End of training
Browse files- README.md +15 -3
- all_results.json +15 -0
- eval_results.json +10 -0
- train_results.json +8 -0
- trainer_state.json +3958 -0
README.md
CHANGED
|
@@ -1,11 +1,23 @@
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
|
|
|
|
|
|
| 4 |
metrics:
|
| 5 |
- accuracy
|
| 6 |
model-index:
|
| 7 |
- name: HBERTv1_emb_compress_48_L12_H64_A2
|
| 8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -13,9 +25,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 13 |
|
| 14 |
# HBERTv1_emb_compress_48_L12_H64_A2
|
| 15 |
|
| 16 |
-
This model is a fine-tuned version of [](https://huggingface.co/) on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
-
- Loss: 6.
|
| 19 |
- Accuracy: 0.1285
|
| 20 |
|
| 21 |
## Model description
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- gokuls/wiki_book_corpus_complete_processed_bert_dataset
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: HBERTv1_emb_compress_48_L12_H64_A2
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Masked Language Modeling
|
| 13 |
+
type: fill-mask
|
| 14 |
+
dataset:
|
| 15 |
+
name: gokuls/wiki_book_corpus_complete_processed_bert_dataset
|
| 16 |
+
type: gokuls/wiki_book_corpus_complete_processed_bert_dataset
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.12850906143802152
|
| 21 |
---
|
| 22 |
|
| 23 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 25 |
|
| 26 |
# HBERTv1_emb_compress_48_L12_H64_A2
|
| 27 |
|
| 28 |
+
This model is a fine-tuned version of [](https://huggingface.co/) on the gokuls/wiki_book_corpus_complete_processed_bert_dataset dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
+
- Loss: 6.4079
|
| 31 |
- Accuracy: 0.1285
|
| 32 |
|
| 33 |
## Model description
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 0.12850906143802152,
|
| 4 |
+
"eval_loss": 6.407903671264648,
|
| 5 |
+
"eval_runtime": 602.4473,
|
| 6 |
+
"eval_samples": 308356,
|
| 7 |
+
"eval_samples_per_second": 511.839,
|
| 8 |
+
"eval_steps_per_second": 5.333,
|
| 9 |
+
"perplexity": 606.6206708599559,
|
| 10 |
+
"train_loss": 6.680038520005282,
|
| 11 |
+
"train_runtime": 123390.5046,
|
| 12 |
+
"train_samples": 5858758,
|
| 13 |
+
"train_samples_per_second": 237.407,
|
| 14 |
+
"train_steps_per_second": 2.473
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 0.12850906143802152,
|
| 4 |
+
"eval_loss": 6.407903671264648,
|
| 5 |
+
"eval_runtime": 602.4473,
|
| 6 |
+
"eval_samples": 308356,
|
| 7 |
+
"eval_samples_per_second": 511.839,
|
| 8 |
+
"eval_steps_per_second": 5.333,
|
| 9 |
+
"perplexity": 606.6206708599559
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 6.680038520005282,
|
| 4 |
+
"train_runtime": 123390.5046,
|
| 5 |
+
"train_samples": 5858758,
|
| 6 |
+
"train_samples_per_second": 237.407,
|
| 7 |
+
"train_steps_per_second": 2.473
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3958 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 6.406748294830322,
|
| 3 |
+
"best_model_checkpoint": "HBERTv1_emb_compress_48_L12_H64_A2/checkpoint-290000",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 10000,
|
| 6 |
+
"global_step": 305145,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01,
|
| 13 |
+
"learning_rate": 5.000000000000001e-07,
|
| 14 |
+
"loss": 10.3179,
|
| 15 |
+
"step": 500
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.02,
|
| 19 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 20 |
+
"loss": 10.2814,
|
| 21 |
+
"step": 1000
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.02,
|
| 25 |
+
"learning_rate": 1.5e-06,
|
| 26 |
+
"loss": 10.2576,
|
| 27 |
+
"step": 1500
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.03,
|
| 31 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 32 |
+
"loss": 10.2271,
|
| 33 |
+
"step": 2000
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.04,
|
| 37 |
+
"learning_rate": 2.5e-06,
|
| 38 |
+
"loss": 10.1895,
|
| 39 |
+
"step": 2500
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.05,
|
| 43 |
+
"learning_rate": 3e-06,
|
| 44 |
+
"loss": 10.1441,
|
| 45 |
+
"step": 3000
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.06,
|
| 49 |
+
"learning_rate": 3.5e-06,
|
| 50 |
+
"loss": 10.0896,
|
| 51 |
+
"step": 3500
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.07,
|
| 55 |
+
"learning_rate": 4.000000000000001e-06,
|
| 56 |
+
"loss": 10.0257,
|
| 57 |
+
"step": 4000
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.07,
|
| 61 |
+
"learning_rate": 4.5e-06,
|
| 62 |
+
"loss": 9.9517,
|
| 63 |
+
"step": 4500
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.08,
|
| 67 |
+
"learning_rate": 5e-06,
|
| 68 |
+
"loss": 9.8689,
|
| 69 |
+
"step": 5000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.09,
|
| 73 |
+
"learning_rate": 5.500000000000001e-06,
|
| 74 |
+
"loss": 9.7764,
|
| 75 |
+
"step": 5500
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.1,
|
| 79 |
+
"learning_rate": 6e-06,
|
| 80 |
+
"loss": 9.6759,
|
| 81 |
+
"step": 6000
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.11,
|
| 85 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 86 |
+
"loss": 9.5681,
|
| 87 |
+
"step": 6500
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.11,
|
| 91 |
+
"learning_rate": 7e-06,
|
| 92 |
+
"loss": 9.4531,
|
| 93 |
+
"step": 7000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.12,
|
| 97 |
+
"learning_rate": 7.500000000000001e-06,
|
| 98 |
+
"loss": 9.3312,
|
| 99 |
+
"step": 7500
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.13,
|
| 103 |
+
"learning_rate": 8.000000000000001e-06,
|
| 104 |
+
"loss": 9.2039,
|
| 105 |
+
"step": 8000
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.14,
|
| 109 |
+
"learning_rate": 8.5e-06,
|
| 110 |
+
"loss": 9.0723,
|
| 111 |
+
"step": 8500
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.15,
|
| 115 |
+
"learning_rate": 9e-06,
|
| 116 |
+
"loss": 8.9366,
|
| 117 |
+
"step": 9000
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.16,
|
| 121 |
+
"learning_rate": 9.5e-06,
|
| 122 |
+
"loss": 8.7964,
|
| 123 |
+
"step": 9500
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.16,
|
| 127 |
+
"learning_rate": 1e-05,
|
| 128 |
+
"loss": 8.6554,
|
| 129 |
+
"step": 10000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.16,
|
| 133 |
+
"eval_accuracy": 0.048290781267576265,
|
| 134 |
+
"eval_loss": 8.584630012512207,
|
| 135 |
+
"eval_runtime": 600.2506,
|
| 136 |
+
"eval_samples_per_second": 513.712,
|
| 137 |
+
"eval_steps_per_second": 5.353,
|
| 138 |
+
"step": 10000
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.17,
|
| 142 |
+
"learning_rate": 9.983059174304157e-06,
|
| 143 |
+
"loss": 8.5166,
|
| 144 |
+
"step": 10500
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.18,
|
| 148 |
+
"learning_rate": 9.966118348608312e-06,
|
| 149 |
+
"loss": 8.3802,
|
| 150 |
+
"step": 11000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.19,
|
| 154 |
+
"learning_rate": 9.949177522912468e-06,
|
| 155 |
+
"loss": 8.2563,
|
| 156 |
+
"step": 11500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.2,
|
| 160 |
+
"learning_rate": 9.932236697216624e-06,
|
| 161 |
+
"loss": 8.1368,
|
| 162 |
+
"step": 12000
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.2,
|
| 166 |
+
"learning_rate": 9.91529587152078e-06,
|
| 167 |
+
"loss": 8.0261,
|
| 168 |
+
"step": 12500
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.21,
|
| 172 |
+
"learning_rate": 9.898355045824935e-06,
|
| 173 |
+
"loss": 7.9253,
|
| 174 |
+
"step": 13000
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.22,
|
| 178 |
+
"learning_rate": 9.881414220129089e-06,
|
| 179 |
+
"loss": 7.8284,
|
| 180 |
+
"step": 13500
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.23,
|
| 184 |
+
"learning_rate": 9.864473394433246e-06,
|
| 185 |
+
"loss": 7.7407,
|
| 186 |
+
"step": 14000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.24,
|
| 190 |
+
"learning_rate": 9.847532568737402e-06,
|
| 191 |
+
"loss": 7.6611,
|
| 192 |
+
"step": 14500
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.25,
|
| 196 |
+
"learning_rate": 9.830591743041556e-06,
|
| 197 |
+
"loss": 7.59,
|
| 198 |
+
"step": 15000
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.25,
|
| 202 |
+
"learning_rate": 9.813650917345712e-06,
|
| 203 |
+
"loss": 7.5243,
|
| 204 |
+
"step": 15500
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.26,
|
| 208 |
+
"learning_rate": 9.796710091649867e-06,
|
| 209 |
+
"loss": 7.4708,
|
| 210 |
+
"step": 16000
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 0.27,
|
| 214 |
+
"learning_rate": 9.779769265954023e-06,
|
| 215 |
+
"loss": 7.4185,
|
| 216 |
+
"step": 16500
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.28,
|
| 220 |
+
"learning_rate": 9.762828440258179e-06,
|
| 221 |
+
"loss": 7.3726,
|
| 222 |
+
"step": 17000
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.29,
|
| 226 |
+
"learning_rate": 9.745887614562334e-06,
|
| 227 |
+
"loss": 7.3442,
|
| 228 |
+
"step": 17500
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.29,
|
| 232 |
+
"learning_rate": 9.72894678886649e-06,
|
| 233 |
+
"loss": 7.3171,
|
| 234 |
+
"step": 18000
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.3,
|
| 238 |
+
"learning_rate": 9.712005963170646e-06,
|
| 239 |
+
"loss": 7.2839,
|
| 240 |
+
"step": 18500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.31,
|
| 244 |
+
"learning_rate": 9.695065137474802e-06,
|
| 245 |
+
"loss": 7.2703,
|
| 246 |
+
"step": 19000
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.32,
|
| 250 |
+
"learning_rate": 9.678124311778957e-06,
|
| 251 |
+
"loss": 7.2537,
|
| 252 |
+
"step": 19500
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.33,
|
| 256 |
+
"learning_rate": 9.661183486083113e-06,
|
| 257 |
+
"loss": 7.2331,
|
| 258 |
+
"step": 20000
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.33,
|
| 262 |
+
"eval_accuracy": 0.05423351427871706,
|
| 263 |
+
"eval_loss": 7.227975845336914,
|
| 264 |
+
"eval_runtime": 605.972,
|
| 265 |
+
"eval_samples_per_second": 508.862,
|
| 266 |
+
"eval_steps_per_second": 5.302,
|
| 267 |
+
"step": 20000
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.34,
|
| 271 |
+
"learning_rate": 9.644242660387269e-06,
|
| 272 |
+
"loss": 7.2207,
|
| 273 |
+
"step": 20500
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.34,
|
| 277 |
+
"learning_rate": 9.627301834691424e-06,
|
| 278 |
+
"loss": 7.2044,
|
| 279 |
+
"step": 21000
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.35,
|
| 283 |
+
"learning_rate": 9.610361008995578e-06,
|
| 284 |
+
"loss": 7.1869,
|
| 285 |
+
"step": 21500
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.36,
|
| 289 |
+
"learning_rate": 9.593420183299736e-06,
|
| 290 |
+
"loss": 7.1736,
|
| 291 |
+
"step": 22000
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.37,
|
| 295 |
+
"learning_rate": 9.576479357603891e-06,
|
| 296 |
+
"loss": 7.1594,
|
| 297 |
+
"step": 22500
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.38,
|
| 301 |
+
"learning_rate": 9.559538531908045e-06,
|
| 302 |
+
"loss": 7.1396,
|
| 303 |
+
"step": 23000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.39,
|
| 307 |
+
"learning_rate": 9.542597706212201e-06,
|
| 308 |
+
"loss": 7.1306,
|
| 309 |
+
"step": 23500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.39,
|
| 313 |
+
"learning_rate": 9.525656880516357e-06,
|
| 314 |
+
"loss": 7.1185,
|
| 315 |
+
"step": 24000
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.4,
|
| 319 |
+
"learning_rate": 9.508716054820512e-06,
|
| 320 |
+
"loss": 7.1029,
|
| 321 |
+
"step": 24500
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.41,
|
| 325 |
+
"learning_rate": 9.491775229124668e-06,
|
| 326 |
+
"loss": 7.0927,
|
| 327 |
+
"step": 25000
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.42,
|
| 331 |
+
"learning_rate": 9.474834403428824e-06,
|
| 332 |
+
"loss": 7.0818,
|
| 333 |
+
"step": 25500
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.43,
|
| 337 |
+
"learning_rate": 9.45789357773298e-06,
|
| 338 |
+
"loss": 7.0696,
|
| 339 |
+
"step": 26000
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.43,
|
| 343 |
+
"learning_rate": 9.440952752037135e-06,
|
| 344 |
+
"loss": 7.0553,
|
| 345 |
+
"step": 26500
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.44,
|
| 349 |
+
"learning_rate": 9.424011926341291e-06,
|
| 350 |
+
"loss": 7.0496,
|
| 351 |
+
"step": 27000
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.45,
|
| 355 |
+
"learning_rate": 9.407071100645447e-06,
|
| 356 |
+
"loss": 7.038,
|
| 357 |
+
"step": 27500
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.46,
|
| 361 |
+
"learning_rate": 9.390130274949602e-06,
|
| 362 |
+
"loss": 7.0289,
|
| 363 |
+
"step": 28000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.47,
|
| 367 |
+
"learning_rate": 9.373189449253758e-06,
|
| 368 |
+
"loss": 7.0244,
|
| 369 |
+
"step": 28500
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.48,
|
| 373 |
+
"learning_rate": 9.356248623557914e-06,
|
| 374 |
+
"loss": 7.0145,
|
| 375 |
+
"step": 29000
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.48,
|
| 379 |
+
"learning_rate": 9.339307797862068e-06,
|
| 380 |
+
"loss": 7.0029,
|
| 381 |
+
"step": 29500
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.49,
|
| 385 |
+
"learning_rate": 9.322366972166225e-06,
|
| 386 |
+
"loss": 7.0014,
|
| 387 |
+
"step": 30000
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.49,
|
| 391 |
+
"eval_accuracy": 0.06774155914288765,
|
| 392 |
+
"eval_loss": 6.992653846740723,
|
| 393 |
+
"eval_runtime": 600.1356,
|
| 394 |
+
"eval_samples_per_second": 513.811,
|
| 395 |
+
"eval_steps_per_second": 5.354,
|
| 396 |
+
"step": 30000
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.5,
|
| 400 |
+
"learning_rate": 9.305426146470379e-06,
|
| 401 |
+
"loss": 6.9903,
|
| 402 |
+
"step": 30500
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.51,
|
| 406 |
+
"learning_rate": 9.288485320774535e-06,
|
| 407 |
+
"loss": 6.9807,
|
| 408 |
+
"step": 31000
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.52,
|
| 412 |
+
"learning_rate": 9.27154449507869e-06,
|
| 413 |
+
"loss": 6.9704,
|
| 414 |
+
"step": 31500
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.52,
|
| 418 |
+
"learning_rate": 9.254603669382846e-06,
|
| 419 |
+
"loss": 6.9613,
|
| 420 |
+
"step": 32000
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.53,
|
| 424 |
+
"learning_rate": 9.237662843687002e-06,
|
| 425 |
+
"loss": 6.9601,
