07,
|
334 |
+
"loss": 1.0406,
|
335 |
+
"step": 1325
|
336 |
+
},
|
337 |
+
{
|
338 |
+
"epoch": 0.12,
|
339 |
+
"learning_rate": 8.122222222222221e-07,
|
340 |
+
"loss": 0.9516,
|
341 |
+
"step": 1350
|
342 |
+
},
|
343 |
+
{
|
344 |
+
"epoch": 0.12,
|
345 |
+
"learning_rate": 8.066666666666666e-07,
|
346 |
+
"loss": 0.9585,
|
347 |
+
"step": 1375
|
348 |
+
},
|
349 |
+
{
|
350 |
+
"epoch": 0.12,
|
351 |
+
"learning_rate": 8.01111111111111e-07,
|
352 |
+
"loss": 0.7759,
|
353 |
+
"step": 1400
|
354 |
+
},
|
355 |
+
{
|
356 |
+
"epoch": 0.12,
|
357 |
+
"learning_rate": 7.955555555555556e-07,
|
358 |
+
"loss": 0.919,
|
359 |
+
"step": 1425
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 0.12,
|
363 |
+
"learning_rate": 7.9e-07,
|
364 |
+
"loss": 0.9281,
|
365 |
+
"step": 1450
|
366 |
+
},
|
367 |
+
{
|
368 |
+
"epoch": 0.13,
|
369 |
+
"learning_rate": 7.844444444444445e-07,
|
370 |
+
"loss": 1.0186,
|
371 |
+
"step": 1475
|
372 |
+
},
|
373 |
+
{
|
374 |
+
"epoch": 0.13,
|
375 |
+
"learning_rate": 7.788888888888889e-07,
|
376 |
+
"loss": 0.8863,
|
377 |
+
"step": 1500
|
378 |
+
},
|
379 |
+
{
|
380 |
+
"epoch": 0.13,
|
381 |
+
"learning_rate": 7.733333333333333e-07,
|
382 |
+
"loss": 0.96,
|
383 |
+
"step": 1525
|
384 |
+
},
|
385 |
+
{
|
386 |
+
"epoch": 0.13,
|
387 |
+
"learning_rate": 7.677777777777778e-07,
|
388 |
+
"loss": 0.7853,
|
389 |
+
"step": 1550
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"epoch": 0.14,
|
393 |
+
"learning_rate": 7.622222222222222e-07,
|
394 |
+
"loss": 0.9674,
|
395 |
+
"step": 1575
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 0.14,
|
399 |
+
"learning_rate": 7.566666666666667e-07,
|
400 |
+
"loss": 0.8892,
|
401 |
+
"step": 1600
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 0.14,
|
405 |
+
"learning_rate": 7.51111111111111e-07,
|
406 |
+
"loss": 0.8938,
|
407 |
+
"step": 1625
|
408 |
+
},
|
409 |
+
{
|
410 |
+
"epoch": 0.14,
|
411 |
+
"learning_rate": 7.455555555555555e-07,
|
412 |
+
"loss": 0.8152,
|
413 |
+
"step": 1650
|
414 |
+
},
|
415 |
+
{
|
416 |
+
"epoch": 0.14,
|
417 |
+
"learning_rate": 7.4e-07,
|
418 |
+
"loss": 0.8955,
|
419 |
+
"step": 1675
|
420 |
+
},
|
421 |
+
{
|
422 |
+
"epoch": 0.15,
|
423 |
+
"learning_rate": 7.344444444444444e-07,
|
424 |
+
"loss": 0.8278,
|
425 |
+
"step": 1700
|
426 |
+
},
|
427 |
+
{
|
428 |
+
"epoch": 0.15,
|
429 |
+
"learning_rate": 7.288888888888889e-07,
|
430 |
+
"loss": 0.9115,
|
431 |
+
"step": 1725
|
432 |
+
},
|
433 |
+
{
|
434 |
+
"epoch": 0.15,
|
435 |
+
"learning_rate": 7.233333333333333e-07,
|
436 |
+
"loss": 0.9066,
|
437 |
+
"step": 1750
|
438 |
+
},
|
439 |
+
{
|
440 |
+
"epoch": 0.15,
|
441 |
+
"learning_rate": 7.177777777777777e-07,
|
442 |
+
"loss": 1.0192,
|
443 |
+
"step": 1775
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 0.15,
|
447 |
+
"learning_rate": 7.122222222222221e-07,
|
448 |
+
"loss": 0.9415,
|
449 |
+
"step": 1800
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 0.16,
|
453 |
+
"learning_rate": 7.066666666666666e-07,
|
454 |
+
