End of training
Browse files- README.md +3 -2
- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +2511 -0
- training_eval_loss.png +0 -0
- training_loss.png +0 -0
README.md
CHANGED
|
@@ -3,6 +3,7 @@ license: llama3
|
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
| 5 |
- llama-factory
|
|
|
|
| 6 |
- generated_from_trainer
|
| 7 |
base_model: meta-llama/Meta-Llama-3-70B-Instruct
|
| 8 |
model-index:
|
|
@@ -15,9 +16,9 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# llama3-70B-lora-pretrain_v2
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-70B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-70B-Instruct) on
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
-
- Loss: 1.
|
| 21 |
|
| 22 |
## Model description
|
| 23 |
|
|
|
|
| 3 |
library_name: peft
|
| 4 |
tags:
|
| 5 |
- llama-factory
|
| 6 |
+
- lora
|
| 7 |
- generated_from_trainer
|
| 8 |
base_model: meta-llama/Meta-Llama-3-70B-Instruct
|
| 9 |
model-index:
|
|
|
|
| 16 |
|
| 17 |
# llama3-70B-lora-pretrain_v2
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-70B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-70B-Instruct) on the sm_artile dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 1.9382
|
| 22 |
|
| 23 |
## Model description
|
| 24 |
|
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.998592210229939,
|
| 3 |
+
"eval_loss": 1.9382482767105103,
|
| 4 |
+
"eval_runtime": 935.7433,
|
| 5 |
+
"eval_samples_per_second": 1.012,
|
| 6 |
+
"eval_steps_per_second": 0.507,
|
| 7 |
+
"perplexity": 6.946571834848492,
|
| 8 |
+
"total_flos": 1.0917373877893988e+19,
|
| 9 |
+
"train_loss": 2.069366264641751,
|
| 10 |
+
"train_runtime": 97122.8592,
|
| 11 |
+
"train_samples_per_second": 0.263,
|
| 12 |
+
"train_steps_per_second": 0.033
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.998592210229939,
|
| 3 |
+
"eval_loss": 1.9382482767105103,
|
| 4 |
+
"eval_runtime": 935.7433,
|
| 5 |
+
"eval_samples_per_second": 1.012,
|
| 6 |
+
"eval_steps_per_second": 0.507,
|
| 7 |
+
"perplexity": 6.946571834848492
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.998592210229939,
|
| 3 |
+
"total_flos": 1.0917373877893988e+19,
|
| 4 |
+
"train_loss": 2.069366264641751,
|
| 5 |
+
"train_runtime": 97122.8592,
|
| 6 |
+
"train_samples_per_second": 0.263,
|
| 7 |
+
"train_steps_per_second": 0.033
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2511 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.998592210229939,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 3195,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.009385265133740028,
|
| 13 |
+
"grad_norm": 0.4665657877922058,
|
| 14 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 15 |
+
"loss": 2.9241,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.018770530267480056,
|
| 20 |
+
"grad_norm": 0.36958208680152893,
|
| 21 |
+
"learning_rate": 3.4000000000000005e-06,
|
| 22 |
+
"loss": 2.9288,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.028155795401220086,
|
| 27 |
+
"grad_norm": 0.3911522626876831,
|
| 28 |
+
"learning_rate": 5.2e-06,
|
| 29 |
+
"loss": 2.8784,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.03754106053496011,
|
| 34 |
+
"grad_norm": 0.6396021842956543,
|
| 35 |
+
"learning_rate": 7.2e-06,
|
| 36 |
+
"loss": 2.9314,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.04692632566870014,
|
| 41 |
+
"grad_norm": 0.5952326059341431,
|
| 42 |
+
"learning_rate": 9.2e-06,
|
| 43 |
+
"loss": 2.8987,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.05631159080244017,
|
| 48 |
+
"grad_norm": 0.570318341255188,
|
| 49 |
+
"learning_rate": 1.1200000000000001e-05,
|
| 50 |
+
"loss": 2.7701,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0656968559361802,
|
| 55 |
+
"grad_norm": 0.5945647358894348,
|
| 56 |
+
"learning_rate": 1.32e-05,
|
| 57 |
+
"loss": 2.7628,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.07508212106992022,
|
| 62 |
+
"grad_norm": 0.5424015522003174,
|
| 63 |
+
"learning_rate": 1.52e-05,
|
| 64 |
+
"loss": 2.8025,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.08446738620366025,
|
| 69 |
+
"grad_norm": 0.5131893157958984,
|
| 70 |
+
"learning_rate": 1.7199999999999998e-05,
|
| 71 |
+
"loss": 2.6489,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.09385265133740028,
|
| 76 |
+
"grad_norm": 0.550221860408783,
|
| 77 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 78 |
+
"loss": 2.6995,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.09385265133740028,
|
| 83 |
+
"eval_loss": 2.63047194480896,
|
| 84 |
+
"eval_runtime": 937.5041,
|
| 85 |
+
"eval_samples_per_second": 1.01,
|
| 86 |
+
"eval_steps_per_second": 0.506,
|
| 87 |
+
"step": 100
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.1032379164711403,
|
| 91 |
+
"grad_norm": 0.42704904079437256,
|
| 92 |
+
"learning_rate": 2.12e-05,
|
| 93 |
+
"loss": 2.6395,
|
| 94 |
+
"step": 110
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.11262318160488034,
|
| 98 |
+
"grad_norm": 0.44245535135269165,
|
| 99 |
+
"learning_rate": 2.32e-05,
|
| 100 |
+
"loss": 2.6048,
|
| 101 |
+
"step": 120
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.12200844673862037,
|
| 105 |
+
"grad_norm": 0.3578210473060608,
|
| 106 |
+
"learning_rate": 2.5200000000000003e-05,
|
| 107 |
+
"loss": 2.5695,
|
| 108 |
+
"step": 130
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.1313937118723604,
|
| 112 |
+
"grad_norm": 0.5043906569480896,
|
| 113 |
+
"learning_rate": 2.7200000000000004e-05,
|
| 114 |
+
"loss": 2.5239,
|
| 115 |
+
"step": 140
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.14077897700610043,
|
| 119 |
+
"grad_norm": 0.4712482988834381,
|
| 120 |
+
"learning_rate": 2.9199999999999998e-05,
|
| 121 |
+
"loss": 2.4199,
|
| 122 |
+
"step": 150
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.15016424213984045,
|
| 126 |
+
"grad_norm": 0.5192570090293884,
|
| 127 |
+
"learning_rate": 3.12e-05,
|
| 128 |
+
"loss": 2.5526,
|
| 129 |
+
"step": 160
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.15954950727358047,
|
| 133 |
+
"grad_norm": 0.48030856251716614,
|
| 134 |
+
"learning_rate": 3.32e-05,
|
| 135 |
+
"loss": 2.4876,
|
| 136 |
+
"step": 170
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.1689347724073205,
|
| 140 |
+
"grad_norm": 0.9590848088264465,
|
| 141 |
+
"learning_rate": 3.52e-05,
|
| 142 |
+
"loss": 2.439,
|
| 143 |
+
"step": 180
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.17832003754106054,
|
| 147 |
+
"grad_norm": 0.5423749685287476,
|
| 148 |
+
"learning_rate": 3.72e-05,
|
| 149 |
+
"loss": 2.3547,
|
| 150 |
+
"step": 190
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.18770530267480057,
|
| 154 |
+
"grad_norm": 0.4835156798362732,
|
| 155 |
+
"learning_rate": 3.9200000000000004e-05,
|
| 156 |
+
"loss": 2.4199,
|
| 157 |
+
"step": 200
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.18770530267480057,
|
| 161 |
+
"eval_loss": 2.3978819847106934,
|
| 162 |
+
"eval_runtime": 937.0546,
|
| 163 |
+
"eval_samples_per_second": 1.011,
|
| 164 |
+
"eval_steps_per_second": 0.506,
|
| 165 |
+
"step": 200
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.1970905678085406,
|
| 169 |
+
"grad_norm": 0.7263341546058655,
|
| 170 |
+
"learning_rate": 4.12e-05,
|
| 171 |
+
"loss": 2.4937,
|
| 172 |
+
"step": 210
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.2064758329422806,
|
| 176 |
+
"grad_norm": 0.6794207096099854,
|
| 177 |
+
"learning_rate": 4.32e-05,
|
| 178 |
+
"loss": 2.3535,
|
| 179 |
+
"step": 220
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.21586109807602064,
|
| 183 |
+
"grad_norm": 0.6615617275238037,
|
| 184 |
+
"learning_rate": 4.52e-05,
|
| 185 |
+
"loss": 2.3753,
|
| 186 |
+
"step": 230
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.22524636320976069,
|
| 190 |
+
"grad_norm": 0.7630943059921265,
|
| 191 |
+
"learning_rate": 4.72e-05,
|
| 192 |
+
"loss": 2.3971,
|
| 193 |
+
"step": 240
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.2346316283435007,
|
| 197 |
+
"grad_norm": 0.7536003589630127,
|
| 198 |
+
"learning_rate": 4.92e-05,
|
| 199 |
+
"loss": 2.3282,
|
| 200 |
+
"step": 250
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.24401689347724073,
|
| 204 |
+
"grad_norm": 1.0338555574417114,
|
| 205 |
+
"learning_rate": 5.1200000000000004e-05,
|
| 206 |
+
"loss": 2.333,
|
| 207 |
+
"step": 260
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.25340215861098075,
|
| 211 |
+
"grad_norm": 0.7328284382820129,
|
| 212 |
+
"learning_rate": 5.3200000000000006e-05,
|
| 213 |
+
"loss": 2.3966,
|
| 214 |
+
"step": 270
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.2627874237447208,
|
| 218 |
+
"grad_norm": 0.8608214855194092,
|
| 219 |
+
"learning_rate": 5.520000000000001e-05,
|
| 220 |
+
"loss": 2.2628,
|
| 221 |
+
"step": 280
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.2721726888784608,
|
| 225 |
+
"grad_norm": 0.8262203931808472,
|
| 226 |
+
"learning_rate": 5.72e-05,
|
| 227 |
+
"loss": 2.3092,
|
| 228 |
+
"step": 290
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.28155795401220085,
|
| 232 |
+
"grad_norm": 0.8401615619659424,
|
| 233 |
+
"learning_rate": 5.92e-05,
|
| 234 |
+
"loss": 2.2722,
|
| 235 |
+
"step": 300
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.28155795401220085,
|
| 239 |
+
"eval_loss": 2.217956066131592,
|
| 240 |
+
"eval_runtime": 936.7717,
|
| 241 |
+
"eval_samples_per_second": 1.011,
|
| 242 |
+
"eval_steps_per_second": 0.506,
|
| 243 |
+
"step": 300
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.29094321914594085,
|
| 247 |
+
"grad_norm": 0.7364845275878906,
|
| 248 |
+
"learning_rate": 6.12e-05,
|
| 249 |
+
"loss": 2.2179,
|
| 250 |
+
"step": 310
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.3003284842796809,
|
| 254 |
+
"grad_norm": 0.8354003429412842,
|
| 255 |
+
"learning_rate": 6.3e-05,
|
| 256 |
+
"loss": 2.3654,
|
| 257 |
+
"step": 320
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.30971374941342095,
|
| 261 |
+
"grad_norm": 1.0776605606079102,
|
| 262 |
+
"learning_rate": 6.500000000000001e-05,
|
| 263 |
+
"loss": 2.2143,
|
| 264 |
+
"step": 330
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.31909901454716094,
|
| 268 |
+
"grad_norm": 0.8842081427574158,
|
| 269 |
+
"learning_rate": 6.7e-05,
|
| 270 |
+
"loss": 2.2705,
|
| 271 |
+
"step": 340
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 0.328484279680901,
