End of training
Browse files- all_results.json +19 -0
- train_results.json +8 -0
- with_eval_results.json +14 -0
all_results.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 15.0,
|
3 |
+
"eval_CLS loss": 3.0579187870025635,
|
4 |
+
"eval_CLS_accuracy": 0.6109510086455331,
|
5 |
+
"eval_CLS_f1": 0.6053901347320909,
|
6 |
+
"eval_CLS_precision": 0.6053977980738959,
|
7 |
+
"eval_CLS_recall": 0.6109510086455331,
|
8 |
+
"eval_LM loss": 3.962585210800171,
|
9 |
+
"eval_Perplexity": "52.59",
|
10 |
+
"eval_loss": 4.727580547332764,
|
11 |
+
"eval_runtime": 147.9435,
|
12 |
+
"eval_samples_per_second": 11.727,
|
13 |
+
"eval_steps_per_second": 5.867,
|
14 |
+
"total_flos": 5.44049307058176e+16,
|
15 |
+
"train_loss": 3.958396730157301,
|
16 |
+
"train_runtime": 23820.1862,
|
17 |
+
"train_samples_per_second": 4.37,
|
18 |
+
"train_steps_per_second": 2.185
|
19 |
+
}
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 15.0,
|
3 |
+
"total_flos": 5.44049307058176e+16,
|
4 |
+
"train_loss": 3.958396730157301,
|
5 |
+
"train_runtime": 23820.1862,
|
6 |
+
"train_samples_per_second": 4.37,
|
7 |
+
"train_steps_per_second": 2.185
|
8 |
+
}
|
with_eval_results.json
ADDED
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 15.0,
|
3 |
+
"eval_CLS loss": 3.0579187870025635,
|
4 |
+
"eval_CLS_accuracy": 0.6109510086455331,
|
5 |
+
"eval_CLS_f1": 0.6053901347320909,
|
6 |
+
"eval_CLS_precision": 0.6053977980738959,
|
7 |
+
"eval_CLS_recall": 0.6109510086455331,
|
8 |
+
"eval_LM loss": 3.962585210800171,
|
9 |
+
"eval_Perplexity": "52.59",
|
10 |
+
"eval_loss": 4.727580547332764,
|
11 |
+
"eval_runtime": 147.9435,
|
12 |
+
"eval_samples_per_second": 11.727,
|
13 |
+
"eval_steps_per_second": 5.867
|
14 |
+
}
|