|
| 426 |
+
"step": 32500
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.54,
|
| 430 |
+
"learning_rate": 9.220722017991158e-06,
|
| 431 |
+
"loss": 6.9596,
|
| 432 |
+
"step": 33000
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.55,
|
| 436 |
+
"learning_rate": 9.203781192295313e-06,
|
| 437 |
+
"loss": 6.9484,
|
| 438 |
+
"step": 33500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.56,
|
| 442 |
+
"learning_rate": 9.186840366599469e-06,
|
| 443 |
+
"loss": 6.9372,
|
| 444 |
+
"step": 34000
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.57,
|
| 448 |
+
"learning_rate": 9.169899540903625e-06,
|
| 449 |
+
"loss": 6.932,
|
| 450 |
+
"step": 34500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.57,
|
| 454 |
+
"learning_rate": 9.15295871520778e-06,
|
| 455 |
+
"loss": 6.925,
|
| 456 |
+
"step": 35000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.58,
|
| 460 |
+
"learning_rate": 9.136017889511936e-06,
|
| 461 |
+
"loss": 6.9252,
|
| 462 |
+
"step": 35500
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 0.59,
|
| 466 |
+
"learning_rate": 9.11907706381609e-06,
|
| 467 |
+
"loss": 6.9177,
|
| 468 |
+
"step": 36000
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.6,
|
| 472 |
+
"learning_rate": 9.102136238120247e-06,
|
| 473 |
+
"loss": 6.906,
|
| 474 |
+
"step": 36500
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.61,
|
| 478 |
+
"learning_rate": 9.085195412424403e-06,
|
| 479 |
+
"loss": 6.8972,
|
| 480 |
+
"step": 37000
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.61,
|
| 484 |
+
"learning_rate": 9.068254586728557e-06,
|
| 485 |
+
"loss": 6.8942,
|
| 486 |
+
"step": 37500
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.62,
|
| 490 |
+
"learning_rate": 9.051313761032714e-06,
|
| 491 |
+
"loss": 6.8912,
|
| 492 |
+
"step": 38000
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.63,
|
| 496 |
+
"learning_rate": 9.034372935336868e-06,
|
| 497 |
+
"loss": 6.8861,
|
| 498 |
+
"step": 38500
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.64,
|
| 502 |
+
"learning_rate": 9.017432109641024e-06,
|
| 503 |
+
"loss": 6.8749,
|
| 504 |
+
"step": 39000
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.65,
|
| 508 |
+
"learning_rate": 9.00049128394518e-06,
|
| 509 |
+
"loss": 6.8718,
|
| 510 |
+
"step": 39500
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 0.66,
|
| 514 |
+
"learning_rate": 8.983550458249335e-06,
|
| 515 |
+
"loss": 6.8699,
|
| 516 |
+
"step": 40000
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.66,
|
| 520 |
+
"eval_accuracy": 0.08556003293560886,
|
| 521 |
+
"eval_loss": 6.8636794090271,
|
| 522 |
+
"eval_runtime": 598.5444,
|
| 523 |
+
"eval_samples_per_second": 515.176,
|
| 524 |
+
"eval_steps_per_second": 5.368,
|
| 525 |
+
"step": 40000
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.66,
|
| 529 |
+
"learning_rate": 8.966609632553491e-06,
|
| 530 |
+
"loss": 6.8626,
|
| 531 |
+
"step": 40500
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.67,
|
| 535 |
+
"learning_rate": 8.949668806857647e-06,
|
| 536 |
+
"loss": 6.8559,
|
| 537 |
+
"step": 41000
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.68,
|
| 541 |
+
"learning_rate": 8.932727981161803e-06,
|
| 542 |
+
"loss": 6.8509,
|
| 543 |
+
"step": 41500
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.69,
|
| 547 |
+
"learning_rate": 8.915787155465958e-06,
|
| 548 |
+
"loss": 6.844,
|
| 549 |
+
"step": 42000
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.7,
|
| 553 |
+
"learning_rate": 8.898846329770114e-06,
|
| 554 |
+
"loss": 6.8383,
|
| 555 |
+
"step": 42500
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.7,
|
| 559 |
+
"learning_rate": 8.88190550407427e-06,
|
| 560 |
+
"loss": 6.8398,
|
| 561 |
+
"step": 43000
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.71,
|
| 565 |
+
"learning_rate": 8.864964678378425e-06,
|
| 566 |
+
"loss": 6.8233,
|
| 567 |
+
"step": 43500
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.72,
|
| 571 |
+
"learning_rate": 8.84802385268258e-06,
|
| 572 |
+
"loss": 6.8242,
|
| 573 |
+
"step": 44000
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.73,
|
| 577 |
+
"learning_rate": 8.831083026986737e-06,
|
| 578 |
+
"loss": 6.82,
|
| 579 |
+
"step": 44500
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.74,
|
| 583 |
+
"learning_rate": 8.814142201290892e-06,
|
| 584 |
+
"loss": 6.8142,
|
| 585 |
+
"step": 45000
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.75,
|
| 589 |
+
"learning_rate": 8.797201375595046e-06,
|
| 590 |
+
"loss": 6.8122,
|
| 591 |
+
"step": 45500
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.75,
|
| 595 |
+
"learning_rate": 8.780260549899204e-06,
|
| 596 |
+
"loss": 6.8017,
|
| 597 |
+
"step": 46000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.76,
|
| 601 |
+
"learning_rate": 8.763319724203358e-06,
|
| 602 |
+
"loss": 6.8021,
|
| 603 |
+
"step": 46500
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.77,
|
| 607 |
+
"learning_rate": 8.746378898507513e-06,
|
| 608 |
+
"loss": 6.8006,
|
| 609 |
+
"step": 47000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.78,
|
| 613 |
+
"learning_rate": 8.729438072811671e-06,
|
| 614 |
+
"loss": 6.7897,
|
| 615 |
+
"step": 47500
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.79,
|
| 619 |
+
"learning_rate": 8.712497247115825e-06,
|
| 620 |
+
"loss": 6.7932,
|
| 621 |
+
"step": 48000
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.79,
|
| 625 |
+
"learning_rate": 8.69555642141998e-06,
|
| 626 |
+
"loss": 6.7881,
|
| 627 |
+
"step": 48500
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.8,
|
| 631 |
+
"learning_rate": 8.678615595724136e-06,
|
| 632 |
+
"loss": 6.7793,
|
| 633 |
+
"step": 49000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.81,
|
| 637 |
+
"learning_rate": 8.661674770028292e-06,
|
| 638 |
+
"loss": 6.7823,
|
| 639 |
+
"step": 49500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.82,
|
| 643 |
+
"learning_rate": 8.644733944332448e-06,
|
| 644 |
+
"loss": 6.7777,
|
| 645 |
+
"step": 50000
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.82,
|
| 649 |
+
"eval_accuracy": 0.09222163830820224,
|
| 650 |
+
"eval_loss": 6.772622108459473,
|
| 651 |
+
"eval_runtime": 600.0695,
|
| 652 |
+
"eval_samples_per_second": 513.867,
|
| 653 |
+
"eval_steps_per_second": 5.354,
|
| 654 |
+
"step": 50000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.83,
|
| 658 |
+
"learning_rate": 8.627793118636603e-06,
|
| 659 |
+
"loss": 6.7669,
|
| 660 |
+
"step": 50500
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.84,
|
| 664 |
+
"learning_rate": 8.610852292940759e-06,
|
| 665 |
+
"loss": 6.7646,
|
| 666 |
+
"step": 51000
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.84,
|
| 670 |
+
"learning_rate": 8.593911467244915e-06,
|
| 671 |
+
"loss": 6.7643,
|
| 672 |
+
"step": 51500
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.85,
|
| 676 |
+
"learning_rate": 8.576970641549069e-06,
|
| 677 |
+
"loss": 6.7579,
|
| 678 |
+
"step": 52000
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.86,
|
| 682 |
+
"learning_rate": 8.560029815853226e-06,
|
| 683 |
+
"loss": 6.753,
|
| 684 |
+
"step": 52500
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.87,
|
| 688 |
+
"learning_rate": 8.543088990157382e-06,
|
| 689 |
+
"loss": 6.7522,
|
| 690 |
+
"step": 53000
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.88,
|
| 694 |
+
"learning_rate": 8.526148164461536e-06,
|
| 695 |
+
"loss": 6.7583,
|
| 696 |
+
"step": 53500
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.88,
|
| 700 |
+
"learning_rate": 8.509207338765693e-06,
|
| 701 |
+
"loss": 6.747,
|
| 702 |
+
"step": 54000
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.89,
|
| 706 |
+
"learning_rate": 8.492266513069847e-06,
|
| 707 |
+
"loss": 6.7433,
|
| 708 |
+
"step": 54500
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 0.9,
|
| 712 |
+
"learning_rate": 8.475325687374003e-06,
|
| 713 |
+
"loss": 6.7394,
|
| 714 |
+
"step": 55000
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.91,
|
| 718 |
+
"learning_rate": 8.45838486167816e-06,
|
| 719 |
+
"loss": 6.7346,
|
| 720 |
+
"step": 55500
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 0.92,
|
| 724 |
+
"learning_rate": 8.441444035982314e-06,
|
| 725 |
+
"loss": 6.7408,
|
| 726 |
+
"step": 56000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 0.93,
|
| 730 |
+
"learning_rate": 8.42450321028647e-06,
|
| 731 |
+
"loss": 6.7341,
|
| 732 |
+
"step": 56500
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.93,
|
| 736 |
+
"learning_rate": 8.407562384590626e-06,
|
| 737 |
+
"loss": 6.7339,
|
| 738 |
+
"step": 57000
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.94,
|
| 742 |
+
"learning_rate": 8.390621558894781e-06,
|
| 743 |
+
"loss": 6.7284,
|
| 744 |
+
"step": 57500
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.95,
|
| 748 |
+
"learning_rate": 8.373680733198937e-06,
|
| 749 |
+
"loss": 6.732,
|
| 750 |
+
"step": 58000
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 0.96,
|
| 754 |
+
"learning_rate": 8.356739907503093e-06,
|
| 755 |
+
"loss": 6.7212,
|
| 756 |
+
"step": 58500
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 0.97,
|
| 760 |
+
"learning_rate": 8.339799081807248e-06,
|
| 761 |
+
"loss": 6.7156,
|
| 762 |
+
"step": 59000
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.97,
|
| 766 |
+
"learning_rate": 8.322858256111404e-06,
|
| 767 |
+
"loss": 6.7182,
|
| 768 |
+
"step": 59500
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.98,
|
| 772 |
+
"learning_rate": 8.305917430415558e-06,
|
| 773 |
+
"loss": 6.7091,
|
| 774 |
+
"step": 60000
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.98,
|
| 778 |
+
"eval_accuracy": 0.09736718543824835,
|
| 779 |
+
"eval_loss": 6.7100605964660645,
|
| 780 |
+
"eval_runtime": 601.1731,
|
| 781 |
+
"eval_samples_per_second": 512.924,
|
| 782 |
+
"eval_steps_per_second": 5.345,
|
| 783 |
+
"step": 60000
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.99,
|
| 787 |
+
"learning_rate": 8.288976604719715e-06,
|
| 788 |
+
"loss": 6.7102,
|
| 789 |
+
"step": 60500
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 1.0,
|
| 793 |
+
"learning_rate": 8.272035779023871e-06,
|
| 794 |
+
"loss": 6.7105,
|
| 795 |
+
"step": 61000
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.01,
|
| 799 |
+
"learning_rate": 8.255094953328025e-06,
|
| 800 |
+
"loss": 6.7048,
|
| 801 |
+
"step": 61500
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.02,
|
| 805 |
+
"learning_rate": 8.238154127632183e-06,
|
| 806 |
+
"loss": 6.7056,
|
| 807 |
+
"step": 62000
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.02,
|
| 811 |
+
"learning_rate": 8.221213301936337e-06,
|
| 812 |
+
"loss": 6.6984,
|
| 813 |
+
"step": 62500
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.03,
|
| 817 |
+
"learning_rate": 8.204272476240492e-06,
|
| 818 |
+
"loss": 6.6931,
|
| 819 |
+
"step": 63000
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 1.04,
|
| 823 |
+
"learning_rate": 8.18733165054465e-06,
|
| 824 |
+
"loss": 6.6938,
|
| 825 |
+
"step": 63500
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.05,
|
| 829 |
+
"learning_rate": 8.170390824848804e-06,
|
| 830 |
+
"loss": 6.6919,
|
| 831 |
+
"step": 64000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.06,
|
| 835 |
+
"learning_rate": 8.15344999915296e-06,
|
| 836 |
+
"loss": 6.6935,
|
| 837 |
+
"step": 64500
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 1.07,
|
| 841 |
+
"learning_rate": 8.136509173457115e-06,
|
| 842 |
+
"loss": 6.6856,
|
| 843 |
+
"step": 65000
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1.07,
|
| 847 |
+
"learning_rate": 8.11956834776127e-06,
|
| 848 |
+
"loss": 6.6865,
|
| 849 |
+
"step": 65500
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.08,
|
| 853 |
+
"learning_rate": 8.102627522065426e-06,
|
| 854 |
+
"loss": 6.6818,
|
| 855 |
+
"step": 66000
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1.09,
|
| 859 |
+
"learning_rate": 8.085686696369582e-06,
|
| 860 |
+
"loss": 6.6852,
|
| 861 |
+
"step": 66500
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 1.1,
|
| 865 |
+
"learning_rate": 8.068745870673738e-06,
|
| 866 |
+
"loss": 6.6818,
|
| 867 |
+
"step": 67000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.11,
|
| 871 |
+
"learning_rate": 8.051805044977893e-06,
|
| 872 |
+
"loss": 6.6742,
|
| 873 |
+
"step": 67500
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.11,
|
| 877 |
+
"learning_rate": 8.034864219282047e-06,
|
| 878 |
+
"loss": 6.6729,
|
| 879 |
+
"step": 68000
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.12,
|
| 883 |
+
"learning_rate": 8.017923393586205e-06,
|
| 884 |
+
"loss": 6.6691,
|
| 885 |
+
"step": 68500
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1.13,
|
| 889 |
+
"learning_rate": 8.00098256789036e-06,
|
| 890 |
+
"loss": 6.6618,
|
| 891 |
+
"step": 69000
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.14,
|
| 895 |
+
"learning_rate": 7.984041742194514e-06,
|
| 896 |
+
"loss": 6.6685,
|
| 897 |
+
"step": 69500
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.15,
|
| 901 |
+
"learning_rate": 7.967100916498672e-06,
|
| 902 |
+
"loss": 6.6626,
|
| 903 |
+
"step": 70000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 1.15,
|
| 907 |
+
"eval_accuracy": 0.1015238380889028,
|
| 908 |
+
"eval_loss": 6.661967754364014,
|
| 909 |
+
"eval_runtime": 605.0085,
|
| 910 |
+
"eval_samples_per_second": 509.672,
|
| 911 |
+
"eval_steps_per_second": 5.311,
|
| 912 |
+
"step": 70000
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.16,
|
| 916 |
+
"learning_rate": 7.950160090802826e-06,
|
| 917 |
+
"loss": 6.6613,
|
| 918 |
+
"step": 70500
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.16,
|
| 922 |
+
"learning_rate": 7.933219265106982e-06,
|
| 923 |
+
"loss": 6.6616,
|
| 924 |
+
"step": 71000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 1.17,
|
| 928 |
+
"learning_rate": 7.916278439411137e-06,
|
| 929 |
+
"loss": 6.6625,
|
| 930 |
+
"step": 71500
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.18,
|
| 934 |
+
"learning_rate": 7.899337613715293e-06,
|
| 935 |
+
"loss": 6.6572,
|
| 936 |
+
"step": 72000
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 1.19,
|
| 940 |
+
"learning_rate": 7.882396788019449e-06,
|
| 941 |
+
"loss": 6.652,
|
| 942 |
+
"step": 72500
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 1.2,
|
| 946 |
+