|
| 275 |
+
"grad_norm": 1.1752427816390991,
|
| 276 |
+
"learning_rate": 6.9e-05,
|
| 277 |
+
"loss": 2.2334,
|
| 278 |
+
"step": 350
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.337869544814641,
|
| 282 |
+
"grad_norm": 0.7227160334587097,
|
| 283 |
+
"learning_rate": 7.1e-05,
|
| 284 |
+
"loss": 2.228,
|
| 285 |
+
"step": 360
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.34725480994838104,
|
| 289 |
+
"grad_norm": 1.1053409576416016,
|
| 290 |
+
"learning_rate": 7.3e-05,
|
| 291 |
+
"loss": 2.2071,
|
| 292 |
+
"step": 370
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 0.3566400750821211,
|
| 296 |
+
"grad_norm": 0.9307537078857422,
|
| 297 |
+
"learning_rate": 7.500000000000001e-05,
|
| 298 |
+
"loss": 2.2334,
|
| 299 |
+
"step": 380
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 0.3660253402158611,
|
| 303 |
+
"grad_norm": 0.9264342188835144,
|
| 304 |
+
"learning_rate": 7.7e-05,
|
| 305 |
+
"loss": 2.2815,
|
| 306 |
+
"step": 390
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.37541060534960113,
|
| 310 |
+
"grad_norm": 1.0281509160995483,
|
| 311 |
+
"learning_rate": 7.900000000000001e-05,
|
| 312 |
+
"loss": 2.0762,
|
| 313 |
+
"step": 400
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.37541060534960113,
|
| 317 |
+
"eval_loss": 2.12506103515625,
|
| 318 |
+
"eval_runtime": 936.7185,
|
| 319 |
+
"eval_samples_per_second": 1.011,
|
| 320 |
+
"eval_steps_per_second": 0.506,
|
| 321 |
+
"step": 400
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.38479587048334113,
|
| 325 |
+
"grad_norm": 0.979566216468811,
|
| 326 |
+
"learning_rate": 8.1e-05,
|
| 327 |
+
"loss": 2.1801,
|
| 328 |
+
"step": 410
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 0.3941811356170812,
|
| 332 |
+
"grad_norm": 1.4208861589431763,
|
| 333 |
+
"learning_rate": 8.3e-05,
|
| 334 |
+
"loss": 2.263,
|
| 335 |
+
"step": 420
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 0.40356640075082123,
|
| 339 |
+
"grad_norm": 1.0267932415008545,
|
| 340 |
+
"learning_rate": 8.5e-05,
|
| 341 |
+
"loss": 2.1086,
|
| 342 |
+
"step": 430
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.4129516658845612,
|
| 346 |
+
"grad_norm": 1.010489583015442,
|
| 347 |
+
"learning_rate": 8.7e-05,
|
| 348 |
+
"loss": 2.1647,
|
| 349 |
+
"step": 440
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.4223369310183013,
|
| 353 |
+
"grad_norm": 0.784968912601471,
|
| 354 |
+
"learning_rate": 8.900000000000001e-05,
|
| 355 |
+
"loss": 2.2024,
|
| 356 |
+
"step": 450
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.43172219615204127,
|
| 360 |
+
"grad_norm": 1.1498106718063354,
|
| 361 |
+
"learning_rate": 9.1e-05,
|
| 362 |
+
"loss": 2.0905,
|
| 363 |
+
"step": 460
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.4411074612857813,
|
| 367 |
+
"grad_norm": 0.7570444345474243,
|
| 368 |
+
"learning_rate": 9.300000000000001e-05,
|
| 369 |
+
"loss": 2.0965,
|
| 370 |
+
"step": 470
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.45049272641952137,
|
| 374 |
+
"grad_norm": 1.162133812904358,
|
| 375 |
+
"learning_rate": 9.5e-05,
|
| 376 |
+
"loss": 2.1354,
|
| 377 |
+
"step": 480
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.45987799155326137,
|
| 381 |
+
"grad_norm": 0.963750422000885,
|
| 382 |
+
"learning_rate": 9.7e-05,
|
| 383 |
+
"loss": 2.1286,
|
| 384 |
+
"step": 490
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.4692632566870014,
|
| 388 |
+
"grad_norm": 0.9272373914718628,
|
| 389 |
+
"learning_rate": 9.900000000000001e-05,
|
| 390 |
+
"loss": 1.9652,
|
| 391 |
+
"step": 500
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.4692632566870014,
|
| 395 |
+
"eval_loss": 2.085766077041626,
|
| 396 |
+
"eval_runtime": 937.6717,
|
| 397 |
+
"eval_samples_per_second": 1.01,
|
| 398 |
+
"eval_steps_per_second": 0.506,
|
| 399 |
+
"step": 500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.4786485218207414,
|
| 403 |
+
"grad_norm": 0.7409124374389648,
|
| 404 |
+
"learning_rate": 9.999915070025401e-05,
|
| 405 |
+
"loss": 2.099,
|
| 406 |
+
"step": 510
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.48803378695448146,
|
| 410 |
+
"grad_norm": 0.8032079339027405,
|
| 411 |
+
"learning_rate": 9.999235647539953e-05,
|
| 412 |
+
"loss": 2.1874,
|
| 413 |
+
"step": 520
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.4974190520882215,
|
| 417 |
+
"grad_norm": 1.00968599319458,
|
| 418 |
+
"learning_rate": 9.997876894893606e-05,
|
| 419 |
+
"loss": 2.2589,
|
| 420 |
+
"step": 530
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.5068043172219615,
|
| 424 |
+
"grad_norm": 0.7752834558486938,
|
| 425 |
+
"learning_rate": 9.995838996722914e-05,
|
| 426 |
+
"loss": 2.1808,
|
| 427 |
+
"step": 540
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.5161895823557016,
|
| 431 |
+
"grad_norm": 0.8999218940734863,
|
| 432 |
+
"learning_rate": 9.993122229951354e-05,
|
| 433 |
+
"loss": 2.2034,
|
| 434 |
+
"step": 550
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.5255748474894416,
|
| 438 |
+
"grad_norm": 0.8991349339485168,
|
| 439 |
+
"learning_rate": 9.989726963751682e-05,
|
| 440 |
+
"loss": 2.1284,
|
| 441 |
+
"step": 560
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.5349601126231815,
|
| 445 |
+
"grad_norm": 0.9258519411087036,
|
| 446 |
+
"learning_rate": 9.985653659495773e-05,
|
| 447 |
+
"loss": 2.0642,
|
| 448 |
+
"step": 570
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.5443453777569216,
|
| 452 |
+
"grad_norm": 0.7985184788703918,
|
| 453 |
+
"learning_rate": 9.980902870691931e-05,
|
| 454 |
+
"loss": 1.9404,
|
| 455 |
+
"step": 580
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 0.5537306428906617,
|
| 459 |
+
"grad_norm": 0.9277619123458862,
|
| 460 |
+
"learning_rate": 9.975475242909667e-05,
|
| 461 |
+
"loss": 1.9928,
|
| 462 |
+
"step": 590
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 0.5631159080244017,
|
| 466 |
+
"grad_norm": 0.8406923413276672,
|
| 467 |
+
"learning_rate": 9.969371513691982e-05,
|
| 468 |
+
"loss": 2.1893,
|
| 469 |
+
"step": 600
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 0.5631159080244017,
|
| 473 |
+
"eval_loss": 2.0628976821899414,
|
| 474 |
+
"eval_runtime": 937.447,
|
| 475 |
+
"eval_samples_per_second": 1.01,
|
| 476 |
+
"eval_steps_per_second": 0.506,
|
| 477 |
+
"step": 600
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.5725011731581418,
|
| 481 |
+
"grad_norm": 0.678254246711731,
|
| 482 |
+
"learning_rate": 9.962592512455138e-05,
|
| 483 |
+
"loss": 2.1181,
|
| 484 |
+
"step": 610
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.5818864382918817,
|
| 488 |
+
"grad_norm": 0.7161825895309448,
|
| 489 |
+
"learning_rate": 9.955139160375959e-05,
|
| 490 |
+
"loss": 2.1328,
|
| 491 |
+
"step": 620
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.5912717034256217,
|
| 495 |
+
"grad_norm": 0.7554607391357422,
|
| 496 |
+
"learning_rate": 9.947012470266645e-05,
|
| 497 |
+
"loss": 2.0865,
|
| 498 |
+
"step": 630
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.6006569685593618,
|
| 502 |
+
"grad_norm": 0.9590722322463989,
|
| 503 |
+
"learning_rate": 9.938213546437154e-05,
|
| 504 |
+
"loss": 2.1012,
|
| 505 |
+
"step": 640
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.6100422336931018,
|
| 509 |
+
"grad_norm": 0.8377043008804321,
|
| 510 |
+
"learning_rate": 9.928743584545132e-05,
|
| 511 |
+
"loss": 2.1155,
|
| 512 |
+
"step": 650
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.6194274988268419,
|
| 516 |
+
"grad_norm": 0.7528682947158813,
|
| 517 |
+
"learning_rate": 9.91860387143345e-05,
|
| 518 |
+
"loss": 2.2455,
|
| 519 |
+
"step": 660
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.6288127639605818,
|
| 523 |
+
"grad_norm": 0.7164850234985352,
|
| 524 |
+
"learning_rate": 9.907795784955327e-05,
|
| 525 |
+
"loss": 2.155,
|
| 526 |
+
"step": 670
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.6381980290943219,
|
| 530 |
+
"grad_norm": 0.7665808200836182,
|
| 531 |
+
"learning_rate": 9.896320793787106e-05,
|
| 532 |
+
"loss": 2.0722,
|
| 533 |
+
"step": 680
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.6475832942280619,
|
| 537 |
+
"grad_norm": 0.8012193441390991,
|
| 538 |
+
"learning_rate": 9.884180457228678e-05,
|
| 539 |
+
"loss": 2.1045,
|
| 540 |
+
"step": 690
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.656968559361802,
|
| 544 |
+
"grad_norm": 0.8847366571426392,
|
| 545 |
+
"learning_rate": 9.871376424991589e-05,
|
| 546 |
+
"loss": 2.0153,
|
| 547 |
+
"step": 700
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.656968559361802,
|
| 551 |
+
"eval_loss": 2.047290802001953,
|
| 552 |
+
"eval_runtime": 937.2712,
|
| 553 |
+
"eval_samples_per_second": 1.01,
|
| 554 |
+
"eval_steps_per_second": 0.506,
|
| 555 |
+
"step": 700
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.666353824495542,
|
| 559 |
+
"grad_norm": 0.7482985854148865,
|
| 560 |
+
"learning_rate": 9.85791043697488e-05,
|
| 561 |
+
"loss": 1.981,
|
| 562 |
+
"step": 710
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.675739089629282,
|
| 566 |
+
"grad_norm": 0.776764988899231,
|
| 567 |
+
"learning_rate": 9.843784323028638e-05,
|
| 568 |
+
"loss": 2.062,
|
| 569 |
+
"step": 720
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.685124354763022,
|
| 573 |
+
"grad_norm": 0.8061379790306091,
|
| 574 |
+
"learning_rate": 9.82900000270536e-05,
|
| 575 |
+
"loss": 1.9919,
|
| 576 |
+
"step": 730
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.6945096198967621,
|
| 580 |
+
"grad_norm": 0.7650445699691772,
|
| 581 |
+
"learning_rate": 9.813559484999102e-05,
|
| 582 |
+
"loss": 2.0104,
|
| 583 |
+
"step": 740
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.7038948850305021,
|
| 587 |
+
"grad_norm": 1.2102171182632446,
|
| 588 |
+
"learning_rate": 9.797464868072488e-05,
|
| 589 |
+
"loss": 2.0407,
|
| 590 |
+
"step": 750
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.7132801501642422,
|
| 594 |
+
"grad_norm": 0.7476488351821899,
|
| 595 |
+
"learning_rate": 9.780718338971591e-05,
|
| 596 |
+
"loss": 1.8859,
|
| 597 |
+
"step": 760
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.7226654152979821,
|
| 601 |
+
"grad_norm": 0.7931784391403198,
|
| 602 |
+
"learning_rate": 9.763322173328753e-05,
|
| 603 |
+
"loss": 2.1804,
