🍻 cheers
Browse files- all_results.json +4 -4
- train_results.json +4 -4
- trainer_state.json +5 -5
all_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 4.0,
|
3 |
"total_flos": 1.4878462006001664e+16,
|
4 |
-
"train_loss": 0.
|
5 |
-
"train_runtime":
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second":
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 4.0,
|
3 |
"total_flos": 1.4878462006001664e+16,
|
4 |
+
"train_loss": 0.4666179021199544,
|
5 |
+
"train_runtime": 21.7213,
|
6 |
+
"train_samples_per_second": 8.839,
|
7 |
+
"train_steps_per_second": 0.552
|
8 |
}
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 4.0,
|
3 |
"total_flos": 1.4878462006001664e+16,
|
4 |
-
"train_loss": 0.
|
5 |
-
"train_runtime":
|
6 |
-
"train_samples_per_second":
|
7 |
-
"train_steps_per_second":
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 4.0,
|
3 |
"total_flos": 1.4878462006001664e+16,
|
4 |
+
"train_loss": 0.4666179021199544,
|
5 |
+
"train_runtime": 21.7213,
|
6 |
+
"train_samples_per_second": 8.839,
|
7 |
+
"train_steps_per_second": 0.552
|
8 |
}
|
trainer_state.json
CHANGED
@@ -10,17 +10,17 @@
|
|
10 |
{
|
11 |
"epoch": 3.33,
|
12 |
"learning_rate": 3.3333333333333335e-05,
|
13 |
-
"loss": 0.
|
14 |
"step": 10
|
15 |
},
|
16 |
{
|
17 |
"epoch": 4.0,
|
18 |
"step": 12,
|
19 |
"total_flos": 1.4878462006001664e+16,
|
20 |
-
"train_loss": 0.
|
21 |
-
"train_runtime":
|
22 |
-
"train_samples_per_second":
|
23 |
-
"train_steps_per_second":
|
24 |
}
|
25 |
],
|
26 |
"max_steps": 12,
|
|
|
10 |
{
|
11 |
"epoch": 3.33,
|
12 |
"learning_rate": 3.3333333333333335e-05,
|
13 |
+
"loss": 0.5032,
|
14 |
"step": 10
|
15 |
},
|
16 |
{
|
17 |
"epoch": 4.0,
|
18 |
"step": 12,
|
19 |
"total_flos": 1.4878462006001664e+16,
|
20 |
+
"train_loss": 0.4666179021199544,
|
21 |
+
"train_runtime": 21.7213,
|
22 |
+
"train_samples_per_second": 8.839,
|
23 |
+
"train_steps_per_second": 0.552
|
24 |
}
|
25 |
],
|
26 |
"max_steps": 12,
|