Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +1710 -0
trainer_state.json
ADDED
@@ -0,0 +1,1710 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 8.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 280,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.03,
|
13 |
+
"learning_rate": 2.2222222222222223e-05,
|
14 |
+
"loss": 1.383,
|
15 |
+
"step": 1
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"epoch": 0.06,
|
19 |
+
"learning_rate": 4.4444444444444447e-05,
|
20 |
+
"loss": 1.3408,
|
21 |
+
"step": 2
|
22 |
+
},
|
23 |
+
{
|
24 |
+
"epoch": 0.09,
|
25 |
+
"learning_rate": 6.666666666666667e-05,
|
26 |
+
"loss": 1.2449,
|
27 |
+
"step": 3
|
28 |
+
},
|
29 |
+
{
|
30 |
+
"epoch": 0.11,
|
31 |
+
"learning_rate": 8.888888888888889e-05,
|
32 |
+
"loss": 0.9588,
|
33 |
+
"step": 4
|
34 |
+
},
|
35 |
+
{
|
36 |
+
"epoch": 0.14,
|
37 |
+
"learning_rate": 0.00011111111111111112,
|
38 |
+
"loss": 0.6465,
|
39 |
+
"step": 5
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 0.17,
|
43 |
+
"learning_rate": 0.00013333333333333334,
|
44 |
+
"loss": 0.4188,
|
45 |
+
"step": 6
|
46 |
+
},
|
47 |
+
{
|
48 |
+
"epoch": 0.2,
|
49 |
+
"learning_rate": 0.00015555555555555556,
|
50 |
+
"loss": 0.272,
|
51 |
+
"step": 7
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.23,
|
55 |
+
"learning_rate": 0.00017777777777777779,
|
56 |
+
"loss": 0.2232,
|
57 |
+
"step": 8
|
58 |
+
},
|
59 |
+
{
|
60 |
+
"epoch": 0.26,
|
61 |
+
"learning_rate": 0.0002,
|
62 |
+
"loss": 0.2167,
|
63 |
+
"step": 9
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"epoch": 0.29,
|
67 |
+
"learning_rate": 0.00019999328066483865,
|
68 |
+
"loss": 0.1756,
|
69 |
+
"step": 10
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 0.31,
|
73 |
+
"learning_rate": 0.00019997312356234386,
|
74 |
+
"loss": 0.1567,
|
75 |
+
"step": 11
|
76 |
+
},
|
77 |
+
{
|
78 |
+
"epoch": 0.34,
|
79 |
+
"learning_rate": 0.0001999395314013622,
|
80 |
+
"loss": 0.1595,
|
81 |
+
"step": 12
|
82 |
+
},
|
83 |
+
{
|
84 |
+
"epoch": 0.37,
|
85 |
+
"learning_rate": 0.00019989250869623343,
|
86 |
+
"loss": 1.0245,
|
87 |
+
"step": 13
|
88 |
+
},
|
89 |
+
{
|
90 |
+
"epoch": 0.4,
|
91 |
+
"learning_rate": 0.00019983206176618388,
|
92 |
+
"loss": 0.1493,
|
93 |
+
"step": 14
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.43,
|
97 |
+
"learning_rate": 0.00019975819873447717,
|
98 |
+
"loss": 0.1426,
|
99 |
+
"step": 15
|
100 |
+
},
|
101 |
+
{
|
102 |
+
"epoch": 0.46,
|
103 |
+
"learning_rate": 0.00019967092952732264,
|
104 |
+
"loss": 0.1397,
|
105 |
+
"step": 16
|
106 |
+
},
|
107 |
+
{
|
108 |
+
"epoch": 0.49,
|
109 |
+
"learning_rate": 0.00019957026587254134,
|
110 |
+
"loss": 0.1395,
|
111 |
+
"step": 17
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"epoch": 0.51,
|
115 |
+
"learning_rate": 0.00019945622129799,
|
116 |
+
"loss": 0.1302,
|
117 |
+
"step": 18
|
118 |
+
},
|
119 |
+
{
|
120 |
+
"epoch": 0.54,
|
121 |
+
"learning_rate": 0.000199328811129743,
|
122 |
+
"loss": 0.1268,
|
123 |
+
"step": 19
|
124 |
+
},
|
125 |
+
{
|
126 |
+
"epoch": 0.57,
|
127 |
+
"learning_rate": 0.0001991880524900327,
|
128 |
+
"loss": 0.1207,
|
129 |
+
"step": 20
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.6,
|
133 |
+
"learning_rate": 0.0001990339642949488,
|
134 |
+
"loss": 0.1221,
|
135 |
+
"step": 21
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.63,
|
139 |
+
"learning_rate": 0.00019886656725189575,
|
140 |
+
"loss": 0.1122,
|
141 |
+
"step": 22
|
142 |
+
},
|
143 |
+
{
|
144 |
+
"epoch": 0.66,
|
145 |
+
"learning_rate": 0.00019868588385681032,
|
146 |
+
"loss": 0.1166,
|
147 |
+
"step": 23
|
148 |
+
},
|
149 |
+
{
|
150 |
+
"epoch": 0.69,
|
151 |
+
"learning_rate": 0.00019849193839113833,
|
152 |
+
"loss": 0.1241,
|
153 |
+
"step": 24
|
154 |
+
},
|
155 |
+
{
|
156 |
+
"epoch": 0.71,
|
157 |
+
"learning_rate": 0.00019828475691857145,
|
158 |
+
"loss": 0.1215,
|
159 |
+
"step": 25
|
160 |
+
},
|
161 |
+
{
|
162 |
+
"epoch": 0.74,
|
163 |
+
"learning_rate": 0.00019806436728154485,
|
164 |
+
"loss": 0.1134,
|
165 |
+
"step": 26
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"epoch": 0.77,
|
169 |
+
"learning_rate": 0.00019783079909749515,
|
170 |
+
"loss": 0.1233,
|
171 |
+
"step": 27
|
172 |
+
},
|
173 |
+
{
|
174 |
+
"epoch": 0.8,
|
175 |
+
"learning_rate": 0.00019758408375488071,
|
176 |
+
"loss": 0.1195,
|
177 |
+
"step": 28
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.83,
|
181 |
+
"learning_rate": 0.00019732425440896297,
|
182 |
+
"loss": 0.1086,
|
183 |
+
"step": 29
|
184 |
+
},
|
185 |
+
{
|
186 |
+
"epoch": 0.86,
|
187 |
+
"learning_rate": 0.00019705134597735113,
|
188 |
+
"loss": 0.118,
|
189 |
+
"step": 30
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"epoch": 0.89,
|
193 |
+
"learning_rate": 0.00019676539513530968,
|
194 |
+
"loss": 0.1141,
|
195 |
+
"step": 31
|
196 |
+
},
|
197 |
+
{
|
198 |
+
"epoch": 0.91,
|
199 |
+
"learning_rate": 0.00019646644031082948,
|
200 |
+
"loss": 0.1065,
|
201 |
+
"step": 32
|
202 |
+
},
|
203 |
+
{
|
204 |
+
"epoch": 0.94,
|
205 |
+
"learning_rate": 0.00019615452167946385,
|
206 |
+
"loss": 0.1077,
|
207 |
+
"step": 33
|
208 |
+
},
|
209 |
+
{
|
210 |
+
"epoch": 0.97,
|
211 |
+
"learning_rate": 0.0001958296811589293,
|
212 |
+
"loss": 0.1033,
|
213 |
+
"step": 34
|
214 |
+
},
|
215 |
+
{
|
216 |
+
"epoch": 1.0,
|
217 |
+
"learning_rate": 0.00019549196240347248,
|
218 |