|
| 604 |
+
"step": 770
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.7320506804317222,
|
| 608 |
+
"grad_norm": 1.1580551862716675,
|
| 609 |
+
"learning_rate": 9.745278735053343e-05,
|
| 610 |
+
"loss": 2.1195,
|
| 611 |
+
"step": 780
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.7414359455654622,
|
| 615 |
+
"grad_norm": 0.7719026803970337,
|
| 616 |
+
"learning_rate": 9.726590476010548e-05,
|
| 617 |
+
"loss": 1.97,
|
| 618 |
+
"step": 790
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.7508212106992023,
|
| 622 |
+
"grad_norm": 0.9482008218765259,
|
| 623 |
+
"learning_rate": 9.707259935688187e-05,
|
| 624 |
+
"loss": 1.9911,
|
| 625 |
+
"step": 800
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.7508212106992023,
|
| 629 |
+
"eval_loss": 2.0318124294281006,
|
| 630 |
+
"eval_runtime": 937.1438,
|
| 631 |
+
"eval_samples_per_second": 1.011,
|
| 632 |
+
"eval_steps_per_second": 0.506,
|
| 633 |
+
"step": 800
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.7602064758329423,
|
| 637 |
+
"grad_norm": 0.7328064441680908,
|
| 638 |
+
"learning_rate": 9.687289740851622e-05,
|
| 639 |
+
"loss": 2.1643,
|
| 640 |
+
"step": 810
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.7695917409666823,
|
| 644 |
+
"grad_norm": 0.7916940450668335,
|
| 645 |
+
"learning_rate": 9.666682605186835e-05,
|
| 646 |
+
"loss": 2.166,
|
| 647 |
+
"step": 820
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.7789770061004223,
|
| 651 |
+
"grad_norm": 0.9480199217796326,
|
| 652 |
+
"learning_rate": 9.645441328931654e-05,
|
| 653 |
+
"loss": 2.108,
|
| 654 |
+
"step": 830
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.7883622712341624,
|
| 658 |
+
"grad_norm": 0.8467483520507812,
|
| 659 |
+
"learning_rate": 9.62356879849525e-05,
|
| 660 |
+
"loss": 2.0794,
|
| 661 |
+
"step": 840
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.7977475363679024,
|
| 665 |
+
"grad_norm": 0.6742168664932251,
|
| 666 |
+
"learning_rate": 9.601067986065909e-05,
|
| 667 |
+
"loss": 2.1227,
|
| 668 |
+
"step": 850
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.8071328015016425,
|
| 672 |
+
"grad_norm": 0.7015961408615112,
|
| 673 |
+
"learning_rate": 9.577941949207146e-05,
|
| 674 |
+
"loss": 2.0288,
|
| 675 |
+
"step": 860
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.8165180666353824,
|
| 679 |
+
"grad_norm": 0.8289031386375427,
|
| 680 |
+
"learning_rate": 9.556596544693951e-05,
|
| 681 |
+
"loss": 2.037,
|
| 682 |
+
"step": 870
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.8259033317691225,
|
| 686 |
+
"grad_norm": 0.8550999760627747,
|
| 687 |
+
"learning_rate": 9.53229130894619e-05,
|
| 688 |
+
"loss": 2.1908,
|
| 689 |
+
"step": 880
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.8352885969028625,
|
| 693 |
+
"grad_norm": 0.7619972229003906,
|
| 694 |
+
"learning_rate": 9.50737019461194e-05,
|
| 695 |
+
"loss": 2.1759,
|
| 696 |
+
"step": 890
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.8446738620366026,
|
| 700 |
+
"grad_norm": 0.7303412556648254,
|
| 701 |
+
"learning_rate": 9.481836588141808e-05,
|
| 702 |
+
"loss": 2.1041,
|
| 703 |
+
"step": 900
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.8446738620366026,
|
| 707 |
+
"eval_loss": 2.019794225692749,
|
| 708 |
+
"eval_runtime": 938.1446,
|
| 709 |
+
"eval_samples_per_second": 1.009,
|
| 710 |
+
"eval_steps_per_second": 0.505,
|
| 711 |
+
"step": 900
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.8540591271703426,
|
| 715 |
+
"grad_norm": 0.6646521687507629,
|
| 716 |
+
"learning_rate": 9.455693959216005e-05,
|
| 717 |
+
"loss": 2.0648,
|
| 718 |
+
"step": 910
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.8634443923040825,
|
| 722 |
+
"grad_norm": 0.910926878452301,
|
| 723 |
+
"learning_rate": 9.428945860272858e-05,
|
| 724 |
+
"loss": 2.0945,
|
| 725 |
+
"step": 920
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 0.8728296574378226,
|
| 729 |
+
"grad_norm": 0.7175999283790588,
|
| 730 |
+
"learning_rate": 9.401595926026077e-05,
|
| 731 |
+
"loss": 2.0488,
|
| 732 |
+
"step": 930
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.8822149225715626,
|
| 736 |
+
"grad_norm": 1.090782642364502,
|
| 737 |
+
"learning_rate": 9.373647872970852e-05,
|
| 738 |
+
"loss": 1.9902,
|
| 739 |
+
"step": 940
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 0.8916001877053027,
|
| 743 |
+
"grad_norm": 1.0074050426483154,
|
| 744 |
+
"learning_rate": 9.345105498878826e-05,
|
| 745 |
+
"loss": 1.9974,
|
| 746 |
+
"step": 950
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.9009854528390427,
|
| 750 |
+
"grad_norm": 0.6742076277732849,
|
| 751 |
+
"learning_rate": 9.315972682282031e-05,
|
| 752 |
+
"loss": 2.0359,
|
| 753 |
+
"step": 960
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.9103707179727827,
|
| 757 |
+
"grad_norm": 0.9036829471588135,
|
| 758 |
+
"learning_rate": 9.286253381945837e-05,
|
| 759 |
+
"loss": 2.1047,
|
| 760 |
+
"step": 970
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 0.9197559831065227,
|
| 764 |
+
"grad_norm": 1.02723228931427,
|
| 765 |
+
"learning_rate": 9.255951636331028e-05,
|
| 766 |
+
"loss": 1.9049,
|
| 767 |
+
"step": 980
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.9291412482402628,
|
| 771 |
+
"grad_norm": 0.9634993076324463,
|
| 772 |
+
"learning_rate": 9.225071563045007e-05,
|
| 773 |
+
"loss": 2.045,
|
| 774 |
+
"step": 990
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.9385265133740028,
|
| 778 |
+
"grad_norm": 0.8380990624427795,
|
| 779 |
+
"learning_rate": 9.193617358282277e-05,
|
| 780 |
+
"loss": 2.0488,
|
| 781 |
+
"step": 1000
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 0.9385265133740028,
|
| 785 |
+
"eval_loss": 2.0117270946502686,
|
| 786 |
+
"eval_runtime": 939.385,
|
| 787 |
+
"eval_samples_per_second": 1.008,
|
| 788 |
+
"eval_steps_per_second": 0.505,
|
| 789 |
+
"step": 1000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.9479117785077429,
|
| 793 |
+
"grad_norm": 0.8317145705223083,
|
| 794 |
+
"learning_rate": 9.161593296254235e-05,
|
| 795 |
+
"loss": 2.0196,
|
| 796 |
+
"step": 1010
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 0.9572970436414828,
|
| 800 |
+
"grad_norm": 0.9297407865524292,
|
| 801 |
+
"learning_rate": 9.129003728608367e-05,
|
| 802 |
+
"loss": 2.0798,
|
| 803 |
+
"step": 1020
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.9666823087752229,
|
| 807 |
+
"grad_norm": 0.6938600540161133,
|
| 808 |
+
"learning_rate": 9.095853083836902e-05,
|
| 809 |
+
"loss": 2.1225,
|
| 810 |
+
"step": 1030
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 0.9760675739089629,
|
| 814 |
+
"grad_norm": 0.7786221504211426,
|
| 815 |
+
"learning_rate": 9.062145866675048e-05,
|
| 816 |
+
"loss": 2.0098,
|
| 817 |
+
"step": 1040
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.985452839042703,
|
| 821 |
+
"grad_norm": 0.7615213394165039,
|
| 822 |
+
"learning_rate": 9.027886657488862e-05,
|
| 823 |
+
"loss": 2.1385,
|
| 824 |
+
"step": 1050
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.994838104176443,
|
| 828 |
+
"grad_norm": 0.9867497086524963,
|
| 829 |
+
"learning_rate": 8.993080111652831e-05,
|
| 830 |
+
"loss": 2.1039,
|
| 831 |
+
"step": 1060
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.004223369310183,
|
| 835 |
+
"grad_norm": 0.747900664806366,
|
| 836 |
+
"learning_rate": 8.95773095891727e-05,
|
| 837 |
+
"loss": 2.104,
|
| 838 |
+
"step": 1070
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.013608634443923,
|
| 842 |
+
"grad_norm": 0.6495920419692993,
|
| 843 |
+
"learning_rate": 8.921844002765613e-05,
|
| 844 |
+
"loss": 2.0333,
|
| 845 |
+
"step": 1080
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.022993899577663,
|
| 849 |
+
"grad_norm": 0.8173234462738037,
|
| 850 |
+
"learning_rate": 8.885424119761684e-05,
|
| 851 |
+
"loss": 2.0524,
|
| 852 |
+
"step": 1090
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.0323791647114031,
|
| 856 |
+
"grad_norm": 0.8507694602012634,
|
| 857 |
+
"learning_rate": 8.848476258887031e-05,
|
| 858 |
+
"loss": 1.897,
|
| 859 |
+
"step": 1100
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.0323791647114031,
|
| 863 |
+
"eval_loss": 2.001809597015381,
|
| 864 |
+
"eval_runtime": 940.385,
|
| 865 |
+
"eval_samples_per_second": 1.007,
|
| 866 |
+
"eval_steps_per_second": 0.504,
|
| 867 |
+
"step": 1100
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.0417644298451432,
|
| 871 |
+
"grad_norm": 1.4220669269561768,
|
| 872 |
+
"learning_rate": 8.814775911179585e-05,
|
| 873 |
+
"loss": 1.9925,
|
| 874 |
+
"step": 1110
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.0511496949788832,
|
| 878 |
+
"grad_norm": 0.7997964024543762,
|
| 879 |
+
"learning_rate": 8.776838783200623e-05,
|
| 880 |
+
"loss": 1.9378,
|
| 881 |
+
"step": 1120
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 1.0605349601126233,
|
| 885 |
+
"grad_norm": 2.4897735118865967,
|
| 886 |
+
"learning_rate": 8.738388432665424e-05,
|
| 887 |
+
"loss": 1.855,
|
| 888 |
+
"step": 1130
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 1.069920225246363,
|
| 892 |
+
"grad_norm": 0.8852785229682922,
|
| 893 |
+
"learning_rate": 8.699430084469276e-05,
|
| 894 |
+
"loss": 2.0958,
|
| 895 |
+
"step": 1140
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 1.0793054903801031,
|
| 899 |
+
"grad_norm": 0.698052167892456,
|
| 900 |
+
"learning_rate": 8.65996903253766e-05,
|
| 901 |
+
"loss": 2.1623,
|
| 902 |
+
"step": 1150
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 1.0886907555138432,
|
| 906 |
+
"grad_norm": 0.8183425068855286,
|
| 907 |
+
"learning_rate": 8.620010639106853e-05,
|
| 908 |
+
"loss": 2.0938,
|
| 909 |
+
"step": 1160
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.0980760206475833,
|
| 913 |
+
"grad_norm": 0.8702631592750549,
|
| 914 |
+
"learning_rate": 8.57956033399528e-05,
|
| 915 |
+
"loss": 1.842,
|
| 916 |
+
"step": 1170
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 1.1074612857813233,
|
| 920 |
+
"grad_norm": 0.7535381317138672,
|
| 921 |
+
"learning_rate": 8.538623613865678e-05,
|
| 922 |
+
"loss": 2.069,
|
| 923 |
+
"step": 1180
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.1168465509150634,
|
| 927 |
+
"grad_norm": 0.7525476813316345,
|
| 928 |
+
"learning_rate": 8.497206041478162e-05,
|
| 929 |
+