"loss": 1.0481,
|
455 |
+
"step": 1825
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"epoch": 0.16,
|
459 |
+
"learning_rate": 7.01111111111111e-07,
|
460 |
+
"loss": 0.8349,
|
461 |
+
"step": 1850
|
462 |
+
},
|
463 |
+
{
|
464 |
+
"epoch": 0.16,
|
465 |
+
"learning_rate": 6.955555555555556e-07,
|
466 |
+
"loss": 0.8967,
|
467 |
+
"step": 1875
|
468 |
+
},
|
469 |
+
{
|
470 |
+
"epoch": 0.16,
|
471 |
+
"learning_rate": 6.9e-07,
|
472 |
+
"loss": 0.8016,
|
473 |
+
"step": 1900
|
474 |
+
},
|
475 |
+
{
|
476 |
+
"epoch": 0.17,
|
477 |
+
"learning_rate": 6.844444444444444e-07,
|
478 |
+
"loss": 0.9607,
|
479 |
+
"step": 1925
|
480 |
+
},
|
481 |
+
{
|
482 |
+
"epoch": 0.17,
|
483 |
+
"learning_rate": 6.788888888888889e-07,
|
484 |
+
"loss": 0.8532,
|
485 |
+
"step": 1950
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 0.17,
|
489 |
+
"learning_rate": 6.733333333333333e-07,
|
490 |
+
"loss": 0.9257,
|
491 |
+
"step": 1975
|
492 |
+
},
|
493 |
+
{
|
494 |
+
"epoch": 0.17,
|
495 |
+
"learning_rate": 6.677777777777778e-07,
|
496 |
+
"loss": 0.8775,
|
497 |
+
"step": 2000
|
498 |
+
},
|
499 |
+
{
|
500 |
+
"epoch": 0.17,
|
501 |
+
"eval_loss": 0.9139208793640137,
|
502 |
+
"eval_runtime": 610.1359,
|
503 |
+
"eval_samples_per_second": 4.017,
|
504 |
+
"eval_steps_per_second": 2.009,
|
505 |
+
"eval_wer": 66.46131898061086,
|
506 |
+
"step": 2000
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 0.17,
|
510 |
+
"learning_rate": 6.622222222222222e-07,
|
511 |
+
"loss": 0.8847,
|
512 |
+
"step": 2025
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 0.18,
|
516 |
+
"learning_rate": 6.566666666666666e-07,
|
517 |
+
"loss": 1.0419,
|
518 |
+
"step": 2050
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"epoch": 0.18,
|
522 |
+
"learning_rate": 6.511111111111111e-07,
|
523 |
+
"loss": 0.8739,
|
524 |
+
"step": 2075
|
525 |
+
},
|
526 |
+
{
|
527 |
+
"epoch": 0.18,
|
528 |
+
"learning_rate": 6.455555555555555e-07,
|
529 |
+
"loss": 0.8261,
|
530 |
+
"step": 2100
|
531 |
+
},
|
532 |
+
{
|
533 |
+
"epoch": 0.18,
|
534 |
+
"learning_rate": 6.4e-07,
|
535 |
+
"loss": 0.9817,
|
536 |
+
"step": 2125
|
537 |
+
},
|
538 |
+
{
|
539 |
+
"epoch": 0.18,
|
540 |
+
"learning_rate": 6.344444444444444e-07,
|
541 |
+
"loss": 1.0232,
|
542 |
+
"step": 2150
|
543 |
+
},
|
544 |
+
{
|
545 |
+
"epoch": 0.19,
|
546 |
+
"learning_rate": 6.288888888888889e-07,
|
547 |
+
"loss": 0.8482,
|
548 |
+
"step": 2175
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 0.19,
|
552 |
+
"learning_rate": 6.233333333333332e-07,
|
553 |
+
"loss": 0.8584,
|
554 |
+
"step": 2200
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 0.19,
|
558 |
+
"learning_rate": 6.177777777777777e-07,
|
559 |
+
"loss": 0.8966,
|
560 |
+
"step": 2225
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 0.19,
|
564 |
+
"learning_rate": 6.122222222222222e-07,
|
565 |
+
"loss": 0.8923,
|
566 |
+
"step": 2250
|
567 |
+
},
|
568 |
+
{
|
569 |
+
"epoch": 0.2,
|
570 |
+
"learning_rate": 6.066666666666666e-07,
|
571 |
+
