End of training
Browse files- all_results.json +6 -6
- eval_results.json +3 -3
- train_results.json +3 -3
- trainer_state.json +0 -0
all_results.json
CHANGED
|
@@ -1,12 +1,12 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1000.0,
|
| 3 |
"eval_loss": 1.7884608507156372,
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second":
|
| 7 |
"total_flos": 3.992166672e+16,
|
| 8 |
"train_loss": 0.9881956270345053,
|
| 9 |
-
"train_runtime":
|
| 10 |
-
"train_samples_per_second": 25.
|
| 11 |
-
"train_steps_per_second": 6.
|
| 12 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1000.0,
|
| 3 |
"eval_loss": 1.7884608507156372,
|
| 4 |
+
"eval_runtime": 0.5864,
|
| 5 |
+
"eval_samples_per_second": 61.397,
|
| 6 |
+
"eval_steps_per_second": 15.349,
|
| 7 |
"total_flos": 3.992166672e+16,
|
| 8 |
"train_loss": 0.9881956270345053,
|
| 9 |
+
"train_runtime": 11653.197,
|
| 10 |
+
"train_samples_per_second": 25.658,
|
| 11 |
+
"train_steps_per_second": 6.436
|
| 12 |
}
|
eval_results.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 1000.0,
|
| 3 |
"eval_loss": 1.7884608507156372,
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second":
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 1000.0,
|
| 3 |
"eval_loss": 1.7884608507156372,
|
| 4 |
+
"eval_runtime": 0.5864,
|
| 5 |
+
"eval_samples_per_second": 61.397,
|
| 6 |
+
"eval_steps_per_second": 15.349
|
| 7 |
}
|
train_results.json
CHANGED
|
@@ -2,7 +2,7 @@
|
|
| 2 |
"epoch": 1000.0,
|
| 3 |
"total_flos": 3.992166672e+16,
|
| 4 |
"train_loss": 0.9881956270345053,
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second": 25.
|
| 7 |
-
"train_steps_per_second": 6.
|
| 8 |
}
|
|
|
|
| 2 |
"epoch": 1000.0,
|
| 3 |
"total_flos": 3.992166672e+16,
|
| 4 |
"train_loss": 0.9881956270345053,
|
| 5 |
+
"train_runtime": 11653.197,
|
| 6 |
+
"train_samples_per_second": 25.658,
|
| 7 |
+
"train_steps_per_second": 6.436
|
| 8 |
}
|
trainer_state.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|