"loss": 2.0564,
|
| 930 |
+
"step": 1190
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.1262318160488034,
|
| 934 |
+
"grad_norm": 0.80061274766922,
|
| 935 |
+
"learning_rate": 8.455313244934324e-05,
|
| 936 |
+
"loss": 2.0298,
|
| 937 |
+
"step": 1200
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 1.1262318160488034,
|
| 941 |
+
"eval_loss": 1.9951938390731812,
|
| 942 |
+
"eval_runtime": 939.3487,
|
| 943 |
+
"eval_samples_per_second": 1.008,
|
| 944 |
+
"eval_steps_per_second": 0.505,
|
| 945 |
+
"step": 1200
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 1.1356170811825435,
|
| 949 |
+
"grad_norm": 0.7788612246513367,
|
| 950 |
+
"learning_rate": 8.412950916912451e-05,
|
| 951 |
+
"loss": 2.1235,
|
| 952 |
+
"step": 1210
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 1.1450023463162835,
|
| 956 |
+
"grad_norm": 0.7296183705329895,
|
| 957 |
+
"learning_rate": 8.370124813893962e-05,
|
| 958 |
+
"loss": 2.1001,
|
| 959 |
+
"step": 1220
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.1543876114500236,
|
| 963 |
+
"grad_norm": 0.9630563855171204,
|
| 964 |
+
"learning_rate": 8.326840755381176e-05,
|
| 965 |
+
"loss": 2.1847,
|
| 966 |
+
"step": 1230
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 1.1637728765837636,
|
| 970 |
+
"grad_norm": 0.8061946630477905,
|
| 971 |
+
"learning_rate": 8.283104623106525e-05,
|
| 972 |
+
"loss": 2.0888,
|
| 973 |
+
"step": 1240
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 1.1731581417175034,
|
| 977 |
+
"grad_norm": 0.8088260889053345,
|
| 978 |
+
"learning_rate": 8.238922360233297e-05,
|
| 979 |
+
"loss": 1.9784,
|
| 980 |
+
"step": 1250
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 1.1825434068512435,
|
| 984 |
+
"grad_norm": 0.8710635900497437,
|
| 985 |
+
"learning_rate": 8.194299970548045e-05,
|
| 986 |
+
"loss": 1.915,
|
| 987 |
+
"step": 1260
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 1.1919286719849835,
|
| 991 |
+
"grad_norm": 0.6845631003379822,
|
| 992 |
+
"learning_rate": 8.149243517644745e-05,
|
| 993 |
+
"loss": 2.2073,
|
| 994 |
+
"step": 1270
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 1.2013139371187236,
|
| 998 |
+
"grad_norm": 0.697823166847229,
|
| 999 |
+
"learning_rate": 8.103759124100839e-05,
|
| 1000 |
+
"loss": 2.0622,
|
| 1001 |
+
"step": 1280
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.2106992022524636,
|
| 1005 |
+
"grad_norm": 0.7662308216094971,
|
| 1006 |
+
"learning_rate": 8.057852970645254e-05,
|
| 1007 |
+
"loss": 2.0764,
|
| 1008 |
+
"step": 1290
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 1.2200844673862037,
|
| 1012 |
+
"grad_norm": 0.7569838166236877,
|
| 1013 |
+
"learning_rate": 8.011531295318526e-05,
|
| 1014 |
+
"loss": 2.0989,
|
| 1015 |
+
"step": 1300
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 1.2200844673862037,
|
| 1019 |
+
"eval_loss": 1.9889544248580933,
|
| 1020 |
+
"eval_runtime": 940.2035,
|
| 1021 |
+
"eval_samples_per_second": 1.007,
|
| 1022 |
+
"eval_steps_per_second": 0.504,
|
| 1023 |
+
"step": 1300
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 1.2294697325199437,
|
| 1027 |
+
"grad_norm": 0.7303356528282166,
|
| 1028 |
+
"learning_rate": 7.964800392625129e-05,
|
| 1029 |
+
"loss": 1.9281,
|
| 1030 |
+
"step": 1310
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 1.2388549976536838,
|
| 1034 |
+
"grad_norm": 0.8145589232444763,
|
| 1035 |
+
"learning_rate": 7.917666612678138e-05,
|
| 1036 |
+
"loss": 2.0838,
|
| 1037 |
+
"step": 1320
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 1.2482402627874238,
|
| 1041 |
+
"grad_norm": 0.9209080934524536,
|
| 1042 |
+
"learning_rate": 7.870136360336328e-05,
|
| 1043 |
+
"loss": 2.0761,
|
| 1044 |
+
"step": 1330
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.2576255279211637,
|
| 1048 |
+
"grad_norm": 0.8099146485328674,
|
| 1049 |
+
"learning_rate": 7.822216094333847e-05,
|
| 1050 |
+
"loss": 2.098,
|
| 1051 |
+
"step": 1340
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.267010793054904,
|
| 1055 |
+
"grad_norm": 0.7984501719474792,
|
| 1056 |
+
"learning_rate": 7.773912326402543e-05,
|
| 1057 |
+
"loss": 1.8043,
|
| 1058 |
+
"step": 1350
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.2763960581886438,
|
| 1062 |
+
"grad_norm": 0.713470995426178,
|
| 1063 |
+
"learning_rate": 7.72523162038713e-05,
|
| 1064 |
+
"loss": 1.9197,
|
| 1065 |
+
"step": 1360
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.2857813233223838,
|
| 1069 |
+
"grad_norm": 0.8801192045211792,
|
| 1070 |
+
"learning_rate": 7.676180591353219e-05,
|
| 1071 |
+
"loss": 2.1053,
|
| 1072 |
+
"step": 1370
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 1.2951665884561239,
|
| 1076 |
+
"grad_norm": 0.7982367873191833,
|
| 1077 |
+
"learning_rate": 7.626765904688447e-05,
|
| 1078 |
+
"loss": 2.2708,
|
| 1079 |
+
"step": 1380
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.304551853589864,
|
| 1083 |
+
"grad_norm": 0.840467095375061,
|
| 1084 |
+
"learning_rate": 7.576994275196712e-05,
|
| 1085 |
+
"loss": 2.0068,
|
| 1086 |
+
"step": 1390
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.313937118723604,
|
| 1090 |
+
"grad_norm": 0.8295313715934753,
|
| 1091 |
+
"learning_rate": 7.526872466185742e-05,
|
| 1092 |
+
"loss": 1.8695,
|
| 1093 |
+
"step": 1400
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 1.313937118723604,
|
| 1097 |
+
"eval_loss": 1.9837820529937744,
|
| 1098 |
+
"eval_runtime": 938.2236,
|
| 1099 |
+
"eval_samples_per_second": 1.009,
|
| 1100 |
+
"eval_steps_per_second": 0.505,
|
| 1101 |
+
"step": 1400
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.323322383857344,
|
| 1105 |
+
"grad_norm": 0.7913984060287476,
|
| 1106 |
+
"learning_rate": 7.476407288548036e-05,
|
| 1107 |
+
"loss": 1.9639,
|
| 1108 |
+
"step": 1410
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 1.332707648991084,
|
| 1112 |
+
"grad_norm": 0.819983720779419,
|
| 1113 |
+
"learning_rate": 7.425605599835361e-05,
|
| 1114 |
+
"loss": 2.0229,
|
| 1115 |
+
"step": 1420
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 1.342092914124824,
|
| 1119 |
+
"grad_norm": 0.8383524417877197,
|
| 1120 |
+
"learning_rate": 7.374474303326896e-05,
|
| 1121 |
+
"loss": 1.9001,
|
| 1122 |
+
"step": 1430
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 1.3514781792585642,
|
| 1126 |
+
"grad_norm": 1.1012581586837769,
|
| 1127 |
+
"learning_rate": 7.323020347091177e-05,
|
| 1128 |
+
"loss": 1.9938,
|
| 1129 |
+
"step": 1440
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 1.360863444392304,
|
| 1133 |
+
"grad_norm": 0.8445969223976135,
|
| 1134 |
+
"learning_rate": 7.271250723041932e-05,
|
| 1135 |
+
"loss": 2.0726,
|
| 1136 |
+
"step": 1450
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.370248709526044,
|
| 1140 |
+
"grad_norm": 0.8882044553756714,
|
| 1141 |
+
"learning_rate": 7.21917246598798e-05,
|
| 1142 |
+
"loss": 2.0695,
|
| 1143 |
+
"step": 1460
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.379633974659784,
|
| 1147 |
+
"grad_norm": 0.7796063423156738,
|
| 1148 |
+
"learning_rate": 7.1667926526773e-05,
|
| 1149 |
+
"loss": 2.0607,
|
| 1150 |
+
"step": 1470
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.3890192397935242,
|
| 1154 |
+
"grad_norm": 0.7294184565544128,
|
| 1155 |
+
"learning_rate": 7.114118400835382e-05,
|
| 1156 |
+
"loss": 2.1345,
|
| 1157 |
+
"step": 1480
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.3984045049272642,
|
| 1161 |
+
"grad_norm": 0.7649526000022888,
|
| 1162 |
+
"learning_rate": 7.061156868198028e-05,
|
| 1163 |
+
"loss": 1.999,
|
| 1164 |
+
"step": 1490
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.4077897700610043,
|
| 1168 |
+
"grad_norm": 0.8375232219696045,
|
| 1169 |
+
"learning_rate": 7.007915251538708e-05,
|
| 1170 |
+
"loss": 2.1573,
|
| 1171 |
+
"step": 1500
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 1.4077897700610043,
|
| 1175 |
+
"eval_loss": 1.976365566253662,
|
| 1176 |
+
"eval_runtime": 937.4612,
|
| 1177 |
+
"eval_samples_per_second": 1.01,
|
| 1178 |
+
"eval_steps_per_second": 0.506,
|
| 1179 |
+
"step": 1500
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.4171750351947443,
|
| 1183 |
+
"grad_norm": 0.7321649193763733,
|
| 1184 |
+
"learning_rate": 6.954400785690622e-05,
|
| 1185 |
+
"loss": 2.0845,
|
| 1186 |
+
"step": 1510
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 1.4265603003284844,
|
| 1190 |
+
"grad_norm": 0.778896152973175,
|
| 1191 |
+
"learning_rate": 6.900620742563562e-05,
|
| 1192 |
+
"loss": 1.9401,
|
| 1193 |
+
"step": 1520
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.4359455654622244,
|
| 1197 |
+
"grad_norm": 0.7842182517051697,
|
| 1198 |
+
"learning_rate": 6.846582430155783e-05,
|
| 1199 |
+
"loss": 1.8992,
|
| 1200 |
+
"step": 1530
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.4453308305959642,
|
| 1204 |
+
"grad_norm": 0.6991093754768372,
|
| 1205 |
+
"learning_rate": 6.792293191560914e-05,
|
| 1206 |
+
"loss": 2.0625,
|
| 1207 |
+
"step": 1540
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 1.4547160957297043,
|
| 1211 |
+
"grad_norm": 0.9950138330459595,
|
| 1212 |
+
"learning_rate": 6.737760403970152e-05,
|
| 1213 |
+
"loss": 2.0905,
|
| 1214 |
+
"step": 1550
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 1.4641013608634443,
|
| 1218 |
+
"grad_norm": 0.6939354538917542,
|
| 1219 |
+
"learning_rate": 6.682991477669781e-05,
|
| 1220 |
+
"loss": 2.2633,
|
| 1221 |
+
"step": 1560
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.4734866259971844,
|
| 1225 |
+
"grad_norm": 0.842707633972168,
|
| 1226 |
+
"learning_rate": 6.627993855034228e-05,
|
| 1227 |
+
"loss": 1.8811,
|
| 1228 |
+
"step": 1570
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 1.4828718911309244,
|
| 1232 |
+
"grad_norm": 0.8008860945701599,
|
| 1233 |
+
"learning_rate": 6.572775009514725e-05,
|
| 1234 |
+
"loss": 1.8528,
|
| 1235 |
+
"step": 1580
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 1.4922571562646645,
|
| 1239 |
+
"grad_norm": 0.7409046292304993,
|
| 1240 |
+
"learning_rate": 6.517342444623784e-05,
|
| 1241 |
+
"loss": 1.9773,
|
| 1242 |
+
"step": 1590
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.5016424213984045,
|
| 1246 |
+
"grad_norm": 0.7854930758476257,
|
| 1247 |
+
"learning_rate": 6.461703692915553e-05,
|
| 1248 |
+