+
"loss": 0.0947,
|
219 |
+
"step": 35
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 1.03,
|
223 |
+
"learning_rate": 0.0001951414107980036,
|
224 |
+
"loss": 0.1081,
|
225 |
+
"step": 36
|
226 |
+
},
|
227 |
+
{
|
228 |
+
"epoch": 1.06,
|
229 |
+
"learning_rate": 0.00019477807345199714,
|
230 |
+
"loss": 0.1021,
|
231 |
+
"step": 37
|
232 |
+
},
|
233 |
+
{
|
234 |
+
"epoch": 1.09,
|
235 |
+
"learning_rate": 0.00019440199919316123,
|
236 |
+
"loss": 0.0916,
|
237 |
+
"step": 38
|
238 |
+
},
|
239 |
+
{
|
240 |
+
"epoch": 1.11,
|
241 |
+
"learning_rate": 0.0001940132385608757,
|
242 |
+
"loss": 0.1015,
|
243 |
+
"step": 39
|
244 |
+
},
|
245 |
+
{
|
246 |
+
"epoch": 1.14,
|
247 |
+
"learning_rate": 0.0001936118437994003,
|
248 |
+
"loss": 0.1068,
|
249 |
+
"step": 40
|
250 |
+
},
|
251 |
+
{
|
252 |
+
"epoch": 1.17,
|
253 |
+
"learning_rate": 0.00019319786885085364,
|
254 |
+
"loss": 0.107,
|
255 |
+
"step": 41
|
256 |
+
},
|
257 |
+
{
|
258 |
+
"epoch": 1.2,
|
259 |
+
"learning_rate": 0.0001927713693479643,
|
260 |
+
"loss": 0.1029,
|
261 |
+
"step": 42
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 1.23,
|
265 |
+
"learning_rate": 0.0001923324026065944,
|
266 |
+
"loss": 0.2492,
|
267 |
+
"step": 43
|
268 |
+
},
|
269 |
+
{
|
270 |
+
"epoch": 1.26,
|
271 |
+
"learning_rate": 0.00019188102761803717,
|
272 |
+
"loss": 0.1002,
|
273 |
+
"step": 44
|
274 |
+
},
|
275 |
+
{
|
276 |
+
"epoch": 1.29,
|
277 |
+
"learning_rate": 0.00019141730504108922,
|
278 |
+
"loss": 0.2045,
|
279 |
+
"step": 45
|
280 |
+
},
|
281 |
+
{
|
282 |
+
"epoch": 1.31,
|
283 |
+
"learning_rate": 0.00019094129719389886,
|
284 |
+
"loss": 0.1099,
|
285 |
+
"step": 46
|
286 |
+
},
|
287 |
+
{
|
288 |
+
"epoch": 1.34,
|
289 |
+
"learning_rate": 0.0001904530680455914,
|
290 |
+
"loss": 0.1143,
|
291 |
+
"step": 47
|
292 |
+
},
|
293 |
+
{
|
294 |
+
"epoch": 1.37,
|
295 |
+
"learning_rate": 0.00018995268320767252,
|
296 |
+
"loss": 0.1081,
|
297 |
+
"step": 48
|
298 |
+
},
|
299 |
+
{
|
300 |
+
"epoch": 1.4,
|
301 |
+
"learning_rate": 0.0001894402099252109,
|
302 |
+
"loss": 0.1011,
|
303 |
+
"step": 49
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 1.43,
|
307 |
+
"learning_rate": 0.00018891571706780146,
|
308 |
+
"loss": 0.1031,
|
309 |
+
"step": 50
|
310 |
+
},
|
311 |
+
{
|
312 |
+
"epoch": 1.46,
|
313 |
+
"learning_rate": 0.0001883792751203102,
|
314 |
+
"loss": 0.106,
|
315 |
+
"step": 51
|
316 |
+
},
|
317 |
+
{
|
318 |
+
"epoch": 1.49,
|
319 |
+
"learning_rate": 0.00018783095617340193,
|
320 |
+
"loss": 0.109,
|
321 |
+
"step": 52
|
322 |
+
},
|
323 |
+
{
|
324 |
+
"epoch": 1.51,
|
325 |
+
"learning_rate": 0.0001872708339138522,
|
326 |
+
"loss": 0.1095,
|
327 |
+
"step": 53
|
328 |
+
},
|
329 |
+
{
|
330 |
+
"epoch": 1.54,
|
331 |
+
"learning_rate": 0.0001866989836146449,
|
332 |
+
"loss": 0.1071,
|
333 |
+
"step": 54
|
334 |
+
},
|
335 |
+
{
|
336 |
+
"epoch": 1.57,
|
337 |
+
"learning_rate": 0.00018611548212485647,
|
338 |
+
"loss": 0.1038,
|
339 |
+
"step": 55
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 1.6,
|
343 |
+
"learning_rate": 0.00018552040785932845,
|
344 |
+
"loss": 0.1111,
|
345 |
+
"step": 56
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 1.63,
|
349 |
+
"learning_rate": 0.00018491384078812959,
|
350 |
+
"loss": 0.0954,
|
351 |
+
"step": 57
|
352 |
+
},
|
353 |
+
{
|
354 |
+
"epoch": 1.66,
|
355 |
+
"learning_rate": 0.00018429586242580884,
|
356 |
+
"loss": 0.1011,
|
357 |
+
"step": 58
|
358 |
+
},
|
359 |
+
{
|
360 |
+
"epoch": 1.69,
|
361 |
+
"learning_rate": 0.00018366655582044094,
|
362 |
+
"loss": 0.1053,
|
363 |
+
"step": 59
|
364 |
+
},
|
365 |
+
{
|
366 |
+
"epoch": 1.71,
|
367 |
+
"learning_rate": 0.00018302600554246601,
|
368 |
+
"loss": 0.1036,
|
369 |
+
"step": 60
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 1.74,
|
373 |
+
"learning_rate": 0.00018237429767332405,
|
374 |
+
"loss": 0.0974,
|
375 |
+
"step": 61
|
376 |
+
},
|
377 |
+
{
|
378 |
+
"epoch": 1.77,
|
379 |
+
"learning_rate": 0.00018171151979388714,
|
380 |
+
"loss": 0.1017,
|
381 |
+
"step": 62
|
382 |
+
},
|
383 |
+
{
|
384 |
+
"epoch": 1.8,
|
385 |
+
"learning_rate": 0.00018103776097268942,
|
386 |
+
"loss": 0.0991,
|
387 |
+
"step": 63
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 1.83,
|
391 |
+
"learning_rate": 0.0001803531117539577,
|
392 |
+
"loss": 0.1039,
|
393 |
+
"step": 64
|
394 |
+
},
|
395 |
+
{
|
396 |
+
"epoch": 1.86,
|
397 |
+
"learning_rate": 0.00017965766414544326,
|
398 |
+
"loss": 0.0924,
|
399 |
+
"step": 65
|
400 |
+
},
|
401 |
+
{
|
402 |
+
"epoch": 1.89,
|
403 |
+
"learning_rate": 0.00017895151160605757,
|
404 |
+
"loss": 0.1002,
|
405 |
+
"step": 66
|
406 |
+
},
|
407 |
+
{
|
408 |
+
"epoch": 1.91,
|
409 |
+
"learning_rate": 0.00017823474903331233,
|
410 |
+
"loss": 0.0999,
|
411 |
+
"step": 67
|
412 |
+
},
|
413 |
+
{
|
414 |
+
"epoch": 1.94,
|
415 |
+
"learning_rate": 0.0001775074727505667,
|
416 |
+
"loss": 0.0942,
|
417 |
+
"step": 68
|
418 |
+
},
|
419 |
+
{
|
420 |
+
"epoch": 1.97,
|
421 |
+
"learning_rate": 0.00017676978049408263,
|
422 |
+
"loss": 0.1004,
|
423 |
+
"step": 69
|
424 |
+
},
|
425 |
+
{
|
426 |
+
"epoch": 2.0,
|
427 |
+
"learning_rate": 0.00017602177139989044,
|
428 |
+
"loss": 0.0808,
|
429 |
+
"step": 70
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 2.03,
|
433 |
+
"learning_rate": 0.00017526354599046635,
|
434 |
+
"loss": 0.0955,