"learning_rate": 7.865455962323604e-06,
|
| 947 |
+
"loss": 6.6498,
|
| 948 |
+
"step": 73000
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.2,
|
| 952 |
+
"learning_rate": 7.84851513662776e-06,
|
| 953 |
+
"loss": 6.6477,
|
| 954 |
+
"step": 73500
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.21,
|
| 958 |
+
"learning_rate": 7.831574310931916e-06,
|
| 959 |
+
"loss": 6.6456,
|
| 960 |
+
"step": 74000
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.22,
|
| 964 |
+
"learning_rate": 7.814633485236071e-06,
|
| 965 |
+
"loss": 6.6418,
|
| 966 |
+
"step": 74500
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 1.23,
|
| 970 |
+
"learning_rate": 7.797692659540227e-06,
|
| 971 |
+
"loss": 6.6447,
|
| 972 |
+
"step": 75000
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 1.24,
|
| 976 |
+
"learning_rate": 7.780751833844383e-06,
|
| 977 |
+
"loss": 6.6428,
|
| 978 |
+
"step": 75500
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.25,
|
| 982 |
+
"learning_rate": 7.763811008148537e-06,
|
| 983 |
+
"loss": 6.635,
|
| 984 |
+
"step": 76000
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 1.25,
|
| 988 |
+
"learning_rate": 7.746870182452694e-06,
|
| 989 |
+
"loss": 6.6335,
|
| 990 |
+
"step": 76500
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.26,
|
| 994 |
+
"learning_rate": 7.729929356756848e-06,
|
| 995 |
+
"loss": 6.637,
|
| 996 |
+
"step": 77000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.27,
|
| 1000 |
+
"learning_rate": 7.712988531061004e-06,
|
| 1001 |
+
"loss": 6.6324,
|
| 1002 |
+
"step": 77500
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.28,
|
| 1006 |
+
"learning_rate": 7.696047705365161e-06,
|
| 1007 |
+
"loss": 6.6344,
|
| 1008 |
+
"step": 78000
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 1.29,
|
| 1012 |
+
"learning_rate": 7.679106879669315e-06,
|
| 1013 |
+
"loss": 6.629,
|
| 1014 |
+
"step": 78500
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.29,
|
| 1018 |
+
"learning_rate": 7.662166053973471e-06,
|
| 1019 |
+
"loss": 6.6266,
|
| 1020 |
+
"step": 79000
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.3,
|
| 1024 |
+
"learning_rate": 7.645225228277627e-06,
|
| 1025 |
+
"loss": 6.6295,
|
| 1026 |
+
"step": 79500
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.31,
|
| 1030 |
+
"learning_rate": 7.628284402581782e-06,
|
| 1031 |
+
"loss": 6.6279,
|
| 1032 |
+
"step": 80000
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 1.31,
|
| 1036 |
+
"eval_accuracy": 0.10400515267929009,
|
| 1037 |
+
"eval_loss": 6.625459671020508,
|
| 1038 |
+
"eval_runtime": 603.1684,
|
| 1039 |
+
"eval_samples_per_second": 511.227,
|
| 1040 |
+
"eval_steps_per_second": 5.327,
|
| 1041 |
+
"step": 80000
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 1.32,
|
| 1045 |
+
"learning_rate": 7.611343576885938e-06,
|
| 1046 |
+
"loss": 6.6241,
|
| 1047 |
+
"step": 80500
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 1.33,
|
| 1051 |
+
"learning_rate": 7.5944027511900945e-06,
|
| 1052 |
+
"loss": 6.6207,
|
| 1053 |
+
"step": 81000
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 1.34,
|
| 1057 |
+
"learning_rate": 7.577461925494249e-06,
|
| 1058 |
+
"loss": 6.6216,
|
| 1059 |
+
"step": 81500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.34,
|
| 1063 |
+
"learning_rate": 7.560521099798405e-06,
|
| 1064 |
+
"loss": 6.619,
|
| 1065 |
+
"step": 82000
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.35,
|
| 1069 |
+
"learning_rate": 7.54358027410256e-06,
|
| 1070 |
+
"loss": 6.6153,
|
| 1071 |
+
"step": 82500
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.36,
|
| 1075 |
+
"learning_rate": 7.5266394484067164e-06,
|
| 1076 |
+
"loss": 6.6181,
|
| 1077 |
+
"step": 83000
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 1.37,
|
| 1081 |
+
"learning_rate": 7.509698622710871e-06,
|
| 1082 |
+
"loss": 6.6137,
|
| 1083 |
+
"step": 83500
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 1.38,
|
| 1087 |
+
"learning_rate": 7.492757797015027e-06,
|
| 1088 |
+
"loss": 6.6157,
|
| 1089 |
+
"step": 84000
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 1.38,
|
| 1093 |
+
"learning_rate": 7.475816971319183e-06,
|
| 1094 |
+
"loss": 6.6114,
|
| 1095 |
+
"step": 84500
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 1.39,
|
| 1099 |
+
"learning_rate": 7.458876145623338e-06,
|
| 1100 |
+
"loss": 6.6068,
|
| 1101 |
+
"step": 85000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.4,
|
| 1105 |
+
"learning_rate": 7.441935319927493e-06,
|
| 1106 |
+
"loss": 6.6081,
|
| 1107 |
+
"step": 85500
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.41,
|
| 1111 |
+
"learning_rate": 7.42499449423165e-06,
|
| 1112 |
+
"loss": 6.6026,
|
| 1113 |
+
"step": 86000
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 1.42,
|
| 1117 |
+
"learning_rate": 7.4080536685358054e-06,
|
| 1118 |
+
"loss": 6.6031,
|
| 1119 |
+
"step": 86500
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.43,
|
| 1123 |
+
"learning_rate": 7.39111284283996e-06,
|
| 1124 |
+
"loss": 6.6086,
|
| 1125 |
+
"step": 87000
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 1.43,
|
| 1129 |
+
"learning_rate": 7.374172017144117e-06,
|
| 1130 |
+
"loss": 6.6048,
|
| 1131 |
+
"step": 87500
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 1.44,
|
| 1135 |
+
"learning_rate": 7.357231191448272e-06,
|
| 1136 |
+
"loss": 6.6031,
|
| 1137 |
+
"step": 88000
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 1.45,
|
| 1141 |
+
"learning_rate": 7.340290365752427e-06,
|
| 1142 |
+
"loss": 6.5961,
|
| 1143 |
+
"step": 88500
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.46,
|
| 1147 |
+
"learning_rate": 7.323349540056584e-06,
|
| 1148 |
+
"loss": 6.5977,
|
| 1149 |
+
"step": 89000
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.47,
|
| 1153 |
+
"learning_rate": 7.306408714360739e-06,
|
| 1154 |
+
"loss": 6.598,
|
| 1155 |
+
"step": 89500
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 1.47,
|
| 1159 |
+
"learning_rate": 7.2894678886648936e-06,
|
| 1160 |
+
"loss": 6.5917,
|
| 1161 |
+
"step": 90000
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 1.47,
|
| 1165 |
+
"eval_accuracy": 0.10684875378443538,
|
| 1166 |
+
"eval_loss": 6.594837188720703,
|
| 1167 |
+
"eval_runtime": 601.7144,
|
| 1168 |
+
"eval_samples_per_second": 512.462,
|
| 1169 |
+
"eval_steps_per_second": 5.34,
|
| 1170 |
+
"step": 90000
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.48,
|
| 1174 |
+
"learning_rate": 7.272527062969049e-06,
|
| 1175 |
+
"loss": 6.5912,
|
| 1176 |
+
"step": 90500
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 1.49,
|
| 1180 |
+
"learning_rate": 7.255586237273206e-06,
|
| 1181 |
+
"loss": 6.5903,
|
| 1182 |
+
"step": 91000
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 1.5,
|
| 1186 |
+
"learning_rate": 7.238645411577361e-06,
|
| 1187 |
+
"loss": 6.5949,
|
| 1188 |
+
"step": 91500
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 1.51,
|
| 1192 |
+
"learning_rate": 7.221704585881516e-06,
|
| 1193 |
+
"loss": 6.5932,
|
| 1194 |
+
"step": 92000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.52,
|
| 1198 |
+
"learning_rate": 7.204763760185672e-06,
|
| 1199 |
+
"loss": 6.5887,
|
| 1200 |
+
"step": 92500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.52,
|
| 1204 |
+
"learning_rate": 7.187822934489828e-06,
|
| 1205 |
+
"loss": 6.5879,
|
| 1206 |
+
"step": 93000
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.53,
|
| 1210 |
+
"learning_rate": 7.1708821087939826e-06,
|
| 1211 |
+
"loss": 6.5892,
|
| 1212 |
+
"step": 93500
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 1.54,
|
| 1216 |
+
"learning_rate": 7.153941283098139e-06,
|
| 1217 |
+
"loss": 6.5873,
|
| 1218 |
+
"step": 94000
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 1.55,
|
| 1222 |
+
"learning_rate": 7.137000457402295e-06,
|
| 1223 |
+
"loss": 6.5908,
|
| 1224 |
+
"step": 94500
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.56,
|
| 1228 |
+
"learning_rate": 7.12005963170645e-06,
|
| 1229 |
+
"loss": 6.5839,
|
| 1230 |
+
"step": 95000
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 1.56,
|
| 1234 |
+
"learning_rate": 7.103118806010606e-06,
|
| 1235 |
+
"loss": 6.5803,
|
| 1236 |
+
"step": 95500
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 1.57,
|
| 1240 |
+
"learning_rate": 7.086177980314761e-06,
|
| 1241 |
+
"loss": 6.5832,
|
| 1242 |
+
"step": 96000
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.58,
|
| 1246 |
+
"learning_rate": 7.069237154618917e-06,
|
| 1247 |
+
"loss": 6.5801,
|
| 1248 |
+
"step": 96500
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.59,
|
| 1252 |
+
"learning_rate": 7.052296328923072e-06,
|
| 1253 |
+
"loss": 6.5743,
|
| 1254 |
+
"step": 97000
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.6,
|
| 1258 |
+
"learning_rate": 7.035355503227228e-06,
|
| 1259 |
+
"loss": 6.576,
|
| 1260 |
+
"step": 97500
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 1.61,
|
| 1264 |
+
"learning_rate": 7.018414677531383e-06,
|
| 1265 |
+
"loss": 6.5739,
|
| 1266 |
+
"step": 98000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 1.61,
|
| 1270 |
+
"learning_rate": 7.001473851835539e-06,
|
| 1271 |
+
"loss": 6.5728,
|
| 1272 |
+
"step": 98500
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 1.62,
|
| 1276 |
+
"learning_rate": 6.984533026139695e-06,
|
| 1277 |
+
"loss": 6.5703,
|
| 1278 |
+
"step": 99000
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 1.63,
|
| 1282 |
+
"learning_rate": 6.96759220044385e-06,
|
| 1283 |
+
"loss": 6.5732,
|
| 1284 |
+
"step": 99500
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.64,
|
| 1288 |
+
"learning_rate": 6.950651374748006e-06,
|
| 1289 |
+
"loss": 6.5691,
|
| 1290 |
+
"step": 100000
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.64,
|
| 1294 |
+
"eval_accuracy": 0.10944561377176425,
|
| 1295 |
+
"eval_loss": 6.569516181945801,
|
| 1296 |
+
"eval_runtime": 601.8291,
|
| 1297 |
+
"eval_samples_per_second": 512.365,
|
| 1298 |
+
"eval_steps_per_second": 5.339,
|
| 1299 |
+
"step": 100000
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.65,
|
| 1303 |
+
"learning_rate": 6.933710549052161e-06,
|
| 1304 |
+
"loss": 6.5714,
|
| 1305 |
+
"step": 100500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.65,
|
| 1309 |
+
"learning_rate": 6.916769723356317e-06,
|
| 1310 |
+
"loss": 6.5635,
|
| 1311 |
+
"step": 101000
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.66,
|
| 1315 |
+
"learning_rate": 6.899828897660472e-06,
|
| 1316 |
+
"loss": 6.572,
|
| 1317 |
+
"step": 101500
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.67,
|
| 1321 |
+
"learning_rate": 6.8828880719646285e-06,
|
| 1322 |
+
"loss": 6.5643,
|
| 1323 |
+
"step": 102000
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 1.68,
|
| 1327 |
+
"learning_rate": 6.865947246268783e-06,
|
| 1328 |
+
"loss": 6.5663,
|
| 1329 |
+
"step": 102500
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.69,
|
| 1333 |
+
"learning_rate": 6.849006420572939e-06,
|
| 1334 |
+
"loss": 6.5591,
|
| 1335 |
+
"step": 103000
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 1.7,
|
| 1339 |
+
"learning_rate": 6.8320655948770955e-06,
|
| 1340 |
+
"loss": 6.5602,
|
| 1341 |
+
"step": 103500
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.7,
|
| 1345 |
+
"learning_rate": 6.81512476918125e-06,
|
| 1346 |
+
"loss": 6.5603,
|
| 1347 |
+
"step": 104000
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.71,
|
| 1351 |
+
"learning_rate": 6.798183943485406e-06,
|
| 1352 |
+
"loss": 6.5588,
|
| 1353 |
+
"step": 104500
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.72,
|
| 1357 |
+
"learning_rate": 6.781243117789562e-06,
|
| 1358 |
+
"loss": 6.5568,
|
| 1359 |
+
"step": 105000
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.73,
|
| 1363 |
+
"learning_rate": 6.7643022920937174e-06,
|
| 1364 |
+
"loss": 6.5571,
|
| 1365 |
+
"step": 105500
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 1.74,
|
| 1369 |
+
"learning_rate": 6.747361466397872e-06,
|
| 1370 |
+
"loss": 6.5582,
|
| 1371 |
+
"step": 106000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 1.75,
|
| 1375 |
+
"learning_rate": 6.730420640702028e-06,
|
| 1376 |
+
"loss": 6.5519,
|
| 1377 |
+
"step": 106500
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.75,
|
| 1381 |
+
"learning_rate": 6.7134798150061845e-06,
|
| 1382 |
+
"loss": 6.5571,
|
| 1383 |
+
"step": 107000
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.76,
|
| 1387 |
+
"learning_rate": 6.696538989310339e-06,
|
| 1388 |
+
"loss": 6.5534,
|
| 1389 |
+
"step": 107500
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.77,
|
| 1393 |
+
"learning_rate": 6.679598163614495e-06,
|
| 1394 |
+
"loss": 6.5516,
|
| 1395 |
+
"step": 108000
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 1.78,
|
| 1399 |
+
"learning_rate": 6.662657337918651e-06,
|
| 1400 |
+
"loss": 6.5497,
|
| 1401 |
+
"step": 108500
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.79,
|
| 1405 |
+
"learning_rate": 6.6457165122228064e-06,
|
| 1406 |
+
"loss": 6.5465,
|
| 1407 |
+
"step": 109000
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 1.79,
|
| 1411 |
+
"learning_rate": 6.628775686526961e-06,
|
| 1412 |
+
"loss": 6.5517,
|
| 1413 |
+
"step": 109500
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.8,
|
| 1417 |
+
"learning_rate": 6.611834860831118e-06,
|
| 1418 |
+
"loss": 6.5486,
|
| 1419 |
+
"step": 110000
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 1.8,
|
| 1423 |
+
"eval_accuracy": 0.11219145006851503,
|
| 1424 |
+
"eval_loss": 6.546039581298828,
|
| 1425 |
+
"eval_runtime": 601.6578,
|
| 1426 |
+
"eval_samples_per_second": 512.511,
|
| 1427 |
+
"eval_steps_per_second": 5.34,
|
| 1428 |
+
"step": 110000
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 1.81,
|
| 1432 |
+
"learning_rate": 6.594894035135273e-06,
|
| 1433 |
+
"loss": 6.5533,
|
| 1434 |
+
"step": 110500
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 1.82,
|
| 1438 |
+
"learning_rate": 6.577953209439428e-06,
|
| 1439 |
+
"loss": 6.5467,
|
| 1440 |
+
"step": 111000
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 1.83,
|
| 1444 |
+
"learning_rate": 6.561012383743585e-06,
|
| 1445 |
+
"loss": 6.5454,
|
| 1446 |
+
"step": 111500
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 1.84,
|
| 1450 |
+
"learning_rate": 6.54407155804774e-06,
|
| 1451 |
+
"loss": 6.5416,
|
| 1452 |
+
"step": 112000