"loss": 2.0183,
|
| 1249 |
+
"step": 1600
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 1.5016424213984045,
|
| 1253 |
+
"eval_loss": 1.9713027477264404,
|
| 1254 |
+
"eval_runtime": 938.8142,
|
| 1255 |
+
"eval_samples_per_second": 1.009,
|
| 1256 |
+
"eval_steps_per_second": 0.505,
|
| 1257 |
+
"step": 1600
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.5110276865321446,
|
| 1261 |
+
"grad_norm": 0.8054217100143433,
|
| 1262 |
+
"learning_rate": 6.405866314962252e-05,
|
| 1263 |
+
"loss": 2.1303,
|
| 1264 |
+
"step": 1610
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 1.5204129516658846,
|
| 1268 |
+
"grad_norm": 0.7017131447792053,
|
| 1269 |
+
"learning_rate": 6.349837898326784e-05,
|
| 1270 |
+
"loss": 2.0846,
|
| 1271 |
+
"step": 1620
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 1.5297982167996245,
|
| 1275 |
+
"grad_norm": 0.8393527865409851,
|
| 1276 |
+
"learning_rate": 6.293626056531693e-05,
|
| 1277 |
+
"loss": 1.8327,
|
| 1278 |
+
"step": 1630
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 1.5391834819333647,
|
| 1282 |
+
"grad_norm": 0.8798466920852661,
|
| 1283 |
+
"learning_rate": 6.237238428024572e-05,
|
| 1284 |
+
"loss": 1.8657,
|
| 1285 |
+
"step": 1640
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 1.5485687470671046,
|
| 1289 |
+
"grad_norm": 0.7530277371406555,
|
| 1290 |
+
"learning_rate": 6.180682675140121e-05,
|
| 1291 |
+
"loss": 2.245,
|
| 1292 |
+
"step": 1650
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 1.5579540122008448,
|
| 1296 |
+
"grad_norm": 0.7642443776130676,
|
| 1297 |
+
"learning_rate": 6.123966483058916e-05,
|
| 1298 |
+
"loss": 1.9058,
|
| 1299 |
+
"step": 1660
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.5673392773345847,
|
| 1303 |
+
"grad_norm": 0.7459161281585693,
|
| 1304 |
+
"learning_rate": 6.067097558763106e-05,
|
| 1305 |
+
"loss": 1.9482,
|
| 1306 |
+
"step": 1670
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 1.5767245424683247,
|
| 1310 |
+
"grad_norm": 0.7460825443267822,
|
| 1311 |
+
"learning_rate": 6.0100836299891314e-05,
|
| 1312 |
+
"loss": 2.127,
|
| 1313 |
+
"step": 1680
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 1.5861098076020648,
|
| 1317 |
+
"grad_norm": 0.710259735584259,
|
| 1318 |
+
"learning_rate": 5.9529324441776314e-05,
|
| 1319 |
+
"loss": 2.1407,
|
| 1320 |
+
"step": 1690
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 1.5954950727358048,
|
| 1324 |
+
"grad_norm": 0.7227075695991516,
|
| 1325 |
+
"learning_rate": 5.8956517674206605e-05,
|
| 1326 |
+
"loss": 1.9229,
|
| 1327 |
+
"step": 1700
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 1.5954950727358048,
|
| 1331 |
+
"eval_loss": 1.967227816581726,
|
| 1332 |
+
"eval_runtime": 937.7597,
|
| 1333 |
+
"eval_samples_per_second": 1.01,
|
| 1334 |
+
"eval_steps_per_second": 0.505,
|
| 1335 |
+
"step": 1700
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 1.6048803378695449,
|
| 1339 |
+
"grad_norm": 1.0246224403381348,
|
| 1340 |
+
"learning_rate": 5.838249383406387e-05,
|
| 1341 |
+
"loss": 2.0563,
|
| 1342 |
+
"step": 1710
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 1.6142656030032847,
|
| 1346 |
+
"grad_norm": 0.8386335968971252,
|
| 1347 |
+
"learning_rate": 5.780733092361388e-05,
|
| 1348 |
+
"loss": 1.8553,
|
| 1349 |
+
"step": 1720
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 1.623650868137025,
|
| 1353 |
+
"grad_norm": 0.7936443090438843,
|
| 1354 |
+
"learning_rate": 5.723110709990707e-05,
|
| 1355 |
+
"loss": 2.1631,
|
| 1356 |
+
"step": 1730
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 1.6330361332707648,
|
| 1360 |
+
"grad_norm": 0.7047923803329468,
|
| 1361 |
+
"learning_rate": 5.6653900664157934e-05,
|
| 1362 |
+
"loss": 1.9989,
|
| 1363 |
+
"step": 1740
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 1.642421398404505,
|
| 1367 |
+
"grad_norm": 0.8624520897865295,
|
| 1368 |
+
"learning_rate": 5.6075790051105023e-05,
|
| 1369 |
+
"loss": 2.0198,
|
| 1370 |
+
"step": 1750
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 1.651806663538245,
|
| 1374 |
+
"grad_norm": 0.8698344826698303,
|
| 1375 |
+
"learning_rate": 5.5496853818352614e-05,
|
| 1376 |
+
"loss": 2.1045,
|
| 1377 |
+
"step": 1760
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.661191928671985,
|
| 1381 |
+
"grad_norm": 0.78273606300354,
|
| 1382 |
+
"learning_rate": 5.491717063569582e-05,
|
| 1383 |
+
"loss": 1.9399,
|
| 1384 |
+
"step": 1770
|
| 1385 |
+
},
|
| 1386 |
+
{
|
| 1387 |
+
"epoch": 1.670577193805725,
|
| 1388 |
+
"grad_norm": 0.7217704057693481,
|
| 1389 |
+
"learning_rate": 5.433681927443043e-05,
|
| 1390 |
+
"loss": 2.0161,
|
| 1391 |
+
"step": 1780
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 1.679962458939465,
|
| 1395 |
+
"grad_norm": 0.7136641144752502,
|
| 1396 |
+
"learning_rate": 5.375587859664885e-05,
|
| 1397 |
+
"loss": 2.1437,
|
| 1398 |
+
"step": 1790
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 1.689347724073205,
|
| 1402 |
+
"grad_norm": 0.7752694487571716,
|
| 1403 |
+
"learning_rate": 5.317442754452379e-05,
|
| 1404 |
+
"loss": 1.9732,
|
| 1405 |
+
"step": 1800
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 1.689347724073205,
|
| 1409 |
+
"eval_loss": 1.9616819620132446,
|
| 1410 |
+
"eval_runtime": 938.0777,
|
| 1411 |
+
"eval_samples_per_second": 1.01,
|
| 1412 |
+
"eval_steps_per_second": 0.505,
|
| 1413 |
+
"step": 1800
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 1.698732989206945,
|
| 1417 |
+
"grad_norm": 0.7387381196022034,
|
| 1418 |
+
"learning_rate": 5.2592545129581185e-05,
|
| 1419 |
+
"loss": 1.8547,
|
| 1420 |
+
"step": 1810
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 1.7081182543406852,
|
| 1424 |
+
"grad_norm": 0.9277381300926208,
|
| 1425 |
+
"learning_rate": 5.2010310421963415e-05,
|
| 1426 |
+
"loss": 1.8679,
|
| 1427 |
+
"step": 1820
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 1.717503519474425,
|
| 1431 |
+
"grad_norm": 0.7474396824836731,
|
| 1432 |
+
"learning_rate": 5.142780253968481e-05,
|
| 1433 |
+
"loss": 2.1186,
|
| 1434 |
+
"step": 1830
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 1.7268887846081653,
|
| 1438 |
+
"grad_norm": 0.8091953992843628,
|
| 1439 |
+
"learning_rate": 5.084510063788056e-05,
|
| 1440 |
+
"loss": 2.0762,
|
| 1441 |
+
"step": 1840
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 1.7362740497419051,
|
| 1445 |
+
"grad_norm": 0.7326928973197937,
|
| 1446 |
+
"learning_rate": 5.02622838980505e-05,
|
| 1447 |
+
"loss": 1.9626,
|
| 1448 |
+
"step": 1850
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 1.7456593148756452,
|
| 1452 |
+
"grad_norm": 0.6803217530250549,
|
| 1453 |
+
"learning_rate": 4.967943151729945e-05,
|
| 1454 |
+
"loss": 1.8606,
|
| 1455 |
+
"step": 1860
|
| 1456 |
+
},
|
| 1457 |
+
{
|
| 1458 |
+
"epoch": 1.7550445800093852,
|
| 1459 |
+
"grad_norm": 0.7370252013206482,
|
| 1460 |
+
"learning_rate": 4.9096622697575394e-05,
|
| 1461 |
+
"loss": 1.9649,
|
| 1462 |
+
"step": 1870
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 1.7644298451431253,
|
| 1466 |
+
"grad_norm": 0.7405309677124023,
|
| 1467 |
+
"learning_rate": 4.851393663490689e-05,
|
| 1468 |
+
"loss": 2.0119,
|
| 1469 |
+
"step": 1880
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.7738151102768653,
|
| 1473 |
+
"grad_norm": 0.8256239891052246,
|
| 1474 |
+
"learning_rate": 4.793145250864151e-05,
|
| 1475 |
+
"loss": 1.9313,
|
| 1476 |
+
"step": 1890
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 1.7832003754106054,
|
| 1480 |
+
"grad_norm": 0.8547274470329285,
|
| 1481 |
+
"learning_rate": 4.7349249470686266e-05,
|
| 1482 |
+
"loss": 1.6835,
|
| 1483 |
+
"step": 1900
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.7832003754106054,
|
| 1487 |
+
"eval_loss": 1.9573733806610107,
|
| 1488 |
+
"eval_runtime": 937.9887,
|
| 1489 |
+
"eval_samples_per_second": 1.01,
|
| 1490 |
+
"eval_steps_per_second": 0.505,
|
| 1491 |
+
"step": 1900
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 1.7925856405443454,
|
| 1495 |
+
"grad_norm": 0.9643909335136414,
|
| 1496 |
+
"learning_rate": 4.676740663475198e-05,
|
| 1497 |
+
"loss": 1.8968,
|
| 1498 |
+
"step": 1910
|
| 1499 |
+
},
|
| 1500 |
+
{
|
| 1501 |
+
"epoch": 1.8019709056780853,
|
| 1502 |
+
"grad_norm": 0.8356963992118835,
|
| 1503 |
+
"learning_rate": 4.6186003065602827e-05,
|
| 1504 |
+
"loss": 1.9651,
|
| 1505 |
+
"step": 1920
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 1.8113561708118255,
|
| 1509 |
+
"grad_norm": 0.7793695330619812,
|
| 1510 |
+
"learning_rate": 4.560511776831235e-05,
|
| 1511 |
+
"loss": 2.0038,
|
| 1512 |
+
"step": 1930
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 1.8207414359455654,
|
| 1516 |
+
"grad_norm": 0.7343499660491943,
|
| 1517 |
+
"learning_rate": 4.502482967752786e-05,
|
| 1518 |
+
"loss": 1.7593,
|
| 1519 |
+
"step": 1940
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 1.8301267010793056,
|
| 1523 |
+
"grad_norm": 0.7215515971183777,
|
| 1524 |
+
"learning_rate": 4.444521764674411e-05,
|
| 1525 |
+
"loss": 2.0668,
|
| 1526 |
+
"step": 1950
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 1.8395119662130455,
|
| 1530 |
+
"grad_norm": 0.8646712303161621,
|
| 1531 |
+
"learning_rate": 4.3866360437588294e-05,
|
| 1532 |
+
"loss": 1.9422,
|
| 1533 |
+
"step": 1960
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 1.8488972313467855,
|
| 1537 |
+
"grad_norm": 0.8201174736022949,
|
| 1538 |
+
"learning_rate": 4.328833670911724e-05,
|
| 1539 |
+
"loss": 2.2159,
|
| 1540 |
+
"step": 1970
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 1.8582824964805256,
|
| 1544 |
+
"grad_norm": 0.8474441766738892,
|
| 1545 |
+
"learning_rate": 4.2711225007128765e-05,
|
| 1546 |
+
"loss": 2.0485,
|
| 1547 |
+
"step": 1980
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 1.8676677616142656,
|
| 1551 |
+
"grad_norm": 0.679102897644043,
|
| 1552 |
+
"learning_rate": 4.213510375348837e-05,
|
| 1553 |
+
"loss": 1.9853,
|
| 1554 |
+
"step": 1990
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 1.8770530267480057,
|
| 1558 |
+
"grad_norm": 0.7910708785057068,
|
| 1559 |
+
"learning_rate": 4.15600512354726e-05,
|
| 1560 |
+
"loss": 1.9874,
|
| 1561 |
+
"step": 2000
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 1.8770530267480057,
|
| 1565 |
+
"eval_loss": 1.9538992643356323,