|
435 |
+
"step": 71
|
436 |
+
},
|
437 |
+
{
|
438 |
+
"epoch": 2.06,
|
439 |
+
"learning_rate": 0.00017449520616122344,
|
440 |
+
"loss": 0.0876,
|
441 |
+
"step": 72
|
442 |
+
},
|
443 |
+
{
|
444 |
+
"epoch": 2.09,
|
445 |
+
"learning_rate": 0.00017371685516681825,
|
446 |
+
"loss": 0.0925,
|
447 |
+
"step": 73
|
448 |
+
},
|
449 |
+
{
|
450 |
+
"epoch": 2.11,
|
451 |
+
"learning_rate": 0.00017292859760727493,
|
452 |
+
"loss": 0.0862,
|
453 |
+
"step": 74
|
454 |
+
},
|
455 |
+
{
|
456 |
+
"epoch": 2.14,
|
457 |
+
"learning_rate": 0.00017213053941392818,
|
458 |
+
"loss": 0.0979,
|
459 |
+
"step": 75
|
460 |
+
},
|
461 |
+
{
|
462 |
+
"epoch": 2.17,
|
463 |
+
"learning_rate": 0.00017132278783518756,
|
464 |
+
"loss": 0.0948,
|
465 |
+
"step": 76
|
466 |
+
},
|
467 |
+
{
|
468 |
+
"epoch": 2.2,
|
469 |
+
"learning_rate": 0.00017050545142212483,
|
470 |
+
"loss": 0.093,
|
471 |
+
"step": 77
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 2.23,
|
475 |
+
"learning_rate": 0.0001696786400138859,
|
476 |
+
"loss": 0.0813,
|
477 |
+
"step": 78
|
478 |
+
},
|
479 |
+
{
|
480 |
+
"epoch": 2.26,
|
481 |
+
"learning_rate": 0.00016884246472293016,
|
482 |
+
"loss": 0.0997,
|
483 |
+
"step": 79
|
484 |
+
},
|
485 |
+
{
|
486 |
+
"epoch": 2.29,
|
487 |
+
"learning_rate": 0.00016799703792009827,
|
488 |
+
"loss": 0.0934,
|
489 |
+
"step": 80
|
490 |
+
},
|
491 |
+
{
|
492 |
+
"epoch": 2.31,
|
493 |
+
"learning_rate": 0.00016714247321951106,
|
494 |
+
"loss": 0.097,
|
495 |
+
"step": 81
|
496 |
+
},
|
497 |
+
{
|
498 |
+
"epoch": 2.34,
|
499 |
+
"learning_rate": 0.00016627888546330138,
|
500 |
+
"loss": 0.0847,
|
501 |
+
"step": 82
|
502 |
+
},
|
503 |
+
{
|
504 |
+
"epoch": 2.37,
|
505 |
+
"learning_rate": 0.0001654063907061807,
|
506 |
+
"loss": 0.0854,
|
507 |
+
"step": 83
|
508 |
+
},
|
509 |
+
{
|
510 |
+
"epoch": 2.4,
|
511 |
+
"learning_rate": 0.000164525106199843,
|
512 |
+
"loss": 0.0854,
|
513 |
+
"step": 84
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 2.43,
|
517 |
+
"learning_rate": 0.00016363515037720773,
|
518 |
+
"loss": 0.0882,
|
519 |
+
"step": 85
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 2.46,
|
523 |
+
"learning_rate": 0.0001627366428365039,
|
524 |
+
"loss": 0.0849,
|
525 |
+
"step": 86
|
526 |
+
},
|
527 |
+
{
|
528 |
+
"epoch": 2.49,
|
529 |
+
"learning_rate": 0.00016182970432519772,
|
530 |
+
"loss": 0.0897,
|
531 |
+
"step": 87
|
532 |
+
},
|
533 |
+
{
|
534 |
+
"epoch": 2.51,
|
535 |
+
"learning_rate": 0.0001609144567237658,
|
536 |
+
"loss": 0.0903,
|
537 |
+
"step": 88
|
538 |
+
},
|
539 |
+
{
|
540 |
+
"epoch": 2.54,
|
541 |
+
"learning_rate": 0.00015999102302931585,
|
542 |
+
"loss": 0.0968,
|
543 |
+
"step": 89
|
544 |
+
},
|
545 |
+
{
|
546 |
+
"epoch": 2.57,
|
547 |
+
"learning_rate": 0.00015905952733905775,
|
548 |
+
"loss": 0.0964,
|
549 |
+
"step": 90
|
550 |
+
},
|
551 |
+
{
|
552 |
+
"epoch": 2.6,
|
553 |
+
"learning_rate": 0.00015812009483362642,
|
554 |
+
"loss": 0.0873,
|
555 |
+
"step": 91
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 2.63,
|
559 |
+
"learning_rate": 0.00015717285176025913,
|
560 |
+
"loss": 0.0919,
|
561 |
+
"step": 92
|
562 |
+
},
|
563 |
+
{
|
564 |
+
"epoch": 2.66,
|
565 |
+
"learning_rate": 0.00015621792541582966,
|
566 |
+
"loss": 0.0851,
|
567 |
+
"step": 93
|
568 |
+
},
|
569 |
+
{
|
570 |
+
"epoch": 2.69,
|
571 |
+
"learning_rate": 0.00015525544412974132,
|
572 |
+
"loss": 0.0887,
|
573 |
+
"step": 94
|
574 |
+
},
|
575 |
+
{
|
576 |
+
"epoch": 2.71,
|
577 |
+
"learning_rate": 0.00015428553724668103,
|
578 |
+
"loss": 0.0812,
|
579 |
+
"step": 95
|
580 |
+
},
|
581 |
+
{
|
582 |
+
"epoch": 2.74,
|
583 |
+
"learning_rate": 0.00015330833510923718,
|
584 |
+
"loss": 0.0919,
|
585 |
+
"step": 96
|
586 |
+
},
|
587 |
+
{
|
588 |
+
"epoch": 2.77,
|
589 |
+
"learning_rate": 0.0001523239690403835,
|
590 |
+
"loss": 0.0857,
|
591 |
+
"step": 97
|
592 |
+
},
|
593 |
+
{
|
594 |
+
"epoch": 2.8,
|
595 |
+
"learning_rate": 0.00015133257132583073,
|
596 |
+
"loss": 0.1003,
|
597 |
+
"step": 98
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 2.83,
|
601 |
+
"learning_rate": 0.0001503342751962493,
|
602 |
+
"loss": 0.0842,
|
603 |
+
"step": 99
|
604 |
+
},
|
605 |
+
{
|
606 |
+
"epoch": 2.86,
|
607 |
+
"learning_rate": 0.0001493292148093649,
|
608 |
+
"loss": 0.0857,
|
609 |
+
"step": 100
|
610 |
+
},
|
611 |
+
{
|
612 |
+
"epoch": 2.89,
|
613 |
+
"learning_rate": 0.00014831752523192948,
|
614 |
+
"loss": 0.0784,
|
615 |
+
"step": 101
|
616 |
+
},
|
617 |
+
{
|
618 |
+
"epoch": 2.91,
|
619 |
+
"learning_rate": 0.00014729934242157004,
|
620 |
+
"loss": 0.0926,
|
621 |
+
"step": 102
|
622 |
+
},
|
623 |
+
{
|
624 |
+
"epoch": 2.94,
|
625 |
+
"learning_rate": 0.00014627480320851774,
|
626 |
+
"loss": 0.077,
|
627 |
+
"step": 103
|
628 |
+
},
|
629 |
+
{
|
630 |
+
"epoch": 2.97,
|
631 |
+
"learning_rate": 0.00014524404527721977,
|
632 |
+
"loss": 0.0893,
|
633 |
+
"step": 104
|
634 |
+
},
|
635 |
+
{
|
636 |
+
"epoch": 3.0,
|
637 |
+
"learning_rate": 0.00014420720714783636,
|
638 |
+
"loss": 0.0834,
|
639 |
+
"step": 105
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 3.03,
|
643 |
+
"learning_rate": 0.00014316442815762544,
|
644 |
+
"loss": 0.0883,
|
645 |
+
"step": 106
|
646 |
+
},
|
647 |
+
{
|
648 |
+
"epoch": 3.06,
|
649 |
+
"learning_rate": 0.0001421158484422177,
|
650 |
+
"loss": 0.08,
|
651 |
+
"step": 107