"loss": 0.9748,
|
572 |
+
"step": 2275
|
573 |
+
},
|
574 |
+
{
|
575 |
+
"epoch": 0.2,
|
576 |
+
"learning_rate": 6.011111111111112e-07,
|
577 |
+
"loss": 1.1212,
|
578 |
+
"step": 2300
|
579 |
+
},
|
580 |
+
{
|
581 |
+
"epoch": 0.2,
|
582 |
+
"learning_rate": 5.955555555555555e-07,
|
583 |
+
"loss": 0.9538,
|
584 |
+
"step": 2325
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 0.2,
|
588 |
+
"learning_rate": 5.9e-07,
|
589 |
+
"loss": 1.1401,
|
590 |
+
"step": 2350
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 0.2,
|
594 |
+
"learning_rate": 5.844444444444444e-07,
|
595 |
+
"loss": 0.9019,
|
596 |
+
"step": 2375
|
597 |
+
},
|
598 |
+
{
|
599 |
+
"epoch": 0.21,
|
600 |
+
"learning_rate": 5.788888888888889e-07,
|
601 |
+
"loss": 0.8402,
|
602 |
+
"step": 2400
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 0.21,
|
606 |
+
"learning_rate": 5.733333333333334e-07,
|
607 |
+
"loss": 0.8792,
|
608 |
+
"step": 2425
|
609 |
+
},
|
610 |
+
{
|
611 |
+
"epoch": 0.21,
|
612 |
+
"learning_rate": 5.677777777777778e-07,
|
613 |
+
"loss": 0.9592,
|
614 |
+
"step": 2450
|
615 |
+
},
|
616 |
+
{
|
617 |
+
"epoch": 0.21,
|
618 |
+
"learning_rate": 5.622222222222222e-07,
|
619 |
+
"loss": 0.9085,
|
620 |
+
"step": 2475
|
621 |
+
},
|
622 |
+
{
|
623 |
+
"epoch": 0.21,
|
624 |
+
"learning_rate": 5.566666666666666e-07,
|
625 |
+
"loss": 1.0189,
|
626 |
+
"step": 2500
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 0.22,
|
630 |
+
"learning_rate": 5.511111111111111e-07,
|
631 |
+
"loss": 0.8341,
|
632 |
+
"step": 2525
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 0.22,
|
636 |
+
"learning_rate": 5.455555555555555e-07,
|
637 |
+
"loss": 0.7564,
|
638 |
+
"step": 2550
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 0.22,
|
642 |
+
"learning_rate": 5.4e-07,
|
643 |
+
"loss": 0.7948,
|
644 |
+
"step": 2575
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 0.22,
|
648 |
+
"learning_rate": 5.344444444444445e-07,
|
649 |
+
"loss": 1.0225,
|
650 |
+
"step": 2600
|
651 |
+
},
|
652 |
+
{
|
653 |
+
"epoch": 0.23,
|
654 |
+
"learning_rate": 5.288888888888888e-07,
|
655 |
+
"loss": 0.9563,
|
656 |
+
"step": 2625
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 0.23,
|
660 |
+
"learning_rate": 5.233333333333333e-07,
|
661 |
+
"loss": 0.7843,
|
662 |
+
"step": 2650
|
663 |
+
},
|
664 |
+
{
|
665 |
+
"epoch": 0.23,
|
666 |
+
"learning_rate": 5.177777777777777e-07,
|
667 |
+
"loss": 1.0565,
|
668 |
+
"step": 2675
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 0.23,
|
672 |
+
"learning_rate": 5.122222222222222e-07,
|
673 |
+
"loss": 0.7467,
|
674 |
+
"step": 2700
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 0.23,
|
678 |
+
"learning_rate": 5.066666666666667e-07,
|
679 |
+
"loss": 0.9035,
|
680 |
+
"step": 2725
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 0.24,
|
684 |
+
"learning_rate": 5.011111111111111e-07,
|
685 |
+
"loss": 0.8723,
|
686 |
+
"step": 2750
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 0.24,
|
690 |
+