|
| 1566 |
+
"eval_runtime": 937.8684,
|
| 1567 |
+
"eval_samples_per_second": 1.01,
|
| 1568 |
+
"eval_steps_per_second": 0.505,
|
| 1569 |
+
"step": 2000
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.8864382918817455,
|
| 1573 |
+
"grad_norm": 0.6856018900871277,
|
| 1574 |
+
"learning_rate": 4.0986145595131055e-05,
|
| 1575 |
+
"loss": 1.9927,
|
| 1576 |
+
"step": 2010
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.8958235570154858,
|
| 1580 |
+
"grad_norm": 0.7743241786956787,
|
| 1581 |
+
"learning_rate": 4.041346481866768e-05,
|
| 1582 |
+
"loss": 1.9437,
|
| 1583 |
+
"step": 2020
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.9052088221492256,
|
| 1587 |
+
"grad_norm": 0.7838016748428345,
|
| 1588 |
+
"learning_rate": 3.9842086725843625e-05,
|
| 1589 |
+
"loss": 1.953,
|
| 1590 |
+
"step": 2030
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.9145940872829659,
|
| 1594 |
+
"grad_norm": 0.7300184965133667,
|
| 1595 |
+
"learning_rate": 3.9272088959402534e-05,
|
| 1596 |
+
"loss": 2.0461,
|
| 1597 |
+
"step": 2040
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.9239793524167057,
|
| 1601 |
+
"grad_norm": 0.8169785141944885,
|
| 1602 |
+
"learning_rate": 3.8703548974519874e-05,
|
| 1603 |
+
"loss": 2.1075,
|
| 1604 |
+
"step": 2050
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.9333646175504458,
|
| 1608 |
+
"grad_norm": 0.724233090877533,
|
| 1609 |
+
"learning_rate": 3.8136544028277894e-05,
|
| 1610 |
+
"loss": 1.897,
|
| 1611 |
+
"step": 2060
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.9427498826841858,
|
| 1615 |
+
"grad_norm": 0.7821764945983887,
|
| 1616 |
+
"learning_rate": 3.757115116916727e-05,
|
| 1617 |
+
"loss": 2.1728,
|
| 1618 |
+
"step": 2070
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.9521351478179259,
|
| 1622 |
+
"grad_norm": 0.7325600981712341,
|
| 1623 |
+
"learning_rate": 3.7007447226617366e-05,
|
| 1624 |
+
"loss": 1.9865,
|
| 1625 |
+
"step": 2080
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.961520412951666,
|
| 1629 |
+
"grad_norm": 0.7352250814437866,
|
| 1630 |
+
"learning_rate": 3.6445508800556036e-05,
|
| 1631 |
+
"loss": 2.0352,
|
| 1632 |
+
"step": 2090
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.970905678085406,
|
| 1636 |
+
"grad_norm": 0.6944911479949951,
|
| 1637 |
+
"learning_rate": 3.5885412251000745e-05,
|
| 1638 |
+
"loss": 1.7607,
|
| 1639 |
+
"step": 2100
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.970905678085406,
|
| 1643 |
+
"eval_loss": 1.9512391090393066,
|
| 1644 |
+
"eval_runtime": 935.8673,
|
| 1645 |
+
"eval_samples_per_second": 1.012,
|
| 1646 |
+
"eval_steps_per_second": 0.506,
|
| 1647 |
+
"step": 2100
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 1.980290943219146,
|
| 1651 |
+
"grad_norm": 1.345037579536438,
|
| 1652 |
+
"learning_rate": 3.532723368768228e-05,
|
| 1653 |
+
"loss": 1.8189,
|
| 1654 |
+
"step": 2110
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 1.9896762083528858,
|
| 1658 |
+
"grad_norm": 0.8876499533653259,
|
| 1659 |
+
"learning_rate": 3.477104895970234e-05,
|
| 1660 |
+
"loss": 2.0414,
|
| 1661 |
+
"step": 2120
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 1.999061473486626,
|
| 1665 |
+
"grad_norm": 0.832295835018158,
|
| 1666 |
+
"learning_rate": 3.4216933645226776e-05,
|
| 1667 |
+
"loss": 1.9307,
|
| 1668 |
+
"step": 2130
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 2.008446738620366,
|
| 1672 |
+
"grad_norm": 0.7709248661994934,
|
| 1673 |
+
"learning_rate": 3.3664963041215406e-05,
|
| 1674 |
+
"loss": 2.0266,
|
| 1675 |
+
"step": 2140
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 2.017832003754106,
|
| 1679 |
+
"grad_norm": 0.8151468634605408,
|
| 1680 |
+
"learning_rate": 3.311521215319021e-05,
|
| 1681 |
+
"loss": 1.9638,
|
| 1682 |
+
"step": 2150
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 2.027217268887846,
|
| 1686 |
+
"grad_norm": 0.731529951095581,
|
| 1687 |
+
"learning_rate": 3.256775568504305e-05,
|
| 1688 |
+
"loss": 1.9209,
|
| 1689 |
+
"step": 2160
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 2.0366025340215863,
|
| 1693 |
+
"grad_norm": 0.7643239498138428,
|
| 1694 |
+
"learning_rate": 3.202266802888439e-05,
|
| 1695 |
+
"loss": 2.0578,
|
| 1696 |
+
"step": 2170
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 2.045987799155326,
|
| 1700 |
+
"grad_norm": 0.7743598222732544,
|
| 1701 |
+
"learning_rate": 3.148002325493445e-05,
|
| 1702 |
+
"loss": 1.9511,
|
| 1703 |
+
"step": 2180
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 2.055373064289066,
|
| 1707 |
+
"grad_norm": 0.7745042443275452,
|
| 1708 |
+
"learning_rate": 3.0939895101457916e-05,
|
| 1709 |
+
"loss": 1.9773,
|
| 1710 |
+
"step": 2190
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 2.0647583294228062,
|
| 1714 |
+
"grad_norm": 0.8307595252990723,
|
| 1715 |
+
"learning_rate": 3.0402356964744027e-05,
|
| 1716 |
+
"loss": 1.9459,
|
| 1717 |
+
"step": 2200
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 2.0647583294228062,
|
| 1721 |
+
"eval_loss": 1.947997808456421,
|
| 1722 |
+
"eval_runtime": 937.6638,
|
| 1723 |
+
"eval_samples_per_second": 1.01,
|
| 1724 |
+
"eval_steps_per_second": 0.506,
|
| 1725 |
+
"step": 2200
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 2.074143594556546,
|
| 1729 |
+
"grad_norm": 0.7969174981117249,
|
| 1730 |
+
"learning_rate": 2.986748188913287e-05,
|
| 1731 |
+
"loss": 1.8188,
|
| 1732 |
+
"step": 2210
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 2.0835288596902863,
|
| 1736 |
+
"grad_norm": 2.2084882259368896,
|
| 1737 |
+
"learning_rate": 2.9335342557089668e-05,
|
| 1738 |
+
"loss": 1.8969,
|
| 1739 |
+
"step": 2220
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 2.092914124824026,
|
| 1743 |
+
"grad_norm": 0.7004146575927734,
|
| 1744 |
+
"learning_rate": 2.8806011279328256e-05,
|
| 1745 |
+
"loss": 1.7638,
|
| 1746 |
+
"step": 2230
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 2.1022993899577664,
|
| 1750 |
+
"grad_norm": 0.6949226260185242,
|
| 1751 |
+
"learning_rate": 2.827955998498482e-05,
|
| 1752 |
+
"loss": 2.1413,
|
| 1753 |
+
"step": 2240
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 2.1116846550915063,
|
| 1757 |
+
"grad_norm": 0.733250081539154,
|
| 1758 |
+
"learning_rate": 2.775606021184396e-05,
|
| 1759 |
+
"loss": 2.0401,
|
| 1760 |
+
"step": 2250
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 2.1210699202252465,
|
| 1764 |
+
"grad_norm": 0.845836341381073,
|
| 1765 |
+
"learning_rate": 2.7235583096617346e-05,
|
| 1766 |
+
"loss": 1.9184,
|
| 1767 |
+
"step": 2260
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 2.1304551853589864,
|
| 1771 |
+
"grad_norm": 0.7574362754821777,
|
| 1772 |
+
"learning_rate": 2.6718199365277397e-05,
|
| 1773 |
+
"loss": 2.0152,
|
| 1774 |
+
"step": 2270
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 2.139840450492726,
|
| 1778 |
+
"grad_norm": 0.9965910911560059,
|
| 1779 |
+
"learning_rate": 2.6203979323446454e-05,
|
| 1780 |
+
"loss": 1.8746,
|
| 1781 |
+
"step": 2280
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 2.1492257156264665,
|
| 1785 |
+
"grad_norm": 0.8181989192962646,
|
| 1786 |
+
"learning_rate": 2.5692992846843206e-05,
|
| 1787 |
+
"loss": 1.8114,
|
| 1788 |
+
"step": 2290
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 2.1586109807602063,
|
| 1792 |
+
"grad_norm": 0.9191744327545166,
|
| 1793 |
+
"learning_rate": 2.5185309371787513e-05,
|
| 1794 |
+
"loss": 1.7611,
|
| 1795 |
+
"step": 2300
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 2.1586109807602063,
|
| 1799 |
+
"eval_loss": 1.9463104009628296,
|
| 1800 |
+
"eval_runtime": 936.8197,
|
| 1801 |
+
"eval_samples_per_second": 1.011,
|
| 1802 |
+
"eval_steps_per_second": 0.506,
|
| 1803 |
+
"step": 2300
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 2.1679962458939466,
|
| 1807 |
+
"grad_norm": 0.7393763065338135,
|
| 1808 |
+
"learning_rate": 2.468099788576482e-05,
|
| 1809 |
+
"loss": 1.9138,
|
| 1810 |
+
"step": 2310
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 2.1773815110276864,
|
| 1814 |
+
"grad_norm": 0.8604278564453125,
|
| 1815 |
+
"learning_rate": 2.418012691805191e-05,
|
| 1816 |
+
"loss": 1.9187,
|
| 1817 |
+
"step": 2320
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 2.1867667761614267,
|
| 1821 |
+
"grad_norm": 0.8256701827049255,
|
| 1822 |
+
"learning_rate": 2.3682764530404365e-05,
|
| 1823 |
+
"loss": 1.9313,
|
| 1824 |
+
"step": 2330
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 2.1961520412951665,
|
| 1828 |
+
"grad_norm": 0.8674435019493103,
|
| 1829 |
+
"learning_rate": 2.3188978307808125e-05,
|
| 1830 |
+
"loss": 2.1127,
|
| 1831 |
+
"step": 2340
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 2.2055373064289068,
|
| 1835 |
+
"grad_norm": 0.8510717153549194,
|
| 1836 |
+
"learning_rate": 2.2698835349295472e-05,
|
| 1837 |
+
"loss": 1.9931,
|
| 1838 |
+
"step": 2350
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 2.2149225715626466,
|
| 1842 |
+
"grad_norm": 0.8951923847198486,
|
| 1843 |
+
"learning_rate": 2.2212402258827115e-05,
|
| 1844 |
+
"loss": 1.8811,
|
| 1845 |
+
"step": 2360
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 2.224307836696387,
|
| 1849 |
+
"grad_norm": 0.7499418258666992,
|
| 1850 |
+
"learning_rate": 2.172974513624176e-05,
|
| 1851 |
+
"loss": 1.9194,
|
| 1852 |
+
"step": 2370
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"epoch": 2.2336931018301267,
|
| 1856 |
+
"grad_norm": 0.7999364137649536,
|
| 1857 |
+
"learning_rate": 2.1250929568273774e-05,
|
| 1858 |
+
"loss": 1.9925,
|
| 1859 |
+
"step": 2380
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 2.2430783669638665,
|
| 1863 |
+
"grad_norm": 0.8435518145561218,
|
| 1864 |
+
"learning_rate": 2.0776020619641024e-05,
|
| 1865 |
+
"loss": 1.9746,
|
| 1866 |
+
"step": 2390
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 2.252463632097607,
|
| 1870 |
+
"grad_norm": 0.8621445298194885,
|
| 1871 |
+
"learning_rate": 2.0305082824203343e-05,
|
| 1872 |
+
"loss": 1.8491,
|
| 1873 |
+
"step": 2400
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 2.252463632097607,
|
| 1877 |
+
"eval_loss": 1.9441428184509277,
|
| 1878 |
+
"eval_runtime": 935.9838,
|
| 1879 |
+
"eval_samples_per_second": 1.012,
|
| 1880 |
+
"eval_steps_per_second": 0.506,