|
652 |
+
},
|
653 |
+
{
|
654 |
+
"epoch": 3.09,
|
655 |
+
"learning_rate": 0.0001410616089167842,
|
656 |
+
"loss": 0.0826,
|
657 |
+
"step": 108
|
658 |
+
},
|
659 |
+
{
|
660 |
+
"epoch": 3.11,
|
661 |
+
"learning_rate": 0.00014000185125709918,
|
662 |
+
"loss": 0.0795,
|
663 |
+
"step": 109
|
664 |
+
},
|
665 |
+
{
|
666 |
+
"epoch": 3.14,
|
667 |
+
"learning_rate": 0.00013893671788050074,
|
668 |
+
"loss": 0.0798,
|
669 |
+
"step": 110
|
670 |
+
},
|
671 |
+
{
|
672 |
+
"epoch": 3.17,
|
673 |
+
"learning_rate": 0.00013786635192675184,
|
674 |
+
"loss": 0.0902,
|
675 |
+
"step": 111
|
676 |
+
},
|
677 |
+
{
|
678 |
+
"epoch": 3.2,
|
679 |
+
"learning_rate": 0.00013679089723880427,
|
680 |
+
"loss": 0.0797,
|
681 |
+
"step": 112
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 3.23,
|
685 |
+
"learning_rate": 0.00013571049834346799,
|
686 |
+
"loss": 0.0765,
|
687 |
+
"step": 113
|
688 |
+
},
|
689 |
+
{
|
690 |
+
"epoch": 3.26,
|
691 |
+
"learning_rate": 0.00013462530043198873,
|
692 |
+
"loss": 0.0748,
|
693 |
+
"step": 114
|
694 |
+
},
|
695 |
+
{
|
696 |
+
"epoch": 3.29,
|
697 |
+
"learning_rate": 0.00013353544934053616,
|
698 |
+
"loss": 0.0773,
|
699 |
+
"step": 115
|
700 |
+
},
|
701 |
+
{
|
702 |
+
"epoch": 3.31,
|
703 |
+
"learning_rate": 0.00013244109153060548,
|
704 |
+
"loss": 0.0767,
|
705 |
+
"step": 116
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 3.34,
|
709 |
+
"learning_rate": 0.00013134237406933492,
|
710 |
+
"loss": 0.0719,
|
711 |
+
"step": 117
|
712 |
+
},
|
713 |
+
{
|
714 |
+
"epoch": 3.37,
|
715 |
+
"learning_rate": 0.00013023944460974183,
|
716 |
+
"loss": 0.0792,
|
717 |
+
"step": 118
|
718 |
+
},
|
719 |
+
{
|
720 |
+
"epoch": 3.4,
|
721 |
+
"learning_rate": 0.00012913245137088024,
|
722 |
+
"loss": 0.0778,
|
723 |
+
"step": 119
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 3.43,
|
727 |
+
"learning_rate": 0.00012802154311792197,
|
728 |
+
"loss": 0.0715,
|
729 |
+
"step": 120
|
730 |
+
},
|
731 |
+
{
|
732 |
+
"epoch": 3.46,
|
733 |
+
"learning_rate": 0.00012690686914216474,
|
734 |
+
"loss": 0.0891,
|
735 |
+
"step": 121
|
736 |
+
},
|
737 |
+
{
|
738 |
+
"epoch": 3.49,
|
739 |
+
"learning_rate": 0.00012578857924096934,
|
740 |
+
"loss": 0.0772,
|
741 |
+
"step": 122
|
742 |
+
},
|
743 |
+
{
|
744 |
+
"epoch": 3.51,
|
745 |
+
"learning_rate": 0.00012466682369762882,
|
746 |
+
"loss": 0.0781,
|
747 |
+
"step": 123
|
748 |
+
},
|
749 |
+
{
|
750 |
+
"epoch": 3.54,
|
751 |
+
"learning_rate": 0.00012354175326117253,
|
752 |
+
"loss": 0.0898,
|
753 |
+
"step": 124
|
754 |
+
},
|
755 |
+
{
|
756 |
+
"epoch": 3.57,
|
757 |
+
"learning_rate": 0.00012241351912610726,
|
758 |
+
"loss": 0.0829,
|
759 |
+
"step": 125
|
760 |
+
},
|
761 |
+
{
|
762 |
+
"epoch": 3.6,
|
763 |
+
"learning_rate": 0.00012128227291209891,
|
764 |
+
"loss": 0.0788,
|
765 |
+
"step": 126
|
766 |
+
},
|
767 |
+
{
|
768 |
+
"epoch": 3.63,
|
769 |
+
"learning_rate": 0.00012014816664359671,
|
770 |
+
"loss": 0.0795,
|
771 |
+
"step": 127
|
772 |
+
},
|
773 |
+
{
|
774 |
+
"epoch": 3.66,
|
775 |
+
"learning_rate": 0.0001190113527294032,
|
776 |
+
"loss": 0.0797,
|
777 |
+
"step": 128
|
778 |
+
},
|
779 |
+
{
|
780 |
+
"epoch": 3.69,
|
781 |
+
"learning_rate": 0.0001178719839421925,
|
782 |
+
"loss": 0.0802,
|
783 |
+
"step": 129
|
784 |
+
},
|
785 |
+
{
|
786 |
+
"epoch": 3.71,
|
787 |
+
"learning_rate": 0.00011673021339797967,
|
788 |
+
"loss": 0.0801,
|
789 |
+
"step": 130
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 3.74,
|
793 |
+
"learning_rate": 0.000115586194535544,
|
794 |
+
"loss": 0.0814,
|
795 |
+
"step": 131
|
796 |
+
},
|
797 |
+
{
|
798 |
+
"epoch": 3.77,
|
799 |
+
"learning_rate": 0.00011444008109580884,
|
800 |
+
"loss": 0.0806,
|
801 |
+
"step": 132
|
802 |
+
},
|
803 |
+
{
|
804 |
+
"epoch": 3.8,
|
805 |
+
"learning_rate": 0.00011329202710118088,
|
806 |
+
"loss": 0.0821,
|
807 |
+
"step": 133
|
808 |
+
},
|
809 |
+
{
|
810 |
+
"epoch": 3.83,
|
811 |
+
"learning_rate": 0.00011214218683485158,
|
812 |
+
"loss": 0.0845,
|
813 |
+
"step": 134
|
814 |
+
},
|
815 |
+
{
|
816 |
+
"epoch": 3.86,
|
817 |
+
"learning_rate": 0.00011099071482006361,
|
818 |
+
"loss": 0.0842,
|
819 |
+
"step": 135
|
820 |
+
},
|
821 |
+
{
|
822 |
+
"epoch": 3.89,
|
823 |
+
"learning_rate": 0.00010983776579934482,
|
824 |
+
"loss": 0.0794,
|
825 |
+
"step": 136
|
826 |
+
},
|
827 |
+
{
|
828 |
+
"epoch": 3.91,
|
829 |
+
"learning_rate": 0.00010868349471371315,
|
830 |
+
"loss": 0.0775,
|
831 |
+
"step": 137
|
832 |
+
},
|
833 |
+
{
|
834 |
+
"epoch": 3.94,
|
835 |
+
"learning_rate": 0.00010752805668185442,
|
836 |
+
"loss": 0.0727,
|
837 |
+
"step": 138
|
838 |
+
},
|
839 |
+
{
|
840 |
+
"epoch": 3.97,
|
841 |
+
"learning_rate": 0.00010637160697927651,
|
842 |
+
"loss": 0.07,
|
843 |
+
"step": 139
|
844 |
+
},
|
845 |
+
{
|
846 |
+
"epoch": 4.0,
|
847 |
+
"learning_rate": 0.00010521430101744239,
|
848 |
+
"loss": 0.0919,
|
849 |
+
"step": 140
|
850 |
+
},
|
851 |
+
{
|
852 |
+
"epoch": 4.03,
|
853 |
+
"learning_rate": 0.00010405629432288488,
|
854 |
+
"loss": 0.0698,
|
855 |
+
"step": 141
|
856 |
+
},
|
857 |
+
{
|
858 |
+
"epoch": 4.06,
|
859 |
+
"learning_rate": 0.00010289774251630602,
|
860 |
+
"loss": 0.0675,
|
861 |
+
"step": 142
|
862 |
+
},
|
863 |
+
{
|
864 |
+
"epoch": 4.09,
|
865 |
+
"learning_rate": 0.00010173880129166358,
|
866 |
+
"loss": 0.0747,
|
867 |
+
"step": 143
|
868 |