"learning_rate": 4.955555555555556e-07,
|
691 |
+
"loss": 0.8477,
|
692 |
+
"step": 2775
|
693 |
+
},
|
694 |
+
{
|
695 |
+
"epoch": 0.24,
|
696 |
+
"learning_rate": 4.9e-07,
|
697 |
+
"loss": 0.8342,
|
698 |
+
"step": 2800
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 0.24,
|
702 |
+
"learning_rate": 4.844444444444445e-07,
|
703 |
+
"loss": 0.7835,
|
704 |
+
"step": 2825
|
705 |
+
},
|
706 |
+
{
|
707 |
+
"epoch": 0.24,
|
708 |
+
"learning_rate": 4.788888888888889e-07,
|
709 |
+
"loss": 1.0936,
|
710 |
+
"step": 2850
|
711 |
+
},
|
712 |
+
{
|
713 |
+
"epoch": 0.25,
|
714 |
+
"learning_rate": 4.733333333333333e-07,
|
715 |
+
"loss": 0.9851,
|
716 |
+
"step": 2875
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 0.25,
|
720 |
+
"learning_rate": 4.677777777777778e-07,
|
721 |
+
"loss": 0.7425,
|
722 |
+
"step": 2900
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 0.25,
|
726 |
+
"learning_rate": 4.622222222222222e-07,
|
727 |
+
"loss": 0.9534,
|
728 |
+
"step": 2925
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 0.25,
|
732 |
+
"learning_rate": 4.5666666666666665e-07,
|
733 |
+
"loss": 0.9309,
|
734 |
+
"step": 2950
|
735 |
+
},
|
736 |
+
{
|
737 |
+
"epoch": 0.26,
|
738 |
+
"learning_rate": 4.511111111111111e-07,
|
739 |
+
"loss": 0.7555,
|
740 |
+
"step": 2975
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 0.26,
|
744 |
+
"learning_rate": 4.455555555555555e-07,
|
745 |
+
"loss": 0.9735,
|
746 |
+
"step": 3000
|
747 |
+
},
|
748 |
+
{
|
749 |
+
"epoch": 0.26,
|
750 |
+
"eval_loss": 0.8845030665397644,
|
751 |
+
"eval_runtime": 580.1372,
|
752 |
+
"eval_samples_per_second": 4.225,
|
753 |
+
"eval_steps_per_second": 2.113,
|
754 |
+
"eval_wer": 58.66675312885027,
|
755 |
+
"step": 3000
|
756 |
+
},
|
757 |
+
{
|
758 |
+
"epoch": 0.26,
|
759 |
+
"learning_rate": 4.3999999999999997e-07,
|
760 |
+
"loss": 0.9725,
|
761 |
+
"step": 3025
|
762 |
+
},
|
763 |
+
{
|
764 |
+
"epoch": 0.26,
|
765 |
+
"learning_rate": 4.344444444444444e-07,
|
766 |
+
"loss": 1.0319,
|
767 |
+
"step": 3050
|
768 |
+
},
|
769 |
+
{
|
770 |
+
"epoch": 0.26,
|
771 |
+
"learning_rate": 4.2888888888888883e-07,
|
772 |
+
"loss": 0.8079,
|
773 |
+
"step": 3075
|
774 |
+
},
|
775 |
+
{
|
776 |
+
"epoch": 0.27,
|
777 |
+
"learning_rate": 4.2333333333333334e-07,
|
778 |
+
"loss": 0.8567,
|
779 |
+
"step": 3100
|
780 |
+
},
|
781 |
+
{
|
782 |
+
"epoch": 0.27,
|
783 |
+
"learning_rate": 4.177777777777778e-07,
|
784 |
+
"loss": 0.9613,
|
785 |
+
"step": 3125
|
786 |
+
},
|
787 |
+
{
|
788 |
+
"epoch": 0.27,
|
789 |
+
"learning_rate": 4.122222222222222e-07,
|
790 |
+
"loss": 0.9328,
|
791 |
+
"step": 3150
|
792 |
+
},
|
793 |
+
{
|
794 |
+
"epoch": 0.27,
|
795 |
+
"learning_rate": 4.0666666666666666e-07,
|
796 |
+
"loss": 0.8802,
|
797 |
+
"step": 3175
|
798 |
+
},
|
799 |
+
{
|
800 |
+
"epoch": 0.27,
|
801 |
+
"learning_rate": 4.0111111111111106e-07,
|
802 |
+
"loss": 0.8767,
|
803 |
+
"step": 3200
|
804 |
+
},
|
805 |
+
{
|
806 |
+
"epoch": 0.28,
|
807 |
+
"learning_rate": 3.955555555555555e-07,
|