|
| 1881 |
+
"step": 2400
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 2.2618488972313466,
|
| 1885 |
+
"grad_norm": 0.8402264714241028,
|
| 1886 |
+
"learning_rate": 1.9838180176193178e-05,
|
| 1887 |
+
"loss": 1.9554,
|
| 1888 |
+
"step": 2410
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"epoch": 2.271234162365087,
|
| 1892 |
+
"grad_norm": 0.8322455883026123,
|
| 1893 |
+
"learning_rate": 1.9375376121519807e-05,
|
| 1894 |
+
"loss": 1.9463,
|
| 1895 |
+
"step": 2420
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 2.2806194274988267,
|
| 1899 |
+
"grad_norm": 0.8249323964118958,
|
| 1900 |
+
"learning_rate": 1.891673354914761e-05,
|
| 1901 |
+
"loss": 1.8215,
|
| 1902 |
+
"step": 2430
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 2.290004692632567,
|
| 1906 |
+
"grad_norm": 0.772278904914856,
|
| 1907 |
+
"learning_rate": 1.8462314782550578e-05,
|
| 1908 |
+
"loss": 1.9064,
|
| 1909 |
+
"step": 2440
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 2.299389957766307,
|
| 1913 |
+
"grad_norm": 0.7047111392021179,
|
| 1914 |
+
"learning_rate": 1.8012181571243097e-05,
|
| 1915 |
+
"loss": 2.0491,
|
| 1916 |
+
"step": 2450
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 2.308775222900047,
|
| 1920 |
+
"grad_norm": 0.9021138548851013,
|
| 1921 |
+
"learning_rate": 1.756639508238922e-05,
|
| 1922 |
+
"loss": 1.9212,
|
| 1923 |
+
"step": 2460
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 2.318160488033787,
|
| 1927 |
+
"grad_norm": 0.9528132677078247,
|
| 1928 |
+
"learning_rate": 1.7125015892490753e-05,
|
| 1929 |
+
"loss": 2.0436,
|
| 1930 |
+
"step": 2470
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 2.327545753167527,
|
| 1934 |
+
"grad_norm": 0.826156735420227,
|
| 1935 |
+
"learning_rate": 1.668810397915568e-05,
|
| 1936 |
+
"loss": 1.9951,
|
| 1937 |
+
"step": 2480
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 2.336931018301267,
|
| 1941 |
+
"grad_norm": 0.7792008519172668,
|
| 1942 |
+
"learning_rate": 1.6255718712948143e-05,
|
| 1943 |
+
"loss": 1.9846,
|
| 1944 |
+
"step": 2490
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.346316283435007,
|
| 1948 |
+
"grad_norm": 0.8387865424156189,
|
| 1949 |
+
"learning_rate": 1.5827918849320567e-05,
|
| 1950 |
+
"loss": 1.9121,
|
| 1951 |
+
"step": 2500
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 2.346316283435007,
|
| 1955 |
+
"eval_loss": 1.9427493810653687,
|
| 1956 |
+
"eval_runtime": 936.6706,
|
| 1957 |
+
"eval_samples_per_second": 1.011,
|
| 1958 |
+
"eval_steps_per_second": 0.506,
|
| 1959 |
+
"step": 2500
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 2.355701548568747,
|
| 1963 |
+
"grad_norm": 0.7230735421180725,
|
| 1964 |
+
"learning_rate": 1.5404762520629724e-05,
|
| 1965 |
+
"loss": 1.8782,
|
| 1966 |
+
"step": 2510
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 2.365086813702487,
|
| 1970 |
+
"grad_norm": 0.7646508812904358,
|
| 1971 |
+
"learning_rate": 1.4986307228237268e-05,
|
| 1972 |
+
"loss": 1.8842,
|
| 1973 |
+
"step": 2520
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 2.3744720788362272,
|
| 1977 |
+
"grad_norm": 0.7921754121780396,
|
| 1978 |
+
"learning_rate": 1.4572609834695971e-05,
|
| 1979 |
+
"loss": 2.1265,
|
| 1980 |
+
"step": 2530
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 2.383857343969967,
|
| 1984 |
+
"grad_norm": 0.8904575109481812,
|
| 1985 |
+
"learning_rate": 1.4163726556023054e-05,
|
| 1986 |
+
"loss": 1.8978,
|
| 1987 |
+
"step": 2540
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 2.3932426091037073,
|
| 1991 |
+
"grad_norm": 0.9155052900314331,
|
| 1992 |
+
"learning_rate": 1.3759712954060921e-05,
|
| 1993 |
+
"loss": 1.8854,
|
| 1994 |
+
"step": 2550
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 2.402627874237447,
|
| 1998 |
+
"grad_norm": 0.8808565139770508,
|
| 1999 |
+
"learning_rate": 1.3360623928927291e-05,
|
| 2000 |
+
"loss": 1.8698,
|
| 2001 |
+
"step": 2560
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 2.4120131393711874,
|
| 2005 |
+
"grad_norm": 0.9437252879142761,
|
| 2006 |
+
"learning_rate": 1.2966513711554744e-05,
|
| 2007 |
+
"loss": 1.7782,
|
| 2008 |
+
"step": 2570
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 2.4213984045049273,
|
| 2012 |
+
"grad_norm": 0.853032112121582,
|
| 2013 |
+
"learning_rate": 1.2577435856321668e-05,
|
| 2014 |
+
"loss": 1.953,
|
| 2015 |
+
"step": 2580
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 2.430783669638667,
|
| 2019 |
+
"grad_norm": 0.8684459328651428,
|
| 2020 |
+
"learning_rate": 1.219344323377482e-05,
|
| 2021 |
+
"loss": 2.2737,
|
| 2022 |
+
"step": 2590
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 2.4401689347724074,
|
| 2026 |
+
"grad_norm": 0.7905233502388,
|
| 2027 |
+
"learning_rate": 1.1814588023444878e-05,
|
| 2028 |
+
"loss": 1.8849,
|
| 2029 |
+
"step": 2600
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 2.4401689347724074,
|
| 2033 |
+
"eval_loss": 1.941327452659607,
|
| 2034 |
+
"eval_runtime": 937.1681,
|
| 2035 |
+
"eval_samples_per_second": 1.01,
|
| 2036 |
+
"eval_steps_per_second": 0.506,
|
| 2037 |
+
"step": 2600
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 2.449554199906147,
|
| 2041 |
+
"grad_norm": 0.769902765750885,
|
| 2042 |
+
"learning_rate": 1.1440921706756092e-05,
|
| 2043 |
+
"loss": 2.057,
|
| 2044 |
+
"step": 2610
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 2.4589394650398875,
|
| 2048 |
+
"grad_norm": 0.8708339333534241,
|
| 2049 |
+
"learning_rate": 1.1072495060030418e-05,
|
| 2050 |
+
"loss": 1.8389,
|
| 2051 |
+
"step": 2620
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 2.4683247301736273,
|
| 2055 |
+
"grad_norm": 0.7585152387619019,
|
| 2056 |
+
"learning_rate": 1.0709358147587884e-05,
|
| 2057 |
+
"loss": 1.9889,
|
| 2058 |
+
"step": 2630
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 2.4777099953073676,
|
| 2062 |
+
"grad_norm": 0.8954083919525146,
|
| 2063 |
+
"learning_rate": 1.0351560314943392e-05,
|
| 2064 |
+
"loss": 2.0466,
|
| 2065 |
+
"step": 2640
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 2.4870952604411074,
|
| 2069 |
+
"grad_norm": 0.8597133755683899,
|
| 2070 |
+
"learning_rate": 9.999150182101319e-06,
|
| 2071 |
+
"loss": 1.8554,
|
| 2072 |
+
"step": 2650
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 2.4964805255748477,
|
| 2076 |
+
"grad_norm": 0.7284257411956787,
|
| 2077 |
+
"learning_rate": 9.652175636948807e-06,
|
| 2078 |
+
"loss": 1.9854,
|
| 2079 |
+
"step": 2660
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 2.5058657907085875,
|
| 2083 |
+
"grad_norm": 0.8639576435089111,
|
| 2084 |
+
"learning_rate": 9.310683828748251e-06,
|
| 2085 |
+
"loss": 1.924,
|
| 2086 |
+
"step": 2670
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 2.5152510558423273,
|
| 2090 |
+
"grad_norm": 0.8123131990432739,
|
| 2091 |
+
"learning_rate": 8.974721161730553e-06,
|
| 2092 |
+
"loss": 1.9737,
|
| 2093 |
+
"step": 2680
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 2.5246363209760676,
|
| 2097 |
+
"grad_norm": 0.8097216486930847,
|
| 2098 |
+
"learning_rate": 8.64433328878917e-06,
|
| 2099 |
+
"loss": 2.0566,
|
| 2100 |
+
"step": 2690
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 2.534021586109808,
|
| 2104 |
+
"grad_norm": 0.8345467448234558,
|
| 2105 |
+
"learning_rate": 8.319565105276678e-06,
|
| 2106 |
+
"loss": 2.0679,
|
| 2107 |
+
"step": 2700
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 2.534021586109808,
|
| 2111 |
+
"eval_loss": 1.9400410652160645,
|
| 2112 |
+
"eval_runtime": 937.8634,
|
| 2113 |
+
"eval_samples_per_second": 1.01,
|
| 2114 |
+
"eval_steps_per_second": 0.505,
|
| 2115 |
+
"step": 2700
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 2.5434068512435477,
|
| 2119 |
+
"grad_norm": 0.8246074318885803,
|
| 2120 |
+
"learning_rate": 8.000460742903987e-06,
|
| 2121 |
+
"loss": 2.0611,
|
| 2122 |
+
"step": 2710
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 2.5527921163772875,
|
| 2126 |
+
"grad_norm": 0.8071450591087341,
|
| 2127 |
+
"learning_rate": 7.687063563743413e-06,
|
| 2128 |
+
"loss": 1.8266,
|
| 2129 |
+
"step": 2720
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 2.562177381511028,
|
| 2133 |
+
"grad_norm": 0.8657225966453552,
|
| 2134 |
+
"learning_rate": 7.379416154336455e-06,
|
| 2135 |
+
"loss": 2.0888,
|
| 2136 |
+
"step": 2730
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 2.5715626466447676,
|
| 2140 |
+
"grad_norm": 0.914188027381897,
|
| 2141 |
+
"learning_rate": 7.077560319906695e-06,
|
| 2142 |
+
"loss": 1.8913,
|
| 2143 |
+
"step": 2740
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 2.580947911778508,
|
| 2147 |
+
"grad_norm": 1.0282979011535645,
|
| 2148 |
+
"learning_rate": 6.781537078679134e-06,
|
| 2149 |
+
"loss": 2.0157,
|
| 2150 |
+
"step": 2750
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 2.5903331769122477,
|
| 2154 |
+
"grad_norm": 0.7493969202041626,
|
| 2155 |
+
"learning_rate": 6.491386656306319e-06,
|
| 2156 |
+
"loss": 2.0123,
|
| 2157 |
+
"step": 2760
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 2.5997184420459876,
|
| 2161 |
+
"grad_norm": 0.8242475986480713,
|
| 2162 |
+
"learning_rate": 6.2071484804021475e-06,
|
| 2163 |
+
"loss": 1.8812,
|
| 2164 |
+
"step": 2770
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 2.609103707179728,
|
| 2168 |
+
"grad_norm": 0.9322003722190857,
|
| 2169 |
+
"learning_rate": 5.928861175184336e-06,
|
| 2170 |
+
"loss": 1.9338,
|
| 2171 |
+
"step": 2780
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 2.618488972313468,
|
| 2175 |
+
"grad_norm": 0.7862038016319275,
|
| 2176 |
+
"learning_rate": 5.656562556225692e-06,
|
| 2177 |
+
"loss": 2.0133,
|
| 2178 |
+
"step": 2790
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 2.627874237447208,
|
| 2182 |
+
"grad_norm": 0.7466335892677307,
|
| 2183 |
+
"learning_rate": 5.3902896253156365e-06,
|
| 2184 |
+
"loss": 1.9908,
|
| 2185 |
+
"step": 2800
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 2.627874237447208,
|
| 2189 |
+
"eval_loss": 1.939355492591858,
|
| 2190 |
+
"eval_runtime": 937.8925,
|
| 2191 |
+
"eval_samples_per_second": 1.01,
|
| 2192 |
+
"eval_steps_per_second": 0.505,
|
| 2193 |
+
"step": 2800
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 2.6372595025809478,
|
| 2197 |
+
"grad_norm": 0.7764114737510681,