+
},
|
869 |
+
{
|
870 |
+
"epoch": 4.11,
|
871 |
+
"learning_rate": 0.00010057962639524798,
|
872 |
+
"loss": 0.0745,
|
873 |
+
"step": 144
|
874 |
+
},
|
875 |
+
{
|
876 |
+
"epoch": 4.14,
|
877 |
+
"learning_rate": 9.942037360475205e-05,
|
878 |
+
"loss": 0.0637,
|
879 |
+
"step": 145
|
880 |
+
},
|
881 |
+
{
|
882 |
+
"epoch": 4.17,
|
883 |
+
"learning_rate": 9.826119870833643e-05,
|
884 |
+
"loss": 0.074,
|
885 |
+
"step": 146
|
886 |
+
},
|
887 |
+
{
|
888 |
+
"epoch": 4.2,
|
889 |
+
"learning_rate": 9.710225748369401e-05,
|
890 |
+
"loss": 0.0718,
|
891 |
+
"step": 147
|
892 |
+
},
|
893 |
+
{
|
894 |
+
"epoch": 4.23,
|
895 |
+
"learning_rate": 9.594370567711513e-05,
|
896 |
+
"loss": 0.0714,
|
897 |
+
"step": 148
|
898 |
+
},
|
899 |
+
{
|
900 |
+
"epoch": 4.26,
|
901 |
+
"learning_rate": 9.478569898255765e-05,
|
902 |
+
"loss": 0.0719,
|
903 |
+
"step": 149
|
904 |
+
},
|
905 |
+
{
|
906 |
+
"epoch": 4.29,
|
907 |
+
"learning_rate": 9.362839302072354e-05,
|
908 |
+
"loss": 0.0744,
|
909 |
+
"step": 150
|
910 |
+
},
|
911 |
+
{
|
912 |
+
"epoch": 4.31,
|
913 |
+
"learning_rate": 9.247194331814562e-05,
|
914 |
+
"loss": 0.068,
|
915 |
+
"step": 151
|
916 |
+
},
|
917 |
+
{
|
918 |
+
"epoch": 4.34,
|
919 |
+
"learning_rate": 9.131650528628687e-05,
|
920 |
+
"loss": 0.067,
|
921 |
+
"step": 152
|
922 |
+
},
|
923 |
+
{
|
924 |
+
"epoch": 4.37,
|
925 |
+
"learning_rate": 9.016223420065519e-05,
|
926 |
+
"loss": 0.0661,
|
927 |
+
"step": 153
|
928 |
+
},
|
929 |
+
{
|
930 |
+
"epoch": 4.4,
|
931 |
+
"learning_rate": 8.900928517993644e-05,
|
932 |
+
"loss": 0.0705,
|
933 |
+
"step": 154
|
934 |
+
},
|
935 |
+
{
|
936 |
+
"epoch": 4.43,
|
937 |
+
"learning_rate": 8.785781316514841e-05,
|
938 |
+
"loss": 0.0711,
|
939 |
+
"step": 155
|
940 |
+
},
|
941 |
+
{
|
942 |
+
"epoch": 4.46,
|
943 |
+
"learning_rate": 8.670797289881915e-05,
|
944 |
+
"loss": 0.0656,
|
945 |
+
"step": 156
|
946 |
+
},
|
947 |
+
{
|
948 |
+
"epoch": 4.49,
|
949 |
+
"learning_rate": 8.555991890419117e-05,
|
950 |
+
"loss": 0.0788,
|
951 |
+
"step": 157
|
952 |
+
},
|
953 |
+
{
|
954 |
+
"epoch": 4.51,
|
955 |
+
"learning_rate": 8.441380546445603e-05,
|
956 |
+
"loss": 0.0753,
|
957 |
+
"step": 158
|
958 |
+
},
|
959 |
+
{
|
960 |
+
"epoch": 4.54,
|
961 |
+
"learning_rate": 8.326978660202034e-05,
|
962 |
+
"loss": 0.0669,
|
963 |
+
"step": 159
|
964 |
+
},
|
965 |
+
{
|
966 |
+
"epoch": 4.57,
|
967 |
+
"learning_rate": 8.212801605780753e-05,
|
968 |
+
"loss": 0.0675,
|
969 |
+
"step": 160
|
970 |
+
},
|
971 |
+
{
|
972 |
+
"epoch": 4.6,
|
973 |
+
"learning_rate": 8.098864727059685e-05,
|
974 |
+
"loss": 0.0689,
|
975 |
+
"step": 161
|
976 |
+
},
|
977 |
+
{
|
978 |
+
"epoch": 4.63,
|
979 |
+
"learning_rate": 7.985183335640331e-05,
|
980 |
+
"loss": 0.0713,
|
981 |
+
"step": 162
|
982 |
+
},
|
983 |
+
{
|
984 |
+
"epoch": 4.66,
|
985 |
+
"learning_rate": 7.871772708790114e-05,
|
986 |
+
"loss": 0.0695,
|
987 |
+
"step": 163
|
988 |
+
},
|
989 |
+
{
|
990 |
+
"epoch": 4.69,
|
991 |
+
"learning_rate": 7.758648087389277e-05,
|
992 |
+
"loss": 0.0736,
|
993 |
+
"step": 164
|
994 |
+
},
|
995 |
+
{
|
996 |
+
"epoch": 4.71,
|
997 |
+
"learning_rate": 7.645824673882748e-05,
|
998 |
+
"loss": 0.0702,
|
999 |
+
"step": 165
|
1000 |
+
},
|
1001 |
+
{
|
1002 |
+
"epoch": 4.74,
|
1003 |
+
"learning_rate": 7.533317630237117e-05,
|
1004 |
+
"loss": 0.0742,
|
1005 |
+
"step": 166
|
1006 |
+
},
|
1007 |
+
{
|
1008 |
+
"epoch": 4.77,
|
1009 |
+
"learning_rate": 7.421142075903067e-05,
|
1010 |
+
"loss": 0.0656,
|
1011 |
+
"step": 167
|
1012 |
+
},
|
1013 |
+
{
|
1014 |
+
"epoch": 4.8,
|
1015 |
+
"learning_rate": 7.309313085783524e-05,
|
1016 |
+
"loss": 0.0717,
|
1017 |
+
"step": 168
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 4.83,
|
1021 |
+
"learning_rate": 7.197845688207805e-05,
|
1022 |
+
"loss": 0.0721,
|
1023 |
+
"step": 169
|
1024 |
+
},
|
1025 |
+
{
|
1026 |
+
"epoch": 4.86,
|
1027 |
+
"learning_rate": 7.086754862911982e-05,
|
1028 |
+
"loss": 0.0698,
|
1029 |
+
"step": 170
|
1030 |
+
},
|
1031 |
+
{
|
1032 |
+
"epoch": 4.89,
|
1033 |
+
"learning_rate": 6.976055539025818e-05,
|
1034 |
+
"loss": 0.0767,
|
1035 |
+
"step": 171
|
1036 |
+
},
|
1037 |
+
{
|
1038 |
+
"epoch": 4.91,
|
1039 |
+
"learning_rate": 6.865762593066513e-05,
|
1040 |
+
"loss": 0.0612,
|
1041 |
+
"step": 172
|
1042 |
+
},
|
1043 |
+
{
|
1044 |
+
"epoch": 4.94,
|
1045 |
+
"learning_rate": 6.755890846939454e-05,
|
1046 |
+
"loss": 0.0679,
|
1047 |
+
"step": 173
|
1048 |
+
},
|
1049 |
+
{
|
1050 |
+
"epoch": 4.97,
|
1051 |
+
"learning_rate": 6.646455065946386e-05,
|
1052 |
+
"loss": 0.0676,
|
1053 |
+
"step": 174
|
1054 |
+
},
|
1055 |
+
{
|
1056 |
+
"epoch": 5.0,
|
1057 |
+
"learning_rate": 6.537469956801128e-05,
|
1058 |
+
"loss": 0.0671,
|
1059 |
+
"step": 175
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 5.03,
|
1063 |
+
"learning_rate": 6.428950165653204e-05,
|
1064 |
+
"loss": 0.0581,
|
1065 |
+
"step": 176
|
1066 |
+
},
|
1067 |
+
{
|
1068 |
+
"epoch": 5.06,
|
1069 |
+
"learning_rate": 6.320910276119576e-05,
|
1070 |
+
"loss": 0.062,
|
1071 |
+
"step": 177
|
1072 |
+
},
|
1073 |
+
{
|
1074 |
+
"epoch": 5.09,
|
1075 |
+
"learning_rate": 6.213364807324818e-05,
|
1076 |
+
"loss": 0.0584,
|
1077 |
+
"step": 178
|
1078 |
+
},
|
1079 |
+
{
|
1080 |
+
"epoch": 5.11,
|
1081 |
+
"learning_rate": 6.106328211949928e-05,
|
1082 |
+
"loss": 0.0629,
|
1083 |
+