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 1.84,
|
| 1456 |
+
"learning_rate": 6.5271307323518954e-06,
|
| 1457 |
+
"loss": 6.5416,
|
| 1458 |
+
"step": 112500
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 1.85,
|
| 1462 |
+
"learning_rate": 6.510189906656051e-06,
|
| 1463 |
+
"loss": 6.543,
|
| 1464 |
+
"step": 113000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.86,
|
| 1468 |
+
"learning_rate": 6.493249080960207e-06,
|
| 1469 |
+
"loss": 6.5443,
|
| 1470 |
+
"step": 113500
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 1.87,
|
| 1474 |
+
"learning_rate": 6.476308255264362e-06,
|
| 1475 |
+
"loss": 6.534,
|
| 1476 |
+
"step": 114000
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 1.88,
|
| 1480 |
+
"learning_rate": 6.459367429568517e-06,
|
| 1481 |
+
"loss": 6.5395,
|
| 1482 |
+
"step": 114500
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 1.88,
|
| 1486 |
+
"learning_rate": 6.442426603872674e-06,
|
| 1487 |
+
"loss": 6.5386,
|
| 1488 |
+
"step": 115000
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 1.89,
|
| 1492 |
+
"learning_rate": 6.425485778176829e-06,
|
| 1493 |
+
"loss": 6.5399,
|
| 1494 |
+
"step": 115500
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 1.9,
|
| 1498 |
+
"learning_rate": 6.4085449524809836e-06,
|
| 1499 |
+
"loss": 6.5382,
|
| 1500 |
+
"step": 116000
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 1.91,
|
| 1504 |
+
"learning_rate": 6.39160412678514e-06,
|
| 1505 |
+
"loss": 6.5309,
|
| 1506 |
+
"step": 116500
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.92,
|
| 1510 |
+
"learning_rate": 6.374663301089296e-06,
|
| 1511 |
+
"loss": 6.533,
|
| 1512 |
+
"step": 117000
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 1.93,
|
| 1516 |
+
"learning_rate": 6.357722475393451e-06,
|
| 1517 |
+
"loss": 6.5269,
|
| 1518 |
+
"step": 117500
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 1.93,
|
| 1522 |
+
"learning_rate": 6.340781649697607e-06,
|
| 1523 |
+
"loss": 6.533,
|
| 1524 |
+
"step": 118000
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 1.94,
|
| 1528 |
+
"learning_rate": 6.323840824001762e-06,
|
| 1529 |
+
"loss": 6.5268,
|
| 1530 |
+
"step": 118500
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 1.95,
|
| 1534 |
+
"learning_rate": 6.306899998305918e-06,
|
| 1535 |
+
"loss": 6.5326,
|
| 1536 |
+
"step": 119000
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 1.96,
|
| 1540 |
+
"learning_rate": 6.289959172610074e-06,
|
| 1541 |
+
"loss": 6.5306,
|
| 1542 |
+
"step": 119500
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 1.97,
|
| 1546 |
+
"learning_rate": 6.273018346914229e-06,
|
| 1547 |
+
"loss": 6.5246,
|
| 1548 |
+
"step": 120000
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.97,
|
| 1552 |
+
"eval_accuracy": 0.11442907817610445,
|
| 1553 |
+
"eval_loss": 6.527458667755127,
|
| 1554 |
+
"eval_runtime": 601.5697,
|
| 1555 |
+
"eval_samples_per_second": 512.586,
|
| 1556 |
+
"eval_steps_per_second": 5.341,
|
| 1557 |
+
"step": 120000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 1.97,
|
| 1561 |
+
"learning_rate": 6.256077521218385e-06,
|
| 1562 |
+
"loss": 6.5326,
|
| 1563 |
+
"step": 120500
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 1.98,
|
| 1567 |
+
"learning_rate": 6.2391366955225405e-06,
|
| 1568 |
+
"loss": 6.535,
|
| 1569 |
+
"step": 121000
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.99,
|
| 1573 |
+
"learning_rate": 6.222195869826696e-06,
|
| 1574 |
+
"loss": 6.5223,
|
| 1575 |
+
"step": 121500
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 2.0,
|
| 1579 |
+
"learning_rate": 6.205255044130851e-06,
|
| 1580 |
+
"loss": 6.5225,
|
| 1581 |
+
"step": 122000
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 2.01,
|
| 1585 |
+
"learning_rate": 6.1883142184350076e-06,
|
| 1586 |
+
"loss": 6.518,
|
| 1587 |
+
"step": 122500
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 2.02,
|
| 1591 |
+
"learning_rate": 6.171373392739162e-06,
|
| 1592 |
+
"loss": 6.5249,
|
| 1593 |
+
"step": 123000
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 2.02,
|
| 1597 |
+
"learning_rate": 6.154432567043318e-06,
|
| 1598 |
+
"loss": 6.5216,
|
| 1599 |
+
"step": 123500
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 2.03,
|
| 1603 |
+
"learning_rate": 6.137491741347473e-06,
|
| 1604 |
+
"loss": 6.5238,
|
| 1605 |
+
"step": 124000
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 2.04,
|
| 1609 |
+
"learning_rate": 6.1205509156516295e-06,
|
| 1610 |
+
"loss": 6.5162,
|
| 1611 |
+
"step": 124500
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 2.05,
|
| 1615 |
+
"learning_rate": 6.103610089955785e-06,
|
| 1616 |
+
"loss": 6.5211,
|
| 1617 |
+
"step": 125000
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 2.06,
|
| 1621 |
+
"learning_rate": 6.08666926425994e-06,
|
| 1622 |
+
"loss": 6.5196,
|
| 1623 |
+
"step": 125500
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 2.06,
|
| 1627 |
+
"learning_rate": 6.0697284385640965e-06,
|
| 1628 |
+
"loss": 6.5158,
|
| 1629 |
+
"step": 126000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 2.07,
|
| 1633 |
+
"learning_rate": 6.052787612868251e-06,
|
| 1634 |
+
"loss": 6.5173,
|
| 1635 |
+
"step": 126500
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 2.08,
|
| 1639 |
+
"learning_rate": 6.035846787172407e-06,
|
| 1640 |
+
"loss": 6.5195,
|
| 1641 |
+
"step": 127000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 2.09,
|
| 1645 |
+
"learning_rate": 6.018905961476564e-06,
|
| 1646 |
+
"loss": 6.525,
|
| 1647 |
+
"step": 127500
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 2.1,
|
| 1651 |
+
"learning_rate": 6.0019651357807185e-06,
|
| 1652 |
+
"loss": 6.5169,
|
| 1653 |
+
"step": 128000
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 2.11,
|
| 1657 |
+
"learning_rate": 5.985024310084873e-06,
|
| 1658 |
+
"loss": 6.5191,
|
| 1659 |
+
"step": 128500
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 2.11,
|
| 1663 |
+
"learning_rate": 5.96808348438903e-06,
|
| 1664 |
+
"loss": 6.5111,
|
| 1665 |
+
"step": 129000
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 2.12,
|
| 1669 |
+
"learning_rate": 5.9511426586931855e-06,
|
| 1670 |
+
"loss": 6.5128,
|
| 1671 |
+
"step": 129500
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 2.13,
|
| 1675 |
+
"learning_rate": 5.93420183299734e-06,
|
| 1676 |
+
"loss": 6.5069,
|
| 1677 |
+
"step": 130000
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 2.13,
|
| 1681 |
+
"eval_accuracy": 0.11622728049791442,
|
| 1682 |
+
"eval_loss": 6.511504650115967,
|
| 1683 |
+
"eval_runtime": 605.5053,
|
| 1684 |
+
"eval_samples_per_second": 509.254,
|
| 1685 |
+
"eval_steps_per_second": 5.306,
|
| 1686 |
+
"step": 130000
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 2.14,
|
| 1690 |
+
"learning_rate": 5.917261007301497e-06,
|
| 1691 |
+
"loss": 6.5104,
|
| 1692 |
+
"step": 130500
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 2.15,
|
| 1696 |
+
"learning_rate": 5.900320181605652e-06,
|
| 1697 |
+
"loss": 6.5106,
|
| 1698 |
+
"step": 131000
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 2.15,
|
| 1702 |
+
"learning_rate": 5.8833793559098074e-06,
|
| 1703 |
+
"loss": 6.5054,
|
| 1704 |
+
"step": 131500
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 2.16,
|
| 1708 |
+
"learning_rate": 5.866438530213962e-06,
|
| 1709 |
+
"loss": 6.5124,
|
| 1710 |
+
"step": 132000
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 2.17,
|
| 1714 |
+
"learning_rate": 5.849497704518119e-06,
|
| 1715 |
+
"loss": 6.508,
|
| 1716 |
+
"step": 132500
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 2.18,
|
| 1720 |
+
"learning_rate": 5.8325568788222745e-06,
|
| 1721 |
+
"loss": 6.5066,
|
| 1722 |
+
"step": 133000
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 2.19,
|
| 1726 |
+
"learning_rate": 5.815616053126429e-06,
|
| 1727 |
+
"loss": 6.5102,
|
| 1728 |
+
"step": 133500
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 2.2,
|
| 1732 |
+
"learning_rate": 5.798675227430586e-06,
|
| 1733 |
+
"loss": 6.5045,
|
| 1734 |
+
"step": 134000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 2.2,
|
| 1738 |
+
"learning_rate": 5.781734401734741e-06,
|
| 1739 |
+
"loss": 6.5027,
|
| 1740 |
+
"step": 134500
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 2.21,
|
| 1744 |
+
"learning_rate": 5.7647935760388964e-06,
|
| 1745 |
+
"loss": 6.502,
|
| 1746 |
+
"step": 135000
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 2.22,
|
| 1750 |
+
"learning_rate": 5.747852750343053e-06,
|
| 1751 |
+
"loss": 6.4952,
|
| 1752 |
+
"step": 135500
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 2.23,
|
| 1756 |
+
"learning_rate": 5.730911924647208e-06,
|
| 1757 |
+
"loss": 6.5016,
|
| 1758 |
+
"step": 136000
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 2.24,
|
| 1762 |
+
"learning_rate": 5.713971098951363e-06,
|
| 1763 |
+
"loss": 6.5039,
|
| 1764 |
+
"step": 136500
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.24,
|
| 1768 |
+
"learning_rate": 5.697030273255519e-06,
|
| 1769 |
+
"loss": 6.5077,
|
| 1770 |
+
"step": 137000
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.25,
|
| 1774 |
+
"learning_rate": 5.680089447559675e-06,
|
| 1775 |
+
"loss": 6.5012,
|
| 1776 |
+
"step": 137500
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 2.26,
|
| 1780 |
+
"learning_rate": 5.66314862186383e-06,
|
| 1781 |
+
"loss": 6.4987,
|
| 1782 |
+
"step": 138000
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 2.27,
|
| 1786 |
+
"learning_rate": 5.646207796167986e-06,
|
| 1787 |
+
"loss": 6.4977,
|
| 1788 |
+
"step": 138500
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 2.28,
|
| 1792 |
+
"learning_rate": 5.629266970472141e-06,
|
| 1793 |
+
"loss": 6.4992,
|
| 1794 |
+
"step": 139000
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.29,
|
| 1798 |
+
"learning_rate": 5.612326144776297e-06,
|
| 1799 |
+
"loss": 6.498,
|
| 1800 |
+
"step": 139500
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 2.29,
|
| 1804 |
+
"learning_rate": 5.595385319080452e-06,
|
| 1805 |
+
"loss": 6.5001,
|
| 1806 |
+
"step": 140000
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 2.29,
|
| 1810 |
+
"eval_accuracy": 0.1180354638977201,
|
| 1811 |
+
"eval_loss": 6.496232509613037,
|
| 1812 |
+
"eval_runtime": 603.9932,
|
| 1813 |
+
"eval_samples_per_second": 510.529,
|
| 1814 |
+
"eval_steps_per_second": 5.32,
|
| 1815 |
+
"step": 140000
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 2.3,
|
| 1819 |
+
"learning_rate": 5.578444493384608e-06,
|
| 1820 |
+
"loss": 6.4969,
|
| 1821 |
+
"step": 140500
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 2.31,
|
| 1825 |
+
"learning_rate": 5.561503667688764e-06,
|
| 1826 |
+
"loss": 6.4959,
|
| 1827 |
+
"step": 141000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 2.32,
|
| 1831 |
+
"learning_rate": 5.544562841992919e-06,
|
| 1832 |
+
"loss": 6.4939,
|
| 1833 |
+
"step": 141500
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 2.33,
|
| 1837 |
+
"learning_rate": 5.527622016297075e-06,
|
| 1838 |
+
"loss": 6.4908,
|
| 1839 |
+
"step": 142000
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 2.33,
|
| 1843 |
+
"learning_rate": 5.51068119060123e-06,
|
| 1844 |
+
"loss": 6.4954,
|
| 1845 |
+
"step": 142500
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 2.34,
|
| 1849 |
+
"learning_rate": 5.493740364905386e-06,
|
| 1850 |
+
"loss": 6.4911,
|
| 1851 |
+
"step": 143000
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 2.35,
|
| 1855 |
+
"learning_rate": 5.4767995392095415e-06,
|
| 1856 |
+
"loss": 6.4886,
|
| 1857 |
+
"step": 143500
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 2.36,
|
| 1861 |
+
"learning_rate": 5.459858713513697e-06,
|
| 1862 |
+
"loss": 6.492,
|
| 1863 |
+
"step": 144000
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 2.37,
|
| 1867 |
+
"learning_rate": 5.442917887817852e-06,
|
| 1868 |
+
"loss": 6.4917,
|
| 1869 |
+
"step": 144500
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 2.38,
|
| 1873 |
+
"learning_rate": 5.4259770621220086e-06,
|
| 1874 |
+
"loss": 6.491,
|
| 1875 |
+
"step": 145000
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 2.38,
|
| 1879 |
+
"learning_rate": 5.409036236426164e-06,
|
| 1880 |
+
"loss": 6.4872,
|
| 1881 |
+
"step": 145500
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 2.39,
|
| 1885 |
+
"learning_rate": 5.392095410730319e-06,
|
| 1886 |
+
"loss": 6.4833,
|
| 1887 |
+
"step": 146000
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 2.4,
|
| 1891 |
+
"learning_rate": 5.375154585034476e-06,
|
| 1892 |
+
"loss": 6.4912,
|
| 1893 |
+
"step": 146500
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 2.41,
|
| 1897 |
+
"learning_rate": 5.3582137593386305e-06,
|
| 1898 |
+
"loss": 6.4895,
|
| 1899 |
+
"step": 147000
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 2.42,
|
| 1903 |
+
"learning_rate": 5.341272933642786e-06,
|
| 1904 |
+
"loss": 6.4869,
|
| 1905 |
+
"step": 147500
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 2.43,
|
| 1909 |
+
"learning_rate": 5.324332107946941e-06,
|
| 1910 |
+
"loss": 6.4875,
|
| 1911 |
+
"step": 148000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 2.43,
|
| 1915 |
+
"learning_rate": 5.3073912822510975e-06,
|
| 1916 |
+
"loss": 6.4895,
|
| 1917 |
+
"step": 148500
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 2.44,
|
| 1921 |
+
"learning_rate": 5.290450456555252e-06,
|
| 1922 |
+
"loss": 6.48,
|
| 1923 |
+
"step": 149000
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 2.45,
|
| 1927 |
+
"learning_rate": 5.273509630859408e-06,
|
| 1928 |
+
"loss": 6.486,
|
| 1929 |
+
"step": 149500
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 2.46,
|
| 1933 |
+
"learning_rate": 5.256568805163565e-06,
|
| 1934 |
+
"loss": 6.4785,
|
| 1935 |
+
"step": 150000
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 2.46,
|
| 1939 |
+
"eval_accuracy": 0.11971311116288914,
|
| 1940 |
+
"eval_loss": 6.4821858406066895,
|
| 1941 |
+
"eval_runtime": 600.3142,
|
| 1942 |
+
"eval_samples_per_second": 513.658,
|
| 1943 |
+
"eval_steps_per_second": 5.352,
|
| 1944 |
+
"step": 150000
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.47,
|
| 1948 |
+
"learning_rate": 5.2396279794677195e-06,
|
| 1949 |
+
"loss": 6.4843,
|
| 1950 |
+
"step": 150500
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 2.47,
|
| 1954 |
+
"learning_rate": 5.222687153771875e-06,
|
| 1955 |
+
"loss": 6.4887,