|
| 2198 |
+
"learning_rate": 5.13007856543209e-06,
|
| 2199 |
+
"loss": 1.9348,
|
| 2200 |
+
"step": 2810
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 2.646644767714688,
|
| 2204 |
+
"grad_norm": 0.8519911766052246,
|
| 2205 |
+
"learning_rate": 4.875964735824645e-06,
|
| 2206 |
+
"loss": 1.9125,
|
| 2207 |
+
"step": 2820
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 2.656030032848428,
|
| 2211 |
+
"grad_norm": 0.8778414726257324,
|
| 2212 |
+
"learning_rate": 4.627982667209818e-06,
|
| 2213 |
+
"loss": 2.0829,
|
| 2214 |
+
"step": 2830
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 2.665415297982168,
|
| 2218 |
+
"grad_norm": 0.7963272929191589,
|
| 2219 |
+
"learning_rate": 4.386166057078639e-06,
|
| 2220 |
+
"loss": 1.989,
|
| 2221 |
+
"step": 2840
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 2.674800563115908,
|
| 2225 |
+
"grad_norm": 0.790169894695282,
|
| 2226 |
+
"learning_rate": 4.150547765117746e-06,
|
| 2227 |
+
"loss": 1.9568,
|
| 2228 |
+
"step": 2850
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 2.684185828249648,
|
| 2232 |
+
"grad_norm": 0.8890179395675659,
|
| 2233 |
+
"learning_rate": 3.921159808744085e-06,
|
| 2234 |
+
"loss": 2.104,
|
| 2235 |
+
"step": 2860
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 2.693571093383388,
|
| 2239 |
+
"grad_norm": 0.8634796738624573,
|
| 2240 |
+
"learning_rate": 3.698033358754205e-06,
|
| 2241 |
+
"loss": 2.0033,
|
| 2242 |
+
"step": 2870
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 2.7029563585171283,
|
| 2246 |
+
"grad_norm": 0.7101777791976929,
|
| 2247 |
+
"learning_rate": 3.481198735088581e-06,
|
| 2248 |
+
"loss": 1.9882,
|
| 2249 |
+
"step": 2880
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 2.712341623650868,
|
| 2253 |
+
"grad_norm": 0.8689360022544861,
|
| 2254 |
+
"learning_rate": 3.270685402711471e-06,
|
| 2255 |
+
"loss": 1.9517,
|
| 2256 |
+
"step": 2890
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 2.721726888784608,
|
| 2260 |
+
"grad_norm": 0.7733302712440491,
|
| 2261 |
+
"learning_rate": 3.0665219676071057e-06,
|
| 2262 |
+
"loss": 1.9557,
|
| 2263 |
+
"step": 2900
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 2.721726888784608,
|
| 2267 |
+
"eval_loss": 1.9387511014938354,
|
| 2268 |
+
"eval_runtime": 937.6722,
|
| 2269 |
+
"eval_samples_per_second": 1.01,
|
| 2270 |
+
"eval_steps_per_second": 0.506,
|
| 2271 |
+
"step": 2900
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 2.7311121539183483,
|
| 2275 |
+
"grad_norm": 0.8051707148551941,
|
| 2276 |
+
"learning_rate": 2.8687361728924056e-06,
|
| 2277 |
+
"loss": 1.9718,
|
| 2278 |
+
"step": 2910
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 2.740497419052088,
|
| 2282 |
+
"grad_norm": 0.8936708569526672,
|
| 2283 |
+
"learning_rate": 2.6773548950471572e-06,
|
| 2284 |
+
"loss": 2.0474,
|
| 2285 |
+
"step": 2920
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 2.7498826841858284,
|
| 2289 |
+
"grad_norm": 0.7301138639450073,
|
| 2290 |
+
"learning_rate": 2.492404140261795e-06,
|
| 2291 |
+
"loss": 1.9602,
|
| 2292 |
+
"step": 2930
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 2.759267949319568,
|
| 2296 |
+
"grad_norm": 0.7741363644599915,
|
| 2297 |
+
"learning_rate": 2.3139090409034946e-06,
|
| 2298 |
+
"loss": 2.0386,
|
| 2299 |
+
"step": 2940
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 2.768653214453308,
|
| 2303 |
+
"grad_norm": 0.771488606929779,
|
| 2304 |
+
"learning_rate": 2.1418938521010954e-06,
|
| 2305 |
+
"loss": 1.9822,
|
| 2306 |
+
"step": 2950
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 2.7780384795870483,
|
| 2310 |
+
"grad_norm": 0.7883902788162231,
|
| 2311 |
+
"learning_rate": 1.9763819484490355e-06,
|
| 2312 |
+
"loss": 2.0061,
|
| 2313 |
+
"step": 2960
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 2.7874237447207886,
|
| 2317 |
+
"grad_norm": 6.553975582122803,
|
| 2318 |
+
"learning_rate": 1.8173958208311526e-06,
|
| 2319 |
+
"loss": 1.9747,
|
| 2320 |
+
"step": 2970
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 2.7968090098545284,
|
| 2324 |
+
"grad_norm": 0.8313648104667664,
|
| 2325 |
+
"learning_rate": 1.6649570733643982e-06,
|
| 2326 |
+
"loss": 2.0416,
|
| 2327 |
+
"step": 2980
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 2.8061942749882682,
|
| 2331 |
+
"grad_norm": 0.867647647857666,
|
| 2332 |
+
"learning_rate": 1.5190864204631672e-06,
|
| 2333 |
+
"loss": 1.8462,
|
| 2334 |
+
"step": 2990
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 2.8155795401220085,
|
| 2338 |
+
"grad_norm": 0.740106999874115,
|
| 2339 |
+
"learning_rate": 1.3798036840244667e-06,
|
| 2340 |
+
"loss": 1.9627,
|
| 2341 |
+
"step": 3000
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 2.8155795401220085,
|
| 2345 |
+
"eval_loss": 1.9384320974349976,
|
| 2346 |
+
"eval_runtime": 936.8374,
|
| 2347 |
+
"eval_samples_per_second": 1.011,
|
| 2348 |
+
"eval_steps_per_second": 0.506,
|
| 2349 |
+
"step": 3000
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 2.8249648052557483,
|
| 2353 |
+
"grad_norm": 0.7124555706977844,
|
| 2354 |
+
"learning_rate": 1.2471277907343703e-06,
|
| 2355 |
+
"loss": 2.1284,
|
| 2356 |
+
"step": 3010
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 2.8343500703894886,
|
| 2360 |
+
"grad_norm": 0.809935986995697,
|
| 2361 |
+
"learning_rate": 1.1210767694961655e-06,
|
| 2362 |
+
"loss": 2.0541,
|
| 2363 |
+
"step": 3020
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 2.8437353355232284,
|
| 2367 |
+
"grad_norm": 0.801823079586029,
|
| 2368 |
+
"learning_rate": 1.0016677489804171e-06,
|
| 2369 |
+
"loss": 1.9914,
|
| 2370 |
+
"step": 3030
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 2.8531206006569687,
|
| 2374 |
+
"grad_norm": 0.7602815628051758,
|
| 2375 |
+
"learning_rate": 8.88916955297453e-07,
|
| 2376 |
+
"loss": 1.9802,
|
| 2377 |
+
"step": 3040
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 2.8625058657907085,
|
| 2381 |
+
"grad_norm": 0.7440112829208374,
|
| 2382 |
+
"learning_rate": 7.8283970979241e-07,
|
| 2383 |
+
"loss": 1.9312,
|
| 2384 |
+
"step": 3050
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 2.871891130924449,
|
| 2388 |
+
"grad_norm": 2.742288112640381,
|
| 2389 |
+
"learning_rate": 6.834504269632835e-07,
|
| 2390 |
+
"loss": 2.03,
|
| 2391 |
+
"step": 3060
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 2.8812763960581886,
|
| 2395 |
+
"grad_norm": 0.815190315246582,
|
| 2396 |
+
"learning_rate": 5.907626125022159e-07,
|
| 2397 |
+
"loss": 1.9215,
|
| 2398 |
+
"step": 3070
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 2.8906616611919285,
|
| 2402 |
+
"grad_norm": 0.8376407027244568,
|
| 2403 |
+
"learning_rate": 5.04788861460187e-07,
|
| 2404 |
+
"loss": 1.8776,
|
| 2405 |
+
"step": 3080
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 2.9000469263256687,
|
| 2409 |
+
"grad_norm": 0.9095739126205444,
|
| 2410 |
+
"learning_rate": 4.255408565355612e-07,
|
| 2411 |
+
"loss": 1.9567,
|
| 2412 |
+
"step": 3090
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 2.9094321914594086,
|
| 2416 |
+
"grad_norm": 0.858132541179657,
|
| 2417 |
+
"learning_rate": 3.530293664865514e-07,
|
| 2418 |
+
"loss": 1.8339,
|
| 2419 |
+
"step": 3100
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 2.9094321914594086,
|
| 2423 |
+
"eval_loss": 1.9382846355438232,
|
| 2424 |
+
"eval_runtime": 937.0394,
|
| 2425 |
+
"eval_samples_per_second": 1.011,
|
| 2426 |
+
"eval_steps_per_second": 0.506,
|
| 2427 |
+
"step": 3100
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 2.918817456593149,
|
| 2431 |
+
"grad_norm": 0.851668119430542,
|
| 2432 |
+
"learning_rate": 2.872642446678897e-07,
|
| 2433 |
+
"loss": 2.1549,
|
| 2434 |
+
"step": 3110
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 2.9282027217268887,
|
| 2438 |
+
"grad_norm": 1.059181571006775,
|
| 2439 |
+
"learning_rate": 2.2825442769188188e-07,
|
| 2440 |
+
"loss": 2.0087,
|
| 2441 |
+
"step": 3120
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 2.937587986860629,
|
| 2445 |
+
"grad_norm": 1.1176784038543701,
|
| 2446 |
+
"learning_rate": 1.7600793421402307e-07,
|
| 2447 |
+
"loss": 1.9762,
|
| 2448 |
+
"step": 3130
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 2.9469732519943688,
|
| 2452 |
+
"grad_norm": 0.7648433446884155,
|
| 2453 |
+
"learning_rate": 1.305318638434083e-07,
|
| 2454 |
+
"loss": 1.9598,
|
| 2455 |
+
"step": 3140
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 2.956358517128109,
|
| 2459 |
+
"grad_norm": 0.8515041470527649,
|
| 2460 |
+
"learning_rate": 9.183239617795436e-08,
|
| 2461 |
+
"loss": 2.0792,
|
| 2462 |
+
"step": 3150
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 2.965743782261849,
|
| 2466 |
+
"grad_norm": 3.4551055431365967,
|
| 2467 |
+
"learning_rate": 5.991478996468236e-08,
|
| 2468 |
+
"loss": 1.9946,
|
| 2469 |
+
"step": 3160
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 2.9751290473955887,
|
| 2473 |
+
"grad_norm": 0.7828194499015808,
|
| 2474 |
+
"learning_rate": 3.4783382385139565e-08,
|
| 2475 |
+
"loss": 1.9363,
|
| 2476 |
+
"step": 3170
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 2.984514312529329,
|
| 2480 |
+
"grad_norm": 0.9214401245117188,
|
| 2481 |
+
"learning_rate": 1.644158846600963e-08,
|
| 2482 |
+
"loss": 1.9451,
|
| 2483 |
+
"step": 3180
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 2.9938995776630692,
|
| 2487 |
+
"grad_norm": 0.8548945188522339,
|
| 2488 |
+
"learning_rate": 4.8919006150727195e-09,
|
| 2489 |
+
"loss": 1.8411,
|
| 2490 |
+
"step": 3190
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 2.998592210229939,
|
| 2494 |
+
"step": 3195,
|
| 2495 |
+
"total_flos": 1.0917373877893988e+19,
|
| 2496 |
+
"train_loss": 2.069366264641751,
|
| 2497 |
+
"train_runtime": 97122.8592,
|
| 2498 |
+
"train_samples_per_second": 0.263,
|
| 2499 |
+
"train_steps_per_second": 0.033
|
| 2500 |
+
}
|
| 2501 |
+
],
|
| 2502 |
+
"logging_steps": 10,
|
| 2503 |
+
"max_steps": 3195,
|
| 2504 |
+
"num_input_tokens_seen": 0,
|
| 2505 |
+
"num_train_epochs": 3,
|
| 2506 |
+
"save_steps": 500,
|
| 2507 |
+
"total_flos": 1.0917373877893988e+19,
|
| 2508 |
+
"train_batch_size": 2,
|
| 2509 |
+
"trial_name": null,
|
| 2510 |
+
"trial_params": null
|
| 2511 |
+
}
|
training_eval_loss.png
ADDED
|
training_loss.png
ADDED
|