marinone94
commited on
Commit
•
9f100c5
1
Parent(s):
695a47b
Training in progress, step 1500
Browse files- checkpoint-1500/optimizer.pt +1 -1
- checkpoint-1500/pytorch_model.bin +1 -1
- checkpoint-1500/scaler.pt +1 -1
- checkpoint-1500/scheduler.pt +1 -1
- checkpoint-1500/trainer_state.json +46 -46
- checkpoint-1500/training_args.bin +1 -1
- pytorch_model.bin +1 -1
checkpoint-1500/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2490362385
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2cd9c62ff94cc2f259c8cbeff5c28cee1c8ef11d0981156e0eae5ca0fddb63a1
|
3 |
size 2490362385
|
checkpoint-1500/pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1262075377
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:acd89b40b183844d2d1cf17d26aeb8e928fe510a53c8c087c715468330a4fb98
|
3 |
size 1262075377
|
checkpoint-1500/scaler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 559
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0b1d791ba720c03cca9f2004a2681a7daf7371671890a4f96b90878506ee97e6
|
3 |
size 559
|
checkpoint-1500/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 623
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:033501fd55355c135dbf9f384ee2274d0c8246e49cc5023499a0046a5993e207
|
3 |
size 623
|
checkpoint-1500/trainer_state.json
CHANGED
@@ -9,124 +9,124 @@
|
|
9 |
"log_history": [
|
10 |
{
|
11 |
"epoch": 0.29,
|
12 |
-
"learning_rate":
|
13 |
-
"loss":
|
14 |
"step": 100
|
15 |
},
|
16 |
{
|
17 |
"epoch": 0.58,
|
18 |
-
"learning_rate":
|
19 |
-
"loss":
|
20 |
"step": 200
|
21 |
},
|
22 |
{
|
23 |
"epoch": 0.87,
|
24 |
-
"learning_rate":
|
25 |
-
"loss":
|
26 |
"step": 300
|
27 |
},
|
28 |
{
|
29 |
"epoch": 1.16,
|
30 |
-
"learning_rate":
|
31 |
-
"loss":
|
32 |
"step": 400
|
33 |
},
|
34 |
{
|
35 |
"epoch": 1.45,
|
36 |
-
"learning_rate":
|
37 |
-
"loss":
|
38 |
"step": 500
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.45,
|
42 |
-
"eval_loss":
|
43 |
-
"eval_runtime":
|
44 |
-
"eval_samples_per_second":
|
45 |
-
"eval_steps_per_second": 4.
|
46 |
"eval_wer": 1.0,
|
47 |
"step": 500
|
48 |
},
|
49 |
{
|
50 |
"epoch": 1.74,
|
51 |
-
"learning_rate":
|
52 |
-
"loss":
|
53 |
"step": 600
|
54 |
},
|
55 |
{
|
56 |
"epoch": 2.03,
|
57 |
-
"learning_rate":
|
58 |
-
"loss":
|
59 |
"step": 700
|
60 |
},
|
61 |
{
|
62 |
"epoch": 2.32,
|
63 |
-
"learning_rate":
|
64 |
-
"loss":
|
65 |
"step": 800
|
66 |
},
|
67 |
{
|
68 |
"epoch": 2.61,
|
69 |
-
"learning_rate": 3.
|
70 |
-
"loss": 2.
|
71 |
"step": 900
|
72 |
},
|
73 |
{
|
74 |
"epoch": 2.91,
|
75 |
-
"learning_rate": 3.
|
76 |
-
"loss": 2.
|
77 |
"step": 1000
|
78 |
},
|
79 |
{
|
80 |
"epoch": 2.91,
|
81 |
-
"eval_loss": 2.
|
82 |
-
"eval_runtime":
|
83 |
-
"eval_samples_per_second":
|
84 |
-
"eval_steps_per_second": 3.
|
85 |
-
"eval_wer":
|
86 |
"step": 1000
|
87 |
},
|
88 |
{
|
89 |
"epoch": 3.2,
|
90 |
-
"learning_rate":
|
91 |
-
"loss": 2.
|
92 |
"step": 1100
|
93 |
},
|
94 |
{
|
95 |
"epoch": 3.49,
|
96 |
-
"learning_rate":
|
97 |
-
"loss": 2.
|
98 |
"step": 1200
|
99 |
},
|
100 |
{
|
101 |
"epoch": 3.78,
|
102 |
-
"learning_rate":
|
103 |
-
"loss": 2.
|
104 |
"step": 1300
|
105 |
},
|
106 |
{
|
107 |
"epoch": 4.07,
|
108 |
-
"learning_rate":
|
109 |
-
"loss": 2.
|
110 |
"step": 1400
|
111 |
},
|
112 |
{
|
113 |
"epoch": 4.36,
|
114 |
-
"learning_rate":
|
115 |
-
"loss": 2.
|
116 |
"step": 1500
|
117 |
},
|
118 |
{
|
119 |
"epoch": 4.36,
|
120 |
-
"eval_loss": 1.
|
121 |
-
"eval_runtime":
|
122 |
-
"eval_samples_per_second":
|
123 |
-
"eval_steps_per_second": 4.
|
124 |
-
"eval_wer": 0.