|
| 1956 |
+
"step": 151000
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 2.48,
|
| 1960 |
+
"learning_rate": 5.205746328076031e-06,
|
| 1961 |
+
"loss": 6.4861,
|
| 1962 |
+
"step": 151500
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.49,
|
| 1966 |
+
"learning_rate": 5.1888055023801865e-06,
|
| 1967 |
+
"loss": 6.4865,
|
| 1968 |
+
"step": 152000
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 2.5,
|
| 1972 |
+
"learning_rate": 5.171864676684341e-06,
|
| 1973 |
+
"loss": 6.4767,
|
| 1974 |
+
"step": 152500
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 2.51,
|
| 1978 |
+
"learning_rate": 5.154923850988498e-06,
|
| 1979 |
+
"loss": 6.4775,
|
| 1980 |
+
"step": 153000
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 2.52,
|
| 1984 |
+
"learning_rate": 5.137983025292654e-06,
|
| 1985 |
+
"loss": 6.4793,
|
| 1986 |
+
"step": 153500
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 2.52,
|
| 1990 |
+
"learning_rate": 5.1210421995968085e-06,
|
| 1991 |
+
"loss": 6.479,
|
| 1992 |
+
"step": 154000
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 2.53,
|
| 1996 |
+
"learning_rate": 5.104101373900965e-06,
|
| 1997 |
+
"loss": 6.4783,
|
| 1998 |
+
"step": 154500
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 2.54,
|
| 2002 |
+
"learning_rate": 5.08716054820512e-06,
|
| 2003 |
+
"loss": 6.479,
|
| 2004 |
+
"step": 155000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 2.55,
|
| 2008 |
+
"learning_rate": 5.0702197225092755e-06,
|
| 2009 |
+
"loss": 6.4804,
|
| 2010 |
+
"step": 155500
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 2.56,
|
| 2014 |
+
"learning_rate": 5.053278896813432e-06,
|
| 2015 |
+
"loss": 6.473,
|
| 2016 |
+
"step": 156000
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 2.56,
|
| 2020 |
+
"learning_rate": 5.036338071117587e-06,
|
| 2021 |
+
"loss": 6.4769,
|
| 2022 |
+
"step": 156500
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 2.57,
|
| 2026 |
+
"learning_rate": 5.019397245421742e-06,
|
| 2027 |
+
"loss": 6.4778,
|
| 2028 |
+
"step": 157000
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 2.58,
|
| 2032 |
+
"learning_rate": 5.0024564197258974e-06,
|
| 2033 |
+
"loss": 6.4764,
|
| 2034 |
+
"step": 157500
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 2.59,
|
| 2038 |
+
"learning_rate": 4.985515594030054e-06,
|
| 2039 |
+
"loss": 6.47,
|
| 2040 |
+
"step": 158000
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 2.6,
|
| 2044 |
+
"learning_rate": 4.968574768334209e-06,
|
| 2045 |
+
"loss": 6.4716,
|
| 2046 |
+
"step": 158500
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 2.61,
|
| 2050 |
+
"learning_rate": 4.9516339426383645e-06,
|
| 2051 |
+
"loss": 6.4754,
|
| 2052 |
+
"step": 159000
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 2.61,
|
| 2056 |
+
"learning_rate": 4.93469311694252e-06,
|
| 2057 |
+
"loss": 6.4741,
|
| 2058 |
+
"step": 159500
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 2.62,
|
| 2062 |
+
"learning_rate": 4.917752291246676e-06,
|
| 2063 |
+
"loss": 6.4706,
|
| 2064 |
+
"step": 160000
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 2.62,
|
| 2068 |
+
"eval_accuracy": 0.12115731725892939,
|
| 2069 |
+
"eval_loss": 6.471449851989746,
|
| 2070 |
+
"eval_runtime": 601.7368,
|
| 2071 |
+
"eval_samples_per_second": 512.443,
|
| 2072 |
+
"eval_steps_per_second": 5.34,
|
| 2073 |
+
"step": 160000
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 2.63,
|
| 2077 |
+
"learning_rate": 4.900811465550832e-06,
|
| 2078 |
+
"loss": 6.4739,
|
| 2079 |
+
"step": 160500
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 2.64,
|
| 2083 |
+
"learning_rate": 4.8838706398549864e-06,
|
| 2084 |
+
"loss": 6.4717,
|
| 2085 |
+
"step": 161000
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 2.65,
|
| 2089 |
+
"learning_rate": 4.866929814159143e-06,
|
| 2090 |
+
"loss": 6.468,
|
| 2091 |
+
"step": 161500
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 2.65,
|
| 2095 |
+
"learning_rate": 4.849988988463299e-06,
|
| 2096 |
+
"loss": 6.4758,
|
| 2097 |
+
"step": 162000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 2.66,
|
| 2101 |
+
"learning_rate": 4.8330481627674535e-06,
|
| 2102 |
+
"loss": 6.4698,
|
| 2103 |
+
"step": 162500
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 2.67,
|
| 2107 |
+
"learning_rate": 4.816107337071609e-06,
|
| 2108 |
+
"loss": 6.4681,
|
| 2109 |
+
"step": 163000
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 2.68,
|
| 2113 |
+
"learning_rate": 4.799166511375765e-06,
|
| 2114 |
+
"loss": 6.47,
|
| 2115 |
+
"step": 163500
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 2.69,
|
| 2119 |
+
"learning_rate": 4.782225685679921e-06,
|
| 2120 |
+
"loss": 6.4711,
|
| 2121 |
+
"step": 164000
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 2.7,
|
| 2125 |
+
"learning_rate": 4.765284859984076e-06,
|
| 2126 |
+
"loss": 6.4707,
|
| 2127 |
+
"step": 164500
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 2.7,
|
| 2131 |
+
"learning_rate": 4.748344034288231e-06,
|
| 2132 |
+
"loss": 6.4717,
|
| 2133 |
+
"step": 165000
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 2.71,
|
| 2137 |
+
"learning_rate": 4.731403208592387e-06,
|
| 2138 |
+
"loss": 6.4677,
|
| 2139 |
+
"step": 165500
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 2.72,
|
| 2143 |
+
"learning_rate": 4.714462382896543e-06,
|
| 2144 |
+
"loss": 6.4677,
|
| 2145 |
+
"step": 166000
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 2.73,
|
| 2149 |
+
"learning_rate": 4.697521557200698e-06,
|
| 2150 |
+
"loss": 6.4642,
|
| 2151 |
+
"step": 166500
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 2.74,
|
| 2155 |
+
"learning_rate": 4.680580731504854e-06,
|
| 2156 |
+
"loss": 6.4587,
|
| 2157 |
+
"step": 167000
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 2.74,
|
| 2161 |
+
"learning_rate": 4.6636399058090096e-06,
|
| 2162 |
+
"loss": 6.4666,
|
| 2163 |
+
"step": 167500
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 2.75,
|
| 2167 |
+
"learning_rate": 4.646699080113165e-06,
|
| 2168 |
+
"loss": 6.4691,
|
| 2169 |
+
"step": 168000
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 2.76,
|
| 2173 |
+
"learning_rate": 4.629758254417321e-06,
|
| 2174 |
+
"loss": 6.4628,
|
| 2175 |
+
"step": 168500
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 2.77,
|
| 2179 |
+
"learning_rate": 4.612817428721476e-06,
|
| 2180 |
+
"loss": 6.4632,
|
| 2181 |
+
"step": 169000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 2.78,
|
| 2185 |
+
"learning_rate": 4.5958766030256315e-06,
|
| 2186 |
+
"loss": 6.4586,
|
| 2187 |
+
"step": 169500
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 2.79,
|
| 2191 |
+
"learning_rate": 4.578935777329788e-06,
|
| 2192 |
+
"loss": 6.4612,
|
| 2193 |
+
"step": 170000
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 2.79,
|
| 2197 |
+
"eval_accuracy": 0.12253875669316174,
|
| 2198 |
+
"eval_loss": 6.461043357849121,
|
| 2199 |
+
"eval_runtime": 602.3343,
|
| 2200 |
+
"eval_samples_per_second": 511.935,
|
| 2201 |
+
"eval_steps_per_second": 5.334,
|
| 2202 |
+
"step": 170000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 2.79,
|
| 2206 |
+
"learning_rate": 4.561994951633943e-06,
|
| 2207 |
+
"loss": 6.4673,
|
| 2208 |
+
"step": 170500
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 2.8,
|
| 2212 |
+
"learning_rate": 4.5450541259380986e-06,
|
| 2213 |
+
"loss": 6.4579,
|
| 2214 |
+
"step": 171000
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 2.81,
|
| 2218 |
+
"learning_rate": 4.528113300242254e-06,
|
| 2219 |
+
"loss": 6.4594,
|
| 2220 |
+
"step": 171500
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 2.82,
|
| 2224 |
+
"learning_rate": 4.51117247454641e-06,
|
| 2225 |
+
"loss": 6.4634,
|
| 2226 |
+
"step": 172000
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 2.83,
|
| 2230 |
+
"learning_rate": 4.494231648850566e-06,
|
| 2231 |
+
"loss": 6.4586,
|
| 2232 |
+
"step": 172500
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 2.83,
|
| 2236 |
+
"learning_rate": 4.4772908231547205e-06,
|
| 2237 |
+
"loss": 6.4587,
|
| 2238 |
+
"step": 173000
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 2.84,
|
| 2242 |
+
"learning_rate": 4.460349997458876e-06,
|
| 2243 |
+
"loss": 6.4599,
|
| 2244 |
+
"step": 173500
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 2.85,
|
| 2248 |
+
"learning_rate": 4.443409171763033e-06,
|
| 2249 |
+
"loss": 6.4596,
|
| 2250 |
+
"step": 174000
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 2.86,
|
| 2254 |
+
"learning_rate": 4.4264683460671875e-06,
|
| 2255 |
+
"loss": 6.4547,
|
| 2256 |
+
"step": 174500
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 2.87,
|
| 2260 |
+
"learning_rate": 4.409527520371343e-06,
|
| 2261 |
+
"loss": 6.4575,
|
| 2262 |
+
"step": 175000
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 2.88,
|
| 2266 |
+
"learning_rate": 4.392586694675499e-06,
|
| 2267 |
+
"loss": 6.4556,
|
| 2268 |
+
"step": 175500
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 2.88,
|
| 2272 |
+
"learning_rate": 4.375645868979655e-06,
|
| 2273 |
+
"loss": 6.4581,
|
| 2274 |
+
"step": 176000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 2.89,
|
| 2278 |
+
"learning_rate": 4.35870504328381e-06,
|
| 2279 |
+
"loss": 6.4535,
|
| 2280 |
+
"step": 176500
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 2.9,
|
| 2284 |
+
"learning_rate": 4.341764217587965e-06,
|
| 2285 |
+
"loss": 6.4617,
|
| 2286 |
+
"step": 177000
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 2.91,
|
| 2290 |
+
"learning_rate": 4.324823391892121e-06,
|
| 2291 |
+
"loss": 6.4535,
|
| 2292 |
+
"step": 177500
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 2.92,
|
| 2296 |
+
"learning_rate": 4.307882566196277e-06,
|
| 2297 |
+
"loss": 6.4543,
|
| 2298 |
+
"step": 178000
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 2.92,
|
| 2302 |
+
"learning_rate": 4.290941740500432e-06,
|
| 2303 |
+
"loss": 6.4535,
|
| 2304 |
+
"step": 178500
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 2.93,
|
| 2308 |
+
"learning_rate": 4.274000914804588e-06,
|
| 2309 |
+
"loss": 6.4523,
|
| 2310 |
+
"step": 179000
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 2.94,
|
| 2314 |
+
"learning_rate": 4.257060089108744e-06,
|
| 2315 |
+
"loss": 6.4542,
|
| 2316 |
+
"step": 179500
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 2.95,
|
| 2320 |
+
"learning_rate": 4.240119263412899e-06,
|
| 2321 |
+
"loss": 6.4485,
|
| 2322 |
+
"step": 180000
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 2.95,
|
| 2326 |
+
"eval_accuracy": 0.1232825828604018,
|
| 2327 |
+
"eval_loss": 6.4530487060546875,
|
| 2328 |
+
"eval_runtime": 601.0649,
|
| 2329 |
+
"eval_samples_per_second": 513.016,
|
| 2330 |
+
"eval_steps_per_second": 5.346,
|
| 2331 |
+
"step": 180000
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 2.96,
|
| 2335 |
+
"learning_rate": 4.223178437717055e-06,
|
| 2336 |
+
"loss": 6.455,
|
| 2337 |
+
"step": 180500
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 2.97,
|
| 2341 |
+
"learning_rate": 4.20623761202121e-06,
|
| 2342 |
+
"loss": 6.4519,
|
| 2343 |
+
"step": 181000
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 2.97,
|
| 2347 |
+
"learning_rate": 4.1892967863253655e-06,
|
| 2348 |
+
"loss": 6.455,
|
| 2349 |
+
"step": 181500
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 2.98,
|
| 2353 |
+
"learning_rate": 4.172355960629522e-06,
|
| 2354 |
+
"loss": 6.4511,
|
| 2355 |
+
"step": 182000
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 2.99,
|
| 2359 |
+
"learning_rate": 4.155415134933677e-06,
|
| 2360 |
+
"loss": 6.4468,
|
| 2361 |
+
"step": 182500
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 3.0,
|
| 2365 |
+
"learning_rate": 4.138474309237833e-06,
|
| 2366 |
+
"loss": 6.4473,
|
| 2367 |
+
"step": 183000
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 3.01,
|
| 2371 |
+
"learning_rate": 4.121533483541988e-06,
|
| 2372 |
+
"loss": 6.4471,
|
| 2373 |
+
"step": 183500
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 3.01,
|
| 2377 |
+
"learning_rate": 4.104592657846144e-06,
|
| 2378 |
+
"loss": 6.4467,
|
| 2379 |
+
"step": 184000
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 3.02,
|
| 2383 |
+
"learning_rate": 4.0876518321503e-06,
|
| 2384 |
+
"loss": 6.4515,
|
| 2385 |
+
"step": 184500
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 3.03,
|
| 2389 |
+
"learning_rate": 4.0707110064544545e-06,
|
| 2390 |
+
"loss": 6.448,
|
| 2391 |
+
"step": 185000
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 3.04,
|
| 2395 |
+
"learning_rate": 4.05377018075861e-06,
|
| 2396 |
+
"loss": 6.4491,
|
| 2397 |
+
"step": 185500
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 3.05,
|
| 2401 |
+
"learning_rate": 4.036829355062766e-06,
|
| 2402 |
+
"loss": 6.4505,
|
| 2403 |
+
"step": 186000
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 3.06,
|
| 2407 |
+
"learning_rate": 4.019888529366922e-06,
|
| 2408 |
+
"loss": 6.4471,
|
| 2409 |
+
"step": 186500
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 3.06,
|
| 2413 |
+
"learning_rate": 4.002947703671077e-06,
|
| 2414 |
+
"loss": 6.4435,
|
| 2415 |
+
"step": 187000
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 3.07,
|
| 2419 |
+
"learning_rate": 3.986006877975233e-06,
|
| 2420 |
+
"loss": 6.4472,
|
| 2421 |
+
"step": 187500
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 3.08,
|
| 2425 |
+
"learning_rate": 3.969066052279389e-06,
|
| 2426 |
+
"loss": 6.4453,
|
| 2427 |
+
"step": 188000
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 3.09,
|
| 2431 |
+
"learning_rate": 3.952125226583544e-06,
|
| 2432 |
+
"loss": 6.449,
|
| 2433 |
+
"step": 188500
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 3.1,
|
| 2437 |
+
"learning_rate": 3.935184400887699e-06,
|
| 2438 |
+
"loss": 6.4475,
|
| 2439 |
+
"step": 189000
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 3.11,
|
| 2443 |
+
"learning_rate": 3.918243575191855e-06,
|
| 2444 |
+
"loss": 6.4411,
|
| 2445 |
+
"step": 189500
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 3.11,
|
| 2449 |
+
"learning_rate": 3.9013027494960106e-06,
|
| 2450 |
+
"loss": 6.4477,
|
| 2451 |
+
"step": 190000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 3.11,
|
| 2455 |
+
"eval_accuracy": 0.1243327551641345,
|
| 2456 |
+
"eval_loss": 6.444092273712158,
|
| 2457 |
+
"eval_runtime": 605.0337,
|
| 2458 |
+