808 |
+
"loss": 0.7442,
|
809 |
+
"step": 3225
|
810 |
+
},
|
811 |
+
{
|
812 |
+
"epoch": 0.28,
|
813 |
+
"learning_rate": 3.8999999999999997e-07,
|
814 |
+
"loss": 0.8729,
|
815 |
+
"step": 3250
|
816 |
+
},
|
817 |
+
{
|
818 |
+
"epoch": 0.28,
|
819 |
+
"learning_rate": 3.8444444444444443e-07,
|
820 |
+
"loss": 0.9029,
|
821 |
+
"step": 3275
|
822 |
+
},
|
823 |
+
{
|
824 |
+
"epoch": 0.28,
|
825 |
+
"learning_rate": 3.788888888888889e-07,
|
826 |
+
"loss": 0.8988,
|
827 |
+
"step": 3300
|
828 |
+
},
|
829 |
+
{
|
830 |
+
"epoch": 0.29,
|
831 |
+
"learning_rate": 3.7333333333333334e-07,
|
832 |
+
"loss": 0.8658,
|
833 |
+
"step": 3325
|
834 |
+
},
|
835 |
+
{
|
836 |
+
"epoch": 0.29,
|
837 |
+
"learning_rate": 3.6777777777777774e-07,
|
838 |
+
"loss": 0.9137,
|
839 |
+
"step": 3350
|
840 |
+
},
|
841 |
+
{
|
842 |
+
"epoch": 0.29,
|
843 |
+
"learning_rate": 3.622222222222222e-07,
|
844 |
+
"loss": 0.8178,
|
845 |
+
"step": 3375
|
846 |
+
},
|
847 |
+
{
|
848 |
+
"epoch": 0.29,
|
849 |
+
"learning_rate": 3.5666666666666666e-07,
|
850 |
+
"loss": 0.8481,
|
851 |
+
"step": 3400
|
852 |
+
},
|
853 |
+
{
|
854 |
+
"epoch": 0.29,
|
855 |
+
"learning_rate": 3.5111111111111106e-07,
|
856 |
+
"loss": 0.8868,
|
857 |
+
"step": 3425
|
858 |
+
},
|
859 |
+
{
|
860 |
+
"epoch": 0.3,
|
861 |
+
"learning_rate": 3.4555555555555557e-07,
|
862 |
+
"loss": 0.7846,
|
863 |
+
"step": 3450
|
864 |
+
},
|
865 |
+
{
|
866 |
+
"epoch": 0.3,
|
867 |
+
"learning_rate": 3.4000000000000003e-07,
|
868 |
+
"loss": 0.8963,
|
869 |
+
"step": 3475
|
870 |
+
},
|
871 |
+
{
|
872 |
+
"epoch": 0.3,
|
873 |
+
"learning_rate": 3.3444444444444443e-07,
|
874 |
+
"loss": 0.7407,
|
875 |
+
"step": 3500
|
876 |
+
},
|
877 |
+
{
|
878 |
+
"epoch": 0.3,
|
879 |
+
"learning_rate": 3.288888888888889e-07,
|
880 |
+
"loss": 0.9731,
|
881 |
+
"step": 3525
|
882 |
+
},
|
883 |
+
{
|
884 |
+
"epoch": 0.3,
|
885 |
+
"learning_rate": 3.233333333333333e-07,
|
886 |
+
"loss": 0.8753,
|
887 |
+
"step": 3550
|
888 |
+
},
|
889 |
+
{
|
890 |
+
"epoch": 0.31,
|
891 |
+
"learning_rate": 3.1777777777777775e-07,
|
892 |
+
"loss": 0.885,
|
893 |
+
"step": 3575
|
894 |
+
},
|
895 |
+
{
|
896 |
+
"epoch": 0.31,
|
897 |
+
"learning_rate": 3.122222222222222e-07,
|
898 |
+
"loss": 0.7584,
|
899 |
+
"step": 3600
|
900 |
+
},
|
901 |
+
{
|
902 |
+
"epoch": 0.31,
|
903 |
+
"learning_rate": 3.066666666666666e-07,
|
904 |
+
"loss": 0.8814,
|
905 |
+
"step": 3625
|
906 |
+
},
|
907 |
+
{
|
908 |
+
"epoch": 0.31,
|
909 |
+
"learning_rate": 3.011111111111111e-07,
|
910 |
+
"loss": 0.9843,
|
911 |
+
"step": 3650
|
912 |
+
},
|
913 |
+
{
|
914 |
+
"epoch": 0.32,
|
915 |
+
"learning_rate": 2.9555555555555557e-07,
|
916 |
+
"loss": 0.9026,
|
917 |
+
"step": 3675
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 0.32,
|
921 |
+
"learning_rate": 2.9e-07,
|
922 |
+
"loss": 0.9259,
|
923 |
+
"step": 3700
|
924 |
+
},
|
925 |
+
{
|
926 |
+
"epoch": 0.32,
|
927 |
+
"learning_rate": 2.8444444444444443e-07,
|
928 |
+
"loss": 0.8303,
|
929 |
+
"step": 3725
|