|
125 |
"step": 1500
|
126 |
}
|
127 |
],
|
128 |
-
"max_steps":
|
129 |
-
"num_train_epochs":
|
130 |
"total_flos": 4.445118706030802e+18,
|
131 |
"trial_name": null,
|
132 |
"trial_params": null
|
|
|
9 |
"log_history": [
|
10 |
{
|
11 |
"epoch": 0.29,
|
12 |
+
"learning_rate": 3.7125e-06,
|
13 |
+
"loss": 12.2713,
|
14 |
"step": 100
|
15 |
},
|
16 |
{
|
17 |
"epoch": 0.58,
|
18 |
+
"learning_rate": 7.4625e-06,
|
19 |
+
"loss": 6.2026,
|
20 |
"step": 200
|
21 |
},
|
22 |
{
|
23 |
"epoch": 0.87,
|
24 |
+
"learning_rate": 1.1212499999999998e-05,
|
25 |
+
"loss": 4.1374,
|
26 |
"step": 300
|
27 |
},
|
28 |
{
|
29 |
"epoch": 1.16,
|
30 |
+
"learning_rate": 1.49625e-05,
|
31 |
+
"loss": 3.6755,
|
32 |
"step": 400
|
33 |
},
|
34 |
{
|
35 |
"epoch": 1.45,
|
36 |
+
"learning_rate": 1.8712499999999997e-05,
|
37 |
+
"loss": 3.3332,
|
38 |
"step": 500
|
39 |
},
|
40 |
{
|
41 |
"epoch": 1.45,
|
42 |
+
"eval_loss": 3.292029857635498,
|
43 |
+
"eval_runtime": 137.2974,
|
44 |
+
"eval_samples_per_second": 33.65,
|
45 |
+
"eval_steps_per_second": 4.21,
|
46 |
"eval_wer": 1.0,
|
47 |
"step": 500
|
48 |
},
|
49 |
{
|
50 |
"epoch": 1.74,
|
51 |
+
"learning_rate": 2.2462499999999997e-05,
|
52 |
+
"loss": 3.1418,
|
53 |
"step": 600
|
54 |
},
|
55 |
{
|
56 |
"epoch": 2.03,
|
57 |
+
"learning_rate": 2.6212499999999997e-05,
|
58 |
+
"loss": 3.0879,
|
59 |
"step": 700
|
60 |
},
|
61 |
{
|
62 |
"epoch": 2.32,
|
63 |
+
"learning_rate": 2.99625e-05,
|
64 |
+
"loss": 3.0216,
|
65 |
"step": 800
|
66 |
},
|
67 |
{
|
68 |
"epoch": 2.61,
|
69 |
+
"learning_rate": 3.37125e-05,
|
70 |
+
"loss": 2.9595,
|
71 |
"step": 900
|
72 |
},
|
73 |
{
|
74 |
"epoch": 2.91,
|
75 |
+
"learning_rate": 3.7462499999999996e-05,
|
76 |
+
"loss": 2.9269,
|
77 |
"step": 1000
|
78 |
},
|
79 |
{
|
80 |
"epoch": 2.91,
|
81 |
+
"eval_loss": 2.941540241241455,
|
82 |
+
"eval_runtime": 178.4388,
|
83 |
+
"eval_samples_per_second": 25.891,
|
84 |
+
"eval_steps_per_second": 3.239,
|
85 |
+
"eval_wer": 0.9966346448020559,
|
86 |
"step": 1000
|
87 |
},
|
88 |
{
|
89 |
"epoch": 3.2,
|
90 |
+
"learning_rate": 4.12125e-05,
|
91 |
+
"loss": 2.914,
|
92 |
"step": 1100
|
93 |
},
|
94 |
{
|
95 |
"epoch": 3.49,
|
96 |
+
"learning_rate": 4.4962499999999995e-05,
|
97 |
+
"loss": 2.8432,
|
98 |
"step": 1200
|
99 |
},
|
100 |
{
|
101 |
"epoch": 3.78,
|
102 |
+
"learning_rate": 4.871249999999999e-05,
|
103 |
+
"loss": 2.6828,
|
104 |
"step": 1300
|
105 |
},
|
106 |
{
|
107 |
"epoch": 4.07,
|
108 |
+
"learning_rate": 5.2462499999999994e-05,
|
109 |
+
"loss": 2.355,
|
110 |
"step": 1400
|
111 |
},
|
112 |
{
|
113 |
"epoch": 4.36,
|
114 |
+
"learning_rate": 5.62125e-05,
|
115 |
+
"loss": 2.0719,
|
116 |
"step": 1500
|
117 |
},
|
118 |
{
|
119 |
"epoch": 4.36,
|
120 |
+
"eval_loss": 1.164096713066101,
|
121 |
+
"eval_runtime": 134.2981,
|
122 |
+
"eval_samples_per_second": 34.401,
|
123 |
+
"eval_steps_per_second": 4.304,
|
124 |
+
"eval_wer": 0.8507923881784251,
|
125 |
"step": 1500
|
126 |
}
|
127 |
],
|
128 |
+
"max_steps": 17200,
|
129 |
+
"num_train_epochs": 50,
|
130 |
"total_flos": 4.445118706030802e+18,
|
131 |
"trial_name": null,
|
132 |
"trial_params": null
|
checkpoint-1500/training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2991
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5a168f126d02648193e16ba893c3e1ef9f2c2de91803928caf5e25532aff1325
|
3 |
size 2991
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1262075377
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:acd89b40b183844d2d1cf17d26aeb8e928fe510a53c8c087c715468330a4fb98
|
3 |
size 1262075377
|