"step": 179
|
1084 |
+
},
|
1085 |
+
{
|
1086 |
+
"epoch": 5.14,
|
1087 |
+
"learning_rate": 5.999814874290084e-05,
|
1088 |
+
"loss": 0.0623,
|
1089 |
+
"step": 180
|
1090 |
+
},
|
1091 |
+
{
|
1092 |
+
"epoch": 5.17,
|
1093 |
+
"learning_rate": 5.893839108321584e-05,
|
1094 |
+
"loss": 0.0586,
|
1095 |
+
"step": 181
|
1096 |
+
},
|
1097 |
+
{
|
1098 |
+
"epoch": 5.2,
|
1099 |
+
"learning_rate": 5.7884151557782305e-05,
|
1100 |
+
"loss": 0.0597,
|
1101 |
+
"step": 182
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 5.23,
|
1105 |
+
"learning_rate": 5.68355718423746e-05,
|
1106 |
+
"loss": 0.0563,
|
1107 |
+
"step": 183
|
1108 |
+
},
|
1109 |
+
{
|
1110 |
+
"epoch": 5.26,
|
1111 |
+
"learning_rate": 5.579279285216369e-05,
|
1112 |
+
"loss": 0.0592,
|
1113 |
+
"step": 184
|
1114 |
+
},
|
1115 |
+
{
|
1116 |
+
"epoch": 5.29,
|
1117 |
+
"learning_rate": 5.475595472278024e-05,
|
1118 |
+
"loss": 0.0507,
|
1119 |
+
"step": 185
|
1120 |
+
},
|
1121 |
+
{
|
1122 |
+
"epoch": 5.31,
|
1123 |
+
"learning_rate": 5.372519679148227e-05,
|
1124 |
+
"loss": 0.051,
|
1125 |
+
"step": 186
|
1126 |
+
},
|
1127 |
+
{
|
1128 |
+
"epoch": 5.34,
|
1129 |
+
"learning_rate": 5.270065757843e-05,
|
1130 |
+
"loss": 0.0542,
|
1131 |
+
"step": 187
|
1132 |
+
},
|
1133 |
+
{
|
1134 |
+
"epoch": 5.37,
|
1135 |
+
"learning_rate": 5.168247476807053e-05,
|
1136 |
+
"loss": 0.0484,
|
1137 |
+
"step": 188
|
1138 |
+
},
|
1139 |
+
{
|
1140 |
+
"epoch": 5.4,
|
1141 |
+
"learning_rate": 5.067078519063514e-05,
|
1142 |
+
"loss": 0.0596,
|
1143 |
+
"step": 189
|
1144 |
+
},
|
1145 |
+
{
|
1146 |
+
"epoch": 5.43,
|
1147 |
+
"learning_rate": 4.9665724803750756e-05,
|
1148 |
+
"loss": 0.0586,
|
1149 |
+
"step": 190
|
1150 |
+
},
|
1151 |
+
{
|
1152 |
+
"epoch": 5.46,
|
1153 |
+
"learning_rate": 4.86674286741693e-05,
|
1154 |
+
"loss": 0.0631,
|
1155 |
+
"step": 191
|
1156 |
+
},
|
1157 |
+
{
|
1158 |
+
"epoch": 5.49,
|
1159 |
+
"learning_rate": 4.7676030959616526e-05,
|
1160 |
+
"loss": 0.0526,
|
1161 |
+
"step": 192
|
1162 |
+
},
|
1163 |
+
{
|
1164 |
+
"epoch": 5.51,
|
1165 |
+
"learning_rate": 4.669166489076283e-05,
|
1166 |
+
"loss": 0.0548,
|
1167 |
+
"step": 193
|
1168 |
+
},
|
1169 |
+
{
|
1170 |
+
"epoch": 5.54,
|
1171 |
+
"learning_rate": 4.571446275331903e-05,
|
1172 |
+
"loss": 0.0545,
|
1173 |
+
"step": 194
|
1174 |
+
},
|
1175 |
+
{
|
1176 |
+
"epoch": 5.57,
|
1177 |
+
"learning_rate": 4.4744555870258694e-05,
|
1178 |
+
"loss": 0.0473,
|
1179 |
+
"step": 195
|
1180 |
+
},
|
1181 |
+
{
|
1182 |
+
"epoch": 5.6,
|
1183 |
+
"learning_rate": 4.378207458417035e-05,
|
1184 |
+
"loss": 0.0606,
|
1185 |
+
"step": 196
|
1186 |
+
},
|
1187 |
+
{
|
1188 |
+
"epoch": 5.63,
|
1189 |
+
"learning_rate": 4.282714823974088e-05,
|
1190 |
+
"loss": 0.061,
|
1191 |
+
"step": 197
|
1192 |
+
},
|
1193 |
+
{
|
1194 |
+
"epoch": 5.66,
|
1195 |
+
"learning_rate": 4.1879905166373614e-05,
|
1196 |
+
"loss": 0.0571,
|
1197 |
+
"step": 198
|
1198 |
+
},
|
1199 |
+
{
|
1200 |
+
"epoch": 5.69,
|
1201 |
+
"learning_rate": 4.094047266094225e-05,
|
1202 |
+
"loss": 0.0578,
|
1203 |
+
"step": 199
|
1204 |
+
},
|
1205 |
+
{
|
1206 |
+
"epoch": 5.71,
|
1207 |
+
"learning_rate": 4.0008976970684176e-05,
|
1208 |
+
"loss": 0.059,
|
1209 |
+
"step": 200
|
1210 |
+
},
|
1211 |
+
{
|
1212 |
+
"epoch": 5.74,
|
1213 |
+
"learning_rate": 3.908554327623425e-05,
|
1214 |
+
"loss": 0.0572,
|
1215 |
+
"step": 201
|
1216 |
+
},
|
1217 |
+
{
|
1218 |
+
"epoch": 5.77,
|
1219 |
+
"learning_rate": 3.817029567480228e-05,
|
1220 |
+
"loss": 0.0551,
|
1221 |
+
"step": 202
|
1222 |
+
},
|
1223 |
+
{
|
1224 |
+
"epoch": 5.8,
|
1225 |
+
"learning_rate": 3.726335716349612e-05,
|
1226 |
+
"loss": 0.057,
|
1227 |
+
"step": 203
|
1228 |
+
},
|
1229 |
+
{
|
1230 |
+
"epoch": 5.83,
|
1231 |
+
"learning_rate": 3.6364849622792266e-05,
|
1232 |
+
"loss": 0.0584,
|
1233 |
+
"step": 204
|
1234 |
+
},
|
1235 |
+
{
|
1236 |
+
"epoch": 5.86,
|
1237 |
+
"learning_rate": 3.547489380015701e-05,
|
1238 |
+
"loss": 0.0592,
|
1239 |
+
"step": 205
|
1240 |
+
},
|
1241 |
+
{
|
1242 |
+
"epoch": 5.89,
|
1243 |
+
"learning_rate": 3.459360929381931e-05,
|
1244 |
+
"loss": 0.0528,
|
1245 |
+
"step": 206
|
1246 |
+
},
|
1247 |
+
{
|
1248 |
+
"epoch": 5.91,
|
1249 |
+
"learning_rate": 3.372111453669864e-05,
|
1250 |
+
"loss": 0.0587,
|
1251 |
+
"step": 207
|
1252 |
+
},
|
1253 |
+
{
|
1254 |
+
"epoch": 5.94,
|
1255 |
+
"learning_rate": 3.285752678048892e-05,
|
1256 |
+
"loss": 0.0527,
|
1257 |
+
"step": 208
|
1258 |
+
},
|
1259 |
+
{
|
1260 |
+
"epoch": 5.97,
|
1261 |
+
"learning_rate": 3.2002962079901744e-05,
|
1262 |
+
"loss": 0.0539,
|
1263 |
+
"step": 209
|
1264 |
+
},
|
1265 |
+
{
|
1266 |
+
"epoch": 6.0,
|
1267 |
+
"learning_rate": 3.115753527706986e-05,
|
1268 |
+
"loss": 0.0479,
|
1269 |
+
"step": 210
|
1270 |
+
},
|
1271 |
+
{
|
1272 |
+
"epoch": 6.03,
|
1273 |
+
"learning_rate": 3.0321359986114096e-05,
|
1274 |
+
"loss": 0.0436,
|
1275 |
+
"step": 211
|
1276 |
+
},
|
1277 |
+
{
|
1278 |
+
"epoch": 6.06,
|
1279 |
+
"learning_rate": 2.9494548577875192e-05,
|
1280 |
+
"loss": 0.0427,
|
1281 |
+
"step": 212
|
1282 |
+
},
|
1283 |
+
{
|
1284 |
+
"epoch": 6.09,
|
1285 |
+
"learning_rate": 2.8677212164812462e-05,
|
1286 |
+
"loss": 0.043,
|
1287 |
+
"step": 213
|
1288 |
+
},
|
1289 |
+
{
|
1290 |
+
"epoch": 6.11,
|
1291 |
+
"learning_rate": 2.7869460586071873e-05,
|
1292 |
+
"loss": 0.0393,
|
1293 |
+
"step": 214