930 |
+
},
|
931 |
+
{
|
932 |
+
"epoch": 0.32,
|
933 |
+
"learning_rate": 2.788888888888889e-07,
|
934 |
+
"loss": 0.7775,
|
935 |
+
"step": 3750
|
936 |
+
},
|
937 |
+
{
|
938 |
+
"epoch": 0.32,
|
939 |
+
"learning_rate": 2.733333333333333e-07,
|
940 |
+
"loss": 0.9112,
|
941 |
+
"step": 3775
|
942 |
+
},
|
943 |
+
{
|
944 |
+
"epoch": 0.33,
|
945 |
+
"learning_rate": 2.6777777777777775e-07,
|
946 |
+
"loss": 0.8211,
|
947 |
+
"step": 3800
|
948 |
+
},
|
949 |
+
{
|
950 |
+
"epoch": 0.33,
|
951 |
+
"learning_rate": 2.6222222222222226e-07,
|
952 |
+
"loss": 0.9493,
|
953 |
+
"step": 3825
|
954 |
+
},
|
955 |
+
{
|
956 |
+
"epoch": 0.33,
|
957 |
+
"learning_rate": 2.5666666666666666e-07,
|
958 |
+
"loss": 0.9079,
|
959 |
+
"step": 3850
|
960 |
+
},
|
961 |
+
{
|
962 |
+
"epoch": 0.33,
|
963 |
+
"learning_rate": 2.511111111111111e-07,
|
964 |
+
"loss": 0.8974,
|
965 |
+
"step": 3875
|
966 |
+
},
|
967 |
+
{
|
968 |
+
"epoch": 0.33,
|
969 |
+
"learning_rate": 2.455555555555555e-07,
|
970 |
+
"loss": 0.9766,
|
971 |
+
"step": 3900
|
972 |
+
},
|
973 |
+
{
|
974 |
+
"epoch": 0.34,
|
975 |
+
"learning_rate": 2.4e-07,
|
976 |
+
"loss": 0.9226,
|
977 |
+
"step": 3925
|
978 |
+
},
|
979 |
+
{
|
980 |
+
"epoch": 0.34,
|
981 |
+
"learning_rate": 2.3444444444444444e-07,
|
982 |
+
"loss": 0.7136,
|
983 |
+
"step": 3950
|
984 |
+
},
|
985 |
+
{
|
986 |
+
"epoch": 0.34,
|
987 |
+
"learning_rate": 2.288888888888889e-07,
|
988 |
+
"loss": 0.9056,
|
989 |
+
"step": 3975
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 0.34,
|
993 |
+
"learning_rate": 2.2333333333333332e-07,
|
994 |
+
"loss": 0.8359,
|
995 |
+
"step": 4000
|
996 |
+
},
|
997 |
+
{
|
998 |
+
"epoch": 0.34,
|
999 |
+
"eval_loss": 0.8695711493492126,
|
1000 |
+
"eval_runtime": 584.3999,
|
1001 |
+
"eval_samples_per_second": 4.194,
|
1002 |
+
"eval_steps_per_second": 2.098,
|
1003 |
+
"eval_wer": 59.58757538421633,
|
1004 |
+
"step": 4000
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 0.35,
|
1008 |
+
"learning_rate": 2.1777777777777775e-07,
|
1009 |
+
"loss": 0.9031,
|
1010 |
+
"step": 4025
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 0.35,
|
1014 |
+
"learning_rate": 2.122222222222222e-07,
|
1015 |
+
"loss": 0.8199,
|
1016 |
+
"step": 4050
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 0.35,
|
1020 |
+
"learning_rate": 2.0666666666666666e-07,
|
1021 |
+
"loss": 0.8729,
|
1022 |
+
"step": 4075
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 0.35,
|
1026 |
+
"learning_rate": 2.011111111111111e-07,
|
1027 |
+
"loss": 1.0134,
|
1028 |
+
"step": 4100
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 0.35,
|
1032 |
+
"learning_rate": 1.9555555555555555e-07,
|
1033 |
+
"loss": 0.8379,
|
1034 |
+
"step": 4125
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 0.36,
|
1038 |
+
"learning_rate": 1.8999999999999998e-07,
|
1039 |
+
"loss": 0.954,
|
1040 |
+
"step": 4150
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 0.36,
|
1044 |
+
"learning_rate": 1.8444444444444444e-07,
|
1045 |
+
"loss": 0.7528,