"eval_samples_per_second": 509.651,
|
| 2459 |
+
"eval_steps_per_second": 5.31,
|
| 2460 |
+
"step": 190000
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 3.12,
|
| 2464 |
+
"learning_rate": 3.884361923800166e-06,
|
| 2465 |
+
"loss": 6.4434,
|
| 2466 |
+
"step": 190500
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 3.13,
|
| 2470 |
+
"learning_rate": 3.867421098104322e-06,
|
| 2471 |
+
"loss": 6.4469,
|
| 2472 |
+
"step": 191000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 3.14,
|
| 2476 |
+
"learning_rate": 3.850480272408478e-06,
|
| 2477 |
+
"loss": 6.4424,
|
| 2478 |
+
"step": 191500
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 3.15,
|
| 2482 |
+
"learning_rate": 3.833539446712633e-06,
|
| 2483 |
+
"loss": 6.4427,
|
| 2484 |
+
"step": 192000
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 3.15,
|
| 2488 |
+
"learning_rate": 3.816598621016789e-06,
|
| 2489 |
+
"loss": 6.4407,
|
| 2490 |
+
"step": 192500
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 3.16,
|
| 2494 |
+
"learning_rate": 3.799657795320944e-06,
|
| 2495 |
+
"loss": 6.4434,
|
| 2496 |
+
"step": 193000
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 3.17,
|
| 2500 |
+
"learning_rate": 3.7827169696251e-06,
|
| 2501 |
+
"loss": 6.442,
|
| 2502 |
+
"step": 193500
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 3.18,
|
| 2506 |
+
"learning_rate": 3.7657761439292557e-06,
|
| 2507 |
+
"loss": 6.438,
|
| 2508 |
+
"step": 194000
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 3.19,
|
| 2512 |
+
"learning_rate": 3.748835318233411e-06,
|
| 2513 |
+
"loss": 6.4435,
|
| 2514 |
+
"step": 194500
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 3.2,
|
| 2518 |
+
"learning_rate": 3.7318944925375666e-06,
|
| 2519 |
+
"loss": 6.4418,
|
| 2520 |
+
"step": 195000
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 3.2,
|
| 2524 |
+
"learning_rate": 3.7149536668417223e-06,
|
| 2525 |
+
"loss": 6.4368,
|
| 2526 |
+
"step": 195500
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 3.21,
|
| 2530 |
+
"learning_rate": 3.6980128411458776e-06,
|
| 2531 |
+
"loss": 6.4424,
|
| 2532 |
+
"step": 196000
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 3.22,
|
| 2536 |
+
"learning_rate": 3.6810720154500333e-06,
|
| 2537 |
+
"loss": 6.4395,
|
| 2538 |
+
"step": 196500
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 3.23,
|
| 2542 |
+
"learning_rate": 3.6641311897541886e-06,
|
| 2543 |
+
"loss": 6.4424,
|
| 2544 |
+
"step": 197000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 3.24,
|
| 2548 |
+
"learning_rate": 3.6471903640583447e-06,
|
| 2549 |
+
"loss": 6.4355,
|
| 2550 |
+
"step": 197500
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 3.24,
|
| 2554 |
+
"learning_rate": 3.6302495383625004e-06,
|
| 2555 |
+
"loss": 6.4401,
|
| 2556 |
+
"step": 198000
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 3.25,
|
| 2560 |
+
"learning_rate": 3.6133087126666556e-06,
|
| 2561 |
+
"loss": 6.4408,
|
| 2562 |
+
"step": 198500
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 3.26,
|
| 2566 |
+
"learning_rate": 3.5963678869708113e-06,
|
| 2567 |
+
"loss": 6.4426,
|
| 2568 |
+
"step": 199000
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 3.27,
|
| 2572 |
+
"learning_rate": 3.579427061274967e-06,
|
| 2573 |
+
"loss": 6.4364,
|
| 2574 |
+
"step": 199500
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 3.28,
|
| 2578 |
+
"learning_rate": 3.5624862355791223e-06,
|
| 2579 |
+
"loss": 6.4373,
|
| 2580 |
+
"step": 200000
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 3.28,
|
| 2584 |
+
"eval_accuracy": 0.12513166304842002,
|
| 2585 |
+
"eval_loss": 6.439530849456787,
|
| 2586 |
+
"eval_runtime": 605.3063,
|
| 2587 |
+
"eval_samples_per_second": 509.421,
|
| 2588 |
+
"eval_steps_per_second": 5.308,
|
| 2589 |
+
"step": 200000
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 3.29,
|
| 2593 |
+
"learning_rate": 3.545545409883278e-06,
|
| 2594 |
+
"loss": 6.4369,
|
| 2595 |
+
"step": 200500
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 3.29,
|
| 2599 |
+
"learning_rate": 3.5286045841874332e-06,
|
| 2600 |
+
"loss": 6.4347,
|
| 2601 |
+
"step": 201000
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 3.3,
|
| 2605 |
+
"learning_rate": 3.511663758491589e-06,
|
| 2606 |
+
"loss": 6.435,
|
| 2607 |
+
"step": 201500
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 3.31,
|
| 2611 |
+
"learning_rate": 3.494722932795745e-06,
|
| 2612 |
+
"loss": 6.4335,
|
| 2613 |
+
"step": 202000
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 3.32,
|
| 2617 |
+
"learning_rate": 3.4777821070999003e-06,
|
| 2618 |
+
"loss": 6.4391,
|
| 2619 |
+
"step": 202500
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 3.33,
|
| 2623 |
+
"learning_rate": 3.460841281404056e-06,
|
| 2624 |
+
"loss": 6.4361,
|
| 2625 |
+
"step": 203000
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 3.33,
|
| 2629 |
+
"learning_rate": 3.4439004557082117e-06,
|
| 2630 |
+
"loss": 6.4367,
|
| 2631 |
+
"step": 203500
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 3.34,
|
| 2635 |
+
"learning_rate": 3.426959630012367e-06,
|
| 2636 |
+
"loss": 6.4366,
|
| 2637 |
+
"step": 204000
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 3.35,
|
| 2641 |
+
"learning_rate": 3.4100188043165226e-06,
|
| 2642 |
+
"loss": 6.4339,
|
| 2643 |
+
"step": 204500
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 3.36,
|
| 2647 |
+
"learning_rate": 3.393077978620678e-06,
|
| 2648 |
+
"loss": 6.4402,
|
| 2649 |
+
"step": 205000
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 3.37,
|
| 2653 |
+
"learning_rate": 3.3761371529248336e-06,
|
| 2654 |
+
"loss": 6.433,
|
| 2655 |
+
"step": 205500
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 3.38,
|
| 2659 |
+
"learning_rate": 3.3591963272289897e-06,
|
| 2660 |
+
"loss": 6.4381,
|
| 2661 |
+
"step": 206000
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 3.38,
|
| 2665 |
+
"learning_rate": 3.342255501533145e-06,
|
| 2666 |
+
"loss": 6.4369,
|
| 2667 |
+
"step": 206500
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 3.39,
|
| 2671 |
+
"learning_rate": 3.3253146758373007e-06,
|
| 2672 |
+
"loss": 6.4371,
|
| 2673 |
+
"step": 207000
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 3.4,
|
| 2677 |
+
"learning_rate": 3.3083738501414564e-06,
|
| 2678 |
+
"loss": 6.4339,
|
| 2679 |
+
"step": 207500
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 3.41,
|
| 2683 |
+
"learning_rate": 3.2914330244456116e-06,
|
| 2684 |
+
"loss": 6.434,
|
| 2685 |
+
"step": 208000
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 3.42,
|
| 2689 |
+
"learning_rate": 3.2744921987497673e-06,
|
| 2690 |
+
"loss": 6.4282,
|
| 2691 |
+
"step": 208500
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 3.42,
|
| 2695 |
+
"learning_rate": 3.2575513730539226e-06,
|
| 2696 |
+
"loss": 6.4306,
|
| 2697 |
+
"step": 209000
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 3.43,
|
| 2701 |
+
"learning_rate": 3.2406105473580783e-06,
|
| 2702 |
+
"loss": 6.4309,
|
| 2703 |
+
"step": 209500
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 3.44,
|
| 2707 |
+
"learning_rate": 3.2236697216622344e-06,
|
| 2708 |
+
"loss": 6.4351,
|
| 2709 |
+
"step": 210000
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 3.44,
|
| 2713 |
+
"eval_accuracy": 0.12585777151942737,
|
| 2714 |
+
"eval_loss": 6.432172775268555,
|
| 2715 |
+
"eval_runtime": 603.5717,
|
| 2716 |
+
"eval_samples_per_second": 510.885,
|
| 2717 |
+
"eval_steps_per_second": 5.323,
|
| 2718 |
+
"step": 210000
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 3.45,
|
| 2722 |
+
"learning_rate": 3.2067288959663897e-06,
|
| 2723 |
+
"loss": 6.4332,
|
| 2724 |
+
"step": 210500
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 3.46,
|
| 2728 |
+
"learning_rate": 3.1897880702705454e-06,
|
| 2729 |
+
"loss": 6.435,
|
| 2730 |
+
"step": 211000
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 3.47,
|
| 2734 |
+
"learning_rate": 3.172847244574701e-06,
|
| 2735 |
+
"loss": 6.4315,
|
| 2736 |
+
"step": 211500
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 3.47,
|
| 2740 |
+
"learning_rate": 3.1559064188788563e-06,
|
| 2741 |
+
"loss": 6.4349,
|
| 2742 |
+
"step": 212000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 3.48,
|
| 2746 |
+
"learning_rate": 3.138965593183012e-06,
|
| 2747 |
+
"loss": 6.4264,
|
| 2748 |
+
"step": 212500
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 3.49,
|
| 2752 |
+
"learning_rate": 3.1220247674871673e-06,
|
| 2753 |
+
"loss": 6.4345,
|
| 2754 |
+
"step": 213000
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 3.5,
|
| 2758 |
+
"learning_rate": 3.105083941791323e-06,
|
| 2759 |
+
"loss": 6.4287,
|
| 2760 |
+
"step": 213500
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 3.51,
|
| 2764 |
+
"learning_rate": 3.088143116095479e-06,
|
| 2765 |
+
"loss": 6.433,
|
| 2766 |
+
"step": 214000
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 3.51,
|
| 2770 |
+
"learning_rate": 3.071202290399634e-06,
|
| 2771 |
+
"loss": 6.4296,
|
| 2772 |
+
"step": 214500
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 3.52,
|
| 2776 |
+
"learning_rate": 3.05426146470379e-06,
|
| 2777 |
+
"loss": 6.4261,
|
| 2778 |
+
"step": 215000
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 3.53,
|
| 2782 |
+
"learning_rate": 3.0373206390079457e-06,
|
| 2783 |
+
"loss": 6.4337,
|
| 2784 |
+
"step": 215500
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 3.54,
|
| 2788 |
+
"learning_rate": 3.020379813312101e-06,
|
| 2789 |
+
"loss": 6.4286,
|
| 2790 |
+
"step": 216000
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 3.55,
|
| 2794 |
+
"learning_rate": 3.0034389876162567e-06,
|
| 2795 |
+
"loss": 6.4267,
|
| 2796 |
+
"step": 216500
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 3.56,
|
| 2800 |
+
"learning_rate": 2.986498161920412e-06,
|
| 2801 |
+
"loss": 6.431,
|
| 2802 |
+
"step": 217000
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 3.56,
|
| 2806 |
+
"learning_rate": 2.9695573362245676e-06,
|
| 2807 |
+
"loss": 6.4325,
|
| 2808 |
+
"step": 217500
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 3.57,
|
| 2812 |
+
"learning_rate": 2.9526165105287238e-06,
|
| 2813 |
+
"loss": 6.4253,
|
| 2814 |
+
"step": 218000
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 3.58,
|
| 2818 |
+
"learning_rate": 2.9356756848328786e-06,
|
| 2819 |
+
"loss": 6.426,
|
| 2820 |
+
"step": 218500
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 3.59,
|
| 2824 |
+
"learning_rate": 2.9187348591370347e-06,
|
| 2825 |
+
"loss": 6.4327,
|
| 2826 |
+
"step": 219000
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 3.6,
|
| 2830 |
+
"learning_rate": 2.9017940334411904e-06,
|
| 2831 |
+
"loss": 6.4291,
|
| 2832 |
+
"step": 219500
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 3.6,
|
| 2836 |
+
"learning_rate": 2.8848532077453457e-06,
|
| 2837 |
+
"loss": 6.4273,
|
| 2838 |
+
"step": 220000
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 3.6,
|
| 2842 |
+
"eval_accuracy": 0.1262452488408161,
|
| 2843 |
+
"eval_loss": 6.426393508911133,
|
| 2844 |
+
"eval_runtime": 606.2126,
|
| 2845 |
+
"eval_samples_per_second": 508.66,
|
| 2846 |
+
"eval_steps_per_second": 5.3,
|
| 2847 |
+
"step": 220000
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 3.61,
|
| 2851 |
+
"learning_rate": 2.8679123820495014e-06,
|
| 2852 |
+
"loss": 6.4263,
|
| 2853 |
+
"step": 220500
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 3.62,
|
| 2857 |
+
"learning_rate": 2.8509715563536566e-06,
|
| 2858 |
+
"loss": 6.4267,
|
| 2859 |
+
"step": 221000
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 3.63,
|
| 2863 |
+
"learning_rate": 2.8340307306578123e-06,
|
| 2864 |
+
"loss": 6.425,
|
| 2865 |
+
"step": 221500
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 3.64,
|
| 2869 |
+
"learning_rate": 2.817089904961968e-06,
|
| 2870 |
+
"loss": 6.4224,
|
| 2871 |
+
"step": 222000
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 3.65,
|
| 2875 |
+
"learning_rate": 2.8001490792661233e-06,
|
| 2876 |
+
"loss": 6.4217,
|
| 2877 |
+
"step": 222500
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 3.65,
|
| 2881 |
+
"learning_rate": 2.7832082535702794e-06,
|
| 2882 |
+
"loss": 6.4258,
|
| 2883 |
+
"step": 223000
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 3.66,
|
| 2887 |
+
"learning_rate": 2.766267427874435e-06,
|
| 2888 |
+
"loss": 6.4253,
|
| 2889 |
+
"step": 223500
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 3.67,
|
| 2893 |
+
"learning_rate": 2.7493266021785904e-06,
|
| 2894 |
+
"loss": 6.4266,
|
| 2895 |
+
"step": 224000
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 3.68,
|
| 2899 |
+
"learning_rate": 2.732385776482746e-06,
|
| 2900 |
+
"loss": 6.4271,
|
| 2901 |
+
"step": 224500
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 3.69,
|
| 2905 |
+
"learning_rate": 2.7154449507869017e-06,
|
| 2906 |
+
"loss": 6.4294,
|
| 2907 |
+
"step": 225000
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 3.69,
|
| 2911 |
+
"learning_rate": 2.698504125091057e-06,
|
| 2912 |
+
"loss": 6.4243,
|
| 2913 |
+
"step": 225500
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 3.7,
|
| 2917 |
+
"learning_rate": 2.6815632993952127e-06,
|
| 2918 |
+
"loss": 6.4242,
|
| 2919 |
+
"step": 226000
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 3.71,
|
| 2923 |
+
"learning_rate": 2.664622473699368e-06,
|
| 2924 |
+
"loss": 6.4242,
|
| 2925 |
+
"step": 226500
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 3.72,
|
| 2929 |
+
"learning_rate": 2.647681648003524e-06,
|
| 2930 |
+
"loss": 6.4262,
|
| 2931 |
+
"step": 227000
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 3.73,
|
| 2935 |
+
"learning_rate": 2.6307408223076798e-06,
|
| 2936 |
+
"loss": 6.4197,
|
| 2937 |
+
"step": 227500
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 3.74,
|
| 2941 |
+
"learning_rate": 2.613799996611835e-06,
|
| 2942 |
+
"loss": 6.4246,
|
| 2943 |
+
"step": 228000
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 3.74,
|
| 2947 |
+
"learning_rate": 2.5968591709159907e-06,
|
| 2948 |
+
"loss": 6.4237,
|
| 2949 |
+
"step": 228500
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 3.75,
|
| 2953 |
+
"learning_rate": 2.5799183452201464e-06,
|
| 2954 |
+
"loss": 6.421,
|
| 2955 |
+
"step": 229000
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 3.76,
|
| 2959 |
+
"learning_rate": 2.5629775195243017e-06,
|
| 2960 |
+