|
1294 |
+
},
|
1295 |
+
{
|
1296 |
+
"epoch": 6.14,
|
1297 |
+
"learning_rate": 2.7071402392725097e-05,
|
1298 |
+
"loss": 0.0385,
|
1299 |
+
"step": 215
|
1300 |
+
},
|
1301 |
+
{
|
1302 |
+
"epoch": 6.17,
|
1303 |
+
"learning_rate": 2.6283144833181783e-05,
|
1304 |
+
"loss": 0.0401,
|
1305 |
+
"step": 216
|
1306 |
+
},
|
1307 |
+
{
|
1308 |
+
"epoch": 6.2,
|
1309 |
+
"learning_rate": 2.5504793838776586e-05,
|
1310 |
+
"loss": 0.0414,
|
1311 |
+
"step": 217
|
1312 |
+
},
|
1313 |
+
{
|
1314 |
+
"epoch": 6.23,
|
1315 |
+
"learning_rate": 2.473645400953366e-05,
|
1316 |
+
"loss": 0.0436,
|
1317 |
+
"step": 218
|
1318 |
+
},
|
1319 |
+
{
|
1320 |
+
"epoch": 6.26,
|
1321 |
+
"learning_rate": 2.3978228600109565e-05,
|
1322 |
+
"loss": 0.0406,
|
1323 |
+
"step": 219
|
1324 |
+
},
|
1325 |
+
{
|
1326 |
+
"epoch": 6.29,
|
1327 |
+
"learning_rate": 2.3230219505917426e-05,
|
1328 |
+
"loss": 0.0464,
|
1329 |
+
"step": 220
|
1330 |
+
},
|
1331 |
+
{
|
1332 |
+
"epoch": 6.31,
|
1333 |
+
"learning_rate": 2.249252724943336e-05,
|
1334 |
+
"loss": 0.0414,
|
1335 |
+
"step": 221
|
1336 |
+
},
|
1337 |
+
{
|
1338 |
+
"epoch": 6.34,
|
1339 |
+
"learning_rate": 2.176525096668769e-05,
|
1340 |
+
"loss": 0.0398,
|
1341 |
+
"step": 222
|
1342 |
+
},
|
1343 |
+
{
|
1344 |
+
"epoch": 6.37,
|
1345 |
+
"learning_rate": 2.1048488393942454e-05,
|
1346 |
+
"loss": 0.0408,
|
1347 |
+
"step": 223
|
1348 |
+
},
|
1349 |
+
{
|
1350 |
+
"epoch": 6.4,
|
1351 |
+
"learning_rate": 2.0342335854556737e-05,
|
1352 |
+
"loss": 0.0419,
|
1353 |
+
"step": 224
|
1354 |
+
},
|
1355 |
+
{
|
1356 |
+
"epoch": 6.43,
|
1357 |
+
"learning_rate": 1.964688824604234e-05,
|
1358 |
+
"loss": 0.0471,
|
1359 |
+
"step": 225
|
1360 |
+
},
|
1361 |
+
{
|
1362 |
+
"epoch": 6.46,
|
1363 |
+
"learning_rate": 1.896223902731058e-05,
|
1364 |
+
"loss": 0.0457,
|
1365 |
+
"step": 226
|
1366 |
+
},
|
1367 |
+
{
|
1368 |
+
"epoch": 6.49,
|
1369 |
+
"learning_rate": 1.828848020611288e-05,
|
1370 |
+
"loss": 0.0375,
|
1371 |
+
"step": 227
|
1372 |
+
},
|
1373 |
+
{
|
1374 |
+
"epoch": 6.51,
|
1375 |
+
"learning_rate": 1.762570232667595e-05,
|
1376 |
+
"loss": 0.0352,
|
1377 |
+
"step": 228
|
1378 |
+
},
|
1379 |
+
{
|
1380 |
+
"epoch": 6.54,
|
1381 |
+
"learning_rate": 1.6973994457534026e-05,
|
1382 |
+
"loss": 0.0403,
|
1383 |
+
"step": 229
|
1384 |
+
},
|
1385 |
+
{
|
1386 |
+
"epoch": 6.57,
|
1387 |
+
"learning_rate": 1.6333444179559078e-05,
|
1388 |
+
"loss": 0.0402,
|
1389 |
+
"step": 230
|
1390 |
+
},
|
1391 |
+
{
|
1392 |
+
"epoch": 6.6,
|
1393 |
+
"learning_rate": 1.5704137574191203e-05,
|
1394 |
+
"loss": 0.0406,
|
1395 |
+
"step": 231
|
1396 |
+
},
|
1397 |
+
{
|
1398 |
+
"epoch": 6.63,
|
1399 |
+
"learning_rate": 1.5086159211870442e-05,
|
1400 |
+
"loss": 0.0407,
|
1401 |
+
"step": 232
|
1402 |
+
},
|
1403 |
+
{
|
1404 |
+
"epoch": 6.66,
|
1405 |
+
"learning_rate": 1.447959214067155e-05,
|
1406 |
+
"loss": 0.0361,
|
1407 |
+
"step": 233
|
1408 |
+
},
|
1409 |
+
{
|
1410 |
+
"epoch": 6.69,
|
1411 |
+
"learning_rate": 1.3884517875143544e-05,
|
1412 |
+
"loss": 0.0389,
|
1413 |
+
"step": 234
|
1414 |
+
},
|
1415 |
+
{
|
1416 |
+
"epoch": 6.71,
|
1417 |
+
"learning_rate": 1.3301016385355092e-05,
|
1418 |
+
"loss": 0.0431,
|
1419 |
+
"step": 235
|
1420 |
+
},
|
1421 |
+
{
|
1422 |
+
"epoch": 6.74,
|
1423 |
+
"learning_rate": 1.2729166086147803e-05,
|
1424 |
+
"loss": 0.0439,
|
1425 |
+
"step": 236
|
1426 |
+
},
|
1427 |
+
{
|
1428 |
+
"epoch": 6.77,
|
1429 |
+
"learning_rate": 1.2169043826598058e-05,
|
1430 |
+
"loss": 0.042,
|
1431 |
+
"step": 237
|
1432 |
+
},
|
1433 |
+
{
|
1434 |
+
"epoch": 6.8,
|
1435 |
+
"learning_rate": 1.1620724879689792e-05,
|
1436 |
+
"loss": 0.0447,
|
1437 |
+
"step": 238
|
1438 |
+
},
|
1439 |
+
{
|
1440 |
+
"epoch": 6.83,
|
1441 |
+
"learning_rate": 1.1084282932198541e-05,
|
1442 |
+
"loss": 0.0386,
|
1443 |
+
"step": 239
|
1444 |
+
},
|
1445 |
+
{
|
1446 |
+
"epoch": 6.86,
|
1447 |
+
"learning_rate": 1.0559790074789133e-05,
|
1448 |
+
"loss": 0.0408,
|
1449 |
+
"step": 240
|
1450 |
+
},
|
1451 |
+
{
|
1452 |
+
"epoch": 6.89,
|
1453 |
+
"learning_rate": 1.0047316792327499e-05,
|
1454 |
+
"loss": 0.0384,
|
1455 |
+
"step": 241
|
1456 |
+
},
|
1457 |
+
{
|
1458 |
+
"epoch": 6.91,
|
1459 |
+
"learning_rate": 9.546931954408622e-06,
|
1460 |
+
"loss": 0.0385,
|
1461 |
+
"step": 242
|
1462 |
+
},
|
1463 |
+
{
|
1464 |
+
"epoch": 6.94,
|
1465 |
+
"learning_rate": 9.058702806101172e-06,
|
1466 |
+
"loss": 0.0403,
|
1467 |
+
"step": 243
|
1468 |
+
},
|
1469 |
+
{
|
1470 |
+
"epoch": 6.97,
|
1471 |
+
"learning_rate": 8.58269495891081e-06,
|
1472 |
+
"loss": 0.0467,
|
1473 |
+
"step": 244
|
1474 |
+
},
|
1475 |
+
{
|
1476 |
+
"epoch": 7.0,
|
1477 |
+
"learning_rate": 8.118972381962853e-06,
|
1478 |
+
"loss": 0.0411,
|
1479 |
+
"step": 245
|
1480 |
+
},
|
1481 |
+
{
|
1482 |
+
"epoch": 7.03,
|
1483 |
+
"learning_rate": 7.6675973934056e-06,
|
1484 |
+
"loss": 0.0347,
|
1485 |
+
"step": 246
|
1486 |
+
},
|
1487 |
+
{
|
1488 |
+
"epoch": 7.06,
|
1489 |
+
"learning_rate": 7.228630652035717e-06,
|
1490 |
+
"loss": 0.0344,
|
1491 |
+
"step": 247
|
1492 |
+
},
|
1493 |
+
{
|
1494 |
+
"epoch": 7.09,
|
1495 |
+
"learning_rate": 6.802131149146373e-06,
|
1496 |
+
"loss": 0.0322,
|
1497 |
+
"step": 248
|
1498 |
+
},
|
1499 |
+
{
|
1500 |
+
"epoch": 7.11,
|
1501 |
+
"learning_rate": 6.388156200599726e-06,
|
1502 |
+
"loss": 0.0304,
|
1503 |
+
"step": 249