|
1046 |
+
"step": 4175
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 0.36,
|
1050 |
+
"learning_rate": 1.7888888888888887e-07,
|
1051 |
+
"loss": 0.8616,
|
1052 |
+
"step": 4200
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 0.36,
|
1056 |
+
"learning_rate": 1.7333333333333332e-07,
|
1057 |
+
"loss": 0.8091,
|
1058 |
+
"step": 4225
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 0.37,
|
1062 |
+
"learning_rate": 1.6777777777777778e-07,
|
1063 |
+
"loss": 0.8108,
|
1064 |
+
"step": 4250
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 0.37,
|
1068 |
+
"learning_rate": 1.622222222222222e-07,
|
1069 |
+
"loss": 0.799,
|
1070 |
+
"step": 4275
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 0.37,
|
1074 |
+
"learning_rate": 1.5666666666666667e-07,
|
1075 |
+
"loss": 1.0137,
|
1076 |
+
"step": 4300
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 0.37,
|
1080 |
+
"learning_rate": 1.511111111111111e-07,
|
1081 |
+
"loss": 0.8332,
|
1082 |
+
"step": 4325
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 0.37,
|
1086 |
+
"learning_rate": 1.4555555555555555e-07,
|
1087 |
+
"loss": 0.9052,
|
1088 |
+
"step": 4350
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 0.38,
|
1092 |
+
"learning_rate": 1.4e-07,
|
1093 |
+
"loss": 0.8797,
|
1094 |
+
"step": 4375
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 0.38,
|
1098 |
+
"learning_rate": 1.3466666666666665e-07,
|
1099 |
+
"loss": 0.8401,
|
1100 |
+
"step": 4400
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 0.38,
|
1104 |
+
"learning_rate": 1.291111111111111e-07,
|
1105 |
+
"loss": 0.9753,
|
1106 |
+
"step": 4425
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 0.38,
|
1110 |
+
"learning_rate": 1.2355555555555554e-07,
|
1111 |
+
"loss": 0.7506,
|
1112 |
+
"step": 4450
|
1113 |
+
},
|
1114 |
+
{
|
1115 |
+
"epoch": 0.38,
|
1116 |
+
"learning_rate": 1.1799999999999998e-07,
|
1117 |
+
"loss": 0.7586,
|
1118 |
+
"step": 4475
|
1119 |
+
},
|
1120 |
+
{
|
1121 |
+
"epoch": 0.39,
|
1122 |
+
"learning_rate": 1.1244444444444444e-07,
|
1123 |
+
"loss": 0.7905,
|
1124 |
+
"step": 4500
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 0.39,
|
1128 |
+
"learning_rate": 1.0688888888888888e-07,
|
1129 |
+
"loss": 0.8575,
|
1130 |
+
"step": 4525
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 0.39,
|
1134 |
+
"learning_rate": 1.0133333333333333e-07,
|
1135 |
+
"loss": 0.6719,
|
1136 |
+
"step": 4550
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 0.39,
|
1140 |
+
"learning_rate": 9.577777777777777e-08,
|
1141 |
+
"loss": 0.9679,
|
1142 |
+
"step": 4575
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 0.4,
|
1146 |
+
"learning_rate": 9.022222222222221e-08,
|
1147 |
+
"loss": 0.884,
|
1148 |
+
"step": 4600
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 0.4,
|
1152 |
+
"learning_rate": 8.466666666666667e-08,
|
1153 |
+
"loss": 0.8952,
|
1154 |
+
"step": 4625
|
1155 |
+
},
|
1156 |
+
{
|
1157 |
+
"epoch": 0.4,
|
1158 |
+
"learning_rate": 7.91111111111111e-08,
|