"loss": 6.4207,
|
| 2961 |
+
"step": 229500
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 3.77,
|
| 2965 |
+
"learning_rate": 2.5460366938284574e-06,
|
| 2966 |
+
"loss": 6.4153,
|
| 2967 |
+
"step": 230000
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 3.77,
|
| 2971 |
+
"eval_accuracy": 0.12687429199787553,
|
| 2972 |
+
"eval_loss": 6.421925067901611,
|
| 2973 |
+
"eval_runtime": 603.0076,
|
| 2974 |
+
"eval_samples_per_second": 511.363,
|
| 2975 |
+
"eval_steps_per_second": 5.328,
|
| 2976 |
+
"step": 230000
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 3.78,
|
| 2980 |
+
"learning_rate": 2.5290958681326126e-06,
|
| 2981 |
+
"loss": 6.4211,
|
| 2982 |
+
"step": 230500
|
| 2983 |
+
},
|
| 2984 |
+
{
|
| 2985 |
+
"epoch": 3.79,
|
| 2986 |
+
"learning_rate": 2.5121550424367688e-06,
|
| 2987 |
+
"loss": 6.4257,
|
| 2988 |
+
"step": 231000
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 3.79,
|
| 2992 |
+
"learning_rate": 2.495214216740924e-06,
|
| 2993 |
+
"loss": 6.4207,
|
| 2994 |
+
"step": 231500
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 3.8,
|
| 2998 |
+
"learning_rate": 2.4782733910450797e-06,
|
| 2999 |
+
"loss": 6.4207,
|
| 3000 |
+
"step": 232000
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 3.81,
|
| 3004 |
+
"learning_rate": 2.4613325653492354e-06,
|
| 3005 |
+
"loss": 6.4252,
|
| 3006 |
+
"step": 232500
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 3.82,
|
| 3010 |
+
"learning_rate": 2.4443917396533907e-06,
|
| 3011 |
+
"loss": 6.4226,
|
| 3012 |
+
"step": 233000
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 3.83,
|
| 3016 |
+
"learning_rate": 2.4274509139575464e-06,
|
| 3017 |
+
"loss": 6.4271,
|
| 3018 |
+
"step": 233500
|
| 3019 |
+
},
|
| 3020 |
+
{
|
| 3021 |
+
"epoch": 3.83,
|
| 3022 |
+
"learning_rate": 2.410510088261702e-06,
|
| 3023 |
+
"loss": 6.4204,
|
| 3024 |
+
"step": 234000
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 3.84,
|
| 3028 |
+
"learning_rate": 2.3935692625658577e-06,
|
| 3029 |
+
"loss": 6.423,
|
| 3030 |
+
"step": 234500
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 3.85,
|
| 3034 |
+
"learning_rate": 2.376628436870013e-06,
|
| 3035 |
+
"loss": 6.4211,
|
| 3036 |
+
"step": 235000
|
| 3037 |
+
},
|
| 3038 |
+
{
|
| 3039 |
+
"epoch": 3.86,
|
| 3040 |
+
"learning_rate": 2.3596876111741687e-06,
|
| 3041 |
+
"loss": 6.4206,
|
| 3042 |
+
"step": 235500
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 3.87,
|
| 3046 |
+
"learning_rate": 2.3427467854783244e-06,
|
| 3047 |
+
"loss": 6.4194,
|
| 3048 |
+
"step": 236000
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 3.88,
|
| 3052 |
+
"learning_rate": 2.32580595978248e-06,
|
| 3053 |
+
"loss": 6.4166,
|
| 3054 |
+
"step": 236500
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 3.88,
|
| 3058 |
+
"learning_rate": 2.3088651340866354e-06,
|
| 3059 |
+
"loss": 6.4259,
|
| 3060 |
+
"step": 237000
|
| 3061 |
+
},
|
| 3062 |
+
{
|
| 3063 |
+
"epoch": 3.89,
|
| 3064 |
+
"learning_rate": 2.291924308390791e-06,
|
| 3065 |
+
"loss": 6.4142,
|
| 3066 |
+
"step": 237500
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 3.9,
|
| 3070 |
+
"learning_rate": 2.2749834826949467e-06,
|
| 3071 |
+
"loss": 6.4215,
|
| 3072 |
+
"step": 238000
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 3.91,
|
| 3076 |
+
"learning_rate": 2.2580426569991024e-06,
|
| 3077 |
+
"loss": 6.4135,
|
| 3078 |
+
"step": 238500
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 3.92,
|
| 3082 |
+
"learning_rate": 2.2411018313032577e-06,
|
| 3083 |
+
"loss": 6.4174,
|
| 3084 |
+
"step": 239000
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 3.92,
|
| 3088 |
+
"learning_rate": 2.2241610056074134e-06,
|
| 3089 |
+
"loss": 6.4231,
|
| 3090 |
+
"step": 239500
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 3.93,
|
| 3094 |
+
"learning_rate": 2.207220179911569e-06,
|
| 3095 |
+
"loss": 6.4188,
|
| 3096 |
+
"step": 240000
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 3.93,
|
| 3100 |
+
"eval_accuracy": 0.12740316889553732,
|
| 3101 |
+
"eval_loss": 6.418206691741943,
|
| 3102 |
+
"eval_runtime": 604.087,
|
| 3103 |
+
"eval_samples_per_second": 510.45,
|
| 3104 |
+
"eval_steps_per_second": 5.319,
|
| 3105 |
+
"step": 240000
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 3.94,
|
| 3109 |
+
"learning_rate": 2.1902793542157248e-06,
|
| 3110 |
+
"loss": 6.4167,
|
| 3111 |
+
"step": 240500
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 3.95,
|
| 3115 |
+
"learning_rate": 2.17333852851988e-06,
|
| 3116 |
+
"loss": 6.4203,
|
| 3117 |
+
"step": 241000
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 3.96,
|
| 3121 |
+
"learning_rate": 2.1563977028240357e-06,
|
| 3122 |
+
"loss": 6.417,
|
| 3123 |
+
"step": 241500
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 3.97,
|
| 3127 |
+
"learning_rate": 2.1394568771281914e-06,
|
| 3128 |
+
"loss": 6.4171,
|
| 3129 |
+
"step": 242000
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 3.97,
|
| 3133 |
+
"learning_rate": 2.122516051432347e-06,
|
| 3134 |
+
"loss": 6.4182,
|
| 3135 |
+
"step": 242500
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 3.98,
|
| 3139 |
+
"learning_rate": 2.1055752257365024e-06,
|
| 3140 |
+
"loss": 6.4181,
|
| 3141 |
+
"step": 243000
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 3.99,
|
| 3145 |
+
"learning_rate": 2.088634400040658e-06,
|
| 3146 |
+
"loss": 6.4159,
|
| 3147 |
+
"step": 243500
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 4.0,
|
| 3151 |
+
"learning_rate": 2.0716935743448138e-06,
|
| 3152 |
+
"loss": 6.4162,
|
| 3153 |
+
"step": 244000
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 4.01,
|
| 3157 |
+
"learning_rate": 2.0547527486489694e-06,
|
| 3158 |
+
"loss": 6.4137,
|
| 3159 |
+
"step": 244500
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 4.01,
|
| 3163 |
+
"learning_rate": 2.0378119229531247e-06,
|
| 3164 |
+
"loss": 6.4163,
|
| 3165 |
+
"step": 245000
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 4.02,
|
| 3169 |
+
"learning_rate": 2.0208710972572804e-06,
|
| 3170 |
+
"loss": 6.4211,
|
| 3171 |
+
"step": 245500
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 4.03,
|
| 3175 |
+
"learning_rate": 2.003930271561436e-06,
|
| 3176 |
+
"loss": 6.4166,
|
| 3177 |
+
"step": 246000
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 4.04,
|
| 3181 |
+
"learning_rate": 1.986989445865592e-06,
|
| 3182 |
+
"loss": 6.416,
|
| 3183 |
+
"step": 246500
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 4.05,
|
| 3187 |
+
"learning_rate": 1.970048620169747e-06,
|
| 3188 |
+
"loss": 6.4162,
|
| 3189 |
+
"step": 247000
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 4.06,
|
| 3193 |
+
"learning_rate": 1.9531077944739027e-06,
|
| 3194 |
+
"loss": 6.4138,
|
| 3195 |
+
"step": 247500
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 4.06,
|
| 3199 |
+
"learning_rate": 1.9361669687780584e-06,
|
| 3200 |
+
"loss": 6.411,
|
| 3201 |
+
"step": 248000
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 4.07,
|
| 3205 |
+
"learning_rate": 1.919226143082214e-06,
|
| 3206 |
+
"loss": 6.4124,
|
| 3207 |
+
"step": 248500
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 4.08,
|
| 3211 |
+
"learning_rate": 1.9022853173863696e-06,
|
| 3212 |
+
"loss": 6.4154,
|
| 3213 |
+
"step": 249000
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 4.09,
|
| 3217 |
+
"learning_rate": 1.885344491690525e-06,
|
| 3218 |
+
"loss": 6.4149,
|
| 3219 |
+
"step": 249500
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 4.1,
|
| 3223 |
+
"learning_rate": 1.8684036659946808e-06,
|
| 3224 |
+
"loss": 6.4128,
|
| 3225 |
+
"step": 250000
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 4.1,
|
| 3229 |
+
"eval_accuracy": 0.12784806575663238,
|
| 3230 |
+
"eval_loss": 6.414950370788574,
|
| 3231 |
+
"eval_runtime": 607.4915,
|
| 3232 |
+
"eval_samples_per_second": 507.589,
|
| 3233 |
+
"eval_steps_per_second": 5.289,
|
| 3234 |
+
"step": 250000
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 4.1,
|
| 3238 |
+
"learning_rate": 1.8514628402988363e-06,
|
| 3239 |
+
"loss": 6.414,
|
| 3240 |
+
"step": 250500
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 4.11,
|
| 3244 |
+
"learning_rate": 1.8345220146029917e-06,
|
| 3245 |
+
"loss": 6.41,
|
| 3246 |
+
"step": 251000
|
| 3247 |
+
},
|
| 3248 |
+
{
|
| 3249 |
+
"epoch": 4.12,
|
| 3250 |
+
"learning_rate": 1.8175811889071474e-06,
|
| 3251 |
+
"loss": 6.4131,
|
| 3252 |
+
"step": 251500
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 4.13,
|
| 3256 |
+
"learning_rate": 1.8006403632113031e-06,
|
| 3257 |
+
"loss": 6.4146,
|
| 3258 |
+
"step": 252000
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 4.14,
|
| 3262 |
+
"learning_rate": 1.7836995375154586e-06,
|
| 3263 |
+
"loss": 6.4126,
|
| 3264 |
+
"step": 252500
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 4.15,
|
| 3268 |
+
"learning_rate": 1.766758711819614e-06,
|
| 3269 |
+
"loss": 6.4164,
|
| 3270 |
+
"step": 253000
|
| 3271 |
+
},
|
| 3272 |
+
{
|
| 3273 |
+
"epoch": 4.15,
|
| 3274 |
+
"learning_rate": 1.7498178861237698e-06,
|
| 3275 |
+
"loss": 6.4147,
|
| 3276 |
+
"step": 253500
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 4.16,
|
| 3280 |
+
"learning_rate": 1.7328770604279255e-06,
|
| 3281 |
+
"loss": 6.4068,
|
| 3282 |
+
"step": 254000
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 4.17,
|
| 3286 |
+
"learning_rate": 1.715936234732081e-06,
|
| 3287 |
+
"loss": 6.4117,
|
| 3288 |
+
"step": 254500
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 4.18,
|
| 3292 |
+
"learning_rate": 1.6989954090362364e-06,
|
| 3293 |
+
"loss": 6.415,
|
| 3294 |
+
"step": 255000
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 4.19,
|
| 3298 |
+
"learning_rate": 1.6820545833403921e-06,
|
| 3299 |
+
"loss": 6.4131,
|
| 3300 |
+
"step": 255500
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 4.19,
|
| 3304 |
+
"learning_rate": 1.6651137576445478e-06,
|
| 3305 |
+
"loss": 6.4125,
|
| 3306 |
+
"step": 256000
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 4.2,
|
| 3310 |
+
"learning_rate": 1.6481729319487033e-06,
|
| 3311 |
+
"loss": 6.4145,
|
| 3312 |
+
"step": 256500
|
| 3313 |
+
},
|
| 3314 |
+
{
|
| 3315 |
+
"epoch": 4.21,
|
| 3316 |
+
"learning_rate": 1.6312321062528588e-06,
|
| 3317 |
+
"loss": 6.4155,
|
| 3318 |
+
"step": 257000
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 4.22,
|
| 3322 |
+
"learning_rate": 1.6142912805570147e-06,
|
| 3323 |
+
"loss": 6.4066,
|
| 3324 |
+
"step": 257500
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 4.23,
|
| 3328 |
+
"learning_rate": 1.5973504548611701e-06,
|
| 3329 |
+
"loss": 6.4129,
|
| 3330 |
+
"step": 258000
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 4.24,
|
| 3334 |
+
"learning_rate": 1.5804096291653256e-06,
|
| 3335 |
+
"loss": 6.4079,
|
| 3336 |
+
"step": 258500
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 4.24,
|
| 3340 |
+
"learning_rate": 1.563468803469481e-06,
|
| 3341 |
+
"loss": 6.4124,
|
| 3342 |
+
"step": 259000
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 4.25,
|
| 3346 |
+
"learning_rate": 1.546527977773637e-06,
|
| 3347 |
+
"loss": 6.4124,
|
| 3348 |
+
"step": 259500
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 4.26,
|
| 3352 |
+
"learning_rate": 1.5295871520777925e-06,
|
| 3353 |
+
"loss": 6.4189,
|
| 3354 |
+
"step": 260000
|
| 3355 |
+
},
|
| 3356 |
+
{
|
| 3357 |
+
"epoch": 4.26,
|
| 3358 |
+
"eval_accuracy": 0.12798065025917943,
|
| 3359 |
+
"eval_loss": 6.412052154541016,
|
| 3360 |
+
"eval_runtime": 604.5339,
|
| 3361 |
+
"eval_samples_per_second": 510.072,
|
| 3362 |
+
"eval_steps_per_second": 5.315,
|
| 3363 |
+
"step": 260000
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 4.27,
|
| 3367 |
+
"learning_rate": 1.512646326381948e-06,
|
| 3368 |
+
"loss": 6.4115,
|
| 3369 |
+
"step": 260500
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 4.28,
|
| 3373 |
+
"learning_rate": 1.4957055006861034e-06,
|
| 3374 |
+
"loss": 6.4086,
|
| 3375 |
+
"step": 261000
|
| 3376 |
+
},
|
| 3377 |
+
{
|
| 3378 |
+
"epoch": 4.28,
|
| 3379 |
+
"learning_rate": 1.4787646749902593e-06,
|
| 3380 |
+
"loss": 6.4153,
|
| 3381 |
+
"step": 261500
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 4.29,
|
| 3385 |
+
"learning_rate": 1.4618238492944148e-06,
|
| 3386 |
+
"loss": 6.4042,
|
| 3387 |
+
"step": 262000
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 4.3,
|
| 3391 |
+
"learning_rate": 1.4448830235985703e-06,
|
| 3392 |
+
"loss": 6.4188,
|
| 3393 |
+
"step": 262500
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 4.31,
|
| 3397 |
+
"learning_rate": 1.4279421979027258e-06,
|
| 3398 |
+
"loss": 6.4107,
|
| 3399 |
+
"step": 263000
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 4.32,
|
| 3403 |
+
"learning_rate": 1.4110013722068817e-06,
|
| 3404 |
+
"loss": 6.4104,
|
| 3405 |
+
"step": 263500
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 4.33,
|
| 3409 |
+
"learning_rate": 1.3940605465110372e-06,
|
| 3410 |
+
"loss": 6.4108,
|
| 3411 |
+
"step": 264000
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 4.33,
|
| 3415 |
+
"learning_rate": 1.3771197208151926e-06,
|
| 3416 |
+
"loss": 6.4096,
|
| 3417 |
+
"step": 264500
|
| 3418 |
+
},
|
| 3419 |
+
{
|
| 3420 |
+
"epoch": 4.34,
|
| 3421 |
+
"learning_rate": 1.3601788951193481e-06,
|
| 3422 |
+
"loss": 6.4139,
|
| 3423 |
+
"step": 265000
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 4.35,
|
| 3427 |
+
"learning_rate": 1.343238069423504e-06,
|
| 3428 |
+
"loss": 6.4142,
|
| 3429 |
+
"step": 265500
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 4.36,
|
| 3433 |
+
"learning_rate": 1.3262972437276595e-06,
|
| 3434 |
+
"loss": 6.4186,
|
| 3435 |
+
"step": 266000
|
| 3436 |
+
},
|
| 3437 |
+
{
|
| 3438 |
+
"epoch": 4.37,
|
| 3439 |
+
"learning_rate": 1.309356418031815e-06,
|
| 3440 |
+
"loss": 6.4126,
|
| 3441 |
+
"step": 266500
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 4.37,
|
| 3445 |
+
"learning_rate": 1.2924155923359705e-06,
|
| 3446 |
+
"loss": 6.408,
|
| 3447 |
+
"step": 267000
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 4.38,
|
| 3451 |
+
"learning_rate": 1.2754747666401264e-06,
|
| 3452 |
+
"loss": 6.4109,
|
| 3453 |
+
"step": 267500
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 4.39,
|
| 3457 |
+
"learning_rate": 1.2585339409442818e-06,
|
| 3458 |
+
"loss": 6.4117,
|
| 3459 |
+
"step": 268000
|
| 3460 |
+
},
|