|
1504 |
+
},
|
1505 |
+
{
|
1506 |
+
"epoch": 7.14,
|
1507 |
+
"learning_rate": 5.986761439124289e-06,
|
1508 |
+
"loss": 0.0346,
|
1509 |
+
"step": 250
|
1510 |
+
},
|
1511 |
+
{
|
1512 |
+
"epoch": 7.17,
|
1513 |
+
"learning_rate": 5.598000806838766e-06,
|
1514 |
+
"loss": 0.0335,
|
1515 |
+
"step": 251
|
1516 |
+
},
|
1517 |
+
{
|
1518 |
+
"epoch": 7.2,
|
1519 |
+
"learning_rate": 5.221926548002876e-06,
|
1520 |
+
"loss": 0.0301,
|
1521 |
+
"step": 252
|
1522 |
+
},
|
1523 |
+
{
|
1524 |
+
"epoch": 7.23,
|
1525 |
+
"learning_rate": 4.858589201996433e-06,
|
1526 |
+
"loss": 0.032,
|
1527 |
+
"step": 253
|
1528 |
+
},
|
1529 |
+
{
|
1530 |
+
"epoch": 7.26,
|
1531 |
+
"learning_rate": 4.508037596527526e-06,
|
1532 |
+
"loss": 0.0318,
|
1533 |
+
"step": 254
|
1534 |
+
},
|
1535 |
+
{
|
1536 |
+
"epoch": 7.29,
|
1537 |
+
"learning_rate": 4.170318841070708e-06,
|
1538 |
+
"loss": 0.0279,
|
1539 |
+
"step": 255
|
1540 |
+
},
|
1541 |
+
{
|
1542 |
+
"epoch": 7.31,
|
1543 |
+
"learning_rate": 3.845478320536178e-06,
|
1544 |
+
"loss": 0.0265,
|
1545 |
+
"step": 256
|
1546 |
+
},
|
1547 |
+
{
|
1548 |
+
"epoch": 7.34,
|
1549 |
+
"learning_rate": 3.53355968917054e-06,
|
1550 |
+
"loss": 0.0322,
|
1551 |
+
"step": 257
|
1552 |
+
},
|
1553 |
+
{
|
1554 |
+
"epoch": 7.37,
|
1555 |
+
"learning_rate": 3.2346048646903494e-06,
|
1556 |
+
"loss": 0.0336,
|
1557 |
+
"step": 258
|
1558 |
+
},
|
1559 |
+
{
|
1560 |
+
"epoch": 7.4,
|
1561 |
+
"learning_rate": 2.9486540226488557e-06,
|
1562 |
+
"loss": 0.0303,
|
1563 |
+
"step": 259
|
1564 |
+
},
|
1565 |
+
{
|
1566 |
+
"epoch": 7.43,
|
1567 |
+
"learning_rate": 2.6757455910370488e-06,
|
1568 |
+
"loss": 0.0298,
|
1569 |
+
"step": 260
|
1570 |
+
},
|
1571 |
+
{
|
1572 |
+
"epoch": 7.46,
|
1573 |
+
"learning_rate": 2.4159162451193097e-06,
|
1574 |
+
"loss": 0.0256,
|
1575 |
+
"step": 261
|
1576 |
+
},
|
1577 |
+
{
|
1578 |
+
"epoch": 7.49,
|
1579 |
+
"learning_rate": 2.1692009025048422e-06,
|
1580 |
+
"loss": 0.0315,
|
1581 |
+
"step": 262
|
1582 |
+
},
|
1583 |
+
{
|
1584 |
+
"epoch": 7.51,
|
1585 |
+
"learning_rate": 1.9356327184551714e-06,
|
1586 |
+
"loss": 0.0266,
|
1587 |
+
"step": 263
|
1588 |
+
},
|
1589 |
+
{
|
1590 |
+
"epoch": 7.54,
|
1591 |
+
"learning_rate": 1.7152430814285303e-06,
|
1592 |
+
"loss": 0.029,
|
1593 |
+
"step": 264
|
1594 |
+
},
|
1595 |
+
{
|
1596 |
+
"epoch": 7.57,
|
1597 |
+
"learning_rate": 1.5080616088616884e-06,
|
1598 |
+
"loss": 0.0289,
|
1599 |
+
"step": 265
|
1600 |
+
},
|
1601 |
+
{
|
1602 |
+
"epoch": 7.6,
|
1603 |
+
"learning_rate": 1.3141161431896808e-06,
|
1604 |
+
"loss": 0.0285,
|
1605 |
+
"step": 266
|
1606 |
+
},
|
1607 |
+
{
|
1608 |
+
"epoch": 7.63,
|
1609 |
+
"learning_rate": 1.1334327481042573e-06,
|
1610 |
+
"loss": 0.0285,
|
1611 |
+
"step": 267
|
1612 |
+
},
|
1613 |
+
{
|
1614 |
+
"epoch": 7.66,
|
1615 |
+
"learning_rate": 9.660357050512158e-07,
|
1616 |
+
"loss": 0.0329,
|
1617 |
+
"step": 268
|
1618 |
+
},
|
1619 |
+
{
|
1620 |
+
"epoch": 7.69,
|
1621 |
+
"learning_rate": 8.119475099673036e-07,
|
1622 |
+
"loss": 0.0303,
|
1623 |
+
"step": 269
|
1624 |
+
},
|
1625 |
+
{
|
1626 |
+
"epoch": 7.71,
|
1627 |
+
"learning_rate": 6.711888702570556e-07,
|
1628 |
+
"loss": 0.0282,
|
1629 |
+
"step": 270
|
1630 |
+
},
|
1631 |
+
{
|
1632 |
+
"epoch": 7.74,
|
1633 |
+
"learning_rate": 5.437787020100115e-07,
|
1634 |
+
"loss": 0.0271,
|
1635 |
+
"step": 271
|
1636 |
+
},
|
1637 |
+
{
|
1638 |
+
"epoch": 7.77,
|
1639 |
+
"learning_rate": 4.297341274586475e-07,
|
1640 |
+
"loss": 0.0303,
|
1641 |
+
"step": 272
|
1642 |
+
},
|
1643 |
+
{
|
1644 |
+
"epoch": 7.8,
|
1645 |
+
"learning_rate": 3.290704726773619e-07,
|
1646 |
+
"loss": 0.0283,
|
1647 |
+
"step": 273
|
1648 |
+
},
|
1649 |
+
{
|
1650 |
+
"epoch": 7.83,
|
1651 |
+
"learning_rate": 2.418012655228452e-07,
|
1652 |
+
"loss": 0.0305,
|
1653 |
+
"step": 274
|
1654 |
+
},
|
1655 |
+
{
|
1656 |
+
"epoch": 7.86,
|
1657 |
+
"learning_rate": 1.6793823381614505e-07,
|
1658 |
+
"loss": 0.0243,
|
1659 |
+
"step": 275
|
1660 |
+
},
|
1661 |
+
{
|
1662 |
+
"epoch": 7.89,
|
1663 |
+
"learning_rate": 1.0749130376659366e-07,
|
1664 |
+
"loss": 0.0271,
|
1665 |
+
"step": 276
|
1666 |
+
},
|
1667 |
+
{
|
1668 |
+
"epoch": 7.91,
|
1669 |
+
"learning_rate": 6.046859863781951e-08,
|
1670 |
+
"loss": 0.0288,
|
1671 |
+
"step": 277
|
1672 |
+
},
|
1673 |
+
{
|
1674 |
+
"epoch": 7.94,
|
1675 |
+
"learning_rate": 2.6876437656153665e-08,
|
1676 |
+
"loss": 0.0278,
|
1677 |
+
"step": 278
|
1678 |
+
},
|
1679 |
+
{
|
1680 |
+
"epoch": 7.97,
|
1681 |
+
"learning_rate": 6.719335161364804e-09,
|
1682 |
+
"loss": 0.0289,
|
1683 |
+
"step": 279
|
1684 |
+
},
|
1685 |
+
{
|
1686 |
+
"epoch": 8.0,
|
1687 |
+
"learning_rate": 0.0,
|
1688 |
+
"loss": 0.0321,
|
1689 |
+
"step": 280
|
1690 |
+
},
|
1691 |
+
{
|
1692 |
+
"epoch": 8.0,
|
1693 |
+
"step": 280,
|
1694 |
+
"total_flos": 71501879181312.0,
|
1695 |
+
"train_loss": 0.09845655608390058,
|
1696 |
+
"train_runtime": 2894.849,
|
1697 |
+
"train_samples_per_second": 12.281,
|
1698 |
+
"train_steps_per_second": 0.097
|
1699 |
+
}
|
1700 |
+
],
|
1701 |
+
"logging_steps": 1.0,
|
1702 |
+
"max_steps": 280,
|
1703 |
+
"num_input_tokens_seen": 0,
|
1704 |
+
"num_train_epochs": 8,
|
1705 |
+
"save_steps": 72,
|
1706 |
+
"total_flos": 71501879181312.0,
|
1707 |
+
"train_batch_size": 16,
|
1708 |
+
"trial_name": null,
|
1709 |
+
"trial_params": null
|
1710 |
+
}
|