1159 |
+
"loss": 0.9371,
|
1160 |
+
"step": 4650
|
1161 |
+
},
|
1162 |
+
{
|
1163 |
+
"epoch": 0.4,
|
1164 |
+
"learning_rate": 7.355555555555556e-08,
|
1165 |
+
"loss": 0.9077,
|
1166 |
+
"step": 4675
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 0.4,
|
1170 |
+
"learning_rate": 6.8e-08,
|
1171 |
+
"loss": 1.0169,
|
1172 |
+
"step": 4700
|
1173 |
+
},
|
1174 |
+
{
|
1175 |
+
"epoch": 0.41,
|
1176 |
+
"learning_rate": 6.244444444444444e-08,
|
1177 |
+
"loss": 0.9629,
|
1178 |
+
"step": 4725
|
1179 |
+
},
|
1180 |
+
{
|
1181 |
+
"epoch": 0.41,
|
1182 |
+
"learning_rate": 5.6888888888888887e-08,
|
1183 |
+
"loss": 0.8887,
|
1184 |
+
"step": 4750
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 0.41,
|
1188 |
+
"learning_rate": 5.133333333333333e-08,
|
1189 |
+
"loss": 0.8295,
|
1190 |
+
"step": 4775
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 0.41,
|
1194 |
+
"learning_rate": 4.577777777777777e-08,
|
1195 |
+
"loss": 0.848,
|
1196 |
+
"step": 4800
|
1197 |
+
},
|
1198 |
+
{
|
1199 |
+
"epoch": 0.41,
|
1200 |
+
"learning_rate": 4.022222222222222e-08,
|
1201 |
+
"loss": 0.7996,
|
1202 |
+
"step": 4825
|
1203 |
+
},
|
1204 |
+
{
|
1205 |
+
"epoch": 0.42,
|
1206 |
+
"learning_rate": 3.4666666666666666e-08,
|
1207 |
+
"loss": 0.9296,
|
1208 |
+
"step": 4850
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 0.42,
|
1212 |
+
"learning_rate": 2.911111111111111e-08,
|
1213 |
+
"loss": 0.928,
|
1214 |
+
"step": 4875
|
1215 |
+
},
|
1216 |
+
{
|
1217 |
+
"epoch": 0.42,
|
1218 |
+
"learning_rate": 2.3555555555555556e-08,
|
1219 |
+
"loss": 0.7704,
|
1220 |
+
"step": 4900
|
1221 |
+
},
|
1222 |
+
{
|
1223 |
+
"epoch": 0.42,
|
1224 |
+
"learning_rate": 1.8e-08,
|
1225 |
+
"loss": 0.917,
|
1226 |
+
"step": 4925
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"epoch": 0.43,
|
1230 |
+
"learning_rate": 1.2444444444444444e-08,
|
1231 |
+
"loss": 0.8176,
|
1232 |
+
"step": 4950
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 0.43,
|
1236 |
+
"learning_rate": 6.888888888888889e-09,
|
1237 |
+
"loss": 0.8245,
|
1238 |
+
"step": 4975
|
1239 |
+
},
|
1240 |
+
{
|
1241 |
+
"epoch": 0.43,
|
1242 |
+
"learning_rate": 1.3333333333333333e-09,
|
1243 |
+
"loss": 0.9089,
|
1244 |
+
"step": 5000
|
1245 |
+
},
|
1246 |
+
{
|
1247 |
+
"epoch": 0.43,
|
1248 |
+
"eval_loss": 0.8638723492622375,
|
1249 |
+
"eval_runtime": 557.1668,
|
1250 |
+
"eval_samples_per_second": 4.399,
|
1251 |
+
"eval_steps_per_second": 2.2,
|
1252 |
+
"eval_wer": 54.70462356526814,
|
1253 |
+
"step": 5000
|
1254 |
+
},
|
1255 |
+
{
|
1256 |
+
"epoch": 0.43,
|
1257 |
+
"step": 5000,
|
1258 |
+
"total_flos": 5.7717080064e+18,
|
1259 |
+
"train_loss": 0.996151273727417,
|
1260 |
+
"train_runtime": 5789.5442,
|
1261 |
+
"train_samples_per_second": 3.455,
|
1262 |
+
"train_steps_per_second": 0.864
|
1263 |
+
}
|
1264 |
+
],
|
1265 |
+
"max_steps": 5000,
|
1266 |
+
"num_train_epochs": 1,
|
1267 |
+
"total_flos": 5.7717080064e+18,
|
1268 |
+
"trial_name": null,
|
1269 |
+
"trial_params": null
|
1270 |
+
}
|