| 3461 |
+
{
|
| 3462 |
+
"epoch": 4.4,
|
| 3463 |
+
"learning_rate": 1.2415931152484373e-06,
|
| 3464 |
+
"loss": 6.4088,
|
| 3465 |
+
"step": 268500
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 4.41,
|
| 3469 |
+
"learning_rate": 1.224652289552593e-06,
|
| 3470 |
+
"loss": 6.4152,
|
| 3471 |
+
"step": 269000
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 4.42,
|
| 3475 |
+
"learning_rate": 1.2077114638567485e-06,
|
| 3476 |
+
"loss": 6.41,
|
| 3477 |
+
"step": 269500
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 4.42,
|
| 3481 |
+
"learning_rate": 1.1907706381609042e-06,
|
| 3482 |
+
"loss": 6.4102,
|
| 3483 |
+
"step": 270000
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 4.42,
|
| 3487 |
+
"eval_accuracy": 0.12817913105652246,
|
| 3488 |
+
"eval_loss": 6.411165237426758,
|
| 3489 |
+
"eval_runtime": 602.0627,
|
| 3490 |
+
"eval_samples_per_second": 512.166,
|
| 3491 |
+
"eval_steps_per_second": 5.337,
|
| 3492 |
+
"step": 270000
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 4.43,
|
| 3496 |
+
"learning_rate": 1.1738298124650597e-06,
|
| 3497 |
+
"loss": 6.4107,
|
| 3498 |
+
"step": 270500
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 4.44,
|
| 3502 |
+
"learning_rate": 1.1568889867692153e-06,
|
| 3503 |
+
"loss": 6.4099,
|
| 3504 |
+
"step": 271000
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 4.45,
|
| 3508 |
+
"learning_rate": 1.1399481610733708e-06,
|
| 3509 |
+
"loss": 6.4162,
|
| 3510 |
+
"step": 271500
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 4.46,
|
| 3514 |
+
"learning_rate": 1.1230073353775265e-06,
|
| 3515 |
+
"loss": 6.4103,
|
| 3516 |
+
"step": 272000
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 4.47,
|
| 3520 |
+
"learning_rate": 1.106066509681682e-06,
|
| 3521 |
+
"loss": 6.4159,
|
| 3522 |
+
"step": 272500
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 4.47,
|
| 3526 |
+
"learning_rate": 1.0891256839858377e-06,
|
| 3527 |
+
"loss": 6.4052,
|
| 3528 |
+
"step": 273000
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 4.48,
|
| 3532 |
+
"learning_rate": 1.0721848582899932e-06,
|
| 3533 |
+
"loss": 6.4079,
|
| 3534 |
+
"step": 273500
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 4.49,
|
| 3538 |
+
"learning_rate": 1.0552440325941489e-06,
|
| 3539 |
+
"loss": 6.4071,
|
| 3540 |
+
"step": 274000
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 4.5,
|
| 3544 |
+
"learning_rate": 1.0383032068983043e-06,
|
| 3545 |
+
"loss": 6.4131,
|
| 3546 |
+
"step": 274500
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 4.51,
|
| 3550 |
+
"learning_rate": 1.02136238120246e-06,
|
| 3551 |
+
"loss": 6.4077,
|
| 3552 |
+
"step": 275000
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 4.51,
|
| 3556 |
+
"learning_rate": 1.0044215555066155e-06,
|
| 3557 |
+
"loss": 6.4113,
|
| 3558 |
+
"step": 275500
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 4.52,
|
| 3562 |
+
"learning_rate": 9.87480729810771e-07,
|
| 3563 |
+
"loss": 6.4141,
|
| 3564 |
+
"step": 276000
|
| 3565 |
+
},
|
| 3566 |
+
{
|
| 3567 |
+
"epoch": 4.53,
|
| 3568 |
+
"learning_rate": 9.705399041149267e-07,
|
| 3569 |
+
"loss": 6.4081,
|
| 3570 |
+
"step": 276500
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 4.54,
|
| 3574 |
+
"learning_rate": 9.535990784190823e-07,
|
| 3575 |
+
"loss": 6.4071,
|
| 3576 |
+
"step": 277000
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 4.55,
|
| 3580 |
+
"learning_rate": 9.366582527232377e-07,
|
| 3581 |
+
"loss": 6.4078,
|
| 3582 |
+
"step": 277500
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 4.56,
|
| 3586 |
+
"learning_rate": 9.197174270273934e-07,
|
| 3587 |
+
"loss": 6.4117,
|
| 3588 |
+
"step": 278000
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 4.56,
|
| 3592 |
+
"learning_rate": 9.027766013315489e-07,
|
| 3593 |
+
"loss": 6.4077,
|
| 3594 |
+
"step": 278500
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 4.57,
|
| 3598 |
+
"learning_rate": 8.858357756357046e-07,
|
| 3599 |
+
"loss": 6.4106,
|
| 3600 |
+
"step": 279000
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 4.58,
|
| 3604 |
+
"learning_rate": 8.688949499398601e-07,
|
| 3605 |
+
"loss": 6.4134,
|
| 3606 |
+
"step": 279500
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 4.59,
|
| 3610 |
+
"learning_rate": 8.519541242440158e-07,
|
| 3611 |
+
"loss": 6.4105,
|
| 3612 |
+
"step": 280000
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 4.59,
|
| 3616 |
+
"eval_accuracy": 0.1284733583672517,
|
| 3617 |
+
"eval_loss": 6.408677577972412,
|
| 3618 |
+
"eval_runtime": 603.8755,
|
| 3619 |
+
"eval_samples_per_second": 510.628,
|
| 3620 |
+
"eval_steps_per_second": 5.321,
|
| 3621 |
+
"step": 280000
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 4.6,
|
| 3625 |
+
"learning_rate": 8.350132985481713e-07,
|
| 3626 |
+
"loss": 6.4046,
|
| 3627 |
+
"step": 280500
|
| 3628 |
+
},
|
| 3629 |
+
{
|
| 3630 |
+
"epoch": 4.6,
|
| 3631 |
+
"learning_rate": 8.180724728523269e-07,
|
| 3632 |
+
"loss": 6.4109,
|
| 3633 |
+
"step": 281000
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 4.61,
|
| 3637 |
+
"learning_rate": 8.011316471564824e-07,
|
| 3638 |
+
"loss": 6.4117,
|
| 3639 |
+
"step": 281500
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 4.62,
|
| 3643 |
+
"learning_rate": 7.841908214606381e-07,
|
| 3644 |
+
"loss": 6.4107,
|
| 3645 |
+
"step": 282000
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 4.63,
|
| 3649 |
+
"learning_rate": 7.672499957647936e-07,
|
| 3650 |
+
"loss": 6.415,
|
| 3651 |
+
"step": 282500
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 4.64,
|
| 3655 |
+
"learning_rate": 7.503091700689493e-07,
|
| 3656 |
+
"loss": 6.4066,
|
| 3657 |
+
"step": 283000
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 4.65,
|
| 3661 |
+
"learning_rate": 7.333683443731048e-07,
|
| 3662 |
+
"loss": 6.4075,
|
| 3663 |
+
"step": 283500
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 4.65,
|
| 3667 |
+
"learning_rate": 7.164275186772605e-07,
|
| 3668 |
+
"loss": 6.4086,
|
| 3669 |
+
"step": 284000
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 4.66,
|
| 3673 |
+
"learning_rate": 6.994866929814159e-07,
|
| 3674 |
+
"loss": 6.4054,
|
| 3675 |
+
"step": 284500
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 4.67,
|
| 3679 |
+
"learning_rate": 6.825458672855715e-07,
|
| 3680 |
+
"loss": 6.4022,
|
| 3681 |
+
"step": 285000
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 4.68,
|
| 3685 |
+
"learning_rate": 6.656050415897271e-07,
|
| 3686 |
+
"loss": 6.406,
|
| 3687 |
+
"step": 285500
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 4.69,
|
| 3691 |
+
"learning_rate": 6.486642158938827e-07,
|
| 3692 |
+
"loss": 6.4032,
|
| 3693 |
+
"step": 286000
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 4.69,
|
| 3697 |
+
"learning_rate": 6.317233901980383e-07,
|
| 3698 |
+
"loss": 6.4105,
|
| 3699 |
+
"step": 286500
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 4.7,
|
| 3703 |
+
"learning_rate": 6.147825645021939e-07,
|
| 3704 |
+
"loss": 6.4073,
|
| 3705 |
+
"step": 287000
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 4.71,
|
| 3709 |
+
"learning_rate": 5.978417388063494e-07,
|
| 3710 |
+
"loss": 6.4058,
|
| 3711 |
+
"step": 287500
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 4.72,
|
| 3715 |
+
"learning_rate": 5.80900913110505e-07,
|
| 3716 |
+
"loss": 6.4123,
|
| 3717 |
+
"step": 288000
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 4.73,
|
| 3721 |
+
"learning_rate": 5.639600874146606e-07,
|
| 3722 |
+
"loss": 6.4132,
|
| 3723 |
+
"step": 288500
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 4.74,
|
| 3727 |
+
"learning_rate": 5.470192617188162e-07,
|
| 3728 |
+
"loss": 6.4072,
|
| 3729 |
+
"step": 289000
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 4.74,
|
| 3733 |
+
"learning_rate": 5.300784360229718e-07,
|
| 3734 |
+
"loss": 6.4085,
|
| 3735 |
+
"step": 289500
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 4.75,
|
| 3739 |
+
"learning_rate": 5.131376103271274e-07,
|
| 3740 |
+
"loss": 6.4065,
|
| 3741 |
+
"step": 290000
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 4.75,
|
| 3745 |
+
"eval_accuracy": 0.12868705268687342,
|
| 3746 |
+
"eval_loss": 6.406748294830322,
|
| 3747 |
+
"eval_runtime": 608.2162,
|
| 3748 |
+
"eval_samples_per_second": 506.984,
|
| 3749 |
+
"eval_steps_per_second": 5.283,
|
| 3750 |
+
"step": 290000
|
| 3751 |
+
},
|
| 3752 |
+
{
|
| 3753 |
+
"epoch": 4.76,
|
| 3754 |
+
"learning_rate": 4.96196784631283e-07,
|
| 3755 |
+
"loss": 6.409,
|
| 3756 |
+
"step": 290500
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 4.77,
|
| 3760 |
+
"learning_rate": 4.792559589354385e-07,
|
| 3761 |
+
"loss": 6.4061,
|
| 3762 |
+
"step": 291000
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 4.78,
|
| 3766 |
+
"learning_rate": 4.623151332395941e-07,
|
| 3767 |
+
"loss": 6.4078,
|
| 3768 |
+
"step": 291500
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 4.78,
|
| 3772 |
+
"learning_rate": 4.453743075437497e-07,
|
| 3773 |
+
"loss": 6.4084,
|
| 3774 |
+
"step": 292000
|
| 3775 |
+
},
|
| 3776 |
+
{
|
| 3777 |
+
"epoch": 4.79,
|
| 3778 |
+
"learning_rate": 4.284334818479053e-07,
|
| 3779 |
+
"loss": 6.412,
|
| 3780 |
+
"step": 292500
|
| 3781 |
+
},
|
| 3782 |
+
{
|
| 3783 |
+
"epoch": 4.8,
|
| 3784 |
+
"learning_rate": 4.114926561520609e-07,
|
| 3785 |
+
"loss": 6.4085,
|
| 3786 |
+
"step": 293000
|
| 3787 |
+
},
|
| 3788 |
+
{
|
| 3789 |
+
"epoch": 4.81,
|
| 3790 |
+
"learning_rate": 3.9455183045621646e-07,
|
| 3791 |
+
"loss": 6.4022,
|
| 3792 |
+
"step": 293500
|
| 3793 |
+
},
|
| 3794 |
+
{
|
| 3795 |
+
"epoch": 4.82,
|
| 3796 |
+
"learning_rate": 3.7761100476037205e-07,
|
| 3797 |
+
"loss": 6.4065,
|
| 3798 |
+
"step": 294000
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 4.83,
|
| 3802 |
+
"learning_rate": 3.6067017906452763e-07,
|
| 3803 |
+
"loss": 6.4114,
|
| 3804 |
+
"step": 294500
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 4.83,
|
| 3808 |
+
"learning_rate": 3.437293533686832e-07,
|
| 3809 |
+
"loss": 6.4085,
|
| 3810 |
+
"step": 295000
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 4.84,
|
| 3814 |
+
"learning_rate": 3.267885276728388e-07,
|
| 3815 |
+
"loss": 6.4117,
|
| 3816 |
+
"step": 295500
|
| 3817 |
+
},
|
| 3818 |
+
{
|
| 3819 |
+
"epoch": 4.85,
|
| 3820 |
+
"learning_rate": 3.098477019769944e-07,
|
| 3821 |
+
"loss": 6.4066,
|
| 3822 |
+
"step": 296000
|
| 3823 |
+
},
|
| 3824 |
+
{
|
| 3825 |
+
"epoch": 4.86,
|
| 3826 |
+
"learning_rate": 2.9290687628114997e-07,
|
| 3827 |
+
"loss": 6.4074,
|
| 3828 |
+
"step": 296500
|
| 3829 |
+
},
|
| 3830 |
+
{
|
| 3831 |
+
"epoch": 4.87,
|
| 3832 |
+
"learning_rate": 2.7596605058530556e-07,
|
| 3833 |
+
"loss": 6.4059,
|
| 3834 |
+
"step": 297000
|
| 3835 |
+
},
|
| 3836 |
+
{
|
| 3837 |
+
"epoch": 4.87,
|
| 3838 |
+
"learning_rate": 2.5902522488946114e-07,
|
| 3839 |
+
"loss": 6.4027,
|
| 3840 |
+
"step": 297500
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 4.88,
|
| 3844 |
+
"learning_rate": 2.4208439919361673e-07,
|
| 3845 |
+
"loss": 6.4034,
|
| 3846 |
+
"step": 298000
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 4.89,
|
| 3850 |
+
"learning_rate": 2.2514357349777229e-07,
|
| 3851 |
+
"loss": 6.4053,
|
| 3852 |
+
"step": 298500
|
| 3853 |
+
},
|
| 3854 |
+
{
|
| 3855 |
+
"epoch": 4.9,
|
| 3856 |
+
"learning_rate": 2.0820274780192787e-07,
|
| 3857 |
+
"loss": 6.4066,
|
| 3858 |
+
"step": 299000
|
| 3859 |
+
},
|
| 3860 |
+
{
|
| 3861 |
+
"epoch": 4.91,
|
| 3862 |
+
"learning_rate": 1.9126192210608346e-07,
|
| 3863 |
+
"loss": 6.4098,
|
| 3864 |
+
"step": 299500
|
| 3865 |
+
},
|
| 3866 |
+
{
|
| 3867 |
+
"epoch": 4.92,
|
| 3868 |
+
"learning_rate": 1.7432109641023904e-07,
|
| 3869 |
+
"loss": 6.4082,
|
| 3870 |
+
"step": 300000
|
| 3871 |
+
},
|
| 3872 |
+
{
|
| 3873 |
+
"epoch": 4.92,
|
| 3874 |
+
"eval_accuracy": 0.12851127491212921,
|
| 3875 |
+
"eval_loss": 6.407046794891357,
|
| 3876 |
+
"eval_runtime": 604.7864,
|
| 3877 |
+
"eval_samples_per_second": 509.859,
|
| 3878 |
+
"eval_steps_per_second": 5.313,
|
| 3879 |
+
"step": 300000
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 4.92,
|
| 3883 |
+
"learning_rate": 1.5738027071439465e-07,
|
| 3884 |
+
"loss": 6.4019,
|
| 3885 |
+
"step": 300500
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 4.93,
|
| 3889 |
+
"learning_rate": 1.404394450185502e-07,
|
| 3890 |
+
"loss": 6.4073,
|
| 3891 |
+
"step": 301000
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 4.94,
|
| 3895 |
+
"learning_rate": 1.234986193227058e-07,
|
| 3896 |
+
"loss": 6.403,
|
| 3897 |
+
"step": 301500
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 4.95,
|
| 3901 |
+
"learning_rate": 1.0655779362686138e-07,
|
| 3902 |
+
"loss": 6.4098,
|
| 3903 |
+
"step": 302000
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 4.96,
|
| 3907 |
+
"learning_rate": 8.961696793101697e-08,
|
| 3908 |
+
"loss": 6.4055,
|
| 3909 |
+
"step": 302500
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 4.96,
|
| 3913 |
+
"learning_rate": 7.267614223517255e-08,
|
| 3914 |
+
"loss": 6.409,
|
| 3915 |
+
"step": 303000
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 4.97,
|
| 3919 |
+
"learning_rate": 5.5735316539328136e-08,
|
| 3920 |
+
"loss": 6.4094,
|
| 3921 |
+
"step": 303500
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 4.98,
|
| 3925 |
+
"learning_rate": 3.8794490843483714e-08,
|
| 3926 |
+
"loss": 6.4038,
|
| 3927 |
+
"step": 304000
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 4.99,
|
| 3931 |
+
"learning_rate": 2.18536651476393e-08,
|
| 3932 |
+
"loss": 6.4017,
|
| 3933 |
+
"step": 304500
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 5.0,
|
| 3937 |
+
"learning_rate": 4.912839451794881e-09,
|
| 3938 |
+
"loss": 6.4089,
|
| 3939 |
+
"step": 305000
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 5.0,
|
| 3943 |
+
"step": 305145,
|
| 3944 |
+
"total_flos": 6.419869954421555e+16,
|
| 3945 |
+
"train_loss": 6.680038520005282,
|
| 3946 |
+
"train_runtime": 123390.5046,
|
| 3947 |
+
"train_samples_per_second": 237.407,
|
| 3948 |
+
"train_steps_per_second": 2.473
|
| 3949 |
+
}
|
| 3950 |
+
],
|
| 3951 |
+
"logging_steps": 500,
|
| 3952 |
+
"max_steps": 305145,
|
| 3953 |
+
"num_train_epochs": 5,
|
| 3954 |
+
"save_steps": 10000,
|
| 3955 |
+
"total_flos": 6.419869954421555e+16,
|
| 3956 |
+
"trial_name": null,
|
| 3957 |
+
"trial_params": null
|
| 3958 |
+
}
|